Split string detection routines into common code.

Subversion-branch: /research
Subversion-revision: 1888

hhe/common.rb

@@ -77,15 +77,96 @@ THING_FLAGS2 = %w{
     MF2_DONTDRAW
 }
 
-def set_config(name)
+# HHE seems to stop when it has found this many strings:
+
+NUM_STRINGS=785
+
+def find_config(name)
 	configs = [ Heretic_1_0, Heretic_1_2, Heretic_1_3 ]
 
 	for config in configs
 		if name == config::NAME
-			include config
-			return
+			return config
 		end
 	end
 
 	raise "Unknown configuration: #{name}"
 end
+
+def set_config(name)
+	config = find_config(name)
+	include config
+end
+
+def read_block(file)
+        result = ""
+
+        4.times do
+                break if file.eof?
+                c = file.readchar
+                result += sprintf("%c", c)
+        end
+
+        end_index = result.index(0)
+        if end_index
+                result = result[0, end_index]
+        end
+
+        result
+end
+
+def bad_string?(str)
+	str.each_byte do |b|
+		if b >= 0x80
+			return true
+		end
+	end
+
+	return false
+end
+
+def find_strings(filename, config)
+
+	strings = []
+
+	File.open(filename) do |file|
+		current_str = ""
+
+		file.seek(config::STRINGS_OFFSET)
+
+		offset = 0
+		start_offset = nil
+
+		while strings.length < NUM_STRINGS
+			block = read_block(file)
+
+			if start_offset == nil
+				start_offset = offset
+			end
+
+			current_str += block
+
+			# End of string?
+
+			if block.length < 4
+
+				# Extended-ASCII characters cannot be output
+				# to JSON.
+
+				if bad_string?(current_str)
+					current_str = nil
+				end
+
+				strings.push([start_offset, current_str])
+
+				current_str = ""
+				start_offset = nil
+			end
+
+			offset += 4
+		end
+	end
+
+	strings
+end
+

hhe/dump-strings

@@ -3,78 +3,17 @@
 require "json"
 require "common"
 
-# HHE seems to stop when it has found this many strings:
+config = find_config(ARGV[1])
+strings = find_strings(ARGV[0], config)
 
-NUM_STRINGS=785
+for string in strings
 
-set_config(ARGV[1])
+	string_text = string[1].to_json
 
-def read_block(file)
-        result = ""
-
-        4.times do
-                break if file.eof?
-                c = file.readchar
-                result += sprintf("%c", c)
-        end
-
-        end_index = result.index(0)
-        if end_index
-                result = result[0, end_index]
-        end
-
-        result
-end
-
-def bad_string?(str)
-	str.each_byte do |b|
-		if b >= 0x80
-			return true
-		end
+	if string[1] == nil
+		string_text = "NULL"
 	end
 
-	return false
+	printf "    { %6i, %s },\n", string[0], string_text
 end
 
-File.open(ARGV[0]) do |file|
-        current_str = ""
-
-	file.seek(STRINGS_OFFSET)
-
-	offset = 0
-	start_offset = nil
-	num_strings = 0
-
-        while num_strings < NUM_STRINGS
-                block = read_block(file)
-
-		if start_offset == nil
-			start_offset = offset
-		end
-
-		current_str += block
-
-		# End of string?
-
-		if block.length < 4
-
-			# Extended-ASCII characters cannot be output
-			# to JSON.
-
-			if bad_string?(current_str)
-				current_str = "XXXXX BAD STRING"
-			end
-
-			printf "    { %6i, %s },\n", start_offset,
-			                             current_str.to_json
-			current_str = ""
-			start_offset = nil
-			num_strings += 1
-		end
-
-		offset += 4
-        end
-end
-
-
-

hhe/v1.0/strings.c

@@ -551,7 +551,7 @@
     {   9972, "P_RemoveActivePlat: can't find plat!" },
     {  10012, "Too many mace spots." },
     {  10036, "" },
-    {  10040, "XXXXX BAD STRING" },
+    {  10040, NULL },
     {  10052, "P_GroupLines: miscounted" },
     {  10080, "-timer" },
     {  10088, "FLTWAWA1" },

hhe/v1.2/strings.c

@@ -581,7 +581,7 @@
     {  11900, "P_RemoveActivePlat: can't find plat!" },
     {  11940, "Too many mace spots." },
     {  11964, "" },
-    {  11968, "XXXXX BAD STRING" },
+    {  11968, NULL },
     {  11980, "P_GroupLines: miscounted" },
     {  12008, "-timer" },
     {  12016, "FLTWAWA1" },

hhe/v1.3/strings.c

@@ -581,7 +581,7 @@
     {  11900, "P_RemoveActivePlat: can't find plat!" },
     {  11940, "Too many mace spots." },
     {  11964, "" },
-    {  11968, "XXXXX BAD STRING" },
+    {  11968, NULL },
     {  11980, "P_GroupLines: miscounted" },
     {  12008, "-timer" },
     {  12016, "FLTWAWA1" },