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 LaTeXFeatures;
  28
  29 class EncodingException : public std::exception {
  30 public:
  31         EncodingException(char_type c);
  32         virtual ~EncodingException() throw() {}
  33         virtual const char * what() const throw();
  34
  35         char_type failed_char;
  36         int par_id;
  37         pos_type pos;
  38 };
  39
  40
  41 ///
  42 class Encoding {
  43 public:
  44         /// Which LaTeX package handles this encoding?
  45         enum Package {
  46                 none,
  47                 inputenc,
  48                 CJK
  49         };
  50         ///
  51         Encoding() {}
  52         ///
  53         Encoding(std::string const & n, std::string const & l,
  54                  std::string const & g, std::string const & i,
  55                  bool f, Package p);
  56         ///
  57         void init() const;
  58         ///
  59         std::string const & name() const { return name_; }
  60         ///
  61         std::string const & latexName() const { return latexName_; }
  62         ///
  63         std::string const & guiName() const { return guiName_; }
  64         ///
  65         std::string const & iconvName() const { return iconvName_; }
  66         ///
  67         bool const & hasFixedWidth() const { return fixedwidth_; }
  68         /**
  69          * Convert \p c to something that LaTeX can understand.
  70          * This is either the character itself (if it is representable
  71          * in this encoding), or a LaTeX macro.
  72          * If the character is not representable in this encoding, but no
  73          * LaTeX macro is known, a warning is given of lyxerr, and the
  74          * character is returned.
  75          */
  76         docstring latexChar(char_type c, bool for_mathed = false) const;
  77         /// Which LaTeX package handles this encoding?
  78         Package package() const { return package_; }
  79         /// A list of all characters usable in this encoding
  80         std::vector<char_type> symbolsList() const;
  81 private:
  82         ///
  83         std::string name_;
  84         ///
  85         std::string latexName_;
  86         ///
  87         std::string guiName_;
  88         ///
  89         std::string iconvName_;
  90         /// Is this a fixed width encoding?
  91         bool fixedwidth_;
  92         ///
  93         typedef std::set<char_type> CharSet;
  94         /// Set of UCS4 characters that we can encode (for singlebyte
  95         /// encodings only)
  96         mutable CharSet encodable_;
  97         /// All code points below this are encodable. This helps us to avoid
  98         /// lokup of ASCII characters in encodable_ and gives about 1 sec
  99         /// speedup on export of the Userguide.
 100         mutable char_type start_encodable_;
 101         /// Which LaTeX package handles this encoding?
 102         Package package_;
 103         /**
 104          * If this is true the stored information about the encoding covers
 105          * all encodable characters. We set this to false initially so that
 106          * we only need to query iconv for the actually used encodings.
 107          * This is needed especially for the multibyte encodings, if we
 108          * complete all encoding info on startup it takes 2-3 minutes.
 109          */
 110         mutable bool complete_;
 111 };
 112
 113 class Encodings {
 114 public:
 115         ///
 116         typedef std::map<std::string, Encoding> EncodingList;
 117         /// iterator to iterate over all encodings.
 118         /// We hide the fact that our encoding list is implemented as a map.
 119         class const_iterator : public EncodingList::const_iterator {
 120                 typedef EncodingList::const_iterator base;
 121         public:
 122                 const_iterator() : base() {}
 123                 const_iterator(base const & b) : base(b) {}
 124                 Encoding const & operator*() const { return base::operator*().second; }
 125                 Encoding const * operator->() const { return &(base::operator*().second); }
 126         };
 127         ///
 128         Encodings();
 129         /// Read the encodings.
 130         /// \param encfile encodings definition file
 131         /// \param symbolsfile unicode->LaTeX mapping file
 132         void read(support::FileName const & encfile,
 133                   support::FileName const & symbolsfile);
 134         /// Get encoding from LyX name \p name
 135         Encoding const * fromLyXName(std::string const & name) const;
 136         /// Get encoding from LaTeX name \p name
 137         Encoding const * fromLaTeXName(std::string const & name) const;
 138
 139         ///
 140         const_iterator begin() const { return encodinglist.begin(); }
 141         ///
 142         const_iterator end() const { return encodinglist.end(); }
 143
 144         ///
 145         enum LetterForm {
 146                 ///
 147                 FORM_ISOLATED,
 148                 ///
 149                 FORM_FINAL,
 150                 ///
 151                 FORM_INITIAL,
 152                 ///
 153                 FORM_MEDIAL
 154         };
 155         ///
 156         static bool isHebrewComposeChar(char_type c);
 157         ///
 158         static bool isArabicComposeChar(char_type c);
 159         ///
 160         static bool isArabicSpecialChar(char_type c);
 161         ///
 162         static bool isArabicChar(char_type c);
 163         ///
 164         static char_type transformChar(char_type c, LetterForm form);
 165         /// Is this a combining char?
 166         static bool isCombiningChar(char_type c);
 167         /**
 168          * Is this a known char from some language?
 169          * If \p preamble is empty and code point \p c is known to belong
 170          * to a supported script, true is returned and \p preamble is set
 171          * to the corresponding entry in the unicodesymbols file.
 172          * If \p preamble is not empty, a check is made whether code point
 173          * \p c is a known character matching the preamble entry.
 174          */
 175         static bool isKnownScriptChar(char_type const c, std::string & preamble);
 176         /**
 177          * Do we have to output this character as LaTeX command in any case?
 178          * This is true if the "forced" flag is set.
 179          * We need this if the inputencoding does not support a certain glyph.
 180          */
 181         static bool isForced(char_type c);
 182         /**
 183          * If \p c cannot be encoded in the given \p encoding, convert
 184          * it to something that LaTeX can understand in math mode.
 185          * \return whether \p command is a math mode command
 186          */
 187         static bool latexMathChar(char_type c, Encoding const * encoding, docstring & command);
 188
 189         /**
 190          * Convert the LaTeX command in \p cmd to the corresponding unicode
 191          * point and set \p combining to true if it is a combining symbol
 192          */
 193         static char_type fromLaTeXCommand(docstring const & cmd, bool & combining);
 194         /**
 195          * Convert the LaTeX commands in \p cmd and \return a docstring
 196          * of corresponding unicode points. The conversion stops at the
 197          * first command which could not be converted, and the remaining
 198          * unconverted commands are returned in \p rem
 199          */
 200         static docstring fromLaTeXCommand(docstring const & cmd, docstring & rem);
 201         /**
 202          * Add the preamble snippet needed for the output of \p c to
 203          * \p features.
 204          * This does not depend on the used encoding, since the inputenc
 205          * package only maps the code point \p c to a command, it does not
 206          * make this command available.
 207          */
 208         static void validate(char_type c, LaTeXFeatures & features, bool for_mathed = false);
 209
 210 private:
 211         ///
 212         EncodingList encodinglist;
 213 };
 214
 215 extern Encodings encodings;
 216
 217
 218 } // namespace lyx
 219
 220 #endif