Mercurial > mplayer.hg
annotate subreader.c @ 3410:3eb15016e454
still fiddling with surround levels
author | steve |
---|---|
date | Sun, 09 Dec 2001 18:54:54 +0000 |
parents | 0cf593b6bab0 |
children | 8519aba2644d |
rev | line source |
---|---|
258 | 1 /* |
2 * Subtitle reader with format autodetection | |
3 * | |
4 * Written by laaz | |
5 * Some code cleanup & realloc() by A'rpi/ESP-team | |
1081 | 6 * dunnowhat sub format by szabi |
258 | 7 */ |
8 | |
9 | |
10 #include <stdio.h> | |
11 #include <stdlib.h> | |
12 #include <string.h> | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
13 #include <ctype.h> |
258 | 14 |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
15 #include "config.h" |
258 | 16 #include "subreader.h" |
17 | |
18 #define ERR (void *)-1 | |
19 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
20 #ifdef USE_ICONV |
2358 | 21 #ifdef __FreeBSD__ |
22 #include <giconv.h> | |
23 #else | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
24 #include <iconv.h> |
2358 | 25 #endif |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
26 char *sub_cp=NULL; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
27 #endif |
258 | 28 |
2912 | 29 /* Maximal length of line of a subtitle */ |
30 #define LINE_LEN 1000 | |
2177 | 31 |
2178 | 32 static float mpsub_position=0; |
2177 | 33 |
258 | 34 int sub_uses_time=0; |
35 int sub_errs=0; | |
624 | 36 int sub_num=0; // number of subtitle structs |
3235
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
37 int sub_slacktime=2000; // 20 seconds |
2912 | 38 |
39 /* Use the SUB_* constant defined in the header file */ | |
40 int sub_format=SUB_INVALID; | |
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) { |
2912 | 55 static char line[LINE_LEN+1]; |
3235
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
56 static char *s = NULL, *slacktime_s; |
2912 | 57 char text[LINE_LEN+1], *p, *q; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
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) |
2912 | 65 if (!(s = fgets(line, LINE_LEN, fd))) return 0; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
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 |
3235
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
70 case 0: /* find "START=" or "Slacktime:" */ |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
71 slacktime_s = strstr (s, "Slacktime:"); |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
72 if (slacktime_s) sub_slacktime = strtol (slacktime_s + 10, NULL, 0) / 10; |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
73 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
74 s = strstr (s, "Start="); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
75 if (s) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
76 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
|
77 state = 1; continue; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
78 } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
79 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
80 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
81 case 1: /* find "<P" */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
82 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
|
83 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
84 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
85 case 2: /* find ">" */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
86 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
|
87 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
88 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
89 case 3: /* get all text until '<' appears */ |
3235
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
90 if (*s == '\0') break; |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
91 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
|
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); |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
95 s += 4; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
96 } |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
97 else if (*s == '<') { state = 4; } |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
98 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
|
99 else if (*s == '\t') { *p++ = ' '; s++; } |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
100 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
|
101 else *p++ = *s++; |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
102 |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
103 /* skip duplicated space */ |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
104 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
|
105 |
624 | 106 continue; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
107 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
108 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
|
109 q = strstr (s, "Start="); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
110 if (q) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
111 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
|
112 *p = '\0'; trail_space (text); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
113 if (text[0] != '\0') |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
114 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
|
115 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
|
116 state = 0; continue; |
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 s = strchr (s, '>'); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
119 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
|
120 break; |
624 | 121 } |
122 | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
123 /* read next line */ |
3235
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
124 if (state != 99 && !(s = fgets (line, LINE_LEN, fd))) |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
125 if (current->start > 0) break; // if it is the last subtitle |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
126 else return 0; |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
127 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
128 } while (state != 99); |
624 | 129 |
3235
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
130 // For the last subtitle |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
131 if (current->end <= 0) { |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
132 current->end = current->start + sub_slacktime; |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
133 *p = '\0'; trail_space (text); |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
134 if (text[0] != '\0') |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
135 current->text[current->lines++] = strdup (text); |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
136 } |
0cf593b6bab0
patch fixes the showing last line of subtitles and support of SAMI Slacktime option by Evgeny Chukreev <codedj@echo.ru>
arpi
parents:
2915
diff
changeset
|
137 |
624 | 138 return current; |
139 } | |
258 | 140 |
141 | |
142 char *sub_readtext(char *source, char **dest) { | |
143 int len=0; | |
932 | 144 char *p=source; |
258 | 145 |
932 | 146 while ( !eol(*p) && *p!= '|' ) { |
147 p++,len++; | |
148 } | |
258 | 149 |
150 *dest= (char *)malloc (len+1); | |
151 if (!dest) {return ERR;} | |
152 | |
153 strncpy(*dest, source, len); | |
154 (*dest)[len]=0; | |
155 | |
156 while (*p=='\r' || *p=='\n' || *p=='|') p++; | |
157 | |
158 if (*p) return p; // not-last text field | |
159 else return NULL; // last text field | |
160 } | |
161 | |
162 subtitle *sub_read_line_microdvd(FILE *fd,subtitle *current) { | |
2912 | 163 char line[LINE_LEN+1]; |
164 char line2[LINE_LEN+1]; | |
258 | 165 char *p, *next; |
166 int i; | |
167 | |
1764 | 168 bzero (current, sizeof(subtitle)); |
258 | 169 |
170 do { | |
2912 | 171 if (!fgets (line, LINE_LEN, fd)) return NULL; |
932 | 172 } while (sscanf (line, "{%ld}{%ld}%[^\r\n]", &(current->start), &(current->end),line2) <3); |
258 | 173 |
932 | 174 p=line2; |
258 | 175 |
176 next=p, i=0; | |
1081 | 177 while ((next =sub_readtext (next, &(current->text[i])))) { |
270 | 178 if (current->text[i]==ERR) {return ERR;} |
258 | 179 i++; |
1081 | 180 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
258 | 181 } |
932 | 182 current->lines= ++i; |
258 | 183 |
184 return current; | |
185 } | |
186 | |
187 subtitle *sub_read_line_subrip(FILE *fd, subtitle *current) { | |
2912 | 188 char line[LINE_LEN+1]; |
258 | 189 int a1,a2,a3,a4,b1,b2,b3,b4; |
190 char *p=NULL, *q=NULL; | |
191 int len; | |
192 | |
1764 | 193 bzero (current, sizeof(subtitle)); |
258 | 194 |
1764 | 195 while (1) { |
2912 | 196 if (!fgets (line, LINE_LEN, fd)) return NULL; |
269 | 197 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue; |
258 | 198 current->start = a1*360000+a2*6000+a3*100+a4; |
199 current->end = b1*360000+b2*6000+b3*100+b4; | |
200 | |
2912 | 201 if (!fgets (line, LINE_LEN, fd)) return NULL; |
258 | 202 |
203 p=q=line; | |
204 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) { | |
205 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && strncmp(p,"[br]",4); p++,len++); | |
206 current->text[current->lines-1]=(char *)malloc (len+1); | |
207 if (!current->text[current->lines-1]) return ERR; | |
208 strncpy (current->text[current->lines-1], q, len); | |
270 | 209 current->text[current->lines-1][len]='\0'; |
258 | 210 if (!*p || *p=='\r' || *p=='\n') break; |
211 while (*p++!=']'); | |
212 } | |
1764 | 213 break; |
258 | 214 } |
215 return current; | |
216 } | |
217 | |
2912 | 218 subtitle *sub_read_line_subviewer(FILE *fd,subtitle *current) { |
219 char line[LINE_LEN+1]; | |
258 | 220 int a1,a2,a3,a4,b1,b2,b3,b4; |
221 char *p=NULL; | |
222 int i,len; | |
223 | |
1764 | 224 bzero (current, sizeof(subtitle)); |
258 | 225 |
226 while (!current->text[0]) { | |
2912 | 227 if (!fgets (line, LINE_LEN, fd)) return NULL; |
269 | 228 if ((len=sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) |
258 | 229 continue; |
230 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
231 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
232 for (i=0; i<SUB_MAX_TEXT;) { | |
2912 | 233 if (!fgets (line, LINE_LEN, fd)) break; |
258 | 234 len=0; |
235 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++); | |
236 if (len) { | |
237 current->text[i]=(char *)malloc (len+1); | |
238 if (!current->text[i]) return ERR; | |
270 | 239 strncpy (current->text[i], line, len); current->text[i][len]='\0'; |
258 | 240 i++; |
241 } else { | |
242 break; | |
243 } | |
244 } | |
245 current->lines=i; | |
246 } | |
247 return current; | |
248 } | |
249 | |
818 | 250 subtitle *sub_read_line_vplayer(FILE *fd,subtitle *current) { |
2912 | 251 char line[LINE_LEN+1]; |
252 char line2[LINE_LEN+1]; | |
818 | 253 int a1,a2,a3,b1,b2,b3; |
1081 | 254 char *p=NULL, *next; |
858 | 255 int i,len,len2,plen; |
818 | 256 |
1764 | 257 bzero (current, sizeof(subtitle)); |
818 | 258 |
259 while (!current->text[0]) { | |
2912 | 260 if (!fgets (line, LINE_LEN, fd)) return NULL; |
858 | 261 if ((len=sscanf (line, "%d:%d:%d:%n",&a1,&a2,&a3,&plen)) < 3) |
818 | 262 continue; |
2912 | 263 if (!fgets (line2, LINE_LEN, fd)) return NULL; |
818 | 264 if ((len2=sscanf (line2, "%d:%d:%d:",&b1,&b2,&b3)) < 3) |
265 continue; | |
266 // przewiñ o linijkê do ty³u: | |
267 fseek(fd,-strlen(line2),SEEK_CUR); | |
268 | |
269 current->start = a1*360000+a2*6000+a3*100; | |
270 current->end = b1*360000+b2*6000+b3*100; | |
896
d46de26aef48
there is another format that get detected as vplayers.
eyck
parents:
892
diff
changeset
|
271 if ((current->end - current->start) > 1000) {current->end = current->start + 1000;} // not too long though. |
818 | 272 // teraz czas na wkopiowanie stringu |
1640
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
273 p=line; |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
274 // finds the body of the subtitle |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
275 for (i=0; i<3; i++){ |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
276 p=strchr(p,':')+1; |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
277 } |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
278 i=0; |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
279 |
818 | 280 if (*p!='|') { |
281 // | |
282 next = p,i=0; | |
283 while ((next =sub_readtext (next, &(current->text[i])))) { | |
284 if (current->text[i]==ERR) {return ERR;} | |
285 i++; | |
1081 | 286 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
818 | 287 } |
288 current->lines=i+1; | |
289 } | |
290 } | |
291 return current; | |
292 } | |
293 | |
850 | 294 subtitle *sub_read_line_rt(FILE *fd,subtitle *current) { |
295 //TODO: This format uses quite rich (sub/super)set of xhtml | |
296 // I couldn't check it since DTD is not included. | |
297 // WARNING: full XML parses can be required for proper parsing | |
2912 | 298 char line[LINE_LEN+1]; |
850 | 299 int a1,a2,a3,a4,b1,b2,b3,b4; |
300 char *p=NULL,*next=NULL; | |
301 int i,len,plen; | |
302 | |
1764 | 303 bzero (current, sizeof(subtitle)); |
850 | 304 |
305 while (!current->text[0]) { | |
2912 | 306 if (!fgets (line, LINE_LEN, fd)) return NULL; |
850 | 307 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0 |
308 //to describe the same moment in time. Maybe there are even more formats in use. | |
309 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) | |
310 plen=a1=a2=a3=a4=b1=b2=b3=b4=0; | |
311 if ( | |
312 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) && | |
313 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) && | |
314 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) && | |
315 ((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) && | |
316 ((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) | |
317 ) | |
318 continue; | |
319 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
320 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
321 p=line; p+=plen;i=0; | |
322 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml? | |
323 next = strstr(line,"<clear/>")+8;i=0; | |
324 while ((next =sub_readtext (next, &(current->text[i])))) { | |
325 if (current->text[i]==ERR) {return ERR;} | |
326 i++; | |
1081 | 327 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
850 | 328 } |
329 current->lines=i+1; | |
330 } | |
331 return current; | |
332 } | |
333 | |
921 | 334 subtitle *sub_read_line_ssa(FILE *fd,subtitle *current) { |
335 int hour1, min1, sec1, hunsec1, | |
336 hour2, min2, sec2, hunsec2, nothing; | |
2141 | 337 int num; |
921 | 338 |
2912 | 339 char line[LINE_LEN+1], |
340 line3[LINE_LEN+1], | |
2140 | 341 *line2; |
2141 | 342 char *tmp; |
343 | |
921 | 344 do { |
2912 | 345 if (!fgets (line, LINE_LEN, fd)) return NULL; |
921 | 346 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d," |
2140 | 347 "%[^\n\r]", ¬hing, |
348 &hour1, &min1, &sec1, &hunsec1, | |
349 &hour2, &min2, &sec2, &hunsec2, | |
350 line3) < 9); | |
351 line2=strstr(line3,",,"); | |
352 if (!line2) return NULL; | |
353 line2 ++; | |
354 line2 ++; | |
355 | |
2141 | 356 current->lines=1;num=0; |
921 | 357 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1; |
358 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2; | |
2141 | 359 |
360 while (tmp=strstr(line2, "\\n")) { | |
361 current->text[num]=(char *)malloc(tmp-line2+1); | |
362 strncpy (current->text[num], line2, tmp-line2); | |
363 current->text[num][tmp-line2]='\0'; | |
364 line2=tmp+2; | |
365 num++; | |
366 current->lines++; | |
367 if (current->lines >= SUB_MAX_TEXT) return current; | |
368 } | |
369 | |
370 | |
371 current->text[num]=(char *) malloc(strlen(line2)+1); | |
372 strcpy(current->text[num],line2); | |
818 | 373 |
921 | 374 return current; |
375 } | |
258 | 376 |
1081 | 377 subtitle *sub_read_line_dunnowhat(FILE *fd,subtitle *current) { |
2912 | 378 char line[LINE_LEN+1]; |
379 char text[LINE_LEN+1]; | |
1081 | 380 |
1764 | 381 bzero (current, sizeof(subtitle)); |
1081 | 382 |
2912 | 383 if (!fgets (line, LINE_LEN, fd)) |
1081 | 384 return NULL; |
385 if (sscanf (line, "%ld,%ld,\"%[^\"]", &(current->start), | |
386 &(current->end), text) <3) | |
387 return ERR; | |
388 current->text[0] = strdup(text); | |
389 current->lines = 1; | |
390 | |
391 return current; | |
392 } | |
393 | |
2177 | 394 subtitle *sub_read_line_mpsub(FILE *fd, subtitle *current) { |
2912 | 395 char line[LINE_LEN+1]; |
2178 | 396 float a,b; |
397 int num=0; | |
2177 | 398 char *p, *q; |
399 | |
400 do | |
401 { | |
2912 | 402 if (!fgets(line, LINE_LEN, fd)) return NULL; |
2178 | 403 } while (sscanf (line, "%f %f", &a, &b) !=2); |
2177 | 404 |
2178 | 405 mpsub_position += (a*100.0); |
406 current->start=(int) mpsub_position; | |
407 mpsub_position += (b*100.0); | |
408 current->end=(int) mpsub_position; | |
2177 | 409 |
410 while (num < SUB_MAX_TEXT) { | |
2912 | 411 if (!fgets (line, LINE_LEN, fd)) return NULL; |
2177 | 412 p=line; |
413 while (isspace(*p)) p++; | |
414 if (eol(*p) && num > 0) return current; | |
415 if (eol(*p)) return NULL; | |
416 | |
417 for (q=p; !eol(*q); q++); | |
418 *q='\0'; | |
419 if (strlen(p)) { | |
420 current->text[num]=strdup(p); | |
2495 | 421 printf (">%s<\n",p); |
2177 | 422 current->lines = ++num; |
423 } else { | |
424 if (num) return current; | |
425 else return NULL; | |
426 } | |
427 } | |
428 } | |
429 | |
2343 | 430 subtitle *previous_aqt_sub = NULL; |
431 | |
432 subtitle *sub_read_line_aqt(FILE *fd,subtitle *current) { | |
2912 | 433 char line[LINE_LEN+1]; |
2343 | 434 |
435 bzero (current, sizeof(subtitle)); | |
436 | |
437 while (1) { | |
438 // try to locate next subtitle | |
2912 | 439 if (!fgets (line, LINE_LEN, fd)) |
2343 | 440 return NULL; |
441 if (!(sscanf (line, "-->> %ld", &(current->start)) <1)) | |
442 break; | |
443 } | |
444 | |
445 if (previous_aqt_sub != NULL) | |
446 previous_aqt_sub->end = current->start-1; | |
447 | |
448 previous_aqt_sub = current; | |
449 | |
2912 | 450 if (!fgets (line, LINE_LEN, fd)) |
2343 | 451 return NULL; |
452 | |
2468 | 453 sub_readtext((char *) &line,¤t->text[0]); |
2343 | 454 current->lines = 1; |
455 current->end = current->start; // will be corrected by next subtitle | |
456 | |
2912 | 457 if (!fgets (line, LINE_LEN, fd)) |
2343 | 458 return current;; |
459 | |
2468 | 460 sub_readtext((char *) &line,¤t->text[1]); |
2343 | 461 current->lines = 2; |
462 | |
463 if ((current->text[0]=="") && (current->text[1]=="")) { | |
464 // void subtitle -> end of previous marked and exit | |
465 previous_aqt_sub = NULL; | |
466 return NULL; | |
467 } | |
468 | |
469 return current; | |
470 } | |
2177 | 471 |
258 | 472 int sub_autodetect (FILE *fd) { |
2912 | 473 char line[LINE_LEN+1]; |
258 | 474 int i,j=0; |
2177 | 475 char p; |
258 | 476 |
624 | 477 while (j < 100) { |
258 | 478 j++; |
2912 | 479 if (!fgets (line, LINE_LEN, fd)) |
480 return SUB_INVALID; | |
258 | 481 |
624 | 482 if (sscanf (line, "{%d}{%d}", &i, &i)==2) |
2912 | 483 {sub_uses_time=0;return SUB_MICRODVD;} |
269 | 484 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
2912 | 485 {sub_uses_time=1;return SUB_SUBRIP;} |
269 | 486 if (sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
2912 | 487 {sub_uses_time=1;return SUB_SUBVIEWER;} |
624 | 488 if (strstr (line, "<SAMI>")) |
2912 | 489 {sub_uses_time=1; return SUB_SAMI;} |
818 | 490 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3) |
2912 | 491 {sub_uses_time=1;return SUB_VPLAYER;} |
850 | 492 //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
|
493 // too weak test for RT |
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
494 // Please someone who knows the format of RT... FIX IT!!! |
921 | 495 // 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
|
496 if ( *line == '<' ) |
2912 | 497 {sub_uses_time=1;return SUB_RT;} |
921 | 498 |
499 if (!memcmp(line, "Dialogue: Marked", 16)) | |
2912 | 500 {sub_uses_time=1; return SUB_SSA;} |
1081 | 501 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3) |
2912 | 502 {sub_uses_time=0;return SUB_DUNNOWHAT;} |
2177 | 503 if (sscanf (line, "FORMAT=%d", &i) == 1) |
2912 | 504 {sub_uses_time=0; return SUB_MPSUB;} |
2177 | 505 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E') |
2912 | 506 {sub_uses_time=1; return SUB_MPSUB;} |
2343 | 507 if (strstr (line, "-->>")) |
2912 | 508 {sub_uses_time=0; return SUB_MPSUB;} |
258 | 509 } |
624 | 510 |
2912 | 511 return SUB_INVALID; // too many bad lines |
258 | 512 } |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
513 |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
514 #ifdef DUMPSUBS |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
515 int sub_utf8=0; |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
516 #else |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
517 extern int sub_utf8; |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
518 #endif |
258 | 519 |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
520 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
521 static iconv_t icdsc; |
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 void subcp_open (void) |
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 char *tocp = "UTF-8"; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
526 icdsc = (iconv_t)(-1); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
527 if (sub_cp){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
528 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
|
529 printf ("SUB: opened iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
530 sub_utf8 = 2; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
531 } else |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
532 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
|
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 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
536 void subcp_close (void) |
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 (icdsc != (iconv_t)(-1)){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
539 (void) iconv_close (icdsc); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
540 printf ("SUB: closed iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
541 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
542 } |
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 #define ICBUFFSIZE 512 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
545 static char icbuffer[ICBUFFSIZE]; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
546 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
547 subtitle* subcp_recode (subtitle *sub) |
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 int l=sub->lines; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
550 size_t ileft, oleft, otlen; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
551 char *op, *ip, *ot; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
552 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
553 while (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
554 op = icbuffer; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
555 ip = sub->text[--l]; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
556 ileft = strlen(ip); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
557 oleft = ICBUFFSIZE - 1; |
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 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
|
560 &op, &oleft) == (size_t)(-1)) { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
561 printf ("SUB: error recoding line.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
562 l++; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
563 break; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
564 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
565 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
|
566 printf ("SUB: error allocating mem.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
567 l++; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
568 break; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
569 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
570 *op='\0' ; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
571 strcpy (ot, icbuffer); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
572 free (sub->text[l]); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
573 sub->text[l] = ot; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
574 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
575 if (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
576 for (l = sub->lines; l;) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
577 free (sub->text[--l]); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
578 return ERR; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
579 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
580 return sub; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
581 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
582 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
583 #endif |
258 | 584 |
585 subtitle* sub_read_file (char *filename) { | |
586 FILE *fd; | |
587 int n_max; | |
588 subtitle *first; | |
1081 | 589 subtitle * (*func[])(FILE *fd,subtitle *dest)= |
258 | 590 { |
591 sub_read_line_microdvd, | |
592 sub_read_line_subrip, | |
2912 | 593 sub_read_line_subviewer, |
818 | 594 sub_read_line_sami, |
850 | 595 sub_read_line_vplayer, |
921 | 596 sub_read_line_rt, |
1081 | 597 sub_read_line_ssa, |
2177 | 598 sub_read_line_dunnowhat, |
2343 | 599 sub_read_line_mpsub, |
600 sub_read_line_aqt | |
601 | |
258 | 602 }; |
2915 | 603 if(filename==NULL) return NULL; //qnx segfault |
258 | 604 fd=fopen (filename, "r"); if (!fd) return NULL; |
605 | |
606 sub_format=sub_autodetect (fd); | |
2912 | 607 if (sub_format==SUB_INVALID) {printf ("SUB: Could not determine file format\n");return NULL;} |
624 | 608 printf ("SUB: Detected subtitle file format: %d\n",sub_format); |
258 | 609 |
610 rewind (fd); | |
611 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
612 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
613 subcp_open(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
614 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
615 |
258 | 616 sub_num=0;n_max=32; |
617 first=(subtitle *)malloc(n_max*sizeof(subtitle)); | |
618 if(!first) return NULL; | |
619 | |
620 while(1){ | |
621 subtitle *sub; | |
622 if(sub_num>=n_max){ | |
623 n_max+=16; | |
624 first=realloc(first,n_max*sizeof(subtitle)); | |
625 } | |
626 sub=func[sub_format](fd,&first[sub_num]); | |
627 if(!sub) break; // EOF | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
628 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
629 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
|
630 #endif |
258 | 631 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid |
632 } | |
633 | |
634 fclose(fd); | |
635 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
636 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
637 subcp_close(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
638 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
639 |
258 | 640 // printf ("SUB: Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); |
269 | 641 printf ("SUB: Read %i subtitles", sub_num); |
624 | 642 if (sub_errs) printf (", %i bad line(s).\n", sub_errs); |
269 | 643 else printf (".\n"); |
258 | 644 |
2880 | 645 if(sub_num<=0){ |
646 free(first); | |
647 return NULL; | |
648 } | |
649 | |
258 | 650 return first; |
651 } | |
652 | |
892 | 653 #if 0 |
509 | 654 char * strreplace( char * in,char * what,char * whereof ) |
655 { | |
656 int i; | |
657 char * tmp; | |
658 | |
659 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL; | |
660 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i]; | |
661 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0; | |
662 return in; | |
663 } | |
892 | 664 #endif |
509 | 665 |
892 | 666 char * sub_filename(char* path, char * fname ) |
509 | 667 { |
892 | 668 char * sub_name1; |
669 char * sub_name2; | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
670 char * aviptr1, * aviptr2, * tmp; |
892 | 671 int i,j; |
672 FILE * f; | |
673 int pos=0; | |
674 char * sub_exts[] = | |
1501
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
675 { ".utf", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
676 ".UTF", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
677 ".sub", |
509 | 678 ".SUB", |
679 ".srt", | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
680 ".SRT", |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
681 ".smi", |
850 | 682 ".SMI", |
683 ".rt", | |
684 ".RT", | |
685 ".txt", | |
1081 | 686 ".TXT", |
687 ".ssa", | |
2343 | 688 ".SSA", |
689 ".aqt", | |
690 ".AQT"}; | |
892 | 691 |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
692 |
509 | 693 if ( fname == NULL ) return NULL; |
892 | 694 |
695 sub_name1=strrchr(fname,'.'); | |
696 if (!sub_name1) return NULL; | |
697 pos=sub_name1-fname; | |
698 | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
699 sub_name1=malloc(strlen(fname)+8); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
700 strcpy(sub_name1,fname); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
701 |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
702 sub_name2=malloc (strlen(path) + strlen(fname) + 8); |
1081 | 703 if ((tmp=strrchr(fname,'/'))) |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
704 sprintf (sub_name2, "%s%s", path, tmp+1); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
705 else |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
706 sprintf (sub_name2, "%s%s", path, fname); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
707 |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
708 aviptr1=strrchr(sub_name1,'.'); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
709 aviptr2=strrchr(sub_name2,'.'); |
892 | 710 |
711 for(j=0;j<=1;j++){ | |
712 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
|
713 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
714 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
|
715 #else |
892 | 716 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
|
717 #endif |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
718 strcpy(j?aviptr1:aviptr2,sub_exts[i]); |
935 | 719 // printf("trying: '%s'\n",sub_name); |
892 | 720 if((f=fopen( sub_name,"rt" ))) { |
509 | 721 fclose( f ); |
722 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
|
723 if (i<2) sub_utf8=1; |
509 | 724 return sub_name; |
892 | 725 } |
509 | 726 } |
892 | 727 } |
728 | |
509 | 729 return NULL; |
730 } | |
731 | |
1761 | 732 void list_sub_file(subtitle* subs){ |
733 int i,j; | |
734 | |
735 for(j=0;j<sub_num;j++){ | |
736 subtitle* egysub=&subs[j]; | |
737 printf ("%i line%c (%li-%li) ", | |
738 egysub->lines, | |
739 (1==egysub->lines)?' ':'s', | |
740 egysub->start, | |
741 egysub->end); | |
742 for (i=0; i<egysub->lines; i++) { | |
743 printf ("%s%s",egysub->text[i], i==egysub->lines-1?"":" <BREAK> "); | |
744 } | |
745 printf ("\n"); | |
746 } | |
747 | |
748 printf ("Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); | |
749 printf ("Read %i subtitles, %i errors.\n", sub_num, sub_errs); | |
750 | |
751 } | |
752 | |
2178 | 753 void dump_mpsub(subtitle* subs){ |
754 int i,j; | |
755 FILE *fd; | |
756 float a,b; | |
757 | |
758 mpsub_position=0.0; | |
759 | |
760 fd=fopen ("dump.mpsub", "w"); | |
761 if (!fd) { | |
762 perror ("dump_mpsub: fopen"); | |
763 return; | |
764 } | |
765 | |
766 | |
767 if (sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n"); | |
2495 | 768 else fprintf (fd, "FORMAT=25\n\n"); // FIXME: fps |
2178 | 769 |
770 for(j=0;j<sub_num;j++){ | |
2495 | 771 subtitle* egysub=&subs[j]; |
772 if (sub_uses_time) { | |
773 a=((egysub->start-mpsub_position)/100.0); | |
774 b=((egysub->end-egysub->start)/100.0); | |
775 if ( (float)((int)a) == a) | |
776 fprintf (fd, "%.0f",a); | |
777 else | |
778 fprintf (fd, "%.2f",a); | |
779 | |
780 if ( (float)((int)b) == b) | |
781 fprintf (fd, " %.0f\n",b); | |
782 else | |
783 fprintf (fd, " %.2f\n",b); | |
784 } else { | |
785 fprintf (fd, "%ld %ld\n", egysub->start-mpsub_position, | |
786 egysub->end-egysub->start); | |
787 } | |
788 | |
789 mpsub_position = egysub->end; | |
790 for (i=0; i<egysub->lines; i++) { | |
791 fprintf (fd, "%s\n",egysub->text[i]); | |
792 } | |
793 fprintf (fd, "\n"); | |
2178 | 794 } |
795 fclose (fd); | |
796 printf ("Subtitles dumped in \'dump.mpsub\'.\n"); | |
797 } | |
798 | |
799 | |
800 | |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
801 #ifdef DUMPSUBS |
258 | 802 int main(int argc, char **argv) { // for testing |
803 | |
804 int i,j; | |
805 subtitle *subs; | |
806 subtitle *egysub; | |
807 | |
808 if(argc<2){ | |
809 printf("\nUsage: subreader filename.sub\n\n"); | |
810 exit(1); | |
811 } | |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
812 sub_cp = argv[2]; |
624 | 813 subs=sub_read_file(argv[1]); |
258 | 814 if(!subs){ |
815 printf("Couldn't load file... let's write a bugreport :)\n"); | |
816 exit(1); | |
817 } | |
1761 | 818 |
819 list_sub_file(subs); | |
258 | 820 |
821 return 0; | |
822 } | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
823 #endif |