Merge pull request #7 from siro53/pysan3

ahocorasick

.gitignore

@@ -1,4 +1,6 @@
 grpwk
+distance
+result
 *.sh
 *.py
 !analyze.py

Makefile

@@ -1,14 +1,7 @@
 PROG = grpwk
-OBJS = test_ahocorasick.o ahocorasick.o ahotrie.o ahotext.o queue.o
 CC = gcc
 CFLAGS = -W -Wall -Wextra -Wconversion -Wshadow
-LDFLAGS = 
 
-.SUFFIXES: .c
-
-$(PROG): $(OBJS)
-	$(CC) $(LDFLAGS) -o $(PROG) $^
-.c.o:
-	$(CC) $(CFLAGS) -c $<
-clean:
-	rm  $(OBJS) $(PROG)
+FILES = $(shell gcc test_distance.c -o distance | find ./ \( -name "*.c" \) \( -not -name "*test*" \))
+default:
+	$(CC) $(CFLAGS) -o $(PROG) $(FILES)

ahotext.c

@@ -1,5 +1,3 @@
-#pragma once
-
 #include <stdlib.h>
 #include <string.h>
 

ahotrie.c

@@ -84,6 +84,8 @@ void trie_connect(aho_trie * restrict t) {
             while (connect_link(tmp, child) == FALSE) tmp = tmp->failure_link;
         }
     }
+
+    que_destroy(&que);
 }
 
 void trie_delete(aho_trie * restrict t) {
@@ -99,6 +101,8 @@ void trie_delete(aho_trie * restrict t) {
 
         if (node->parent == NULL) continue;
     }
+
+    que_destroy(&que);
 }
 
 int find_node(aho_node ** restrict node, const char text) {
@@ -137,4 +141,6 @@ void trie_print(aho_trie * restrict t) {
 
         printf("%c, refs:%d, fail: %p, output:%p\n", node->data + 'a', node->ref_count, node->failure_link, node->output_link);
     }
+
+    que_destroy(&que);
 }

example/grpwk.c

@@ -0,0 +1,11 @@
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "grpwk.h"
+
+char *grpwk(const string_s t, const string_s s[], int len) {
+    char *str = (char *)malloc(sizeof(char) * 100);
+    sprintf(str, "I got %d data!", len);
+
+    return str;
+}

example/grpwk.h

@@ -0,0 +1,5 @@
+#pragma once
+
+#include "../string_s.h"
+
+char *grpwk(const string_s t, const string_s s[], int len);

input_win.c

@@ -0,0 +1,45 @@
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include "string_s.h"
+
+// change here to include your files!!
+#include "example/grpwk.h"
+
+int main_prg(int, char **);
+
+int main(int argc, char **argv)
+{
+    __clock_t c_start, c_end;
+
+    c_start = clock();
+    main_prg(argc, argv);
+    c_end = clock();
+
+    printf("%f\n", (double)(c_end - c_start) / CLOCKS_PER_SEC);
+    return (0);
+}
+
+int main_prg(int argc, char **argv)
+{
+    assert(argc == 3);
+    FILE *fp_in = fopen(argv[1], "r");
+    assert(fp_in != NULL);
+    FILE *fp_out = fopen(argv[2], "w");
+    assert(fp_out != NULL);
+
+    string_s t, s[50000];
+
+    // input t
+    fscanf(fp_in, "%s", t.str);
+    t.len = strlen(t.str);
+
+    // input s[]
+    int i = 0;
+    while (~fscanf(fp_in, "%s", s[i].str)) s[i++].len = strlen(s[i].str);
+
+    fprintf(fp_out, "%s\n", grpwk(t, s, i));
+
+    return 0;
+}

queue.c

@@ -5,14 +5,15 @@
 
 void que_init(queue *que)
 {
+    que->data = (QUE_TYPE *)malloc(sizeof(QUE_TYPE) * MAX_SIZE);
     que->front = 0;
     que->back = 0;
     que->size = 0;
 }
 
 void que_push(queue *que, QUE_TYPE value)
 {
-    if (que->back >= MAX_SIZE)
+    if (que->size >= MAX_SIZE)
     {
         printf("キューの容量がいっぱいです\n");
         return;
@@ -24,7 +25,7 @@ void que_push(queue *que, QUE_TYPE value)
 
 QUE_TYPE que_pop(queue *que)
 {
-    if (que->front >= que->back)
+    if (que->size == 0)
     {
         printf("キューの中身は空です\n");
         return NULL;
@@ -35,13 +36,17 @@ QUE_TYPE que_pop(queue *que)
     return top;
 }
 
-void printQue(queue *que)
-{
-    int i;
-    printf("data : ");
-    for (i = que->front; i < que->back; i++)
-    {
-        printf("%p ", que->data[i]);
-    }
-    printf("\n");
+void que_destroy(queue *que) {
+    free(que->data);
 }
+
+// void printQue(queue *que)
+// {
+//     int i;
+//     printf("data : ");
+//     for (i = que->front; i < que->back; i++)
+//     {
+//         printf("%p ", que->data[i]);
+//     }
+//     printf("\n");
+// }

queue.h

@@ -10,7 +10,7 @@
 
 typedef struct
 {
-    QUE_TYPE data[MAX_SIZE];
+    QUE_TYPE *data;
     int front;
     int back;
     int size;
@@ -19,6 +19,7 @@ typedef struct
 void que_init(queue *);
 void que_push(queue *, QUE_TYPE);
 QUE_TYPE que_pop(queue *);
-void printQue(queue *);
+void que_destroy(queue *);
+// void printQue(queue *);
 
 #endif

string_s.h

@@ -0,0 +1,6 @@
+#pragma once
+
+typedef struct {
+    char str[120];
+    int len;
+} string_s;

test_ahocorasick.c

@@ -5,21 +5,21 @@
 
 int test_trie(void) {
     char *s[] = {
-        // "ab",
-        // "abc",
-        // "aaaaabddbdaaacccaacbaabacbaadb",
+        "ab",
+        "abc",
+        "aaaaabddbdaaacccaacbaabacbaadb",
         "aaacdbdbcbcdbdbadaacbaadbdbdaacaaaaaadacababdadddacaacbaaaabdacdadadbabbbddaaddaaaaa",
         "abacbadaadbcaaabaaacbbaabadbababdbcadbd",
-        // "dbbbdaabaaabaabab",
-        // "daaadaaa",
-        // "dbaac",
-        // "ad",
-        // "bdadaabbaaadaabdd",
-        // "ddaabdd",
-        // "bdbabb",
-        // "abdb",
-        // "adbab",
-        // "aaabaaabcadba",
+        "dbbbdaabaaabaabab",
+        "daaadaaa",
+        "dbaac",
+        "ad",
+        "bdadaabbaaadaabdd",
+        "ddaabdd",
+        "bdbabb",
+        "abdb",
+        "adbab",
+        "aaabaaabcadba",
     };
     aho_text text[sizeof(s)/sizeof(s[0])];
     for (int i=0; i<sizeof(s)/sizeof(s[0]); i++) {
@@ -32,7 +32,7 @@ int test_trie(void) {
     for (int i=0; i<sizeof(text)/sizeof(text[0]); i++) if (!trie_add(t, &text[i])) printf("error (unexpected input [^a-d]\n");
     trie_connect(t);
 
-    // trie_print(t);
+    trie_print(t);
 
     return 0;
 }
@@ -65,7 +65,6 @@ int test_ahocora(void) {
     printf("total match: %d\n", aho_search(&aho, test, strlen(test)));
 
     aho_destroy(&aho);
-    printf("end of func\n");
 
     return 0;
 }
@@ -78,8 +77,8 @@ int test_input(void) {
     FILE *fp = fopen("data/dat0_in", "r");
     fscanf(fp, "%s", s);
     while (~fscanf(fp, "%s", s))
-        printf("%d\n", aho_add_match_text(&aho, s, strlen(s)));
-    printf("here\n");
+        if (strlen(s) >= 60)
+            aho_add_match_text(&aho, s, strlen(s));
 
     fp = fopen("data/dat0_ref", "r");
     fscanf(fp, "%s", s);

test_distance.c

@@ -0,0 +1,23 @@
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+int main(int argc, char *argv[]) {
+    FILE *fp_in = fopen(argv[1], "r");
+    assert(fp_in != NULL);
+    FILE *fp_out = fopen(argv[2], "r");
+    assert(fp_out != NULL);
+
+    char in[500000], out[500000];
+    fscanf(fp_in, "%s", in);
+    fscanf(fp_out, "%s", out);
+
+    int counter = 0;
+    for (int i=0; i<strlen(in); i++) counter += (in[i] != out[i]);
+    counter += abs(strlen(in) - strlen(out));
+
+    printf("edit distance: %d\n", counter);
+
+    return 0;
+}