From 7277f89a5f0be91f1639c5cec0e414d2be69cf03 Mon Sep 17 00:00:00 2001
From: Michael Tremer <michael.tremer@ipfire.org>
Date: Sat, 11 Jan 2025 14:05:56 +0000
Subject: [PATCH] parser: Compile regular expressions only once

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
---
 src/pakfire/parser.c | 87 +++++++++++++++++++-------------------------
 1 file changed, 37 insertions(+), 50 deletions(-)

diff --git a/src/pakfire/parser.c b/src/pakfire/parser.c
index 77c1491b0..b6a862f14 100644
--- a/src/pakfire/parser.c
+++ b/src/pakfire/parser.c
@@ -45,6 +45,12 @@
 // Enable to get more debugging output
 //#define PAKFIRE_DEBUG_PARSER
 
+// Regular expressions
+static struct pakfire_parser_regexes {
+	pcre2_code* variable;
+	pcre2_code* command;
+} regexes = {};
+
 struct pakfire_parser {
 	struct pakfire_ctx* ctx;
 	struct pakfire* pakfire;
@@ -56,34 +62,8 @@ struct pakfire_parser {
 
 	struct pakfire_parser_declaration** declarations;
 	size_t num_declarations;
-
-	// Regular expressions
-	pcre2_code* regex_command;
-	pcre2_code* regex_variable;
 };
 
-static int pakfire_parser_compile_regexes(struct pakfire_parser* parser) {
-	int r;
-
-	// Commands
-	if (!parser->regex_command && (parser->flags & PAKFIRE_PARSER_FLAGS_EXPAND_COMMANDS)) {
-		r = pakfire_compile_regex(parser->ctx, &parser->regex_command,
-			"%(\\(((?>[^()]|(?1))*)\\))");
-		if (r)
-			return r;
-	}
-
-	// Variables
-	if (!parser->regex_variable) {
-		r = pakfire_compile_regex(parser->ctx, &parser->regex_variable,
-			"%\\{([A-Za-z0-9_\\-]+)\\}");
-		if (r)
-			return r;
-	}
-
-	return 0;
-}
-
 static void pakfire_parser_free_declaration(struct pakfire_parser_declaration* d) {
 	if (d->value)
 		free(d->value);
@@ -101,12 +81,6 @@ static void pakfire_parser_free_declarations(struct pakfire_parser* parser) {
 }
 
 static void pakfire_parser_free(struct pakfire_parser* parser) {
-	// Release regular expressions
-	if (parser->regex_variable)
-		pcre2_code_free(parser->regex_variable);
-	if (parser->regex_command)
-		pcre2_code_free(parser->regex_command);
-
 	pakfire_parser_free_declarations(parser);
 
 	if (parser->namespace)
@@ -503,26 +477,35 @@ int pakfire_parser_append(struct pakfire_parser* parser,
 }
 
 static int pakfire_parser_expand_commands(struct pakfire_parser* parser, char** buffer) {
-	int r = 0;
+	pcre2_match_data* match = NULL;
 	PCRE2_UCHAR* command = NULL;
 	PCRE2_SIZE command_length;
 	PCRE2_UCHAR* pattern = NULL;
 	PCRE2_SIZE pattern_length;
+	int r = 0;
 
 	DEBUG(parser->ctx, "Searching for commands in:\n%s\n", *buffer);
 
+	// Compile the regular expression
+	if (!regexes.command) {
+		r = pakfire_compile_regex(parser->ctx, &regexes.command, "%(\\(((?>[^()]|(?1))*)\\))");
+		if (r < 0) {
+			ERROR(parser->ctx, "Failed to compile the command regex: %s\n", strerror(-r));
+			goto ERROR;
+		}
+	}
+
 	// Allocate memory for results
-	pcre2_match_data* match = pcre2_match_data_create_from_pattern(
-		parser->regex_command, NULL);
+	match = pcre2_match_data_create_from_pattern(regexes.command, NULL);
 
 	// Arguments passed to pakfire_execute
 	const char* argv[4] = {
 		"/bin/sh", "-c", NULL /* will be replaced by command later */, NULL,
 	};
 
-	while (1) {
+	for (;;) {
 		// Perform matching
-		r = pcre2_jit_match(parser->regex_command,
+		r = pcre2_jit_match(regexes.command,
 			(PCRE2_UCHAR*)*buffer, strlen(*buffer), 0, 0, match, NULL);
 
 		// End loop when we have expanded all variables
@@ -593,8 +576,8 @@ static int pakfire_parser_expand_commands(struct pakfire_parser* parser, char**
 	}
 
 ERROR:
-	pcre2_match_data_free(match);
-
+	if (match)
+		pcre2_match_data_free(match);
 	if (command)
 		pcre2_substring_free(command);
 	if (pattern)
@@ -605,6 +588,7 @@ ERROR:
 
 static int pakfire_parser_expand_variables(struct pakfire_parser* parser,
 		const char* namespace, char** buffer) {
+	pcre2_match_data* match = NULL;
 	PCRE2_UCHAR* variable = NULL;
 	PCRE2_SIZE variable_length = 0;
 	PCRE2_UCHAR* pattern = NULL;
@@ -619,14 +603,22 @@ static int pakfire_parser_expand_variables(struct pakfire_parser* parser,
 	else if (!*buffer)
 		return 0;
 
+	// Compile the regular expression
+	if (!regexes.variable) {
+		r = pakfire_compile_regex(parser->ctx, &regexes.variable, "%\\{([A-Za-z0-9_\\-]+)\\}");
+		if (r < 0) {
+			ERROR(parser->ctx, "Failed to compile the variable regex: %s\n", strerror(-r));
+			goto ERROR;
+		}
+	}
+
 	// Allocate memory for results
-	pcre2_match_data* match = pcre2_match_data_create_from_pattern(
-		parser->regex_variable, NULL);
+	match = pcre2_match_data_create_from_pattern(regexes.variable, NULL);
 
 	// Search for any variables
-	while (1) {
+	for (;;) {
 		// Perform matching
-		r = pcre2_jit_match(parser->regex_variable,
+		r = pcre2_jit_match(regexes.variable,
 			(PCRE2_UCHAR*)*buffer, strlen(*buffer), 0, 0, match, NULL);
 
 		// End loop when we have expanded all variables
@@ -711,6 +703,8 @@ ERROR:
 
 char* pakfire_parser_expand(struct pakfire_parser* parser,
 		const char* namespace, const char* value) {
+	int r;
+
 	// Return NULL when the value is NULL
 	if (!value)
 		return NULL;
@@ -723,13 +717,6 @@ char* pakfire_parser_expand(struct pakfire_parser* parser,
 	if (!pos)
 		return buffer;
 
-	// Compile all regular expressions
-	int r = pakfire_parser_compile_regexes(parser);
-	if (r) {
-		DEBUG(parser->ctx, "Could not compile regular expressions: %m\n");
-		goto ERROR;
-	}
-
 	// Expand all variables
 	r = pakfire_parser_expand_variables(parser, namespace, &buffer);
 	if (r) {
-- 
2.47.3