Mercurial > mplayer.hg
annotate subreader.c @ 2343:da8dbcfb89d0
AQT type support patch by Jiri.Svoboda@seznam.cz
| author | arpi |
|---|---|
| date | Sun, 21 Oct 2001 17:19:59 +0000 |
| parents | e509abdbf195 |
| children | 8b971caf9e03 |
| 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 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
21 #include <iconv.h> |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
22 char *sub_cp=NULL; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
23 #endif |
| 258 | 24 |
| 2177 | 25 |
| 2178 | 26 static float mpsub_position=0; |
| 2177 | 27 |
| 258 | 28 int sub_uses_time=0; |
| 29 int sub_errs=0; | |
| 624 | 30 int sub_num=0; // number of subtitle structs |
| 31 int sub_format=-1; // 0 for microdvd | |
| 32 // 1 for SubRip | |
| 921 | 33 // 2 for SubViewer |
| 624 | 34 // 3 for SAMI (smi) |
| 818 | 35 // 4 for vplayer format |
| 850 | 36 // 5 for RT format |
| 921 | 37 // 6 for ssa (Sub Station Alpha) |
| 1081 | 38 // 7 for ... erm ... dunnowhat. tell me if you know |
| 2177 | 39 // 8 for the glorious MPsub |
| 2343 | 40 // 9 for AQTitle |
| 624 | 41 |
| 42 int eol(char p) { | |
| 43 return (p=='\r' || p=='\n' || p=='\0'); | |
| 44 } | |
| 45 | |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
46 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
|
47 int i; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
48 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
|
49 i = strlen(s) - 1; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
50 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
|
51 } |
| 624 | 52 |
| 2343 | 53 |
| 624 | 54 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
|
55 static char line[1001]; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
56 static char *s = NULL; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
57 char text[1000], *p, *q; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
58 int state; |
| 624 | 59 |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
60 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
|
61 state = 0; |
| 624 | 62 |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
63 /* read the first line */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
64 if (!s) |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
65 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
|
66 |
| 624 | 67 do { |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
68 switch (state) { |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
69 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
70 case 0: /* find "START=" */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
71 s = strstr (s, "Start="); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
72 if (s) { |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
73 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
|
74 state = 1; continue; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
75 } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
76 break; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
77 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
78 case 1: /* find "<P" */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
79 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
|
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 2: /* find ">" */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
83 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
|
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 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
|
87 if (*s == '\0') { break; } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
88 else if (*s == '<') { state = 4; } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
89 else if (!strncasecmp (s, " ", 6)) { *p++ = ' '; s += 6; } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
90 else if (*s == '\r') { s++; } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
91 else if (!strncasecmp (s, "<br>", 4) || *s == '\n') { |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
92 *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
|
93 if (text[0] != '\0') |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
94 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
|
95 if (*s == '\n') s++; else s += 4; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
96 } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
97 else *p++ = *s++; |
| 624 | 98 continue; |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
99 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
100 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
|
101 q = strstr (s, "Start="); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
102 if (q) { |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
103 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
|
104 *p = '\0'; trail_space (text); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
105 if (text[0] != '\0') |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
106 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
|
107 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
|
108 state = 0; continue; |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
109 } |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
110 s = strchr (s, '>'); |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
111 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
|
112 break; |
| 624 | 113 } |
| 114 | |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
115 /* read next line */ |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
116 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
|
117 |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
118 } while (state != 99); |
| 624 | 119 |
| 120 return current; | |
| 121 } | |
| 258 | 122 |
| 123 | |
| 124 char *sub_readtext(char *source, char **dest) { | |
| 125 int len=0; | |
| 932 | 126 char *p=source; |
| 258 | 127 |
| 932 | 128 while ( !eol(*p) && *p!= '|' ) { |
| 129 p++,len++; | |
| 130 } | |
| 258 | 131 |
| 132 *dest= (char *)malloc (len+1); | |
| 133 if (!dest) {return ERR;} | |
| 134 | |
| 135 strncpy(*dest, source, len); | |
| 136 (*dest)[len]=0; | |
| 137 | |
| 138 while (*p=='\r' || *p=='\n' || *p=='|') p++; | |
| 139 | |
| 140 if (*p) return p; // not-last text field | |
| 141 else return NULL; // last text field | |
| 142 } | |
| 143 | |
| 144 subtitle *sub_read_line_microdvd(FILE *fd,subtitle *current) { | |
| 145 char line[1001]; | |
| 146 char line2[1001]; | |
| 147 char *p, *next; | |
| 148 int i; | |
| 149 | |
| 1764 | 150 bzero (current, sizeof(subtitle)); |
| 258 | 151 |
| 152 do { | |
| 153 if (!fgets (line, 1000, fd)) return NULL; | |
| 932 | 154 } while (sscanf (line, "{%ld}{%ld}%[^\r\n]", &(current->start), &(current->end),line2) <3); |
| 258 | 155 |
| 932 | 156 p=line2; |
| 258 | 157 |
| 158 next=p, i=0; | |
| 1081 | 159 while ((next =sub_readtext (next, &(current->text[i])))) { |
| 270 | 160 if (current->text[i]==ERR) {return ERR;} |
| 258 | 161 i++; |
| 1081 | 162 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
| 258 | 163 } |
| 932 | 164 current->lines= ++i; |
| 258 | 165 |
| 166 return current; | |
| 167 } | |
| 168 | |
| 169 subtitle *sub_read_line_subrip(FILE *fd, subtitle *current) { | |
| 170 char line[1001]; | |
| 171 int a1,a2,a3,a4,b1,b2,b3,b4; | |
| 172 char *p=NULL, *q=NULL; | |
| 173 int len; | |
| 174 | |
| 1764 | 175 bzero (current, sizeof(subtitle)); |
| 258 | 176 |
| 1764 | 177 while (1) { |
| 258 | 178 if (!fgets (line, 1000, fd)) return NULL; |
| 269 | 179 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue; |
| 258 | 180 current->start = a1*360000+a2*6000+a3*100+a4; |
| 181 current->end = b1*360000+b2*6000+b3*100+b4; | |
| 182 | |
| 183 if (!fgets (line, 1000, fd)) return NULL; | |
| 184 | |
| 185 p=q=line; | |
| 186 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) { | |
| 187 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && strncmp(p,"[br]",4); p++,len++); | |
| 188 current->text[current->lines-1]=(char *)malloc (len+1); | |
| 189 if (!current->text[current->lines-1]) return ERR; | |
| 190 strncpy (current->text[current->lines-1], q, len); | |
| 270 | 191 current->text[current->lines-1][len]='\0'; |
| 258 | 192 if (!*p || *p=='\r' || *p=='\n') break; |
| 193 while (*p++!=']'); | |
| 194 } | |
| 1764 | 195 break; |
| 258 | 196 } |
| 197 return current; | |
| 198 } | |
| 199 | |
| 200 subtitle *sub_read_line_third(FILE *fd,subtitle *current) { | |
| 201 char line[1001]; | |
| 202 int a1,a2,a3,a4,b1,b2,b3,b4; | |
| 203 char *p=NULL; | |
| 204 int i,len; | |
| 205 | |
| 1764 | 206 bzero (current, sizeof(subtitle)); |
| 258 | 207 |
| 208 while (!current->text[0]) { | |
| 209 if (!fgets (line, 1000, fd)) return NULL; | |
| 269 | 210 if ((len=sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) |
| 258 | 211 continue; |
| 212 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
| 213 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
| 214 for (i=0; i<SUB_MAX_TEXT;) { | |
| 269 | 215 if (!fgets (line, 1000, fd)) break; |
| 258 | 216 len=0; |
| 217 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++); | |
| 218 if (len) { | |
| 219 current->text[i]=(char *)malloc (len+1); | |
| 220 if (!current->text[i]) return ERR; | |
| 270 | 221 strncpy (current->text[i], line, len); current->text[i][len]='\0'; |
| 258 | 222 i++; |
| 223 } else { | |
| 224 break; | |
| 225 } | |
| 226 } | |
| 227 current->lines=i; | |
| 228 } | |
| 229 return current; | |
| 230 } | |
| 231 | |
| 818 | 232 subtitle *sub_read_line_vplayer(FILE *fd,subtitle *current) { |
| 233 char line[1001]; | |
| 234 char line2[1001]; | |
| 235 int a1,a2,a3,b1,b2,b3; | |
| 1081 | 236 char *p=NULL, *next; |
| 858 | 237 int i,len,len2,plen; |
| 818 | 238 |
| 1764 | 239 bzero (current, sizeof(subtitle)); |
| 818 | 240 |
| 241 while (!current->text[0]) { | |
| 242 if (!fgets (line, 1000, fd)) return NULL; | |
| 858 | 243 if ((len=sscanf (line, "%d:%d:%d:%n",&a1,&a2,&a3,&plen)) < 3) |
| 818 | 244 continue; |
| 245 if (!fgets (line2, 1000, fd)) return NULL; | |
| 246 if ((len2=sscanf (line2, "%d:%d:%d:",&b1,&b2,&b3)) < 3) | |
| 247 continue; | |
| 248 // przewiñ o linijkê do ty³u: | |
| 249 fseek(fd,-strlen(line2),SEEK_CUR); | |
| 250 | |
| 251 current->start = a1*360000+a2*6000+a3*100; | |
| 252 current->end = b1*360000+b2*6000+b3*100; | |
|
896
d46de26aef48
there is another format that get detected as vplayers.
eyck
parents:
892
diff
changeset
|
253 if ((current->end - current->start) > 1000) {current->end = current->start + 1000;} // not too long though. |
| 818 | 254 // teraz czas na wkopiowanie stringu |
|
1640
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
255 p=line; |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
256 // finds the body of the subtitle |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
257 for (i=0; i<3; i++){ |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
258 p=strchr(p,':')+1; |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
259 } |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
260 i=0; |
|
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
261 |
| 818 | 262 if (*p!='|') { |
| 263 // | |
| 264 next = p,i=0; | |
| 265 while ((next =sub_readtext (next, &(current->text[i])))) { | |
| 266 if (current->text[i]==ERR) {return ERR;} | |
| 267 i++; | |
| 1081 | 268 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
| 818 | 269 } |
| 270 current->lines=i+1; | |
| 271 } | |
| 272 } | |
| 273 return current; | |
| 274 } | |
| 275 | |
| 850 | 276 subtitle *sub_read_line_rt(FILE *fd,subtitle *current) { |
| 277 //TODO: This format uses quite rich (sub/super)set of xhtml | |
| 278 // I couldn't check it since DTD is not included. | |
| 279 // WARNING: full XML parses can be required for proper parsing | |
| 280 char line[1001]; | |
| 281 int a1,a2,a3,a4,b1,b2,b3,b4; | |
| 282 char *p=NULL,*next=NULL; | |
| 283 int i,len,plen; | |
| 284 | |
| 1764 | 285 bzero (current, sizeof(subtitle)); |
| 850 | 286 |
| 287 while (!current->text[0]) { | |
| 288 if (!fgets (line, 1000, fd)) return NULL; | |
| 289 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0 | |
| 290 //to describe the same moment in time. Maybe there are even more formats in use. | |
| 291 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) | |
| 292 plen=a1=a2=a3=a4=b1=b2=b3=b4=0; | |
| 293 if ( | |
| 294 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) && | |
| 295 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) && | |
| 296 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) && | |
| 297 ((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) && | |
| 298 ((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) | |
| 299 ) | |
| 300 continue; | |
| 301 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
| 302 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
| 303 p=line; p+=plen;i=0; | |
| 304 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml? | |
| 305 next = strstr(line,"<clear/>")+8;i=0; | |
| 306 while ((next =sub_readtext (next, &(current->text[i])))) { | |
| 307 if (current->text[i]==ERR) {return ERR;} | |
| 308 i++; | |
| 1081 | 309 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
| 850 | 310 } |
| 311 current->lines=i+1; | |
| 312 } | |
| 313 return current; | |
| 314 } | |
| 315 | |
| 921 | 316 subtitle *sub_read_line_ssa(FILE *fd,subtitle *current) { |
| 317 int hour1, min1, sec1, hunsec1, | |
| 318 hour2, min2, sec2, hunsec2, nothing; | |
| 2141 | 319 int num; |
| 921 | 320 |
| 321 char line[1000], | |
| 2140 | 322 line3[1000], |
| 323 *line2; | |
| 2141 | 324 char *tmp; |
| 325 | |
| 921 | 326 do { |
| 327 if (!fgets (line, 1000, fd)) return NULL; | |
| 328 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d," | |
| 2140 | 329 "%[^\n\r]", ¬hing, |
| 330 &hour1, &min1, &sec1, &hunsec1, | |
| 331 &hour2, &min2, &sec2, &hunsec2, | |
| 332 line3) < 9); | |
| 333 line2=strstr(line3,",,"); | |
| 334 if (!line2) return NULL; | |
| 335 line2 ++; | |
| 336 line2 ++; | |
| 337 | |
| 2141 | 338 current->lines=1;num=0; |
| 921 | 339 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1; |
| 340 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2; | |
| 2141 | 341 |
| 342 while (tmp=strstr(line2, "\\n")) { | |
| 343 current->text[num]=(char *)malloc(tmp-line2+1); | |
| 344 strncpy (current->text[num], line2, tmp-line2); | |
| 345 current->text[num][tmp-line2]='\0'; | |
| 346 line2=tmp+2; | |
| 347 num++; | |
| 348 current->lines++; | |
| 349 if (current->lines >= SUB_MAX_TEXT) return current; | |
| 350 } | |
| 351 | |
| 352 | |
| 353 current->text[num]=(char *) malloc(strlen(line2)+1); | |
| 354 strcpy(current->text[num],line2); | |
| 818 | 355 |
| 921 | 356 return current; |
| 357 } | |
| 258 | 358 |
| 1081 | 359 subtitle *sub_read_line_dunnowhat(FILE *fd,subtitle *current) { |
| 360 char line[1001]; | |
| 361 char text[1001]; | |
| 362 | |
| 1764 | 363 bzero (current, sizeof(subtitle)); |
| 1081 | 364 |
| 365 if (!fgets (line, 1000, fd)) | |
| 366 return NULL; | |
| 367 if (sscanf (line, "%ld,%ld,\"%[^\"]", &(current->start), | |
| 368 &(current->end), text) <3) | |
| 369 return ERR; | |
| 370 current->text[0] = strdup(text); | |
| 371 current->lines = 1; | |
| 372 | |
| 373 return current; | |
| 374 } | |
| 375 | |
| 2177 | 376 subtitle *sub_read_line_mpsub(FILE *fd, subtitle *current) { |
| 377 char line[1000]; | |
| 2178 | 378 float a,b; |
| 379 int num=0; | |
| 2177 | 380 char *p, *q; |
| 381 | |
| 382 do | |
| 383 { | |
| 384 if (!fgets(line, 1000, fd)) return NULL; | |
| 2178 | 385 } while (sscanf (line, "%f %f", &a, &b) !=2); |
| 2177 | 386 |
| 2178 | 387 mpsub_position += (a*100.0); |
| 388 current->start=(int) mpsub_position; | |
| 389 mpsub_position += (b*100.0); | |
| 390 current->end=(int) mpsub_position; | |
| 2177 | 391 |
| 392 while (num < SUB_MAX_TEXT) { | |
| 393 if (!fgets (line, 1000, fd)) return NULL; | |
| 394 p=line; | |
| 395 while (isspace(*p)) p++; | |
| 396 if (eol(*p) && num > 0) return current; | |
| 397 if (eol(*p)) return NULL; | |
| 398 | |
| 399 for (q=p; !eol(*q); q++); | |
| 400 *q='\0'; | |
| 401 if (strlen(p)) { | |
| 402 current->text[num]=strdup(p); | |
| 403 current->lines = ++num; | |
| 404 } else { | |
| 405 if (num) return current; | |
| 406 else return NULL; | |
| 407 } | |
| 408 } | |
| 409 } | |
| 410 | |
| 2343 | 411 subtitle *previous_aqt_sub = NULL; |
| 412 | |
| 413 subtitle *sub_read_line_aqt(FILE *fd,subtitle *current) { | |
| 414 char line[1001]; | |
| 415 | |
| 416 bzero (current, sizeof(subtitle)); | |
| 417 | |
| 418 while (1) { | |
| 419 // try to locate next subtitle | |
| 420 if (!fgets (line, 1000, fd)) | |
| 421 return NULL; | |
| 422 if (!(sscanf (line, "-->> %ld", &(current->start)) <1)) | |
| 423 break; | |
| 424 } | |
| 425 | |
| 426 if (previous_aqt_sub != NULL) | |
| 427 previous_aqt_sub->end = current->start-1; | |
| 428 | |
| 429 previous_aqt_sub = current; | |
| 430 | |
| 431 if (!fgets (line, 1000, fd)) | |
| 432 return NULL; | |
| 433 | |
| 434 sub_readtext(&line,¤t->text[0]); | |
| 435 current->lines = 1; | |
| 436 current->end = current->start; // will be corrected by next subtitle | |
| 437 | |
| 438 if (!fgets (line, 1000, fd)) | |
| 439 return current;; | |
| 440 | |
| 441 sub_readtext(&line,¤t->text[1]); | |
| 442 current->lines = 2; | |
| 443 | |
| 444 if ((current->text[0]=="") && (current->text[1]=="")) { | |
| 445 // void subtitle -> end of previous marked and exit | |
| 446 previous_aqt_sub = NULL; | |
| 447 return NULL; | |
| 448 } | |
| 449 | |
| 450 return current; | |
| 451 } | |
| 2177 | 452 |
| 258 | 453 int sub_autodetect (FILE *fd) { |
| 454 char line[1001]; | |
| 455 int i,j=0; | |
| 2177 | 456 char p; |
| 258 | 457 |
| 624 | 458 while (j < 100) { |
| 258 | 459 j++; |
| 460 if (!fgets (line, 1000, fd)) | |
| 461 return -1; | |
| 462 | |
| 624 | 463 if (sscanf (line, "{%d}{%d}", &i, &i)==2) |
| 258 | 464 {sub_uses_time=0;return 0;} |
| 269 | 465 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
| 258 | 466 {sub_uses_time=1;return 1;} |
| 269 | 467 if (sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
| 258 | 468 {sub_uses_time=1;return 2;} |
| 624 | 469 if (strstr (line, "<SAMI>")) |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
470 {sub_uses_time=1; return 3;} |
| 818 | 471 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3) |
| 472 {sub_uses_time=1;return 4;} | |
| 850 | 473 //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
|
474 // too weak test for RT |
|
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
475 // Please someone who knows the format of RT... FIX IT!!! |
| 921 | 476 // 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
|
477 if ( *line == '<' ) |
| 850 | 478 {sub_uses_time=1;return 5;} |
| 921 | 479 |
| 480 // I have only seen only 1 piece of .ssa file. | |
| 481 // It may be not correct (tell me if it's not) | |
| 482 if (!memcmp(line, "Dialogue: Marked", 16)) | |
| 483 {sub_uses_time=1; return 6;} | |
| 1081 | 484 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3) |
| 485 {sub_uses_time=0;return 7;} | |
| 2177 | 486 if (sscanf (line, "FORMAT=%d", &i) == 1) |
| 487 {sub_uses_time=0; return 8;} | |
| 488 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E') | |
| 489 {sub_uses_time=1; return 8;} | |
| 2343 | 490 if (strstr (line, "-->>")) |
| 491 {sub_uses_time=0; return 9;} | |
| 258 | 492 } |
| 624 | 493 |
| 494 return -1; // too many bad lines | |
| 258 | 495 } |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
496 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
497 extern int sub_utf8; |
| 258 | 498 |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
499 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
500 static iconv_t icdsc; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
501 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
502 void subcp_open (void) |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
503 { |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
504 char *tocp = "UTF-8"; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
505 icdsc = (iconv_t)(-1); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
506 if (sub_cp){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
507 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
|
508 printf ("SUB: opened iconv descriptor.\n"); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
509 sub_utf8 = 2; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
510 } else |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
511 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
|
512 } |
|
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 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
515 void subcp_close (void) |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
516 { |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
517 if (icdsc != (iconv_t)(-1)){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
518 (void) iconv_close (icdsc); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
519 printf ("SUB: closed iconv descriptor.\n"); |
|
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 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
523 #define ICBUFFSIZE 512 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
524 static char icbuffer[ICBUFFSIZE]; |
|
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 subtitle* subcp_recode (subtitle *sub) |
|
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 int l=sub->lines; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
529 size_t ileft, oleft, otlen; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
530 char *op, *ip, *ot; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
531 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
532 while (l){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
533 op = icbuffer; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
534 ip = sub->text[--l]; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
535 ileft = strlen(ip); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
536 oleft = ICBUFFSIZE - 1; |
|
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 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
|
539 &op, &oleft) == (size_t)(-1)) { |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
540 printf ("SUB: error recoding line.\n"); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
541 l++; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
542 break; |
|
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 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
|
545 printf ("SUB: error allocating mem.\n"); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
546 l++; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
547 break; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
548 } |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
549 *op='\0' ; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
550 strcpy (ot, icbuffer); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
551 free (sub->text[l]); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
552 sub->text[l] = ot; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
553 } |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
554 if (l){ |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
555 for (l = sub->lines; l;) |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
556 free (sub->text[--l]); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
557 return ERR; |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
558 } |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
559 return sub; |
|
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 |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
562 #endif |
| 258 | 563 |
| 564 subtitle* sub_read_file (char *filename) { | |
| 565 FILE *fd; | |
| 566 int n_max; | |
| 567 subtitle *first; | |
| 1081 | 568 subtitle * (*func[])(FILE *fd,subtitle *dest)= |
| 258 | 569 { |
| 570 sub_read_line_microdvd, | |
| 571 sub_read_line_subrip, | |
| 624 | 572 sub_read_line_third, |
| 818 | 573 sub_read_line_sami, |
| 850 | 574 sub_read_line_vplayer, |
| 921 | 575 sub_read_line_rt, |
| 1081 | 576 sub_read_line_ssa, |
| 2177 | 577 sub_read_line_dunnowhat, |
| 2343 | 578 sub_read_line_mpsub, |
| 579 sub_read_line_aqt | |
| 580 | |
| 258 | 581 }; |
| 582 | |
| 583 fd=fopen (filename, "r"); if (!fd) return NULL; | |
| 584 | |
| 585 sub_format=sub_autodetect (fd); | |
| 586 if (sub_format==-1) {printf ("SUB: Could not determine file format\n");return NULL;} | |
| 624 | 587 printf ("SUB: Detected subtitle file format: %d\n",sub_format); |
| 258 | 588 |
| 589 rewind (fd); | |
| 590 | |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
591 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
592 subcp_open(); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
593 #endif |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
594 |
| 258 | 595 sub_num=0;n_max=32; |
| 596 first=(subtitle *)malloc(n_max*sizeof(subtitle)); | |
| 597 if(!first) return NULL; | |
| 598 | |
| 599 while(1){ | |
| 600 subtitle *sub; | |
| 601 if(sub_num>=n_max){ | |
| 602 n_max+=16; | |
| 603 first=realloc(first,n_max*sizeof(subtitle)); | |
| 604 } | |
| 605 sub=func[sub_format](fd,&first[sub_num]); | |
| 606 if(!sub) break; // EOF | |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
607 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
608 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
|
609 #endif |
| 258 | 610 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid |
| 611 } | |
| 612 | |
| 613 fclose(fd); | |
| 614 | |
|
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
615 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
616 subcp_close(); |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
617 #endif |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
618 |
| 258 | 619 // printf ("SUB: Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); |
| 269 | 620 printf ("SUB: Read %i subtitles", sub_num); |
| 624 | 621 if (sub_errs) printf (", %i bad line(s).\n", sub_errs); |
| 269 | 622 else printf (".\n"); |
| 258 | 623 |
| 624 return first; | |
| 625 } | |
| 626 | |
| 892 | 627 #if 0 |
| 509 | 628 char * strreplace( char * in,char * what,char * whereof ) |
| 629 { | |
| 630 int i; | |
| 631 char * tmp; | |
| 632 | |
| 633 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL; | |
| 634 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i]; | |
| 635 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0; | |
| 636 return in; | |
| 637 } | |
| 892 | 638 #endif |
| 509 | 639 |
| 892 | 640 char * sub_filename(char* path, char * fname ) |
| 509 | 641 { |
| 892 | 642 char * sub_name1; |
| 643 char * sub_name2; | |
|
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
644 char * aviptr1, * aviptr2, * tmp; |
| 892 | 645 int i,j; |
| 646 FILE * f; | |
| 647 int pos=0; | |
| 648 char * sub_exts[] = | |
|
1501
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
649 { ".utf", |
|
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
650 ".UTF", |
|
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
651 ".sub", |
| 509 | 652 ".SUB", |
| 653 ".srt", | |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
654 ".SRT", |
|
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
655 ".smi", |
| 850 | 656 ".SMI", |
| 657 ".rt", | |
| 658 ".RT", | |
| 659 ".txt", | |
| 1081 | 660 ".TXT", |
| 661 ".ssa", | |
| 2343 | 662 ".SSA", |
| 663 ".aqt", | |
| 664 ".AQT"}; | |
| 892 | 665 |
|
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
666 |
| 509 | 667 if ( fname == NULL ) return NULL; |
| 892 | 668 |
| 669 sub_name1=strrchr(fname,'.'); | |
| 670 if (!sub_name1) return NULL; | |
| 671 pos=sub_name1-fname; | |
| 672 | |
|
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
673 sub_name1=malloc(strlen(fname)+8); |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
674 strcpy(sub_name1,fname); |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
675 |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
676 sub_name2=malloc (strlen(path) + strlen(fname) + 8); |
| 1081 | 677 if ((tmp=strrchr(fname,'/'))) |
|
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
678 sprintf (sub_name2, "%s%s", path, tmp+1); |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
679 else |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
680 sprintf (sub_name2, "%s%s", path, fname); |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
681 |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
682 aviptr1=strrchr(sub_name1,'.'); |
|
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
683 aviptr2=strrchr(sub_name2,'.'); |
| 892 | 684 |
| 685 for(j=0;j<=1;j++){ | |
| 686 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
|
687 #ifdef USE_ICONV |
|
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
688 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
|
689 #else |
| 892 | 690 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
|
691 #endif |
|
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
692 strcpy(j?aviptr1:aviptr2,sub_exts[i]); |
| 935 | 693 // printf("trying: '%s'\n",sub_name); |
| 892 | 694 if((f=fopen( sub_name,"rt" ))) { |
| 509 | 695 fclose( f ); |
| 696 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
|
697 if (i<2) sub_utf8=1; |
| 509 | 698 return sub_name; |
| 892 | 699 } |
| 509 | 700 } |
| 892 | 701 } |
| 702 | |
| 509 | 703 return NULL; |
| 704 } | |
| 705 | |
| 1761 | 706 void list_sub_file(subtitle* subs){ |
| 707 int i,j; | |
| 708 | |
| 709 for(j=0;j<sub_num;j++){ | |
| 710 subtitle* egysub=&subs[j]; | |
| 711 printf ("%i line%c (%li-%li) ", | |
| 712 egysub->lines, | |
| 713 (1==egysub->lines)?' ':'s', | |
| 714 egysub->start, | |
| 715 egysub->end); | |
| 716 for (i=0; i<egysub->lines; i++) { | |
| 717 printf ("%s%s",egysub->text[i], i==egysub->lines-1?"":" <BREAK> "); | |
| 718 } | |
| 719 printf ("\n"); | |
| 720 } | |
| 721 | |
| 722 printf ("Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); | |
| 723 printf ("Read %i subtitles, %i errors.\n", sub_num, sub_errs); | |
| 724 | |
| 725 } | |
| 726 | |
| 2178 | 727 void dump_mpsub(subtitle* subs){ |
| 728 int i,j; | |
| 729 FILE *fd; | |
| 730 float a,b; | |
| 731 | |
| 732 mpsub_position=0.0; | |
| 733 | |
| 734 fd=fopen ("dump.mpsub", "w"); | |
| 735 if (!fd) { | |
| 736 perror ("dump_mpsub: fopen"); | |
| 737 return; | |
| 738 } | |
| 739 | |
| 740 | |
| 741 if (sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n"); | |
| 742 else fprintf (fd, "FORMAT=25"); // FIXME: fps | |
| 743 | |
| 744 for(j=0;j<sub_num;j++){ | |
| 745 subtitle* egysub=&subs[j]; | |
| 746 a=((egysub->start-mpsub_position)/100.0); | |
| 747 b=((egysub->end-egysub->start)/100.0); | |
| 748 if ( (float)((int)a) == a) | |
| 749 fprintf (fd, "%.0f",a); | |
| 750 else | |
| 751 fprintf (fd, "%.2f",a); | |
| 752 | |
| 753 if ( (float)((int)b) == b) | |
| 754 fprintf (fd, " %.0f\n",b); | |
| 755 else | |
| 756 fprintf (fd, " %.2f\n",b); | |
| 757 | |
| 758 mpsub_position = egysub->end; | |
| 759 for (i=0; i<egysub->lines; i++) { | |
| 760 fprintf (fd, "%s\n",egysub->text[i]); | |
| 761 } | |
| 762 fprintf (fd, "\n"); | |
| 763 } | |
| 764 fclose (fd); | |
| 765 printf ("Subtitles dumped in \'dump.mpsub\'.\n"); | |
| 766 } | |
| 767 | |
| 768 | |
| 769 | |
| 625 | 770 #if 0 |
| 258 | 771 int main(int argc, char **argv) { // for testing |
| 772 | |
| 773 int i,j; | |
| 774 subtitle *subs; | |
| 775 subtitle *egysub; | |
| 776 | |
| 777 if(argc<2){ | |
| 778 printf("\nUsage: subreader filename.sub\n\n"); | |
| 779 exit(1); | |
| 780 } | |
| 781 | |
| 624 | 782 subs=sub_read_file(argv[1]); |
| 258 | 783 if(!subs){ |
| 784 printf("Couldn't load file... let's write a bugreport :)\n"); | |
| 785 exit(1); | |
| 786 } | |
| 1761 | 787 |
| 788 list_sub_file(subs); | |
| 258 | 789 |
| 790 return 0; | |
| 791 } | |
|
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
792 #endif |
