]> git.lyx.org Git - lyx.git/blobdiff - src/tex2lyx/text.cpp
Cmake build: support tests
[lyx.git] / src / tex2lyx / text.cpp
index 6d32c8b742be4bf3d21222f11fa9145e0c53e305..7f7603c198b244057a8495edc381fd763e47db23 100644 (file)
@@ -117,43 +117,11 @@ char const * const known_ref_commands[] = { "ref", "pageref", "vref",
 char const * const known_coded_ref_commands[] = { "ref", "pageref", "vref",
  "vpageref", "formatted", "eqref", 0 };
 
-/**
- * known polyglossia language names (including variants)
- */
-const char * const polyglossia_languages[] = {
-"albanian", "croatian", "hebrew", "norsk", "swedish", "amharic", "czech", "hindi",
-"nynorsk", "syriac", "arabic", "danish", "icelandic", "occitan", "tamil",
-"armenian", "divehi", "interlingua", "polish", "telugu", "asturian", "dutch",
-"irish", "portuges", "thai", "bahasai", "english", "italian", "romanian", "turkish",
-"bahasam", "esperanto", "lao", "russian", "turkmen", "basque", "estonian", "latin",
-"samin", "ukrainian", "bengali", "farsi", "latvian", "sanskrit", "urdu", "brazil",
-"brazilian", "finnish", "lithuanian", "scottish", "usorbian", "breton", "french",
-"lsorbian", "serbian", "vietnamese", "bulgarian", "galician", "magyar", "slovak",
-"welsh", "catalan", "german", "malayalam", "slovenian", "coptic", "greek",
-"marathi", "spanish",
-"american", "ancient", "australian", "british", "monotonic", "newzealand",
-"polytonic", 0};
-
-/**
- * the same as polyglossia_languages with .lyx names
- * please keep this in sync with polyglossia_languages line by line!
- */
-const char * const coded_polyglossia_languages[] = {
-"albanian", "croatian", "hebrew", "norsk", "swedish", "amharic", "czech", "hindi",
-"nynorsk", "syriac", "arabic_arabi", "danish", "icelandic", "occitan", "tamil",
-"armenian", "divehi", "interlingua", "polish", "telugu", "asturian", "dutch",
-"irish", "portuges", "thai", "bahasa", "english", "italian", "romanian", "turkish",
-"bahasam", "esperanto", "lao", "russian", "turkmen", "basque", "estonian", "latin",
-"samin", "ukrainian", "bengali", "farsi", "latvian", "sanskrit", "urdu", "brazilian",
-"brazilian", "finnish", "lithuanian", "scottish", "uppersorbian", "breton", "french",
-"lowersorbian", "serbian", "vietnamese", "bulgarian", "galician", "magyar", "slovak",
-"welsh", "catalan", "ngerman", "malayalam", "slovene", "coptic", "greek",
-"marathi", "spanish",
-"american", "ancientgreek", "australian", "british", "greek", "newzealand",
-"polutonikogreek", 0};
-
 /**
  * supported CJK encodings
+ * SJIS and Bg5 cannot be supported as they are not
+ * supported by iconv
+ * JIS does not work with LyX's encoding conversion
  */
 const char * const supported_CJK_encodings[] = {
 "EUC-JP", "KS", "GB", "UTF8", 0};
@@ -162,17 +130,9 @@ const char * const supported_CJK_encodings[] = {
  * the same as supported_CJK_encodings with their corresponding LyX language name
  * please keep this in sync with supported_CJK_encodings line by line!
  */
-const char * const coded_supported_CJK_encodings[] = {
+const char * const supported_CJK_languages[] = {
 "japanese-cjk", "korean", "chinese-simplified", "chinese-traditional", 0};
 
-string CJK2lyx(string const & encoding)
-{
-       char const * const * where = is_known(encoding, supported_CJK_encodings);
-       if (where)
-               return coded_supported_CJK_encodings[where - supported_CJK_encodings];
-       return encoding;
-}
-
 /*!
  * natbib commands.
  * The starred forms are also known except for "citefullauthor",
@@ -854,8 +814,18 @@ void parse_box(Parser & p, ostream & os, unsigned outer_flags,
                        if (inner_type != "makebox") {
                                latex_height = p.getArg('[', ']');
                                translate_box_len(latex_height, height_value, height_unit, height_special);
-                       } else
-                               hor_pos = p.getArg('[', ']');
+                       } else {
+                               string const opt = p.getArg('[', ']');
+                               if (!opt.empty()) {
+                                       hor_pos = opt;
+                                       if (hor_pos != "l" && hor_pos != "c" &&
+                                           hor_pos != "r" && hor_pos != "s") {
+                                               cerr << "invalid hor_pos " << hor_pos
+                                                    << " for " << inner_type << endl;
+                                               hor_pos = "c";
+                                       }
+                               }
+                       }
 
                        if (p.hasOpt()) {
                                inner_pos = p.getArg('[', ']');
@@ -879,7 +849,7 @@ void parse_box(Parser & p, ostream & os, unsigned outer_flags,
                        if (!opt.empty()) {
                                hor_pos = opt;
                                if (hor_pos != "l" && hor_pos != "c" &&
-                                   hor_pos != "r") {
+                                   hor_pos != "r" && hor_pos != "s") {
                                        cerr << "invalid hor_pos " << hor_pos
                                             << " for " << outer_type << endl;
                                        hor_pos = "c";
@@ -1268,7 +1238,7 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                }
        }
 
-       else if (is_known(name, polyglossia_languages)) {
+       else if (is_known(name, preamble.polyglossia_languages)) {
                // We must begin a new paragraph if not already done
                if (! parent_context.atParagraphStart()) {
                        parent_context.check_end_layout(os);
@@ -1276,7 +1246,7 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                }
                // save the language in the context so that it is
                // handled by parse_text
-               parent_context.font.language = polyglossia2lyx(name);
+               parent_context.font.language = preamble.polyglossia2lyx(name);
                parse_text(p, os, FLAG_END, outer, parent_context);
                // Just in case the environment is empty
                parent_context.extra_stuff.erase();
@@ -1461,26 +1431,26 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                eat_whitespace(p, os, parent_context, false);
                parent_context.check_end_layout(os);
                // store the encoding to be able to reset it
-               string const encoding_old = p.encoding_latex_;
+               string const encoding_old = p.getEncoding();
                string const encoding = p.getArg('{', '}');
-               // SJIS and BIG5 don't work with LaTeX according to the comment in unicode.cpp
+               // SJIS and Bg5 cannot be handled by iconv
                // JIS does not work with LyX's encoding conversion
-               if (encoding != "SJIS" && encoding != "BIG5" && encoding != "JIS")
+               if (encoding != "Bg5" && encoding != "JIS" && encoding != "SJIS")
                        p.setEncoding(encoding);
                else
                        p.setEncoding("utf8");
                // LyX doesn't support the second argument so if
                // this is used we need to output everything as ERT
                string const mapping = p.getArg('{', '}');
-               if ( (!mapping.empty() && mapping != " ")
-                       || (!is_known(encoding, supported_CJK_encodings))) {
+               char const * const * const where =
+                       is_known(encoding, supported_CJK_encodings);
+               if ((!mapping.empty() && mapping != " ") || !where) {
                        parent_context.check_layout(os);
                        handle_ert(os, "\\begin{" + name + "}{" + encoding + "}{" + mapping + "}",
                                       parent_context);
-                       // we must parse the content as verbatim because e.g. SJIS can contain
+                       // we must parse the content as verbatim because e.g. JIS can contain
                        // normally invalid characters
                        string const s = p.plainEnvironment("CJK");
-                       string::const_iterator it2 = s.begin();
                        for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
                                if (*it == '\\')
                                        handle_ert(os, "\\", parent_context);
@@ -1489,11 +1459,11 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                                else 
                                        os << *it;
                        }
-                       p.skip_spaces();
                        handle_ert(os, "\\end{" + name + "}",
                                       parent_context);
                } else {
-                       string const lang = CJK2lyx(encoding);
+                       string const lang =
+                               supported_CJK_languages[where - supported_CJK_encodings];
                        // store the language because we must reset it at the end
                        string const lang_old = parent_context.font.language;
                        parent_context.font.language = lang;
@@ -1501,7 +1471,7 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                        parent_context.font.language = lang_old;
                        parent_context.new_paragraph(os);
                }
-               p.encoding_latex_ = encoding_old;
+               p.setEncoding(encoding_old);
                p.skip_spaces();
        }
 
@@ -1868,13 +1838,85 @@ string const normalize_filename(string const & name)
 
 /// Convert \p name from TeX convention (relative to master file) to LyX
 /// convention (relative to .lyx file) if it is relative
-void fix_relative_filename(string & name)
+void fix_child_filename(string & name)
 {
-       if (FileName::isAbsolute(name))
-               return;
+       string const absMasterTeX = getMasterFilePath(true);
+       bool const isabs = FileName::isAbsolute(name);
+       // convert from "relative to .tex master" to absolute original path
+       if (!isabs)
+               name = makeAbsPath(name, absMasterTeX).absFileName();
+       bool copyfile = copyFiles();
+       string const absParentLyX = getParentFilePath(false);
+       string abs = name;
+       if (copyfile) {
+               // convert from absolute original path to "relative to master file"
+               string const rel = to_utf8(makeRelPath(from_utf8(name),
+                                                      from_utf8(absMasterTeX)));
+               // re-interpret "relative to .tex file" as "relative to .lyx file"
+               // (is different if the master .lyx file resides in a
+               // different path than the master .tex file)
+               string const absMasterLyX = getMasterFilePath(false);
+               abs = makeAbsPath(rel, absMasterLyX).absFileName();
+               // Do not copy if the new path is impossible to create. Example:
+               // absMasterTeX = "/foo/bar/"
+               // absMasterLyX = "/bar/"
+               // name = "/baz.eps" => new absolute name would be "/../baz.eps"
+               if (contains(name, "/../"))
+                       copyfile = false;
+       }
+       if (copyfile) {
+               if (isabs)
+                       name = abs;
+               else {
+                       // convert from absolute original path to
+                       // "relative to .lyx file"
+                       name = to_utf8(makeRelPath(from_utf8(abs),
+                                                  from_utf8(absParentLyX)));
+               }
+       }
+       else if (!isabs) {
+               // convert from absolute original path to "relative to .lyx file"
+               name = to_utf8(makeRelPath(from_utf8(name),
+                                          from_utf8(absParentLyX)));
+       }
+}
+
 
-       name = to_utf8(makeRelPath(from_utf8(makeAbsPath(name, getMasterFilePath()).absFileName()),
-                                  from_utf8(getParentFilePath())));
+void copy_file(FileName const & src, string dstname)
+{
+       if (!copyFiles())
+               return;
+       string const absParent = getParentFilePath(false);
+       FileName dst;
+       if (FileName::isAbsolute(dstname))
+               dst = FileName(dstname);
+       else
+               dst = makeAbsPath(dstname, absParent);
+       string const absMaster = getMasterFilePath(false);
+       FileName const srcpath = src.onlyPath();
+       FileName const dstpath = dst.onlyPath();
+       if (equivalent(srcpath, dstpath))
+               return;
+       if (!dstpath.isDirectory()) {
+               if (!dstpath.createPath()) {
+                       cerr << "Warning: Could not create directory for file `"
+                            << dst.absFileName() << "´." << endl;
+                       return;
+               }
+       }
+       if (dst.isReadableFile()) {
+               if (overwriteFiles())
+                       cerr << "Warning: Overwriting existing file `"
+                            << dst.absFileName() << "´." << endl;
+               else {
+                       cerr << "Warning: Not overwriting existing file `"
+                            << dst.absFileName() << "´." << endl;
+                       return;
+               }
+       }
+       if (!src.copyTo(dst))
+               cerr << "Warning: Could not copy file `" << src.absFileName()
+                    << "´ to `" << dst.absFileName() << "´." << endl;
 }
 
 
@@ -2101,24 +2143,6 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
        while (p.good()) {
                Token const & t = p.get_token();
 
-       // it is impossible to determine the correct document language if CJK is used.
-       // Therefore write a note at the beginning of the document
-       if (have_CJK) {
-               context.check_layout(os);
-               begin_inset(os, "Note Note\n");
-               os << "status open\n\\begin_layout Plain Layout\n"
-                  << "\\series bold\n"
-                  << "Important information:\n"
-                  << "\\end_layout\n\n"
-                  << "\\begin_layout Plain Layout\n"
-                  << "This document contains text in Chinese, Japanese or Korean.\n"
-                  << " It was therefore impossible for tex2lyx to set the correct document langue for your document."
-                  << " Please set in the document settings by yourself!\n"
-                  << "\\end_layout\n";
-               end_inset(os);
-               have_CJK = false;
-       }
-
        // it is impossible to determine the correct encoding for non-CJK Japanese.
        // Therefore write a note at the beginning of the document
        if (is_nonCJKJapanese) {
@@ -2565,7 +2589,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                                skip_braces(p);
                                                p.get_token();
                                                string name = normalize_filename(p.verbatim_item());
-                                               string const path = getMasterFilePath();
+                                               string const path = getMasterFilePath(true);
                                                // We want to preserve relative / absolute filenames,
                                                // therefore path is only used for testing
                                                // The file extension is in every case ".tex".
@@ -2580,9 +2604,11 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                                        if (!Gnumeric_name.empty())
                                                                name = Gnumeric_name;
                                                }
-                                               if (makeAbsPath(name, path).exists())
-                                                       fix_relative_filename(name);
-                                               else
+                                               FileName const absname = makeAbsPath(name, path);
+                                               if (absname.exists()) {
+                                                       fix_child_filename(name);
+                                                       copy_file(absname, name);
+                                               } else
                                                        cerr << "Warning: Could not find file '"
                                                             << name << "'." << endl;
                                                context.check_layout(os);
@@ -2592,7 +2618,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                                end_inset(os);
                                                context.check_layout(os);
                                                macro = false;
-                                               // register the packages that are automatically reloaded
+                                               // register the packages that are automatically loaded
                                                // by the Gnumeric template
                                                registerExternalTemplatePackages("GnumericSpreadsheet");
                                        }
@@ -2696,8 +2722,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                        context.check_layout(os);
                        p.skip_spaces();
                        begin_inset(os, "Caption\n");
-                       Context newcontext(true, context.textclass);
-                       newcontext.font = context.font;
+                       Context newcontext(true, context.textclass, 0, 0, context.font);
                        newcontext.check_layout(os);
                        if (p.next_token().cat() != catEscape &&
                            p.next_token().character() == '[') {
@@ -2747,8 +2772,8 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                        os << "\n\\begin_layout Plain Layout";
                                        p.skip_spaces();
                                        begin_inset(os, "Caption\n");
-                                       Context newcontext(true, context.textclass);
-                                       newcontext.font = context.font;
+                                       Context newcontext(true, context.textclass,
+                                                          0, 0, context.font);
                                        newcontext.check_layout(os);
                                        os << caption << "\n";
                                        newcontext.check_end_layout(os);
@@ -2792,7 +2817,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                opts["clip"] = string();
                        string name = normalize_filename(p.verbatim_item());
 
-                       string const path = getMasterFilePath();
+                       string const path = getMasterFilePath(true);
                        // We want to preserve relative / absolute filenames,
                        // therefore path is only used for testing
                        if (!makeAbsPath(name, path).exists()) {
@@ -2827,9 +2852,11 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                }
                        }
 
-                       if (makeAbsPath(name, path).exists())
-                               fix_relative_filename(name);
-                       else
+                       FileName const absname = makeAbsPath(name, path);
+                       if (absname.exists()) {
+                               fix_child_filename(name);
+                               copy_file(absname, name);
+                       } else
                                cerr << "Warning: Could not find graphics file '"
                                     << name << "'." << endl;
 
@@ -3165,8 +3192,8 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
 
                else if (t.cs() == "href") {
                        context.check_layout(os);
-                       string target = p.getArg('{', '}');
-                       string name = p.getArg('{', '}');
+                       string target = convert_command_inset_arg(p.verbatim_item());
+                       string name = convert_command_inset_arg(p.verbatim_item());
                        string type;
                        size_t i = target.find(':');
                        if (i != string::npos) {
@@ -3574,7 +3601,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                }
                
                else if (prefixIs(t.cs(), "text") 
-                        && is_known(t.cs().substr(4), polyglossia_languages)) {
+                        && is_known(t.cs().substr(4), preamble.polyglossia_languages)) {
                        // scheme is \textLANGUAGE{text} where LANGUAGE is in polyglossia_languages[]
                        string lang;
                        // We have to output the whole command if it has an option
@@ -3589,14 +3616,14 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                if (pos_var != string::npos && i == string::npos) {
                                        string variant;
                                        variant = langopts.substr(k + 1, langopts.length() - k - 2);
-                                       lang = polyglossia2lyx(variant);
+                                       lang = preamble.polyglossia2lyx(variant);
                                        parse_text_attributes(p, os, FLAG_ITEM, outer,
                                                                  context, "\\lang",
                                                                  context.font.language, lang);
                                } else
                                        handle_ert(os, t.asInput() + langopts, context);
                        } else {
-                               lang = polyglossia2lyx(t.cs().substr(4, string::npos));
+                               lang = preamble.polyglossia2lyx(t.cs().substr(4, string::npos));
                                parse_text_attributes(p, os, FLAG_ITEM, outer,
                                                          context, "\\lang",
                                                          context.font.language, lang);
@@ -3684,7 +3711,11 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                else if (t.cs() == "verb") {
                        context.check_layout(os);
                        char const delimiter = p.next_token().character();
-                       string const arg = p.getArg(delimiter, delimiter);
+                       // \verb is special: The usual escaping rules do not
+                       // apply, e.g. "\verb+\+" is valid and denotes a single
+                       // backslash (bug #4468). Therefore we do not allow
+                       // escaping in getArg().
+                       string const arg = p.getArg(delimiter, delimiter, false);
                        ostringstream oss;
                        oss << "\\verb" << delimiter << arg << delimiter;
                        handle_ert(os, oss.str(), context);
@@ -3757,7 +3788,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                name += p.get_token().asInput();
                        context.check_layout(os);
                        string filename(normalize_filename(p.getArg('{', '}')));
-                       string const path = getMasterFilePath();
+                       string const path = getMasterFilePath(true);
                        // We want to preserve relative / absolute filenames,
                        // therefore path is only used for testing
                        if ((t.cs() == "include" || t.cs() == "input") &&
@@ -3775,13 +3806,13 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                        if (makeAbsPath(filename, path).exists()) {
                                string const abstexname =
                                        makeAbsPath(filename, path).absFileName();
-                               string const abslyxname =
-                                       changeExtension(abstexname, ".lyx");
                                string const absfigname =
                                        changeExtension(abstexname, ".fig");
-                               fix_relative_filename(filename);
+                               fix_child_filename(filename);
                                string const lyxname =
                                        changeExtension(filename, ".lyx");
+                               string const abslyxname = makeAbsPath(
+                                       lyxname, getParentFilePath(false)).absFileName();
                                bool xfig = false;
                                external = FileName(absfigname).exists();
                                if (t.cs() == "input") {
@@ -3827,16 +3858,24 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                }
                                if (external) {
                                        outname = changeExtension(filename, ".fig");
+                                       FileName abssrc(changeExtension(abstexname, ".fig"));
+                                       copy_file(abssrc, outname);
                                } else if (xfig) {
                                        // Don't try to convert, the result
                                        // would be full of ERT.
                                        outname = filename;
+                                       FileName abssrc(abstexname);
+                                       copy_file(abssrc, outname);
                                } else if (t.cs() != "verbatiminput" &&
                                    tex2lyx(abstexname, FileName(abslyxname),
                                            p.getEncoding())) {
                                        outname = lyxname;
+                                       // no need to call copy_file
+                                       // tex2lyx creates the file
                                } else {
                                        outname = filename;
+                                       FileName abssrc(abstexname);
+                                       copy_file(abssrc, outname);
                                }
                        } else {
                                cerr << "Warning: Could not find included file '"
@@ -3945,7 +3984,6 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                arg += p.getFullOpt();
                                eat_whitespace(p, os, context, false);
                                handle_ert(os, arg + '{', context);
-                               eat_whitespace(p, os, context, false);
                                parse_text(p, os, FLAG_ITEM, outer, context);
                                handle_ert(os, "}", context);
                        } else {
@@ -3958,7 +3996,6 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                else {
                                        eat_whitespace(p, os, context, false);
                                        handle_ert(os, "\\framebox{", context);
-                                       eat_whitespace(p, os, context, false);
                                        parse_text(p, os, FLAG_ITEM, outer, context);
                                        handle_ert(os, "}", context);
                                }
@@ -3975,7 +4012,6 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                arg += p.getFullOpt();
                                eat_whitespace(p, os, context, false);
                                handle_ert(os, arg + '{', context);
-                               eat_whitespace(p, os, context, false);
                                parse_text(p, os, FLAG_ITEM, outer, context);
                                handle_ert(os, "}", context);
                        } else
@@ -4220,7 +4256,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                        vector<string> keys;
                        split_map(arg, opts, keys);
                        string name = normalize_filename(p.verbatim_item());
-                       string const path = getMasterFilePath();
+                       string const path = getMasterFilePath(true);
                        // We want to preserve relative / absolute filenames,
                        // therefore path is only used for testing
                        if (!makeAbsPath(name, path).exists()) {
@@ -4234,9 +4270,12 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                        pdflatex = true;
                                }
                        }
-                       if (makeAbsPath(name, path).exists())
-                               fix_relative_filename(name);
-                       else
+                       FileName const absname = makeAbsPath(name, path);
+                       if (absname.exists())
+                       {
+                               fix_child_filename(name);
+                               copy_file(absname, name);
+                       } else
                                cerr << "Warning: Could not find file '"
                                     << name << "'." << endl;
                        // write output
@@ -4285,7 +4324,7 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                else if (t.cs() == "loadgame") {
                        p.skip_spaces();
                        string name = normalize_filename(p.verbatim_item());
-                       string const path = getMasterFilePath();
+                       string const path = getMasterFilePath(true);
                        // We want to preserve relative / absolute filenames,
                        // therefore path is only used for testing
                        if (!makeAbsPath(name, path).exists()) {
@@ -4297,9 +4336,12 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                if (!lyxskak_name.empty())
                                        name = lyxskak_name;
                        }
-                       if (makeAbsPath(name, path).exists())
-                               fix_relative_filename(name);
-                       else
+                       FileName const absname = makeAbsPath(name, path);
+                       if (absname.exists())
+                       {
+                               fix_child_filename(name);
+                               copy_file(absname, name);
+                       } else
                                cerr << "Warning: Could not find file '"
                                     << name << "'." << endl;
                        context.check_layout(os);
@@ -4367,6 +4409,79 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
        }
 }
 
+
+string guessLanguage(Parser & p, string const & lang)
+{
+       typedef std::map<std::string, size_t> LangMap;
+       // map from language names to number of characters
+       LangMap used;
+       used[lang] = 0;
+       for (char const * const * i = supported_CJK_languages; *i; i++)
+               used[string(*i)] = 0;
+
+       while (p.good()) {
+               Token const t = p.get_token();
+               // comments are not counted for any language
+               if (t.cat() == catComment)
+                       continue;
+               // commands are not counted as well, but we need to detect
+               // \begin{CJK} and switch encoding if needed
+               if (t.cat() == catEscape) {
+                       if (t.cs() == "inputencoding") {
+                               string const enc = subst(p.verbatim_item(), "\n", " ");
+                               p.setEncoding(enc);
+                               continue;
+                       }
+                       if (t.cs() != "begin")
+                               continue;
+               } else {
+                       // Non-CJK content is counted for lang.
+                       // We do not care about the real language here:
+                       // If we have more non-CJK contents than CJK contents,
+                       // we simply use the language that was specified as
+                       // babel main language.
+                       used[lang] += t.asInput().length();
+                       continue;
+               }
+               // Now we are starting an environment
+               p.pushPosition();
+               string const name = p.getArg('{', '}');
+               if (name != "CJK") {
+                       p.popPosition();
+                       continue;
+               }
+               // It is a CJK environment
+               p.popPosition();
+               /* name = */ p.getArg('{', '}');
+               string const encoding = p.getArg('{', '}');
+               /* mapping = */ p.getArg('{', '}');
+               string const encoding_old = p.getEncoding();
+               char const * const * const where =
+                       is_known(encoding, supported_CJK_encodings);
+               if (where)
+                       p.setEncoding(encoding);
+               else
+                       p.setEncoding("utf8");
+               string const text = p.verbatimEnvironment("CJK");
+               p.setEncoding(encoding_old);
+               p.skip_spaces();
+               if (!where) {
+                       // ignore contents in unknown CJK encoding
+                       continue;
+               }
+               // the language of the text
+               string const cjk =
+                       supported_CJK_languages[where - supported_CJK_encodings];
+               used[cjk] += text.length();
+       }
+       LangMap::const_iterator use = used.begin();
+       for (LangMap::const_iterator it = used.begin(); it != used.end(); ++it) {
+               if (it->second > use->second)
+                       use = it;
+       }
+       return use->first;
+}
+
 // }])