src/Encoding.h

   1 // -*- C++ -*-
   2 /**
   3  * \file Encoding.h
   4  * This file is part of LyX, the document processor.
   5  * Licence details can be found in the file COPYING.
   6  *
   7  * \author Lars Gullik Bjønnes
   8  * \author Jean-Marc Lasgouttes
   9  *
  10  * Full author contact details are available in file CREDITS.
  11  */
  12
  13 #ifndef ENCODING_H
  14 #define ENCODING_H
  15
  16 #include "support/docstring.h"
  17 #include "support/types.h"
  18
  19 #include <map>
  20 #include <set>
  21 #include <vector>
  22
  23 namespace lyx {
  24
  25 namespace support { class FileName; }
  26
  27 class Buffer;
  28 class LaTeXFeatures;
  29
  30 class EncodingException : public std::exception {
  31 public:
  32         EncodingException(char_type c);
  33         virtual ~EncodingException() throw() {}
  34         virtual const char * what() const throw();
  35
  36         char_type failed_char;
  37         int par_id;
  38         pos_type pos;
  39 };
  40
  41
  42 ///
  43 class Encoding {
  44 public:
  45         /// Which LaTeX package handles this encoding?
  46         enum Package {
  47                 none,
  48                 inputenc,
  49                 CJK,
  50                 japanese
  51         };
  52         ///
  53         Encoding() {}
  54         ///
  55         Encoding(std::string const & n, std::string const & l,
  56                  std::string const & g, std::string const & i,
  57                  bool f, Package p);
  58         ///
  59         void init() const;
  60         ///
  61         std::string const & name() const { return name_; }
  62         ///
  63         std::string const & latexName() const { return latexName_; }
  64         ///
  65         std::string const & guiName() const { return guiName_; }
  66         ///
  67         std::string const & iconvName() const { return iconvName_; }
  68         ///
  69         bool const & hasFixedWidth() const { return fixedwidth_; }
  70         /**
  71          * Convert \p c to something that LaTeX can understand.
  72          * This is either the character itself (if it is representable
  73          * in this encoding), or a LaTeX macro.
  74          * If the character is not representable in this encoding, but no
  75          * LaTeX macro is known, a warning is given of lyxerr, and the
  76          * character is returned.
  77          */
  78         docstring latexChar(char_type c, bool for_mathed = false) const;
  79         /// Which LaTeX package handles this encoding?
  80         Package package() const { return package_; }
  81         /// A list of all characters usable in this encoding
  82         std::vector<char_type> symbolsList() const;
  83 private:
  84         ///
  85         std::string name_;
  86         ///
  87         std::string latexName_;
  88         ///
  89         std::string guiName_;
  90         ///
  91         std::string iconvName_;
  92         /// Is this a fixed width encoding?
  93         bool fixedwidth_;
  94         ///
  95         typedef std::set<char_type> CharSet;
  96         /// Set of UCS4 characters that we can encode (for singlebyte
  97         /// encodings only)
  98         mutable CharSet encodable_;
  99         /// All code points below this are encodable. This helps us to avoid
 100         /// lokup of ASCII characters in encodable_ and gives about 1 sec
 101         /// speedup on export of the Userguide.
 102         mutable char_type start_encodable_;
 103         /// Which LaTeX package handles this encoding?
 104         Package package_;
 105         /**
 106          * If this is true the stored information about the encoding covers
 107          * all encodable characters. We set this to false initially so that
 108          * we only need to query iconv for the actually used encodings.
 109          * This is needed especially for the multibyte encodings, if we
 110          * complete all encoding info on startup it takes 2-3 minutes.
 111          */
 112         mutable bool complete_;
 113 };
 114
 115 class Encodings {
 116 public:
 117         ///
 118         typedef std::set<char_type> MathCommandSet;
 119         ///
 120         typedef std::set<char_type> TextCommandSet;
 121         ///
 122         typedef std::set<char_type> MathSymbolSet;
 123         ///
 124         typedef std::map<std::string, Encoding> EncodingList;
 125         /// iterator to iterate over all encodings.
 126         /// We hide the fact that our encoding list is implemented as a map.
 127         class const_iterator : public EncodingList::const_iterator {
 128                 typedef EncodingList::const_iterator base;
 129         public:
 130                 const_iterator() : base() {}
 131                 const_iterator(base const & b) : base(b) {}
 132                 Encoding const & operator*() const { return base::operator*().second; }
 133                 Encoding const * operator->() const { return &(base::operator*().second); }
 134         };
 135         ///
 136         Encodings();
 137         /// Read the encodings.
 138         /// \param encfile encodings definition file
 139         /// \param symbolsfile unicode->LaTeX mapping file
 140         void read(support::FileName const & encfile,
 141                   support::FileName const & symbolsfile);
 142         /// Get encoding from LyX name \p name
 143         Encoding const * fromLyXName(std::string const & name) const;
 144         /// Get encoding from LaTeX name \p name
 145         Encoding const * fromLaTeXName(std::string const & name) const;
 146
 147         ///
 148         const_iterator begin() const { return encodinglist.begin(); }
 149         ///
 150         const_iterator end() const { return encodinglist.end(); }
 151
 152         ///
 153         enum LetterForm {
 154                 ///
 155                 FORM_ISOLATED,
 156                 ///
 157                 FORM_FINAL,
 158                 ///
 159                 FORM_INITIAL,
 160                 ///
 161                 FORM_MEDIAL
 162         };
 163         ///
 164         static bool isHebrewComposeChar(char_type c);
 165         ///
 166         static bool isArabicComposeChar(char_type c);
 167         ///
 168         static bool isArabicSpecialChar(char_type c);
 169         ///
 170         static bool isArabicChar(char_type c);
 171         ///
 172         static char_type transformChar(char_type c, LetterForm form);
 173         /// Is this a combining char?
 174         static bool isCombiningChar(char_type c);
 175         /**
 176          * Is this a known char from some language?
 177          * If \p preamble is empty and code point \p c is known to belong
 178          * to a supported script, true is returned and \p preamble is set
 179          * to the corresponding entry in the unicodesymbols file.
 180          * If \p preamble is not empty, a check is made whether code point
 181          * \p c is a known character matching the preamble entry.
 182          */
 183         static bool isKnownScriptChar(char_type const c, std::string & preamble);
 184         /**
 185          * Do we have to output this character as LaTeX command in any case?
 186          * This is true if the "force" flag is set.
 187          * We need this if the inputencoding does not support a certain glyph.
 188          */
 189         static bool isForced(char_type c);
 190         /**
 191          * Do we have to display in italics this character when in mathmode?
 192          * This is true if the "mathalpha" flag is set. We use this for
 193          * letters and accented characters that are output as math commands.
 194          */
 195         static bool isMathAlpha(char_type c);
 196         /**
 197          * Register \p c as a mathmode command.
 198          */
 199         static void addMathCmd(char_type c) { mathcmd.insert(c); }
 200         /**
 201          * Register \p c as a textmode command.
 202          */
 203         static void addTextCmd(char_type c) { textcmd.insert(c); }
 204         /**
 205          * Register \p c as a mathmode symbol.
 206          */
 207         static void addMathSym(char_type c) { mathsym.insert(c); }
 208         /**
 209          * Tell whether \p c is registered as a mathmode command.
 210          */
 211         static bool isMathCmd(char_type c) { return mathcmd.count(c); }
 212         /**
 213          * Tell whether \p c is registered as a textmode command.
 214          */
 215         static bool isTextCmd(char_type c) { return textcmd.count(c); }
 216         /**
 217          * Tell whether \p c is registered as a mathmode symbol.
 218          */
 219         static bool isMathSym(char_type c) { return mathsym.count(c); }
 220         /**
 221          * Initialize mathcmd, textcmd, and mathsym sets.
 222          */
 223         static void initUnicodeMath(Buffer const & buffer, bool clear_sets = true);
 224         /**
 225          * If \p c cannot be encoded in the given \p encoding, convert
 226          * it to something that LaTeX can understand in mathmode.
 227          * \return whether \p command is a mathmode command
 228          */
 229         static bool latexMathChar(char_type c, bool mathmode,
 230                         Encoding const * encoding, docstring & command);
 231         /**
 232          * Convert the LaTeX command in \p cmd to the corresponding unicode
 233          * point and set \p combining to true if it is a combining symbol
 234          */
 235         static char_type fromLaTeXCommand(docstring const & cmd, bool & combining);
 236         ///
 237         enum LatexCmd {
 238                 ///
 239                 MATH_CMD = 1,
 240                 ///
 241                 TEXT_CMD = 2
 242         };
 243         /**
 244          * Convert the LaTeX commands in \p cmd and \return a docstring
 245          * of corresponding unicode points. The conversion stops at the
 246          * first command which could not be converted, and the remaining
 247          * unconverted commands are returned in \p rem.
 248          * The \p cmdtype parameter can be used to limit recognized
 249          * commands to math or text mode commands only.
 250          */
 251         static docstring fromLaTeXCommand(docstring const & cmd,
 252                         docstring & rem, int cmdtype = MATH_CMD | TEXT_CMD);
 253         /**
 254          * Add the preamble snippet needed for the output of \p c to
 255          * \p features.
 256          * This does not depend on the used encoding, since the inputenc
 257          * package only maps the code point \p c to a command, it does not
 258          * make this command available.
 259          */
 260         static void validate(char_type c, LaTeXFeatures & features, bool for_mathed = false);
 261
 262 private:
 263         ///
 264         EncodingList encodinglist;
 265         ///
 266         static MathCommandSet mathcmd;
 267         ///
 268         static TextCommandSet textcmd;
 269         ///
 270         static MathSymbolSet mathsym;
 271 };
 272
 273 extern Encodings encodings;
 274
 275
 276 } // namespace lyx
 277
 278 #endif