Inline EncodingCorpus::get_pl_builder(). Saves ~2% CPU.

[plocate] / plocate.cpp

diff --git a/plocate.cpp b/plocate.cpp
index e863edbc4ab9294c89bcbd45e822c560190e0f7f..c47c6480465a7f9e70472e7cbac984e3c6ddc78d 100644 (file)
--- a/plocate.cpp
+++ b/plocate.cpp
@@ -1,4 +1,5 @@
 #include "access_rx_cache.h"
+#include "complete_pread.h"
 #include "db.h"
 #include "dprintf.h"
 #include "io_uring_engine.h"
@@ -42,9 +43,7 @@
 using namespace std;
 using namespace std::chrono;
 
-#define DEFAULT_DBPATH "/var/lib/mlocate/plocate.db"
-
-const char *dbpath = DEFAULT_DBPATH;
+const char *dbpath = DBFILE;
 bool ignore_case = false;
 bool only_count = false;
 bool print_nul = false;
@@ -55,6 +54,7 @@ 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();
+bool stdout_is_tty = false;
 
 steady_clock::time_point start;
 ZSTD_DDict *ddict = nullptr;
@@ -105,6 +105,10 @@ Corpus::Corpus(int fd, IOUringEngine *engine)
                hdr.zstd_dictionary_offset_bytes = 0;
                hdr.zstd_dictionary_length_bytes = 0;
        }
+       if (hdr.max_version < 2) {
+               // This too. (We ignore the other max_version 2 fields.)
+               hdr.check_visibility = true;
+       }
 }
 
 Corpus::~Corpus()
@@ -230,7 +234,7 @@ void scan_file_block(const vector<Needle> &needles, string_view compressed,
 size_t scan_docids(const vector<Needle> &needles, const vector<uint32_t> &docids, const Corpus &corpus, IOUringEngine *engine)
 {
        Serializer docids_in_order;
-       AccessRXCache access_rx_cache(engine);
+       AccessRXCache access_rx_cache(engine, corpus.get_hdr().check_visibility);
        atomic<uint64_t> matched{ 0 };
        for (size_t i = 0; i < docids.size(); ++i) {
                uint32_t docid = docids[i];
@@ -306,7 +310,7 @@ uint64_t scan_all_docids(const vector<Needle> &needles, int fd, const Corpus &co
                }
        }
 
-       AccessRXCache access_rx_cache(nullptr);
+       AccessRXCache access_rx_cache(nullptr, corpus.get_hdr().check_visibility);
        Serializer serializer;
        uint32_t num_blocks = corpus.get_num_filename_blocks();
        unique_ptr<uint64_t[]> offsets(new uint64_t[num_blocks + 1]);
@@ -501,7 +505,7 @@ void do_search_file(const vector<Needle> &needles, const char *filename)
                // 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);
+                       1e3 * duration<float>(steady_clock::now() - start).count(), matched);
                if (only_count) {
                        printf("%" PRId64 "\n", matched);
                }
@@ -601,7 +605,7 @@ void do_search_file(const vector<Needle> &needles, const char *filename)
                                if (done)
                                        return;
 
-                               uint32_t trgm __attribute__((unused)) = trgmptr.trgm;
+                               uint32_t trgm = trgmptr.trgm;
                                const unsigned char *pldata = reinterpret_cast<const unsigned char *>(s.data());
                                size_t num = trgmptr.num_docids;
                                decoded.resize(num);
@@ -654,7 +658,7 @@ void usage()
                "  -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"
+               "                         (default is " DBFILE ")\n"
                "  -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"
@@ -667,7 +671,7 @@ void usage()
 
 void version()
 {
-       printf("plocate %s\n", PLOCATE_VERSION);
+       printf("%s %s\n", PACKAGE_NAME, PACKAGE_VERSION);
        printf("Copyright 2020 Steinar H. Gunderson\n");
        printf("License GPLv2+: GNU GPL version 2 or later <https://gnu.org/licenses/gpl.html>.\n");
        printf("This is free software: you are free to change and redistribute it.\n");
@@ -700,7 +704,7 @@ int main(int argc, char **argv)
        setlocale(LC_ALL, "");
        for (;;) {
                int option_index = 0;
-               int c = getopt_long(argc, argv, "bcd:hil:n:0wVD", long_options, &option_index);
+               int c = getopt_long(argc, argv, "bcd:hil:n:0rwVD", long_options, &option_index);
                if (c == -1) {
                        break;
                }
@@ -767,6 +771,10 @@ int main(int argc, char **argv)
                }
        }
 
+       if (!print_nul) {
+               stdout_is_tty = isatty(1);
+       }
+
        vector<Needle> needles;
        for (int i = optind; i < argc; ++i) {
                Needle needle;