]> git.lyx.org Git - lyx.git/blobdiff - src/tex2lyx/text.cpp
Cmake build: support tests
[lyx.git] / src / tex2lyx / text.cpp
index 24694d3da27dce83260f1c0440240288b303975b..7f7603c198b244057a8495edc381fd763e47db23 100644 (file)
@@ -117,6 +117,22 @@ char const * const known_ref_commands[] = { "ref", "pageref", "vref",
 char const * const known_coded_ref_commands[] = { "ref", "pageref", "vref",
  "vpageref", "formatted", "eqref", 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};
+
+/**
+ * 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 supported_CJK_languages[] = {
+"japanese-cjk", "korean", "chinese-simplified", "chinese-traditional", 0};
+
 /*!
  * natbib commands.
  * The starred forms are also known except for "citefullauthor",
@@ -461,7 +477,9 @@ docstring convert_unicodesymbols(docstring s)
                set<string> req;
                docstring parsed = encodings.fromLaTeXCommand(s,
                                Encodings::TEXT_CMD, termination, rem, &req);
-               for (set<string>::const_iterator it = req.begin(); it != req.end(); ++it)
+               set<string>::const_iterator it = req.begin();
+               set<string>::const_iterator en = req.end();
+               for (; it != en; ++it)
                        preamble.registerAutomaticallyLoadedPackage(*it);
                os << parsed;
                s = rem;
@@ -796,8 +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('[', ']');
@@ -821,7 +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";
@@ -1210,6 +1238,23 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                }
        }
 
+       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);
+                       parent_context.new_paragraph(os);
+               }
+               // save the language in the context so that it is
+               // handled by parse_text
+               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();
+               // We must begin a new paragraph to reset the language
+               parent_context.new_paragraph(os);
+               p.skip_spaces();
+       }
+
        else if (unstarred_name == "tabular" || name == "longtable") {
                eat_whitespace(p, os, parent_context, false);
                string width = "0pt";
@@ -1379,6 +1424,57 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                os << "\n\\begin_layout Standard\n";
        }
 
+       else if (name == "CJK") {
+               // the scheme is \begin{CJK}{encoding}{mapping}{text}
+               // It is impossible to decide if a CJK environment was in its own paragraph or within
+               // a line. We therefore always assume a paragraph since the latter is a rare case.
+               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.getEncoding();
+               string const encoding = p.getArg('{', '}');
+               // SJIS and Bg5 cannot be handled by iconv
+               // JIS does not work with LyX's encoding conversion
+               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('{', '}');
+               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. JIS can contain
+                       // normally invalid characters
+                       string const s = p.plainEnvironment("CJK");
+                       for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
+                               if (*it == '\\')
+                                       handle_ert(os, "\\", parent_context);
+                               else if (*it == '$')
+                                       handle_ert(os, "$", parent_context);
+                               else 
+                                       os << *it;
+                       }
+                       handle_ert(os, "\\end{" + name + "}",
+                                      parent_context);
+               } else {
+                       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;
+                       parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
+                       parent_context.font.language = lang_old;
+                       parent_context.new_paragraph(os);
+               }
+               p.setEncoding(encoding_old);
+               p.skip_spaces();
+       }
+
        else if (name == "lyxgreyedout") {
                eat_whitespace(p, os, parent_context, false);
                parent_context.check_layout(os);
@@ -1556,7 +1652,9 @@ void parse_environment(Parser & p, ostream & os, bool outer,
                if (!preamble.titleLayoutFound())
                        preamble.titleLayoutFound(newlayout->intitle);
                set<string> const & req = newlayout->requires();
-               for (set<string>::const_iterator it = req.begin(); it != req.end(); ++it)
+               set<string>::const_iterator it = req.begin();
+               set<string>::const_iterator en = req.end();
+               for (; it != en; ++it)
                        preamble.registerAutomaticallyLoadedPackage(*it);
        }
 
@@ -1740,13 +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;
 }
 
 
@@ -1973,6 +2143,25 @@ 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 encoding for non-CJK Japanese.
+       // Therefore write a note at the beginning of the document
+       if (is_nonCJKJapanese) {
+               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 is in Japanese (non-CJK).\n"
+                  << " It was therefore impossible for tex2lyx to determine the correct encoding."
+                  << " The encoding EUC-JP was assumed. If this is incorrect, please set the correct"
+                  << " encoding in the document settings.\n"
+                  << "\\end_layout\n";
+               end_inset(os);
+               is_nonCJKJapanese = false;
+       }
+
 #ifdef FILEDEBUG
                debugToken(cerr, t, flags);
 #endif
@@ -2400,7 +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".
@@ -2415,9 +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);
@@ -2427,7 +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");
                                        }
@@ -2486,8 +2677,9 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                        if (!preamble.titleLayoutFound())
                                                preamble.titleLayoutFound(newlayout->intitle);
                                        set<string> const & req = newlayout->requires();
-                                       for (set<string>::const_iterator it = req.begin();
-                                            it != req.end(); ++it)
+                                       set<string>::const_iterator it = req.begin();
+                                       set<string>::const_iterator en = req.end();
+                                       for (; it != en; ++it)
                                                preamble.registerAutomaticallyLoadedPackage(*it);
                                } else
                                        handle_ert(os,
@@ -2530,8 +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() == '[') {
@@ -2581,8 +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);
@@ -2626,7 +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()) {
@@ -2661,9 +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;
 
@@ -2999,8 +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) {
@@ -3396,7 +3589,6 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                        context.check_layout(os);
                        // save the language for the case that a
                        // \foreignlanguage is used
-
                        context.font.language = babel2lyx(p.verbatim_item());
                        os << "\n\\lang " << context.font.language << "\n";
                }
@@ -3407,6 +3599,36 @@ void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
                                              context, "\\lang",
                                              context.font.language, lang);
                }
+               
+               else if (prefixIs(t.cs(), "text") 
+                        && 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
+                       // because LyX doesn't support this yet, see bug #8214,
+                       // only if there is a single option specifying a variant, we can handle it.
+                       if (p.hasOpt()) {
+                               string langopts = p.getOpt();
+                               // check if the option contains a variant, if yes, extract it
+                               string::size_type pos_var = langopts.find("variant");
+                               string::size_type i = langopts.find(',');
+                               string::size_type k = langopts.find('=', pos_var);
+                               if (pos_var != string::npos && i == string::npos) {
+                                       string variant;
+                                       variant = langopts.substr(k + 1, langopts.length() - k - 2);
+                                       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 = preamble.polyglossia2lyx(t.cs().substr(4, string::npos));
+                               parse_text_attributes(p, os, FLAG_ITEM, outer,
+                                                         context, "\\lang",
+                                                         context.font.language, lang);
+                       }
+               }
 
                else if (t.cs() == "inputencoding") {
                        // nothing to write here
@@ -3489,7 +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);
@@ -3562,7 +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") &&
@@ -3580,13 +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") {
@@ -3632,16 +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 '"
@@ -3750,14 +3984,21 @@ 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 {
                                string special = p.getFullOpt();
                                special += p.getOpt();
-                               parse_outer_box(p, os, FLAG_ITEM, outer,
-                                               context, t.cs(), special);
+                               // LyX does not yet support \framebox without any option
+                               if (!special.empty())
+                                       parse_outer_box(p, os, FLAG_ITEM, outer,
+                                                       context, t.cs(), special);
+                               else {
+                                       eat_whitespace(p, os, context, false);
+                                       handle_ert(os, "\\framebox{", context);
+                                       parse_text(p, os, FLAG_ITEM, outer, context);
+                                       handle_ert(os, "}", context);
+                               }
                        }
                }
 
@@ -3771,7 +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
@@ -4016,7 +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()) {
@@ -4030,9 +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
@@ -4081,7 +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()) {
@@ -4093,9 +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);
@@ -4163,6 +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;
+}
+
 // }])