]> git.lyx.org Git - lyx.git/blobdiff - src/tex2lyx/tex2lyx.cpp
Merge branch 'master' of git.lyx.org:lyx
[lyx.git] / src / tex2lyx / tex2lyx.cpp
index 5467b5e015d2ce83f2d27889199ba5659401da42..37bcf7d038d9e76edf47d7bd90cacac9f7f5bb84 100644 (file)
 // {[(
 
 #include <config.h>
+#include <version.h>
 
 #include "tex2lyx.h"
 
 #include "Context.h"
 #include "Encoding.h"
 #include "Layout.h"
+#include "LayoutFile.h"
+#include "LayoutModuleList.h"
+#include "ModuleList.h"
+#include "Preamble.h"
 #include "TextClass.h"
 
 #include "support/convert.h"
@@ -55,6 +60,22 @@ namespace Alert {
 }
 
 
+// Dummy texrow support
+void TexRow::newline()
+{}
+
+
+void TexRow::newlines(int)
+{}
+
+
+// Dummy LyXRC support
+class LyXRC {
+public:
+       string icon_set;
+} lyxrc;
+
+
 // Dummy translation support
 Messages messages_;
 Messages const & getMessages(std::string const &)
@@ -74,21 +95,9 @@ void lyx_exit(int)
 {}
 
 
-string const trim(string const & a, char const * p)
+string const trimSpaceAndEol(string const & a)
 {
-       // LASSERT(p, /**/);
-
-       if (a.empty() || !*p)
-               return a;
-
-       size_t r = a.find_last_not_of(p);
-       size_t l = a.find_first_not_of(p);
-
-       // Is this the minimal test? (lgb)
-       if (r == string::npos && l == string::npos)
-               return string();
-
-       return a.substr(l, r - l + 1);
+       return trim(a, " \t\n\r");
 }
 
 
@@ -135,13 +144,19 @@ string active_environment()
 }
 
 
+TeX2LyXDocClass textclass;
 CommandMap known_commands;
 CommandMap known_environments;
 CommandMap known_math_environments;
+FullCommandMap possible_textclass_commands;
+FullEnvironmentMap possible_textclass_environments;
+int const LYX_FORMAT = LYX_FORMAT_TEX2LYX;
 
+/// used modules
+LayoutModuleList used_modules;
 
-void add_known_command(string const & command, string const & o1,
-                      bool o2)
+
+void convertArgs(string const & o1, bool o2, vector<ArgumentType> & arguments)
 {
        // We have to handle the following cases:
        // definition                      o1    o2    invocation result
@@ -151,7 +166,6 @@ void add_known_command(string const & command, string const & o1,
        // \newcommand{\foo}[1][]{bar #1}  "[1]" true  \foo[x]    bar x
        // \newcommand{\foo}[1][x]{bar #1} "[1]" true  \foo[x]    bar x
        unsigned int nargs = 0;
-       vector<ArgumentType> arguments;
        string const opt1 = rtrim(ltrim(o1, "["), "]");
        if (isStrUnsignedInt(opt1)) {
                // The command has arguments
@@ -164,12 +178,161 @@ void add_known_command(string const & command, string const & o1,
        }
        for (unsigned int i = 0; i < nargs; ++i)
                arguments.push_back(required);
+}
+
+
+void add_known_command(string const & command, string const & o1,
+                       bool o2, docstring const & definition)
+{
+       vector<ArgumentType> arguments;
+       convertArgs(o1, o2, arguments);
        known_commands[command] = arguments;
+       if (!definition.empty())
+               possible_textclass_commands[command] =
+                       FullCommand(arguments, definition);
+}
+
+
+void add_known_environment(string const & environment, string const & o1,
+                           bool o2, docstring const & beg, docstring const &end)
+{
+       vector<ArgumentType> arguments;
+       convertArgs(o1, o2, arguments);
+       known_environments[environment] = arguments;
+       if (!beg.empty() || ! end.empty())
+               possible_textclass_environments[environment] =
+                       FullEnvironment(arguments, beg, end);
+}
+
+
+Layout const * findLayoutWithoutModule(TextClass const & textclass,
+                                       string const & name, bool command)
+{
+       DocumentClass::const_iterator it = textclass.begin();
+       DocumentClass::const_iterator en = textclass.end();
+       for (; it != en; ++it) {
+               if (it->latexname() == name &&
+                   ((command && it->isCommand()) || (!command && it->isEnvironment())))
+                       return &*it;
+       }
+       return 0;
+}
+
+
+InsetLayout const * findInsetLayoutWithoutModule(TextClass const & textclass,
+                                                 string const & name, bool command)
+{
+       DocumentClass::InsetLayouts::const_iterator it = textclass.insetLayouts().begin();
+       DocumentClass::InsetLayouts::const_iterator en = textclass.insetLayouts().end();
+       for (; it != en; ++it) {
+               if (it->second.latexname() == name &&
+                   ((command && it->second.latextype() == InsetLayout::COMMAND) ||
+                    (!command && it->second.latextype() == InsetLayout::ENVIRONMENT)))
+                       return &(it->second);
+       }
+       return 0;
+}
+
+
+bool checkModule(string const & name, bool command)
+{
+       // Cache to avoid slowdown by repated searches
+       static set<string> failed[2];
+
+       // Only add the module if the command was actually defined in the LyX preamble
+       if (command) {
+               if (possible_textclass_commands.find('\\' + name) == possible_textclass_commands.end())
+                       return false;
+       } else {
+               if (possible_textclass_environments.find(name) == possible_textclass_environments.end())
+                       return false;
+       }
+       if (failed[command].find(name) != failed[command].end())
+               return false;
+
+       // Create list of dummy document classes if not already done.
+       // This is needed since a module cannot be read on its own, only as
+       // part of a document class.
+       LayoutFile const & baseClass = LayoutFileList::get()[textclass.name()];
+       typedef map<string, DocumentClassPtr > ModuleMap;
+       static ModuleMap modules;
+       static bool init = true;
+       if (init) {
+               baseClass.load();
+               LyXModuleList::const_iterator const end = theModuleList.end();
+               LyXModuleList::const_iterator it = theModuleList.begin();
+               for (; it != end; ++it) {
+                       string const module = it->getID();
+                       LayoutModuleList m;
+                       // FIXME this excludes all modules that depend on another one
+                       if (!m.moduleCanBeAdded(module, &baseClass))
+                               continue;
+                       m.push_back(module);
+                       modules[module] = getDocumentClass(baseClass, m);
+               }
+               init = false;
+       }
+
+       // Try to find a module that defines the command.
+       // Only add it if the definition can be found in the preamble of the
+       // style that corresponds to the command. This is a heuristic and
+       // different from the way how we parse the builtin commands of the
+       // text class (in that case we only compare the name), but it is
+       // needed since it is not unlikely that two different modules define a
+       // command with the same name.
+       ModuleMap::iterator const end = modules.end();
+       for (ModuleMap::iterator it = modules.begin(); it != end; ++it) {
+               string const module = it->first;
+               if (!used_modules.moduleCanBeAdded(module, &baseClass))
+                       continue;
+               if (findLayoutWithoutModule(textclass, name, command))
+                       continue;
+               if (findInsetLayoutWithoutModule(textclass, name, command))
+                       continue;
+               DocumentClassConstPtr  c = it->second;
+               Layout const * layout = findLayoutWithoutModule(*c, name, command);
+               InsetLayout const * insetlayout = layout ? 0 :
+                       findInsetLayoutWithoutModule(*c, name, command);
+               docstring preamble;
+               if (layout)
+                       preamble = layout->preamble();
+               else if (insetlayout)
+                       preamble = insetlayout->preamble();
+               if (preamble.empty())
+                       continue;
+               bool add = false;
+               if (command) {
+                       FullCommand const & cmd =
+                               possible_textclass_commands['\\' + name];
+                       if (preamble.find(cmd.def) != docstring::npos)
+                               add = true;
+               } else {
+                       FullEnvironment const & env =
+                               possible_textclass_environments[name];
+                       if (preamble.find(env.beg) != docstring::npos &&
+                           preamble.find(env.end) != docstring::npos)
+                               add = true;
+               }
+               if (add) {
+                       FileName layout_file = libFileSearch("layouts", module, "module");
+                       if (textclass.read(layout_file, TextClass::MODULE)) {
+                               used_modules.push_back(module);
+                               // speed up further searches:
+                               // the module does not need to be checked anymore.
+                               modules.erase(it);
+                               return true;
+                       }
+               }
+       }
+       failed[command].insert(name);
+       return false;
 }
 
 
 bool noweb_mode = false;
 bool pdflatex = false;
+bool xetex = false;
+bool have_CJK = false;
 bool roundtrip = false;
 
 
@@ -192,13 +355,20 @@ void read_command(Parser & p, string command, CommandMap & commands)
                        string const arg = p.getArg('{', '}');
                        if (arg == "translate")
                                arguments.push_back(required);
+                       else if (arg == "group")
+                               arguments.push_back(req_group);
                        else if (arg == "item")
                                arguments.push_back(item);
+                       else if (arg == "displaymath")
+                               arguments.push_back(displaymath);
                        else
                                arguments.push_back(verbatim);
                } else {
-                       p.getArg('[', ']');
-                       arguments.push_back(optional);
+                       string const arg = p.getArg('[', ']');
+                       if (arg == "group")
+                               arguments.push_back(opt_group);
+                       else
+                               arguments.push_back(optional);
                }
        }
        commands[command] = arguments;
@@ -293,13 +463,33 @@ int parse_help(string const &, string const &)
                "\t-n                 translate a noweb (aka literate programming) file.\n"
                "\t-roundtrip         re-export created .lyx file infile.lyx.lyx to infile.lyx.tex.\n"
                "\t-s syntaxfile      read additional syntax file.\n"
-               "\t-sysdir dir        Set system directory to DIR.\n"
-               "\t-userdir DIR       Set user directory to DIR."
+               "\t-sysdir SYSDIR     Set system directory to SYSDIR.\n"
+               "\t                   Default: " << package().system_support() << "\n"
+               "\t-userdir USERDIR   Set user directory to USERDIR.\n"
+               "\t                   Default: " << package().user_support() << "\n"
+               "\t-version           Summarize version and build info.\n"
+               "Paths:\n"
+               "\tThe program searches for the files \"encodings\", \"lyxmodules.lst\",\n"
+               "\t\"textclass.lst\", \"syntax.default\", and \"unicodesymbols\", first in\n"
+               "\t\"USERDIR\", then in \"SYSDIR\". The subdirectories \"USERDIR/layouts\"\n"
+               "\tand \"SYSDIR/layouts\" are searched for layout and module files.\n"
+               "Check the tex2lyx man page for more details."
             << endl;
        exit(error_code);
 }
 
 
+int parse_version(string const &, string const &)
+{
+       lyxerr << "tex2lyx " << lyx_version
+              << " (" << lyx_release_date << ")" << endl;
+       lyxerr << "Built on " << __DATE__ << ", " << __TIME__ << endl;
+
+       lyxerr << lyx_version_info << endl;
+       exit(error_code);
+}
+
+
 void error_message(string const & message)
 {
        cerr << "tex2lyx: " << message << "\n\n";
@@ -384,12 +574,16 @@ void easyParse(int & argc, char * argv[])
 {
        map<string, cmd_helper> cmdmap;
 
+       cmdmap["-h"] = parse_help;
+       cmdmap["-help"] = parse_help;
+       cmdmap["--help"] = parse_help;
+       cmdmap["-v"] = parse_version;
+       cmdmap["-version"] = parse_version;
+       cmdmap["--version"] = parse_version;
        cmdmap["-c"] = parse_class;
        cmdmap["-e"] = parse_encoding;
        cmdmap["-f"] = parse_force;
        cmdmap["-s"] = parse_syntaxfile;
-       cmdmap["-help"] = parse_help;
-       cmdmap["--help"] = parse_help;
        cmdmap["-n"] = parse_noweb;
        cmdmap["-sysdir"] = parse_sysdir;
        cmdmap["-userdir"] = parse_userdir;
@@ -453,19 +647,29 @@ namespace {
  *  You must ensure that \p parentFilePath is properly set before calling
  *  this function!
  */
-void tex2lyx(idocstream & is, ostream & os, string const & encoding)
+bool tex2lyx(idocstream & is, ostream & os, string encoding)
 {
+       // Set a sensible default encoding.
+       // This is used until an encoding command is found.
+       // For child documents use the encoding of the master, else latin1,
+       // since latin1 does not cause an iconv error if the actual encoding
+       // is different (bug 7509).
+       if (encoding.empty()) {
+               if (preamble.inputencoding() == "auto")
+                       encoding = "latin1";
+               else
+                       encoding = preamble.inputencoding();
+       }
+
        Parser p(is);
-       if (!encoding.empty())
-               p.setEncoding(encoding);
+       p.setEncoding(encoding);
        //p.dump();
 
-       stringstream ss;
-       TeX2LyXDocClass textclass;
-       parse_preamble(p, ss, documentclass, textclass);
+       preamble.parse(p, documentclass, textclass);
 
        active_environments.push_back("document");
        Context context(true, textclass);
+       stringstream ss;
        parse_text(p, ss, FLAG_END, true, context);
        if (Context::empty)
                // Empty document body. LyX needs at least one paragraph.
@@ -473,6 +677,19 @@ void tex2lyx(idocstream & is, ostream & os, string const & encoding)
        context.check_end_layout(ss);
        ss << "\n\\end_body\n\\end_document\n";
        active_environments.pop_back();
+
+       // We know the used modules only after parsing the full text
+       if (!used_modules.empty()) {
+               LayoutModuleList::const_iterator const end = used_modules.end();
+               LayoutModuleList::const_iterator it = used_modules.begin();
+               for (; it != end; ++it)
+                       preamble.addModule(*it);
+       }
+       if (!preamble.writeLyXHeader(os, !active_environments.empty())) {
+               cerr << "Could write LyX file header." << endl;
+               return false;
+       }
+
        ss.seekg(0);
        os << ss.str();
 #ifdef TEST_PARSER
@@ -482,6 +699,7 @@ void tex2lyx(idocstream & is, ostream & os, string const & encoding)
                parsertest << p.get_token().asInput();
        // <origfile> and parsertest.tex should now have identical content
 #endif
+       return true;
 }
 
 
@@ -499,9 +717,9 @@ bool tex2lyx(FileName const & infilename, ostream & os, string const & encoding)
        }
        string const oldParentFilePath = parentFilePath;
        parentFilePath = onlyPath(infilename.absFileName());
-       tex2lyx(is, os, encoding);
+       bool retval = tex2lyx(is, os, encoding);
        parentFilePath = oldParentFilePath;
-       return true;
+       return retval;
 }
 
 } // anonymous namespace
@@ -571,6 +789,15 @@ int main(int argc, char * argv[])
 
        os::init(argc, argv);
 
+       try {
+               init_package(internal_path(os::utf8_argv(0)), string(), string());
+       } catch (ExceptionMessage const & message) {
+               cerr << to_utf8(message.title_) << ":\n"
+                    << to_utf8(message.details_) << endl;
+               if (message.type_ == ErrorException)
+                       return EXIT_FAILURE;
+       }
+
        easyParse(argc, argv);
 
        if (argc <= 1) 
@@ -578,8 +805,7 @@ int main(int argc, char * argv[])
 
        try {
                init_package(internal_path(os::utf8_argv(0)),
-                            cl_system_support, cl_user_support,
-                            top_build_dir_is_two_levels_up);
+                            cl_system_support, cl_user_support);
        } catch (ExceptionMessage const & message) {
                cerr << to_utf8(message.title_) << ":\n"
                     << to_utf8(message.details_) << endl;
@@ -637,6 +863,11 @@ int main(int argc, char * argv[])
        if (!default_encoding.empty() && !encodings.fromLaTeXName(default_encoding))
                error_message("Unknown LaTeX encoding `" + default_encoding + "'");
 
+       // Load the layouts
+       LayoutFileList::get().read();
+       //...and the modules
+       theModuleList.read();
+
        // The real work now.
        masterFilePath = onlyPath(infilename);
        parentFilePath = masterFilePath;