X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=plocate.cpp;h=a1cd97a4a4645c08880b014d4799f9d4c33182f6;hb=fd6198891d6fd9642effc0843fef6f23b991af3e;hp=7cc4f5a4eb2587f4bb685aaa83155e8951f06dd2;hpb=cb87fdb5ede1d5a9fa0a5a309d9ccf98f42fe0b7;p=plocate

diff --git a/plocate.cpp b/plocate.cpp
index 7cc4f5a..a1cd97a 100644
--- a/plocate.cpp
+++ b/plocate.cpp
@@ -31,6 +31,7 @@
 #include <string.h>
 #include <string>
 #include <string_view>
+#include <sys/stat.h>
 #include <sys/types.h>
 #include <sys/wait.h>
 #include <thread>
@@ -57,6 +58,7 @@ bool check_existence = false;
 int64_t limit_matches = numeric_limits<int64_t>::max();
 int64_t limit_left = numeric_limits<int64_t>::max();
 bool stdout_is_tty = false;
+bool literal_printing = false;
 static bool in_forked_child = false;
 
 steady_clock::time_point start;
@@ -64,7 +66,7 @@ ZSTD_DDict *ddict = nullptr;
 
 class Corpus {
 public:
-	Corpus(int fd, IOUringEngine *engine);
+	Corpus(int fd, const char *filename_for_errors, IOUringEngine *engine);
 	~Corpus();
 	void find_trigram(uint32_t trgm, function<void(const Trigram *trgmptr, size_t len)> cb);
 	void get_compressed_filename_block(uint32_t docid, function<void(string_view)> cb) const;
@@ -82,7 +84,7 @@ public:
 	Header hdr;
 };
 
-Corpus::Corpus(int fd, IOUringEngine *engine)
+Corpus::Corpus(int fd, const char *filename_for_errors, IOUringEngine *engine)
 	: fd(fd), engine(engine)
 {
 	if (flush_cache) {
@@ -96,11 +98,11 @@ Corpus::Corpus(int fd, IOUringEngine *engine)
 
 	complete_pread(fd, &hdr, sizeof(hdr), /*offset=*/0);
 	if (memcmp(hdr.magic, "\0plocate", 8) != 0) {
-		fprintf(stderr, "plocate.db is corrupt or an old version; please rebuild it.\n");
+		fprintf(stderr, "%s: database is corrupt or not a plocate database; please rebuild it.\n", filename_for_errors);
 		exit(1);
 	}
 	if (hdr.version != 0 && hdr.version != 1) {
-		fprintf(stderr, "plocate.db has version %u, expected 0 or 1; please rebuild it.\n", hdr.version);
+		fprintf(stderr, "%s: has version %u, expected 0 or 1; please rebuild it.\n", filename_for_errors, hdr.version);
 		exit(1);
 	}
 	if (hdr.version == 0) {
@@ -347,7 +349,7 @@ uint64_t scan_all_docids(const vector<Needle> &needles, int fd, const Corpus &co
 	dprintf("Using %u worker threads for linear scan.\n", num_threads);
 	unique_ptr<WorkerThread[]> threads(new WorkerThread[num_threads]);
 	for (unsigned i = 0; i < num_threads; ++i) {
-		threads[i].t = thread([&threads, &mu, &queue_added, &queue_removed, &work_queue, &done, &offsets, &needles, &access_rx_cache, engine{ corpus.engine }, &matched, i] {
+		threads[i].t = thread([&threads, &mu, &queue_added, &queue_removed, &work_queue, &done, &offsets, &needles, &access_rx_cache, &matched, i] {
 			// regcomp() takes a lock on the regex, so each thread will need its own.
 			const vector<Needle> *use_needles = &needles;
 			vector<Needle> recompiled_needles;
@@ -378,7 +380,8 @@ uint64_t scan_all_docids(const vector<Needle> &needles, int fd, const Corpus &co
 				for (uint32_t docid = io_docid; docid < last_docid; ++docid) {
 					size_t relative_offset = offsets[docid] - offsets[io_docid];
 					size_t len = offsets[docid + 1] - offsets[docid];
-					scan_file_block(*use_needles, { &compressed[relative_offset], len }, engine, &access_rx_cache, docid, &receiver, &matched);
+					// IOUringEngine isn't thread-safe, so we do any needed stat()s synchronously (nullptr engine).
+					scan_file_block(*use_needles, { &compressed[relative_offset], len }, /*engine=*/nullptr, &access_rx_cache, docid, &receiver, &matched);
 				}
 			}
 		});
@@ -486,7 +489,7 @@ uint64_t do_search_file(const vector<Needle> &needles, const std::string &filena
 	}
 
 	IOUringEngine engine(/*slop_bytes=*/16);  // 16 slop bytes as described in turbopfor.h.
-	Corpus corpus(fd, &engine);
+	Corpus corpus(fd, filename.c_str(), &engine);
 	dprintf("Corpus init done after %.1f ms.\n", 1e3 * duration<float>(steady_clock::now() - start).count());
 
 	vector<TrigramDisjunction> trigram_groups;
@@ -807,6 +810,7 @@ void usage()
 		"  -i, --ignore-case      search case-insensitively\n"
 		"  -l, --limit LIMIT      stop after LIMIT matches\n"
 		"  -0, --null             delimit matches by NUL instead of newline\n"
+		"  -N, --literal          do not quote filenames, even if printing to a tty\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"
@@ -833,11 +837,13 @@ int main(int argc, char **argv)
 	static const struct option long_options[] = {
 		{ "help", no_argument, 0, 'h' },
 		{ "count", no_argument, 0, 'c' },
+		{ "all", no_argument, 0, 'A' },
 		{ "basename", no_argument, 0, 'b' },
 		{ "database", required_argument, 0, 'd' },
 		{ "existing", no_argument, 0, 'e' },
 		{ "ignore-case", no_argument, 0, 'i' },
 		{ "limit", required_argument, 0, 'l' },
+		{ "literal", no_argument, 0, 'N' },
 		{ "null", no_argument, 0, '0' },
 		{ "version", no_argument, 0, 'V' },
 		{ "regexp", no_argument, 0, 'r' },
@@ -852,11 +858,14 @@ int main(int argc, char **argv)
 	setlocale(LC_ALL, "");
 	for (;;) {
 		int option_index = 0;
-		int c = getopt_long(argc, argv, "bcd:ehil:n:0rwVD", long_options, &option_index);
+		int c = getopt_long(argc, argv, "Abcd:ehil:n:N0rwVD", long_options, &option_index);
 		if (c == -1) {
 			break;
 		}
 		switch (c) {
+		case 'A':
+			// Ignored.
+			break;
 		case 'b':
 			match_basename = true;
 			break;
@@ -883,6 +892,9 @@ int main(int argc, char **argv)
 				exit(1);
 			}
 			break;
+		case 'N':
+			literal_printing = true;
+			break;
 		case '0':
 			print_nul = true;
 			break;