set things up for UTF-8 PCRE2

6 files changed, 34 insertions, 31 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 238b113..74816ce 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -27,10 +27,10 @@ if(MSVC)
 	set(CMAKE_C_FLAGS_DEBUG "/WX /Zi /Ob0 /Od /RTC1")
 	set(SDL2_LIB_DIR ${CMAKE_SOURCE_DIR}/SDL2/lib/x64)
 	target_link_libraries(ted ${SDL2_LIB_DIR}/SDL2.lib)
-	target_link_libraries(ted ${CMAKE_SOURCE_DIR}/pcre2-32-static.lib)
+	target_link_libraries(ted ${CMAKE_SOURCE_DIR}/pcre2-32-static.lib ${CMAKE_SOURCE_DIR}/pcre2-8-static.lib)
 else()
 	# NOTE: -gdwarf-4 is needed for valgrind to work
 	set(CMAKE_C_FLAGS "-Wall -Wextra -Wshadow -Wconversion -Wpedantic -pedantic -std=gnu11 -gdwarf-4 -Wno-unused-function -Wno-fixed-enum-extension -Wimplicit-fallthrough -Wno-format-truncation -Wno-unknown-warning-option")
 	target_link_libraries(ted m SDL2)
-	target_link_libraries(ted ${CMAKE_SOURCE_DIR}/libpcre2-32.a)
+	target_link_libraries(ted ${CMAKE_SOURCE_DIR}/libpcre2-32.a ${CMAKE_SOURCE_DIR}/libpcre2-8.a)
 endif()
diff --git a/Makefile b/Makefile
index 651384d..50861cd 100644
--- a/Makefile
+++ b/Makefile
@@ -1,7 +1,7 @@
 ALL_CFLAGS=$(CFLAGS) -Wall -Wextra -Wshadow -Wconversion -Wpedantic -pedantic -std=gnu11 \
 	-Wno-unused-function -Wno-fixed-enum-extension -Wimplicit-fallthrough -Wno-format-truncation -Wno-unknown-warning-option \
 	-Ipcre2
-LIBS=-lSDL2 -lGL -lm libpcre2-32.a
+LIBS=-lSDL2 -lGL -lm libpcre2-32.a libpcre2-8.a
 RELEASE_CFLAGS=$(ALL_CFLAGS) -O3
 PROFILE_CFLAGS=$(ALL_CFLAGS) -O3 -g -DPROFILE=1
 # if you change the directories below, ted won't work.
@@ -17,15 +17,15 @@ ted: debug/ted
 compile_commands.json: debug/ted
 	rm -f compile_commands.json
 	cp debug/compile_commands.json .
-debug/ted: *.[ch] libpcre2-32.a CMakeLists.txt
+debug/ted: *.[ch] pcre-lib CMakeLists.txt
 	mkdir -p debug
 	cd debug && cmake -DCMAKE_EXPORT_COMPILE_COMMANDS=1 -DCMAKE_BUILD_TYPE=Debug -GNinja ..
 	ninja -C debug
-release: *.[ch] libpcre2-32.a
+release: *.[ch] pcre-lib
 	$(CC) main.c -o ted $(RELEASE_CFLAGS) $(LIBS)
-release_debug: *.[ch] libpcre2-32.a
+release_debug: *.[ch] pcre-lib
 	$(CC) main.c -g -o ted $(RELEASE_CFLAGS) $(LIBS)
-profile: *.[ch] libpcre2-32.a
+profile: *.[ch] pcre-lib
 	$(CC) main.c -o ted $(PROFILE_CFLAGS) $(LIBS)
 clean:
 	rm -f ted *.o *.a
@@ -39,9 +39,11 @@ install: release
 	cp -r themes $(GLOBAL_DATA_DIR)
 	install -m 644 ted.cfg $(GLOBAL_DATA_DIR)
 	install ted $(INSTALL_BIN_DIR)
-libpcre2-32.a: pcre2
-	cd pcre2 && cmake -DPCRE2_BUILD_PCRE2_32=ON . && $(MAKE) -j8
-	cp pcre2/libpcre2-32.a ./
+pcre-lib:
+	@if [ '(' '!' -f libpcre2-32.a ')' -o '(' '!' -f libpcre2-8.a ')' ]; then \
+		cd pcre2 && cmake -DPCRE2_BUILD_PCRE2_32=ON . && $(MAKE) -j8 && \
+		cp libpcre2-32.a libpcre2-8.a ../ ; \
+	fi
 keywords.h: keywords.py
 	python3 keywords.py
 ted.deb: release
diff --git a/find.c b/find.c
index b5fee08..992f508 100644
--- a/find.c
+++ b/find.c
@@ -31,7 +31,7 @@ TextBuffer *find_search_buffer(Ted *ted) {
 
 static void ted_error_from_pcre2_error(Ted *ted, int err) {
 	char32_t buf[256] = {0};
-	size_t len = (size_t)pcre2_get_error_message(err, buf, arr_count(buf) - 1);
+	size_t len = (size_t)pcre2_get_error_message_32(err, buf, arr_count(buf) - 1);
 	char *error_cstr = str32_to_utf8_cstr(str32(buf, len));
 	if (error_cstr) {
 		ted_error(ted, "Search error: %s.", error_cstr);
@@ -43,11 +43,11 @@ static bool find_compile_pattern(Ted *ted) {
 	TextBuffer *find_buffer = ted->find_buffer;
 	String32 term = buffer_get_line(find_buffer, 0);
 	if (term.len) {
-		pcre2_match_data *match_data = pcre2_match_data_create(FIND_MAX_GROUPS, NULL);
+		pcre2_match_data_32 *match_data = pcre2_match_data_create_32(FIND_MAX_GROUPS, NULL);
 		if (match_data) {
 			int error = 0;
 			PCRE2_SIZE error_pos = 0;
-			pcre2_code *code = pcre2_compile(term.str, term.len, find_compilation_flags(ted), &error, &error_pos, NULL);
+			pcre2_code_32 *code = pcre2_compile_32(term.str, term.len, find_compilation_flags(ted), &error, &error_pos, NULL);
 			if (code) {
 				ted->find_code = code;
 				ted->find_match_data = match_data;
@@ -56,7 +56,7 @@ static bool find_compile_pattern(Ted *ted) {
 			} else {
 				ted->find_invalid_pattern = true;
 			}
-			pcre2_match_data_free(match_data);
+			pcre2_match_data_free_32(match_data);
 		} else {
 			ted_error(ted, "Out of memory.");
 		}
@@ -68,11 +68,11 @@ static bool find_compile_pattern(Ted *ted) {
 
 static void find_free_pattern(Ted *ted) {
 	if (ted->find_code) {
-		pcre2_code_free(ted->find_code);
+		pcre2_code_free_32(ted->find_code);
 		ted->find_code = NULL;
 	}
 	if (ted->find_match_data) {
-		pcre2_match_data_free(ted->find_match_data);
+		pcre2_match_data_free_32(ted->find_match_data);
 		ted->find_match_data = NULL;
 	}
 	arr_clear(ted->find_results);
@@ -97,19 +97,19 @@ static WarnUnusedResult bool find_match(Ted *ted, BufferPos *pos, u32 *match_sta
 	TextBuffer *buffer = find_search_buffer(ted);
 	if (!buffer) return false;
 	String32 str = buffer_get_line(buffer, pos->line);
-	PCRE2_SIZE *groups = pcre2_get_ovector_pointer(ted->find_match_data);
+	PCRE2_SIZE *groups = pcre2_get_ovector_pointer_32(ted->find_match_data);
 
 	u32 match_flags = PCRE2_NOTEMPTY;
 	
 	int ret;
 	if (direction == +1)
-		ret = pcre2_match(ted->find_code, str.str, str.len, pos->index, match_flags, ted->find_match_data, NULL);
+		ret = pcre2_match_32(ted->find_code, str.str, str.len, pos->index, match_flags, ted->find_match_data, NULL);
 	else {
 		// unfortunately PCRE does not have a backwards option, so we need to do the search multiple times
 		u32 last_pos = 0;
 		ret = -1;
 		while (1) {
-			int next_ret = pcre2_match(ted->find_code, str.str, pos->index, last_pos, match_flags, ted->find_match_data, NULL);
+			int next_ret = pcre2_match_32(ted->find_code, str.str, pos->index, last_pos, match_flags, ted->find_match_data, NULL);
 			if (next_ret > 0) {
 				ret = next_ret;
 				last_pos = (u32)groups[1];
@@ -243,14 +243,14 @@ static bool find_replace_match(Ted *ted, u32 match_idx) {
 	char32_t *str = line.str + match.start.index;
 	u32 len = match.end.index - match.start.index;
 	
-	int ret = pcre2_substitute(ted->find_code, str, len, 0,
+	int ret = pcre2_substitute_32(ted->find_code, str, len, 0,
 		PCRE2_SUBSTITUTE_OVERFLOW_LENGTH|flags, ted->find_match_data, NULL, replacement.str,
 		replacement.len, NULL, &output_size);
 	char32_t *output_buffer = output_size
 		? calloc(output_size, sizeof *output_buffer)
 		: NULL;
 	if (output_buffer || !output_size) {
-		ret = pcre2_substitute(ted->find_code, str, len, 0,
+		ret = pcre2_substitute_32(ted->find_code, str, len, 0,
 			flags, ted->find_match_data, NULL, replacement.str,
 			replacement.len, output_buffer, &output_size);
 		if (ret > 0) {
diff --git a/make.bat b/make.bat
index 30de8a4..16dc9db 100644
--- a/make.bat
+++ b/make.bat
@@ -3,14 +3,15 @@ if _%VCVARS% == _ (
 	set VCVARS=1
 	call vcvarsall x64
 )
-if not exist pcre2-32-static.lib (
+if not exist pcre2-8-static.lib (
 	pushd pcre2
 	cmake -D PCRE2_BUILD_PCRE2_8=OFF -D PCRE2_BUILD_TESTS=OFF -D PCRE2_BUILD_PCRE2_32=ON -D CMAKE_BUILD_TYPE=Release -D CMAKE_GENERATOR_PLATFORM=x64 -D PCRE2_STATIC=ON .
 	cmake --build . --config Release
 	popd
-	copy pcre2\Release\pcre2-32-static.lib
+	copy /y pcre2\Release\pcre2-32-static.lib
+	copy /y pcre2\Release\pcre2-8-static.lib
 )
-SET C_FLAGS=/nologo /W4 /MD /wd4200 /wd4204 /wd4221 /wd4706 /wd4214 /D_CRT_SECURE_NO_WARNINGS /I SDL2/include /I pcre2 User32.lib SDL2/lib/x64/SDL2main.lib SDL2/lib/x64/SDL2.lib pcre2-32-static.lib
+SET C_FLAGS=/nologo /W4 /MD /wd4200 /wd4204 /wd4221 /wd4706 /wd4214 /D_CRT_SECURE_NO_WARNINGS /I SDL2/include /I pcre2 User32.lib SDL2/lib/x64/SDL2main.lib SDL2/lib/x64/SDL2.lib pcre2-8-static.lib pcre2-32-static.lib
 rc /nologo ted.rc 
 if _%1 == _ (
 	if not exist debug mkdir debug
diff --git a/pcre-inc.h b/pcre-inc.h
index 3d8f3b9..6c36f4a 100644
--- a/pcre-inc.h
+++ b/pcre-inc.h
@@ -5,7 +5,7 @@
 #define PCRE_INC_H_
 
 #define PCRE2_STATIC
-#define PCRE2_CODE_UNIT_WIDTH 32
+#define PCRE2_CODE_UNIT_WIDTH 0
 #include <pcre2.h>
 
 #endif // PCRE_INC_H_
diff --git a/tags.c b/tags.c
index 7d8086b..68a3f2c 100644
--- a/tags.c
+++ b/tags.c
@@ -324,18 +324,18 @@ top:;
 								if (end_anchored) options |= PCRE2_ENDANCHORED;
 								int error_code;
 								PCRE2_SIZE error_offset;
-								pcre2_code *code = pcre2_compile(pattern32.str, pattern32.len,
+								pcre2_code_32 *code = pcre2_compile_32(pattern32.str, pattern32.len,
 									options, &error_code, &error_offset, NULL);
 								if (code) {
-									pcre2_match_data *match_data = pcre2_match_data_create(10, NULL);
+									pcre2_match_data_32 *match_data = pcre2_match_data_create_32(10, NULL);
 									if (match_data) {
 										for (u32 line_idx = 0, line_count = buffer_line_count(buffer); line_idx < line_count; ++line_idx) {
 											String32 line = buffer_get_line(buffer, line_idx);
-											int n = pcre2_match(code, line.str, line.len, 0, PCRE2_NOTEMPTY,
+											int n = pcre2_match_32(code, line.str, line.len, 0, PCRE2_NOTEMPTY,
 												match_data, NULL);
 											if (n == 1) {
 												// found it!
-												PCRE2_SIZE *ovector = pcre2_get_ovector_pointer(match_data);
+												PCRE2_SIZE *ovector = pcre2_get_ovector_pointer_32(match_data);
 												PCRE2_SIZE index = ovector[0];
 												BufferPos pos = {line_idx, (u32)index};
 												buffer_cursor_move_to_pos(buffer, pos);
@@ -344,9 +344,9 @@ top:;
 												break;
 											}
 										}
-										pcre2_match_data_free(match_data);
+										pcre2_match_data_free_32(match_data);
 									}
-									pcre2_code_free(code);
+									pcre2_code_free_32(code);
 								}
 								str32_free(&pattern32);
 								free(pattern);