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
  22 namespace lyx {
  23
  24 namespace support { class FileName; }
  25
  26 class LaTeXFeatures;
  27
  28 class EncodingException : public std::exception {
  29 public:
  30         EncodingException(char_type c);
  31         virtual ~EncodingException() throw() {}
  32         virtual const char * what() const throw();
  33
  34         char_type failed_char;
  35         int par_id;
  36         pos_type pos;
  37 };
  38
  39
  40 ///
  41 class Encoding {
  42 public:
  43         /// Which LaTeX package handles this encoding?
  44         enum Package {
  45                 none,
  46                 inputenc,
  47                 CJK
  48         };
  49         ///
  50         Encoding() {}
  51         ///
  52         Encoding(std::string const & n, std::string const & l,
  53                  std::string const & i, bool f, Package p);
  54         ///
  55         void init() const;
  56         ///
  57         std::string const & name() const { return Name_; }
  58         ///
  59         std::string const & latexName() const { return LatexName_; }
  60         ///
  61         std::string const & iconvName() const { return iconvName_; }
  62         /**
  63          * Convert \p c to something that LaTeX can understand.
  64          * This is either the character itself (if it is representable
  65          * in this encoding), or a LaTeX macro.
  66          * If the character is not representable in this encoding, but no
  67          * LaTeX macro is known, a warning is given of lyxerr, and the
  68          * character is returned.
  69          */
  70         docstring const latexChar(char_type c) const;
  71         /// Which LaTeX package handles this encoding?
  72         Package package() const { return package_; }
  73         /// A list of all characters usable in this encoding
  74         std::set<char_type> getSymbolsList() const;
  75 private:
  76         ///
  77         std::string Name_;
  78         ///
  79         std::string LatexName_;
  80         ///
  81         std::string iconvName_;
  82         /// Is this a fixed width encoding?
  83         bool fixedwidth_;
  84         ///
  85         typedef std::set<char_type> CharSet;
  86         /// Set of UCS4 characters that we can encode (for singlebyte
  87         /// encodings only)
  88         mutable CharSet encodable_;
  89         /// All code points below this are encodable. This helps us to avoid
  90         /// lokup of ASCII characters in encodable_ and gives about 1 sec
  91         /// speedup on export of the Userguide.
  92         mutable char_type start_encodable_;
  93         /// Which LaTeX package handles this encoding?
  94         Package package_;
  95         /**
  96          * If this is true the stored information about the encoding covers
  97          * all encodable characters. We set this to false initially so that
  98          * we only need to query iconv for the actually used encodings.
  99          * This is needed especially for the multibyte encodings, if we
 100          * complete all encoding info on startup it takes 2-3 minutes.
 101          */
 102         mutable bool complete_;
 103 };
 104
 105 class Encodings {
 106 public:
 107         ///
 108         typedef std::map<std::string, Encoding> EncodingList;
 109         /// iterator to iterate over all encodings.
 110         /// We hide the fact that our encoding list is implemented as a map.
 111         class const_iterator : public EncodingList::const_iterator {
 112                 typedef EncodingList::const_iterator base;
 113         public:
 114                 const_iterator() : base() {}
 115                 const_iterator(base const & b) : base(b) {}
 116                 Encoding const & operator*() const { return base::operator*().second; }
 117                 Encoding const * operator->() const { return &(base::operator*().second); }
 118         };
 119         ///
 120         Encodings();
 121         /// Read the encodings.
 122         /// \param encfile encodings definition file
 123         /// \param symbolsfile unicode->LaTeX mapping file
 124         void read(support::FileName const & encfile,
 125                   support::FileName const & symbolsfile);
 126         /// Get encoding from LyX name \p name
 127         Encoding const * getFromLyXName(std::string const & name) const;
 128         /// Get encoding from LaTeX name \p name
 129         Encoding const * getFromLaTeXName(std::string const & name) const;
 130
 131         ///
 132         const_iterator begin() const { return encodinglist.begin(); }
 133         ///
 134         const_iterator end() const { return encodinglist.end(); }
 135
 136         ///
 137         enum Letter_Form {
 138                 ///
 139                 FORM_ISOLATED,
 140                 ///
 141                 FORM_FINAL,
 142                 ///
 143                 FORM_INITIAL,
 144                 ///
 145                 FORM_MEDIAL
 146         };
 147         ///
 148         static bool isComposeChar_hebrew(char_type c);
 149         ///
 150         static bool isComposeChar_arabic(char_type c);
 151         ///
 152         static bool is_arabic_special(char_type c);
 153         ///
 154         static bool is_arabic(char_type c);
 155         ///
 156         static char_type transformChar(char_type c, Letter_Form form);
 157         /// Is this a combining char?
 158         static bool isCombiningChar(char_type c);
 159         /**
 160          * Is this a known char from some language?
 161          * If \p preamble is empty and code point \p c is known to belong
 162          * to a supported script, true is returned and \p preamble is set
 163          * to the corresponding entry in the unicodesymbols file.
 164          * If \p preamble is not empty, a check is made whether code point
 165          * \p c is a known character matching the preamble entry.
 166          */
 167         static bool isKnownScriptChar(char_type const c, std::string & preamble);
 168         /**
 169          * Add the preamble snippet needed for the output of \p c to
 170          * \p features.
 171          * This does not depend on the used encoding, since the inputenc
 172          * package only maps the code point \p c to a command, it does not
 173          * make this command available.
 174          */
 175         static void validate(char_type c, LaTeXFeatures & features);
 176
 177 private:
 178         ///
 179         EncodingList encodinglist;
 180 };
 181
 182 extern Encodings encodings;
 183
 184
 185 } // namespace lyx
 186
 187 #endif