Merge pull request libgit2#5226 from pks-t/pks/regexp-api

regexp: implement a new regular expression API

Author: ethomson

cmake/Modules/FindPCRE2.cmake

@@ -20,15 +20,14 @@ FIND_PATH(PCRE2_INCLUDE_DIR NAMES pcre2posix.h)
 
 # Look for the library.
 FIND_LIBRARY(PCRE2_LIBRARY NAMES pcre2-8)
-FIND_LIBRARY(PCRE2_POSIX_LIBRARY NAMES pcre2-posix)
 
 # Handle the QUIETLY and REQUIRED arguments and set PCRE2_FOUND to TRUE if all listed variables are TRUE.
 INCLUDE(FindPackageHandleStandardArgs)
-FIND_PACKAGE_HANDLE_STANDARD_ARGS(PCRE2 DEFAULT_MSG PCRE2_LIBRARY PCRE2_POSIX_LIBRARY PCRE2_INCLUDE_DIR)
+FIND_PACKAGE_HANDLE_STANDARD_ARGS(PCRE2 DEFAULT_MSG PCRE2_LIBRARY PCRE2_INCLUDE_DIR)
 
 # Copy the results to the output variables.
 IF(PCRE2_FOUND)
-	SET(PCRE2_LIBRARIES ${PCRE2_LIBRARY} ${PCRE2_POSIX_LIBRARY})
+	SET(PCRE2_LIBRARIES ${PCRE2_LIBRARY})
 	SET(PCRE2_INCLUDE_DIRS ${PCRE2_INCLUDE_DIR})
 ELSE(PCRE2_FOUND)
 	SET(PCRE2_LIBRARIES)

src/common.h

@@ -88,7 +88,6 @@
 #include "git2/deprecated.h"
 
 #include "posix.h"
-#include "posix_regex.h"
 
 #define DEFAULT_BUFSIZE 65536
 #define FILEIO_BUFSIZE DEFAULT_BUFSIZE

src/config.c

@@ -7,13 +7,15 @@
 
 #include "config.h"
 
-#include "sysdir.h"
 #include "git2/config.h"
 #include "git2/sys/config.h"
-#include "vector.h"
+
 #include "buf_text.h"
 #include "config_backend.h"
+#include "regexp.h"
+#include "sysdir.h"
 #include "transaction.h"
+#include "vector.h"
 #if GIT_WIN32
 # include <windows.h>
 #endif
@@ -345,7 +347,7 @@ typedef struct {
 	git_config_iterator parent;
 	git_config_iterator *current;
 	const git_config *cfg;
-	p_regex_t regex;
+	git_regexp regex;
 	size_t i;
 } all_iter;
 
@@ -423,7 +425,7 @@ static int all_iter_glob_next(git_config_entry **entry, git_config_iterator *_it
 	 */
 	while ((error = all_iter_next(entry, _iter)) == 0) {
 		/* skip non-matching keys if regexp was provided */
-		if (p_regexec(&iter->regex, (*entry)->name, 0, NULL, 0) != 0)
+		if (git_regexp_match(&iter->regex, (*entry)->name) != 0)
 			continue;
 
 		/* and simply return if we like the entry's name */
@@ -447,7 +449,7 @@ static void all_iter_glob_free(git_config_iterator *_iter)
 {
 	all_iter *iter = (all_iter *) _iter;
 
-	p_regfree(&iter->regex);
+	git_regexp_dispose(&iter->regex);
 	all_iter_free(_iter);
 }
 
@@ -480,8 +482,7 @@ int git_config_iterator_glob_new(git_config_iterator **out, const git_config *cf
 	iter = git__calloc(1, sizeof(all_iter));
 	GIT_ERROR_CHECK_ALLOC(iter);
 
-	if ((result = p_regcomp(&iter->regex, regexp, P_REG_EXTENDED)) != 0) {
-		git_error_set_regex(&iter->regex, result);
+	if ((result = git_regexp_compile(&iter->regex, regexp, 0)) < 0) {
 		git__free(iter);
 		return -1;
 	}
@@ -510,18 +511,13 @@ int git_config_backend_foreach_match(
 {
 	git_config_entry *entry;
 	git_config_iterator* iter;
-	p_regex_t regex;
+	git_regexp regex;
 	int error = 0;
 
 	assert(backend && cb);
 
-	if (regexp != NULL) {
-		if ((error = p_regcomp(&regex, regexp, P_REG_EXTENDED)) != 0) {
-			git_error_set_regex(&regex, error);
-			p_regfree(&regex);
-			return -1;
-		}
-	}
+	if (regexp && git_regexp_compile(&regex, regexp, 0) < 0)
+		return -1;
 
 	if ((error = backend->iterator(&iter, backend)) < 0) {
 		iter = NULL;
@@ -530,7 +526,7 @@ int git_config_backend_foreach_match(
 
 	while (!(iter->next(&entry, iter) < 0)) {
 		/* skip non-matching keys if regexp was provided */
-		if (regexp && p_regexec(&regex, entry->name, 0, NULL, 0) != 0)
+		if (regexp && git_regexp_match(&regex, entry->name) != 0)
 			continue;
 
 		/* abort iterator on non-zero return value */
@@ -541,7 +537,7 @@ int git_config_backend_foreach_match(
 	}
 
 	if (regexp != NULL)
-		p_regfree(&regex);
+		git_regexp_dispose(&regex);
 
 	iter->free(iter);
 
@@ -981,7 +977,7 @@ typedef struct {
 	git_config_iterator parent;
 	git_config_iterator *iter;
 	char *name;
-	p_regex_t regex;
+	git_regexp regex;
 	int have_regex;
 } multivar_iter;
 
@@ -997,7 +993,7 @@ static int multivar_iter_next(git_config_entry **entry, git_config_iterator *_it
 		if (!iter->have_regex)
 			return 0;
 
-		if (p_regexec(&iter->regex, (*entry)->value, 0, NULL, 0) == 0)
+		if (git_regexp_match(&iter->regex, (*entry)->value) == 0)
 			return 0;
 	}
 
@@ -1012,7 +1008,7 @@ void multivar_iter_free(git_config_iterator *_iter)
 
 	git__free(iter->name);
 	if (iter->have_regex)
-		p_regfree(&iter->regex);
+		git_regexp_dispose(&iter->regex);
 	git__free(iter);
 }
 
@@ -1032,13 +1028,8 @@ int git_config_multivar_iterator_new(git_config_iterator **out, const git_config
 		goto on_error;
 
 	if (regexp != NULL) {
-		error = p_regcomp(&iter->regex, regexp, P_REG_EXTENDED);
-		if (error != 0) {
-			git_error_set_regex(&iter->regex, error);
-			error = -1;
-			p_regfree(&iter->regex);
+		if ((error = git_regexp_compile(&iter->regex, regexp, 0)) < 0)
 			goto on_error;
-		}
 
 		iter->have_regex = 1;
 	}

src/config_file.c

@@ -18,6 +18,7 @@
 #include "config_entries.h"
 #include "config_parse.h"
 #include "filebuf.h"
+#include "regexp.h"
 #include "strmap.h"
 #include "sysdir.h"
 #include "wildmatch.h"
@@ -61,7 +62,7 @@ typedef struct {
 
 static int config_read(git_config_entries *entries, const git_repository *repo, config_file *file, git_config_level_t level, int depth);
 static int config_read_buffer(git_config_entries *entries, const git_repository *repo, config_file *file, git_config_level_t level, int depth, const char *buf, size_t buflen);
-static int config_write(config_file_backend *cfg, const char *orig_key, const char *key, const p_regex_t *preg, const char *value);
+static int config_write(config_file_backend *cfg, const char *orig_key, const char *key, const git_regexp *preg, const char *value);
 static char *escape_value(const char *ptr);
 
 /**
@@ -350,29 +351,25 @@ static int config_set_multivar(
 	git_config_backend *cfg, const char *name, const char *regexp, const char *value)
 {
 	config_file_backend *b = GIT_CONTAINER_OF(cfg, config_file_backend, parent);
-	char *key;
-	p_regex_t preg;
+	git_regexp preg;
 	int result;
+	char *key;
 
 	assert(regexp);
 
 	if ((result = git_config__normalize_name(name, &key)) < 0)
 		return result;
 
-	result = p_regcomp(&preg, regexp, P_REG_EXTENDED);
-	if (result != 0) {
-		git_error_set_regex(&preg, result);
-		result = -1;
+	if ((result = git_regexp_compile(&preg, regexp, 0)) < 0)
 		goto out;
-	}
 
 	/* If we do have it, set call config_write() and reload */
 	if ((result = config_write(b, name, key, &preg, value)) < 0)
 		goto out;
 
 out:
 	git__free(key);
-	p_regfree(&preg);
+	git_regexp_dispose(&preg);
 
 	return result;
 }
@@ -412,7 +409,7 @@ static int config_delete_multivar(git_config_backend *cfg, const char *name, con
 	config_file_backend *b = GIT_CONTAINER_OF(cfg, config_file_backend, parent);
 	git_config_entries *entries = NULL;
 	git_config_entry *entry = NULL;
-	p_regex_t preg = { 0 };
+	git_regexp preg = GIT_REGEX_INIT;
 	char *key = NULL;
 	int result;
 
@@ -430,19 +427,16 @@ static int config_delete_multivar(git_config_backend *cfg, const char *name, con
 		goto out;
 	}
 
-	if ((result = p_regcomp(&preg, regexp, P_REG_EXTENDED)) != 0) {
-		git_error_set_regex(&preg, result);
-		result = -1;
+	if ((result = git_regexp_compile(&preg, regexp, 0)) < 0)
 		goto out;
-	}
 
 	if ((result = config_write(b, name, key, &preg, NULL)) < 0)
 		goto out;
 
 out:
 	git_config_entries_free(entries);
 	git__free(key);
-	p_regfree(&preg);
+	git_regexp_dispose(&preg);
 	return result;
 }
 
@@ -928,7 +922,7 @@ struct write_data {
 	const char *section;
 	const char *orig_name;
 	const char *name;
-	const p_regex_t *preg;
+	const git_regexp *preg;
 	const char *value;
 };
 
@@ -1033,7 +1027,7 @@ static int write_on_variable(
 
 	/* If we have a regex to match the value, see if it matches */
 	if (has_matched && write_data->preg != NULL)
-		has_matched = (p_regexec(write_data->preg, var_value, 0, NULL, 0) == 0);
+		has_matched = (git_regexp_match(write_data->preg, var_value) == 0);
 
 	/* If this isn't the name/value we're looking for, simply dump the
 	 * existing data back out and continue on.
@@ -1094,7 +1088,8 @@ static int write_on_eof(
 /*
  * This is pretty much the parsing, except we write out anything we don't have
  */
-static int config_write(config_file_backend *cfg, const char *orig_key, const char *key, const p_regex_t *preg, const char* value)
+static int config_write(config_file_backend *cfg, const char *orig_key, const char *key, const git_regexp *preg, const char* value)
+
 {
 	char *orig_section = NULL, *section = NULL, *orig_name, *name, *ldot;
 	git_buf buf = GIT_BUF_INIT, contents = GIT_BUF_INIT;

src/diff_driver.c

@@ -15,6 +15,7 @@
 #include "map.h"
 #include "buf_text.h"
 #include "config.h"
+#include "regexp.h"
 #include "repository.h"
 
 typedef enum {
@@ -25,7 +26,7 @@ typedef enum {
 } git_diff_driver_t;
 
 typedef struct {
-	p_regex_t re;
+	git_regexp re;
 	int flags;
 } git_diff_driver_pattern;
 
@@ -39,7 +40,7 @@ struct git_diff_driver {
 	uint32_t binary_flags;
 	uint32_t other_flags;
 	git_array_t(git_diff_driver_pattern) fn_patterns;
-	p_regex_t  word_pattern;
+	git_regexp  word_pattern;
 	char name[GIT_FLEX_ARRAY];
 };
 
@@ -113,7 +114,7 @@ static int diff_driver_add_patterns(
 		if (error < 0)
 			break;
 
-		if ((error = p_regcomp(&pat->re, buf.ptr, regex_flags)) != 0) {
+		if ((error = git_regexp_compile(&pat->re, buf.ptr, regex_flags)) != 0) {
 			/*
 			 * TODO: issue a warning
 			 */
@@ -130,7 +131,7 @@ static int diff_driver_add_patterns(
 
 static int diff_driver_xfuncname(const git_config_entry *entry, void *payload)
 {
-	return diff_driver_add_patterns(payload, entry->value, P_REG_EXTENDED);
+	return diff_driver_add_patterns(payload, entry->value, 0);
 }
 
 static int diff_driver_funcname(const git_config_entry *entry, void *payload)
@@ -205,16 +206,12 @@ static int git_diff_driver_builtin(
 
 	if (ddef->fns &&
 		(error = diff_driver_add_patterns(
-			drv, ddef->fns, ddef->flags | P_REG_EXTENDED)) < 0)
+			drv, ddef->fns, ddef->flags)) < 0)
 		goto done;
 
 	if (ddef->words &&
-		(error = p_regcomp(
-			&drv->word_pattern, ddef->words, ddef->flags | P_REG_EXTENDED)))
-	{
-		error = git_error_set_regex(&drv->word_pattern, error);
+	    (error = git_regexp_compile(&drv->word_pattern, ddef->words, ddef->flags)) < 0)
 		goto done;
-	}
 
 	if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
 		goto done;
@@ -316,11 +313,10 @@ static int git_diff_driver_load(
 		goto done;
 	if (!ce || !ce->value)
 		/* no diff.<driver>.wordregex, so just continue */;
-	else if (!(error = p_regcomp(&drv->word_pattern, ce->value, P_REG_EXTENDED)))
+	else if (!(error = git_regexp_compile(&drv->word_pattern, ce->value, 0)))
 		found_driver = true;
 	else {
 		/* TODO: warn about bad regex instead of failure */
-		error = git_error_set_regex(&drv->word_pattern, error);
 		goto done;
 	}
 
@@ -400,10 +396,10 @@ void git_diff_driver_free(git_diff_driver *driver)
 		return;
 
 	for (i = 0; i < git_array_size(driver->fn_patterns); ++i)
-		p_regfree(& git_array_get(driver->fn_patterns, i)->re);
+		git_regexp_dispose(& git_array_get(driver->fn_patterns, i)->re);
 	git_array_clear(driver->fn_patterns);
 
-	p_regfree(&driver->word_pattern);
+	git_regexp_dispose(&driver->word_pattern);
 
 	git__free(driver);
 }
@@ -451,19 +447,19 @@ static int diff_context_line__pattern_match(
 	git_diff_driver *driver, git_buf *line)
 {
 	size_t i, maxi = git_array_size(driver->fn_patterns);
-	p_regmatch_t pmatch[2];
+	git_regmatch pmatch[2];
 
 	for (i = 0; i < maxi; ++i) {
 		git_diff_driver_pattern *pat = git_array_get(driver->fn_patterns, i);
 
-		if (!p_regexec(&pat->re, line->ptr, 2, pmatch, 0)) {
+		if (!git_regexp_search(&pat->re, line->ptr, 2, pmatch)) {
 			if (pat->flags & REG_NEGATE)
 				return false;
 
 			/* use pmatch data to trim line data */
-			i = (pmatch[1].rm_so >= 0) ? 1 : 0;
-			git_buf_consume(line, git_buf_cstr(line) + pmatch[i].rm_so);
-			git_buf_truncate(line, pmatch[i].rm_eo - pmatch[i].rm_so);
+			i = (pmatch[1].start >= 0) ? 1 : 0;
+			git_buf_consume(line, git_buf_cstr(line) + pmatch[i].start);
+			git_buf_truncate(line, pmatch[i].end - pmatch[i].start);
 			git_buf_rtrim(line);
 
 			return true;