Move several needle/searching related functions into its own file.

diff --git a/meson.build b/meson.build
index d6d0e99b55fbe4127c82e38dda78e81bc2b35fc8..913fc0f8a98b04d5a2c690652eb29e8ba2a8d340 100644 (file)
--- a/meson.build
+++ b/meson.build
@@ -12,7 +12,7 @@ if not uringdep.found()
        add_project_arguments('-DWITHOUT_URING', language: 'cpp')
 endif
 
-executable('plocate', ['plocate.cpp', 'io_uring_engine.cpp', 'turbopfor.cpp', 'parse_trigrams.cpp', 'serializer.cpp', 'access_rx_cache.cpp'],
+executable('plocate', ['plocate.cpp', 'io_uring_engine.cpp', 'turbopfor.cpp', 'parse_trigrams.cpp', 'serializer.cpp', 'access_rx_cache.cpp', 'needle.cpp'],
        dependencies: [uringdep, zstddep, threaddep],
        install: true,
        install_mode: ['rwxr-sr-x', 'root', 'mlocate'])

diff --git a/needle.cpp b/needle.cpp
new file mode 100644 (file)
index 0000000..60f1698
--- /dev/null
+++ b/needle.cpp
@@ -0,0 +1,62 @@
+#include "needle.h"
+
+#include "options.h"
+#include "parse_trigrams.h"
+
+#include <assert.h>
+#include <fnmatch.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <utility>
+
+using namespace std;
+
+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;
+       }
+}
+
+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;
+}

diff --git a/needle.h b/needle.h
new file mode 100644 (file)
index 0000000..ce443f2
--- /dev/null
+++ b/needle.h
@@ -0,0 +1,19 @@
+#ifndef _NEEDLE_H
+#define _NEEDLE_H 1
+
+#include <regex.h>
+#include <string>
+
+struct Needle {
+       enum { STRSTR,
+              REGEX,
+              GLOB } type;
+       std::string str;  // Filled in no matter what.
+       regex_t re;  // For REGEX.
+};
+
+bool matches(const Needle &needle, const char *haystack);
+std::string unescape_glob_to_plain_string(const std::string &needle);
+regex_t compile_regex(const std::string &needle);
+
+#endif  // !defined(_NEEDLE_H)

diff --git a/plocate.cpp b/plocate.cpp
index 0d59076e8fc99efb4829384da1894e99c4399867..4141a316bb91a5381bfd01a28163f58341e26fd9 100644 (file)
--- a/plocate.cpp
+++ b/plocate.cpp
@@ -2,6 +2,7 @@
 #include "db.h"
 #include "dprintf.h"
 #include "io_uring_engine.h"
+#include "needle.h"
 #include "parse_trigrams.h"
 #include "serializer.h"
 #include "turbopfor.h"
@@ -14,7 +15,6 @@
 #include <condition_variable>
 #include <deque>
 #include <fcntl.h>
-#include <fnmatch.h>
 #include <functional>
 #include <getopt.h>
 #include <inttypes.h>
@@ -57,29 +57,6 @@ int64_t limit_left = 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);
@@ -656,41 +633,6 @@ void do_search_file(const vector<Needle> &needles, const char *filename)
        }
 }
 
-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(