#include "db.h"
#include "dprintf.h"
#include "io_uring_engine.h"
+#include "needle.h"
#include "parse_trigrams.h"
#include "serializer.h"
#include "turbopfor.h"
#include <condition_variable>
#include <deque>
#include <fcntl.h>
-#include <fnmatch.h>
#include <functional>
#include <getopt.h>
#include <inttypes.h>
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;
int64_t limit_matches = numeric_limits<int64_t>::max();
steady_clock::time_point start;
ZSTD_DDict *ddict = nullptr;
-regex_t compile_regex(const string &needle);
-
-struct Needle {
- enum { STRSTR,
- REGEX,
- GLOB } type;
- string str; // Filled in no matter what.
- regex_t re; // For REGEX.
-};
-
-bool matches(const Needle &needle, const char *haystack)
-{
- if (needle.type == Needle::STRSTR) {
- return strstr(haystack, needle.str.c_str()) != nullptr;
- } else if (needle.type == Needle::GLOB) {
- int flags = ignore_case ? FNM_CASEFOLD : 0;
- return fnmatch(needle.str.c_str(), haystack, flags) == 0;
- } else {
- assert(needle.type == Needle::REGEX);
- return regexec(&needle.re, haystack, /*nmatch=*/0, /*pmatch=*/nullptr, /*flags=*/0) == 0;
- }
-}
-
class Corpus {
public:
Corpus(int fd, IOUringEngine *engine);
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");
}
}
-string unescape_glob_to_plain_string(const string &needle)
-{
- string unescaped;
- for (size_t i = 0; i < needle.size(); i += read_unigram(needle, i).second) {
- uint32_t ch = read_unigram(needle, i).first;
- assert(ch != WILDCARD_UNIGRAM);
- if (ch == PREMATURE_END_UNIGRAM) {
- fprintf(stderr, "Pattern '%s' ended prematurely\n", needle.c_str());
- exit(1);
- }
- unescaped.push_back(ch);
- }
- return unescaped;
-}
-
-regex_t compile_regex(const string &needle)
-{
- regex_t re;
- int flags = REG_NOSUB;
- if (ignore_case) {
- flags |= REG_ICASE;
- }
- if (use_extended_regex) {
- flags |= REG_EXTENDED;
- }
- int err = regcomp(&re, needle.c_str(), flags);
- if (err != 0) {
- char errbuf[256];
- regerror(err, &re, errbuf, sizeof(errbuf));
- fprintf(stderr, "Error when compiling regex '%s': %s\n", needle.c_str(), errbuf);
- exit(1);
- }
- return re;
-}
-
void usage()
{
printf(
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' },
{ "regexp", no_argument, 0, 'r' },
{ "regex", no_argument, 0, EXTENDED_REGEX },
{ "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 }
};
case 'D':
use_debug = true;
break;
+ case FLUSH_CACHE:
+ flush_cache = true;
+ break;
case 'V':
version();
break;
}
}
- 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);