Mercurial > mplayer.hg
annotate subreader.c @ 2836:ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
| author | atmos4 |
|---|---|
| date | Sun, 11 Nov 2001 15:41:17 +0000 |
| parents | d767086efcda |
| children | d6343a243515 |
| 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 |
| 2177 | 29 |
| 2178 | 30 static float mpsub_position=0; |
| 2177 | 31 |
| 258 | 32 int sub_uses_time=0; |
| 33 int sub_errs=0; | |
| 624 | 34 int sub_num=0; // number of subtitle structs |
| 35 int sub_format=-1; // 0 for microdvd | |
| 36 // 1 for SubRip | |
| 921 | 37 // 2 for SubViewer |
| 624 | 38 // 3 for SAMI (smi) |
| 818 | 39 // 4 for vplayer format |
| 850 | 40 // 5 for RT format |
| 921 | 41 // 6 for ssa (Sub Station Alpha) |
| 1081 | 42 // 7 for ... erm ... dunnowhat. tell me if you know |
| 2177 | 43 // 8 for the glorious MPsub |
| 2343 | 44 // 9 for AQTitle |
| 624 | 45 |
| 46 int eol(char p) { | |
| 47 return (p=='\r' || p=='\n' || p=='\0'); | |
| 48 } | |
| 49 | |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
50 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
|
51 int i; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
52 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
|
53 i = strlen(s) - 1; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
54 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
|
55 } |
| 624 | 56 |
| 2343 | 57 |
| 624 | 58 subtitle *sub_read_line_sami(FILE *fd, subtitle *current) { |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
59 static char line[1001]; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
60 static char *s = NULL; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
61 char text[1000], *p, *q; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
62 int state; |
| 624 | 63 |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
64 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
|
65 state = 0; |
| 624 | 66 |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
67 /* read the first line */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
68 if (!s) |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
69 if (!(s = fgets(line, 1000, fd))) return 0; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
70 |
| 624 | 71 do { |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
72 switch (state) { |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
73 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
74 case 0: /* find "START=" */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
75 s = strstr (s, "Start="); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
76 if (s) { |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
77 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
|
78 state = 1; continue; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
79 } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
80 break; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
81 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
82 case 1: /* find "<P" */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
83 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
|
84 break; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
85 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
86 case 2: /* find ">" */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
87 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
|
88 break; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
89 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
90 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
|
91 if (*s == '\0') { break; } |
|
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
92 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
|
93 *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
|
94 if (text[0] != '\0') |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
95 current->text[current->lines++] = strdup (text); |
|
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
96 s += 4; |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
97 } |
|
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
98 else if (*s == '<') { state = 4; } |
|
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
99 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
|
100 else if (*s == '\t') { *p++ = ' '; s++; } |
|
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
101 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
|
102 else *p++ = *s++; |
|
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
103 |
|
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
104 /* skip duplicated space */ |
|
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
105 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
|
106 |
| 624 | 107 continue; |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
108 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
109 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
|
110 q = strstr (s, "Start="); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
111 if (q) { |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
112 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
|
113 *p = '\0'; trail_space (text); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
114 if (text[0] != '\0') |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
115 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
|
116 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
|
117 state = 0; continue; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
118 } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
119 s = strchr (s, '>'); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
120 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
|
121 break; |
| 624 | 122 } |
| 123 | |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
124 /* read next line */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
125 if (state != 99 && !(s = fgets (line, 1000, fd))) return 0; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
126 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
127 } while (state != 99); |
| 624 | 128 |
| 129 return current; | |
| 130 } | |
| 258 | 131 |
| 132 | |
| 133 char *sub_readtext(char *source, char **dest) { | |
| 134 int len=0; | |
| 932 | 135 char *p=source; |
| 258 | 136 |
| 932 | 137 while ( !eol(*p) && *p!= '|' ) { |
| 138 p++,len++; | |
| 139 } | |
| 258 | 140 |
| 141 *dest= (char *)malloc (len+1); | |
| 142 if (!dest) {return ERR;} | |
| 143 | |
| 144 strncpy(*dest, source, len); | |
| 145 (*dest)[len]=0; | |
| 146 | |
| 147 while (*p=='\r' || *p=='\n' || *p=='|') p++; | |
| 148 | |
| 149 if (*p) return p; // not-last text field | |
| 150 else return NULL; // last text field | |
| 151 } | |
| 152 | |
| 153 subtitle *sub_read_line_microdvd(FILE *fd,subtitle *current) { | |
| 154 char line[1001]; | |
| 155 char line2[1001]; | |
| 156 char *p, *next; | |
| 157 int i; | |
| 158 | |
| 1764 | 159 bzero (current, sizeof(subtitle)); |
| 258 | 160 |
| 161 do { | |
| 162 if (!fgets (line, 1000, fd)) return NULL; | |
| 932 | 163 } while (sscanf (line, "{%ld}{%ld}%[^\r\n]", &(current->start), &(current->end),line2) <3); |
| 258 | 164 |
| 932 | 165 p=line2; |
| 258 | 166 |
| 167 next=p, i=0; | |
| 1081 | 168 while ((next =sub_readtext (next, &(current->text[i])))) { |
| 270 | 169 if (current->text[i]==ERR) {return ERR;} |
| 258 | 170 i++; |
| 1081 | 171 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
| 258 | 172 } |
| 932 | 173 current->lines= ++i; |
| 258 | 174 |
| 175 return current; | |
| 176 } | |
| 177 | |
| 178 subtitle *sub_read_line_subrip(FILE *fd, subtitle *current) { | |
| 179 char line[1001]; | |
| 180 int a1,a2,a3,a4,b1,b2,b3,b4; | |
| 181 char *p=NULL, *q=NULL; | |
| 182 int len; | |
| 183 | |
| 1764 | 184 bzero (current, sizeof(subtitle)); |
| 258 | 185 |
| 1764 | 186 while (1) { |
| 258 | 187 if (!fgets (line, 1000, fd)) return NULL; |
| 269 | 188 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue; |
| 258 | 189 current->start = a1*360000+a2*6000+a3*100+a4; |
| 190 current->end = b1*360000+b2*6000+b3*100+b4; | |
| 191 | |
| 192 if (!fgets (line, 1000, fd)) return NULL; | |
| 193 | |
| 194 p=q=line; | |
| 195 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) { | |
| 196 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && strncmp(p,"[br]",4); p++,len++); | |
| 197 current->text[current->lines-1]=(char *)malloc (len+1); | |
| 198 if (!current->text[current->lines-1]) return ERR; | |
| 199 strncpy (current->text[current->lines-1], q, len); | |
| 270 | 200 current->text[current->lines-1][len]='\0'; |
| 258 | 201 if (!*p || *p=='\r' || *p=='\n') break; |
| 202 while (*p++!=']'); | |
| 203 } | |
| 1764 | 204 break; |
| 258 | 205 } |
| 206 return current; | |
| 207 } | |
| 208 | |
| 209 subtitle *sub_read_line_third(FILE *fd,subtitle *current) { | |
| 210 char line[1001]; | |
| 211 int a1,a2,a3,a4,b1,b2,b3,b4; | |
| 212 char *p=NULL; | |
| 213 int i,len; | |
| 214 | |
| 1764 | 215 bzero (current, sizeof(subtitle)); |
| 258 | 216 |
| 217 while (!current->text[0]) { | |
| 218 if (!fgets (line, 1000, fd)) return NULL; | |
| 269 | 219 if ((len=sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) |
| 258 | 220 continue; |
| 221 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
| 222 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
| 223 for (i=0; i<SUB_MAX_TEXT;) { | |
| 269 | 224 if (!fgets (line, 1000, fd)) break; |
| 258 | 225 len=0; |
| 226 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++); | |
| 227 if (len) { | |
| 228 current->text[i]=(char *)malloc (len+1); | |
| 229 if (!current->text[i]) return ERR; | |
| 270 | 230 strncpy (current->text[i], line, len); current->text[i][len]='\0'; |
| 258 | 231 i++; |
| 232 } else { | |
| 233 break; | |
| 234 } | |
| 235 } | |
| 236 current->lines=i; | |
| 237 } | |
| 238 return current; | |
| 239 } | |
| 240 | |
| 818 | 241 subtitle *sub_read_line_vplayer(FILE *fd,subtitle *current) { |
| 242 char line[1001]; | |
| 243 char line2[1001]; | |
| 244 int a1,a2,a3,b1,b2,b3; | |
| 1081 | 245 char *p=NULL, *next; |
| 858 | 246 int i,len,len2,plen; |
| 818 | 247 |
| 1764 | 248 bzero (current, sizeof(subtitle)); |
| 818 | 249 |
| 250 while (!current->text[0]) { | |
| 251 if (!fgets (line, 1000, fd)) return NULL; | |
| 858 | 252 if ((len=sscanf (line, "%d:%d:%d:%n",&a1,&a2,&a3,&plen)) < 3) |
| 818 | 253 continue; |
| 254 if (!fgets (line2, 1000, fd)) return NULL; | |
| 255 if ((len2=sscanf (line2, "%d:%d:%d:",&b1,&b2,&b3)) < 3) | |
| 256 continue; | |
| 257 // przewiñ o linijkê do ty³u: | |
| 258 fseek(fd,-strlen(line2),SEEK_CUR); | |
| 259 | |
| 260 current->start = a1*360000+a2*6000+a3*100; | |
| 261 current->end = b1*360000+b2*6000+b3*100; | |
|
896
d46de26aef48
there is another format that get detected as vplayers.
eyck
parents:
892
diff
changeset
|
262 if ((current->end - current->start) > 1000) {current->end = current->start + 1000;} // not too long though. |
| 818 | 263 // teraz czas na wkopiowanie stringu |
|
1640
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
264 p=line; |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
265 // finds the body of the subtitle |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
266 for (i=0; i<3; i++){ |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
267 p=strchr(p,':')+1; |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
268 } |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
269 i=0; |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
270 |
| 818 | 271 if (*p!='|') { |
| 272 // | |
| 273 next = p,i=0; | |
| 274 while ((next =sub_readtext (next, &(current->text[i])))) { | |
| 275 if (current->text[i]==ERR) {return ERR;} | |
| 276 i++; | |
| 1081 | 277 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
| 818 | 278 } |
| 279 current->lines=i+1; | |
| 280 } | |
| 281 } | |
| 282 return current; | |
| 283 } | |
| 284 | |
| 850 | 285 subtitle *sub_read_line_rt(FILE *fd,subtitle *current) { |
| 286 //TODO: This format uses quite rich (sub/super)set of xhtml | |
| 287 // I couldn't check it since DTD is not included. | |
| 288 // WARNING: full XML parses can be required for proper parsing | |
| 289 char line[1001]; | |
| 290 int a1,a2,a3,a4,b1,b2,b3,b4; | |
| 291 char *p=NULL,*next=NULL; | |
| 292 int i,len,plen; | |
| 293 | |
| 1764 | 294 bzero (current, sizeof(subtitle)); |
| 850 | 295 |
| 296 while (!current->text[0]) { | |
| 297 if (!fgets (line, 1000, fd)) return NULL; | |
| 298 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0 | |
| 299 //to describe the same moment in time. Maybe there are even more formats in use. | |
| 300 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) | |
| 301 plen=a1=a2=a3=a4=b1=b2=b3=b4=0; | |
| 302 if ( | |
| 303 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) && | |
| 304 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) && | |
| 305 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) && | |
| 306 ((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) && | |
| 307 ((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) | |
| 308 ) | |
| 309 continue; | |
| 310 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
| 311 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
| 312 p=line; p+=plen;i=0; | |
| 313 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml? | |
| 314 next = strstr(line,"<clear/>")+8;i=0; | |
| 315 while ((next =sub_readtext (next, &(current->text[i])))) { | |
| 316 if (current->text[i]==ERR) {return ERR;} | |
| 317 i++; | |
| 1081 | 318 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
| 850 | 319 } |
| 320 current->lines=i+1; | |
| 321 } | |
| 322 return current; | |
| 323 } | |
| 324 | |
| 921 | 325 subtitle *sub_read_line_ssa(FILE *fd,subtitle *current) { |
| 326 int hour1, min1, sec1, hunsec1, | |
| 327 hour2, min2, sec2, hunsec2, nothing; | |
| 2141 | 328 int num; |
| 921 | 329 |
| 330 char line[1000], | |
| 2140 | 331 line3[1000], |
| 332 *line2; | |
| 2141 | 333 char *tmp; |
| 334 | |
| 921 | 335 do { |
| 336 if (!fgets (line, 1000, fd)) return NULL; | |
| 337 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d," | |
| 2140 | 338 "%[^\n\r]", ¬hing, |
| 339 &hour1, &min1, &sec1, &hunsec1, | |
| 340 &hour2, &min2, &sec2, &hunsec2, | |
| 341 line3) < 9); | |
| 342 line2=strstr(line3,",,"); | |
| 343 if (!line2) return NULL; | |
| 344 line2 ++; | |
| 345 line2 ++; | |
| 346 | |
| 2141 | 347 current->lines=1;num=0; |
| 921 | 348 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1; |
| 349 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2; | |
| 2141 | 350 |
| 351 while (tmp=strstr(line2, "\\n")) { | |
| 352 current->text[num]=(char *)malloc(tmp-line2+1); | |
| 353 strncpy (current->text[num], line2, tmp-line2); | |
| 354 current->text[num][tmp-line2]='\0'; | |
| 355 line2=tmp+2; | |
| 356 num++; | |
| 357 current->lines++; | |
| 358 if (current->lines >= SUB_MAX_TEXT) return current; | |
| 359 } | |
| 360 | |
| 361 | |
| 362 current->text[num]=(char *) malloc(strlen(line2)+1); | |
| 363 strcpy(current->text[num],line2); | |
| 818 | 364 |
| 921 | 365 return current; |
| 366 } | |
| 258 | 367 |
| 1081 | 368 subtitle *sub_read_line_dunnowhat(FILE *fd,subtitle *current) { |
| 369 char line[1001]; | |
| 370 char text[1001]; | |
| 371 | |
| 1764 | 372 bzero (current, sizeof(subtitle)); |
| 1081 | 373 |
| 374 if (!fgets (line, 1000, fd)) | |
| 375 return NULL; | |
| 376 if (sscanf (line, "%ld,%ld,\"%[^\"]", &(current->start), | |
| 377 &(current->end), text) <3) | |
| 378 return ERR; | |
| 379 current->text[0] = strdup(text); | |
| 380 current->lines = 1; | |
| 381 | |
| 382 return current; | |
| 383 } | |
| 384 | |
| 2177 | 385 subtitle *sub_read_line_mpsub(FILE *fd, subtitle *current) { |
| 386 char line[1000]; | |
| 2178 | 387 float a,b; |
| 388 int num=0; | |
| 2177 | 389 char *p, *q; |
| 390 | |
| 391 do | |
| 392 { | |
| 393 if (!fgets(line, 1000, fd)) return NULL; | |
| 2178 | 394 } while (sscanf (line, "%f %f", &a, &b) !=2); |
| 2177 | 395 |
| 2178 | 396 mpsub_position += (a*100.0); |
| 397 current->start=(int) mpsub_position; | |
| 398 mpsub_position += (b*100.0); | |
| 399 current->end=(int) mpsub_position; | |
| 2177 | 400 |
| 401 while (num < SUB_MAX_TEXT) { | |
| 402 if (!fgets (line, 1000, fd)) return NULL; | |
| 403 p=line; | |
| 404 while (isspace(*p)) p++; | |
| 405 if (eol(*p) && num > 0) return current; | |
| 406 if (eol(*p)) return NULL; | |
| 407 | |
| 408 for (q=p; !eol(*q); q++); | |
| 409 *q='\0'; | |
| 410 if (strlen(p)) { | |
| 411 current->text[num]=strdup(p); | |
| 2495 | 412 printf (">%s<\n",p); |
| 2177 | 413 current->lines = ++num; |
| 414 } else { | |
| 415 if (num) return current; | |
| 416 else return NULL; | |
| 417 } | |
| 418 } | |
| 419 } | |
| 420 | |
| 2343 | 421 subtitle *previous_aqt_sub = NULL; |
| 422 | |
| 423 subtitle *sub_read_line_aqt(FILE *fd,subtitle *current) { | |
| 424 char line[1001]; | |
| 425 | |
| 426 bzero (current, sizeof(subtitle)); | |
| 427 | |
| 428 while (1) { | |
| 429 // try to locate next subtitle | |
| 430 if (!fgets (line, 1000, fd)) | |
| 431 return NULL; | |
| 432 if (!(sscanf (line, "-->> %ld", &(current->start)) <1)) | |
| 433 break; | |
| 434 } | |
| 435 | |
| 436 if (previous_aqt_sub != NULL) | |
| 437 previous_aqt_sub->end = current->start-1; | |
| 438 | |
| 439 previous_aqt_sub = current; | |
| 440 | |
| 441 if (!fgets (line, 1000, fd)) | |
| 442 return NULL; | |
| 443 | |
| 2468 | 444 sub_readtext((char *) &line,¤t->text[0]); |
| 2343 | 445 current->lines = 1; |
| 446 current->end = current->start; // will be corrected by next subtitle | |
| 447 | |
| 448 if (!fgets (line, 1000, fd)) | |
| 449 return current;; | |
| 450 | |
| 2468 | 451 sub_readtext((char *) &line,¤t->text[1]); |
| 2343 | 452 current->lines = 2; |
| 453 | |
| 454 if ((current->text[0]=="") && (current->text[1]=="")) { | |
| 455 // void subtitle -> end of previous marked and exit | |
| 456 previous_aqt_sub = NULL; | |
| 457 return NULL; | |
| 458 } | |
| 459 | |
| 460 return current; | |
| 461 } | |
| 2177 | 462 |
| 258 | 463 int sub_autodetect (FILE *fd) { |
| 464 char line[1001]; | |
| 465 int i,j=0; | |
| 2177 | 466 char p; |
| 258 | 467 |
| 624 | 468 while (j < 100) { |
| 258 | 469 j++; |
| 470 if (!fgets (line, 1000, fd)) | |
| 471 return -1; | |
| 472 | |
| 624 | 473 if (sscanf (line, "{%d}{%d}", &i, &i)==2) |
| 258 | 474 {sub_uses_time=0;return 0;} |
| 269 | 475 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
| 258 | 476 {sub_uses_time=1;return 1;} |
| 269 | 477 if (sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
| 258 | 478 {sub_uses_time=1;return 2;} |
| 624 | 479 if (strstr (line, "<SAMI>")) |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
480 {sub_uses_time=1; return 3;} |
| 818 | 481 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3) |
| 482 {sub_uses_time=1;return 4;} | |
| 850 | 483 //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
|
484 // too weak test for RT |
|
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
485 // Please someone who knows the format of RT... FIX IT!!! |
| 921 | 486 // 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
|
487 if ( *line == '<' ) |
| 850 | 488 {sub_uses_time=1;return 5;} |
| 921 | 489 |
| 490 if (!memcmp(line, "Dialogue: Marked", 16)) | |
| 491 {sub_uses_time=1; return 6;} | |
| 1081 | 492 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3) |
| 493 {sub_uses_time=0;return 7;} | |
| 2177 | 494 if (sscanf (line, "FORMAT=%d", &i) == 1) |
| 495 {sub_uses_time=0; return 8;} | |
| 496 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E') | |
| 497 {sub_uses_time=1; return 8;} | |
| 2343 | 498 if (strstr (line, "-->>")) |
| 499 {sub_uses_time=0; return 9;} | |
| 258 | 500 } |
| 624 | 501 |
| 502 return -1; // too many bad lines | |
| 258 | 503 } |
|
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
504 |
|
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
505 #ifdef DUMPSUBS |
|
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
506 int sub_utf8=0; |
|
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
507 #else |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
508 extern int sub_utf8; |
|
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
509 #endif |
| 258 | 510 |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
511 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
512 static iconv_t icdsc; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
513 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
514 void subcp_open (void) |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
515 { |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
516 char *tocp = "UTF-8"; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
517 icdsc = (iconv_t)(-1); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
518 if (sub_cp){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
519 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
|
520 printf ("SUB: opened iconv descriptor.\n"); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
521 sub_utf8 = 2; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
522 } else |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
523 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
|
524 } |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
525 } |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
526 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
527 void subcp_close (void) |
|
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 if (icdsc != (iconv_t)(-1)){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
530 (void) iconv_close (icdsc); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
531 printf ("SUB: closed iconv descriptor.\n"); |
|
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 } |
|
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 #define ICBUFFSIZE 512 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
536 static char icbuffer[ICBUFFSIZE]; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
537 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
538 subtitle* subcp_recode (subtitle *sub) |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
539 { |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
540 int l=sub->lines; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
541 size_t ileft, oleft, otlen; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
542 char *op, *ip, *ot; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
543 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
544 while (l){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
545 op = icbuffer; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
546 ip = sub->text[--l]; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
547 ileft = strlen(ip); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
548 oleft = ICBUFFSIZE - 1; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
549 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
550 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
|
551 &op, &oleft) == (size_t)(-1)) { |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
552 printf ("SUB: error recoding line.\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 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
|
557 printf ("SUB: error allocating mem.\n"); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
558 l++; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
559 break; |
|
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 *op='\0' ; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
562 strcpy (ot, icbuffer); |
|
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 sub->text[l] = ot; |
|
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 if (l){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
567 for (l = sub->lines; l;) |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
568 free (sub->text[--l]); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
569 return ERR; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
570 } |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
571 return sub; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
572 } |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
573 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
574 #endif |
| 258 | 575 |
| 576 subtitle* sub_read_file (char *filename) { | |
| 577 FILE *fd; | |
| 578 int n_max; | |
| 579 subtitle *first; | |
| 1081 | 580 subtitle * (*func[])(FILE *fd,subtitle *dest)= |
| 258 | 581 { |
| 582 sub_read_line_microdvd, | |
| 583 sub_read_line_subrip, | |
| 624 | 584 sub_read_line_third, |
| 818 | 585 sub_read_line_sami, |
| 850 | 586 sub_read_line_vplayer, |
| 921 | 587 sub_read_line_rt, |
| 1081 | 588 sub_read_line_ssa, |
| 2177 | 589 sub_read_line_dunnowhat, |
| 2343 | 590 sub_read_line_mpsub, |
| 591 sub_read_line_aqt | |
| 592 | |
| 258 | 593 }; |
| 594 | |
| 595 fd=fopen (filename, "r"); if (!fd) return NULL; | |
| 596 | |
| 597 sub_format=sub_autodetect (fd); | |
| 598 if (sub_format==-1) {printf ("SUB: Could not determine file format\n");return NULL;} | |
| 624 | 599 printf ("SUB: Detected subtitle file format: %d\n",sub_format); |
| 258 | 600 |
| 601 rewind (fd); | |
| 602 | |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
603 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
604 subcp_open(); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
605 #endif |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
606 |
| 258 | 607 sub_num=0;n_max=32; |
| 608 first=(subtitle *)malloc(n_max*sizeof(subtitle)); | |
| 609 if(!first) return NULL; | |
| 610 | |
| 611 while(1){ | |
| 612 subtitle *sub; | |
| 613 if(sub_num>=n_max){ | |
| 614 n_max+=16; | |
| 615 first=realloc(first,n_max*sizeof(subtitle)); | |
| 616 } | |
| 617 sub=func[sub_format](fd,&first[sub_num]); | |
| 618 if(!sub) break; // EOF | |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
619 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
620 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
|
621 #endif |
| 258 | 622 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid |
| 623 } | |
| 624 | |
| 625 fclose(fd); | |
| 626 | |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
627 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
628 subcp_close(); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
629 #endif |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
630 |
| 258 | 631 // printf ("SUB: Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); |
| 269 | 632 printf ("SUB: Read %i subtitles", sub_num); |
| 624 | 633 if (sub_errs) printf (", %i bad line(s).\n", sub_errs); |
| 269 | 634 else printf (".\n"); |
| 258 | 635 |
| 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 |
