0
|
1 |
/**
|
|
2 |
* $Id: sentence.c 53 2008-01-10 00:19:41Z mbroeker $
|
|
3 |
* $URL: http://localhost/svn/c/VirtualReader/trunk/src/sentence.c $
|
|
4 |
*/
|
|
5 |
|
|
6 |
#include <sentence.h>
|
|
7 |
|
|
8 |
int words (char *ptr)
|
|
9 |
{
|
|
10 |
char *buffer = NULL;
|
|
11 |
char *token;
|
|
12 |
int count = 0;
|
|
13 |
|
|
14 |
if (ptr != NULL)
|
|
15 |
buffer = strdup (ptr);
|
|
16 |
|
|
17 |
if (buffer == NULL)
|
|
18 |
return -1;
|
|
19 |
|
|
20 |
token = strtok (buffer, " ");
|
|
21 |
while (token != NULL) {
|
|
22 |
token = strtok (NULL, " ");
|
|
23 |
count++;
|
|
24 |
}
|
|
25 |
|
|
26 |
if (buffer != NULL)
|
|
27 |
free (buffer);
|
|
28 |
|
|
29 |
return count;
|
|
30 |
}
|
|
31 |
|
|
32 |
int *parse (char *buf)
|
|
33 |
{
|
|
34 |
int count = 0;
|
|
35 |
int i;
|
|
36 |
int *s_ptr = NULL;
|
|
37 |
|
|
38 |
s_ptr = malloc (2 * sizeof (int));
|
|
39 |
s_ptr[count++] = 0;
|
|
40 |
|
|
41 |
for (i = 0; i < strlen (buf); i++)
|
|
42 |
switch (buf[i]) {
|
|
43 |
case '\n':
|
|
44 |
case '\t':
|
|
45 |
case '"':
|
|
46 |
case '\'':
|
|
47 |
buf[i] = ' ';
|
|
48 |
break;
|
|
49 |
case '.':
|
|
50 |
case '!':
|
|
51 |
case '?':
|
|
52 |
s_ptr = realloc (s_ptr, (count + 2) * sizeof (int));
|
|
53 |
if (s_ptr == NULL) {
|
|
54 |
printf ("Not enough memory\n");
|
|
55 |
return NULL;
|
|
56 |
}
|
|
57 |
s_ptr[count++] = ++i;
|
|
58 |
break;
|
|
59 |
default:
|
|
60 |
break;
|
|
61 |
}
|
|
62 |
|
|
63 |
s_ptr[count] = 0;
|
|
64 |
|
|
65 |
return s_ptr;
|
|
66 |
}
|
|
67 |
|
|
68 |
char *readbuffer (char *fname)
|
|
69 |
{
|
|
70 |
FILE *f;
|
|
71 |
char *text = NULL;
|
|
72 |
char buffer[201];
|
|
73 |
|
|
74 |
if ((f = fopen (fname, "r")) == NULL) {
|
|
75 |
fprintf (stderr, "ERROR: Cannot open %s for reading!\n", fname);
|
|
76 |
return NULL;
|
|
77 |
}
|
|
78 |
|
|
79 |
text = malloc (2);
|
|
80 |
*text = 0;
|
|
81 |
|
|
82 |
while (fgets (buffer, 200, f) != 0) {
|
|
83 |
if ((text = realloc (text, strlen (text) + strlen (buffer) + 1)) == NULL) {
|
|
84 |
printf ("Not enough memory\n");
|
|
85 |
return NULL;
|
|
86 |
};
|
|
87 |
strcat (text, buffer);
|
|
88 |
*buffer = 0;
|
|
89 |
}
|
|
90 |
|
|
91 |
fclose (f);
|
|
92 |
return text;
|
|
93 |
}
|
|
94 |
|
|
95 |
char *getSentence (char *text, int start, int end)
|
|
96 |
{
|
|
97 |
int i;
|
|
98 |
char *value = NULL;
|
|
99 |
|
|
100 |
if ((i = end - start) < 0)
|
|
101 |
return NULL;
|
|
102 |
|
|
103 |
value = malloc (i + 1);
|
|
104 |
if (value == NULL)
|
|
105 |
return NULL;
|
|
106 |
|
|
107 |
i = 1;
|
|
108 |
while (i) {
|
|
109 |
switch (text[start]) {
|
|
110 |
case '.':
|
|
111 |
case ',':
|
|
112 |
case '!':
|
|
113 |
case '?':
|
|
114 |
case '(':
|
|
115 |
case ')':
|
|
116 |
case '\n':
|
|
117 |
case ' ':
|
|
118 |
start++;
|
|
119 |
break;
|
|
120 |
default:
|
|
121 |
i = 0;
|
|
122 |
break;
|
|
123 |
}
|
|
124 |
}
|
|
125 |
|
|
126 |
for (i = start; i < end; i++) {
|
|
127 |
value[i - start] = text[i];
|
|
128 |
}
|
|
129 |
|
|
130 |
value[i - start] = 0;
|
|
131 |
|
|
132 |
return value;
|
|
133 |
}
|
|
134 |
|
|
135 |
char **getstrings (char *Text, int *value)
|
|
136 |
{
|
|
137 |
char *ptr;
|
|
138 |
int i = 0;
|
|
139 |
int count = 0;
|
|
140 |
char **sentences = NULL;
|
|
141 |
int *s_ptr;
|
|
142 |
|
|
143 |
if ((s_ptr = parse (Text)) == NULL)
|
|
144 |
return NULL;
|
|
145 |
|
|
146 |
while (s_ptr[i + 1] > 0) {
|
|
147 |
if ((ptr = getSentence (Text, s_ptr[i], s_ptr[i + 1])) == NULL)
|
|
148 |
return NULL;
|
|
149 |
if (strlen (ptr) > 0) {
|
|
150 |
if ((sentences = realloc (sentences, (count + 1) * sizeof (char *) + 1)) == NULL)
|
|
151 |
return NULL;
|
|
152 |
sentences[count++] = strdup (ptr);
|
|
153 |
}
|
|
154 |
if (ptr != NULL)
|
|
155 |
free (ptr);
|
|
156 |
i++;
|
|
157 |
}
|
|
158 |
|
|
159 |
if (s_ptr)
|
|
160 |
free (s_ptr);
|
|
161 |
|
|
162 |
*value = count;
|
|
163 |
return sentences;
|
|
164 |
}
|