src/subrip_reader.cc

   1 /*
   2     Copyright (C) 2014-2020 Carl Hetherington <cth@carlh.net>
   3
   4     This program is free software; you can redistribute it and/or modify
   5     it under the terms of the GNU General Public License as published by
   6     the Free Software Foundation; either version 2 of the License, or
   7     (at your option) any later version.
   8
   9     This program is distributed in the hope that it will be useful,
  10     but WITHOUT ANY WARRANTY; without even the implied warranty of
  11     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12     GNU General Public License for more details.
  13
  14     You should have received a copy of the GNU General Public License
  15     along with this program; if not, write to the Free Software
  16     Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  17
  18 */
  19
  20 /** @file  src/subrip_reader.cc
  21  *  @brief SubripReader class.
  22  */
  23
  24 #include "subrip_reader.h"
  25 #include "exceptions.h"
  26 #include "util.h"
  27 #include "sub_assert.h"
  28 #include "raw_convert.h"
  29 #include "ssa_reader.h"
  30 #include <boost/algorithm/string.hpp>
  31 #include <boost/lexical_cast.hpp>
  32 #include <boost/regex.hpp>
  33 #include <boost/bind.hpp>
  34 #include <cstdio>
  35 #include <vector>
  36 #include <iostream>
  37
  38 using std::string;
  39 using std::vector;
  40 using std::cout;
  41 using std::hex;
  42 using boost::lexical_cast;
  43 using boost::to_upper;
  44 using boost::optional;
  45 using boost::function;
  46 using boost::algorithm::replace_all;
  47 using namespace sub;
  48
  49 /** @param s Subtitle string encoded in UTF-8 */
  50 SubripReader::SubripReader (string s)
  51 {
  52         this->read (boost::bind(&get_line_string, &s));
  53 }
  54
  55 /** @param f Subtitle file encoded in UTF-8 */
  56 SubripReader::SubripReader (FILE* f)
  57 {
  58         this->read (boost::bind (&get_line_file, f));
  59 }
  60
  61 void
  62 SubripReader::read (function<optional<string> ()> get_line)
  63 {
  64         enum {
  65                 COUNTER,
  66                 METADATA,
  67                 CONTENT
  68         } state = COUNTER;
  69
  70         RawSubtitle rs;
  71
  72         rs.vertical_position.line = 0;
  73         rs.vertical_position.reference = TOP_OF_SUBTITLE;
  74
  75         while (true) {
  76                 auto line = get_line ();
  77                 if (!line) {
  78                         break;
  79                 }
  80
  81                 trim_right_if (*line, boost::is_any_of ("\n\r"));
  82                 remove_unicode_bom (line);
  83
  84                 /* Keep some history in case there is an error to report */
  85                 _context.push_back (*line);
  86                 if (_context.size() > 5) {
  87                         _context.pop_front ();
  88                 }
  89
  90                 switch (state) {
  91                 case COUNTER:
  92                 {
  93                         if (line->empty ()) {
  94                                 /* a blank line at the start is ok */
  95                                 break;
  96                         }
  97
  98                         state = METADATA;
  99
 100                         /* Reset stuff that should not persist across separate subtitles */
 101                         rs.bold = false;
 102                         rs.italic = false;
 103                         rs.underline = false;
 104                         rs.vertical_position.line = 0;
 105                         rs.vertical_position.reference = TOP_OF_SUBTITLE;
 106                 }
 107                 break;
 108                 case METADATA:
 109                 {
 110                         vector<string> p;
 111
 112                         /* Further trim this line, removing spaces from the end */
 113                         trim_right_if (*line, boost::is_any_of (" "));
 114
 115                         boost::algorithm::split (p, *line, boost::algorithm::is_any_of (" "), boost::token_compress_on);
 116                         if (p.size() != 3 && p.size() != 7) {
 117                                 for (int i = 0; i < 2; ++i) {
 118                                         optional<string> ex = get_line ();
 119                                         if (ex) {
 120                                                 _context.push_back (*ex);
 121                                         }
 122                                 }
 123                                 throw SubripError (*line, "a time/position line", _context);
 124                         }
 125
 126                         rs.from = convert_time (p[0]);
 127                         rs.to = convert_time (p[2]);
 128
 129                         /* XXX: should not ignore coordinate specifications */
 130
 131                         state = CONTENT;
 132                         break;
 133                 }
 134                 case CONTENT:
 135                         if (line->empty ()) {
 136                                 state = COUNTER;
 137                         } else {
 138                                 convert_line (*line, rs);
 139                                 rs.vertical_position.line = rs.vertical_position.line.get() + 1;
 140                         }
 141                         break;
 142                 }
 143         }
 144 }
 145
 146 Time
 147 SubripReader::convert_time (string t)
 148 {
 149         vector<string> a;
 150         boost::algorithm::split (a, t, boost::is_any_of (":"));
 151         if (a.size() != 3) {
 152                 throw SubripError (t, "time in the format h:m:s,ms", _context);
 153         }
 154
 155         vector<string> b;
 156         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 157         if (b.size() != 2) {
 158                 throw SubripError (t, "time in the format h:m:s,ms", _context);
 159         }
 160
 161         int h, m, s, ms;
 162
 163         try {
 164                 h = lexical_cast<int>(a[0]);
 165         } catch (boost::bad_lexical_cast &) {
 166                 throw SubripError (t, "integer hour value", _context);
 167         }
 168
 169         try {
 170                 m = lexical_cast<int>(a[1]);
 171         } catch (boost::bad_lexical_cast &) {
 172                 throw SubripError (t, "integer minute value", _context);
 173         }
 174
 175         try {
 176                 s = lexical_cast<int>(b[0]);
 177         } catch (boost::bad_lexical_cast &) {
 178                 throw SubripError (t, "integer second value", _context);
 179         }
 180
 181         try {
 182                 ms = lexical_cast<int>(b[1]);
 183         } catch (boost::bad_lexical_cast &) {
 184                 throw SubripError (t, "integer millisecond value", _context);
 185         }
 186
 187         return Time::from_hms (h, m, s, ms);
 188 }
 189
 190 void
 191 SubripReader::convert_line (string t, RawSubtitle& p)
 192 {
 193         enum {
 194                 TEXT,
 195                 TAG
 196         } state = TEXT;
 197
 198         string tag;
 199
 200         vector<Colour> colours;
 201         colours.push_back (Colour (1, 1, 1));
 202
 203         /* XXX: missing <font> support */
 204         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 205            not work, I think.
 206         */
 207
 208         for (size_t i = 0; i < t.size(); ++i) {
 209                 switch (state) {
 210                 case TEXT:
 211                         if (t[i] == '<' || t[i] == '{') {
 212                                 state = TAG;
 213                         } else {
 214                                 p.text += t[i];
 215                         }
 216                         break;
 217                 case TAG:
 218                         if (t[i] == '>' || t[i] == '}') {
 219                                 if (tag == "b") {
 220                                         maybe_content (p);
 221                                         p.bold = true;
 222                                 } else if (tag == "/b") {
 223                                         maybe_content (p);
 224                                         p.bold = false;
 225                                 } else if (tag == "i") {
 226                                         maybe_content (p);
 227                                         p.italic = true;
 228                                 } else if (tag == "/i") {
 229                                         maybe_content (p);
 230                                         p.italic = false;
 231                                 } else if (tag == "u") {
 232                                         maybe_content (p);
 233                                         p.underline = true;
 234                                 } else if (tag == "/u") {
 235                                         maybe_content (p);
 236                                         p.underline = false;
 237                                 } else if (boost::starts_with (tag, "font")) {
 238                                         maybe_content (p);
 239                                         boost::regex re (".*color=\"?#([[:xdigit:]]+)\"?");
 240                                         boost::smatch match;
 241                                         if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
 242                                                 p.colour = Colour::from_rgb_hex (match[1]);
 243                                                 colours.push_back (p.colour);
 244                                         } else {
 245                                                 re = boost::regex (
 246                                                         ".*color=\"rgba\\("
 247                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 248                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 249                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 250                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*"
 251                                                         "\\)\""
 252                                                         );
 253                                                 if (boost::regex_search (tag, match, re) && match.size() == 5) {
 254                                                         p.colour.r = raw_convert<int>(string(match[1])) / 255.0;
 255                                                         p.colour.g = raw_convert<int>(string(match[2])) / 255.0;
 256                                                         p.colour.b = raw_convert<int>(string(match[3])) / 255.0;
 257                                                         colours.push_back (p.colour);
 258                                                 } else {
 259                                                         throw SubripError (tag, "a colour in the format #rrggbb or rgba(rr,gg,bb,aa)", _context);
 260                                                 }
 261                                         }
 262                                 } else if (tag == "/font") {
 263                                         maybe_content (p);
 264                                         SUB_ASSERT (!colours.empty());
 265                                         colours.pop_back ();
 266                                         p.colour = colours.back ();
 267                                 } else if (tag.size() > 0 && tag[0] == '\\') {
 268                                         SSAReader::parse_style (p, tag, 288, 288);
 269                                 }
 270                                 tag.clear ();
 271                                 state = TEXT;
 272                         } else {
 273                                 tag += tolower (t[i]);
 274                         }
 275                         break;
 276                 }
 277         }
 278
 279         /* Strip Unicode U+202B (right-to-left embedding) as sometimes it is rendered
 280            as a missing character.  This may be a hack.
 281         */
 282         replace_all (p.text, "\xe2\x80\xab", "");
 283
 284         maybe_content (p);
 285 }
 286
 287 /* Push p into _subs if it has some text, and clear the text out of p */
 288 void
 289 SubripReader::maybe_content (RawSubtitle& p)
 290 {
 291         if (!p.text.empty ()) {
 292                 _subs.push_back (p);
 293                 p.text.clear ();
 294         }
 295 }