Use an enum for the SRT parser states

This commit is contained in:
Thomas Goyne 2012-12-05 07:16:10 -08:00
parent c1ed667025
commit c7c270cf12
1 changed files with 31 additions and 33 deletions

View File

@ -375,6 +375,14 @@ wxArrayString SRTSubtitleFormat::GetWriteWildcards() const {
return GetReadWildcards(); return GetReadWildcards();
} }
enum ParseState {
STATE_INITIAL,
STATE_TIMESTAMP,
STATE_FIRST_LINE_OF_BODY,
STATE_REST_OF_BODY,
STATE_LAST_WAS_BLANK
};
void SRTSubtitleFormat::ReadFile(AssFile *target, wxString const& filename, wxString const& encoding) const { void SRTSubtitleFormat::ReadFile(AssFile *target, wxString const& filename, wxString const& encoding) const {
using namespace std; using namespace std;
@ -390,7 +398,7 @@ void SRTSubtitleFormat::ReadFile(AssFile *target, wxString const& filename, wxSt
SrtTagParser tag_parser; SrtTagParser tag_parser;
int state = 1; ParseState state = STATE_INITIAL;
int line_num = 0; int line_num = 0;
int linebreak_debt = 0; int linebreak_debt = 0;
AssDialogue *line = 0; AssDialogue *line = 0;
@ -400,24 +408,20 @@ void SRTSubtitleFormat::ReadFile(AssFile *target, wxString const& filename, wxSt
text_line.Trim(true).Trim(false); text_line.Trim(true).Trim(false);
switch (state) { switch (state) {
case 1: case STATE_INITIAL:
// start of file, no subtitles found yet // ignore leading blank lines
if (text_line.empty()) if (text_line.empty()) break;
// ignore blank lines
break;
if (text_line.IsNumber()) { if (text_line.IsNumber()) {
// found the line number, throw it away and hope for timestamps // found the line number, throw it away and hope for timestamps
state = 2; state = STATE_TIMESTAMP;
break; break;
} }
if (timestamp_regex.Matches(text_line)) if (timestamp_regex.Matches(text_line))
goto found_timestamps; goto found_timestamps;
throw SRTParseError(STD_STR(wxString::Format("Parsing SRT: Expected subtitle index at line %d", line_num)), 0); throw SRTParseError(STD_STR(wxString::Format("Parsing SRT: Expected subtitle index at line %d", line_num)), 0);
case 2: case STATE_TIMESTAMP:
// want timestamps if (!timestamp_regex.Matches(text_line))
if (timestamp_regex.Matches(text_line) == false)
// bad format
throw SRTParseError(STD_STR(wxString::Format("Parsing SRT: Expected timestamp pair at line %d", line_num)), 0); throw SRTParseError(STD_STR(wxString::Format("Parsing SRT: Expected timestamp pair at line %d", line_num)), 0);
found_timestamps: found_timestamps:
if (line) { if (line) {
@ -432,41 +436,37 @@ found_timestamps:
// store pointer to subtitle, we'll continue working on it // store pointer to subtitle, we'll continue working on it
target->Line.push_back(*line); target->Line.push_back(*line);
// next we're reading the text // next we're reading the text
state = 3; state = STATE_FIRST_LINE_OF_BODY;
break; break;
case 3: case STATE_FIRST_LINE_OF_BODY:
// reading first line of subtitle text
if (text_line.empty()) { if (text_line.empty()) {
// that's not very interesting... blank subtitle? // that's not very interesting... blank subtitle?
state = 5; state = STATE_LAST_WAS_BLANK;
// no previous line that needs a line break after // no previous line that needs a line break after
linebreak_debt = 0; linebreak_debt = 0;
break; break;
} }
line->Text.Append(text_line); line->Text.Append(text_line);
state = 4; state = STATE_REST_OF_BODY;
break; break;
case 4: case STATE_REST_OF_BODY:
// reading following line of subtitle text
if (text_line.empty()) { if (text_line.empty()) {
// blank line, next may begin a new subtitle // Might be either the gap between two subtitles or just a
state = 5; // blank line in the middle of a subtitle, so defer adding
// previous line needs a line break after // the line break until we check what's on the next line
state = STATE_LAST_WAS_BLANK;
linebreak_debt = 1; linebreak_debt = 1;
break; break;
} }
line->Text.Append("\\N").Append(text_line); line->Text.Append("\\N").Append(text_line);
break; break;
case 5: case STATE_LAST_WAS_BLANK:
// blank line in subtitle text ++linebreak_debt;
linebreak_debt++; if (text_line.empty()) break;
if (text_line.empty())
// multiple blank lines in a row, just add a line break...
break;
if (text_line.IsNumber()) { if (text_line.IsNumber()) {
// must be a subtitle index! // Hopefully it's the start of a new subtitle, and the
// go for timestamps next // previous blank line(s) were the gap between subtitles
state = 2; state = STATE_TIMESTAMP;
break; break;
} }
if (timestamp_regex.Matches(text_line)) if (timestamp_regex.Matches(text_line))
@ -477,10 +477,8 @@ found_timestamps:
while (linebreak_debt-- > 0) while (linebreak_debt-- > 0)
line->Text.Append("\\N"); line->Text.Append("\\N");
line->Text.Append(text_line); line->Text.Append(text_line);
state = 4; state = STATE_REST_OF_BODY;
break; break;
default:
throw agi::InternalError(STD_STR(wxString::Format("Parsing SRT: Reached unexpected state %d", state)), 0);
} }
} }