kmp_enter_raw_string uses zero-terminated array of kmp_char_t characters

instead of bytes

diff --git a/lib/kmp.c b/lib/kmp.c
index c2e2dd10f8437a11c9ea8d7ac2d067277813fa5c..e30eb62cc6486768c9a26cb5e1ddef9ae1b12c0e 100644 (file)
--- a/lib/kmp.c
+++ b/lib/kmp.c
@@ -10,7 +10,7 @@
 #include "lib/lists.h"
 #include "lib/unicode.h"
 
-#define KMP_GET_CHAR KMP_GET_RAW
+#define KMP_GET_CHAR(pos, c, flags) do{}while(0)
 #include "lib/kmp.h"
 
 #include <stdlib.h>
@@ -65,16 +65,15 @@ new_output(struct kmp *kmp, uns id, uns len)
 }
  
 void
-kmp_enter_raw_string(struct kmp *kmp, const byte *str, uns id)
+kmp_enter_raw_string(struct kmp *kmp, kmp_char_t *str, uns id)
 {
        struct kmp_transition tr = { .next=NULL, .from=0 }, **prev;
        struct kmp_output *new_out;
-       const byte *orig_str = str;
        uns len = 0;
        kmp_char_t c = 'a';
 
-       TRACE(20, "kmp.c: Entering string %s", str);
-       kmp_get_char(&str, &c, 0);
+       TRACE(20, "kmp.c: Entering string");
+       c = *str++;
        len++;
        if (!c)
                return;
@@ -85,7 +84,7 @@ kmp_enter_raw_string(struct kmp *kmp, const byte *str, uns id)
                if (!*prev)
                        break;
                tr.from = (*prev)->to;
-               kmp_get_char(&str, &c, 0);
+               c = *str++;
                len++;
        }
        while (c)
@@ -95,7 +94,7 @@ kmp_enter_raw_string(struct kmp *kmp, const byte *str, uns id)
                **prev = tr;
                add_tail(kmp->g.sons + tr.from, &(*prev)->n);
                init_list(kmp->g.sons + tr.to);
-               kmp_get_char(&str, &c, 0);
+               c = *str++;
                len++;
                tr.from = tr.to;
                tr.c = c;
@@ -103,7 +102,7 @@ kmp_enter_raw_string(struct kmp *kmp, const byte *str, uns id)
                ASSERT(!*prev);
        }
        if (kmp->out[tr.from])
-               TRACE(5, "kmp.c: string %s is inserted more than once", orig_str);
+               TRACE(5, "kmp.c: string is inserted more than once");
        new_out = new_output(kmp, id, len-1);
        merge_output(kmp->out + tr.from, new_out);
 }

diff --git a/lib/kmp.h b/lib/kmp.h
index 7d392bf95b4dbf87c338a0676443bc8a922ca969..7264b9545df0db8f41eda66425e4c9dd2177ed19 100644 (file)
--- a/lib/kmp.h
+++ b/lib/kmp.h
@@ -33,7 +33,7 @@
 
 /* Pre-defined input functions */
 
-#define KMP_GET_RAW(pos, c, flags) do { uns cc; pos = utf8_get(pos, &cc); c=cc; } while(0)
+#define KMP_GET_UTF8(pos, c, flags) do { uns cc; pos = utf8_get(pos, &cc); c = cc; } while(0)
 
 #define KMP_GET_ASCII(pos, c, flags) do { \
        c = *pos++; \
@@ -86,11 +86,11 @@ struct kmp_result {
 
 /* kmp.c */
 struct kmp *kmp_new(struct mempool *mp, int words_len, uns modify_flags);
-void kmp_enter_raw_string(struct kmp *kmp, const byte *str, uns id);
+void kmp_enter_raw_string(struct kmp *kmp, kmp_char_t *str, uns id);
 void kmp_build(struct kmp *kmp);
 
 static inline void
-kmp_get_char(const byte **str, kmp_char_t *c, uns modify_flags UNUSED)
+kmp_get_char(const byte **str UNUSED, kmp_char_t *c, uns modify_flags UNUSED)
 {
        while (1)
        {
@@ -112,12 +112,11 @@ kmp_enter_string(struct kmp *kmp, const byte *str, uns id)
         * to a conversion wrapper (this function) and the rest, which resides in kmp.c
         * and uses KMP_GET_RAW to read its input.
         */
-       byte buf[3*strlen(str)+1], *str2 = buf;
-       kmp_char_t c = 0;
+       kmp_char_t buf[strlen(str)+1], *str2 = buf, c = 0;
        do
        {
                kmp_get_char(&str, &c, kmp->modify_flags);
-               str2 = utf8_put(str2, c);
+               *str2++ = c;
        }
        while (c);
        kmp_enter_raw_string(kmp, buf, id);