comparison mov.c @ 5790:dba191dda275 libavformat

In mov demuxer, convert mac encoded strings to utf-8
author bcoudurier
date Tue, 09 Mar 2010 01:03:42 +0000
parents 8539cb618ae7
children 536e5527c1e0
comparison
equal deleted inserted replaced
5789:2c96c4c8b6e9 5790:dba191dda275
91 get_be16(pb); // total tracks 91 get_be16(pb); // total tracks
92 92
93 return 0; 93 return 0;
94 } 94 }
95 95
96 static const uint32_t mac_to_unicode[128] = {
97 0x00C4,0x00C5,0x00C7,0x00C9,0x00D1,0x00D6,0x00DC,0x00E1,
98 0x00E0,0x00E2,0x00E4,0x00E3,0x00E5,0x00E7,0x00E9,0x00E8,
99 0x00EA,0x00EB,0x00ED,0x00EC,0x00EE,0x00EF,0x00F1,0x00F3,
100 0x00F2,0x00F4,0x00F6,0x00F5,0x00FA,0x00F9,0x00FB,0x00FC,
101 0x2020,0x00B0,0x00A2,0x00A3,0x00A7,0x2022,0x00B6,0x00DF,
102 0x00AE,0x00A9,0x2122,0x00B4,0x00A8,0x2260,0x00C6,0x00D8,
103 0x221E,0x00B1,0x2264,0x2265,0x00A5,0x00B5,0x2202,0x2211,
104 0x220F,0x03C0,0x222B,0x00AA,0x00BA,0x03A9,0x00E6,0x00F8,
105 0x00BF,0x00A1,0x00AC,0x221A,0x0192,0x2248,0x2206,0x00AB,
106 0x00BB,0x2026,0x00A0,0x00C0,0x00C3,0x00D5,0x0152,0x0153,
107 0x2013,0x2014,0x201C,0x201D,0x2018,0x2019,0x00F7,0x25CA,
108 0x00FF,0x0178,0x2044,0x20AC,0x2039,0x203A,0xFB01,0xFB02,
109 0x2021,0x00B7,0x201A,0x201E,0x2030,0x00C2,0x00CA,0x00C1,
110 0x00CB,0x00C8,0x00CD,0x00CE,0x00CF,0x00CC,0x00D3,0x00D4,
111 0xF8FF,0x00D2,0x00DA,0x00DB,0x00D9,0x0131,0x02C6,0x02DC,
112 0x00AF,0x02D8,0x02D9,0x02DA,0x00B8,0x02DD,0x02DB,0x02C7,
113 };
114
115 static int mov_read_mac_string(MOVContext *c, ByteIOContext *pb, int len,
116 char *dst, int dstlen)
117 {
118 char *p = dst;
119 char *end = dst+dstlen-1;
120 int i;
121
122 for (i = 0; i < len; i++) {
123 uint8_t t, c = get_byte(pb);
124 if (c < 0x80 && p < end)
125 *p++ = c;
126 else
127 PUT_UTF8(mac_to_unicode[c-0x80], t, if (p < end) *p++ = t;);
128 }
129 *p = 0;
130 return p - dst;
131 }
132
96 static int mov_read_udta_string(MOVContext *c, ByteIOContext *pb, MOVAtom atom) 133 static int mov_read_udta_string(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
97 { 134 {
98 #ifdef MOV_EXPORT_ALL_METADATA 135 #ifdef MOV_EXPORT_ALL_METADATA
99 char tmp_key[5]; 136 char tmp_key[5];
100 #endif 137 #endif
101 char str[1024], key2[16], language[4] = {0}; 138 char str[1024], key2[16], language[4] = {0};
102 const char *key = NULL; 139 const char *key = NULL;
103 uint16_t str_size; 140 uint16_t str_size, langcode = 0;
141 uint32_t data_type = 0;
104 int (*parse)(MOVContext*, ByteIOContext*, unsigned) = NULL; 142 int (*parse)(MOVContext*, ByteIOContext*, unsigned) = NULL;
105 143
106 switch (atom.type) { 144 switch (atom.type) {
107 case MKTAG(0xa9,'n','a','m'): key = "title"; break; 145 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
108 case MKTAG(0xa9,'a','u','t'): 146 case MKTAG(0xa9,'a','u','t'):
128 166
129 if (c->itunes_metadata && atom.size > 8) { 167 if (c->itunes_metadata && atom.size > 8) {
130 int data_size = get_be32(pb); 168 int data_size = get_be32(pb);
131 int tag = get_le32(pb); 169 int tag = get_le32(pb);
132 if (tag == MKTAG('d','a','t','a')) { 170 if (tag == MKTAG('d','a','t','a')) {
133 get_be32(pb); // type 171 data_type = get_be32(pb); // type
134 get_be32(pb); // unknown 172 get_be32(pb); // unknown
135 str_size = data_size - 16; 173 str_size = data_size - 16;
136 atom.size -= 16; 174 atom.size -= 16;
137 } else return 0; 175 } else return 0;
138 } else if (atom.size > 4 && key && !c->itunes_metadata) { 176 } else if (atom.size > 4 && key && !c->itunes_metadata) {
139 str_size = get_be16(pb); // string length 177 str_size = get_be16(pb); // string length
140 ff_mov_lang_to_iso639(get_be16(pb), language); 178 langcode = get_be16(pb);
179 ff_mov_lang_to_iso639(langcode, language);
141 atom.size -= 4; 180 atom.size -= 4;
142 } else 181 } else
143 str_size = atom.size; 182 str_size = atom.size;
144 183
145 #ifdef MOV_EXPORT_ALL_METADATA 184 #ifdef MOV_EXPORT_ALL_METADATA
157 str_size = FFMIN3(sizeof(str)-1, str_size, atom.size); 196 str_size = FFMIN3(sizeof(str)-1, str_size, atom.size);
158 197
159 if (parse) 198 if (parse)
160 parse(c, pb, str_size); 199 parse(c, pb, str_size);
161 else { 200 else {
162 get_buffer(pb, str, str_size); 201 if (data_type == 3 || (data_type == 0 && langcode < 0x800)) { // MAC Encoded
163 str[str_size] = 0; 202 mov_read_mac_string(c, pb, str_size, str, sizeof(str));
203 } else {
204 get_buffer(pb, str, str_size);
205 str[str_size] = 0;
206 }
164 av_metadata_set(&c->fc->metadata, key, str); 207 av_metadata_set(&c->fc->metadata, key, str);
165 if (*language && strcmp(language, "und")) { 208 if (*language && strcmp(language, "und")) {
166 snprintf(key2, sizeof(key2), "%s-%s", key, language); 209 snprintf(key2, sizeof(key2), "%s-%s", key, language);
167 av_metadata_set(&c->fc->metadata, key2, str); 210 av_metadata_set(&c->fc->metadata, key2, str);
168 } 211 }
909 dprintf(c->fc, "size=%d 4CC= %c%c%c%c codec_type=%d\n", size, 952 dprintf(c->fc, "size=%d 4CC= %c%c%c%c codec_type=%d\n", size,
910 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff, 953 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff,
911 (format >> 24) & 0xff, st->codec->codec_type); 954 (format >> 24) & 0xff, st->codec->codec_type);
912 955
913 if(st->codec->codec_type==CODEC_TYPE_VIDEO) { 956 if(st->codec->codec_type==CODEC_TYPE_VIDEO) {
914 uint8_t codec_name[32]; 957 unsigned int color_depth, len;
915 unsigned int color_depth;
916 int color_greyscale; 958 int color_greyscale;
917 959
918 st->codec->codec_id = id; 960 st->codec->codec_id = id;
919 get_be16(pb); /* version */ 961 get_be16(pb); /* version */
920 get_be16(pb); /* revision level */ 962 get_be16(pb); /* revision level */
928 get_be32(pb); /* horiz resolution */ 970 get_be32(pb); /* horiz resolution */
929 get_be32(pb); /* vert resolution */ 971 get_be32(pb); /* vert resolution */
930 get_be32(pb); /* data size, always 0 */ 972 get_be32(pb); /* data size, always 0 */
931 get_be16(pb); /* frames per samples */ 973 get_be16(pb); /* frames per samples */
932 974
933 get_buffer(pb, codec_name, 32); /* codec name, pascal string */ 975 len = get_byte(pb); /* codec name, pascal string */
934 if (codec_name[0] <= 31) { 976 if (len > 31)
935 int i; 977 len = 31;
936 int pos = 0; 978 mov_read_mac_string(c, pb, len, st->codec->codec_name, 32);
937 for (i = 0; i < codec_name[0] && pos < sizeof(st->codec->codec_name) - 3; i++) { 979 if (len < 31)
938 uint8_t tmp; 980 url_fskip(pb, 31 - len);
939 PUT_UTF8(codec_name[i+1], tmp, st->codec->codec_name[pos++] = tmp;) 981 /* codec_tag YV12 triggers an UV swap in rawdec.c */
940 } 982 if (!memcmp(st->codec->codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25))
941 st->codec->codec_name[pos] = 0; 983 st->codec->codec_tag=MKTAG('I', '4', '2', '0');
942 /* codec_tag YV12 triggers an UV swap in rawdec.c */
943 if (!memcmp(st->codec->codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25))
944 st->codec->codec_tag=MKTAG('I', '4', '2', '0');
945 }
946 984
947 st->codec->bits_per_coded_sample = get_be16(pb); /* depth */ 985 st->codec->bits_per_coded_sample = get_be16(pb); /* depth */
948 st->codec->color_table_id = get_be16(pb); /* colortable id */ 986 st->codec->color_table_id = get_be16(pb); /* colortable id */
949 dprintf(c->fc, "depth %d, ctab id %d\n", 987 dprintf(c->fc, "depth %d, ctab id %d\n",
950 st->codec->bits_per_coded_sample, st->codec->color_table_id); 988 st->codec->bits_per_coded_sample, st->codec->color_table_id);