1
mirror of https://github.com/mpv-player/mpv synced 2024-12-28 06:03:45 +01:00

subassconvert: handle unquoted attributes in subrip font tags

Previously, mplayer didn't convert tags like <font color=#00FF00>. But
such subtitles exist in the wild, and should be handled.
This commit is contained in:
wm4 2012-04-16 23:00:19 +02:00 committed by Uoti Urpala
parent aadf1002f8
commit f0ce95607f
3 changed files with 56 additions and 27 deletions

8
bstr.c
View File

@ -185,6 +185,14 @@ struct bstr *bstr_splitlines(void *talloc_ctx, struct bstr str)
return r;
}
bool bstr_eatstart(struct bstr *s, struct bstr prefix)
{
if (!bstr_startswith(*s, prefix))
return false;
*s = bstr_cut(*s, prefix.len);
return true;
}
void bstr_lower(struct bstr str)
{
for (int i = 0; i < str.len; i++)

3
bstr.h
View File

@ -82,6 +82,9 @@ int bstr_decode_utf8(struct bstr str, struct bstr *out_next);
// On error, -1 is returned. On success, it returns a value in the range [1, 4].
int bstr_parse_utf8_code_length(unsigned char b);
// If s starts with prefix, return true and return the rest of the string in s.
bool bstr_eatstart(struct bstr *s, struct bstr prefix);
static inline struct bstr bstr_cut(struct bstr str, int n)
{
if (n > str.len)

View File

@ -106,6 +106,27 @@ static const struct {
#define SUBRIP_MAX_STACKED_FONT_TAGS 16
/* Read the attribute value starting at *s, and skip *s past the value.
* Set out_value to the parsed value, with possible '"' stripped.
* Return whether the attribute is well formed. */
static bool read_value(char **s, struct bstr *out_value)
{
char term = 0;
if (**s == '"') {
term = '"';
(*s)++;
}
out_value->start = *s;
out_value->len = 0;
unsigned char *start = *s;
unsigned char *end = term ? strchr(start, term) : strpbrk(start, " >");
if (!end)
return false;
out_value->len = end - out_value->start;
*s = end + (term ? 1 : 0);
return true;
}
void subassconvert_subrip(const char *orig, char *dest, int dest_buffer_size)
{
/* line is not const to avoid warnings with strtol, etc.
@ -174,34 +195,35 @@ void subassconvert_subrip(const char *orig, char *dest, int dest_buffer_size)
line += 6;
while (*line && *line != '>') {
if (strncmp(line, "size=\"", 6) == 0) {
line += 6;
tag->size = strtol(line, &line, 10);
if (*line != '"')
if (strncmp(line, "size=", 5) == 0) {
line += 5;
struct bstr val;
if (!read_value(&line, &val))
break;
tag->size = bstrtoll(val, &val, 10);
if (val.len)
break;
append_text(&new_line, "{\\fs%d}", tag->size);
tag->has_size = true;
has_valid_attr = true;
} else if (strncmp(line, "color=\"", 7) == 0) {
line += 7;
if (*line == '#') {
} else if (strncmp(line, "color=", 6) == 0) {
line += 6;
struct bstr val;
if (!read_value(&line, &val))
break;
if (bstr_eatstart(&val, bstr("#"))) {
// #RRGGBB format
line++;
tag->color = strtol(line, &line, 16) & 0x00ffffff;
if (*line != '"')
tag->color = bstrtoll(val, &val, 16) & 0x00ffffff;
if (val.len)
break;
tag->color = ((tag->color & 0xff) << 16)
| (tag->color & 0xff00)
| ((tag->color & 0xff0000) >> 16);
} else {
// Standard web colors
int len = indexof(line, '"');
if (len <= 0)
break;
for (int i = 0; i < FF_ARRAY_ELEMS(subrip_web_colors); i++) {
char *color = subrip_web_colors[i].s;
if (strlen(color) == len
&& strncasecmp(line, color, len) == 0) {
if (bstrcasecmp(val, bstr(color)) == 0) {
tag->color = subrip_web_colors[i].v;
goto foundcolor;
}
@ -211,29 +233,25 @@ void subassconvert_subrip(const char *orig, char *dest, int dest_buffer_size)
mp_tmsg(MSGT_SUBREADER, MSGL_WARN,
"SubRip: unknown font color in subtitle: %s\n", orig);
append_text(&new_line, "{\\c}");
line += len + 1;
continue;
foundcolor:
line += len;
foundcolor: ;
}
append_text(&new_line, "{\\c&H%06X&}", tag->color);
tag->has_color = true;
has_valid_attr = true;
} else if (strncmp(line, "face=\"", 6) == 0) {
} else if (strncmp(line, "face=", 5) == 0) {
/* Font face attribute */
line += 6;
int len = indexof(line, '"');
if (len <= 0)
line += 5;
struct bstr val;
if (!read_value(&line, &val))
break;
tag->face.start = line;
tag->face.len = len;
line += len;
tag->face = val;
append_text(&new_line, "{\\fn%.*s}", BSTR_P(tag->face));
tag->has_face = true;
has_valid_attr = true;
}
line++;
} else
line++;
}
if (!has_valid_attr || *line != '>') { /* Not valid font tag */