#include <string.h>
#include <string>
#include <string_view>
+#include <sys/stat.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <thread>
bool patterns_are_regex = false;
bool use_extended_regex = false;
bool match_basename = false;
+bool check_existence = false;
+bool ignore_visibility = 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;
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;
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) {
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) {
// This too. (We ignore the other max_version 2 fields.)
hdr.check_visibility = true;
}
+ if (ignore_visibility) {
+ hdr.check_visibility = false;
+ }
}
Corpus::~Corpus()
return hdr.num_docids;
}
+template<class T>
+void stat_if_needed(const char *filename, bool access_ok, IOUringEngine *engine, T cb)
+{
+ if (!access_ok || !check_existence) {
+ // Doesn't have access or doesn't care about existence, so no need to stat.
+ cb(access_ok);
+ } else if (engine == nullptr || !engine->get_supports_stat()) {
+ // Do a synchronous stat.
+ struct stat buf;
+ bool ok = lstat(filename, &buf) == 0;
+ cb(ok);
+ } else {
+ engine->submit_stat(filename, cb);
+ }
+}
+
void scan_file_block(const vector<Needle> &needles, string_view compressed,
- AccessRXCache *access_rx_cache, uint64_t seq, ResultReceiver *serializer,
+ IOUringEngine *engine, AccessRXCache *access_rx_cache, uint64_t seq, ResultReceiver *serializer,
atomic<uint64_t> *matched)
{
unsigned long long uncompressed_len = ZSTD_getFrameContentSize(compressed.data(), compressed.size());
block[block.size() - 1] = '\0';
auto test_candidate = [&](const char *filename, uint64_t local_seq, uint64_t next_seq) {
- access_rx_cache->check_access(filename, /*allow_async=*/true, [matched, serializer, local_seq, next_seq, filename{ strdup(filename) }](bool ok) {
- if (ok) {
- ++*matched;
- serializer->print(local_seq, next_seq - local_seq, filename);
- } else {
- serializer->print(local_seq, next_seq - local_seq, "");
- }
- free(filename);
+ access_rx_cache->check_access(filename, /*allow_async=*/true, [matched, engine, serializer, local_seq, next_seq, filename{ strdup(filename) }](bool ok) {
+ stat_if_needed(filename, ok, engine, [matched, serializer, local_seq, next_seq, filename](bool ok) {
+ if (ok) {
+ ++*matched;
+ serializer->print(local_seq, next_seq - local_seq, filename);
+ } else {
+ serializer->print(local_seq, next_seq - local_seq, "");
+ }
+ free(filename);
+ });
});
};
atomic<uint64_t> matched{ 0 };
for (size_t i = 0; i < docids.size(); ++i) {
uint32_t docid = docids[i];
- corpus.get_compressed_filename_block(docid, [i, &matched, &needles, &access_rx_cache, &docids_in_order](string_view compressed) {
- scan_file_block(needles, compressed, &access_rx_cache, i, &docids_in_order, &matched);
+ corpus.get_compressed_filename_block(docid, [i, &matched, &needles, &access_rx_cache, engine, &docids_in_order](string_view compressed) {
+ scan_file_block(needles, compressed, engine, &access_rx_cache, i, &docids_in_order, &matched);
});
}
engine->finish();
string compressed;
{
- unique_lock<mutex> lock(mu);
+ unique_lock lock(mu);
queue_added.wait(lock, [&work_queue, &done] { return !work_queue.empty() || done; });
if (done && work_queue.empty()) {
return;
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 }, &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);
}
}
});
complete_pread(fd, &compressed[0], io_len, offsets[io_docid]);
{
- unique_lock<mutex> lock(mu);
+ unique_lock lock(mu);
queue_removed.wait(lock, [&work_queue] { return work_queue.size() < 256; }); // Allow ~2MB of data queued up.
work_queue.emplace_back(io_docid, last_docid, move(compressed));
queue_added.notify_one(); // Avoid the thundering herd.
}
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;
if (only_count) {
printf("0\n");
}
- exit(0);
+ exit(1);
}
}
}
" -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"
constexpr int EXTENDED_REGEX = 1000;
constexpr int FLUSH_CACHE = 1001;
+ constexpr int IGNORE_VISIBILITY = 1002;
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' },
{ "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 },
+ // Mostly useful to dump out the entire database, even if the given directories
+ // are gone. Disables sgid due to security. Not documented.
+ { "ignore-visibility", no_argument, 0, IGNORE_VISIBILITY },
{ 0, 0, 0, 0 }
};
setlocale(LC_ALL, "");
for (;;) {
int option_index = 0;
- int c = getopt_long(argc, argv, "bcd:hil: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;
case 'd':
parse_dbpaths(optarg, &dbpaths);
break;
+ case 'e':
+ check_existence = true;
+ break;
case 'h':
usage();
exit(0);
exit(1);
}
break;
+ case 'N':
+ literal_printing = true;
+ break;
case '0':
print_nul = true;
break;
case 'V':
version();
break;
+ case IGNORE_VISIBILITY:
+ ignore_visibility = true;
+ break;
default:
exit(1);
}
}
- if (use_debug || flush_cache) {
+ if (use_debug || flush_cache || ignore_visibility) {
// 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. Doing the same thing for
// flush_cache is mostly paranoia, in an attempt to prevent random users
// from making plocate slow for everyone else.
+ // --ignore-visibility is obvious; if we allowed to keep sgid with
+ // that flag on, it would subvert the entire security model.
if (setgid(getgid()) != 0) {
perror("setgid");
exit(EXIT_FAILURE);
}
if (needles.empty()) {
fprintf(stderr, "plocate: no pattern to search for specified\n");
- exit(0);
+ exit(1);
}
if (dbpaths.empty()) {
if (only_count) {
printf("%" PRId64 "\n", matched);
}
+
+ return matched == 0;
}
projects / plocate / blobdiff