Mercurial > mplayer.hg
annotate subreader.c @ 2978:6be2873a5a91
banner + version printed
author | arpi |
---|---|
date | Sun, 18 Nov 2001 19:12:25 +0000 |
parents | 80954dfeabed |
children | 0cf593b6bab0 |
rev | line source |
---|---|
258 | 1 /* |
2 * Subtitle reader with format autodetection | |
3 * | |
4 * Written by laaz | |
5 * Some code cleanup & realloc() by A'rpi/ESP-team | |
1081 | 6 * dunnowhat sub format by szabi |
258 | 7 */ |
8 | |
9 | |
10 #include <stdio.h> | |
11 #include <stdlib.h> | |
12 #include <string.h> | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
13 #include <ctype.h> |
258 | 14 |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
15 #include "config.h" |
258 | 16 #include "subreader.h" |
17 | |
18 #define ERR (void *)-1 | |
19 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
20 #ifdef USE_ICONV |
2358 | 21 #ifdef __FreeBSD__ |
22 #include <giconv.h> | |
23 #else | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
24 #include <iconv.h> |
2358 | 25 #endif |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
26 char *sub_cp=NULL; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
27 #endif |
258 | 28 |
2912 | 29 /* Maximal length of line of a subtitle */ |
30 #define LINE_LEN 1000 | |
2177 | 31 |
2178 | 32 static float mpsub_position=0; |
2177 | 33 |
258 | 34 int sub_uses_time=0; |
35 int sub_errs=0; | |
624 | 36 int sub_num=0; // number of subtitle structs |
2912 | 37 |
38 /* Use the SUB_* constant defined in the header file */ | |
39 int sub_format=SUB_INVALID; | |
624 | 40 |
41 int eol(char p) { | |
42 return (p=='\r' || p=='\n' || p=='\0'); | |
43 } | |
44 | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
45 static inline void trail_space(char *s) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
46 int i; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
47 while (isspace(*s)) strcpy(s, s + 1); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
48 i = strlen(s) - 1; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
49 while (i > 0 && isspace(s[i])) s[i--] = '\0'; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
50 } |
624 | 51 |
2343 | 52 |
624 | 53 subtitle *sub_read_line_sami(FILE *fd, subtitle *current) { |
2912 | 54 static char line[LINE_LEN+1]; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
55 static char *s = NULL; |
2912 | 56 char text[LINE_LEN+1], *p, *q; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
57 int state; |
624 | 58 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
59 current->lines = current->start = current->end = 0; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
60 state = 0; |
624 | 61 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
62 /* read the first line */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
63 if (!s) |
2912 | 64 if (!(s = fgets(line, LINE_LEN, fd))) return 0; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
65 |
624 | 66 do { |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
67 switch (state) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
68 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
69 case 0: /* find "START=" */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
70 s = strstr (s, "Start="); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
71 if (s) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
72 current->start = strtol (s + 6, &s, 0) / 10; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
73 state = 1; continue; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
74 } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
75 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
76 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
77 case 1: /* find "<P" */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
78 if ((s = strstr (s, "<P"))) { s += 2; state = 2; continue; } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
79 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
80 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
81 case 2: /* find ">" */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
82 if ((s = strchr (s, '>'))) { s++; state = 3; p = text; continue; } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
83 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
84 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
85 case 3: /* get all text until '<' appears */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
86 if (*s == '\0') { break; } |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
87 else if (!strncasecmp (s, "<br>", 4)) { |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
88 *p = '\0'; p = text; trail_space (text); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
89 if (text[0] != '\0') |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
90 current->text[current->lines++] = strdup (text); |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
91 s += 4; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
92 } |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
93 else if (*s == '<') { state = 4; } |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
94 else if (!strncasecmp (s, " ", 6)) { *p++ = ' '; s += 6; } |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
95 else if (*s == '\t') { *p++ = ' '; s++; } |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
96 else if (*s == '\r' || *s == '\n') { s++; } |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
97 else *p++ = *s++; |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
98 |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
99 /* skip duplicated space */ |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
100 if (p > text + 2) if (*(p-1) == ' ' && *(p-2) == ' ') p--; |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
101 |
624 | 102 continue; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
103 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
104 case 4: /* get current->end or skip <TAG> */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
105 q = strstr (s, "Start="); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
106 if (q) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
107 current->end = strtol (q + 6, &q, 0) / 10 - 1; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
108 *p = '\0'; trail_space (text); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
109 if (text[0] != '\0') |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
110 current->text[current->lines++] = strdup (text); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
111 if (current->lines > 0) { state = 99; break; } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
112 state = 0; continue; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
113 } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
114 s = strchr (s, '>'); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
115 if (s) { s++; state = 3; continue; } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
116 break; |
624 | 117 } |
118 | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
119 /* read next line */ |
2912 | 120 if (state != 99 && !(s = fgets (line, LINE_LEN, fd))) return 0; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
121 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
122 } while (state != 99); |
624 | 123 |
124 return current; | |
125 } | |
258 | 126 |
127 | |
128 char *sub_readtext(char *source, char **dest) { | |
129 int len=0; | |
932 | 130 char *p=source; |
258 | 131 |
932 | 132 while ( !eol(*p) && *p!= '|' ) { |
133 p++,len++; | |
134 } | |
258 | 135 |
136 *dest= (char *)malloc (len+1); | |
137 if (!dest) {return ERR;} | |
138 | |
139 strncpy(*dest, source, len); | |
140 (*dest)[len]=0; | |
141 | |
142 while (*p=='\r' || *p=='\n' || *p=='|') p++; | |
143 | |
144 if (*p) return p; // not-last text field | |
145 else return NULL; // last text field | |
146 } | |
147 | |
148 subtitle *sub_read_line_microdvd(FILE *fd,subtitle *current) { | |
2912 | 149 char line[LINE_LEN+1]; |
150 char line2[LINE_LEN+1]; | |
258 | 151 char *p, *next; |
152 int i; | |
153 | |
1764 | 154 bzero (current, sizeof(subtitle)); |
258 | 155 |
156 do { | |
2912 | 157 if (!fgets (line, LINE_LEN, fd)) return NULL; |
932 | 158 } while (sscanf (line, "{%ld}{%ld}%[^\r\n]", &(current->start), &(current->end),line2) <3); |
258 | 159 |
932 | 160 p=line2; |
258 | 161 |
162 next=p, i=0; | |
1081 | 163 while ((next =sub_readtext (next, &(current->text[i])))) { |
270 | 164 if (current->text[i]==ERR) {return ERR;} |
258 | 165 i++; |
1081 | 166 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
258 | 167 } |
932 | 168 current->lines= ++i; |
258 | 169 |
170 return current; | |
171 } | |
172 | |
173 subtitle *sub_read_line_subrip(FILE *fd, subtitle *current) { | |
2912 | 174 char line[LINE_LEN+1]; |
258 | 175 int a1,a2,a3,a4,b1,b2,b3,b4; |
176 char *p=NULL, *q=NULL; | |
177 int len; | |
178 | |
1764 | 179 bzero (current, sizeof(subtitle)); |
258 | 180 |
1764 | 181 while (1) { |
2912 | 182 if (!fgets (line, LINE_LEN, fd)) return NULL; |
269 | 183 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue; |
258 | 184 current->start = a1*360000+a2*6000+a3*100+a4; |
185 current->end = b1*360000+b2*6000+b3*100+b4; | |
186 | |
2912 | 187 if (!fgets (line, LINE_LEN, fd)) return NULL; |
258 | 188 |
189 p=q=line; | |
190 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) { | |
191 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && strncmp(p,"[br]",4); p++,len++); | |
192 current->text[current->lines-1]=(char *)malloc (len+1); | |
193 if (!current->text[current->lines-1]) return ERR; | |
194 strncpy (current->text[current->lines-1], q, len); | |
270 | 195 current->text[current->lines-1][len]='\0'; |
258 | 196 if (!*p || *p=='\r' || *p=='\n') break; |
197 while (*p++!=']'); | |
198 } | |
1764 | 199 break; |
258 | 200 } |
201 return current; | |
202 } | |
203 | |
2912 | 204 subtitle *sub_read_line_subviewer(FILE *fd,subtitle *current) { |
205 char line[LINE_LEN+1]; | |
258 | 206 int a1,a2,a3,a4,b1,b2,b3,b4; |
207 char *p=NULL; | |
208 int i,len; | |
209 | |
1764 | 210 bzero (current, sizeof(subtitle)); |
258 | 211 |
212 while (!current->text[0]) { | |
2912 | 213 if (!fgets (line, LINE_LEN, fd)) return NULL; |
269 | 214 if ((len=sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) |
258 | 215 continue; |
216 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
217 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
218 for (i=0; i<SUB_MAX_TEXT;) { | |
2912 | 219 if (!fgets (line, LINE_LEN, fd)) break; |
258 | 220 len=0; |
221 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++); | |
222 if (len) { | |
223 current->text[i]=(char *)malloc (len+1); | |
224 if (!current->text[i]) return ERR; | |
270 | 225 strncpy (current->text[i], line, len); current->text[i][len]='\0'; |
258 | 226 i++; |
227 } else { | |
228 break; | |
229 } | |
230 } | |
231 current->lines=i; | |
232 } | |
233 return current; | |
234 } | |
235 | |
818 | 236 subtitle *sub_read_line_vplayer(FILE *fd,subtitle *current) { |
2912 | 237 char line[LINE_LEN+1]; |
238 char line2[LINE_LEN+1]; | |
818 | 239 int a1,a2,a3,b1,b2,b3; |
1081 | 240 char *p=NULL, *next; |
858 | 241 int i,len,len2,plen; |
818 | 242 |
1764 | 243 bzero (current, sizeof(subtitle)); |
818 | 244 |
245 while (!current->text[0]) { | |
2912 | 246 if (!fgets (line, LINE_LEN, fd)) return NULL; |
858 | 247 if ((len=sscanf (line, "%d:%d:%d:%n",&a1,&a2,&a3,&plen)) < 3) |
818 | 248 continue; |
2912 | 249 if (!fgets (line2, LINE_LEN, fd)) return NULL; |
818 | 250 if ((len2=sscanf (line2, "%d:%d:%d:",&b1,&b2,&b3)) < 3) |
251 continue; | |
252 // przewiñ o linijkê do ty³u: | |
253 fseek(fd,-strlen(line2),SEEK_CUR); | |
254 | |
255 current->start = a1*360000+a2*6000+a3*100; | |
256 current->end = b1*360000+b2*6000+b3*100; | |
896
d46de26aef48
there is another format that get detected as vplayers.
eyck
parents:
892
diff
changeset
|
257 if ((current->end - current->start) > 1000) {current->end = current->start + 1000;} // not too long though. |
818 | 258 // teraz czas na wkopiowanie stringu |
1640
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
259 p=line; |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
260 // finds the body of the subtitle |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
261 for (i=0; i<3; i++){ |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
262 p=strchr(p,':')+1; |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
263 } |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
264 i=0; |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
265 |
818 | 266 if (*p!='|') { |
267 // | |
268 next = p,i=0; | |
269 while ((next =sub_readtext (next, &(current->text[i])))) { | |
270 if (current->text[i]==ERR) {return ERR;} | |
271 i++; | |
1081 | 272 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
818 | 273 } |
274 current->lines=i+1; | |
275 } | |
276 } | |
277 return current; | |
278 } | |
279 | |
850 | 280 subtitle *sub_read_line_rt(FILE *fd,subtitle *current) { |
281 //TODO: This format uses quite rich (sub/super)set of xhtml | |
282 // I couldn't check it since DTD is not included. | |
283 // WARNING: full XML parses can be required for proper parsing | |
2912 | 284 char line[LINE_LEN+1]; |
850 | 285 int a1,a2,a3,a4,b1,b2,b3,b4; |
286 char *p=NULL,*next=NULL; | |
287 int i,len,plen; | |
288 | |
1764 | 289 bzero (current, sizeof(subtitle)); |
850 | 290 |
291 while (!current->text[0]) { | |
2912 | 292 if (!fgets (line, LINE_LEN, fd)) return NULL; |
850 | 293 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0 |
294 //to describe the same moment in time. Maybe there are even more formats in use. | |
295 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) | |
296 plen=a1=a2=a3=a4=b1=b2=b3=b4=0; | |
297 if ( | |
298 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) && | |
299 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) && | |
300 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) && | |
301 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&b4,&plen)) < 6) && | |
302 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\" %*[Ee]nd=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4,&plen)) < 8) | |
303 ) | |
304 continue; | |
305 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
306 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
307 p=line; p+=plen;i=0; | |
308 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml? | |
309 next = strstr(line,"<clear/>")+8;i=0; | |
310 while ((next =sub_readtext (next, &(current->text[i])))) { | |
311 if (current->text[i]==ERR) {return ERR;} | |
312 i++; | |
1081 | 313 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
850 | 314 } |
315 current->lines=i+1; | |
316 } | |
317 return current; | |
318 } | |
319 | |
921 | 320 subtitle *sub_read_line_ssa(FILE *fd,subtitle *current) { |
321 int hour1, min1, sec1, hunsec1, | |
322 hour2, min2, sec2, hunsec2, nothing; | |
2141 | 323 int num; |
921 | 324 |
2912 | 325 char line[LINE_LEN+1], |
326 line3[LINE_LEN+1], | |
2140 | 327 *line2; |
2141 | 328 char *tmp; |
329 | |
921 | 330 do { |
2912 | 331 if (!fgets (line, LINE_LEN, fd)) return NULL; |
921 | 332 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d," |
2140 | 333 "%[^\n\r]", ¬hing, |
334 &hour1, &min1, &sec1, &hunsec1, | |
335 &hour2, &min2, &sec2, &hunsec2, | |
336 line3) < 9); | |
337 line2=strstr(line3,",,"); | |
338 if (!line2) return NULL; | |
339 line2 ++; | |
340 line2 ++; | |
341 | |
2141 | 342 current->lines=1;num=0; |
921 | 343 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1; |
344 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2; | |
2141 | 345 |
346 while (tmp=strstr(line2, "\\n")) { | |
347 current->text[num]=(char *)malloc(tmp-line2+1); | |
348 strncpy (current->text[num], line2, tmp-line2); | |
349 current->text[num][tmp-line2]='\0'; | |
350 line2=tmp+2; | |
351 num++; | |
352 current->lines++; | |
353 if (current->lines >= SUB_MAX_TEXT) return current; | |
354 } | |
355 | |
356 | |
357 current->text[num]=(char *) malloc(strlen(line2)+1); | |
358 strcpy(current->text[num],line2); | |
818 | 359 |
921 | 360 return current; |
361 } | |
258 | 362 |
1081 | 363 subtitle *sub_read_line_dunnowhat(FILE *fd,subtitle *current) { |
2912 | 364 char line[LINE_LEN+1]; |
365 char text[LINE_LEN+1]; | |
1081 | 366 |
1764 | 367 bzero (current, sizeof(subtitle)); |
1081 | 368 |
2912 | 369 if (!fgets (line, LINE_LEN, fd)) |
1081 | 370 return NULL; |
371 if (sscanf (line, "%ld,%ld,\"%[^\"]", &(current->start), | |
372 &(current->end), text) <3) | |
373 return ERR; | |
374 current->text[0] = strdup(text); | |
375 current->lines = 1; | |
376 | |
377 return current; | |
378 } | |
379 | |
2177 | 380 subtitle *sub_read_line_mpsub(FILE *fd, subtitle *current) { |
2912 | 381 char line[LINE_LEN+1]; |
2178 | 382 float a,b; |
383 int num=0; | |
2177 | 384 char *p, *q; |
385 | |
386 do | |
387 { | |
2912 | 388 if (!fgets(line, LINE_LEN, fd)) return NULL; |
2178 | 389 } while (sscanf (line, "%f %f", &a, &b) !=2); |
2177 | 390 |
2178 | 391 mpsub_position += (a*100.0); |
392 current->start=(int) mpsub_position; | |
393 mpsub_position += (b*100.0); | |
394 current->end=(int) mpsub_position; | |
2177 | 395 |
396 while (num < SUB_MAX_TEXT) { | |
2912 | 397 if (!fgets (line, LINE_LEN, fd)) return NULL; |
2177 | 398 p=line; |
399 while (isspace(*p)) p++; | |
400 if (eol(*p) && num > 0) return current; | |
401 if (eol(*p)) return NULL; | |
402 | |
403 for (q=p; !eol(*q); q++); | |
404 *q='\0'; | |
405 if (strlen(p)) { | |
406 current->text[num]=strdup(p); | |
2495 | 407 printf (">%s<\n",p); |
2177 | 408 current->lines = ++num; |
409 } else { | |
410 if (num) return current; | |
411 else return NULL; | |
412 } | |
413 } | |
414 } | |
415 | |
2343 | 416 subtitle *previous_aqt_sub = NULL; |
417 | |
418 subtitle *sub_read_line_aqt(FILE *fd,subtitle *current) { | |
2912 | 419 char line[LINE_LEN+1]; |
2343 | 420 |
421 bzero (current, sizeof(subtitle)); | |
422 | |
423 while (1) { | |
424 // try to locate next subtitle | |
2912 | 425 if (!fgets (line, LINE_LEN, fd)) |
2343 | 426 return NULL; |
427 if (!(sscanf (line, "-->> %ld", &(current->start)) <1)) | |
428 break; | |
429 } | |
430 | |
431 if (previous_aqt_sub != NULL) | |
432 previous_aqt_sub->end = current->start-1; | |
433 | |
434 previous_aqt_sub = current; | |
435 | |
2912 | 436 if (!fgets (line, LINE_LEN, fd)) |
2343 | 437 return NULL; |
438 | |
2468 | 439 sub_readtext((char *) &line,¤t->text[0]); |
2343 | 440 current->lines = 1; |
441 current->end = current->start; // will be corrected by next subtitle | |
442 | |
2912 | 443 if (!fgets (line, LINE_LEN, fd)) |
2343 | 444 return current;; |
445 | |
2468 | 446 sub_readtext((char *) &line,¤t->text[1]); |
2343 | 447 current->lines = 2; |
448 | |
449 if ((current->text[0]=="") && (current->text[1]=="")) { | |
450 // void subtitle -> end of previous marked and exit | |
451 previous_aqt_sub = NULL; | |
452 return NULL; | |
453 } | |
454 | |
455 return current; | |
456 } | |
2177 | 457 |
258 | 458 int sub_autodetect (FILE *fd) { |
2912 | 459 char line[LINE_LEN+1]; |
258 | 460 int i,j=0; |
2177 | 461 char p; |
258 | 462 |
624 | 463 while (j < 100) { |
258 | 464 j++; |
2912 | 465 if (!fgets (line, LINE_LEN, fd)) |
466 return SUB_INVALID; | |
258 | 467 |
624 | 468 if (sscanf (line, "{%d}{%d}", &i, &i)==2) |
2912 | 469 {sub_uses_time=0;return SUB_MICRODVD;} |
269 | 470 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
2912 | 471 {sub_uses_time=1;return SUB_SUBRIP;} |
269 | 472 if (sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
2912 | 473 {sub_uses_time=1;return SUB_SUBVIEWER;} |
624 | 474 if (strstr (line, "<SAMI>")) |
2912 | 475 {sub_uses_time=1; return SUB_SAMI;} |
818 | 476 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3) |
2912 | 477 {sub_uses_time=1;return SUB_VPLAYER;} |
850 | 478 //TODO: just checking if first line of sub starts with "<" is WAY |
913
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
479 // too weak test for RT |
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
480 // Please someone who knows the format of RT... FIX IT!!! |
921 | 481 // It may conflict with other sub formats in the future (actually it doesn't) |
913
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
482 if ( *line == '<' ) |
2912 | 483 {sub_uses_time=1;return SUB_RT;} |
921 | 484 |
485 if (!memcmp(line, "Dialogue: Marked", 16)) | |
2912 | 486 {sub_uses_time=1; return SUB_SSA;} |
1081 | 487 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3) |
2912 | 488 {sub_uses_time=0;return SUB_DUNNOWHAT;} |
2177 | 489 if (sscanf (line, "FORMAT=%d", &i) == 1) |
2912 | 490 {sub_uses_time=0; return SUB_MPSUB;} |
2177 | 491 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E') |
2912 | 492 {sub_uses_time=1; return SUB_MPSUB;} |
2343 | 493 if (strstr (line, "-->>")) |
2912 | 494 {sub_uses_time=0; return SUB_MPSUB;} |
258 | 495 } |
624 | 496 |
2912 | 497 return SUB_INVALID; // too many bad lines |
258 | 498 } |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
499 |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
500 #ifdef DUMPSUBS |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
501 int sub_utf8=0; |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
502 #else |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
503 extern int sub_utf8; |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
504 #endif |
258 | 505 |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
506 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
507 static iconv_t icdsc; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
508 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
509 void subcp_open (void) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
510 { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
511 char *tocp = "UTF-8"; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
512 icdsc = (iconv_t)(-1); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
513 if (sub_cp){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
514 if ((icdsc = iconv_open (tocp, sub_cp)) != (iconv_t)(-1)){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
515 printf ("SUB: opened iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
516 sub_utf8 = 2; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
517 } else |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
518 printf ("SUB: error opening iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
519 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
520 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
521 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
522 void subcp_close (void) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
523 { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
524 if (icdsc != (iconv_t)(-1)){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
525 (void) iconv_close (icdsc); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
526 printf ("SUB: closed iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
527 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
528 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
529 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
530 #define ICBUFFSIZE 512 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
531 static char icbuffer[ICBUFFSIZE]; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
532 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
533 subtitle* subcp_recode (subtitle *sub) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
534 { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
535 int l=sub->lines; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
536 size_t ileft, oleft, otlen; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
537 char *op, *ip, *ot; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
538 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
539 while (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
540 op = icbuffer; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
541 ip = sub->text[--l]; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
542 ileft = strlen(ip); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
543 oleft = ICBUFFSIZE - 1; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
544 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
545 if (iconv(icdsc, (const char **) &ip, &ileft, |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
546 &op, &oleft) == (size_t)(-1)) { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
547 printf ("SUB: error recoding line.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
548 l++; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
549 break; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
550 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
551 if (!(ot = (char *)malloc(op - icbuffer + 1))){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
552 printf ("SUB: error allocating mem.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
553 l++; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
554 break; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
555 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
556 *op='\0' ; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
557 strcpy (ot, icbuffer); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
558 free (sub->text[l]); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
559 sub->text[l] = ot; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
560 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
561 if (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
562 for (l = sub->lines; l;) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
563 free (sub->text[--l]); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
564 return ERR; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
565 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
566 return sub; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
567 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
568 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
569 #endif |
258 | 570 |
571 subtitle* sub_read_file (char *filename) { | |
572 FILE *fd; | |
573 int n_max; | |
574 subtitle *first; | |
1081 | 575 subtitle * (*func[])(FILE *fd,subtitle *dest)= |
258 | 576 { |
577 sub_read_line_microdvd, | |
578 sub_read_line_subrip, | |
2912 | 579 sub_read_line_subviewer, |
818 | 580 sub_read_line_sami, |
850 | 581 sub_read_line_vplayer, |
921 | 582 sub_read_line_rt, |
1081 | 583 sub_read_line_ssa, |
2177 | 584 sub_read_line_dunnowhat, |
2343 | 585 sub_read_line_mpsub, |
586 sub_read_line_aqt | |
587 | |
258 | 588 }; |
2915 | 589 if(filename==NULL) return NULL; //qnx segfault |
258 | 590 fd=fopen (filename, "r"); if (!fd) return NULL; |
591 | |
592 sub_format=sub_autodetect (fd); | |
2912 | 593 if (sub_format==SUB_INVALID) {printf ("SUB: Could not determine file format\n");return NULL;} |
624 | 594 printf ("SUB: Detected subtitle file format: %d\n",sub_format); |
258 | 595 |
596 rewind (fd); | |
597 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
598 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
599 subcp_open(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
600 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
601 |
258 | 602 sub_num=0;n_max=32; |
603 first=(subtitle *)malloc(n_max*sizeof(subtitle)); | |
604 if(!first) return NULL; | |
605 | |
606 while(1){ | |
607 subtitle *sub; | |
608 if(sub_num>=n_max){ | |
609 n_max+=16; | |
610 first=realloc(first,n_max*sizeof(subtitle)); | |
611 } | |
612 sub=func[sub_format](fd,&first[sub_num]); | |
613 if(!sub) break; // EOF | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
614 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
615 if ((sub!=ERR) && (sub_utf8 & 2)) sub=subcp_recode(sub); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
616 #endif |
258 | 617 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid |
618 } | |
619 | |
620 fclose(fd); | |
621 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
622 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
623 subcp_close(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
624 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
625 |
258 | 626 // printf ("SUB: Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); |
269 | 627 printf ("SUB: Read %i subtitles", sub_num); |
624 | 628 if (sub_errs) printf (", %i bad line(s).\n", sub_errs); |
269 | 629 else printf (".\n"); |
258 | 630 |
2880 | 631 if(sub_num<=0){ |
632 free(first); | |
633 return NULL; | |
634 } | |
635 | |
258 | 636 return first; |
637 } | |
638 | |
892 | 639 #if 0 |
509 | 640 char * strreplace( char * in,char * what,char * whereof ) |
641 { | |
642 int i; | |
643 char * tmp; | |
644 | |
645 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL; | |
646 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i]; | |
647 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0; | |
648 return in; | |
649 } | |
892 | 650 #endif |
509 | 651 |
892 | 652 char * sub_filename(char* path, char * fname ) |
509 | 653 { |
892 | 654 char * sub_name1; |
655 char * sub_name2; | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
656 char * aviptr1, * aviptr2, * tmp; |
892 | 657 int i,j; |
658 FILE * f; | |
659 int pos=0; | |
660 char * sub_exts[] = | |
1501
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
661 { ".utf", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
662 ".UTF", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
663 ".sub", |
509 | 664 ".SUB", |
665 ".srt", | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
666 ".SRT", |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
667 ".smi", |
850 | 668 ".SMI", |
669 ".rt", | |
670 ".RT", | |
671 ".txt", | |
1081 | 672 ".TXT", |
673 ".ssa", | |
2343 | 674 ".SSA", |
675 ".aqt", | |
676 ".AQT"}; | |
892 | 677 |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
678 |
509 | 679 if ( fname == NULL ) return NULL; |
892 | 680 |
681 sub_name1=strrchr(fname,'.'); | |
682 if (!sub_name1) return NULL; | |
683 pos=sub_name1-fname; | |
684 | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
685 sub_name1=malloc(strlen(fname)+8); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
686 strcpy(sub_name1,fname); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
687 |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
688 sub_name2=malloc (strlen(path) + strlen(fname) + 8); |
1081 | 689 if ((tmp=strrchr(fname,'/'))) |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
690 sprintf (sub_name2, "%s%s", path, tmp+1); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
691 else |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
692 sprintf (sub_name2, "%s%s", path, fname); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
693 |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
694 aviptr1=strrchr(sub_name1,'.'); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
695 aviptr2=strrchr(sub_name2,'.'); |
892 | 696 |
697 for(j=0;j<=1;j++){ | |
698 char* sub_name=j?sub_name1:sub_name2; | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
699 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
700 for ( i=(sub_cp?2:0);i<(sizeof(sub_exts)/sizeof(char*));i++ ) { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
701 #else |
892 | 702 for ( i=0;i<(sizeof(sub_exts)/sizeof(char*));i++ ) { |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
703 #endif |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
704 strcpy(j?aviptr1:aviptr2,sub_exts[i]); |
935 | 705 // printf("trying: '%s'\n",sub_name); |
892 | 706 if((f=fopen( sub_name,"rt" ))) { |
509 | 707 fclose( f ); |
708 printf( "SUB: Detected sub file: %s\n",sub_name ); | |
1501
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
709 if (i<2) sub_utf8=1; |
509 | 710 return sub_name; |
892 | 711 } |
509 | 712 } |
892 | 713 } |
714 | |
509 | 715 return NULL; |
716 } | |
717 | |
1761 | 718 void list_sub_file(subtitle* subs){ |
719 int i,j; | |
720 | |
721 for(j=0;j<sub_num;j++){ | |
722 subtitle* egysub=&subs[j]; | |
723 printf ("%i line%c (%li-%li) ", | |
724 egysub->lines, | |
725 (1==egysub->lines)?' ':'s', | |
726 egysub->start, | |
727 egysub->end); | |
728 for (i=0; i<egysub->lines; i++) { | |
729 printf ("%s%s",egysub->text[i], i==egysub->lines-1?"":" <BREAK> "); | |
730 } | |
731 printf ("\n"); | |
732 } | |
733 | |
734 printf ("Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); | |
735 printf ("Read %i subtitles, %i errors.\n", sub_num, sub_errs); | |
736 | |
737 } | |
738 | |
2178 | 739 void dump_mpsub(subtitle* subs){ |
740 int i,j; | |
741 FILE *fd; | |
742 float a,b; | |
743 | |
744 mpsub_position=0.0; | |
745 | |
746 fd=fopen ("dump.mpsub", "w"); | |
747 if (!fd) { | |
748 perror ("dump_mpsub: fopen"); | |
749 return; | |
750 } | |
751 | |
752 | |
753 if (sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n"); | |
2495 | 754 else fprintf (fd, "FORMAT=25\n\n"); // FIXME: fps |
2178 | 755 |
756 for(j=0;j<sub_num;j++){ | |
2495 | 757 subtitle* egysub=&subs[j]; |
758 if (sub_uses_time) { | |
759 a=((egysub->start-mpsub_position)/100.0); | |
760 b=((egysub->end-egysub->start)/100.0); | |
761 if ( (float)((int)a) == a) | |
762 fprintf (fd, "%.0f",a); | |
763 else | |
764 fprintf (fd, "%.2f",a); | |
765 | |
766 if ( (float)((int)b) == b) | |
767 fprintf (fd, " %.0f\n",b); | |
768 else | |
769 fprintf (fd, " %.2f\n",b); | |
770 } else { | |
771 fprintf (fd, "%ld %ld\n", egysub->start-mpsub_position, | |
772 egysub->end-egysub->start); | |
773 } | |
774 | |
775 mpsub_position = egysub->end; | |
776 for (i=0; i<egysub->lines; i++) { | |
777 fprintf (fd, "%s\n",egysub->text[i]); | |
778 } | |
779 fprintf (fd, "\n"); | |
2178 | 780 } |
781 fclose (fd); | |
782 printf ("Subtitles dumped in \'dump.mpsub\'.\n"); | |
783 } | |
784 | |
785 | |
786 | |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
787 #ifdef DUMPSUBS |
258 | 788 int main(int argc, char **argv) { // for testing |
789 | |
790 int i,j; | |
791 subtitle *subs; | |
792 subtitle *egysub; | |
793 | |
794 if(argc<2){ | |
795 printf("\nUsage: subreader filename.sub\n\n"); | |
796 exit(1); | |
797 } | |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
798 sub_cp = argv[2]; |
624 | 799 subs=sub_read_file(argv[1]); |
258 | 800 if(!subs){ |
801 printf("Couldn't load file... let's write a bugreport :)\n"); | |
802 exit(1); | |
803 } | |
1761 | 804 |
805 list_sub_file(subs); | |
258 | 806 |
807 return 0; | |
808 } | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
809 #endif |