Scintilla

Scintilla Documentation

Last edited 20/June/2007 NH

There is an overview of the internal design of Scintilla.
Some notes on using Scintilla.
How to use the Scintilla Edit Control on Windows.
A simple sample using Scintilla from C++ on Windows.
A simple sample using Scintilla from Visual Basic.
Bait is a tiny sample using Scintilla on GTK+.
A detailed description of how to write a lexer, including a discussion of folding.
How to implement a lexer in the container.
How to implement folding.
The coding style used in Scintilla and SciTE is worth following if you want to contribute code to Scintilla but is not compulsory.

Introduction

The Windows version of Scintilla is a Windows Control. As such, its primary programming interface is through Windows messages. Early versions of Scintilla emulated much of the API defined by the standard Windows Edit and RichEdit controls but those APIs are now deprecated in favour of Scintilla's own, more consistent API. In addition to messages performing the actions of a normal Edit control, Scintilla allows control of syntax styling, folding, markers, autocompletion and call tips.

The GTK+ version also uses messages in a similar way to the Windows version. This is different to normal GTK+ practice but made it easier to implement rapidly.

This documentation describes the individual messages and notifications used by Scintilla. It does not describe how to link them together to form a useful editor. For now, the best way to work out how to develop using Scintilla is to see how SciTE uses it. SciTE exercises most of Scintilla's facilities.

In the descriptions that follow, the messages are described as function calls with zero, one or two arguments. These two arguments are the standard wParam and lParam familiar to Windows programmers. These parameters are integers that are large enough to hold pointers, and the return value is also an integer large enough to contain a pointer. Although the commands only use the arguments described, because all messages have two arguments whether Scintilla uses them or not, it is strongly recommended that any unused arguments are set to 0. This allows future enhancement of messages without the risk of breaking existing code. Common argument types are:

Contents

Messages with names of the form SCI_SETxxxxx often have a companion SCI_GETxxxxx. To save tedious repetition, if the SCI_GETxxxxx message returns the value set by the SCI_SETxxxxx message, the SET routine is described and the GET routine is left to your imagination.

Text retrieval and modification

Each character in a Scintilla document is followed by an associated byte of styling information. The combination of a character byte and a style byte is called a cell. Style bytes are interpreted an index into an array of styles. Style bytes may be split into an index and a set of indicator bits but this use is discouraged and indicators should now use SCI_INDICATORFILLRANGE and related calls. The default split is with the index in the low 5 bits and 3 high bits as indicators. This allows 32 fundamental styles, which is enough for most languages, and three independent indicators so that, for example, syntax errors, deprecated names and bad indentation could all be displayed at once. The number of bits used for styles can be altered with SCI_SETSTYLEBITS up to a maximum of 7 bits. The remaining bits can be used for indicators.

Positions within the Scintilla document refer to a character or the gap before that character. The first character in a document is 0, the second 1 and so on. If a document contains nLen characters, the last character is numbered nLen-1. The caret exists between character positions and can be located from before the first character (0) to after the last character (nLen).

There are places where the caret can not go where two character bytes make up one character. This occurs when a DBCS character from a language like Japanese is included in the document or when line ends are marked with the CP/M standard of a carriage return followed by a line feed. The INVALID_POSITION constant (-1) represents an invalid position within the document.

All lines of text in Scintilla are the same height, and this height is calculated from the largest font in any current style. This restriction is for performance; if lines differed in height then calculations involving positioning of text would require the text to be styled first.

SCI_GETTEXT(int length, char *text)
This returns length-1 characters of text from the start of the document plus one terminating 0 character. To collect all the text in a document, use SCI_GETLENGTH to get the number of characters in the document (nLen), allocate a character buffer of length nLen+1 bytes, then call SCI_GETTEXT(nLen+1, char *text). If the text argument is 0 then the length that should be allocated to store the entire document is returned. If you then save the text, you should use SCI_SETSAVEPOINT to mark the text as unmodified.

See also: SCI_GETSELTEXT, SCI_GETCURLINE, SCI_GETLINE, SCI_GETSTYLEDTEXT, SCI_GETTEXTRANGE

SCI_SETTEXT(<unused>, const char *text)
This replaces all the text in the document with the zero terminated text string you pass in.

SCI_SETSAVEPOINT
This message tells Scintilla that the current state of the document is unmodified. This is usually done when the file is saved or loaded, hence the name "save point". As Scintilla performs undo and redo operations, it notifies the container that it has entered or left the save point with SCN_SAVEPOINTREACHED and SCN_SAVEPOINTLEFT notification messages, allowing the container to know if the file should be considered dirty or not.

See also: SCI_EMPTYUNDOBUFFER, SCI_GETMODIFY

SCI_GETLINE(int line, char *text)
This fills the buffer defined by text with the contents of the nominated line (lines start at 0). The buffer is not terminated by a 0 character. It is up to you to make sure that the buffer is long enough for the text, use SCI_LINELENGTH(int line). The returned value is the number of characters copied to the buffer. The returned text includes any end of line characters. If you ask for a line number outside the range of lines in the document, 0 characters are copied. If the text argument is 0 then the length that should be allocated to store the entire line is returned.

See also: SCI_GETCURLINE, SCI_GETSELTEXT, SCI_GETTEXTRANGE, SCI_GETSTYLEDTEXT, SCI_GETTEXT

SCI_REPLACESEL(<unused>, const char *text)
The currently selected text between the anchor and the current position is replaced by the 0 terminated text string. If the anchor and current position are the same, the text is inserted at the caret position. The caret is positioned after the inserted text and the caret is scrolled into view.

SCI_SETREADONLY(bool readOnly)
SCI_GETREADONLY
These messages set and get the read-only flag for the document. If you mark a document as read only, attempts to modify the text cause the SCN_MODIFYATTEMPTRO notification.

SCI_GETTEXTRANGE(<unused>, TextRange *tr)
This collects the text between the positions cpMin and cpMax and copies it to lpstrText (see struct TextRange in Scintilla.h). If cpMax is -1, text is returned to the end of the document. The text is 0 terminated, so you must supply a buffer that is at least 1 character longer than the number of characters you wish to read. The return value is the length of the returned text not including the terminating 0.

See also: SCI_GETSELTEXT, SCI_GETLINE, SCI_GETCURLINE, SCI_GETSTYLEDTEXT, SCI_GETTEXT

SCI_GETSTYLEDTEXT(<unused>, TextRange *tr)
This collects styled text into a buffer using two bytes for each cell, with the character at the lower address of each pair and the style byte at the upper address. Characters between the positions cpMin and cpMax are copied to lpstrText (see struct TextRange in Scintilla.h). Two 0 bytes are added to the end of the text, so the buffer that lpstrText points at must be at least 2*(cpMax-cpMin)+2 bytes long. No check is made for sensible values of cpMin or cpMax. Positions outside the document return character codes and style bytes of 0.

See also: SCI_GETSELTEXT, SCI_GETLINE, SCI_GETCURLINE, SCI_GETTEXTRANGE, SCI_GETTEXT

SCI_ALLOCATE(int bytes, <unused>)
Allocate a document buffer large enough to store a given number of bytes. The document will not be made smaller than its current contents.

SCI_ADDTEXT(int length, const char *s)
This inserts the first length characters from the string s at the current position. This will include any 0's in the string that you might have expected to stop the insert operation. The current position is set at the end of the inserted text, but it is not scrolled into view.

SCI_ADDSTYLEDTEXT(int length, cell *s)
This behaves just like SCI_ADDTEXT, but inserts styled text.

SCI_APPENDTEXT(int length, const char *s)
This adds the first length characters from the string s to the end of the document. This will include any 0's in the string that you might have expected to stop the operation. The current selection is not changed and the new text is not scrolled into view.

SCI_INSERTTEXT(int pos, const char *text)
This inserts the zero terminated text string at position pos or at the current position if pos is -1. If the current position is after the insertion point then it is moved along with its surrounding text but no scrolling is performed.

SCI_CLEARALL
Unless the document is read-only, this deletes all the text.

SCI_CLEARDOCUMENTSTYLE
When wanting to completely restyle the document, for example after choosing a lexer, the SCI_CLEARDOCUMENTSTYLE can be used to clear all styling information and reset the folding state.

SCI_GETCHARAT(int pos)
This returns the character at pos in the document or 0 if pos is negative or past the end of the document.

SCI_GETSTYLEAT(int pos)
This returns the style at pos in the document, or 0 if pos is negative or past the end of the document.

SCI_SETSTYLEBITS(int bits)
SCI_GETSTYLEBITS
This pair of routines sets and reads back the number of bits in each cell to use for styling, to a maximum of 7 style bits. The remaining bits can be used as indicators. The standard setting is SCI_SETSTYLEBITS(5). The number of styling bits needed by the current lexer can be found with SCI_GETSTYLEBITSNEEDED.

TextRange and CharacterRange
These structures are defined to be exactly the same shape as the Win32 TEXTRANGE and CHARRANGE, so that older code that treats Scintilla as a RichEdit will work.

struct CharacterRange {
    long cpMin;
    long cpMax;
};

struct TextRange {
    struct CharacterRange chrg;
    char *lpstrText;
};

GTK+-specific: Access to encoded text

SCI_TARGETASUTF8(<unused>, char *s)
This method retrieves the value of the target encoded as UTF-8 which is the default encoding of GTK+ so is useful for retrieving text for use in other parts of the user interface, such as find and replace dialogs. The length of the encoded text in bytes is returned.

SCI_ENCODEDFROMUTF8(const char *utf8, char *encoded)
SCI_SETLENGTHFORENCODE(int bytes)
SCI_ENCODEDFROMUTF8 converts a UTF-8 string into the document's encoding which is useful for taking the results of a find dialog, for example, and receiving a string of bytes that can be searched for in the document. Since the text can contain nul bytes, the SCI_SETLENGTHFORENCODE method can be used to set the length that will be converted. If set to -1, the length is determined by finding a nul byte. The length of the converted string is returned.

Searching

There are methods to search for text and for regular expressions. The regular expression support is limited and should only be used for simple cases and initial development. A different regular expression library can be integrated into Scintilla or can be called from the container using direct access to the buffer contents through SCI_GETCHARACTERPOINTER.

searchFlags
Several of the search routines use flag options, which include a simple regular expression search. Combine the flag options by adding them:

If SCFIND_REGEXP is not included in the searchFlags, you can search backwards to find the previous occurrence of a search string by setting the end of the search range before the start. If SCFIND_REGEXP is included, searches are always from a lower position to a higher position, even if the search range is backwards.

In a regular expression, special characters interpreted are:

SCI_FINDTEXT(int searchFlags, TextToFind *ttf)
This message searches for text in the document. It does not use or move the current selection. The searchFlags argument controls the search type, which includes regular expression searches.

The TextToFind structure is defined in Scintilla.h; set chrg.cpMin and chrg.cpMax with the range of positions in the document to search. If SCFIND_REGEXP is not included in the flags, you can search backwards by setting chrg.cpMax less than chrg.cpMin. If SCFIND_REGEXP is included, the search is always forwards (even if chrg.cpMax is less than chrg.cpMin). Set the lpstrText member of TextToFind to point at a zero terminated text string holding the search pattern. If your language makes the use of TextToFind difficult, you should consider using SCI_SEARCHINTARGET instead.

The return value is -1 if the search fails or the position of the start of the found text if it succeeds. The chrgText.cpMin and chrgText.cpMax members of TextToFind are filled in with the start and end positions of the found text.

See also: SCI_SEARCHINTARGET

TextToFind
This structure is defined to have exactly the same shape as the Win32 structure FINDTEXTEX for old code that treated Scintilla as a RichEdit control.

struct TextToFind {
    struct CharacterRange chrg;     // range to search
    char *lpstrText;                // the search pattern (zero terminated)
    struct CharacterRange chrgText; // returned as position of matching text
};

SCI_SEARCHANCHOR
SCI_SEARCHNEXT(int searchFlags, const char *text)
SCI_SEARCHPREV(int searchFlags, const char *text)
These messages provide relocatable search support. This allows multiple incremental interactive searches to be macro recorded while still setting the selection to found text so the find/select operation is self-contained. These three messages send SCN_MACRORECORD notifications if macro recording is enabled.

SCI_SEARCHANCHOR sets the search start point used by SCI_SEARCHNEXT and SCI_SEARCHPREV to the start of the current selection, that is, the end of the selection that is nearer to the start of the document. You should always call this before calling either of SCI_SEARCHNEXT or SCI_SEARCHPREV.

SCI_SEARCHNEXT and SCI_SEARCHPREV search for the next and previous occurrence of the zero terminated search string pointed at by text. The search is modified by the searchFlags. If you request a regular expression, SCI_SEARCHPREV finds the first occurrence of the search string in the document, not the previous one before the anchor point.

The return value is -1 if nothing is found, otherwise the return value is the start position of the matching text. The selection is updated to show the matched text, but is not scrolled into view.

See also: SCI_SEARCHINTARGET, SCI_FINDTEXT

Search and replace using the target

Using SCI_REPLACESEL, modifications cause scrolling and other visible changes, which may take some time and cause unwanted display updates. If performing many changes, such as a replace all command, the target can be used instead. First, set the target, ie. the range to be replaced. Then call SCI_REPLACETARGET or SCI_REPLACETARGETRE.

Searching can be performed within the target range with SCI_SEARCHINTARGET, which uses a counted string to allow searching for null characters. It returns the length of range or -1 for failure, in which case the target is not moved. The flags used by SCI_SEARCHINTARGET such as SCFIND_MATCHCASE, SCFIND_WHOLEWORD, SCFIND_WORDSTART, and SCFIND_REGEXP can be set with SCI_SETSEARCHFLAGS. SCI_SEARCHINTARGET may be simpler for some clients to use than SCI_FINDTEXT, as that requires using a pointer to a structure.

SCI_SETTARGETSTART(int pos)
SCI_GETTARGETSTART
SCI_SETTARGETEND(int pos)
SCI_GETTARGETEND
These functions set and return the start and end of the target. When searching in non-regular expression mode, you can set start greater than end to find the last matching text in the target rather than the first matching text. The target is also set by a successful SCI_SEARCHINTARGET.

SCI_TARGETFROMSELECTION
Set the target start and end to the start and end positions of the selection.

SCI_SETSEARCHFLAGS(int searchFlags)
SCI_GETSEARCHFLAGS
These get and set the searchFlags used by SCI_SEARCHINTARGET. There are several option flags including a simple regular expression search.

SCI_SEARCHINTARGET(int length, const char *text)
This searches for the first occurrence of a text string in the target defined by SCI_SETTARGETSTART and SCI_SETTARGETEND. The text string is not zero terminated; the size is set by length. The search is modified by the search flags set by SCI_SETSEARCHFLAGS. If the search succeeds, the target is set to the found text and the return value is the position of the start of the matching text. If the search fails, the result is -1.

SCI_REPLACETARGET(int length, const char *text)
If length is -1, text is a zero terminated string, otherwise length sets the number of character to replace the target with. After replacement, the target range refers to the replacement text. The return value is the length of the replacement string.
Note that the recommended way to delete text in the document is to set the target to the text to be removed, and to perform a replace target with an empty string.

SCI_REPLACETARGETRE(int length, const char *text)
This replaces the target using regular expressions. If length is -1, text is a zero terminated string, otherwise length is the number of characters to use. The replacement string is formed from the text string with any sequences of \1 through \9 replaced by tagged matches from the most recent regular expression search. After replacement, the target range refers to the replacement text. The return value is the length of the replacement string.

See also: SCI_FINDTEXT

Overtype

SCI_SETOVERTYPE(bool overType)
SCI_GETOVERTYPE
When overtype is enabled, each typed character replaces the character to the right of the text caret. When overtype is disabled, characters are inserted at the caret. SCI_GETOVERTYPE returns TRUE (1) if overtyping is active, otherwise FALSE (0) will be returned. Use SCI_SETOVERTYPE to set the overtype mode.

Cut, copy and paste

SCI_CUT
SCI_COPY
SCI_PASTE
SCI_CLEAR
SCI_CANPASTE
SCI_COPYALLOWLINE
These commands perform the standard tasks of cutting and copying data to the clipboard, pasting from the clipboard into the document, and clearing the document. SCI_CANPASTE returns non-zero if the document isn't read-only and if the selection doesn't contain protected text. If you need a "can copy" or "can cut", use SCI_GETSELECTIONSTART()-SCI_GETSELECTIONEND(), which will be non-zero if you can copy or cut to the clipboard.

GTK+ does not really support SCI_CANPASTE and always returns TRUE unless the document is read-only.

On X, the clipboard is asynchronous and may require several messages between the destination and source applications. Data from SCI_PASTE will not arrive in the document immediately.

SCI_COPYALLOWLINE works the same as SCI_COPY except that if the selection is empty then the current line is copied. On Windows, an extra "MSDEVLineSelect" marker is added to the clipboard which is then used in SCI_PASTE to paste the whole line before the current line.

SCI_COPYRANGE copies a range of text from the document to the system clipboard and SCI_COPYTEXT copies a supplied piece of text to the system clipboard.

SCI_SETPASTECONVERTENDINGS(bool convert)
SCI_GETPASTECONVERTENDINGS
If this property is set then when text is pasted any line ends are converted to match the document's end of line mode as set with SCI_SETEOLMODE. Currently only changeable on Windows. On GTK+ pasted text is always converted.

Error handling

SCI_SETSTATUS(int status)
SCI_GETSTATUS
If an error occurs, Scintilla may set an internal error number that can be retrieved with SCI_GETSTATUS. Not currently used but will be in the future. To clear the error status call SCI_SETSTATUS(0).

Undo and Redo

Scintilla has multiple level undo and redo. It will continue to collect undoable actions until memory runs out. Scintilla saves actions that change the document. Scintilla does not save caret and selection movements, view scrolling and the like. Sequences of typing or deleting are compressed into single transactions to make it easier to undo and redo at a sensible level of detail. Sequences of actions can be combined into transactions that are undone as a unit. These sequences occur between SCI_BEGINUNDOACTION and SCI_ENDUNDOACTION messages. These transactions can be nested and only the top-level sequences are undone as units.

SCI_UNDO
SCI_CANUNDO
SCI_UNDO undoes one action, or if the undo buffer has reached a SCI_ENDUNDOACTION point, all the actions back to the corresponding SCI_BEGINUNDOACTION.

SCI_CANUNDO returns 0 if there is nothing to undo, and 1 if there is. You would typically use the result of this message to enable/disable the Edit menu Undo command.

SCI_REDO
SCI_CANREDO
SCI_REDO undoes the effect of the last SCI_UNDO operation.

SCI_CANREDO returns 0 if there is no action to redo and 1 if there are undo actions to redo. You could typically use the result of this message to enable/disable the Edit menu Redo command.

SCI_EMPTYUNDOBUFFER
This command tells Scintilla to forget any saved undo or redo history. It also sets the save point to the start of the undo buffer, so the document will appear to be unmodified. This does not cause the SCN_SAVEPOINTREACHED notification to be sent to the container.

See also: SCI_SETSAVEPOINT

SCI_SETUNDOCOLLECTION(bool collectUndo)
SCI_GETUNDOCOLLECTION
You can control whether Scintilla collects undo information with SCI_SETUNDOCOLLECTION. Pass in true (1) to collect information and false (0) to stop collecting. If you stop collection, you should also use SCI_EMPTYUNDOBUFFER to avoid the undo buffer being unsynchronized with the data in the buffer.

You might wish to turn off saving undo information if you use the Scintilla to store text generated by a program (a Log view) or in a display window where text is often deleted and regenerated.

SCI_BEGINUNDOACTION
SCI_ENDUNDOACTION
Send these two messages to Scintilla to mark the beginning and end of a set of operations that you want to undo all as one operation but that you have to generate as several operations. Alternatively, you can use these to mark a set of operations that you do not want to have combined with the preceding or following operations if they are undone.

Selection and information

Scintilla maintains a selection that stretches between two points, the anchor and the current position. If the anchor and the current position are the same, there is no selected text. Positions in the document range from 0 (before the first character), to the document size (after the last character). If you use messages, there is nothing to stop you setting a position that is in the middle of a CRLF pair, or in the middle of a 2 byte character. However, keyboard commands will not move the caret into such positions.

SCI_GETTEXTLENGTH
SCI_GETLENGTH
Both these messages return the length of the document in characters.

SCI_GETLINECOUNT
This returns the number of lines in the document. An empty document contains 1 line. A document holding only an end of line sequence has 2 lines.

SCI_GETFIRSTVISIBLELINE
This returns the line number of the first visible line in the Scintilla view. The first line in the document is numbered 0. The value is a visible line rather than a document line.

SCI_LINESONSCREEN
This returns the number of complete lines visible on the screen. With a constant line height, this is the vertical space available divided by the line separation. Unless you arrange to size your window to an integral number of lines, there may be a partial line visible at the bottom of the view.

SCI_GETMODIFY
This returns non-zero if the document is modified and 0 if it is unmodified. The modified status of a document is determined by the undo position relative to the save point. The save point is set by SCI_SETSAVEPOINT, usually when you have saved data to a file.

If you need to be notified when the document becomes modified, Scintilla notifies the container that it has entered or left the save point with the SCN_SAVEPOINTREACHED and SCN_SAVEPOINTLEFT notification messages.

SCI_SETSEL(int anchorPos, int currentPos)
This message sets both the anchor and the current position. If currentPos is negative, it means the end of the document. If anchorPos is negative, it means remove any selection (i.e. set the anchor to the same position as currentPos). The caret is scrolled into view after this operation.

SCI_GOTOPOS(int pos)
This removes any selection, sets the caret at pos and scrolls the view to make the caret visible, if necessary. It is equivalent to SCI_SETSEL(pos, pos). The anchor position is set the same as the current position.

SCI_GOTOLINE(int line)
This removes any selection and sets the caret at the start of line number line and scrolls the view (if needed) to make it visible. The anchor position is set the same as the current position. If line is outside the lines in the document (first line is 0), the line set is the first or last.

SCI_SETCURRENTPOS(int pos)
This sets the current position and creates a selection between the anchor and the current position. The caret is not scrolled into view.

See also: SCI_SCROLLCARET

SCI_GETCURRENTPOS
This returns the current position.

SCI_SETANCHOR(int pos)
This sets the anchor position and creates a selection between the anchor position and the current position. The caret is not scrolled into view.

See also: SCI_SCROLLCARET

SCI_GETANCHOR
This returns the current anchor position.

SCI_SETSELECTIONSTART(int pos)
SCI_SETSELECTIONEND(int pos)
These set the selection based on the assumption that the anchor position is less than the current position. They do not make the caret visible. The table shows the positions of the anchor and the current position after using these messages.

See also: SCI_SCROLLCARET

SCI_GETSELECTIONSTART
SCI_GETSELECTIONEND
These return the start and end of the selection without regard to which end is the current position and which is the anchor. SCI_GETSELECTIONSTART returns the smaller of the current position or the anchor position. SCI_GETSELECTIONEND returns the larger of the two values.

SCI_SELECTALL
This selects all the text in the document. The current position is not scrolled into view.

SCI_LINEFROMPOSITION(int pos)
This message returns the line that contains the position pos in the document. The return value is 0 if pos <= 0. The return value is the last line if pos is beyond the end of the document.

SCI_POSITIONFROMLINE(int line)
This returns the document position that corresponds with the start of the line. If line is negative, the position of the line holding the start of the selection is returned. If line is greater than the lines in the document, the return value is -1. If line is equal to the number of lines in the document (i.e. 1 line past the last line), the return value is the end of the document.

SCI_GETLINEENDPOSITION(int line)
This returns the position at the end of the line, before any line end characters. If line is the last line in the document (which does not have any end of line characters), the result is the size of the document. If line is negative or line >= SCI_GETLINECOUNT(), the result is undefined.

SCI_LINELENGTH(int line)
This returns the length of the line, including any line end characters. If line is negative or beyond the last line in the document, the result is 0. If you want the length of the line not including any end of line characters, use SCI_GETLINEENDPOSITION(line) - SCI_POSITIONFROMLINE(line).

See also: SCI_GETCURLINE, SCI_GETLINE, SCI_GETTEXT, SCI_GETSTYLEDTEXT, SCI_GETTEXTRANGE

SCI_GETCURLINE(int textLen, char *text)
This retrieves the text of the line containing the caret and returns the position within the line of the caret. Pass in char* text pointing at a buffer large enough to hold the text you wish to retrieve and a terminating 0 character. Set textLen to the length of the buffer which must be at least 1 to hold the terminating 0 character. If the text argument is 0 then the length that should be allocated to store the entire current line is returned.

See also: SCI_GETSELTEXT, SCI_GETLINE, SCI_GETTEXT, SCI_GETSTYLEDTEXT, SCI_GETTEXTRANGE

SCI_SELECTIONISRECTANGLE
This returns 1 if the current selection is in rectangle mode, 0 if not.

SCI_SETSELECTIONMODE(int mode)
SCI_GETSELECTIONMODE
The two functions set and get the selection mode, which can be stream (SC_SEL_STREAM=0) or rectangular (SC_SEL_RECTANGLE=1) or by lines (SC_SEL_LINES=2). When set in these modes, regular caret moves will extend or reduce the selection, until the mode is cancelled by a call with same value or with SCI_CANCEL. The get function returns the current mode even if the selection was made by mouse or with regular extended moves.

SCI_GETLINESELSTARTPOSITION(int line)
SCI_GETLINESELENDPOSITION(int line)
Retrieve the position of the start and end of the selection at the given line with INVALID_POSITION returned if no selection on this line.

SCI_MOVECARETINSIDEVIEW
If the caret is off the top or bottom of the view, it is moved to the nearest line that is visible to its current position. Any selection is lost.

SCI_WORDENDPOSITION(int position, bool onlyWordCharacters)
SCI_WORDSTARTPOSITION(int position, bool onlyWordCharacters)
These messages return the start and end of words using the same definition of words as used internally within Scintilla. You can set your own list of characters that count as words with SCI_SETWORDCHARS. The position sets the start or the search, which is forwards when searching for the end and backwards when searching for the start.

Set onlyWordCharacters to true (1) to stop searching at the first non-word character in the search direction. If onlyWordCharacters is false (0), the first character in the search direction sets the type of the search as word or non-word and the search stops at the first non-matching character. Searches are also terminated by the start or end of the document.

If "w" represents word characters and "." represents non-word characters and "|" represents the position and true or false is the state of onlyWordCharacters:

SCI_POSITIONBEFORE(int position)
SCI_POSITIONAFTER(int position)
These messages return the position before and after another position in the document taking into account the current code page. The minimum position returned is 0 and the maximum is the last position in the document. If called with a position within a multi byte character will return the position of the start/end of that character.

SCI_TEXTWIDTH(int styleNumber, const char *text)
This returns the pixel width of a string drawn in the given styleNumber which can be used, for example, to decide how wide to make the line number margin in order to display a given number of numerals.

SCI_TEXTHEIGHT(int line)
This returns the height in pixels of a particular line. Currently all lines are the same height.

SCI_GETCOLUMN(int pos)
This message returns the column number of a position pos within the document taking the width of tabs into account. This returns the column number of the last tab on the line before pos, plus the number of characters between the last tab and pos. If there are no tab characters on the line, the return value is the number of characters up to the position on the line. In both cases, double byte characters count as a single character. This is probably only useful with monospaced fonts.

SCI_FINDCOLUMN(int line, int column)
This message returns the position of a column on a line taking the width of tabs into account. It treats a multi-byte character as a single column. Column numbers, like lines start at 0.

SCI_POSITIONFROMPOINT(int x, int y)
SCI_POSITIONFROMPOINTCLOSE(int x, int y)
SCI_POSITIONFROMPOINT finds the closest character position to a point and SCI_POSITIONFROMPOINTCLOSE is similar but returns -1 if the point is outside the window or not close to any characters.

SCI_POINTXFROMPOSITION(<unused>, int pos)
SCI_POINTYFROMPOSITION(<unused>, int pos)
These messages return the x and y display pixel location of text at position pos in the document.

SCI_HIDESELECTION(bool hide)
The normal state is to make the selection visible by drawing it as set by SCI_SETSELFORE and SCI_SETSELBACK. However, if you hide the selection, it is drawn as normal text.

SCI_CHOOSECARETX
Scintilla remembers the x value of the last position horizontally moved to explicitly by the user and this value is then used when moving vertically such as by using the up and down keys. This message sets the current x position of the caret as the remembered value.

Scrolling and automatic scrolling

SCI_LINESCROLL(int column, int line)
This will attempt to scroll the display by the number of columns and lines that you specify. Positive line values increase the line number at the top of the screen (i.e. they move the text upwards as far as the user is concerned), Negative line values do the reverse.

The column measure is the width of a space in the default style. Positive values increase the column at the left edge of the view (i.e. they move the text leftwards as far as the user is concerned). Negative values do the reverse.

See also: SCI_SETXOFFSET

SCI_SCROLLCARET
If the current position (this is the caret if there is no selection) is not visible, the view is scrolled to make it visible according to the current caret policy.

SCI_SETXCARETPOLICY(int caretPolicy, int caretSlop)
SCI_SETYCARETPOLICY(int caretPolicy, int caretSlop)
These set the caret policy. The value of caretPolicy is a combination of CARET_SLOP, CARET_STRICT, CARET_JUMPS and CARET_EVEN.

SCI_SETVISIBLEPOLICY(int caretPolicy, int caretSlop)
This determines how the vertical positioning is determined when SCI_ENSUREVISIBLEENFORCEPOLICY is called. It takes VISIBLE_SLOP and VISIBLE_STRICT flags for the policy parameter. It is similar in operation to SCI_SETYCARETPOLICY(int caretPolicy, int caretSlop).

SCI_SETHSCROLLBAR(bool visible)
SCI_GETHSCROLLBAR
The horizontal scroll bar is only displayed if it is needed for the assumed width. If you never wish to see it, call SCI_SETHSCROLLBAR(0). Use SCI_SETHSCROLLBAR(1) to enable it again. SCI_GETHSCROLLBAR returns the current state. The default state is to display it when needed.

See also: SCI_SETSCROLLWIDTH.

SCI_SETVSCROLLBAR(bool visible)
SCI_GETVSCROLLBAR
By default, the vertical scroll bar is always displayed when required. You can choose to hide or show it with SCI_SETVSCROLLBAR and get the current state with SCI_GETVSCROLLBAR.

SCI_SETXOFFSET(int xOffset)
SCI_GETXOFFSET
The xOffset is the horizontal scroll position in pixels of the start of the text view. A value of 0 is the normal position with the first text column visible at the left of the view.

See also: SCI_LINESCROLL

SCI_SETSCROLLWIDTH(int pixelWidth)
SCI_GETSCROLLWIDTH
For performance, Scintilla does not measure the display width of the document to determine the properties of the horizontal scroll bar. Instead, an assumed width is used. These messages set and get the document width in pixels assumed by Scintilla. The default value is 2000. To ensure the width of the currently visible lines can be scrolled use SCI_SETSCROLLWIDTHTRACKING

SCI_SETSCROLLWIDTHTRACKING(bool tracking)
SCI_GETSCROLLWIDTHTRACKING
If scroll width tracking is enabled then the scroll width is adjusted to ensure that all of the lines currently displayed can be completely scrolled. This mode never adjusts the scroll width to be narrower.

SCI_SETENDATLASTLINE(bool endAtLastLine)
SCI_GETENDATLASTLINE
SCI_SETENDATLASTLINE sets the scroll range so that maximum scroll position has the last line at the bottom of the view (default). Setting this to false allows scrolling one page below the last line.

White space

SCI_SETVIEWWS(int wsMode)
SCI_GETVIEWWS
White space can be made visible which may useful for languages in which white space is significant, such as Python. Space characters appear as small centred dots and tab characters as light arrows pointing to the right. There are also ways to control the display of end of line characters. The two messages set and get the white space display mode. The wsMode argument can be one of:

The effect of using any other wsMode value is undefined.

SCI_SETWHITESPACEFORE(bool useWhitespaceForeColour, int colour)
SCI_SETWHITESPACEBACK(bool useWhitespaceBackColour, int colour)
By default, the colour of visible white space is determined by the lexer in use. The foreground and/or background colour of all visible white space can be set globally, overriding the lexer's colours with SCI_SETWHITESPACEFORE and SCI_SETWHITESPACEBACK.

Cursor

SCI_SETCURSOR(int curType)
SCI_GETCURSOR
The cursor is normally chosen in a context sensitive way, so it will be different over the margin than when over the text. When performing a slow action, you may wish to change to a wait cursor. You set the cursor type with SCI_SETCURSOR. The curType argument can be:

Cursor values 1 through 7 have defined cursors, but only SC_CURSORWAIT is usefully controllable. Other values of curType cause a pointer to be displayed. The SCI_GETCURSOR message returns the last cursor type you set, or SC_CURSORNORMAL (-1) if you have not set a cursor type.

Mouse capture

SCI_SETMOUSEDOWNCAPTURES(bool captures)
SCI_GETMOUSEDOWNCAPTURES
When the mouse is pressed inside Scintilla, it is captured so future mouse movement events are sent to Scintilla. This behavior may be turned off with SCI_SETMOUSEDOWNCAPTURES(0).

Line endings

Scintilla can interpret any of the three major line end conventions, Macintosh (\r), Unix (\n) and CP/M / DOS / Windows (\r\n). When the user presses the Enter key, one of these line end strings is inserted into the buffer. The default is \r\n in Windows and \n in Unix, but this can be changed with the SCI_SETEOLMODE message. You can also convert the entire document to one of these line endings with SCI_CONVERTEOLS. Finally, you can choose to display the line endings with SCI_SETVIEWEOL.

SCI_SETEOLMODE(int eolMode)
SCI_GETEOLMODE
SCI_SETEOLMODE sets the characters that are added into the document when the user presses the Enter key. You can set eolMode to one of SC_EOL_CRLF (0), SC_EOL_CR (1), or SC_EOL_LF (2). The SCI_GETEOLMODE message retrieves the current state.

SCI_CONVERTEOLS(int eolMode)
This message changes all the end of line characters in the document to match eolMode. Valid values are: SC_EOL_CRLF (0), SC_EOL_CR (1), or SC_EOL_LF (2).

SCI_SETVIEWEOL(bool visible)
SCI_GETVIEWEOL
Normally, the end of line characters are hidden, but SCI_SETVIEWEOL allows you to display (or hide) them by setting visible true (or false). The visible rendering of the end of line characters is similar to (CR), (LF), or (CR)(LF). SCI_GETVIEWEOL returns the current state.

Styling

The styling messages allow you to assign styles to text. The standard Scintilla settings divide the 8 style bits available for each character into 5 bits (0 to 4 = styles 0 to 31) that set a style and three bits (5 to 7) that define indicators. You can change the balance between styles and indicators with SCI_SETSTYLEBITS. If your styling needs can be met by one of the standard lexers, or if you can write your own, then a lexer is probably the easiest way to style your document. If you choose to use the container to do the styling you can use the SCI_SETLEXER command to select SCLEX_CONTAINER, in which case the container is sent a SCN_STYLENEEDED notification each time text needs styling for display. As another alternative, you might use idle time to style the document. Even if you use a lexer, you might use the styling commands to mark errors detected by a compiler. The following commands can be used.

SCI_GETENDSTYLED
Scintilla keeps a record of the last character that is likely to be styled correctly. This is moved forwards when characters after it are styled and moved backwards if changes are made to the text of the document before it. Before drawing text, this position is checked to see if any styling is needed and, if so, a SCN_STYLENEEDED notification message is sent to the container. The container can send SCI_GETENDSTYLED to work out where it needs to start styling. Scintilla will always ask to style whole lines.

SCI_STARTSTYLING(int pos, int mask)
This prepares for styling by setting the styling position pos to start at and a mask indicating which bits of the style bytes can be set. The mask allows styling to occur over several passes, with, for example, basic styling done on an initial pass to ensure that the text of the code is seen quickly and correctly, and then a second slower pass, detecting syntax errors and using indicators to show where these are. For example, with the standard settings of 5 style bits and 3 indicator bits, you would use a mask value of 31 (0x1f) if you were setting text styles and did not want to change the indicators. After SCI_STARTSTYLING, send multiple SCI_SETSTYLING messages for each lexical entity to style.

SCI_SETSTYLING(int length, int style)
This message sets the style of length characters starting at the styling position and then increases the styling position by length, ready for the next call. If sCell is the style byte, the operation is:
if ((sCell & mask) != style) sCell = (sCell & ~mask) | (style & mask);

SCI_SETSTYLINGEX(int length, const char *styles)
As an alternative to SCI_SETSTYLING, which applies the same style to each byte, you can use this message which specifies the styles for each of length bytes from the styling position and then increases the styling position by length, ready for the next call. The length styling bytes pointed at by styles should not contain any bits not set in mask.

SCI_SETLINESTATE(int line, int value)
SCI_GETLINESTATE(int line)
As well as the 8 bits of lexical state stored for each character there is also an integer stored for each line. This can be used for longer lived parse states such as what the current scripting language is in an ASP page. Use SCI_SETLINESTATE to set the integer value and SCI_GETLINESTATE to get the value. Changing the value produces a SC_MOD_CHANGELINESTATE notification.

SCI_GETMAXLINESTATE
This returns the last line that has any line state.

Style definition

While the style setting messages mentioned above change the style numbers associated with text, these messages define how those style numbers are interpreted visually. There are 256 lexer styles that can be set, numbered 0 to STYLE_MAX (255). Unless you use SCI_SETSTYLEBITS to change the number of style bits, styles 0 to 31 are used to set the text attributes. There are also some predefined numbered styles starting at 32, The following STYLE_* constants are defined.

For each style you can set the font name, size and use of bold, italic and underline, foreground and background colour and the character set. You can also choose to hide text with a given style, display all characters as upper or lower case and fill from the last character on a line to the end of the line (for embedded languages). There is also an experimental attribute to make text read-only.

It is entirely up to you how you use styles. If you want to use syntax colouring you might use style 0 for white space, style 1 for numbers, style 2 for keywords, style 3 for strings, style 4 for preprocessor, style 5 for operators, and so on.

SCI_STYLERESETDEFAULT
This message resets STYLE_DEFAULT to its state when Scintilla was initialised.

SCI_STYLECLEARALL
This message sets all styles to have the same attributes as STYLE_DEFAULT. If you are setting up Scintilla for syntax colouring, it is likely that the lexical styles you set will be very similar. One way to set the styles is to:
1. Set STYLE_DEFAULT to the common features of all styles.
2. Use SCI_STYLECLEARALL to copy this to all styles.
3. Set the style attributes that make your lexical styles different.

SCI_STYLESETFONT(int styleNumber, const char *fontName)
SCI_STYLEGETFONT(int styleNumber, char *fontName)
SCI_STYLESETSIZE(int styleNumber, int sizeInPoints)
SCI_STYLEGETSIZE(int styleNumber)
SCI_STYLESETBOLD(int styleNumber, bool bold)
SCI_STYLEGETBOLD(int styleNumber)
SCI_STYLESETITALIC(int styleNumber, bool italic)
SCI_STYLEGETITALIC(int styleNumber)
These messages (plus SCI_STYLESETCHARACTERSET) set the font attributes that are used to match the fonts you request to those available. The fontName is a zero terminated string holding the name of a font. Under Windows, only the first 32 characters of the name are used and the name is not case sensitive. For internal caching, Scintilla tracks fonts by name and does care about the casing of font names, so please be consistent. On GTK+ 2.x, either GDK or Pango can be used to display text. Pango antialiases text, works well with Unicode and is better supported in recent versions of GTK+ but GDK is faster. Prepend a '!' character to the font name to use Pango.

SCI_STYLESETUNDERLINE(int styleNumber, bool underline)
SCI_STYLEGETUNDERLINE(int styleNumber)
You can set a style to be underlined. The underline is drawn in the foreground colour. All characters with a style that includes the underline attribute are underlined, even if they are white space.

SCI_STYLESETFORE(int styleNumber, int colour)
SCI_STYLEGETFORE(int styleNumber)
SCI_STYLESETBACK(int styleNumber, int colour)
SCI_STYLEGETBACK(int styleNumber)
Text is drawn in the foreground colour. The space in each character cell that is not occupied by the character is drawn in the background colour.

SCI_STYLESETEOLFILLED(int styleNumber, bool eolFilled)
SCI_STYLEGETEOLFILLED(int styleNumber)
If the last character in the line has a style with this attribute set, the remainder of the line up to the right edge of the window is filled with the background colour set for the last character. This is useful when a document contains embedded sections in another language such as HTML pages with embedded JavaScript. By setting eolFilled to true and a consistent background colour (different from the background colour set for the HTML styles) to all JavaScript styles then JavaScript sections will be easily distinguished from HTML.

SCI_STYLESETCHARACTERSET(int styleNumber, int charSet)
SCI_STYLEGETCHARACTERSET(int styleNumber)
You can set a style to use a different character set than the default. The places where such characters sets are likely to be useful are comments and literal strings. For example, SCI_STYLESETCHARACTERSET(SCE_C_STRING, SC_CHARSET_RUSSIAN) would ensure that strings in Russian would display correctly in C and C++ (SCE_C_STRING is the style number used by the C and C++ lexer to display literal strings; it has the value 6). This feature works differently on Windows and GTK+.

The character sets supported on Windows are:
SC_CHARSET_ANSI, SC_CHARSET_ARABIC, SC_CHARSET_BALTIC, SC_CHARSET_CHINESEBIG5, SC_CHARSET_DEFAULT, SC_CHARSET_EASTEUROPE, SC_CHARSET_GB2312, SC_CHARSET_GREEK, SC_CHARSET_HANGUL, SC_CHARSET_HEBREW, SC_CHARSET_JOHAB, SC_CHARSET_MAC, SC_CHARSET_OEM, SC_CHARSET_RUSSIAN (code page 1251), SC_CHARSET_SHIFTJIS, SC_CHARSET_SYMBOL, SC_CHARSET_THAI, SC_CHARSET_TURKISH, and SC_CHARSET_VIETNAMESE.

The character sets supported on GTK+ are:
SC_CHARSET_ANSI, SC_CHARSET_CYRILLIC (code page 1251), SC_CHARSET_EASTEUROPE, SC_CHARSET_GB2312, SC_CHARSET_HANGUL, SC_CHARSET_RUSSIAN (KOI8-R), SC_CHARSET_SHIFTJIS, and SC_CHARSET_8859_15.

SCI_STYLESETCASE(int styleNumber, int caseMode)
SCI_STYLEGETCASE(int styleNumber)
The value of caseMode determines how text is displayed. You can set upper case (SC_CASE_UPPER, 1) or lower case (SC_CASE_LOWER, 2) or display normally (SC_CASE_MIXED, 0). This does not change the stored text, only how it is displayed.

SCI_STYLESETVISIBLE(int styleNumber, bool visible)
SCI_STYLEGETVISIBLE(int styleNumber)
Text is normally visible. However, you can completely hide it by giving it a style with the visible set to 0. This could be used to hide embedded formatting instructions or hypertext keywords in HTML or XML.

SCI_STYLESETCHANGEABLE(int styleNumber, bool changeable)
SCI_STYLEGETCHANGEABLE(int styleNumber)
This is an experimental and incompletely implemented style attribute. The default setting is changeable set true but when set false it makes text read-only. Currently it only stops the caret from being within not-changeable text and does not yet stop deleting a range that contains not-changeable text.

SCI_STYLESETHOTSPOT(int styleNumber, bool hotspot)
SCI_STYLEGETHOTSPOT(int styleNumber)
This style is used to mark ranges of text that can detect mouse clicks. The cursor changes to a hand over hotspots, and the foreground, and background colours may change and an underline appear to indicate that these areas are sensitive to clicking. This may be used to allow hyperlinks to other documents.

Caret, selection, and hotspot styles

The selection is shown by changing the foreground and/or background colours. If one of these is not set then that attribute is not changed for the selection. The default is to show the selection by changing the background to light gray and leaving the foreground the same as when it was not selected. When there is no selection, the current insertion point is marked by the text caret. This is a vertical line that is normally blinking on and off to attract the users attention.

SCI_SETSELFORE(bool useSelectionForeColour, int colour)
SCI_SETSELBACK(bool useSelectionBackColour, int colour)
You can choose to override the default selection colouring with these two messages. The colour you provide is used if you set useSelection*Colour to true. If it is set to false, the default styled colouring is used and the colour argument has no effect.

SCI_SETSELALPHA(int alpha)
SCI_GETSELALPHA
The selection can be drawn translucently in the selection background colour by setting an alpha value.

SCI_SETSELEOLFILLED(bool filled)
SCI_GETSELEOLFILLED
The selection can be drawn up to the right hand border by setting this property.

SCI_SETCARETFORE(int colour)
SCI_GETCARETFORE
The colour of the caret can be set with SCI_SETCARETFORE and retrieved with SCI_GETCARETFORE.

SCI_SETCARETLINEVISIBLE(bool show)
SCI_GETCARETLINEVISIBLE
SCI_SETCARETLINEBACK(int colour)
SCI_GETCARETLINEBACK
SCI_SETCARETLINEBACKALPHA(int alpha)
SCI_GETCARETLINEBACKALPHA
You can choose to make the background colour of the line containing the caret different with these messages. To do this, set the desired background colour with SCI_SETCARETLINEBACK, then use SCI_SETCARETLINEVISIBLE(true) to enable the effect. You can cancel the effect with SCI_SETCARETLINEVISIBLE(false). The two SCI_GETCARET* functions return the state and the colour. This form of background colouring has highest priority when a line has markers that would otherwise change the background colour. The caret line may also be drawn translucently which allows other background colours to show through. This is done by setting the alpha (translucency) value by calling SCI_SETCARETLINEBACKALPHA. When the alpha is not SC_ALPHA_NOALPHA, the caret line is drawn after all other features so will affect the colour of all other features.

SCI_SETCARETPERIOD(int milliseconds)
SCI_GETCARETPERIOD
The rate at which the caret blinks can be set with SCI_SETCARETPERIOD which determines the time in milliseconds that the caret is visible or invisible before changing state. Setting the period to 0 stops the caret blinking. The default value is 500 milliseconds. SCI_GETCARETPERIOD returns the current setting.

SCI_SETCARETSTYLE(int style)
SCI_GETCARETSTYLE
The style of the caret can be set with SCI_SETCARETSTYLE to be a line caret (CARETSTYLE_LINE=1), a block caret (CARETSTYLE_BLOCK=2) or to not draw at all (CARETSTYLE_INVISIBLE=0). The default value is the line caret (CARETSTYLE_LINE=1). You can determine the current caret style setting using SCI_GETCARETSTYLE.

The block character draws most combining and multibyte character sequences successfully, though some fonts like Thai Fonts (and possibly others) can sometimes appear strange when the cursor is positioned at these characters, which may result in only drawing a part of the cursor character sequence. This is most notable on Windows platforms.

SCI_SETCARETWIDTH(int pixels)
SCI_GETCARETWIDTH
The width of the line caret can be set with SCI_SETCARETWIDTH to a value of 0, 1, 2 or 3 pixels. The default width is 1 pixel. You can read back the current width with SCI_GETCARETWIDTH. A width of 0 makes the caret invisible (added at version 1.50), similar to setting the caret style to CARETSTYLE_INVISIBLE (though not interchangable). This setting only affects the width of the cursor when the cursor style is set to line caret mode, it does not affect the width for a block caret.

SCI_SETHOTSPOTACTIVEFORE(bool useHotSpotForeColour, int colour)
SCI_GETHOTSPOTACTIVEFORE
SCI_SETHOTSPOTACTIVEBACK(bool useHotSpotBackColour, int colour)
SCI_GETHOTSPOTACTIVEBACK
SCI_SETHOTSPOTACTIVEUNDERLINE(bool underline)
SCI_GETHOTSPOTACTIVEUNDERLINE
SCI_SETHOTSPOTSINGLELINE(bool singleLine)
SCI_GETHOTSPOTSINGLELINE
While the cursor hovers over text in a style with the hotspot attribute set, the default colouring can be modified and an underline drawn with these settings. Single line mode stops a hotspot from wrapping onto next line.

SCI_SETCONTROLCHARSYMBOL(int symbol)
SCI_GETCONTROLCHARSYMBOL
By default, Scintilla displays control characters (characters with codes less than 32) in a rounded rectangle as ASCII mnemonics: "NUL", "SOH", "STX", "ETX", "EOT", "ENQ", "ACK", "BEL", "BS", "HT", "LF", "VT", "FF", "CR", "SO", "SI", "DLE", "DC1", "DC2", "DC3", "DC4", "NAK", "SYN", "ETB", "CAN", "EM", "SUB", "ESC", "FS", "GS", "RS", "US". These mnemonics come from the early days of signaling, though some are still used (LF = Line Feed, BS = Back Space, CR = Carriage Return, for example).

You can choose to replace these mnemonics by a nominated symbol with an ASCII code in the range 32 to 255. If you set a symbol value less than 32, all control characters are displayed as mnemonics. The symbol you set is rendered in the font of the style set for the character. You can read back the current symbol with the SCI_GETCONTROLCHARSYMBOL message. The default symbol value is 0.

SCI_SETCARETSTICKY(bool useCaretStickyBehaviour)
SCI_GETCARETSTICKY
SCI_TOGGLECARETSTICKY
These messages set, get or toggle the caretSticky flag which controls when the last position of the caret on the line is saved. When set to true, the position is not saved when you type a character, a tab, paste the clipboard content or press backspace.

Margins

There may be up to five margins to the left of the text display, plus a gap either side of the text. Each margin can be set to display either symbols or line numbers with SCI_SETMARGINTYPEN. The markers that can be displayed in each margin are set with SCI_SETMARGINMASKN. Any markers not associated with a visible margin will be displayed as changes in background colour in the text. A width in pixels can be set for each margin. Margins with a zero width are ignored completely. You can choose if a mouse click in a margin sends a SCN_MARGINCLICK notification to the container or selects a line of text.

The margins are numbered 0 to 4. Using a margin number outside the valid range has no effect. By default, margin 0 is set to display line numbers, but is given a width of 0, so it is hidden. Margin 1 is set to display non-folding symbols and is given a width of 16 pixels, so it is visible. Margin 2 is set to display the folding symbols, but is given a width of 0, so it is hidden. Of course, you can set the margins to be whatever you wish.

SCI_SETMARGINTYPEN(int margin, int iType)
SCI_GETMARGINTYPEN(int margin)
These two routines set and get the type of a margin. The margin argument should be 0, 1, 2, 3 or 4. You can use the predefined constants SC_MARGIN_SYMBOL (0) and SC_MARGIN_NUMBER (1) to set a margin as either a line number or a symbol margin. By convention, margin 0 is used for line numbers and the next two are used for symbols. You can also use the constants SC_MARGIN_BACK (2) and SC_MARGIN_FORE (3) for symbol margins that set their background colour to match the STYLE_DEFAULT background and foreground colours.

SCI_SETMARGINWIDTHN(int margin, int pixelWidth)
SCI_GETMARGINWIDTHN(int margin)
These routines set and get the width of a margin in pixels. A margin with zero width is invisible. By default, Scintilla sets margin 1 for symbols with a width of 16 pixels, so this is a reasonable guess if you are not sure what would be appropriate. Line number margins widths should take into account the number of lines in the document and the line number style. You could use something like SCI_TEXTWIDTH(STYLE_LINENUMBER, "_99999") to get a suitable width.

SCI_SETMARGINMASKN(int margin, int mask)
SCI_GETMARGINMASKN(int margin)
The mask is a 32-bit value. Each bit corresponds to one of 32 logical symbols that can be displayed in a margin that is enabled for symbols. There is a useful constant, SC_MASK_FOLDERS (0xFE000000 or -33554432), that is a mask for the 7 logical symbols used to denote folding. You can assign a wide range of symbols and colours to each of the 32 logical symbols, see Markers for more information. If (mask & SC_MASK_FOLDERS)==0, the margin background colour is controlled by style 33 (STYLE_LINENUMBER).

You add logical markers to a line with SCI_MARKERADD. If a line has an associated marker that does not appear in the mask of any margin with a non-zero width, the marker changes the background colour of the line. For example, suppose you decide to use logical marker 10 to mark lines with a syntax error and you want to show such lines by changing the background colour. The mask for this marker is 1 shifted left 10 times (1<<10) which is 0x400. If you make sure that no symbol margin includes 0x400 in its mask, any line with the marker gets the background colour changed.

To set a non-folding margin 1 use SCI_SETMARGINMASKN(1, ~SC_MASK_FOLDERS); to set a folding margin 2 use SCI_SETMARGINMASKN(2, SC_MASK_FOLDERS). This is the default set by Scintilla. ~SC_MASK_FOLDERS is 0x1FFFFFF in hexadecimal or 33554431 decimal. Of course, you may need to display all 32 symbols in a margin, in which case use SCI_SETMARGINMASKN(margin, -1).

SCI_SETMARGINSENSITIVEN(int margin, bool sensitive)
SCI_GETMARGINSENSITIVEN(int margin)
Each of the five margins can be set sensitive or insensitive to mouse clicks. A click in a sensitive margin sends a SCN_MARGINCLICK notification to the container. Margins that are not sensitive act as selection margins which make it easy to select ranges of lines. By default, all margins are insensitive.

SCI_SETMARGINLEFT(<unused>, int pixels)
SCI_GETMARGINLEFT
SCI_SETMARGINRIGHT(<unused>, int pixels)
SCI_GETMARGINRIGHT
These messages set and get the width of the blank margin on both sides of the text in pixels. The default is to one pixel on each side.

SCI_SETFOLDMARGINCOLOUR(bool useSetting, int colour)
SCI_SETFOLDMARGINHICOLOUR(bool useSetting, int colour)
These messages allow changing the colour of the fold margin and fold margin highlight. On Windows the fold margin colour defaults to ::GetSysColor(COLOR_3DFACE) and the fold margin highlight colour to ::GetSysColor(COLOR_3DHIGHLIGHT).

Other settings

SCI_SETUSEPALETTE(bool allowPaletteUse)
SCI_GETUSEPALETTE
On 8 bit displays, which can only display a maximum of 256 colours, the graphics environment mediates between the colour needs of applications through the use of palettes. On GTK+, Scintilla always uses a palette.

On Windows, there are some problems with visual flashing when switching between applications with palettes and it is also necessary for the application containing the Scintilla control to forward some messages to Scintilla for its palette code to work. Because of this, by default, the palette is not used and the application must tell Scintilla to use one. If Scintilla is not using a palette, it will only display in those colours already available, which are often the 20 Windows system colours.

To see an example of how to enable palette support in Scintilla, search the text of SciTE for WM_PALETTECHANGED, WM_QUERYNEWPALETTE and SCI_SETUSEPALETTE. The Windows messages to forward are:
WM_SYSCOLORCHANGE, WM_PALETTECHANGED, WM_QUERYNEWPALETTE (should return TRUE).

To forward a message (WM_XXXX, WPARAM, LPARAM) to Scintilla, you can use SendMessage(hScintilla, WM_XXXX, WPARAM, LPARAM) where hScintilla is the handle to the Scintilla window you created as your editor.

While we are on the subject of forwarding messages in Windows, the top level window should forward any WM_SETTINGCHANGE messages to Scintilla (this is currently used to collect changes to mouse settings, but could be used for other user interface items in the future).

SCI_SETBUFFEREDDRAW(bool isBuffered)
SCI_GETBUFFEREDDRAW
These messages turn buffered drawing on or off and report the buffered drawing state. Buffered drawing draws each line into a bitmap rather than directly to the screen and then copies the bitmap to the screen. This avoids flickering although it does take longer. The default is for drawing to be buffered.

SCI_SETTWOPHASEDRAW(bool twoPhase)
SCI_GETTWOPHASEDRAW
Two phase drawing is a better but slower way of drawing text. In single phase drawing each run of characters in one style is drawn along with its background. If a character overhangs the end of a run, such as in "V_" where the "V" is in a different style from the "_", then this can cause the right hand side of the "V" to be overdrawn by the background of the "_" which cuts it off. Two phase drawing fixes this by drawing all the backgrounds first and then drawing the text in transparent mode. Two phase drawing may flicker more than single phase unless buffered drawing is on. The default is for drawing to be two phase.

SCI_SETCODEPAGE(int codePage)
SCI_GETCODEPAGE
Scintilla has some support for Japanese, Chinese and Korean DBCS. Use this message with codePage set to the code page number to set Scintilla to use code page information to ensure double byte characters are treated as one character rather than two. This also stops the caret from moving between the two bytes in a double byte character. Do not use this message to choose between different single byte character sets: it doesn't do that. Call with codePage set to zero to disable DBCS support. The default is SCI_SETCODEPAGE(0).

Code page SC_CP_UTF8 (65001) sets Scintilla into Unicode mode with the document treated as a sequence of characters expressed in UTF-8. The text is converted to the platform's normal Unicode encoding before being drawn by the OS and thus can display Hebrew, Arabic, Cyrillic, and Han characters. Languages which can use two characters stacked vertically in one horizontal space, such as Thai, will mostly work but there are some issues where the characters are drawn separately leading to visual glitches. Bi-directional text is not supported. Characters outside the Basic Multilingual Plane are unlikely to work.

On Windows, code page can be set to 932 (Japanese Shift-JIS), 936 (Simplified Chinese GBK), 949 (Korean Unified Hangul Code), 950 (Traditional Chinese Big5), or 1361 (Korean Johab) although these may require installation of language specific support.

On GTK+, code page SC_CP_DBCS (1) sets Scintilla into multi byte character mode as is required for Japanese language processing with the EUC encoding.

For GTK+ 1.x, the locale should be set to a Unicode locale with a call similar to setlocale(LC_CTYPE, "en_US.UTF-8"). Fonts with an "iso10646" registry should be used in a font set. Font sets are a comma separated list of partial font specifications where each partial font specification can be in the form: foundry-fontface-charsetregistry-encoding or fontface-charsetregistry-encoding or foundry-fontface or fontface. An example is "misc-fixed-iso10646-1,*". On GTK+ 2.x, Pango fonts should be used rather than font sets.

Setting codePage to a non-zero value that is not SC_CP_UTF8 is operating system dependent.

SCI_SETKEYSUNICODE(bool keysUnicode)
SCI_GETKEYSUNICODE
On Windows, character keys are normally handled differently depending on whether Scintilla is a wide or narrow character window with character messages treated as Unicode when wide and as 8 bit otherwise. Set this property to always treat as Unicode. This option is needed for Delphi.

SCI_SETWORDCHARS(<unused>, const char *chars)
Scintilla has several functions that operate on words, which are defined to be contiguous sequences of characters from a particular set of characters. This message defines which characters are members of that set. The character sets are set to default values before processing this function. For example, if you don't allow '_' in your set of characters use:
SCI_SETWORDCHARS(0, "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789");

SCI_SETWHITESPACECHARS(<unused>, const char *chars)
Similar to SCI_SETWORDCHARS, this message allows the user to define which chars Scintilla considers as whitespace. Setting the whitespace chars allows the user to fine-tune Scintilla's behaviour doing such things as moving the cursor to the start or end of a word; for example, by defining punctuation chars as whitespace, they will be skipped over when the user presses ctrl+left or ctrl+right. This function should be called after SCI_SETWORDCHARS as it will reset the whitespace characters to the default set.

SCI_SETCHARSDEFAULT
Use the default sets of word and whitespace characters. This sets whitespace to space, tab and other characters with codes less than 0x20, with word characters set to alphanumeric and '_'.

SCI_GRABFOCUS
SCI_SETFOCUS(bool focus)
SCI_GETFOCUS
Scintilla can be told to grab the focus with this message. This is needed more on GTK+ where focus handling is more complicated than on Windows.

The internal focus flag can be set with SCI_SETFOCUS. This is used by clients that have complex focus requirements such as having their own window that gets the real focus but with the need to indicate that Scintilla has the logical focus.

Brace highlighting

SCI_BRACEHIGHLIGHT(int pos1, int pos2)
Up to two characters can be highlighted in a 'brace highlighting style', which is defined as style number STYLE_BRACELIGHT (34). If you have enabled indent guides, you may also wish to highlight the indent that corresponds with the brace. You can locate the column with SCI_GETCOLUMN and highlight the indent with SCI_SETHIGHLIGHTGUIDE.

SCI_BRACEBADLIGHT(int pos1)
If there is no matching brace then the brace badlighting style, style BRACE_BADLIGHT (35), can be used to show the brace that is unmatched. Using a position of INVALID_POSITION (-1) removes the highlight.

SCI_BRACEMATCH(int pos, int maxReStyle)
The SCI_BRACEMATCH message finds a corresponding matching brace given pos, the position of one brace. The brace characters handled are '(', ')', '[', ']', '{', '}', '<', and '>'. The search is forwards from an opening brace and backwards from a closing brace. If the character at position is not a brace character, or a matching brace cannot be found, the return value is -1. Otherwise, the return value is the position of the matching brace.

A match only occurs if the style of the matching brace is the same as the starting brace or the matching brace is beyond the end of styling. Nested braces are handled correctly. The maxReStyle parameter must currently be 0 - it may be used in the future to limit the length of brace searches.

Tabs and Indentation Guides

Indentation (the white space at the start of a line) is often used by programmers to clarify program structure and in some languages, for example Python, it may be part of the language syntax. Tabs are normally used in editors to insert a tab character or to pad text with spaces up to the next tab.

Scintilla can be set to treat tab and backspace in the white space at the start of a line in a special way: inserting a tab indents the line to the next indent position rather than just inserting a tab at the current character position and backspace unindents the line rather than deleting a character. Scintilla can also display indentation guides (vertical lines) to help you to generate code.

SCI_SETTABWIDTH(int widthInChars)
SCI_GETTABWIDTH
SCI_SETTABWIDTH sets the size of a tab as a multiple of the size of a space character in STYLE_DEFAULT. The default tab width is 8 characters. There are no limits on tab sizes, but values less than 1 or large values may have undesirable effects.

SCI_SETUSETABS(bool useTabs)
SCI_GETUSETABS
SCI_SETUSETABS determines whether indentation should be created out of a mixture of tabs and spaces or be based purely on spaces. Set useTabs to false (0) to create all tabs and indents out of spaces. The default is true. You can use SCI_GETCOLUMN to get the column of a position taking the width of a tab into account.

SCI_SETINDENT(int widthInChars)
SCI_GETINDENT
SCI_SETINDENT sets the size of indentation in terms of the width of a space in STYLE_DEFAULT. If you set a width of 0, the indent size is the same as the tab size. There are no limits on indent sizes, but values less than 0 or large values may have undesirable effects.

SCI_SETTABINDENTS(bool tabIndents)
SCI_GETTABINDENTS
SCI_SETBACKSPACEUNINDENTS(bool bsUnIndents)
SCI_GETBACKSPACEUNINDENTS

Inside indentation white space, the tab and backspace keys can be made to indent and unindent rather than insert a tab character or delete a character with the SCI_SETTABINDENTS and SCI_SETBACKSPACEUNINDENTS functions.

SCI_SETLINEINDENTATION(int line, int indentation)
SCI_GETLINEINDENTATION(int line)
The amount of indentation on a line can be discovered and set with SCI_GETLINEINDENTATION and SCI_SETLINEINDENTATION. The indentation is measured in character columns, which correspond to the width of space characters.

SCI_GETLINEINDENTPOSITION(int line)
This returns the position at the end of indentation of a line.

SCI_SETINDENTATIONGUIDES(int indentView)
SCI_GETINDENTATIONGUIDES
Indentation guides are dotted vertical lines that appear within indentation white space every indent size columns. They make it easy to see which constructs line up especially when they extend over multiple pages. Style STYLE_INDENTGUIDE (37) is used to specify the foreground and background colour of the indentation guides.

There are 4 indentation guide views. SC_IV_NONE turns the feature off but the other 3 states determine how far the guides appear on empty lines.

SCI_SETHIGHLIGHTGUIDE(int column)
SCI_GETHIGHLIGHTGUIDE
When brace highlighting occurs, the indentation guide corresponding to the braces may be highlighted with the brace highlighting style, STYLE_BRACELIGHT (34). Set column to 0 to cancel this highlight.

Markers

There are 32 markers, numbered 0 to 31, and you can assign any combination of them to each line in the document. Markers appear in the selection margin to the left of the text. If the selection margin is set to zero width, the background colour of the whole line is changed instead. Marker numbers 25 to 31 are used by Scintilla in folding margins, and have symbolic names of the form SC_MARKNUM_*, for example SC_MARKNUM_FOLDEROPEN.

Marker numbers 0 to 24 have no pre-defined function; you can use them to mark syntax errors or the current point of execution, break points, or whatever you need marking. If you do not need folding, you can use all 32 for any purpose you wish.

Each marker number has a symbol associated with it. You can also set the foreground and background colour for each marker number, so you can use the same symbol more than once with different colouring for different uses. Scintilla has a set of symbols you can assign (SC_MARK_*) or you can use characters. By default, all 32 markers are set to SC_MARK_CIRCLE with a black foreground and a white background.

The markers are drawn in the order of their numbers, so higher numbered markers appear on top of lower numbered ones. Markers try to move with their text by tracking where the start of their line moves. When a line is deleted, its markers are combined, by an OR operation, with the markers of the previous line.

SCI_MARKERDEFINE(int markerNumber, int markerSymbols)
This message associates a marker number in the range 0 to 31 with one of the marker symbols or an ASCII character. The general-purpose marker symbols currently available are:
SC_MARK_CIRCLE, SC_MARK_ROUNDRECT, SC_MARK_ARROW, SC_MARK_SMALLRECT, SC_MARK_SHORTARROW, SC_MARK_EMPTY, SC_MARK_ARROWDOWN, SC_MARK_MINUS, SC_MARK_PLUS, SC_MARK_ARROWS, SC_MARK_DOTDOTDOT, SC_MARK_EMPTY, SC_MARK_BACKGROUND, SC_MARK_LEFTRECT and SC_MARK_FULLRECT.

The SC_MARK_BACKGROUND marker changes the background colour of the line only. The SC_MARK_FULLRECT symbol mirrors this, changing only the margin background colour. The SC_MARK_EMPTY symbol is invisible, allowing client code to track the movement of lines. You would also use it if you changed the folding style and wanted one or more of the SC_FOLDERNUM_* markers to have no associated symbol.

There are also marker symbols designed for use in the folding margin in a flattened tree style.
SC_MARK_BOXMINUS, SC_MARK_BOXMINUSCONNECTED, SC_MARK_BOXPLUS, SC_MARK_BOXPLUSCONNECTED, SC_MARK_CIRCLEMINUS, SC_MARK_CIRCLEMINUSCONNECTED, SC_MARK_CIRCLEPLUS, SC_MARK_CIRCLEPLUSCONNECTED, SC_MARK_LCORNER, SC_MARK_LCORNERCURVE, SC_MARK_TCORNER, SC_MARK_TCORNERCURVE, and SC_MARK_VLINE.

The marker numbers SC_MARKNUM_FOLDER and SC_MARKNUM_FOLDEROPEN are used for showing that a fold is present and open or closed. Any symbols may be assigned for this purpose although the (SC_MARK_PLUS, SC_MARK_MINUS) pair or the (SC_MARK_ARROW, SC_MARK_ARROWDOWN) pair are good choices. As well as these two, more assignments are needed for the flattened tree style: SC_MARKNUM_FOLDEREND, SC_MARKNUM_FOLDERMIDTAIL, SC_MARKNUM_FOLDEROPENMID, SC_MARKNUM_FOLDERSUB, and SC_MARKNUM_FOLDERTAIL. The bits used for folding are specified by SC_MASK_FOLDERS, which is commonly used as an argument to SCI_SETMARGINMASKN when defining a margin to be used for folding.

This table shows which SC_MARK_* symbols should be assigned to which SC_MARKNUM_* marker numbers to obtain four folding styles: Arrow (mimics Macintosh), plus/minus shows folded lines as '+' and opened folds as '-', Circle tree, Box tree.

SCI_MARKERDEFINEPIXMAP(int markerNumber, const char *xpm)
Markers can be set to pixmaps with this message. The XPM format is used for the pixmap and it is limited to pixmaps that use one character per pixel with no named colours. The transparent colour may be named 'None'. The data should be null terminated. Pixmaps use the SC_MARK_PIXMAP marker symbol. You can find the full description of the XPM format here.

SCI_MARKERSETFORE(int markerNumber, int colour)
SCI_MARKERSETBACK(int markerNumber, int colour)
These two messages set the foreground and background colour of a marker number.

SCI_MARKERSETALPHA(int markerNumber, int alpha)
When markers are drawn in the content area, either because there is no margin for them or they are of SC_MARK_BACKGROUND type, they may be drawn translucently by setting an alpha value.

SCI_MARKERADD(int line, int markerNumber)
This message adds marker number markerNumber to a line. The message returns -1 if this fails (illegal line number, out of memory) or it returns a marker handle number that identifies the added marker. You can use this returned handle with SCI_MARKERLINEFROMHANDLE to find where a marker is after moving or combining lines and with SCI_MARKERDELETEHANDLE to delete the marker based on its handle. The message does not check the value of markerNumber, nor does it check if the line already contains the marker.

SCI_MARKERADDSET(int line, int markerMask)
This message can add one or more markers to a line with a single call, specified in the same "one-bit-per-marker" 32-bit integer format returned by SCI_MARKERGET (and used by the mask-based marker search functions SCI_MARKERNEXT and SCI_MARKERPREVIOUS). As with SCI_MARKERADD, no check is made to see if any of the markers are already present on the targeted line.

SCI_MARKERDELETE(int line, int markerNumber)
This searches the given line number for the given marker number and deletes it if it is present. If you added the same marker more than once to the line, this will delete one copy each time it is used. If you pass in a marker number of -1, all markers are deleted from the line.

SCI_MARKERDELETEALL(int markerNumber)
This removes markers of the given number from all lines. If markerNumber is -1, it deletes all markers from all lines.

SCI_MARKERGET(int line)
This returns a 32-bit integer that indicates which markers were present on the line. Bit 0 is set if marker 0 is present, bit 1 for marker 1 and so on.

SCI_MARKERNEXT(int lineStart, int markerMask)
SCI_MARKERPREVIOUS(int lineStart, int markerMask)
These messages search efficiently for lines that include a given set of markers. The search starts at line number lineStart and continues forwards to the end of the file (SCI_MARKERNEXT) or backwards to the start of the file (SCI_MARKERPREVIOUS). The markerMask argument should have one bit set for each marker you wish to find. Set bit 0 to find marker 0, bit 1 for marker 1 and so on. The message returns the line number of the first line that contains one of the markers in markerMask or -1 if no marker is found.

SCI_MARKERLINEFROMHANDLE(int markerHandle)
The markerHandle argument is an identifier for a marker returned by SCI_MARKERADD. This function searches the document for the marker with this handle and returns the line number that contains it or -1 if it is not found.

SCI_MARKERDELETEHANDLE(int markerHandle)
The markerHandle argument is an identifier for a marker returned by SCI_MARKERADD. This function searches the document for the marker with this handle and deletes the marker if it is found.

Indicators

Indicators are used to display additional information over the top of styling. They can be used to show, for example, syntax errors, deprecated names and bad indentation by drawing underlines under text or boxes around text. Originally, Scintilla stored indicator information in the style bytes but this has proved limiting, so now up to 32 separately stored indicators may be used. While style byte indicators currently still work, they will soon be removed so all the bits in each style byte can be used for lexical states.

Indicators may be displayed as simple underlines, squiggly underlines, a line of small 'T' shapes, a line of diagonal hatching, a strike-out or a rectangle around the text.

The SCI_INDIC* messages allow you to get and set the visual appearance of the indicators. They all use an indicatorNumber argument in the range 0 to INDIC_MAX(31) to set the indicator to style. To prevent interference the set of indicators is divided up into a range for use by lexers (0..7) and a range for use by containers (8=INDIC_CONTAINER .. 31=INDIC_MAX).

SCI_INDICSETSTYLE(int indicatorNumber, int indicatorStyle)
SCI_INDICGETSTYLE(int indicatorNumber)
These two messages set and get the style for a particular indicator. The indicator styles currently available are:

The default indicator styles are equivalent to:
SCI_INDICSETSTYLE(0, INDIC_SQUIGGLE);
SCI_INDICSETSTYLE(1, INDIC_TT);
SCI_INDICSETSTYLE(2, INDIC_PLAIN);

SCI_INDICSETFORE(int indicatorNumber, int colour)
SCI_INDICGETFORE(int indicatorNumber)
These two messages set and get the colour used to draw an indicator. The default indicator colours are equivalent to:
SCI_INDICSETFORE(0, 0x007f00); (dark green)
SCI_INDICSETFORE(1, 0xff0000); (light blue)
SCI_INDICSETFORE(2, 0x0000ff); (light red)

SCI_INDICSETUNDER(int indicatorNumber, bool under)
SCI_INDICGETUNDER(int indicatorNumber)
These two messages set and get whether an indicator is drawn under text or over(default). Drawing under text works only for modern indicators when two phase drawing is enabled.

Modern Indicators

Modern indicators are stored in a format similar to run length encoding which is efficient in both speed and storage for sparse information.

An indicator may store different values for each range but currently all values are drawn the same. In the future, it may be possible to draw different values in different styles.

SCI_SETINDICATORCURRENT(int indicator)
SCI_GETINDICATORCURRENT
These two messages set and get the indicator that will be affected by calls to SCI_INDICATORFILLRANGE and SCI_INDICATORCLEARRANGE.

SCI_SETINDICATORVALUE(int value)
SCI_GETINDICATORVALUE
These two messages set and get the value that will be set by calls to SCI_INDICATORFILLRANGE.

SCI_INDICATORFILLRANGE(int position, int fillLength)
SCI_INDICATORCLEARRANGE(int position, int clearLength)
These two messages fill or clear a range for the current indicator. SCI_INDICATORFILLRANGE fills with the the current value.

SCI_INDICATORALLONFOR(int position)
Retrieve a bitmap value representing which indicators are non-zero at a position.

SCI_INDICATORVALUEAT(int indicator, int position)
Retrieve the value of a particular indicator at a position.

SCI_INDICATORSTART(int indicator, int position)
SCI_INDICATOREND(int indicator, int position)
Find the start or end of a range with one value from a position within the range. Can be used to iterate through the document to discover all the indicator positions.

Style Byte Indicators (deprecated)

By default, Scintilla organizes the style byte associated with each text byte as 5 bits of style information (for 32 styles) and 3 bits of indicator information for 3 independent indicators so that, for example, syntax errors, deprecated names and bad indentation could all be displayed at once.

The indicators are set using SCI_STARTSTYLING with a INDICS_MASK mask and SCI_SETSTYLING with the values INDIC0_MASK, INDIC1_MASK and INDIC2_MASK.

If you are using indicators in a buffer that has a lexer active (see SCI_SETLEXER), you must save lexing state information before setting any indicators and restore it afterwards. Use SCI_GETENDSTYLED to retrieve the current "styled to" position and SCI_STARTSTYLING to reset the styling position and mask (0x1f in the default layout of 5 style bits and 3 indicator bits) when you are done.

The number of bits used for styles can be altered with SCI_SETSTYLEBITS from 0 to 7 bits. The remaining bits can be used for indicators, so there can be from 1 to 8 indicators. However, the INDIC*_MASK constants defined in Scintilla.h all assume 5 bits of styling information and 3 indicators. If you use a different arrangement, you must define your own constants.

Autocompletion

Autocompletion displays a list box showing likely identifiers based upon the user's typing. The user chooses the currently selected item by pressing the tab character or another character that is a member of the fillup character set defined with SCI_AUTOCSETFILLUPS. Autocompletion is triggered by your application. For example, in C if you detect that the user has just typed fred. you could look up fred, and if it has a known list of members, you could offer them in an autocompletion list. Alternatively, you could monitor the user's typing and offer a list of likely items once their typing has narrowed down the choice to a reasonable list. As yet another alternative, you could define a key code to activate the list.

When the user makes a selection from the list the container is sent a SCN_AUTOCSELECTION notification message. On return from the notification Scintilla will insert the selected text unless the autocompletion list has been cancelled, for example by the container sending SCI_AUTOCCANCEL.

To make use of autocompletion you must monitor each character added to the document. See SciTEBase::CharAdded() in SciTEBase.cxx for an example of autocompletion.

SCI_AUTOCSHOW(int lenEntered, const char *list)
This message causes a list to be displayed. lenEntered is the number of characters of the word already entered and list is the list of words separated by separator characters. The initial separator character is a space but this can be set or got with SCI_AUTOCSETSEPARATOR and SCI_AUTOCGETSEPARATOR.

The list of words should be in sorted order. If set to ignore case mode with SCI_AUTOCSETIGNORECASE, then strings are matched after being converted to upper case. One result of this is that the list should be sorted with the punctuation characters '[', '\', ']', '^', '_', and '`' sorted after letters.

SCI_AUTOCCANCEL
This message cancels any displayed autocompletion list. When in autocompletion mode, the list should disappear when the user types a character that can not be part of the autocompletion, such as '.', '(' or '[' when typing an identifier. A set of characters that will cancel autocompletion can be specified with SCI_AUTOCSTOPS.

SCI_AUTOCACTIVE
This message returns non-zero if there is an active autocompletion list and zero if there is not.

SCI_AUTOCPOSSTART
This returns the value of the current position when SCI_AUTOCSHOW started display of the list.

SCI_AUTOCCOMPLETE
This message triggers autocompletion. This has the same effect as the tab key.

SCI_AUTOCSTOPS(<unused>, const char *chars)
The chars argument is a string containing a list of characters that will automatically cancel the autocompletion list. When you start the editor, this list is empty.

SCI_AUTOCSETSEPARATOR(char separator)
SCI_AUTOCGETSEPARATOR
These two messages set and get the separator character used to separate words in the SCI_AUTOCSHOW list. The default is the space character.

SCI_AUTOCSELECT(<unused>, const char *select)
SCI_AUTOCGETCURRENT
This message selects an item in the autocompletion list. It searches the list of words for the first that matches select. By default, comparisons are case sensitive, but you can change this with SCI_AUTOCSETIGNORECASE. The match is character by character for the length of the select string. That is, if select is "Fred" it will match "Frederick" if this is the first item in the list that begins with "Fred". If an item is found, it is selected. If the item is not found, the autocompletion list closes if auto-hide is true (see SCI_AUTOCSETAUTOHIDE).
The current selection can be retrieved with SCI_AUTOCGETCURRENT

SCI_AUTOCSETCANCELATSTART(bool cancel)
SCI_AUTOCGETCANCELATSTART
The default behavior is for the list to be cancelled if the caret moves before the location it was at when the list was displayed. By calling this message with a false argument, the list is not cancelled until the caret moves before the first character of the word being completed.

SCI_AUTOCSETFILLUPS(<unused>, const char *chars)
If a fillup character is typed with an autocompletion list active, the currently selected item in the list is added into the document, then the fillup character is added. Common fillup characters are '(', '[' and '.' but others are possible depending on the language. By default, no fillup characters are set.

SCI_AUTOCSETCHOOSESINGLE(bool chooseSingle)
SCI_AUTOCGETCHOOSESINGLE
If you use SCI_AUTOCSETCHOOSESINGLE(1) and a list has only one item, it is automatically added and no list is displayed. The default is to display the list even if there is only a single item.

SCI_AUTOCSETIGNORECASE(bool ignoreCase)
SCI_AUTOCGETIGNORECASE
By default, matching of characters to list members is case sensitive. These messages let you set and get case sensitivity.

SCI_AUTOCSETAUTOHIDE(bool autoHide)
SCI_AUTOCGETAUTOHIDE
By default, the list is cancelled if there are no viable matches (the user has typed characters that no longer match a list entry). If you want to keep displaying the original list, set autoHide to false. This also effects SCI_AUTOCSELECT.

SCI_AUTOCSETDROPRESTOFWORD(bool dropRestOfWord)
SCI_AUTOCGETDROPRESTOFWORD
When an item is selected, any word characters following the caret are first erased if dropRestOfWord is set true. The default is false.

SCI_REGISTERIMAGE(int type, const char *xpmData)
SCI_CLEARREGISTEREDIMAGES
SCI_AUTOCSETTYPESEPARATOR(char separatorCharacter)
SCI_AUTOCGETTYPESEPARATOR
Autocompletion list items may display an image as well as text. Each image is first registered with an integer type. Then this integer is included in the text of the list separated by a '?' from the text. For example, "fclose?2 fopen" displays image 2 before the string "fclose" and no image before "fopen". The images are in XPM format as is described for SCI_MARKERDEFINEPIXMAP The set of registered images can be cleared with SCI_CLEARREGISTEREDIMAGES and the '?' separator changed with SCI_AUTOCSETTYPESEPARATOR.

SCI_AUTOCSETMAXHEIGHT(int rowCount)
SCI_AUTOCGETMAXHEIGHT
Get or set the maximum number of rows that will be visible in an autocompletion list. If there are more rows in the list, then a vertical scrollbar is shown. The default is 5.

SCI_AUTOCSETMAXWIDTH(int characterCount)
SCI_AUTOCGETMAXWIDTH
Get or set the maximum width of an autocompletion list expressed as the number of characters in the longest item that will be totally visible. If zero (the default) then the list's width is calculated to fit the item with the most characters. Any items that cannot be fully displayed within the available width are indicated by the presence of ellipsis.

User lists

User lists use the same internal mechanisms as autocompletion lists, and all the calls listed for autocompletion work on them; you cannot display a user list at the same time as an autocompletion list is active. They differ in the following respects:

o The SCI_AUTOCSETCHOOSESINGLE message has no effect.
o When the user makes a selection you are sent a SCN_USERLISTSELECTION notification message rather than SCN_AUTOCSELECTION.

BEWARE: if you have set fillup characters or stop characters, these will still be active with the user list, and may result in items being selected or the user list cancelled due to the user typing into the editor.

SCI_USERLISTSHOW(int listType, const char *list)
The listType parameter is returned to the container as the wParam field of the SCNotification structure. It must be greater than 0 as this is how Scintilla tells the difference between an autocompletion list and a user list. If you have different types of list, for example a list of buffers and a list of macros, you can use listType to tell which one has returned a selection.

Call tips

Call tips are small windows displaying the arguments to a function and are displayed after the user has typed the name of the function. They normally display characters using the font facename, size and character set defined by STYLE_DEFAULT. You can choose to use STYLE_CALLTIP to define the facename, size, foreground and background colours and character set with SCI_CALLTIPUSESTYLE. This also enables support for Tab characters. There is some interaction between call tips and autocompletion lists in that showing a call tip cancels any active autocompletion list, and vice versa.

Call tips can highlight part of the text within them. You could use this to highlight the current argument to a function by counting the number of commas (or whatever separator your language uses). See SciTEBase::CharAdded() in SciTEBase.cxx for an example of call tip use.

The mouse may be clicked on call tips and this causes a SCN_CALLTIPCLICK notification to be sent to the container. Small up and down arrows may be displayed within a call tip by, respectively, including the characters '\001', or '\002'. This is useful for showing that there are overloaded variants of one function name and that the user can click on the arrows to cycle through the overloads.

Alternatively, call tips can be displayed when you leave the mouse pointer for a while over a word in response to the SCN_DWELLSTART notification and cancelled in response to SCN_DWELLEND. This method could be used in a debugger to give the value of a variable, or during editing to give information about the word under the pointer.

SCI_CALLTIPSHOW(int posStart, const char *definition)
This message starts the process by displaying the call tip window. If a call tip is already active, this has no effect.
posStart is the position in the document at which to align the call tip. The call tip text is aligned to start 1 line below this character unless you have included up and/or down arrows in the call tip text in which case the tip is aligned to the right-hand edge of the rightmost arrow. The assumption is that you will start the text with something like "\001 1 of 3 \002".
definition is the call tip text. This can contain multiple lines separated by '\n' (Line Feed, ASCII code 10) characters. Do not include '\r' (Carriage Return, ASCII code 13), as this will most likely print as an empty box. '\t' (Tab, ASCII code 9) is supported if you set a tabsize with SCI_CALLTIPUSESTYLE.

SCI_CALLTIPCANCEL
This message cancels any displayed call tip. Scintilla will also cancel call tips for you if you use any keyboard commands that are not compatible with editing the argument list of a function.

SCI_CALLTIPACTIVE
This returns 1 if a call tip is active and 0 if it is not active.

SCI_CALLTIPPOSSTART
This message returns the value of the current position when SCI_CALLTIPSHOW started to display the tip.

SCI_CALLTIPSETHLT(int hlStart, int hlEnd)
This sets the region of the call tips text to display in a highlighted style. hlStart is the zero-based index into the string of the first character to highlight and hlEnd is the index of the first character after the highlight. hlEnd must be greater than hlStart; hlEnd-hlStart is the number of characters to highlight. Highlights can extend over line ends if this is required.

Unhighlighted text is drawn in a mid gray. Selected text is drawn in a dark blue. The background is white. These can be changed with SCI_CALLTIPSETBACK, SCI_CALLTIPSETFORE, and SCI_CALLTIPSETFOREHLT.

SCI_CALLTIPSETBACK(int colour)
The background colour of call tips can be set with this message; the default colour is white. It is not a good idea to set a dark colour as the background as the default colour for normal calltip text is mid gray and the defaultcolour for highlighted text is dark blue. This also sets the background colour of STYLE_CALLTIP.

SCI_CALLTIPSETFORE(int colour)
The colour of call tip text can be set with this message; the default colour is mid gray. This also sets the foreground colour of STYLE_CALLTIP.

SCI_CALLTIPSETFOREHLT(int colour)
The colour of highlighted call tip text can be set with this message; the default colour is dark blue.

SCI_CALLTIPUSESTYLE(int tabsize)
This message changes the style used for call tips from STYLE_DEFAULT to STYLE_CALLTIP and sets a tab size in screen pixels. If tabsize is less than 1, Tab characters are not treated specially. Once this call has been used, the call tip foreground and background colours are also taken from the style.

Keyboard commands

To allow the container application to perform any of the actions available to the user with keyboard, all the keyboard actions are messages. They do not take any parameters. These commands are also used when redefining the key bindings with the SCI_ASSIGNCMDKEY message.

The SCI_*EXTEND messages extend the selection.

The SCI_*RECTEXTEND messages extend the rectangular selection (and convert regular selection to rectangular one, if any).

The SCI_WORDPART* commands are used to move between word segments marked by capitalisation (aCamelCaseIdentifier) or underscores (an_under_bar_ident).

The SCI_HOME* commands move the caret to the start of the line, while the SCI_VCHOME*commands move the caret to the first non-blank character of the line (ie. just after the indentation) unless it is already there; in this case, it acts as SCI_HOME*.

The SCI_[HOME|LINEEND]DISPLAY* commands are used when in line wrap mode to allow movement to the start or end of display lines as opposed to the normal SCI_[HOME|LINEEND] commands which move to the start or end of document lines.

The SCI_[[VC]HOME|LINEEND]WRAP* commands are like their namesakes SCI_[[VC]HOME|LINEEND]* except they behave differently when word-wrap is enabled: They go first to the start / end of the display line, like SCI_[HOME|LINEEND]DISPLAY*, but if the cursor is already at the point, it goes on to the start or end of the document line, as appropriate for SCI_[[VC]HOME|LINEEND]*.

Key bindings

There is a default binding of keys to commands that is defined in the Scintilla source in the file KeyMap.cxx by the constant KeyMap::MapDefault[]. This table maps key definitions to SCI_* messages with no parameters (mostly the keyboard commands discussed above, but any Scintilla command that has no arguments can be mapped). You can change the mapping to suit your own requirements.

keyDefinition
A key definition contains the key code in the low 16-bits and the key modifiers in the high 16-bits. To combine keyCode and keyMod set:

keyDefinition = keyCode + (keyMod << 16)

The key code is a visible or control character or a key from the SCK_* enumeration, which contains:
SCK_ADD, SCK_BACK, SCK_DELETE, SCK_DIVIDE, SCK_DOWN, SCK_END, SCK_ESCAPE, SCK_HOME, SCK_INSERT, SCK_LEFT, SCK_MENU, SCK_NEXT (Page Down), SCK_PRIOR (Page Up), SCK_RETURN, SCK_RIGHT, SCK_RWIN, SCK_SUBTRACT, SCK_TAB, SCK_UP, and SCK_WIN.

The modifiers are a combination of zero or more of SCMOD_ALT, SCMOD_CTRL, and SCMOD_SHIFT. If you are building a table, you might want to use SCMOD_NORM, which has the value 0, to mean no modifiers.

SCI_ASSIGNCMDKEY(int keyDefinition, int sciCommand)
This assigns the given key definition to a Scintilla command identified by sciCommand. sciCommand can be any SCI_* command that has no arguments.

SCI_CLEARCMDKEY(int keyDefinition)
This makes the given key definition do nothing by assigning the action SCI_NULL to it.

SCI_CLEARALLCMDKEYS
This command removes all keyboard command mapping by setting an empty mapping table.

SCI_NULL
The SCI_NULL does nothing and is the value assigned to keys that perform no action. SCI_NULL ensures that keys do not propagate to the parent window as that may cause focus to move. If you want the standard platform behaviour use the constant 0 instead.

Popup edit menu

SCI_USEPOPUP(bool bEnablePopup)
Clicking the wrong button on the mouse pops up a short default editing menu. This may be turned off with SCI_USEPOPUP(0). If you turn it off, context menu commands (in Windows, WM_CONTEXTMENU) will not be handled by Scintilla, so the parent of the Scintilla window will have the opportunity to handle the message.

Macro recording

Start and stop macro recording mode. In macro recording mode, actions are reported to the container through SCN_MACRORECORD notifications. It is then up to the container to record these actions for future replay.

SCI_STARTRECORD
SCI_STOPRECORD
These two messages turn macro recording on and off.

Printing

On Windows SCI_FORMATRANGE can be used to draw the text onto a display context which can include a printer display context. Printed output shows text styling as on the screen, but it hides all margins except a line number margin. All special marker effects are removed and the selection and caret are hidden.

SCI_FORMATRANGE(bool bDraw, RangeToFormat *pfr)
This call allows Windows users to render a range of text into a device context. If you use this for printing, you will probably want to arrange a page header and footer; Scintilla does not do this for you. See SciTEWin::Print() in SciTEWinDlg.cxx for an example. Each use of this message renders a range of text into a rectangular area and returns the position in the document of the next character to print.

bDraw controls if any output is done. Set this to false if you are paginating (for example, if you use this with MFC you will need to paginate in OnBeginPrinting() before you output each page.

struct RangeToFormat {
    SurfaceID hdc;        // The HDC (device context) we print to
    SurfaceID hdcTarget;  // The HDC we use for measuring (may be same as hdc)
    PRectangle rc;        // Rectangle in which to print
    PRectangle rcPage;    // Physically printable page size
    CharacterRange chrg;  // Range of characters to print
};

hdc and hdcTarget should both be set to the device context handle of the output device (usually a printer). If you print to a metafile these will not be the same as Windows metafiles (unlike extended metafiles) do not implement the full API for returning information. In this case, set hdcTarget to the screen DC.
rcPage is the rectangle {0, 0, maxX, maxY} where maxX+1 and maxY+1 are the number of physically printable pixels in x and y.
rc is the rectangle to render the text in (which will, of course, fit within the rectangle defined by rcPage).
chrg.cpMin and chrg.cpMax define the start position and maximum position of characters to output. All of each line within this character range is drawn.

When printing, the most tedious part is always working out what the margins should be to allow for the non-printable area of the paper and printing a header and footer. If you look at the printing code in SciTE, you will find that most of it is taken up with this. The loop that causes Scintilla to render text is quite simple if you strip out all the margin, non-printable area, header and footer code.

SCI_SETPRINTMAGNIFICATION(int magnification)
SCI_GETPRINTMAGNIFICATION
SCI_GETPRINTMAGNIFICATION lets you to print at a different size than the screen font. magnification is the number of points to add to the size of each screen font. A value of -3 or -4 gives reasonably small print. You can get this value with SCI_GETPRINTMAGNIFICATION.

SCI_SETPRINTCOLOURMODE(int mode)
SCI_GETPRINTCOLOURMODE
These two messages set and get the method used to render coloured text on a printer that is probably using white paper. It is especially important to consider the treatment of colour if you use a dark or black screen background. Printing white on black uses up toner and ink very many times faster than the other way around. You can set the mode to one of:

SCI_SETPRINTWRAPMODE(int wrapMode)
SCI_GETPRINTWRAPMODE
These two functions get and set the printer wrap mode. wrapMode can be set to SC_WRAP_NONE (0), SC_WRAP_WORD (1) or SC_WRAP_CHAR (2). The default is SC_WRAP_WORD, which wraps printed output so that all characters fit into the print rectangle. If you set SC_WRAP_NONE, each line of text generates one line of output and the line is truncated if it is too long to fit into the print area.
SC_WRAP_WORD tries to wrap only between words as indicated by white space or style changes although if a word is longer than a line, it will be wrapped before the line end. SC_WRAP_CHAR is preferred to SC_WRAP_WORD for Asian languages where there is no white space between words.

Direct access

On Windows, the message-passing scheme used to communicate between the container and Scintilla is mediated by the operating system SendMessage function and can lead to bad performance when calling intensively. To avoid this overhead, Scintilla provides messages that allow you to call the Scintilla message function directly. The code to do this in C/C++ is of the form:

#include "Scintilla.h"
SciFnDirect pSciMsg = (SciFnDirect)SendMessage(hSciWnd, SCI_GETDIRECTFUNCTION, 0, 0);
sptr_t pSciWndData = (sptr_t)SendMessage(hSciWnd, SCI_GETDIRECTPOINTER, 0, 0);

// now a wrapper to call Scintilla directly
sptr_t CallScintilla(unsigned int iMessage, uptr_t wParam, sptr_t lParam){
    return pSciMsg(pSciWndData, iMessage, wParam, lParam);
}

SciFnDirect, sptr_t and uptr_t are declared in Scintilla.h. hSciWnd is the window handle returned when you created the Scintilla window.

While faster, this direct calling will cause problems if performed from a different thread to the native thread of the Scintilla window in which case SendMessage(hSciWnd, SCI_*, wParam, lParam) should be used to synchronize with the window's thread.

This feature also works on GTK+ but has no significant impact on speed.

From version 1.47 on Windows, Scintilla exports a function called Scintilla_DirectFunction that can be used the same as the function returned by SCI_GETDIRECTFUNCTION. This saves you the call to SCI_GETDIRECTFUNCTION and the need to call Scintilla indirectly via the function pointer.

SCI_GETDIRECTFUNCTION
This message returns the address of the function to call to handle Scintilla messages without the overhead of passing through the Windows messaging system. You need only call this once, regardless of the number of Scintilla windows you create.

SCI_GETDIRECTPOINTER
This returns a pointer to data that identifies which Scintilla window is in use. You must call this once for each Scintilla window you create. When you call the direct function, you must pass in the direct pointer associated with the target window.

SCI_GETCHARACTERPOINTER
Move the gap within Scintilla so that the text of the document is stored consecutively and ensure there is a NUL character after the text, then return a pointer to the first character. Applications may then pass this to a function that accepts a character pointer such as a regular expression search or a parser. The pointer should not be written to as that may desynchronize the internal state of Scintilla.

Since any action in Scintilla may change its internal state this pointer becomes invalid after any call or by allowing user interface activity. The application should reacquire the pointer after making any call to Scintilla or performing any user-interface calls such as modifying a progress indicator.

This call takes similar time to inserting a character at the end of the document and this may include moving the document contents. Specifically, all the characters after the document gap are moved to before the gap. This compacted state should persist over calls and user interface actions that do not change the document contents so reacquiring the pointer afterwards is very quick. If this call is used to implement a global replace operation, then each replacement will move the gap so if SCI_GETCHARACTERPOINTER is called after each replacement then the operation will become O(n^2) rather than O(n). Instead, all matches should be found and remembered, then all the replacements performed.

Multiple views

A Scintilla window and the document that it displays are separate entities. When you create a new window, you also create a new, empty document. Each document has a reference count that is initially set to 1. The document also has a list of the Scintilla windows that are linked to it so when any window changes the document, all other windows in which it appears are notified to cause them to update. The system is arranged in this way so that you can work with many documents in a single Scintilla window and so you can display a single document in multiple windows (for use with splitter windows).

Although these messages use document *pDoc, to ensure compatibility with future releases of Scintilla you should treat pDoc as an opaque void*. That is, you can use and store the pointer as described in this section but you should not dereference it.

SCI_GETDOCPOINTER
This returns a pointer to the document currently in use by the window. It has no other effect.

SCI_SETDOCPOINTER(<unused>, document *pDoc)
This message does the following:
1. It removes the current window from the list held by the current document.
2. It reduces the reference count of the current document by 1.
3. If the reference count reaches 0, the document is deleted.
4. pDoc is set as the new document for the window.
5. If pDoc was 0, a new, empty document is created and attached to the window.
6. If pDoc was not 0, its reference count is increased by 1.

SCI_CREATEDOCUMENT
This message creates a new, empty document and returns a pointer to it. This document is not selected into the editor and starts with a reference count of 1. This means that you have ownership of it and must either reduce its reference count by 1 after using SCI_SETDOCPOINTER so that the Scintilla window owns it or you must make sure that you reduce the reference count by 1 with SCI_RELEASEDOCUMENT before you close the application to avoid memory leaks.

SCI_ADDREFDOCUMENT(<unused>, document *pDoc)
This increases the reference count of a document by 1. If you want to replace the current document in the Scintilla window and take ownership of the current document, for example if you are editing many documents in one window, do the following:
1. Use SCI_GETDOCPOINTER to get a pointer to the document, pDoc.
2. Use SCI_ADDREFDOCUMENT(0, pDoc) to increment the reference count.
3. Use SCI_SETDOCPOINTER(0, pNewDoc) to set a different document or SCI_SETDOCPOINTER(0, 0) to set a new, empty document.

SCI_RELEASEDOCUMENT(<unused>, document *pDoc)
This message reduces the reference count of the document identified by pDoc. pDoc must be the result of SCI_GETDOCPOINTER or SCI_CREATEDOCUMENT and must point at a document that still exists. If you call this on a document with a reference count of 1 that is still attached to a Scintilla window, bad things will happen. To keep the world spinning in its orbit you must balance each call to SCI_CREATEDOCUMENT or SCI_ADDREFDOCUMENT with a call to SCI_RELEASEDOCUMENT.

Folding

The fundamental operation in folding is making lines invisible or visible. Line visibility is a property of the view rather than the document so each view may be displaying a different set of lines. From the point of view of the user, lines are hidden and displayed using fold points. Generally, the fold points of a document are based on the hierarchical structure of the document contents. In Python, the hierarchy is determined by indentation and in C++ by brace characters. This hierarchy can be represented within a Scintilla document object by attaching a numeric "fold level" to each line. The fold level is most easily set by a lexer, but you can also set it with messages.

It is up to your code to set the connection between user actions and folding and unfolding. The best way to see how this is done is to search the SciTE source code for the messages used in this section of the documentation and see how they are used. You will also need to use markers and a folding margin to complete your folding implementation. The "fold" property should be set to "1" with SCI_SETPROPERTY("fold", "1") to enable folding.

SCI_VISIBLEFROMDOCLINE(int docLine)
When some lines are folded, then a particular line in the document may be displayed at a different position to its document position. If no lines are folded, this message returns docLine. Otherwise, this returns the display line (counting the very first visible line as 0). The display line of an invisible line is the same as the previous visible line. The display line number of the first line in the document is 0. If there is folding and docLine is outside the range of lines in the document, the return value is -1. Lines can occupy more than one display line if they wrap.

SCI_DOCLINEFROMVISIBLE(int displayLine)
When some lines are hidden, then a particular line in the document may be displayed at a different position to its document position. This message returns the document line number that corresponds to a display line (counting the display line of the first line in the document as 0). If displayLine is less than or equal to 0, the result is 0. If displayLine is greater than or equal to the number of displayed lines, the result is the number of lines in the document.

SCI_SHOWLINES(int lineStart, int lineEnd)
SCI_HIDELINES(int lineStart, int lineEnd)
SCI_GETLINEVISIBLE(int line)
The first two messages mark a range of lines as visible or invisible and then redraw the display. The third message reports on the visible state of a line and returns 1 if it is visible and 0 if it is not visible. These messages have no effect on fold levels or fold flags. The first line can not be hidden.

SCI_SETFOLDLEVEL(int line, int level)
SCI_GETFOLDLEVEL(int line)
These two messages set and get a 32-bit value that contains the fold level of a line and some flags associated with folding. The fold level is a number in the range 0 to SC_FOLDLEVELNUMBERMASK (4095). However, the initial fold level is set to SC_FOLDLEVELBASE (1024) to allow unsigned arithmetic on folding levels. There are two addition flag bits. SC_FOLDLEVELWHITEFLAG indicates that the line is blank and allows it to be treated slightly different then its level may indicate. For example, blank lines should generally not be fold points and will be considered part of the preceding section even though they may have a lesser fold level. SC_FOLDLEVELHEADERFLAG indicates that the line is a header (fold point).

Use SCI_GETFOLDLEVEL(line) & SC_FOLDLEVELNUMBERMASK to get the fold level of a line. Likewise, use SCI_GETFOLDLEVEL(line) & SC_FOLDLEVEL*FLAG to get the state of the flags. To set the fold level you must or in the associated flags. For instance, to set the level to thisLevel and mark a line as being a fold point use: SCI_SETFOLDLEVEL(line, thisLevel | SC_FOLDLEVELHEADERFLAG).

SCI_SETFOLDFLAGS(int flags)
In addition to showing markers in the folding margin, you can indicate folds to the user by drawing lines in the text area. The lines are drawn in the foreground colour set for STYLE_DEFAULT. Bits set in flags determine where folding lines are drawn:

This message causes the display to redraw.

SCI_GETLASTCHILD(int startLine, int level)
This message searches for the next line after startLine, that has a folding level that is less than or equal to level and then returns the previous line number. If you set level to -1, level is set to the folding level of line startLine. If from is a fold point, SCI_GETLASTCHILD(from, -1) returns the last line that would be in made visible or hidden by toggling the fold state.

SCI_GETFOLDPARENT(int startLine)
This message returns the line number of the first line before startLine that is marked as a fold point with SC_FOLDLEVELHEADERFLAG and has a fold level less than the startLine. If no line is found, or if the header flags and fold levels are inconsistent, the return value is -1.

SCI_TOGGLEFOLD(int line)
Each fold point may be either expanded, displaying all its child lines, or contracted, hiding all the child lines. This message toggles the folding state of the given line as long as it has the SC_FOLDLEVELHEADERFLAG set. This message takes care of folding or expanding all the lines that depend on the line. The display updates after this message.

SCI_SETFOLDEXPANDED(int line, bool expanded)
SCI_GETFOLDEXPANDED(int line)
These messages set and get the expanded state of a single line. The set message has no effect on the visible state of the line or any lines that depend on it. It does change the markers in the folding margin. If you ask for the expansion state of a line that is outside the document, the result is false (0).

If you just want to toggle the fold state of one line and handle all the lines that are dependent on it, it is much easier to use SCI_TOGGLEFOLD. You would use the SCI_SETFOLDEXPANDED message to process many folds without updating the display until you had finished. See SciTEBase::FoldAll() and SciTEBase::Expand() for examples of the use of these messages.

SCI_ENSUREVISIBLE(int line)
SCI_ENSUREVISIBLEENFORCEPOLICY(int line)
A line may be hidden because more than one of its parent lines is contracted. Both these message travels up the fold hierarchy, expanding any contracted folds until they reach the top level. The line will then be visible. If you use SCI_ENSUREVISIBLEENFORCEPOLICY, the vertical caret policy set by SCI_SETVISIBLEPOLICY is then applied.

Line wrapping

By default, Scintilla does not wrap lines of text. If you enable line wrapping, lines wider than the window width are continued on the following lines. Lines are broken after space or tab characters or between runs of different styles. If this is not possible because a word in one style is wider than the window then the break occurs after the last character that completely fits on the line. The horizontal scroll bar does not appear when wrap mode is on.

For wrapped lines Scintilla can draw visual flags (little arrows) at end of a a subline of a wrapped line and at begin of the next subline. These can be enabled individually, but if Scintilla draws the visual flag at begin of the next subline this subline will be indented by one char. Independent from drawing a visual flag at the begin the subline can have an indention.

Much of the time used by Scintilla is spent on laying out and drawing text. The same text layout calculations may be performed many times even when the data used in these calculations does not change. To avoid these unnecessary calculations in some circumstances, the line layout cache can store the results of the calculations. The cache is invalidated whenever the underlying data, such as the contents or styling of the document changes. Caching the layout of the whole document has the most effect, making dynamic line wrap as much as 20 times faster but this requires 7 times the memory required by the document contents plus around 80 bytes per line.

Wrapping is not performed immediately there is a change but is delayed until the display is redrawn. This delay improves peformance by allowing a set of changes to be performed and then wrapped and displayed once. Because of this, some operations may not occur as expected. If a file is read and the scroll position moved to a particular line in the text, such as occurs when a container tries to restore a previous editing session, then the scroll position will have been determined before wrapping so an unexpected range of text will be displayed. To scroll to the position correctly, delay the scroll until the wrapping has been performed by waiting for an initial SCN_PAINTED notification.

SCI_SETWRAPMODE(int wrapMode)
SCI_GETWRAPMODE
Set wrapMode to SC_WRAP_WORD (1) to enable wrapping on word boundaries, SC_WRAP_CHAR (2) to enable wrapping between any characters, and to SC_WRAP_NONE (0) to disable line wrapping. SC_WRAP_CHAR is preferred to SC_WRAP_WORD for Asian languages where there is no white space between words.

SCI_SETWRAPVISUALFLAGS(int wrapVisualFlags)
SCI_GETWRAPVISUALFLAGS
You can enable the drawing of visual flags to indicate a line is wrapped. Bits set in wrapVisualFlags determine which visual flags are drawn.

SCI_SETWRAPVISUALFLAGSLOCATION(int wrapVisualFlagsLocation)
SCI_GETWRAPVISUALFLAGSLOCATION
You can set wether the visual flags to indicate a line is wrapped are drawn near the border or near the text. Bits set in wrapVisualFlagsLocation set the location to near the text for the corresponding visual flag.

SCI_SETWRAPSTARTINDENT(int indent)
SCI_GETWRAPSTARTINDENT
SCI_SETWRAPSTARTINDENT sets the size of indentation of sublines for wrapped lines in terms of the width of a space in STYLE_DEFAULT. There are no limits on indent sizes, but values less than 0 or large values may have undesirable effects.
The indention of sublines is independent of visual flags, but if SC_WRAPVISUALFLAG_START is set an indent of at least 1 is used.

SCI_SETLAYOUTCACHE(int cacheMode)
SCI_GETLAYOUTCACHE
You can set cacheMode to one of the symbols in the table:

SCI_SETPOSITIONCACHE(int size)
SCI_GETPOSITIONCACHE
The position cache stores position information for short runs of text so that their layout can be determined more quickly if the run recurs. The size in entries of this cache can be set with SCI_SETPOSITIONCACHE.

SCI_LINESSPLIT(int pixelWidth)
Split a range of lines indicated by the target into lines that are at most pixelWidth wide. Splitting occurs on word boundaries wherever possible in a similar manner to line wrapping. When pixelWidth is 0 then the width of the window is used.

SCI_LINESJOIN
Join a range of lines indicated by the target into one line by removing line end characters. Where this would lead to no space between words, an extra space is inserted.

SCI_WRAPCOUNT(int docLine)
Document lines can occupy more than one display line if they wrap and this returns the number of display lines needed to wrap a document line.

Zooming

Scintilla incorporates a "zoom factor" that lets you make all the text in the document larger or smaller in steps of one point. The displayed point size never goes below 2, whatever zoom factor you set. You can set zoom factors in the range -10 to +20 points.

SCI_ZOOMIN
SCI_ZOOMOUT
SCI_ZOOMIN increases the zoom factor by one point if the current zoom factor is less than 20 points. SCI_ZOOMOUT decreases the zoom factor by one point if the current zoom factor is greater than -10 points.

SCI_SETZOOM(int zoomInPoints)
SCI_GETZOOM
These messages let you set and get the zoom factor directly. There is no limit set on the factors you can set, so limiting yourself to -10 to +20 to match the incremental zoom functions is a good idea.

Long lines

You can choose to mark lines that exceed a given length by drawing a vertical line or by colouring the background of characters that exceed the set length.

SCI_SETEDGEMODE(int edgeMode)
SCI_GETEDGEMODE
These two messages set and get the mode used to display long lines. You can set one of the values in the table:

SCI_SETEDGECOLUMN(int column)
SCI_GETEDGECOLUMN
These messages set and get the column number at which to display the long line marker. When drawing lines, the column sets a position in units of the width of a space character in STYLE_DEFAULT. When setting the background colour, the column is a character count (allowing for tabs) into the line.

SCI_SETEDGECOLOUR(int colour)
SCI_GETEDGECOLOUR
These messages set and get the colour of the marker used to show that a line has exceeded the length set by SCI_SETEDGECOLUMN.

Lexer

If you define the symbol SCI_LEXER when building Scintilla, (this is sometimes called the SciLexer version of Scintilla), lexing support for a wide range programming languages is included and the messages in this section are supported. If you want to set styling and fold points for an unsupported language you can either do this in the container or better still, write your own lexer following the pattern of one of the existing ones.

Scintilla also supports external lexers. These are DLLs (on Windows) or .so modules (on GTK+/Linux) that export four functions: GetLexerCount, GetLexerName, Lex and Fold. See externalLexer.cxx for more.

SCI_SETLEXER(int lexer)
SCI_GETLEXER
You can select the lexer to use with an integer code from the SCLEX_* enumeration in Scintilla.h. There are two codes in this sequence that do not use lexers: SCLEX_NULL to select no lexing action and SCLEX_CONTAINER which sends the SCN_STYLENEEDED notification to the container whenever a range of text needs to be styled. You cannot use the SCLEX_AUTOMATIC value; this identifies additional external lexers that Scintilla assigns unused lexer numbers to.

SCI_SETLEXERLANGUAGE(<unused>, const char *name)
This message lets you select a lexer by name, and is the only method if you are using an external lexer or if you have written a lexer module for a language of your own and do not wish to assign it an explicit lexer number. To select an existing lexer, set name to match the (case sensitive) name given to the module, for example "ada" or "python", not "Ada" or "Python". To locate the name for the built-in lexers, open the relevant Lex*.cxx file and search for LexerModule. The third argument in the LexerModule constructor is the name to use.

To test if your lexer assignment worked, use SCI_GETLEXER before and after setting the new lexer to see if the lexer number changed.

SCI_LOADLEXERLIBRARY(<unused>, const char *path)
Load a lexer implemented in a shared library. This is a .so file on GTK+/Linux or a .DLL file on Windows.

SCI_COLOURISE(int startPos, int endPos)
This requests the current lexer or the container (if the lexer is set to SCLEX_CONTAINER) to style the document between startPos and endPos. If endPos is -1, the document is styled from startPos to the end. If the "fold" property is set to "1" and your lexer or container supports folding, fold levels are also set. This message causes a redraw.

SCI_SETPROPERTY(const char *key, const char *value)
You can communicate settings to lexers with keyword:value string pairs. There is no limit to the number of keyword pairs you can set, other than available memory. key is a case sensitive keyword, value is a string that is associated with the keyword. If there is already a value string associated with the keyword, it is replaced. If you pass a zero length string, the message does nothing. Both key and value are used without modification; extra spaces at the beginning or end of key are significant.

The value string can refer to other keywords. For example, SCI_SETPROPERTY("foldTimes10", "$(fold)0") stores the string "$(fold)0", but when this is accessed, the $(fold) is replaced by the value of the "fold" keyword (or by nothing if this keyword does not exist).

Currently the "fold" property is defined for most of the lexers to set the fold structure if set to "1". SCLEX_PYTHON understands "tab.timmy.whinge.level" as a setting that determines how to indicate bad indentation. Most keywords have values that are interpreted as integers. Search the lexer sources for GetPropertyInt to see how properties are used.

There is a convention for naming properties used by lexers so that the set of properties can be found by scripts. Property names should start with "lexer.<lexer>." or "fold.<lexer>." when they apply to one lexer or start with "lexer." or "fold." if they apply to multiple lexers.

SCI_GETPROPERTY(const char *key, char *value)
Lookup a keyword:value pair using the specified key; if found, copy the value to the user-supplied buffer and return the length (not including the terminating 0). If not found, copy an empty string to the buffer and return 0.

Note that "keyword replacement" as described in SCI_SETPROPERTY will not be performed.

If the value argument is 0 then the length that should be allocated to store the value is returned; again, the terminating 0 is not included.

SCI_GETPROPERTYEXPANDED(const char *key, char *value)
Lookup a keyword:value pair using the specified key; if found, copy the value to the user-supplied buffer and return the length (not including the terminating 0). If not found, copy an empty string to the buffer and return 0.

Note that "keyword replacement" as described in SCI_SETPROPERTY will be performed.

If the value argument is 0 then the length that should be allocated to store the value (including any indicated keyword replacement) is returned; again, the terminating 0 is not included.

SCI_GETPROPERTYINT(const char *key, int default)
Lookup a keyword:value pair using the specified key; if found, interpret the value as an integer and return it. If not found (or the value is an empty string) then return the supplied default. If the keyword:value pair is found but is not a number, then return 0.

Note that "keyword replacement" as described in SCI_SETPROPERTY will be performed before any numeric interpretation.

SCI_SETKEYWORDS(int keyWordSet, const char *keyWordList)
You can set up to 9 lists of keywords for use by the current lexer. This was increased from 6 at revision 1.50. keyWordSet can be 0 to 8 (actually 0 to KEYWORDSET_MAX) and selects which keyword list to replace. keyWordList is a list of keywords separated by spaces, tabs, "\n" or "\r" or any combination of these. It is expected that the keywords will be composed of standard ASCII printing characters, but there is nothing to stop you using any non-separator character codes from 1 to 255 (except common sense).

How these keywords are used is entirely up to the lexer. Some languages, such as HTML may contain embedded languages, VBScript and JavaScript are common for HTML. For HTML, key word set 0 is for HTML, 1 is for JavaScript and 2 is for VBScript, 3 is for Python, 4 is for PHP and 5 is for SGML and DTD keywords. Review the lexer code to see examples of keyword list. A fully conforming lexer sets the fourth argument of the LexerModule constructor to be a list of strings that describe the uses of the keyword lists.

Alternatively, you might use set 0 for general keywords, set 1 for keywords that cause indentation and set 2 for keywords that cause unindentation. Yet again, you might have a simple lexer that colours keywords and you could change languages by changing the keywords in set 0. There is nothing to stop you building your own keyword lists into the lexer, but this means that the lexer must be rebuilt if more keywords are added.

SCI_GETSTYLEBITSNEEDED
Retrieve the number of bits the current lexer needs for styling. This should normally be the argument to SCI_SETSTYLEBITS.

Notifications

Notifications are sent (fired) from the Scintilla control to its container when an event has occurred that may interest the container. Notifications are sent using the WM_NOTIFY message on Windows and the "notify" signal on GTK+. The container is passed a SCNotification structure containing information about the event.

struct NotifyHeader {   // This matches the Win32 NMHDR structure
    void *hwndFrom;     // environment specific window handle/pointer
    uptr_t idFrom;        // CtrlID of the window issuing the notification
    unsigned int code;  // The SCN_* notification code
};

struct SCNotification {
    struct NotifyHeader nmhdr;
    int position;
    // SCN_STYLENEEDED, SCN_DOUBLECLICK, SCN_MODIFIED, SCN_DWELLSTART,
    // SCN_DWELLEND, SCN_CALLTIPCLICK,
    // SCN_HOTSPOTCLICK, SCN_HOTSPOTDOUBLECLICK
    int ch;             // SCN_CHARADDED, SCN_KEY
    int modifiers;      // SCN_KEY, SCN_DOUBLECLICK, SCN_HOTSPOTCLICK, SCN_HOTSPOTDOUBLECLICK
    int modificationType; // SCN_MODIFIED
    const char *text;   // SCN_MODIFIED, SCN_USERLISTSELECTION, SCN_AUTOCSELECTION
    int length;         // SCN_MODIFIED
    int linesAdded;     // SCN_MODIFIED
    int message;        // SCN_MACRORECORD
    uptr_t wParam;      // SCN_MACRORECORD
    sptr_t lParam;      // SCN_MACRORECORD
    int line;           // SCN_MODIFIED, SCN_DOUBLECLICK
    int foldLevelNow;   // SCN_MODIFIED
    int foldLevelPrev;  // SCN_MODIFIED
    int margin;         // SCN_MARGINCLICK
    int listType;       // SCN_USERLISTSELECTION, SCN_AUTOCSELECTION
    int x;              // SCN_DWELLSTART, SCN_DWELLEND
    int y;              // SCN_DWELLSTART, SCN_DWELLEND
};

The notification messages that your container can choose to handle and the messages associated with them are:

The following SCI_* messages are associated with these notifications:

The following additional notifications are sent using the WM_COMMAND message on Windows and the "Command" signal on GTK+. This emulates the Windows Edit control. Only the lower 16 bits of the control's ID is passed in these notifications.

SCN_STYLENEEDED
If you used SCI_SETLEXER(SCLEX_CONTAINER) to make the container act as the lexer, you will receive this notification when Scintilla is about to display or print text that requires styling. You are required to style the text from the line that contains the position returned by SCI_GETENDSTYLED up to the position passed in SCNotification.position. Symbolically, you need code of the form:

    startPos = SCI_GETENDSTYLED()
    lineNumber = SCI_LINEFROMPOSITION(startPos);
    startPos = SCI_POSITIONFROMLINE(lineNumber);
    MyStyleRoutine(startPos, SCNotification.position);

SCN_CHARADDED
This is sent when the user types an ordinary text character (as opposed to a command character) that is entered into the text. The container can use this to decide to display a call tip or an auto completion list. The character is in SCNotification.ch. This notification is sent before the character has been styled so processing that depends on styling should instead be performed in the SCN_UPDATEUI notification.

SCN_SAVEPOINTREACHED
SCN_SAVEPOINTLEFT
Sent to the container when the save point is entered or left, allowing the container to display a "document dirty" indicator and change its menus.
See also: SCI_SETSAVEPOINT, SCI_GETMODIFY

SCN_MODIFYATTEMPTRO
When in read-only mode, this notification is sent to the container if the user tries to change the text. This can be used to check the document out of a version control system. You can set the read-only state of a document with SCI_SETREADONLY.

SCN_KEY
Reports all keys pressed but not consumed by Scintilla. Used on GTK+ because of some problems with keyboard focus and is not sent by the Windows version. SCNotification.ch holds the key code and SCNotification.modifiers holds the modifiers. This notification is sent if the modifiers include SCMOD_ALT or SCMOD_CTRL and the key code is less than 256.

SCN_DOUBLECLICK
The mouse button was double clicked in editor. The position field is set to the text position of the double click and the line field is set to the line of the double click.

SCN_UPDATEUI
Either the text or styling of the document has changed or the selection range has changed. Now would be a good time to update any container UI elements that depend on document or view state. This was previously called SCN_CHECKBRACE because a common use is to check whether the caret is next to a brace and set highlights on this brace and its corresponding matching brace. This also replaces SCN_POSCHANGED, which is now deprecated.

SCN_MODIFIED
This notification is sent when the text or styling of the document changes or is about to change. You can set a mask for the notifications that are sent to the container with SCI_SETMODEVENTMASK. The notification structure contains information about what changed, how the change occurred and whether this changed the number of lines in the document. No modifications may be performed while in a SCN_MODIFIED event. The SCNotification fields used are:

The SCNotification.modificationType field has bits set to tell you what has been done. The SC_MOD_* bits correspond to actions. The SC_PERFORMED_* bits tell you if the action was done by the user, or the result of Undo or Redo of a previous action.

SCEN_CHANGE
SCEN_CHANGE (768) is fired when the text (not the style) of the document changes. This notification is sent using the WM_COMMAND message on Windows and the "Command" signal on GTK+ as this is the behavior of the standard Edit control (SCEN_CHANGE has the same value as the Windows Edit control EN_CHANGE). No other information is sent. If you need more detailed information use SCN_MODIFIED. You can filter the types of changes you are notified about with SCI_SETMODEVENTMASK.

SCI_SETMODEVENTMASK(int eventMask)
SCI_GETMODEVENTMASK
These messages set and get an event mask that determines which document change events are notified to the container with SCN_MODIFIED and SCEN_CHANGE. For example, a container may decide to see only notifications about changes to text and not styling changes by calling SCI_SETMODEVENTMASK(SC_MOD_INSERTTEXT|SC_MOD_DELETETEXT).

The possible notification types are the same as the modificationType bit flags used by SCN_MODIFIED: SC_MOD_INSERTTEXT, SC_MOD_DELETETEXT, SC_MOD_CHANGESTYLE, SC_MOD_CHANGEFOLD, SC_PERFORMED_USER, SC_PERFORMED_UNDO, SC_PERFORMED_REDO, SC_MULTISTEPUNDOREDO, SC_LASTSTEPINUNDOREDO, SC_MOD_CHANGEMARKER, SC_MOD_BEFOREINSERT, SC_MOD_BEFOREDELETE, SC_MULTILINEUNDOREDO, and SC_MODEVENTMASKALL.

SCEN_SETFOCUS
SCEN_KILLFOCUS
SCEN_SETFOCUS (512) is fired when Scintilla receives focus and SCEN_KILLFOCUS (256) when it loses focus. These notifications are sent using the WM_COMMAND message on Windows and the "Command" signal on GTK+ as this is the behavior of the standard Edit control. Unfortunately, these codes do not match the Windows Edit notification codes EN_SETFOCUS (256) and EN_KILLFOCUS (512). It is now too late to change the Scintilla codes as clients depend on the current values.

SCN_MACRORECORD
The SCI_STARTRECORD and SCI_STOPRECORD messages enable and disable macro recording. When enabled, each time a recordable change occurs, the SCN_MACRORECORD notification is sent to the container. It is up to the container to record the action. To see the complete list of SCI_* messages that are recordable, search the Scintilla source Editor.cxx for Editor::NotifyMacroRecord. The fields of SCNotification set in this notification are:

SCN_MARGINCLICK
This notification tells the container that the mouse was clicked inside a margin that was marked as sensitive (see SCI_SETMARGINSENSITIVEN). This can be used to perform folding or to place breakpoints. The following SCNotification fields are used:

SCN_NEEDSHOWN
Scintilla has determined that a range of lines that is currently invisible should be made visible. An example of where this may be needed is if the end of line of a contracted fold point is deleted. This message is sent to the container in case it wants to make the line visible in some unusual way such as making the whole document visible. Most containers will just ensure each line in the range is visible by calling SCI_ENSUREVISIBLE. The position and length fields of SCNotification indicate the range of the document that should be made visible. The container code will be similar to the following code skeleton:

firstLine = SCI_LINEFROMPOSITION(scn.position)
lastLine = SCI_LINEFROMPOSITION(scn.position+scn.length-1)
for line = lineStart to lineEnd do SCI_ENSUREVISIBLE(line) next

SCN_PAINTED
Painting has just been done. Useful when you want to update some other widgets based on a change in Scintilla, but want to have the paint occur first to appear more responsive. There is no other information in SCNotification.

SCN_USERLISTSELECTION
The user has selected an item in a user list. The SCNotification fields used are:

SCN_URIDROPPED
Only on the GTK+ version. Indicates that the user has dragged a URI such as a file name or Web address onto Scintilla. The container could interpret this as a request to open the file. The text field of SCNotification points at the URI text.

SCN_DWELLSTART
SCN_DWELLEND
SCN_DWELLSTART is generated when the user keeps the mouse in one position for the dwell period (see SCI_SETMOUSEDWELLTIME). SCN_DWELLEND is generated after a SCN_DWELLSTART and the mouse is moved or other activity such as key press indicates the dwell is over. Both notifications set the same fields in SCNotification:

SCI_SETMOUSEDWELLTIME
SCI_GETMOUSEDWELLTIME
These two messages set and get the time the mouse must sit still, in milliseconds, to generate a SCN_DWELLSTART notification. If set to SC_TIME_FOREVER, the default, no dwell events are generated.

SCN_ZOOM
This notification is generated when the user zooms the display using the keyboard or the SCI_SETZOOM method is called. This notification can be used to recalculate positions, such as the width of the line number margin to maintain sizes in terms of characters rather than pixels. SCNotification has no additional information.

SCN_HOTSPOTCLICK
SCN_HOTSPOTDOUBLECLICK
These notifications are generated when the user clicks or double clicks on text that is in a style with the hotspot attribute set. This notification can be used to link to variable definitions or web pages. The position field is set the text position of the click or double click and the modifiers field set to the key modifiers held down in a similar manner to SCN_KEY.

SCN_INDICATORCLICK
SCN_INDICATORRELEASE
These notifications are generated when the user clicks or releases the mouse on text that has an indicator. The position field is set the text position of the click or double click and the modifiers field set to the key modifiers held down in a similar manner to SCN_KEY.

SCN_CALLTIPCLICK
This notification is generated when the user clicks on a calltip. This notification can be used to display the next function prototype when a function name is overloaded with different arguments. The position field is set to 1 if the click is in an up arrow, 2 if in a down arrow, and 0 if elsewhere.

SCN_AUTOCSELECTION
The user has selected an item in an autocompletion list. The notification is sent before the selection is inserted. Automatic insertion can be cancelled by sending a SCI_AUTOCCANCEL message before returning from the notification. The SCNotification fields used are:

SCN_AUTOCCANCELLED
The user has cancelled an autocompletion list. There is no other information in SCNotification.

GTK+

On GTK+, the following functions create a Scintilla widget, communicate with it and allow resources to be released after all Scintilla widgets have been destroyed.

GtkWidget *scintilla_new()
Create a new Scintilla widget. The returned pointer can be added to a container and displayed in the same way as other widgets.

void scintilla_set_id(ScintillaObject *sci, uptr_t id)
Set the control ID which will be used in the idFrom field of the NotifyHeader structure of all notifications for this instance. When an application creates multiple Scintilla widgets, this allows the source of each notification to be found. The value should be small, preferrably less than 16 bits, rather than a pointer as some of the functions will only transmit 16 or 32 bits.

sptr_t scintilla_send_message(ScintillaObject *sci,unsigned int iMessage, uptr_t wParam, sptr_t lParam)
The main entry point allows sending any of the messages described in this document.

void scintilla_release_resources()
Call this to free any remaining resources after all the Scintilla widgets have been destroyed.

Deprecated messages and notifications

The following messages are currently supported to emulate existing Windows controls, but they will be removed in future versions of Scintilla. If you use these messages you should replace them with the Scintilla equivalent.

WM_GETTEXT(int length, char *text)
WM_SETTEXT(<unused>, const char *text)
EM_GETLINE(int line, char *text)
EM_REPLACESEL(<unused>, const char *text)
EM_SETREADONLY
EM_GETTEXTRANGE(<unused>, TEXTRANGE *tr)
WM_CUT
WM_COPY
WM_PASTE
WM_CLEAR
WM_UNDO
EM_CANUNDO
EM_EMPTYUNDOBUFFER
WM_GETTEXTLENGTH
EM_GETFIRSTVISIBLELINE
EM_GETLINECOUNT
EM_GETMODIFY
EM_SETMODIFY(bool isModified)
EM_GETRECT(RECT *rect)
EM_GETSEL(int *start, int *end)
EM_EXGETSEL(<unused>, CHARRANGE *cr)
EM_SETSEL(int start, int end)
EM_EXSETSEL(<unused>, CHARRANGE *cr)
EM_GETSELTEXT(<unused>, char *text)
EM_LINEFROMCHAR(int position)
EM_EXLINEFROMCHAR(int position)
EM_LINEINDEX(int line)
EM_LINELENGTH(int position)
EM_SCROLL(int line)
EM_LINESCROLL(int column, int line)
EM_SCROLLCARET()
EM_CANPASTE
EM_CHARFROMPOS(<unused>, POINT *location)
EM_POSFROMCHAR(int position, POINT *location)
EM_SELECTIONTYPE
EM_HIDESELECTION(bool hide)
EM_FINDTEXT(int flags, FINDTEXTEX *ft)
EM_FINDTEXTEX(int flags, FINDTEXTEX *ft)
EM_GETMARGINS
EM_SETMARGINS(EC_LEFTMARGIN or EC_RIGHTMARGIN or EC_USEFONTINFO, int val)
EM_FORMATRANGE

The following are features that are only included if you define INCLUDE_DEPRECATED_FEATURES in Scintilla.h. To ensure future compatibility you should change them as indicated.

SCN_POSCHANGED() Deprecated
Fired when the user moves the cursor to a different position in the text. Use SCN_UPDATEUI instead.

SCN_CHECKBRACE Deprecated
Either the text or styling of the document has changed or the selection range has changed. This is replaced by SCN_UPDATEUI. You can also use SCN_MODIFIED for more detailed information on text and styling changes,

Edit messages never supported by Scintilla

EM_GETWORDBREAKPROC EM_GETWORDBREAKPROCEX
EM_SETWORDBREAKPROC EM_SETWORDBREAKPROCEX
EM_GETWORDWRAPMODE EM_SETWORDWRAPMODE
EM_LIMITTEXT EM_EXLIMITTEXT
EM_SETRECT EM_SETRECTNP
EM_FMTLINES
EM_GETHANDLE EM_SETHANDLE
EM_GETPASSWORDCHAR EM_SETPASSWORDCHAR
EM_SETTABSTOPS
EM_FINDWORDBREAK
EM_GETCHARFORMAT EM_SETCHARFORMAT
EM_GETOLEINTERFACE EM_SETOLEINTERFACE
EM_SETOLECALLBACK
EM_GETPARAFORMAT EM_SETPARAFORMAT
EM_PASTESPECIAL
EM_REQUESTRESIZE
EM_GETBKGNDCOLOR EM_SETBKGNDCOLOR
EM_STREAMIN EM_STREAMOUT
EM_GETIMECOLOR EM_SETIMECOLOR
EM_GETIMEOPTIONS EM_SETIMEOPTIONS
EM_GETOPTIONS EM_SETOPTIONS
EM_GETPUNCTUATION EM_SETPUNCTUATION
EM_GETTHUMB
EM_GETEVENTMASK
EM_SETEVENTMASK
EM_DISPLAYBAND
EM_SETTARGETDEVICE

Scintilla tries to be a superset of the standard windows Edit and RichEdit controls wherever that makes sense. As it is not intended for use in a word processor, some edit messages can not be sensibly handled. Unsupported messages have no effect.

Building Scintilla

To build Scintilla or SciTE, see the README file present in both the Scintilla and SciTE directories. For Windows, GCC 3.2, Borland C++ or Microsoft Visual Studio .NET can be used for building. There is a make file for building Scintilla but not SciTE with Visual C++ 6 at scintilla/win32/scintilla_vc6.mak. For GTK+, GCC 3.1 should be used. GTK+ 1.2x and 2.0x are supported. The version of GTK+ installed should be detected automatically. When both GTK+ 1 and GTK+ 2 are present, building for GTK+ 1.x requires defining GTK1 on the command line.

Static linking

On Windows, Scintilla is normally used as a dynamic library as a .DLL file. If you want to link Scintilla directly into your application .EXE or .DLL file, then the STATIC_BUILD preprocessor symbol should be defined and Scintilla_RegisterClasses called. STATIC_BUILD prevents compiling the DllMain function which will conflict with any DllMain defined in your code. Scintilla_RegisterClasses takes the HINSTANCE of your application and ensures that the "Scintilla" window class is registered. To make sure that the right pointing arrow cursor used in the margin is displayed by Scintilla add the scintilla/win32/Margin.cur file to your application's resources with the ID IDC_MARGIN which is defined in scintilla/win32/platfromRes.h as 400.

Ensuring lexers are linked into Scintilla

Depending on the compiler and linker used, the lexers may be stripped out. This is most often caused when building a static library. To ensure the lexers are linked in, the Scintilla_LinkLexers() function may be called.

Changing set of lexers

To change the set of lexers in Scintilla, add and remove lexer source files (Lex*.cxx) from the scintilla/src directory and run the src/LexGen.py script from the src directory to update the make files and KeyWords.cxx. LexGen.py requires Python 2.1 or later. If you do not have access to Python, you can hand edit KeyWords.cxx in a simple-minded way, following the patterns of other lexers. The important thing is to include LINK_LEXER(lmMyLexer); to correspond with the LexerModule lmMyLexer(...); in your lexer source code.

Building with an alternative Regular Expression implementation

A simple interface provides support for switching the Regular Expressions engine at compile time. You must implement RegexSearchBase for your chosen engine, look at the built-in implementation BuiltinRegex to see how this is done. You then need to implement the factory method CreateRegexSearch to create an instance of your class. You must disable the built-in implementation by defining SCI_OWNREGEX.