Mercurial > mplayer.hg
annotate subreader.c @ 3475:390388c75209
Applied the patch from Alban Bedel <albeu@free.fr> to
make the asf network streaming work for some links.
author | bertrand |
---|---|
date | Wed, 12 Dec 2001 22:35:51 +0000 |
parents | 8519aba2644d |
children | 5a11cbf8791f |
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++){ |
3433 | 276 p=strchr(p,':'); |
277 if (p==NULL) break; | |
278 ++p; | |
279 } | |
280 if (p==NULL) { | |
281 printf("SUB: Skipping incorrect subtitle line!\n"); | |
282 continue; | |
283 } | |
1640
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
284 i=0; |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
285 |
818 | 286 if (*p!='|') { |
287 // | |
288 next = p,i=0; | |
289 while ((next =sub_readtext (next, &(current->text[i])))) { | |
290 if (current->text[i]==ERR) {return ERR;} | |
291 i++; | |
1081 | 292 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
818 | 293 } |
294 current->lines=i+1; | |
295 } | |
296 } | |
297 return current; | |
298 } | |
299 | |
850 | 300 subtitle *sub_read_line_rt(FILE *fd,subtitle *current) { |
301 //TODO: This format uses quite rich (sub/super)set of xhtml | |
302 // I couldn't check it since DTD is not included. | |
303 // WARNING: full XML parses can be required for proper parsing | |
2912 | 304 char line[LINE_LEN+1]; |
850 | 305 int a1,a2,a3,a4,b1,b2,b3,b4; |
306 char *p=NULL,*next=NULL; | |
307 int i,len,plen; | |
308 | |
1764 | 309 bzero (current, sizeof(subtitle)); |
850 | 310 |
311 while (!current->text[0]) { | |
2912 | 312 if (!fgets (line, LINE_LEN, fd)) return NULL; |
850 | 313 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0 |
314 //to describe the same moment in time. Maybe there are even more formats in use. | |
315 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) | |
316 plen=a1=a2=a3=a4=b1=b2=b3=b4=0; | |
317 if ( | |
318 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) && | |
319 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) && | |
320 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) && | |
321 ((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) && | |
322 ((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) | |
323 ) | |
324 continue; | |
325 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
326 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
327 p=line; p+=plen;i=0; | |
328 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml? | |
329 next = strstr(line,"<clear/>")+8;i=0; | |
330 while ((next =sub_readtext (next, &(current->text[i])))) { | |
331 if (current->text[i]==ERR) {return ERR;} | |
332 i++; | |
1081 | 333 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
850 | 334 } |
335 current->lines=i+1; | |
336 } | |
337 return current; | |
338 } | |
339 | |
921 | 340 subtitle *sub_read_line_ssa(FILE *fd,subtitle *current) { |
341 int hour1, min1, sec1, hunsec1, | |
342 hour2, min2, sec2, hunsec2, nothing; | |
2141 | 343 int num; |
921 | 344 |
2912 | 345 char line[LINE_LEN+1], |
346 line3[LINE_LEN+1], | |
2140 | 347 *line2; |
2141 | 348 char *tmp; |
349 | |
921 | 350 do { |
2912 | 351 if (!fgets (line, LINE_LEN, fd)) return NULL; |
921 | 352 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d," |
2140 | 353 "%[^\n\r]", ¬hing, |
354 &hour1, &min1, &sec1, &hunsec1, | |
355 &hour2, &min2, &sec2, &hunsec2, | |
356 line3) < 9); | |
357 line2=strstr(line3,",,"); | |
358 if (!line2) return NULL; | |
359 line2 ++; | |
360 line2 ++; | |
361 | |
2141 | 362 current->lines=1;num=0; |
921 | 363 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1; |
364 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2; | |
2141 | 365 |
366 while (tmp=strstr(line2, "\\n")) { | |
367 current->text[num]=(char *)malloc(tmp-line2+1); | |
368 strncpy (current->text[num], line2, tmp-line2); | |
369 current->text[num][tmp-line2]='\0'; | |
370 line2=tmp+2; | |
371 num++; | |
372 current->lines++; | |
373 if (current->lines >= SUB_MAX_TEXT) return current; | |
374 } | |
375 | |
376 | |
377 current->text[num]=(char *) malloc(strlen(line2)+1); | |
378 strcpy(current->text[num],line2); | |
818 | 379 |
921 | 380 return current; |
381 } | |
258 | 382 |
1081 | 383 subtitle *sub_read_line_dunnowhat(FILE *fd,subtitle *current) { |
2912 | 384 char line[LINE_LEN+1]; |
385 char text[LINE_LEN+1]; | |
1081 | 386 |
1764 | 387 bzero (current, sizeof(subtitle)); |
1081 | 388 |
2912 | 389 if (!fgets (line, LINE_LEN, fd)) |
1081 | 390 return NULL; |
391 if (sscanf (line, "%ld,%ld,\"%[^\"]", &(current->start), | |
392 &(current->end), text) <3) | |
393 return ERR; | |
394 current->text[0] = strdup(text); | |
395 current->lines = 1; | |
396 | |
397 return current; | |
398 } | |
399 | |
2177 | 400 subtitle *sub_read_line_mpsub(FILE *fd, subtitle *current) { |
2912 | 401 char line[LINE_LEN+1]; |
2178 | 402 float a,b; |
403 int num=0; | |
2177 | 404 char *p, *q; |
405 | |
406 do | |
407 { | |
2912 | 408 if (!fgets(line, LINE_LEN, fd)) return NULL; |
2178 | 409 } while (sscanf (line, "%f %f", &a, &b) !=2); |
2177 | 410 |
2178 | 411 mpsub_position += (a*100.0); |
412 current->start=(int) mpsub_position; | |
413 mpsub_position += (b*100.0); | |
414 current->end=(int) mpsub_position; | |
2177 | 415 |
416 while (num < SUB_MAX_TEXT) { | |
2912 | 417 if (!fgets (line, LINE_LEN, fd)) return NULL; |
2177 | 418 p=line; |
419 while (isspace(*p)) p++; | |
420 if (eol(*p) && num > 0) return current; | |
421 if (eol(*p)) return NULL; | |
422 | |
423 for (q=p; !eol(*q); q++); | |
424 *q='\0'; | |
425 if (strlen(p)) { | |
426 current->text[num]=strdup(p); | |
2495 | 427 printf (">%s<\n",p); |
2177 | 428 current->lines = ++num; |
429 } else { | |
430 if (num) return current; | |
431 else return NULL; | |
432 } | |
433 } | |
434 } | |
435 | |
2343 | 436 subtitle *previous_aqt_sub = NULL; |
437 | |
438 subtitle *sub_read_line_aqt(FILE *fd,subtitle *current) { | |
2912 | 439 char line[LINE_LEN+1]; |
2343 | 440 |
441 bzero (current, sizeof(subtitle)); | |
442 | |
443 while (1) { | |
444 // try to locate next subtitle | |
2912 | 445 if (!fgets (line, LINE_LEN, fd)) |
2343 | 446 return NULL; |
447 if (!(sscanf (line, "-->> %ld", &(current->start)) <1)) | |
448 break; | |
449 } | |
450 | |
451 if (previous_aqt_sub != NULL) | |
452 previous_aqt_sub->end = current->start-1; | |
453 | |
454 previous_aqt_sub = current; | |
455 | |
2912 | 456 if (!fgets (line, LINE_LEN, fd)) |
2343 | 457 return NULL; |
458 | |
2468 | 459 sub_readtext((char *) &line,¤t->text[0]); |
2343 | 460 current->lines = 1; |
461 current->end = current->start; // will be corrected by next subtitle | |
462 | |
2912 | 463 if (!fgets (line, LINE_LEN, fd)) |
2343 | 464 return current;; |
465 | |
2468 | 466 sub_readtext((char *) &line,¤t->text[1]); |
2343 | 467 current->lines = 2; |
468 | |
469 if ((current->text[0]=="") && (current->text[1]=="")) { | |
470 // void subtitle -> end of previous marked and exit | |
471 previous_aqt_sub = NULL; | |
472 return NULL; | |
473 } | |
474 | |
475 return current; | |
476 } | |
2177 | 477 |
258 | 478 int sub_autodetect (FILE *fd) { |
2912 | 479 char line[LINE_LEN+1]; |
258 | 480 int i,j=0; |
2177 | 481 char p; |
258 | 482 |
624 | 483 while (j < 100) { |
258 | 484 j++; |
2912 | 485 if (!fgets (line, LINE_LEN, fd)) |
486 return SUB_INVALID; | |
258 | 487 |
624 | 488 if (sscanf (line, "{%d}{%d}", &i, &i)==2) |
2912 | 489 {sub_uses_time=0;return SUB_MICRODVD;} |
269 | 490 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
2912 | 491 {sub_uses_time=1;return SUB_SUBRIP;} |
269 | 492 if (sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d", &i, &i, &i, &i, &i, &i, &i, &i)==8) |
2912 | 493 {sub_uses_time=1;return SUB_SUBVIEWER;} |
624 | 494 if (strstr (line, "<SAMI>")) |
2912 | 495 {sub_uses_time=1; return SUB_SAMI;} |
818 | 496 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3) |
2912 | 497 {sub_uses_time=1;return SUB_VPLAYER;} |
850 | 498 //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
|
499 // too weak test for RT |
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
500 // Please someone who knows the format of RT... FIX IT!!! |
921 | 501 // 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
|
502 if ( *line == '<' ) |
2912 | 503 {sub_uses_time=1;return SUB_RT;} |
921 | 504 |
505 if (!memcmp(line, "Dialogue: Marked", 16)) | |
2912 | 506 {sub_uses_time=1; return SUB_SSA;} |
1081 | 507 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3) |
2912 | 508 {sub_uses_time=0;return SUB_DUNNOWHAT;} |
2177 | 509 if (sscanf (line, "FORMAT=%d", &i) == 1) |
2912 | 510 {sub_uses_time=0; return SUB_MPSUB;} |
2177 | 511 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E') |
2912 | 512 {sub_uses_time=1; return SUB_MPSUB;} |
2343 | 513 if (strstr (line, "-->>")) |
2912 | 514 {sub_uses_time=0; return SUB_MPSUB;} |
258 | 515 } |
624 | 516 |
2912 | 517 return SUB_INVALID; // too many bad lines |
258 | 518 } |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
519 |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
520 #ifdef DUMPSUBS |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
521 int sub_utf8=0; |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
522 #else |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
523 extern int sub_utf8; |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
524 #endif |
258 | 525 |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
526 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
527 static iconv_t icdsc; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
528 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
529 void subcp_open (void) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
530 { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
531 char *tocp = "UTF-8"; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
532 icdsc = (iconv_t)(-1); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
533 if (sub_cp){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
534 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
|
535 printf ("SUB: opened iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
536 sub_utf8 = 2; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
537 } else |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
538 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
|
539 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
540 } |
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 void subcp_close (void) |
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 (icdsc != (iconv_t)(-1)){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
545 (void) iconv_close (icdsc); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
546 printf ("SUB: closed iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
547 } |
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 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
550 #define ICBUFFSIZE 512 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
551 static char icbuffer[ICBUFFSIZE]; |
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 subtitle* subcp_recode (subtitle *sub) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
554 { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
555 int l=sub->lines; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
556 size_t ileft, oleft, otlen; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
557 char *op, *ip, *ot; |
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 while (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
560 op = icbuffer; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
561 ip = sub->text[--l]; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
562 ileft = strlen(ip); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
563 oleft = ICBUFFSIZE - 1; |
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 (iconv(icdsc, (const char **) &ip, &ileft, |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
566 &op, &oleft) == (size_t)(-1)) { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
567 printf ("SUB: error recoding line.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
568 l++; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
569 break; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
570 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
571 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
|
572 printf ("SUB: error allocating mem.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
573 l++; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
574 break; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
575 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
576 *op='\0' ; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
577 strcpy (ot, icbuffer); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
578 free (sub->text[l]); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
579 sub->text[l] = ot; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
580 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
581 if (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
582 for (l = sub->lines; l;) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
583 free (sub->text[--l]); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
584 return ERR; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
585 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
586 return sub; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
587 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
588 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
589 #endif |
258 | 590 |
591 subtitle* sub_read_file (char *filename) { | |
592 FILE *fd; | |
593 int n_max; | |
594 subtitle *first; | |
1081 | 595 subtitle * (*func[])(FILE *fd,subtitle *dest)= |
258 | 596 { |
597 sub_read_line_microdvd, | |
598 sub_read_line_subrip, | |
2912 | 599 sub_read_line_subviewer, |
818 | 600 sub_read_line_sami, |
850 | 601 sub_read_line_vplayer, |
921 | 602 sub_read_line_rt, |
1081 | 603 sub_read_line_ssa, |
2177 | 604 sub_read_line_dunnowhat, |
2343 | 605 sub_read_line_mpsub, |
606 sub_read_line_aqt | |
607 | |
258 | 608 }; |
2915 | 609 if(filename==NULL) return NULL; //qnx segfault |
258 | 610 fd=fopen (filename, "r"); if (!fd) return NULL; |
611 | |
612 sub_format=sub_autodetect (fd); | |
2912 | 613 if (sub_format==SUB_INVALID) {printf ("SUB: Could not determine file format\n");return NULL;} |
624 | 614 printf ("SUB: Detected subtitle file format: %d\n",sub_format); |
258 | 615 |
616 rewind (fd); | |
617 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
618 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
619 subcp_open(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
620 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
621 |
258 | 622 sub_num=0;n_max=32; |
623 first=(subtitle *)malloc(n_max*sizeof(subtitle)); | |
624 if(!first) return NULL; | |
625 | |
626 while(1){ | |
627 subtitle *sub; | |
628 if(sub_num>=n_max){ | |
629 n_max+=16; | |
630 first=realloc(first,n_max*sizeof(subtitle)); | |
631 } | |
632 sub=func[sub_format](fd,&first[sub_num]); | |
633 if(!sub) break; // EOF | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
634 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
635 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
|
636 #endif |
258 | 637 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid |
638 } | |
639 | |
640 fclose(fd); | |
641 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
642 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
643 subcp_close(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
644 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
645 |
258 | 646 // printf ("SUB: Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); |
269 | 647 printf ("SUB: Read %i subtitles", sub_num); |
624 | 648 if (sub_errs) printf (", %i bad line(s).\n", sub_errs); |
269 | 649 else printf (".\n"); |
258 | 650 |
2880 | 651 if(sub_num<=0){ |
652 free(first); | |
653 return NULL; | |
654 } | |
655 | |
258 | 656 return first; |
657 } | |
658 | |
892 | 659 #if 0 |
509 | 660 char * strreplace( char * in,char * what,char * whereof ) |
661 { | |
662 int i; | |
663 char * tmp; | |
664 | |
665 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL; | |
666 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i]; | |
667 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0; | |
668 return in; | |
669 } | |
892 | 670 #endif |
509 | 671 |
892 | 672 char * sub_filename(char* path, char * fname ) |
509 | 673 { |
892 | 674 char * sub_name1; |
675 char * sub_name2; | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
676 char * aviptr1, * aviptr2, * tmp; |
892 | 677 int i,j; |
678 FILE * f; | |
679 int pos=0; | |
680 char * sub_exts[] = | |
1501
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
681 { ".utf", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
682 ".UTF", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
683 ".sub", |
509 | 684 ".SUB", |
685 ".srt", | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
686 ".SRT", |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
687 ".smi", |
850 | 688 ".SMI", |
689 ".rt", | |
690 ".RT", | |
691 ".txt", | |
1081 | 692 ".TXT", |
693 ".ssa", | |
2343 | 694 ".SSA", |
695 ".aqt", | |
696 ".AQT"}; | |
892 | 697 |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
698 |
509 | 699 if ( fname == NULL ) return NULL; |
892 | 700 |
701 sub_name1=strrchr(fname,'.'); | |
702 if (!sub_name1) return NULL; | |
703 pos=sub_name1-fname; | |
704 | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
705 sub_name1=malloc(strlen(fname)+8); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
706 strcpy(sub_name1,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 sub_name2=malloc (strlen(path) + strlen(fname) + 8); |
1081 | 709 if ((tmp=strrchr(fname,'/'))) |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
710 sprintf (sub_name2, "%s%s", path, tmp+1); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
711 else |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
712 sprintf (sub_name2, "%s%s", path, fname); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
713 |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
714 aviptr1=strrchr(sub_name1,'.'); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
715 aviptr2=strrchr(sub_name2,'.'); |
892 | 716 |
717 for(j=0;j<=1;j++){ | |
718 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
|
719 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
720 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
|
721 #else |
892 | 722 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
|
723 #endif |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
724 strcpy(j?aviptr1:aviptr2,sub_exts[i]); |
935 | 725 // printf("trying: '%s'\n",sub_name); |
892 | 726 if((f=fopen( sub_name,"rt" ))) { |
509 | 727 fclose( f ); |
728 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
|
729 if (i<2) sub_utf8=1; |
509 | 730 return sub_name; |
892 | 731 } |
509 | 732 } |
892 | 733 } |
734 | |
509 | 735 return NULL; |
736 } | |
737 | |
1761 | 738 void list_sub_file(subtitle* subs){ |
739 int i,j; | |
740 | |
741 for(j=0;j<sub_num;j++){ | |
742 subtitle* egysub=&subs[j]; | |
743 printf ("%i line%c (%li-%li) ", | |
744 egysub->lines, | |
745 (1==egysub->lines)?' ':'s', | |
746 egysub->start, | |
747 egysub->end); | |
748 for (i=0; i<egysub->lines; i++) { | |
749 printf ("%s%s",egysub->text[i], i==egysub->lines-1?"":" <BREAK> "); | |
750 } | |
751 printf ("\n"); | |
752 } | |
753 | |
754 printf ("Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); | |
755 printf ("Read %i subtitles, %i errors.\n", sub_num, sub_errs); | |
756 | |
757 } | |
758 | |
2178 | 759 void dump_mpsub(subtitle* subs){ |
760 int i,j; | |
761 FILE *fd; | |
762 float a,b; | |
763 | |
764 mpsub_position=0.0; | |
765 | |
766 fd=fopen ("dump.mpsub", "w"); | |
767 if (!fd) { | |
768 perror ("dump_mpsub: fopen"); | |
769 return; | |
770 } | |
771 | |
772 | |
773 if (sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n"); | |
2495 | 774 else fprintf (fd, "FORMAT=25\n\n"); // FIXME: fps |
2178 | 775 |
776 for(j=0;j<sub_num;j++){ | |
2495 | 777 subtitle* egysub=&subs[j]; |
778 if (sub_uses_time) { | |
779 a=((egysub->start-mpsub_position)/100.0); | |
780 b=((egysub->end-egysub->start)/100.0); | |
781 if ( (float)((int)a) == a) | |
782 fprintf (fd, "%.0f",a); | |
783 else | |
784 fprintf (fd, "%.2f",a); | |
785 | |
786 if ( (float)((int)b) == b) | |
787 fprintf (fd, " %.0f\n",b); | |
788 else | |
789 fprintf (fd, " %.2f\n",b); | |
790 } else { | |
791 fprintf (fd, "%ld %ld\n", egysub->start-mpsub_position, | |
792 egysub->end-egysub->start); | |
793 } | |
794 | |
795 mpsub_position = egysub->end; | |
796 for (i=0; i<egysub->lines; i++) { | |
797 fprintf (fd, "%s\n",egysub->text[i]); | |
798 } | |
799 fprintf (fd, "\n"); | |
2178 | 800 } |
801 fclose (fd); | |
802 printf ("Subtitles dumped in \'dump.mpsub\'.\n"); | |
803 } | |
804 | |
805 | |
806 | |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
807 #ifdef DUMPSUBS |
258 | 808 int main(int argc, char **argv) { // for testing |
809 | |
810 int i,j; | |
811 subtitle *subs; | |
812 subtitle *egysub; | |
813 | |
814 if(argc<2){ | |
815 printf("\nUsage: subreader filename.sub\n\n"); | |
816 exit(1); | |
817 } | |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
818 sub_cp = argv[2]; |
624 | 819 subs=sub_read_file(argv[1]); |
258 | 820 if(!subs){ |
821 printf("Couldn't load file... let's write a bugreport :)\n"); | |
822 exit(1); | |
823 } | |
1761 | 824 |
825 list_sub_file(subs); | |
258 | 826 |
827 return 0; | |
828 } | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
829 #endif |