GF_List *gf_webvtt_parse_iso_cues(GF_ISOSample *iso_sample, u64 start)
{
GF_List *cues;
GF_WebVTTCue *cue;
GF_VTTCueBox *cuebox;
GF_BitStream *bs;
cues = gf_list_new();
bs = gf_bs_new(iso_sample->data, iso_sample->dataLength, GF_BITSTREAM_READ);
while(gf_bs_available(bs))
{
GF_Err e;
GF_Box *box;
e = gf_isom_parse_box(&box, bs);
if (e) return NULL;
if (box->type == GF_ISOM_BOX_TYPE_VTCU) {
cuebox = (GF_VTTCueBox *)box;
cue = gf_webvtt_cue_new();
gf_list_add(cues, cue);
gf_webvtt_timestamp_set(&cue->start, start);
if (cuebox->id) {
gf_webvtt_cue_add_property(cue, WEBVTT_ID, cuebox->id->string, (u32) strlen(cuebox->id->string));
}
if (cuebox->settings) {
gf_webvtt_cue_add_property(cue, WEBVTT_SETTINGS, cuebox->settings->string, (u32) strlen(cuebox->settings->string));
}
if (cuebox->payload) {
gf_webvtt_cue_add_property(cue, WEBVTT_PAYLOAD, cuebox->payload->string, (u32) strlen(cuebox->payload->string));
}
}
gf_isom_box_del(box);
}
gf_bs_del(bs);
return cues;
}
GF_EXPORT
GF_List *gf_webvtt_parse_cues_from_data(const char *data, u32 dataLength, u64 start)
{
GF_List *cues;
GF_WebVTTCue *cue;
GF_VTTCueBox *cuebox;
GF_BitStream *bs;
char *pre_text;
cue = NULL;
pre_text = NULL;
cues = gf_list_new();
bs = gf_bs_new(data, dataLength, GF_BITSTREAM_READ);
while(gf_bs_available(bs))
{
GF_Err e;
GF_Box *box;
e = gf_isom_parse_box(&box, bs);
if (e) return NULL;
if (box->type == GF_ISOM_BOX_TYPE_VTCU) {
cuebox = (GF_VTTCueBox *)box;
cue = gf_webvtt_cue_new();
if (pre_text) {
gf_webvtt_cue_add_property(cue, WEBVTT_PRECUE_TEXT, pre_text, (u32) strlen(pre_text));
gf_free(pre_text);
pre_text = NULL;
}
gf_list_add(cues, cue);
gf_webvtt_timestamp_set(&cue->start, start);
if (cuebox->id) {
gf_webvtt_cue_add_property(cue, WEBVTT_ID, cuebox->id->string, (u32) strlen(cuebox->id->string));
}
if (cuebox->settings) {
gf_webvtt_cue_add_property(cue, WEBVTT_SETTINGS, cuebox->settings->string, (u32) strlen(cuebox->settings->string));
}
if (cuebox->payload) {
gf_webvtt_cue_add_property(cue, WEBVTT_PAYLOAD, cuebox->payload->string, (u32) strlen(cuebox->payload->string));
}
} else if (box->type == GF_ISOM_BOX_TYPE_VTTA) {
GF_StringBox *sbox = (GF_StringBox *)box;
if (cue) {
gf_webvtt_cue_add_property(cue, WEBVTT_POSTCUE_TEXT, sbox->string, (u32) strlen(sbox->string));
} else {
pre_text = gf_strdup(sbox->string);
}
}
gf_isom_box_del(box);
}
gf_bs_del(bs);
return cues;
}
GF_Err gf_webvtt_parser_parse(GF_WebVTTParser *parser, u32 duration)
{
char szLine[2048];
char *sOK;
u32 len;
GF_Err e;
Bool do_parse = GF_TRUE;
GF_WebVTTCue *cue = NULL;
u32 start = 0;
u32 end = 0;
char *prevLine = NULL;
char *header = NULL;
u32 header_len = 0;
Bool had_marks = GF_FALSE;
if (!parser) return GF_BAD_PARAM;
if (parser->is_srt) {
parser->on_header_parsed(parser->user, gf_strdup("WEBVTT\n"));
}
while (do_parse) {
sOK = gf_text_get_utf8_line(szLine, 2048, parser->vtt_in, parser->unicode_type);
REM_TRAIL_MARKS(szLine, "\r\n")
len = (u32) strlen(szLine);
switch (parser->state) {
case WEBVTT_PARSER_STATE_WAITING_SIGNATURE:
if (!sOK || len < 6 || strnicmp(szLine, "WEBVTT", 6) || (len > 6 && szLine[6] != ' ' && szLine[6] != '\t')) {
e = GF_CORRUPTED_DATA;
parser->report_message(parser->user, e, "Bad WEBVTT file signature %s", szLine);
goto exit;
} else {
if (had_marks) {
szLine[len] = '\n';
len++;
}
header = gf_strdup(szLine);
header_len = len;
parser->state = WEBVTT_PARSER_STATE_WAITING_HEADER;
}
break; /* proceed to next line */
case WEBVTT_PARSER_STATE_WAITING_HEADER:
if (prevLine) {
u32 prev_len = (u32) strlen(prevLine);
header = (char *)gf_realloc(header, header_len + prev_len + 1);
strcpy(header+header_len,prevLine);
header_len += prev_len;
gf_free(prevLine);
prevLine = NULL;
}
if (sOK && len) {
if (strstr(szLine, "-->")) {
parser->on_header_parsed(parser->user, header);
/* continue to the next state without breaking */
parser->state = WEBVTT_PARSER_STATE_WAITING_CUE_TIMESTAMP;
/* no break, continue to the next state*/
} else {
if (had_marks) {
szLine[len] = '\n';
len++;
}
prevLine = gf_strdup(szLine);
break; /* proceed to next line */
}
} else {
parser->on_header_parsed(parser->user, header);
if (!sOK) {
/* end of file, parsing is done */
do_parse = GF_FALSE;
break;
} else {
/* empty line means end of header */
parser->state = WEBVTT_PARSER_STATE_WAITING_CUE;
/* no break, continue to the next state*/
}
}
case WEBVTT_PARSER_STATE_WAITING_CUE:
if (sOK && len) {
if (strstr(szLine, "-->")) {
parser->state = WEBVTT_PARSER_STATE_WAITING_CUE_TIMESTAMP;
/* continue to the next state without breaking */
} else {
/* discard the previous line */
/* should we do something with it ? callback ?*/
if (prevLine) {
gf_free(prevLine);
prevLine = NULL;
}
/* save this new line */
if (had_marks) {
szLine[len] = '\n';
len++;
}
prevLine = gf_strdup(szLine);
/* stay in the same state */
break;
}
} else {
/* discard the previous line */
/* should we do something with it ? callback ?*/
if (prevLine) {
gf_free(prevLine);
prevLine = NULL;
}
if (!sOK) {
do_parse = GF_FALSE;
break;
} else {
/* remove empty lines and stay in the same state */
break;
}
}
case WEBVTT_PARSER_STATE_WAITING_CUE_TIMESTAMP:
if (sOK && len) {
if (cue == NULL) {
cue = gf_webvtt_cue_new();
}
if (prevLine) {
gf_webvtt_cue_add_property(cue, WEBVTT_ID, prevLine, (u32) strlen(prevLine));
gf_free(prevLine);
prevLine = NULL;
}
e = gf_webvtt_parser_parse_timings_settings(parser, cue, szLine, len);
if (e) {
if (cue) gf_webvtt_cue_del(cue);
cue = NULL;
parser->state = WEBVTT_PARSER_STATE_WAITING_CUE;
} else {
start = (u32)gf_webvtt_timestamp_get(&cue->start);
end = (u32)gf_webvtt_timestamp_get(&cue->end);
parser->state = WEBVTT_PARSER_STATE_WAITING_CUE_PAYLOAD;
}
} else {
/* not possible */
assert(0);
}
break;
case WEBVTT_PARSER_STATE_WAITING_CUE_PAYLOAD:
if (sOK && len) {
if (had_marks) {
szLine[len] = '\n';
len++;
}
gf_webvtt_cue_add_property(cue, WEBVTT_PAYLOAD, szLine, len);
/* remain in the same state as a cue payload can have multiple lines */
break;
} else {
/* end of the current cue */
gf_webvtt_add_cue_to_samples(parser, parser->samples, cue);
cue = NULL;
gf_set_progress("Importing WebVTT", gf_ftell(parser->vtt_in), parser->file_size);
if ((duration && (end >= duration)) || !sOK) {
do_parse = GF_FALSE;
break;
} else {
/* empty line, move to next cue */
parser->state = WEBVTT_PARSER_STATE_WAITING_CUE;
break;
}
}
}
if (duration && (start >= duration)) {
do_parse = GF_FALSE;
break;
}
}
/* no more cues to come, flush everything */
if (cue) {
gf_webvtt_add_cue_to_samples(parser, parser->samples, cue);
cue = NULL;
}
while (gf_list_count(parser->samples) > 0) {
GF_WebVTTSample *sample = (GF_WebVTTSample *)gf_list_get(parser->samples, 0);
parser->last_duration = sample->end - sample->start;
gf_list_rem(parser->samples, 0);
parser->on_sample_parsed(parser->user, sample);
}
gf_set_progress("Importing WebVTT", parser->file_size, parser->file_size);
e = GF_OK;
exit:
if (cue) gf_webvtt_cue_del(cue);
if (prevLine) gf_free(prevLine);
if (header) gf_free(header);
return e;
}