ssaparse: enhance SSA text lines parsing.
some parser will pass in the original ssa text line which starts with "Dialog:"
and there's are maybe multiple Dialog lines in one input buffer.
Upstream Status: Inappropriate [i.MX specific]
Signed-off-by: Mingke Wang <mingke.wang@freescale.com>
diff --git a/gst/subparse/gstssaparse.c b/gst/subparse/gstssaparse.c
old mode 100644
new mode 100755
index c849c08..4b9636c
--- a/gst/subparse/gstssaparse.c
+++ b/gst/subparse/gstssaparse.c
@@ -262,6 +262,7 @@
* gst_ssa_parse_push_line:
* @parse: caller element
* @txt: text to push
+ * @size: text size need to be parse
* @start: timestamp for the buffer
* @duration: duration for the buffer
*
@@ -271,27 +272,133 @@
* Returns: result of the push of the created buffer
*/
static GstFlowReturn
-gst_ssa_parse_push_line (GstSsaParse * parse, gchar * txt,
+gst_ssa_parse_push_line (GstSsaParse * parse, gchar * txt, gint size,
GstClockTime start, GstClockTime duration)
{
GstFlowReturn ret;
GstBuffer *buf;
- gchar *t, *escaped;
+ gchar *t, *text, *p, *escaped, *p_start, *p_end;
gint num, i, len;
+ GstClockTime start_time = G_MAXUINT64, end_time = 0;
- num = atoi (txt);
- GST_LOG_OBJECT (parse, "Parsing line #%d at %" GST_TIME_FORMAT,
- num, GST_TIME_ARGS (start));
-
- /* skip all non-text fields before the actual text */
+ p = text = g_malloc(size + 1);
+ *p = '\0';
t = txt;
- for (i = 0; i < 8; ++i) {
- t = strchr (t, ',');
+
+ /* there are may have multiple dialogue lines at a time */
+ while (*t) {
+ /* ignore leading white space characters */
+ while (isspace(*t))
+ t++;
+
+ /* ignore Format: and Style: lines */
+ if (strncmp(t, "Format:", 7) == 0 || strncmp(t, "Style:", 6) == 0) {
+ while (*t != '\0' && *t != '\n') {
+ t++;
+ }
+ }
+
+ if (*t == '\0')
+ break;
+
+ /* continue with next line */
+ if (*t == '\n') {
+ t++;
+ continue;
+ }
+
+ if(strncmp(t, "Dialogue:", 9) != 0) {
+ /* not started with "Dialogue:", it must be a line trimmed by demuxer */
+ num = atoi (t);
+ GST_LOG_OBJECT (parse, "Parsing line #%d at %" GST_TIME_FORMAT,
+ num, GST_TIME_ARGS (start));
+
+ /* skip all non-text fields before the actual text */
+ for (i = 0; i < 8; ++i) {
+ t = strchr (t, ',');
+ if (t == NULL)
+ break;
+ ++t;
+ }
+ } else {
+ /* started with "Dialogue:", update timestamp and duration */
+ /* time format are like Dialog:Mark,0:00:01.02,0:00:03.04,xx,xxx,... */
+ guint hour, min, sec, msec, len;
+ GstClockTime tmp;
+ gchar t_str[12] = {0};
+
+ /* find the first ',' */
+ p_start = strchr (t, ',');
+ if (p_start)
+ p_end = strchr (++p_start, ',');
+
+ if (p_start && p_end) {
+ /* copy text between first ',' and second ',' */
+ strncpy(t_str, p_start, p_end - p_start);
+ if (sscanf (t_str, "%u:%u:%u.%u", &hour, &min, &sec, &msec) == 4) {
+ tmp = ((hour*3600) + (min*60) + sec) * GST_SECOND + msec*GST_MSECOND;
+ GST_DEBUG_OBJECT (parse, "Get start time:%02d:%02d:%02d:%03d\n",
+ hour, min, sec, msec);
+ if (start_time > tmp)
+ start_time = tmp;
+ } else {
+ GST_WARNING_OBJECT (parse,
+ "failed to parse ssa start timestamp string :%s", t_str);
+ }
+
+ p_start = p_end;
+ p_end = strchr (++p_start, ',');
+ if (p_end) {
+ /* copy text between second ',' and third ',' */
+ strncpy(t_str, p_start, p_end - p_start);
+ if (sscanf (t_str, "%u:%u:%u.%u", &hour, &min, &sec, &msec) == 4) {
+ tmp = ((hour*3600) + (min*60) + sec)*GST_SECOND + msec*GST_MSECOND;
+ GST_DEBUG_OBJECT(parse, "Get end time:%02d:%02d:%02d:%03d\n",
+ hour, min, sec, msec);
+ if (end_time < tmp)
+ end_time = tmp;
+ } else {
+ GST_WARNING_OBJECT (parse,
+ "failed to parse ssa end timestamp string :%s", t_str);
+ }
+ }
+ }
+
+ /* now skip all non-text fields before the actual text */
+ for (i = 0; i <= 8; ++i) {
+ t = strchr (t, ',');
+ if (t == NULL)
+ break;
+ ++t;
+ }
+ }
+
+ /* line end before expected number of ',', not a Dialogue line */
if (t == NULL)
- return GST_FLOW_ERROR;
- ++t;
+ break;
+
+ /* if not the first line, and the last character of previous line is '\0',
+ * then replace it with '\N' */
+ if (p != text && *p == '\0') {
+ *p++ = '\\';
+ *p++ = 'N';
+ }
+
+ /* copy all actual text of this line */
+ while ((*t != '\0') && (*t != '\n'))
+ *p++ = *t++;
+
+ /* add a terminator at the end */
+ *p = '\0';
}
+ /* not valid text found in this buffer return OK to let caller unref buffer */
+ if (strlen(text) <= 0) {
+ GST_WARNING_OBJECT (parse, "Not valid text found in this buffer\n");
+ return GST_FLOW_ERROR;
+ }
+
+ t = text;
GST_LOG_OBJECT (parse, "Text : %s", t);
if (gst_ssa_parse_remove_override_codes (parse, t)) {
@@ -309,13 +416,22 @@
gst_buffer_fill (buf, 0, escaped, len + 1);
gst_buffer_set_size (buf, len);
g_free (escaped);
+ g_free(t);
- GST_BUFFER_TIMESTAMP (buf) = start;
- GST_BUFFER_DURATION (buf) = duration;
+ if (start_time != G_MAXUINT64)
+ GST_BUFFER_TIMESTAMP (buf) = start_time;
+ else
+ GST_BUFFER_TIMESTAMP (buf) = start;
+
+ if (end_time > start_time)
+ GST_BUFFER_DURATION (buf) = end_time - start_time;
+ else
+ GST_BUFFER_DURATION (buf) = duration;
GST_LOG_OBJECT (parse, "Pushing buffer with timestamp %" GST_TIME_FORMAT
- " and duration %" GST_TIME_FORMAT, GST_TIME_ARGS (start),
- GST_TIME_ARGS (duration));
+ " and duration %" GST_TIME_FORMAT,
+ GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (buf)),
+ GST_TIME_ARGS (GST_BUFFER_DURATION (buf)));
ret = gst_pad_push (parse->srcpad, buf);
@@ -335,6 +451,7 @@
GstClockTime ts;
gchar *txt;
GstMapInfo map;
+ gint size;
if (G_UNLIKELY (!parse->framed))
goto not_framed;
@@ -352,13 +469,14 @@
/* make double-sure it's 0-terminated and all */
gst_buffer_map (buf, &map, GST_MAP_READ);
txt = g_strndup ((gchar *) map.data, map.size);
+ size = map.size;
gst_buffer_unmap (buf, &map);
if (txt == NULL)
goto empty_text;
ts = GST_BUFFER_TIMESTAMP (buf);
- ret = gst_ssa_parse_push_line (parse, txt, ts, GST_BUFFER_DURATION (buf));
+ ret = gst_ssa_parse_push_line (parse, txt, size, ts, GST_BUFFER_DURATION (buf));
if (ret != GST_FLOW_OK && GST_CLOCK_TIME_IS_VALID (ts)) {
GstSegment segment;
