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::list;
  41 using std::cout;
  42 using std::hex;
  43 using boost::lexical_cast;
  44 using boost::to_upper;
  45 using boost::optional;
  46 using boost::function;
  47 using boost::algorithm::replace_all;
  48 using namespace sub;
  49
  50 /** @param s Subtitle string encoded in UTF-8 */
  51 SubripReader::SubripReader (string s)
  52 {
  53         this->read (boost::bind(&get_line_string, &s));
  54 }
  55
  56 /** @param f Subtitle file encoded in UTF-8 */
  57 SubripReader::SubripReader (FILE* f)
  58 {
  59         this->read (boost::bind (&get_line_file, f));
  60 }
  61
  62 void
  63 SubripReader::read (function<optional<string> ()> get_line)
  64 {
  65         enum {
  66                 COUNTER,
  67                 METADATA,
  68                 CONTENT
  69         } state = COUNTER;
  70
  71         RawSubtitle rs;
  72
  73         rs.vertical_position.line = 0;
  74         rs.vertical_position.reference = TOP_OF_SUBTITLE;
  75
  76         while (true) {
  77                 optional<string> line = get_line ();
  78                 if (!line) {
  79                         break;
  80                 }
  81
  82                 trim_right_if (*line, boost::is_any_of ("\n\r"));
  83                 remove_unicode_bom (line);
  84
  85                 /* Keep some history in case there is an error to report */
  86                 _context.push_back (*line);
  87                 if (_context.size() > 5) {
  88                         _context.pop_front ();
  89                 }
  90
  91                 switch (state) {
  92                 case COUNTER:
  93                 {
  94                         if (line->empty ()) {
  95                                 /* a blank line at the start is ok */
  96                                 break;
  97                         }
  98
  99                         state = METADATA;
 100
 101                         /* Reset stuff that should not persist across separate subtitles */
 102                         rs.bold = false;
 103                         rs.italic = false;
 104                         rs.underline = false;
 105                         rs.vertical_position.line = 0;
 106                         rs.vertical_position.reference = TOP_OF_SUBTITLE;
 107                 }
 108                 break;
 109                 case METADATA:
 110                 {
 111                         vector<string> p;
 112
 113                         /* Further trim this line, removing spaces from the end */
 114                         trim_right_if (*line, boost::is_any_of (" "));
 115
 116                         boost::algorithm::split (p, *line, boost::algorithm::is_any_of (" "), boost::token_compress_on);
 117                         if (p.size() != 3 && p.size() != 7) {
 118                                 for (int i = 0; i < 2; ++i) {
 119                                         optional<string> ex = get_line ();
 120                                         if (ex) {
 121                                                 _context.push_back (*ex);
 122                                         }
 123                                 }
 124                                 throw SubripError (*line, "a time/position line", _context);
 125                         }
 126
 127                         rs.from = convert_time (p[0]);
 128                         rs.to = convert_time (p[2]);
 129
 130                         /* XXX: should not ignore coordinate specifications */
 131
 132                         state = CONTENT;
 133                         break;
 134                 }
 135                 case CONTENT:
 136                         if (line->empty ()) {
 137                                 state = COUNTER;
 138                         } else {
 139                                 convert_line (*line, rs);
 140                                 rs.vertical_position.line = rs.vertical_position.line.get() + 1;
 141                         }
 142                         break;
 143                 }
 144         }
 145 }
 146
 147 Time
 148 SubripReader::convert_time (string t)
 149 {
 150         vector<string> a;
 151         boost::algorithm::split (a, t, boost::is_any_of (":"));
 152         if (a.size() != 3) {
 153                 throw SubripError (t, "time in the format h:m:s,ms", _context);
 154         }
 155
 156         vector<string> b;
 157         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 158         if (b.size() != 2) {
 159                 throw SubripError (t, "time in the format h:m:s,ms", _context);
 160         }
 161
 162         int h, m, s, ms;
 163
 164         try {
 165                 h = lexical_cast<int>(a[0]);
 166         } catch (boost::bad_lexical_cast &) {
 167                 throw SubripError (t, "integer hour value", _context);
 168         }
 169
 170         try {
 171                 m = lexical_cast<int>(a[1]);
 172         } catch (boost::bad_lexical_cast &) {
 173                 throw SubripError (t, "integer minute value", _context);
 174         }
 175
 176         try {
 177                 s = lexical_cast<int>(b[0]);
 178         } catch (boost::bad_lexical_cast &) {
 179                 throw SubripError (t, "integer second value", _context);
 180         }
 181
 182         try {
 183                 ms = lexical_cast<int>(b[1]);
 184         } catch (boost::bad_lexical_cast &) {
 185                 throw SubripError (t, "integer millisecond value", _context);
 186         }
 187
 188         return Time::from_hms (h, m, s, ms);
 189 }
 190
 191 void
 192 SubripReader::convert_line (string t, RawSubtitle& p)
 193 {
 194         enum {
 195                 TEXT,
 196                 TAG
 197         } state = TEXT;
 198
 199         string tag;
 200
 201         list<Colour> colours;
 202         colours.push_back (Colour (1, 1, 1));
 203
 204         /* XXX: missing <font> support */
 205         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 206            not work, I think.
 207         */
 208
 209         for (size_t i = 0; i < t.size(); ++i) {
 210                 switch (state) {
 211                 case TEXT:
 212                         if (t[i] == '<' || t[i] == '{') {
 213                                 state = TAG;
 214                         } else {
 215                                 p.text += t[i];
 216                         }
 217                         break;
 218                 case TAG:
 219                         if (t[i] == '>' || t[i] == '}') {
 220                                 if (tag == "b") {
 221                                         maybe_content (p);
 222                                         p.bold = true;
 223                                 } else if (tag == "/b") {
 224                                         maybe_content (p);
 225                                         p.bold = false;
 226                                 } else if (tag == "i") {
 227                                         maybe_content (p);
 228                                         p.italic = true;
 229                                 } else if (tag == "/i") {
 230                                         maybe_content (p);
 231                                         p.italic = false;
 232                                 } else if (tag == "u") {
 233                                         maybe_content (p);
 234                                         p.underline = true;
 235                                 } else if (tag == "/u") {
 236                                         maybe_content (p);
 237                                         p.underline = false;
 238                                 } else if (boost::starts_with (tag, "font")) {
 239                                         maybe_content (p);
 240                                         boost::regex re (".*color=\"?#([[:xdigit:]]+)\"?");
 241                                         boost::smatch match;
 242                                         if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
 243                                                 p.colour = Colour::from_rgb_hex (match[1]);
 244                                                 colours.push_back (p.colour);
 245                                         } else {
 246                                                 re = boost::regex (
 247                                                         ".*color=\"rgba\\("
 248                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 249                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 250                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 251                                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*"
 252                                                         "\\)\""
 253                                                         );
 254                                                 if (boost::regex_search (tag, match, re) && match.size() == 5) {
 255                                                         p.colour.r = raw_convert<int>(string(match[1])) / 255.0;
 256                                                         p.colour.g = raw_convert<int>(string(match[2])) / 255.0;
 257                                                         p.colour.b = raw_convert<int>(string(match[3])) / 255.0;
 258                                                         colours.push_back (p.colour);
 259                                                 } else {
 260                                                         throw SubripError (tag, "a colour in the format #rrggbb or rgba(rr,gg,bb,aa)", _context);
 261                                                 }
 262                                         }
 263                                 } else if (tag == "/font") {
 264                                         maybe_content (p);
 265                                         SUB_ASSERT (!colours.empty());
 266                                         colours.pop_back ();
 267                                         p.colour = colours.back ();
 268                                 } else if (tag.size() > 0 && tag[0] == '\\') {
 269                                         SSAReader::parse_style (p, tag, 288, 288);
 270                                 }
 271                                 tag.clear ();
 272                                 state = TEXT;
 273                         } else {
 274                                 tag += tolower (t[i]);
 275                         }
 276                         break;
 277                 }
 278         }
 279
 280         /* Strip Unicode U+202B (right-to-left embedding) as sometimes it is rendered
 281            as a missing character.  This may be a hack.
 282         */
 283         replace_all (p.text, "\xe2\x80\xab", "");
 284
 285         maybe_content (p);
 286 }
 287
 288 /* Push p into _subs if it has some text, and clear the text out of p */
 289 void
 290 SubripReader::maybe_content (RawSubtitle& p)
 291 {
 292         if (!p.text.empty ()) {
 293                 _subs.push_back (p);
 294                 p.text.clear ();
 295         }
 296 }