+Buffered reading (6.3x times faster)

diff --git a/cmc-find.c b/cmc-find.c
index c06cd269b8ae35202f34cd2eece19300c7d7338c..5fa4093407f5b8c3c8f33a650ff92162c1ca6417 100644 (file)
--- a/cmc-find.c
+++ b/cmc-find.c
@@ -11,6 +11,10 @@
 #define MAX_LEN 64
 #endif
 
+#ifndef BUF_LEN
+#define BUF_LEN (64*1024)
+#endif
+
 typedef struct cs_t {
        char buf[MAX_LEN];
 
@@ -19,11 +23,24 @@ typedef struct cs_t {
        size_t start;           /* index of firs char */
 } *cs;
 
+typedef struct buf_t {
+       char buf[BUF_LEN];
+
+       FILE *fp;
+       size_t cap /* max string size; <= MAX_LEN */ ;
+       size_t cur;
+} *buf;
+
 static void cs_init(cs S, size_t cap);
 static void cs_add(cs S, char ch);
 static int cs_cmp(cs S, char *ch);
 static void cs_print(cs S);
 
+static void buf_init(buf B, FILE *);
+static void buf_read(buf B);
+static char buf_next_ch(buf B);
+static int buf_eof(buf B);
+
 int main(int argc, char **argv)
 {
        int ch = 0;
@@ -31,7 +48,8 @@ int main(int argc, char **argv)
        time_t now;
 
        char str[MAX_LEN];
-       struct cs_t matcher;
+       struct cs_t S;
+       struct buf_t B;
 
        if (argc < 2) {
                printf("Usage: %s <string to find>\n", argv[0]);
@@ -41,15 +59,20 @@ int main(int argc, char **argv)
        printf("Check against %s\n", str);
 
        now = time(0);
-       cs_init(&matcher, strlen(str));
+       buf_init(&B, stdin);
+       cs_init(&S, strlen(str));
+
+       for (n = 1;; ++n) {
+               ch = buf_next_ch(&B);
+               if (!ch && buf_eof(&B))
+                       break;
 
-       for (n = 1; (ch = fgetc(stdin)) != EOF; ++n) {
-               cs_add(&matcher, ch);
+               cs_add(&S, ch);
 
-               if (matcher.len == matcher.cap) {
-                       if (cs_cmp(&matcher, str)) {
-                               printf("Found at %llu\t%lu\n", n - matcher.cap, time(0) - now);
-                               /* cs_print(&matcher); printf("\n"); */
+               if (S.len == S.cap) {
+                       if (cs_cmp(&S, str)) {
+                               printf("Found at %llu\t%lu\n", n - S.cap, time(0) - now);
+                               /* cs_print(&S); printf("\n"); */
                        }
                }
        }
@@ -132,3 +155,31 @@ static void cs_print(cs S)
                ++count;
        }
 }
+
+static void buf_init(buf B, FILE * fp)
+{
+       B->cap = 0;
+       B->cur = 0;
+       B->fp = fp;
+}
+
+static void buf_read(buf B)
+{
+       B->cap = fread(B->buf, 1, sizeof(B->buf), B->fp);
+       B->cur = 0;
+}
+
+static char buf_next_ch(buf B)
+{
+       if (B->cur >= B->cap) {
+               buf_read(B);
+               if (B->cap == 0)
+                       return '\0';
+       }
+       return B->buf[B->cur++];
+}
+
+static int buf_eof(buf B)
+{
+       return feof(B->fp) && B->cur >= B->cap;
+}