src/sgml.cpp

   1 /**
   2  * \file sgml.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author José Matos
   7  * \author John Levon
   8  *
   9  * Full author contact details are available in file CREDITS.
  10  */
  11
  12 #include <config.h>
  13
  14 #include "sgml.h"
  15
  16 #include "Buffer.h"
  17 #include "BufferParams.h"
  18 #include "Counters.h"
  19 #include "Layout.h"
  20 #include "OutputParams.h"
  21 #include "Paragraph.h"
  22 #include "Text.h"
  23 #include "TextClass.h"
  24
  25 #include "support/convert.h"
  26 #include "support/docstream.h"
  27 #include "support/lstrings.h"
  28 #include "support/textutils.h"
  29
  30 #include <map>
  31
  32 using namespace std;
  33 using namespace lyx::support;
  34
  35 namespace lyx {
  36
  37
  38 docstring sgml::escapeChar(char_type c)
  39 {
  40         docstring str;
  41         switch (c) {
  42         case ' ':
  43                 str += " ";
  44                 break;
  45         case '&':
  46                 str += "&amp;";
  47                 break;
  48         case '<':
  49                 str += "&lt;";
  50                 break;
  51         case '>':
  52                 str += "&gt;";
  53                 break;
  54 #if 0
  55         case '$':
  56                 str += "&dollar;";
  57                 break;
  58         case '#':
  59                 str += "&num;";
  60                 break;
  61         case '%':
  62                 str += "&percnt;";
  63                 break;
  64         case '[':
  65                 str += "&lsqb;";
  66                 break;
  67         case ']':
  68                 str += "&rsqb;";
  69                 break;
  70         case '{':
  71                 str += "&lcub;";
  72                 break;
  73         case '}':
  74                 str += "&rcub;";
  75                 break;
  76         case '~':
  77                 str += "&tilde;";
  78                 break;
  79         case '"':
  80                 str += "&quot;";
  81                 break;
  82         case '\\':
  83                 str += "&bsol;";
  84                 break;
  85 #endif
  86         default:
  87                 str += c;
  88                 break;
  89         }
  90         return str;
  91 }
  92
  93
  94 docstring sgml::escapeString(docstring const & raw)
  95 {
  96         docstring bin;
  97         bin.reserve(raw.size() * 2); // crude approximation is sufficient
  98         for (size_t i = 0; i != raw.size(); ++i)
  99                 bin += sgml::escapeChar(raw[i]);
 100
 101         return bin;
 102 }
 103
 104
 105 docstring const sgml::uniqueID(docstring const label)
 106 {
 107         static unsigned int seed = 1000;
 108         return label + convert<docstring>(++seed);
 109 }
 110
 111
 112 docstring sgml::cleanID(Buffer const & buf, OutputParams const & runparams,
 113         docstring const & orig)
 114 {
 115         // The standard DocBook SGML declaration only allows letters,
 116         // digits, '-' and '.' in a name.
 117         // Since users might change that declaration one has to cater
 118         // for additional allowed characters.
 119         // This routine replaces illegal characters by '-' or '.'
 120         // and adds a number for uniqueness.
 121         // If you know what you are doing, you can set allowed==""
 122         // to disable this mangling.
 123         DocumentClass const & tclass = buf.params().documentClass();
 124         docstring const allowed = from_ascii(
 125                 runparams.flavor == OutputParams::XML ? ".-_:" : tclass.options());
 126
 127         if (allowed.empty())
 128                 return orig;
 129
 130         docstring::const_iterator it  = orig.begin();
 131         docstring::const_iterator end = orig.end();
 132
 133         docstring content;
 134
 135         typedef map<docstring, docstring> MangledMap;
 136         static MangledMap mangledNames;
 137         static int mangleID = 1;
 138
 139         MangledMap::const_iterator const known = mangledNames.find(orig);
 140         if (known != mangledNames.end())
 141                 return known->second;
 142
 143         // make sure it starts with a letter
 144         if (!isAlphaASCII(*it) && allowed.find(*it) >= allowed.size())
 145                 content += "x";
 146
 147         bool mangle = false;
 148         for (; it != end; ++it) {
 149                 char_type c = *it;
 150                 if (isAlphaASCII(c) || isDigitASCII(c) || c == '-' || c == '.'
 151                       || allowed.find(c) < allowed.size())
 152                         content += c;
 153                 else if (c == '_' || c == ' ') {
 154                         mangle = true;
 155                         content += "-";
 156                 }
 157                 else if (c == ':' || c == ',' || c == ';' || c == '!') {
 158                         mangle = true;
 159                         content += ".";
 160                 }
 161                 else {
 162                         mangle = true;
 163                 }
 164         }
 165
 166         if (mangle)
 167                 content += "-" + convert<docstring>(mangleID++);
 168         else if (isDigitASCII(content[content.size() - 1]))
 169                 content += ".";
 170
 171         mangledNames[orig] = content;
 172
 173         return content;
 174 }
 175
 176
 177 void sgml::openTag(odocstream & os, string const & name, string const & attribute)
 178 {
 179         // FIXME UNICODE
 180         // This should be fixed in layout files later.
 181         string param = subst(attribute, "<", "\"");
 182         param = subst(param, ">", "\"");
 183
 184         if (!name.empty() && name != "!-- --") {
 185                 os << '<' << from_ascii(name);
 186                 if (!param.empty())
 187                         os << ' ' << from_ascii(param);
 188                 os << '>';
 189         }
 190 }
 191
 192
 193 void sgml::closeTag(odocstream & os, string const & name)
 194 {
 195         if (!name.empty() && name != "!-- --")
 196                 os << "</" << from_ascii(name) << '>';
 197 }
 198
 199
 200 void sgml::openTag(Buffer const & buf, odocstream & os,
 201         OutputParams const & runparams, Paragraph const & par)
 202 {
 203         Layout const & style = par.layout();
 204         string const & name = style.latexname();
 205         string param = style.latexparam();
 206         Counters & counters = buf.params().documentClass().counters();
 207
 208         string id = par.getID(buf, runparams);
 209
 210         string attribute;
 211         if (!id.empty()) {
 212                 if (param.find('#') != string::npos) {
 213                         string::size_type pos = param.find("id=<");
 214                         string::size_type end = param.find(">");
 215                         if( pos != string::npos && end != string::npos)
 216                                 param.erase(pos, end-pos + 1);
 217                 }
 218                 attribute = id + ' ' + param;
 219         } else {
 220                 if (param.find('#') != string::npos) {
 221                         // FIXME UNICODE
 222                         if (!style.counter.empty())
 223                                 counters.step(style.counter);
 224                         else
 225                                 counters.step(from_ascii(name));
 226                         int i = counters.value(from_ascii(name));
 227                         attribute = subst(param, "#", convert<string>(i));
 228                 } else {
 229                         attribute = param;
 230                 }
 231         }
 232         openTag(os, name, attribute);
 233 }
 234
 235
 236 void sgml::closeTag(odocstream & os, Paragraph const & par)
 237 {
 238         Layout const & style = par.layout();
 239         closeTag(os, style.latexname());
 240 }
 241
 242
 243 } // namespace lyx