src/subrip_reader.cc

   1 /*
   2     Copyright (C) 2014 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 #include "subrip_reader.h"
  21 #include "exceptions.h"
  22 #include <boost/algorithm/string.hpp>
  23 #include <boost/lexical_cast.hpp>
  24 #include <cstdio>
  25 #include <vector>
  26
  27 using std::string;
  28 using std::vector;
  29 using boost::lexical_cast;
  30 using namespace sub;
  31
  32 SubripReader::SubripReader (FILE* f)
  33 {
  34         enum {
  35                 COUNTER,
  36                 METADATA,
  37                 CONTENT
  38         } state = COUNTER;
  39
  40         char buffer[256];
  41
  42         TimePair from;
  43         TimePair to;
  44
  45         string line;
  46         int line_number = 0;
  47
  48         while (!feof (f)) {
  49                 char* r = fgets (buffer, sizeof (buffer), f);
  50                 if (r == 0 || feof (f)) {
  51                         break;
  52                 }
  53
  54                 line = string (buffer);
  55                 trim_right_if (line, boost::is_any_of ("\n\r"));
  56
  57                 switch (state) {
  58                 case COUNTER:
  59                 {
  60                         if (line.empty ()) {
  61                                 /* a blank line at the start is ok */
  62                                 break;
  63                         }
  64
  65                         state = METADATA;
  66                 }
  67                 break;
  68                 case METADATA:
  69                 {
  70                         vector<string> p;
  71                         boost::algorithm::split (p, line, boost::algorithm::is_any_of (" "));
  72                         if (p.size() != 3 && p.size() != 7) {
  73                                 throw SubripError (line, "a time/position line");
  74                         }
  75
  76                         from = convert_time (p[0]);
  77                         to = convert_time (p[2]);
  78
  79                         /* XXX: should not ignore coordinate specifications */
  80
  81                         state = CONTENT;
  82                         break;
  83                 }
  84                 case CONTENT:
  85                         if (line.empty ()) {
  86                                 state = COUNTER;
  87                                 line_number = 0;
  88                         } else {
  89                                 convert_line (line, line_number, from, to);
  90                                 line_number++;
  91                         }
  92                         break;
  93                 }
  94         }
  95 }
  96
  97 TimePair
  98 SubripReader::convert_time (string t)
  99 {
 100         vector<string> a;
 101         boost::algorithm::split (a, t, boost::is_any_of (":"));
 102         if (a.size() != 3) {
 103                 throw SubripError (t, "time in the format h:m:s,ms");
 104         }
 105
 106         vector<string> b;
 107         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 108
 109         return TimePair (
 110                 MetricTime (
 111                         lexical_cast<int> (a[0]),
 112                         lexical_cast<int> (a[1]),
 113                         lexical_cast<int> (b[0]),
 114                         lexical_cast<int> (b[1])
 115                         )
 116                 );
 117 }
 118
 119 void
 120 SubripReader::convert_line (string t, int line_number, TimePair from, TimePair to)
 121 {
 122         enum {
 123                 TEXT,
 124                 TAG
 125         } state = TEXT;
 126
 127         string tag;
 128
 129         RawSubtitle p;
 130         p.font = "Arial";
 131         p.font_size.set_points (48);
 132         p.from = from;
 133         p.to = to;
 134         p.vertical_position.proportional = 0.7 + line_number * 0.1;
 135         p.vertical_position.reference = TOP;
 136
 137         /* XXX: missing <font> support */
 138         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 139            not work, I think.
 140         */
 141
 142         for (size_t i = 0; i < t.size(); ++i) {
 143                 switch (state) {
 144                 case TEXT:
 145                         if (t[i] == '<' || t[i] == '{') {
 146                                 state = TAG;
 147                         } else {
 148                                 p.text += t[i];
 149                         }
 150                         break;
 151                 case TAG:
 152                         if (t[i] == '>' || t[i] == '}') {
 153                                 if (tag == "b") {
 154                                         maybe_content (p);
 155                                         p.bold = true;
 156                                 } else if (tag == "/b") {
 157                                         maybe_content (p);
 158                                         p.bold = false;
 159                                 } else if (tag == "i") {
 160                                         maybe_content (p);
 161                                         p.italic = true;
 162                                 } else if (tag == "/i") {
 163                                         maybe_content (p);
 164                                         p.italic = false;
 165                                 } else if (tag == "u") {
 166                                         maybe_content (p);
 167                                         p.underline = true;
 168                                 } else if (tag == "/u") {
 169                                         maybe_content (p);
 170                                         p.underline = false;
 171                                 }
 172                                 tag.clear ();
 173                                 state = TEXT;
 174                         } else {
 175                                 tag += t[i];
 176                         }
 177                         break;
 178                 }
 179         }
 180
 181         maybe_content (p);
 182 }
 183
 184 void
 185 SubripReader::maybe_content (RawSubtitle& p)
 186 {
 187         if (!p.text.empty ()) {
 188                 _subs.push_back (p);
 189                 p.text.clear ();
 190         }
 191 }