X-Git-Url: http://git.madism.org/?a=blobdiff_plain;f=common%2Ftst-trie.c;h=2b7bdee64e7bb7e8cac5f9bbb80573249341337b;hb=76226c6f7b43aaa6480f776f70fb62f4a3e937df;hp=d2ccc14d21bb56d0705416a27fc4dcfc421f3c1e;hpb=fb519b0c3a78183d3fad3778d7d57bc2693598fc;p=apps%2Fpfixtools.git

diff --git a/common/tst-trie.c b/common/tst-trie.c
index d2ccc14..2b7bdee 100644
--- a/common/tst-trie.c
+++ b/common/tst-trie.c
@@ -33,54 +33,78 @@
  * Copyright © 2008 Florent Bruneau
  */
 
-#include <arpa/inet.h>
-#include <fcntl.h>
-#include <netinet/in.h>
-#include <sys/mman.h>
-#include <sys/stat.h>
-
+#include <time.h>
+#include <sys/time.h>
 #include "common.h"
 #include "str.h"
 #include "trie.h"
+#include "file.h"
 
 static trie_t *create_trie_from_file(const char *file)
 {
     trie_t *db;
-    const char *map, *p, *end;
-    struct stat st;
-    int fd;
+    file_map_t map;
+    const char *p, *end;
     char line[BUFSIZ];
 
-    fd = open(file, O_RDONLY, 0000);
-    if (fd < 0) {
-        UNIXERR("open");
+    if (!file_map_open(&map, file, false)) {
         return NULL;
     }
-
-    if (fstat(fd, &st) < 0) {
-        UNIXERR("fstat");
-        close(fd);
-        return NULL;
+    p   = map.map;
+    end = map.end;
+    while (end > p && end[-1] != '\n') {
+        --end;
+    }
+    if (end != map.end) {
+        warn("file %s miss a final \\n, ignoring last line", file);
     }
 
-    p = map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
-    if (map == MAP_FAILED) {
-        UNIXERR("mmap");
-        close(fd);
-        return NULL;
+    db = trie_new();
+    while (p < end && p != NULL) {
+        const char *eol = (char *)memchr(p, '\n', end - p);
+        if (eol == NULL) {
+            eol = end;
+        }
+        if (eol - p > BUFSIZ) {
+            p = eol - BUFSIZ;
+        }
+        int i = 0;
+#if 1
+        for (const char *s = eol - 1 ; s >= p ; --s) {
+            line[i++] = ascii_tolower(*s);
+        }
+#else
+        memcpy(line, p, eol - p);
+        i = eol - p;
+#endif
+        line[i] = '\0';
+        trie_insert(db, line);
+        p = eol + 1;
     }
-    close(fd);
+    file_map_close(&map);
+    trie_compile(db, false);
+    return db;
+}
 
-    end = map + st.st_size;
-    while (end > map && end[-1] != '\n') {
+__attribute__((used))
+static void check_trie_with_file(const trie_t *db, const char *file)
+{
+    file_map_t map;
+    const char *p, *end;
+    char line[BUFSIZ];
+
+    if (!file_map_open(&map, file, false)) {
+        return;
+    }
+    p   = map.map;
+    end = map.end;
+    while (end > p && end[-1] != '\n') {
         --end;
     }
-    if (end != map + st.st_size) {
-        syslog(LOG_WARNING, "file %s miss a final \\n, ignoring last line",
-               file);
+    if (end != map.end) {
+        warn("file %s miss a final \\n, ignoring last line", file);
     }
 
-    db = trie_new();
     while (p < end && p != NULL) {
         const char *eol = (char *)memchr(p, '\n', end - p);
         if (eol == NULL) {
@@ -99,11 +123,12 @@ static trie_t *create_trie_from_file(const char *file)
         i = eol - p;
 #endif
         line[i] = '\0';
-        trie_insert(db, line);
+        if (!trie_lookup(db, line)) {
+          warn("'%s' not found in the trie", line);
+        }
         p = eol + 1;
     }
-    munmap((void*)map, st.st_size);
-    return db;
+    file_map_close(&map);
 }
 
 
@@ -112,12 +137,14 @@ int main(int argc, char *argv[])
     /* Trivial tests
      */
     trie_t *trie = trie_new();
-    trie_insert(trie, "abcdefghi");
-    trie_insert(trie, "abcde123654789");
     trie_insert(trie, "abcde123456789");
     trie_insert(trie, "abcde123654789");
+    trie_insert(trie, "abcdefghi");
     trie_insert(trie, "coucou");
     trie_insert(trie, "coucou chez vous");
+    trie_insert(trie, "debout !");
+    trie_compile(trie, false);
+    trie_inspect(trie, true);
 
 #define ASSERT_TRUE(str)                            \
     if (!trie_lookup(trie, str)) {                  \
@@ -133,16 +160,36 @@ int main(int argc, char *argv[])
     ASSERT_FALSE("coucou ");
     ASSERT_FALSE("abcde123");
     ASSERT_FALSE("abcde");
+    ASSERT_FALSE("coucou chez vous tous");
+    ASSERT_TRUE("abcde123456789");
+    ASSERT_TRUE("abcde123456789");
+    ASSERT_TRUE("abcde123654789");
     ASSERT_TRUE("abcdefghi");
     ASSERT_TRUE("coucou");
-    ASSERT_FALSE("coucou chez vous tous");
+    ASSERT_TRUE("coucou chez vous");
+    ASSERT_TRUE("debout !");
+
     trie_delete(&trie);
 
     /* Perf test
      */
     if (argc > 1) {
         trie = create_trie_from_file(argv[1]);
-//        trie_inspect(trie);
+        trie_inspect(trie, false);
+        check_trie_with_file(trie, argv[1]);
+        if (argc > 2) {
+            const uint32_t how_many = 8 * 1000 * 1000;
+            struct timeval start, end;
+            double diff;
+
+            gettimeofday(&start, NULL);
+            for (uint32_t i = 0 ; i < how_many ; ++i) {
+                trie_lookup(trie, argv[2]);
+            }
+            gettimeofday(&end, NULL);
+            diff = (end.tv_sec - start.tv_sec) + (double)(end.tv_usec - start.tv_usec) / 10e6;
+            printf("%u lookups per second\n", (int)(how_many / diff));
+        }
         trie_delete(&trie);
     }
     return 0;