Mercurial > mplayer.hg
annotate subreader.c @ 5406:7a8e6b6418f5
10l
author | arpi |
---|---|
date | Sat, 30 Mar 2002 18:09:30 +0000 |
parents | 1f068f4bb6e7 |
children | 880008901169 |
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 | |
3701 | 18 #define ERR ((void *) -1) |
258 | 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 |
3701 | 42 static int eol(char p) { |
624 | 43 return (p=='\r' || p=='\n' || p=='\0'); |
44 } | |
45 | |
3701 | 46 /* Remove leading and trailing space */ |
47 static void trail_space(char *s) { | |
48 int i = 0; | |
3924
9f18722fafe9
tail_space infinite loop fix by jeon_goon@lycos.co.kr
arpi
parents:
3735
diff
changeset
|
49 while (isspace(s[i])) ++i; |
3701 | 50 if (i) strcpy(s, s + i); |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
51 i = strlen(s) - 1; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
52 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
|
53 } |
624 | 54 |
2343 | 55 |
624 | 56 subtitle *sub_read_line_sami(FILE *fd, subtitle *current) { |
2912 | 57 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
|
58 static char *s = NULL, *slacktime_s; |
2912 | 59 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
|
60 int state; |
624 | 61 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
62 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
|
63 state = 0; |
624 | 64 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
65 /* read the first line */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
66 if (!s) |
2912 | 67 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
|
68 |
624 | 69 do { |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
70 switch (state) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
71 |
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
|
72 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
|
73 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
|
74 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
|
75 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
76 s = strstr (s, "Start="); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
77 if (s) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
78 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
|
79 state = 1; continue; |
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 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
82 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
83 case 1: /* find "<P" */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
84 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
|
85 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
86 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
87 case 2: /* find ">" */ |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
88 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
|
89 break; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
90 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
91 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
|
92 if (*s == '\0') break; |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
93 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
|
94 *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
|
95 if (text[0] != '\0') |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
96 current->text[current->lines++] = strdup (text); |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
97 s += 4; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
98 } |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
99 else if (*s == '<') { state = 4; } |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
100 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
|
101 else if (*s == '\t') { *p++ = ' '; s++; } |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
102 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
|
103 else *p++ = *s++; |
2836
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
104 |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
105 /* skip duplicated space */ |
ec672ea5ac2c
Applied SAMI patch by Evgeny Chukreev <codedj at echo dot ru>
atmos4
parents:
2495
diff
changeset
|
106 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
|
107 |
624 | 108 continue; |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
109 |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
110 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
|
111 q = strstr (s, "Start="); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
112 if (q) { |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
113 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
|
114 *p = '\0'; trail_space (text); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
115 if (text[0] != '\0') |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
116 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
|
117 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
|
118 state = 0; continue; |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
119 } |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
120 s = strchr (s, '>'); |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
121 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
|
122 break; |
624 | 123 } |
124 | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
125 /* read next line */ |
3735 | 126 if (state != 99 && !(s = fgets (line, LINE_LEN, fd))) { |
127 if (current->start > 0) { | |
128 break; // if it is the last subtitle | |
129 } else { | |
130 return 0; | |
131 } | |
132 } | |
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
|
133 |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
134 } while (state != 99); |
624 | 135 |
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
|
136 // 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
|
137 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
|
138 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
|
139 *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
|
140 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
|
141 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
|
142 } |
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
|
143 |
624 | 144 return current; |
145 } | |
258 | 146 |
147 | |
148 char *sub_readtext(char *source, char **dest) { | |
149 int len=0; | |
932 | 150 char *p=source; |
258 | 151 |
932 | 152 while ( !eol(*p) && *p!= '|' ) { |
153 p++,len++; | |
154 } | |
258 | 155 |
156 *dest= (char *)malloc (len+1); | |
157 if (!dest) {return ERR;} | |
158 | |
159 strncpy(*dest, source, len); | |
160 (*dest)[len]=0; | |
161 | |
162 while (*p=='\r' || *p=='\n' || *p=='|') p++; | |
163 | |
164 if (*p) return p; // not-last text field | |
165 else return NULL; // last text field | |
166 } | |
167 | |
168 subtitle *sub_read_line_microdvd(FILE *fd,subtitle *current) { | |
2912 | 169 char line[LINE_LEN+1]; |
170 char line2[LINE_LEN+1]; | |
258 | 171 char *p, *next; |
172 int i; | |
173 | |
174 do { | |
2912 | 175 if (!fgets (line, LINE_LEN, fd)) return NULL; |
4048
654419a9a228
changed subreader.c to read microdvd lines in form "{%ld}{}[^\r\n]" too
atlka
parents:
3924
diff
changeset
|
176 } while ((sscanf (line, |
654419a9a228
changed subreader.c to read microdvd lines in form "{%ld}{}[^\r\n]" too
atlka
parents:
3924
diff
changeset
|
177 "{%ld}{}%[^\r\n]", |
654419a9a228
changed subreader.c to read microdvd lines in form "{%ld}{}[^\r\n]" too
atlka
parents:
3924
diff
changeset
|
178 &(current->start), line2) < 2) && |
654419a9a228
changed subreader.c to read microdvd lines in form "{%ld}{}[^\r\n]" too
atlka
parents:
3924
diff
changeset
|
179 (sscanf (line, |
654419a9a228
changed subreader.c to read microdvd lines in form "{%ld}{}[^\r\n]" too
atlka
parents:
3924
diff
changeset
|
180 "{%ld}{%ld}%[^\r\n]", |
654419a9a228
changed subreader.c to read microdvd lines in form "{%ld}{}[^\r\n]" too
atlka
parents:
3924
diff
changeset
|
181 &(current->start), &(current->end), line2) < 3)); |
654419a9a228
changed subreader.c to read microdvd lines in form "{%ld}{}[^\r\n]" too
atlka
parents:
3924
diff
changeset
|
182 |
932 | 183 p=line2; |
258 | 184 |
185 next=p, i=0; | |
1081 | 186 while ((next =sub_readtext (next, &(current->text[i])))) { |
270 | 187 if (current->text[i]==ERR) {return ERR;} |
258 | 188 i++; |
1081 | 189 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
258 | 190 } |
932 | 191 current->lines= ++i; |
258 | 192 |
193 return current; | |
194 } | |
195 | |
196 subtitle *sub_read_line_subrip(FILE *fd, subtitle *current) { | |
2912 | 197 char line[LINE_LEN+1]; |
258 | 198 int a1,a2,a3,a4,b1,b2,b3,b4; |
199 char *p=NULL, *q=NULL; | |
200 int len; | |
201 | |
1764 | 202 while (1) { |
2912 | 203 if (!fgets (line, LINE_LEN, fd)) return NULL; |
269 | 204 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue; |
258 | 205 current->start = a1*360000+a2*6000+a3*100+a4; |
206 current->end = b1*360000+b2*6000+b3*100+b4; | |
207 | |
2912 | 208 if (!fgets (line, LINE_LEN, fd)) return NULL; |
258 | 209 |
210 p=q=line; | |
211 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) { | |
212 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && strncmp(p,"[br]",4); p++,len++); | |
213 current->text[current->lines-1]=(char *)malloc (len+1); | |
214 if (!current->text[current->lines-1]) return ERR; | |
215 strncpy (current->text[current->lines-1], q, len); | |
270 | 216 current->text[current->lines-1][len]='\0'; |
258 | 217 if (!*p || *p=='\r' || *p=='\n') break; |
218 while (*p++!=']'); | |
219 } | |
1764 | 220 break; |
258 | 221 } |
222 return current; | |
223 } | |
224 | |
2912 | 225 subtitle *sub_read_line_subviewer(FILE *fd,subtitle *current) { |
226 char line[LINE_LEN+1]; | |
258 | 227 int a1,a2,a3,a4,b1,b2,b3,b4; |
228 char *p=NULL; | |
229 int i,len; | |
230 | |
231 while (!current->text[0]) { | |
2912 | 232 if (!fgets (line, LINE_LEN, fd)) return NULL; |
269 | 233 if ((len=sscanf (line, "%d:%d:%d,%d --> %d:%d:%d,%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) |
258 | 234 continue; |
235 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
236 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
237 for (i=0; i<SUB_MAX_TEXT;) { | |
2912 | 238 if (!fgets (line, LINE_LEN, fd)) break; |
258 | 239 len=0; |
240 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++); | |
241 if (len) { | |
242 current->text[i]=(char *)malloc (len+1); | |
243 if (!current->text[i]) return ERR; | |
270 | 244 strncpy (current->text[i], line, len); current->text[i][len]='\0'; |
258 | 245 i++; |
246 } else { | |
247 break; | |
248 } | |
249 } | |
250 current->lines=i; | |
251 } | |
252 return current; | |
253 } | |
254 | |
818 | 255 subtitle *sub_read_line_vplayer(FILE *fd,subtitle *current) { |
2912 | 256 char line[LINE_LEN+1]; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
257 int a1,a2,a3; |
3735 | 258 char *p=NULL, *next,separator; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
259 int i,len,plen; |
818 | 260 |
261 while (!current->text[0]) { | |
2912 | 262 if (!fgets (line, LINE_LEN, fd)) return NULL; |
3724
a2325883c46c
vplayer format - no longer crashes on slightly broken subs.
eyck
parents:
3701
diff
changeset
|
263 if ((len=sscanf (line, "%d:%d:%d%c%n",&a1,&a2,&a3,&separator,&plen)) < 4) |
818 | 264 continue; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
265 |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
266 if (!(current->start = a1*360000+a2*6000+a3*100)) |
818 | 267 continue; |
5363
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
268 /* removed by wodzu |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
269 p=line; |
1640
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
270 // finds the body of the subtitle |
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
271 for (i=0; i<3; i++){ |
3433 | 272 p=strchr(p,':'); |
273 if (p==NULL) break; | |
274 ++p; | |
275 } | |
276 if (p==NULL) { | |
277 printf("SUB: Skipping incorrect subtitle line!\n"); | |
278 continue; | |
279 } | |
5363
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
280 */ |
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
281 // by wodzu: hey! this time we know what length it has! what is |
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
282 // that magic for? it can't deal with space instead of third |
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
283 // colon! look, what simple it can be: |
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
284 p = &line[ plen ]; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
285 |
1640
cbedcfab877b
Fixup to vplayer subtitle submitted to sourceforge by Igor Wojnicki
eyck
parents:
1501
diff
changeset
|
286 i=0; |
818 | 287 if (*p!='|') { |
288 // | |
289 next = p,i=0; | |
290 while ((next =sub_readtext (next, &(current->text[i])))) { | |
291 if (current->text[i]==ERR) {return ERR;} | |
292 i++; | |
1081 | 293 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
818 | 294 } |
295 current->lines=i+1; | |
296 } | |
297 } | |
298 return current; | |
299 } | |
300 | |
850 | 301 subtitle *sub_read_line_rt(FILE *fd,subtitle *current) { |
302 //TODO: This format uses quite rich (sub/super)set of xhtml | |
303 // I couldn't check it since DTD is not included. | |
304 // WARNING: full XML parses can be required for proper parsing | |
2912 | 305 char line[LINE_LEN+1]; |
850 | 306 int a1,a2,a3,a4,b1,b2,b3,b4; |
307 char *p=NULL,*next=NULL; | |
308 int i,len,plen; | |
309 | |
310 while (!current->text[0]) { | |
2912 | 311 if (!fgets (line, LINE_LEN, fd)) return NULL; |
850 | 312 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0 |
313 //to describe the same moment in time. Maybe there are even more formats in use. | |
314 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8) | |
315 plen=a1=a2=a3=a4=b1=b2=b3=b4=0; | |
316 if ( | |
317 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) && | |
318 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) && | |
319 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) && | |
320 ((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) && | |
321 ((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) | |
322 ) | |
323 continue; | |
324 current->start = a1*360000+a2*6000+a3*100+a4/10; | |
325 current->end = b1*360000+b2*6000+b3*100+b4/10; | |
326 p=line; p+=plen;i=0; | |
327 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml? | |
328 next = strstr(line,"<clear/>")+8;i=0; | |
329 while ((next =sub_readtext (next, &(current->text[i])))) { | |
330 if (current->text[i]==ERR) {return ERR;} | |
331 i++; | |
1081 | 332 if (i>=SUB_MAX_TEXT) { printf ("Too many lines in a subtitle\n");current->lines=i;return current;} |
850 | 333 } |
334 current->lines=i+1; | |
335 } | |
336 return current; | |
337 } | |
338 | |
921 | 339 subtitle *sub_read_line_ssa(FILE *fd,subtitle *current) { |
340 int hour1, min1, sec1, hunsec1, | |
341 hour2, min2, sec2, hunsec2, nothing; | |
2141 | 342 int num; |
921 | 343 |
2912 | 344 char line[LINE_LEN+1], |
345 line3[LINE_LEN+1], | |
2140 | 346 *line2; |
2141 | 347 char *tmp; |
348 | |
921 | 349 do { |
2912 | 350 if (!fgets (line, LINE_LEN, fd)) return NULL; |
921 | 351 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d," |
2140 | 352 "%[^\n\r]", ¬hing, |
353 &hour1, &min1, &sec1, &hunsec1, | |
354 &hour2, &min2, &sec2, &hunsec2, | |
355 line3) < 9); | |
356 line2=strstr(line3,",,"); | |
357 if (!line2) return NULL; | |
358 line2 ++; | |
359 line2 ++; | |
360 | |
2141 | 361 current->lines=1;num=0; |
921 | 362 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1; |
363 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2; | |
2141 | 364 |
3735 | 365 while ((tmp=strstr(line2, "\\n")) != NULL) { |
2141 | 366 current->text[num]=(char *)malloc(tmp-line2+1); |
367 strncpy (current->text[num], line2, tmp-line2); | |
368 current->text[num][tmp-line2]='\0'; | |
369 line2=tmp+2; | |
370 num++; | |
371 current->lines++; | |
372 if (current->lines >= SUB_MAX_TEXT) return current; | |
373 } | |
374 | |
375 | |
3701 | 376 current->text[num]=strdup(line2); |
818 | 377 |
921 | 378 return current; |
379 } | |
258 | 380 |
1081 | 381 subtitle *sub_read_line_dunnowhat(FILE *fd,subtitle *current) { |
2912 | 382 char line[LINE_LEN+1]; |
383 char text[LINE_LEN+1]; | |
1081 | 384 |
2912 | 385 if (!fgets (line, LINE_LEN, fd)) |
1081 | 386 return NULL; |
387 if (sscanf (line, "%ld,%ld,\"%[^\"]", &(current->start), | |
388 &(current->end), text) <3) | |
389 return ERR; | |
390 current->text[0] = strdup(text); | |
391 current->lines = 1; | |
392 | |
393 return current; | |
394 } | |
395 | |
2177 | 396 subtitle *sub_read_line_mpsub(FILE *fd, subtitle *current) { |
2912 | 397 char line[LINE_LEN+1]; |
2178 | 398 float a,b; |
399 int num=0; | |
2177 | 400 char *p, *q; |
401 | |
402 do | |
403 { | |
2912 | 404 if (!fgets(line, LINE_LEN, fd)) return NULL; |
2178 | 405 } while (sscanf (line, "%f %f", &a, &b) !=2); |
2177 | 406 |
2178 | 407 mpsub_position += (a*100.0); |
408 current->start=(int) mpsub_position; | |
409 mpsub_position += (b*100.0); | |
410 current->end=(int) mpsub_position; | |
2177 | 411 |
412 while (num < SUB_MAX_TEXT) { | |
4098 | 413 if (!fgets (line, LINE_LEN, fd)) { |
414 if (num == 0) return NULL; | |
415 else return current; | |
416 } | |
2177 | 417 p=line; |
418 while (isspace(*p)) p++; | |
419 if (eol(*p) && num > 0) return current; | |
420 if (eol(*p)) return NULL; | |
421 | |
422 for (q=p; !eol(*q); q++); | |
423 *q='\0'; | |
424 if (strlen(p)) { | |
425 current->text[num]=strdup(p); | |
4098 | 426 // printf (">%s<\n",p); |
2177 | 427 current->lines = ++num; |
428 } else { | |
429 if (num) return current; | |
430 else return NULL; | |
431 } | |
432 } | |
3735 | 433 return NULL; // we should have returned before if it's OK |
2177 | 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 while (1) { | |
442 // try to locate next subtitle | |
2912 | 443 if (!fgets (line, LINE_LEN, fd)) |
2343 | 444 return NULL; |
445 if (!(sscanf (line, "-->> %ld", &(current->start)) <1)) | |
446 break; | |
447 } | |
448 | |
449 if (previous_aqt_sub != NULL) | |
450 previous_aqt_sub->end = current->start-1; | |
451 | |
452 previous_aqt_sub = current; | |
453 | |
2912 | 454 if (!fgets (line, LINE_LEN, fd)) |
2343 | 455 return NULL; |
456 | |
2468 | 457 sub_readtext((char *) &line,¤t->text[0]); |
2343 | 458 current->lines = 1; |
459 current->end = current->start; // will be corrected by next subtitle | |
460 | |
2912 | 461 if (!fgets (line, LINE_LEN, fd)) |
2343 | 462 return current;; |
463 | |
2468 | 464 sub_readtext((char *) &line,¤t->text[1]); |
2343 | 465 current->lines = 2; |
466 | |
467 if ((current->text[0]=="") && (current->text[1]=="")) { | |
468 // void subtitle -> end of previous marked and exit | |
469 previous_aqt_sub = NULL; | |
470 return NULL; | |
471 } | |
472 | |
473 return current; | |
474 } | |
2177 | 475 |
258 | 476 int sub_autodetect (FILE *fd) { |
2912 | 477 char line[LINE_LEN+1]; |
258 | 478 int i,j=0; |
2177 | 479 char p; |
258 | 480 |
624 | 481 while (j < 100) { |
258 | 482 j++; |
2912 | 483 if (!fgets (line, LINE_LEN, fd)) |
484 return SUB_INVALID; | |
258 | 485 |
624 | 486 if (sscanf (line, "{%d}{%d}", &i, &i)==2) |
2912 | 487 {sub_uses_time=0;return SUB_MICRODVD;} |
4519 | 488 if (sscanf (line, "{%d}{}", &i)==1) |
4444 | 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;} |
5363
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
498 if (sscanf (line, "%d:%d:%d ", &i, &i, &i )==3) |
1f068f4bb6e7
vplayer sub fix by Arkadiusz Podgorski <wodzu@softomat.com.pl>
arpi
parents:
4886
diff
changeset
|
499 {sub_uses_time=1;return SUB_VPLAYER;} |
850 | 500 //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
|
501 // too weak test for RT |
18c43d261c35
corrected strcmp() bug, now it works again with every subs (it was broken)
laaz
parents:
896
diff
changeset
|
502 // Please someone who knows the format of RT... FIX IT!!! |
921 | 503 // 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
|
504 if ( *line == '<' ) |
2912 | 505 {sub_uses_time=1;return SUB_RT;} |
921 | 506 |
507 if (!memcmp(line, "Dialogue: Marked", 16)) | |
2912 | 508 {sub_uses_time=1; return SUB_SSA;} |
1081 | 509 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3) |
2912 | 510 {sub_uses_time=0;return SUB_DUNNOWHAT;} |
2177 | 511 if (sscanf (line, "FORMAT=%d", &i) == 1) |
2912 | 512 {sub_uses_time=0; return SUB_MPSUB;} |
2177 | 513 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E') |
2912 | 514 {sub_uses_time=1; return SUB_MPSUB;} |
2343 | 515 if (strstr (line, "-->>")) |
2912 | 516 {sub_uses_time=0; return SUB_MPSUB;} |
258 | 517 } |
624 | 518 |
2912 | 519 return SUB_INVALID; // too many bad lines |
258 | 520 } |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
521 |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
522 #ifdef DUMPSUBS |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
523 int sub_utf8=0; |
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
524 #else |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
525 extern int sub_utf8; |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
526 #endif |
258 | 527 |
4886 | 528 extern float sub_delay; |
529 extern float sub_fps; | |
530 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
531 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
532 static iconv_t icdsc; |
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 void subcp_open (void) |
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 char *tocp = "UTF-8"; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
537 icdsc = (iconv_t)(-1); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
538 if (sub_cp){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
539 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
|
540 printf ("SUB: opened iconv descriptor.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
541 sub_utf8 = 2; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
542 } else |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
543 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
|
544 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
545 } |
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 void subcp_close (void) |
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 if (icdsc != (iconv_t)(-1)){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
550 (void) iconv_close (icdsc); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
551 printf ("SUB: closed iconv descriptor.\n"); |
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 } |
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 #define ICBUFFSIZE 512 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
556 static char icbuffer[ICBUFFSIZE]; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
557 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
558 subtitle* subcp_recode (subtitle *sub) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
559 { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
560 int l=sub->lines; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
561 size_t ileft, oleft, otlen; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
562 char *op, *ip, *ot; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
563 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
564 while (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
565 op = icbuffer; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
566 ip = sub->text[--l]; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
567 ileft = strlen(ip); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
568 oleft = ICBUFFSIZE - 1; |
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 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
|
571 &op, &oleft) == (size_t)(-1)) { |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
572 printf ("SUB: error recoding line.\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 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
|
577 printf ("SUB: error allocating mem.\n"); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
578 l++; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
579 break; |
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 *op='\0' ; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
582 strcpy (ot, icbuffer); |
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 sub->text[l] = ot; |
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 if (l){ |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
587 for (l = sub->lines; l;) |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
588 free (sub->text[--l]); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
589 return ERR; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
590 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
591 return sub; |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
592 } |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
593 |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
594 #endif |
258 | 595 |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
596 static void adjust_subs_time(subtitle* sub, float subtime, float fps){ |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
597 int n,m; |
4052
505f206d80d1
corrections to adjust_subs_time function which now uses fps if needed
atlka
parents:
4051
diff
changeset
|
598 subtitle* nextsub; |
4051
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
599 int i = sub_num; |
4052
505f206d80d1
corrections to adjust_subs_time function which now uses fps if needed
atlka
parents:
4051
diff
changeset
|
600 unsigned long subfms = (sub_uses_time ? 100 : fps) * subtime; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
601 |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
602 n=m=0; |
4052
505f206d80d1
corrections to adjust_subs_time function which now uses fps if needed
atlka
parents:
4051
diff
changeset
|
603 if (i) for (;;){ |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
604 if (sub->end <= sub->start){ |
4052
505f206d80d1
corrections to adjust_subs_time function which now uses fps if needed
atlka
parents:
4051
diff
changeset
|
605 sub->end = sub->start + subfms; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
606 m++; |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
607 n++; |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
608 } |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
609 if (!--i) break; |
4051
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
610 nextsub = sub + 1; |
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
611 if (sub->end >= nextsub->start){ |
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
612 sub->end = nextsub->start - 1; |
4052
505f206d80d1
corrections to adjust_subs_time function which now uses fps if needed
atlka
parents:
4051
diff
changeset
|
613 if (sub->end - sub->start > subfms) |
505f206d80d1
corrections to adjust_subs_time function which now uses fps if needed
atlka
parents:
4051
diff
changeset
|
614 sub->end = sub->start + subfms; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
615 if (!m) |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
616 n++; |
4051
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
617 } |
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
618 sub = nextsub; |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
619 m = 0; |
4051
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
620 } |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
621 if (n) printf ("SUB: Adjusted %d subtitle(s).\n", n); |
4051
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
622 } |
0e7c382bc53a
added adjust_subs_time function which corrects bad sub->end time
atlka
parents:
4048
diff
changeset
|
623 |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
624 subtitle* sub_read_file (char *filename, float fps) { |
258 | 625 FILE *fd; |
626 int n_max; | |
627 subtitle *first; | |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
628 char *fmtname[] = { "microdvd", "subrip", "subviewer", "sami", "vplayer", |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
629 "rt", "ssa", "dunnowhat", "mpsub", "aqt" }; |
1081 | 630 subtitle * (*func[])(FILE *fd,subtitle *dest)= |
258 | 631 { |
632 sub_read_line_microdvd, | |
633 sub_read_line_subrip, | |
2912 | 634 sub_read_line_subviewer, |
818 | 635 sub_read_line_sami, |
850 | 636 sub_read_line_vplayer, |
921 | 637 sub_read_line_rt, |
1081 | 638 sub_read_line_ssa, |
2177 | 639 sub_read_line_dunnowhat, |
2343 | 640 sub_read_line_mpsub, |
641 sub_read_line_aqt | |
642 | |
258 | 643 }; |
2915 | 644 if(filename==NULL) return NULL; //qnx segfault |
258 | 645 fd=fopen (filename, "r"); if (!fd) return NULL; |
646 | |
647 sub_format=sub_autodetect (fd); | |
2912 | 648 if (sub_format==SUB_INVALID) {printf ("SUB: Could not determine file format\n");return NULL;} |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
649 printf ("SUB: Detected subtitle file format: %s\n", fmtname[sub_format]); |
258 | 650 |
651 rewind (fd); | |
652 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
653 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
654 subcp_open(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
655 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
656 |
258 | 657 sub_num=0;n_max=32; |
658 first=(subtitle *)malloc(n_max*sizeof(subtitle)); | |
659 if(!first) return NULL; | |
660 | |
661 while(1){ | |
662 subtitle *sub; | |
663 if(sub_num>=n_max){ | |
664 n_max+=16; | |
665 first=realloc(first,n_max*sizeof(subtitle)); | |
666 } | |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
667 sub = &first[sub_num]; |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
668 memset(sub, '\0', sizeof(subtitle)); |
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
669 sub=func[sub_format](fd,sub); |
258 | 670 if(!sub) break; // EOF |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
671 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
672 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
|
673 #endif |
258 | 674 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid |
675 } | |
676 | |
677 fclose(fd); | |
678 | |
2151
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
679 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
680 subcp_close(); |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
681 #endif |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
682 |
258 | 683 // printf ("SUB: Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); |
269 | 684 printf ("SUB: Read %i subtitles", sub_num); |
624 | 685 if (sub_errs) printf (", %i bad line(s).\n", sub_errs); |
269 | 686 else printf (".\n"); |
258 | 687 |
2880 | 688 if(sub_num<=0){ |
689 free(first); | |
690 return NULL; | |
691 } | |
692 | |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
693 adjust_subs_time(first, 6.0, fps); /* ~6 secs AST */ |
258 | 694 return first; |
695 } | |
696 | |
892 | 697 #if 0 |
509 | 698 char * strreplace( char * in,char * what,char * whereof ) |
699 { | |
700 int i; | |
701 char * tmp; | |
702 | |
703 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL; | |
704 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i]; | |
705 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0; | |
706 return in; | |
707 } | |
892 | 708 #endif |
509 | 709 |
892 | 710 char * sub_filename(char* path, char * fname ) |
509 | 711 { |
892 | 712 char * sub_name1; |
713 char * sub_name2; | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
714 char * aviptr1, * aviptr2, * tmp; |
892 | 715 int i,j; |
716 FILE * f; | |
717 int pos=0; | |
718 char * sub_exts[] = | |
1501
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
719 { ".utf", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
720 ".UTF", |
d40f2b686846
changes according to -utf8 option, draw_osd() function added
atlka
parents:
1081
diff
changeset
|
721 ".sub", |
509 | 722 ".SUB", |
723 ".srt", | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
724 ".SRT", |
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
725 ".smi", |
850 | 726 ".SMI", |
727 ".rt", | |
728 ".RT", | |
729 ".txt", | |
1081 | 730 ".TXT", |
731 ".ssa", | |
2343 | 732 ".SSA", |
733 ".aqt", | |
734 ".AQT"}; | |
892 | 735 |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
736 |
509 | 737 if ( fname == NULL ) return NULL; |
892 | 738 |
739 sub_name1=strrchr(fname,'.'); | |
740 if (!sub_name1) return NULL; | |
741 pos=sub_name1-fname; | |
742 | |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
743 sub_name1=malloc(strlen(fname)+8); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
744 strcpy(sub_name1,fname); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
745 |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
746 sub_name2=malloc (strlen(path) + strlen(fname) + 8); |
1081 | 747 if ((tmp=strrchr(fname,'/'))) |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
748 sprintf (sub_name2, "%s%s", path, tmp+1); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
749 else |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
750 sprintf (sub_name2, "%s%s", path, fname); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
751 |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
752 aviptr1=strrchr(sub_name1,'.'); |
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
753 aviptr2=strrchr(sub_name2,'.'); |
892 | 754 |
755 for(j=0;j<=1;j++){ | |
756 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
|
757 #ifdef USE_ICONV |
a9d91476085a
modifications to use iconv(3) function to recode text of subs (autodetect)
atlka
parents:
2141
diff
changeset
|
758 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
|
759 #else |
892 | 760 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
|
761 #endif |
934
b2c7c4b49948
Gabucino (CGA user)'s request (finds default.subs well)
laaz
parents:
932
diff
changeset
|
762 strcpy(j?aviptr1:aviptr2,sub_exts[i]); |
935 | 763 // printf("trying: '%s'\n",sub_name); |
892 | 764 if((f=fopen( sub_name,"rt" ))) { |
509 | 765 fclose( f ); |
766 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
|
767 if (i<2) sub_utf8=1; |
509 | 768 return sub_name; |
892 | 769 } |
509 | 770 } |
892 | 771 } |
772 | |
509 | 773 return NULL; |
774 } | |
775 | |
1761 | 776 void list_sub_file(subtitle* subs){ |
777 int i,j; | |
778 | |
779 for(j=0;j<sub_num;j++){ | |
780 subtitle* egysub=&subs[j]; | |
781 printf ("%i line%c (%li-%li) ", | |
782 egysub->lines, | |
783 (1==egysub->lines)?' ':'s', | |
784 egysub->start, | |
785 egysub->end); | |
786 for (i=0; i<egysub->lines; i++) { | |
787 printf ("%s%s",egysub->text[i], i==egysub->lines-1?"":" <BREAK> "); | |
788 } | |
789 printf ("\n"); | |
790 } | |
791 | |
792 printf ("Subtitle format %s time.\n", sub_uses_time?"uses":"doesn't use"); | |
793 printf ("Read %i subtitles, %i errors.\n", sub_num, sub_errs); | |
794 | |
795 } | |
796 | |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
797 void dump_mpsub(subtitle* subs, float fps){ |
2178 | 798 int i,j; |
799 FILE *fd; | |
800 float a,b; | |
801 | |
4886 | 802 mpsub_position=sub_uses_time?(sub_delay*100):(sub_delay*fps); |
803 if (sub_fps==0) sub_fps=fps; | |
2178 | 804 |
805 fd=fopen ("dump.mpsub", "w"); | |
806 if (!fd) { | |
807 perror ("dump_mpsub: fopen"); | |
808 return; | |
809 } | |
810 | |
811 | |
812 if (sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n"); | |
4064
3c747168eb6e
1. subs know are readed after reading AVI header so we already know fps
atlka
parents:
4052
diff
changeset
|
813 else fprintf (fd, "FORMAT=%5.2f\n\n", fps); |
2178 | 814 |
815 for(j=0;j<sub_num;j++){ | |
2495 | 816 subtitle* egysub=&subs[j]; |
817 if (sub_uses_time) { | |
818 a=((egysub->start-mpsub_position)/100.0); | |
819 b=((egysub->end-egysub->start)/100.0); | |
820 if ( (float)((int)a) == a) | |
821 fprintf (fd, "%.0f",a); | |
822 else | |
823 fprintf (fd, "%.2f",a); | |
824 | |
825 if ( (float)((int)b) == b) | |
826 fprintf (fd, " %.0f\n",b); | |
827 else | |
828 fprintf (fd, " %.2f\n",b); | |
829 } else { | |
4886 | 830 fprintf (fd, "%ld %ld\n", (long)((egysub->start*(fps/sub_fps))-((mpsub_position*(fps/sub_fps)))), |
831 (long)(((egysub->end)-(egysub->start))*(fps/sub_fps))); | |
2495 | 832 } |
833 | |
834 mpsub_position = egysub->end; | |
835 for (i=0; i<egysub->lines; i++) { | |
836 fprintf (fd, "%s\n",egysub->text[i]); | |
837 } | |
838 fprintf (fd, "\n"); | |
2178 | 839 } |
840 fclose (fd); | |
4880 | 841 printf ("SUB: Subtitles dumped in \'dump.mpsub\'.\n"); |
2178 | 842 } |
843 | |
3543 | 844 void sub_free( subtitle * subs ) |
845 { | |
846 int i; | |
847 | |
848 if ( !subs ) return; | |
849 | |
850 sub_num=0; | |
851 sub_errs=0; | |
852 for ( i=0;i<subs->lines;i++ ) free( subs->text[i] ); | |
853 free( subs ); | |
854 subs=NULL; | |
855 } | |
2178 | 856 |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
857 #ifdef DUMPSUBS |
258 | 858 int main(int argc, char **argv) { // for testing |
859 | |
860 int i,j; | |
861 subtitle *subs; | |
862 subtitle *egysub; | |
863 | |
864 if(argc<2){ | |
865 printf("\nUsage: subreader filename.sub\n\n"); | |
866 exit(1); | |
867 } | |
2449
7ef89d9b06ed
added DUMPSUBS recognition if we need compile only subreader.c alone
atlka
parents:
2358
diff
changeset
|
868 sub_cp = argv[2]; |
624 | 869 subs=sub_read_file(argv[1]); |
258 | 870 if(!subs){ |
4886 | 871 printf("Couldn't load file.\n"); |
258 | 872 exit(1); |
873 } | |
1761 | 874 |
875 list_sub_file(subs); | |
258 | 876 |
877 return 0; | |
878 } | |
706
8a7666a78f83
better .smi support and display two-byte characters- patch by Sunjin Yang
arpi_esp
parents:
678
diff
changeset
|
879 #endif |