From: Michael Tremer <michael.tremer@ipfire.org>
Date: Sat, 4 Jan 2025 14:14:02 +0000 (+0000)
Subject: filelist: Implement binary search for paths
X-Git-Tag: 0.9.30~540
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=73f23b5f0b22109bea1523c235125d6517d1bb34;p=pakfire.git

filelist: Implement binary search for paths

This is a fast path if we are looking for an exact match which will be a
lot faster than the linear pattern matching search.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
---

diff --git a/src/pakfire/filelist.c b/src/pakfire/filelist.c
index 552023fe5..dfc2a12c2 100644
--- a/src/pakfire/filelist.c
+++ b/src/pakfire/filelist.c
@@ -431,10 +431,47 @@ ERROR:
 	return r;
 }
 
-int pakfire_filelist_contains(struct pakfire_filelist* list, const char* pattern) {
-	if (!pattern)
-		return -EINVAL;
+static int pakfire_filelist_contains_path(
+		struct pakfire_filelist* list, const char* path) {
+	const char* p = NULL;
+	int i;
+	int r;
+
+	// Set starting points
+	int lo = 0;
+	int hi = list->num_files - 1;
 
+	while (lo <= hi) {
+		// Find the middle
+		i = lo + (hi - lo) / 2;
+
+		// Fetch the package path
+		p = pakfire_file_get_path(list->files[i]);
+		if (!p)
+			return -EINVAL;
+
+		// Compare the paths
+		r = strcmp(p, path);
+
+		// We found a match
+		if (r == 0)
+			return 1;
+
+		// Search the right half
+		else if (r < 0)
+			lo = i + 1;
+
+		// Search the left half
+		else if (r > 0)
+			hi = i - 1;
+	}
+
+	// No match
+	return 0;
+}
+
+static int pakfire_filelist_contains_pattern(
+		struct pakfire_filelist* list, const char* pattern) {
 	for (unsigned int i = 0; i < list->num_files; i++) {
 		const char* path = pakfire_file_get_path(list->files[i]);
 		if (!path)
@@ -447,6 +484,18 @@ int pakfire_filelist_contains(struct pakfire_filelist* list, const char* pattern
 	return 0;
 }
 
+int pakfire_filelist_contains(struct pakfire_filelist* list, const char* pattern) {
+	if (!pattern)
+		return -EINVAL;
+
+	// If this is a pattern, we perform a naive search
+	if (pakfire_path_is_pattern(pattern))
+		return pakfire_filelist_contains_pattern(list, pattern);
+
+	// Otherwise we perform a binary search to be faster
+	return pakfire_filelist_contains_path(list, pattern);
+}
+
 int pakfire_filelist_walk(struct pakfire_filelist* list,
 		pakfire_filelist_walk_callback callback, void* data, int flags, const char* title) {
 	struct pakfire_progress* progress = NULL;
diff --git a/src/pakfire/path.c b/src/pakfire/path.c
index d0123e65c..f90017a85 100644
--- a/src/pakfire/path.c
+++ b/src/pakfire/path.c
@@ -640,6 +640,25 @@ int pakfire_path_match(const char* p, const char* s) {
 	return 1;
 }
 
+// Returns true if the path contains patterns (i.e. ? and *)
+int pakfire_path_is_pattern(const char* path) {
+	if (!path)
+		return -EINVAL;
+
+	for (const char* p = path; *p; p++) {
+		switch (*p) {
+			case '*':
+			case '?':
+				return 1;
+
+			default:
+				break;
+		}
+	}
+
+	return 0;
+}
+
 static int pakfire_path_strip_extension(char* path) {
 	if (!path)
 		return -EINVAL;
diff --git a/src/pakfire/path.h b/src/pakfire/path.h
index 906824f81..b58e871a7 100644
--- a/src/pakfire/path.h
+++ b/src/pakfire/path.h
@@ -55,6 +55,7 @@ int pakfire_path_is_absolute(const char* s);
 int __pakfire_path_absolute(char* buffer, const size_t length, const char* s);
 
 int pakfire_path_match(const char* p, const char* s);
+int pakfire_path_is_pattern(const char* path);
 
 #define pakfire_path_replace_extension(path, extension) \
 	__pakfire_path_replace_extension(path, sizeof(path), extension)