X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=plocate.cpp;h=c2bb17ca1452e5b63468c7e45ae5a305db1dc059;hb=97c57726bc81c979386c623ad7586c2677fdd865;hp=4141a316bb91a5381bfd01a28163f58341e26fd9;hpb=601e6fef22bc018be1ca3b19215e54a2c2edcb9d;p=plocate

diff --git a/plocate.cpp b/plocate.cpp
index 4141a31..c2bb17c 100644
--- a/plocate.cpp
+++ b/plocate.cpp
@@ -49,8 +49,10 @@ bool ignore_case = false;
 bool only_count = false;
 bool print_nul = false;
 bool use_debug = false;
+bool flush_cache = false;
 bool patterns_are_regex = false;
 bool use_extended_regex = false;
+bool match_basename = false;
 int64_t limit_matches = numeric_limits<int64_t>::max();
 int64_t limit_left = numeric_limits<int64_t>::max();
 
@@ -80,8 +82,7 @@ public:
 Corpus::Corpus(int fd, IOUringEngine *engine)
 	: fd(fd), engine(engine)
 {
-	// Enable to test cold-cache behavior (except for access()).
-	if (false) {
+	if (flush_cache) {
 		off_t len = lseek(fd, 0, SEEK_END);
 		if (len == -1) {
 			perror("lseek");
@@ -194,9 +195,19 @@ void scan_file_block(const vector<Needle> &needles, string_view compressed,
 	for (const char *filename = block.data();
 	     filename != block.data() + block.size();
 	     filename += strlen(filename) + 1) {
+		const char *haystack = filename;
+		if (match_basename) {
+			haystack = strrchr(filename, '/');
+			if (haystack == nullptr) {
+				haystack = filename;
+			} else {
+				++haystack;
+			}
+		}
+
 		bool found = true;
 		for (const Needle &needle : needles) {
-			if (!matches(needle, filename)) {
+			if (!matches(needle, haystack)) {
 				found = false;
 				break;
 			}
@@ -489,6 +500,8 @@ void do_search_file(const vector<Needle> &needles, const char *filename)
 		// the pattern and done a union of them, but that's a lot of
 		// work for fairly unclear gain.)
 		uint64_t matched = scan_all_docids(needles, fd, corpus);
+		dprintf("Done in %.1f ms, found %" PRId64 " matches.\n",
+		        1e3 * duration<float>(steady_clock::now() - start).count(), matched);
 		if (only_count) {
 			printf("%" PRId64 "\n", matched);
 		}
@@ -638,6 +651,7 @@ void usage()
 	printf(
 		"Usage: plocate [OPTION]... PATTERN...\n"
 		"\n"
+		"  -b, --basename         search only the file name portion of path names\n"
 		"  -c, --count            print number of matches instead of the matches\n"
 		"  -d, --database DBPATH  search for files in DBPATH\n"
 		"                         (default is " DEFAULT_DBPATH ")\n"
@@ -646,6 +660,7 @@ void usage()
 		"  -0, --null             delimit matches by NUL instead of newline\n"
 		"  -r, --regexp           interpret patterns as basic regexps (slow)\n"
 		"      --regex            interpret patterns as extended regexps (slow)\n"
+		"  -w, --wholename        search the entire path name (default; see -b)\n"
 		"      --help             print this help\n"
 		"      --version          print version information\n");
 }
@@ -663,9 +678,11 @@ void version()
 int main(int argc, char **argv)
 {
 	constexpr int EXTENDED_REGEX = 1000;
+	constexpr int FLUSH_CACHE = 1001;
 	static const struct option long_options[] = {
 		{ "help", no_argument, 0, 'h' },
 		{ "count", no_argument, 0, 'c' },
+		{ "basename", no_argument, 0, 'b' },
 		{ "database", required_argument, 0, 'd' },
 		{ "ignore-case", no_argument, 0, 'i' },
 		{ "limit", required_argument, 0, 'l' },
@@ -673,18 +690,24 @@ int main(int argc, char **argv)
 		{ "version", no_argument, 0, 'V' },
 		{ "regexp", no_argument, 0, 'r' },
 		{ "regex", no_argument, 0, EXTENDED_REGEX },
+		{ "wholename", no_argument, 0, 'w' },
 		{ "debug", no_argument, 0, 'D' },  // Not documented.
+		// Enable to test cold-cache behavior (except for access()). Not documented.
+		{ "flush-cache", no_argument, 0, FLUSH_CACHE },
 		{ 0, 0, 0, 0 }
 	};
 
 	setlocale(LC_ALL, "");
 	for (;;) {
 		int option_index = 0;
-		int c = getopt_long(argc, argv, "cd:hil:n:0VD", long_options, &option_index);
+		int c = getopt_long(argc, argv, "bcd:hil:n:0rwVD", long_options, &option_index);
 		if (c == -1) {
 			break;
 		}
 		switch (c) {
+		case 'b':
+			match_basename = true;
+			break;
 		case 'c':
 			only_count = true;
 			break;
@@ -715,9 +738,15 @@ int main(int argc, char **argv)
 			patterns_are_regex = true;
 			use_extended_regex = true;
 			break;
+		case 'w':
+			match_basename = false;  // No-op unless -b is given first.
+			break;
 		case 'D':
 			use_debug = true;
 			break;
+		case FLUSH_CACHE:
+			flush_cache = true;
+			break;
 		case 'V':
 			version();
 			break;
@@ -726,10 +755,12 @@ int main(int argc, char **argv)
 		}
 	}
 
-	if (use_debug) {
+	if (use_debug || flush_cache) {
 		// Debug information would leak information about which files exist,
 		// so drop setgid before we open the file; one would either need to run
-		// as root, or use a locally-built file.
+		// as root, or use a locally-built file. Doing the same thing for
+		// flush_cache is mostly paranoia, in an attempt to prevent random users
+		// from making plocate slow for everyone else.
 		if (setgid(getgid()) != 0) {
 			perror("setgid");
 			exit(EXIT_FAILURE);