]> git.lyx.org Git - lyx.git/blobdiff - src/Format.cpp
Run codespell on tex2lyx/, client/, convert/ and graphics/
[lyx.git] / src / Format.cpp
index 2bc7ac534b745139e79e6282dbf0bd291c4a12c1..ee4f7f464b82d3cc3d64bca4915dddb00e3791c4 100644 (file)
 #include "frontends/alert.h" //to be removed?
 
 #include "support/debug.h"
+#include "support/docstream.h"
 #include "support/filetools.h"
 #include "support/gettext.h"
 #include "support/lstrings.h"
+#include "support/lyxmagic.h"
+#include "support/mutex.h"
 #include "support/os.h"
 #include "support/PathChanger.h"
 #include "support/Systemcall.h"
 #include "support/Translator.h"
 
 #include <algorithm>
+#include <functional>
 #include <map>
 #include <ctime>
 
-// FIXME: Q_WS_MACX is not available, it's in Qt
+// FIXME: Q_OS_MAC is not available, it's in Qt
 #ifdef USE_MACOSX_PACKAGING
 #include "support/linkback/LinkBackProxy.h"
 #endif
 
-#ifdef HAVE_MAGIC_H
-#include <magic.h>
-#endif
-
 using namespace std;
 using namespace lyx::support;
 
@@ -55,65 +55,23 @@ string const token_from_format("$$i");
 string const token_path_format("$$p");
 string const token_socket_format("$$a");
 
-
-class FormatNamesEqual : public unary_function<Format, bool> {
-public:
-       FormatNamesEqual(string const & name)
-               : name_(name)
-       {}
-       bool operator()(Format const & f) const
-       {
-               return f.name() == name_;
-       }
-private:
-       string name_;
-};
-
-
-class FormatExtensionsEqual : public unary_function<Format, bool> {
-public:
-       FormatExtensionsEqual(string const & extension)
-               : extension_(extension)
-       {}
-       bool operator()(Format const & f) const
-       {
-               return f.hasExtension(extension_);
-       }
-private:
-       string extension_;
-};
+} // namespace
 
 
-class FormatMimeEqual : public unary_function<Format, bool> {
-public:
-       FormatMimeEqual(string const & mime)
-               : mime_(mime)
-       {}
-       bool operator()(Format const & f) const
-       {
-               // The test for empty mime strings is needed since we allow
-               // formats with empty mime types.
-               return f.mime() == mime_ && !mime_.empty();
-       }
-private:
-       string mime_;
-};
-
-
-} //namespace anon
-
 bool Format::formatSorter(Format const * lhs, Format const * rhs)
 {
-       return _(lhs->prettyname()) < _(rhs->prettyname());
+       return compare_locale(translateIfPossible(lhs->prettyname()),
+                             translateIfPossible(rhs->prettyname())) < 0;
 }
 
 bool operator<(Format const & a, Format const & b)
 {
-       return _(a.prettyname()) < _(b.prettyname());
+       return compare_locale(translateIfPossible(a.prettyname()),
+                             translateIfPossible(b.prettyname())) < 0;
 }
 
 
-Format::Format(string const & n, string const & e, string const & p,
+Format::Format(string const & n, string const & e, docstring const & p,
               string const & s, string const & v, string const & ed,
               string const & m, int flags)
        : name_(n), prettyname_(p), shortcut_(s), viewer_(v),
@@ -163,17 +121,39 @@ void Format::setExtensions(string const & e)
 }
 
 
+namespace {
+
+std::function<bool (Format const &)> FormatNameIs(string const & name)
+{
+       return [name](Format const & f){ return f.name() == name; };
+}
+
+}
+
 // This method should return a reference, and throw an exception
 // if the format named name cannot be found (Lgb)
 Format const * Formats::getFormat(string const & name) const
 {
        FormatList::const_iterator cit =
-               find_if(formatlist.begin(), formatlist.end(),
-                       FormatNamesEqual(name));
-       if (cit != formatlist.end())
+               find_if(formatlist_.begin(), formatlist_.end(),
+                       FormatNameIs(name));
+       if (cit != formatlist_.end())
                return &(*cit);
        else
-               return 0;
+               return nullptr;
+}
+
+
+Format * Formats::getFormat(string const & name)
+{
+       FormatList::iterator it =
+               find_if(formatlist_.begin(), formatlist_.end(),
+                               FormatNameIs(name));
+
+       if (it != formatlist_.end())
+               return &(*it);
+
+       return nullptr;
 }
 
 
@@ -228,11 +208,15 @@ string guessFormatFromContents(FileName const & fn)
        static string const zipStamp = "PK";
 
        // ZIP containers (koffice, openoffice.org etc).
-       static string const nonzipStamp = "\008\0\0\0mimetypeapplication/";
+       static string const nonzipStamp = "\010\0\0\0mimetypeapplication/";
 
        // compress
        static string const compressStamp = "\037\235";
 
+       // DOS binary EPS according to Adobe TN-5002
+       static string const binEPSStamp = "\xC5\xD0\xD3\xC6";
+
+
        // Maximum strings to read
        int const max_count = 50;
        int count = 0;
@@ -241,8 +225,9 @@ string guessFormatFromContents(FileName const & fn)
        string format;
        bool firstLine = true;
        bool backslash = false;
+       bool maybelatex = false;
        int dollars = 0;
-       while ((count++ < max_count) && format.empty()) {
+       while ((count++ < max_count) && format.empty() && !maybelatex) {
                if (ifs.eof())
                        break;
 
@@ -272,6 +257,8 @@ string guessFormatFromContents(FileName const & fn)
 
                        } else if (stamp == "\001\332") {
                                format =  "sgi";
+                       } else if (prefixIs(str, binEPSStamp)) {
+                               format =  "eps";
 
                        // PBM family
                        // Don't need to use str.at(0), str.at(1) because
@@ -329,6 +316,9 @@ string guessFormatFromContents(FileName const & fn)
                        // autodetect pdf format for graphics inclusion
                        format = "pdf6";
 
+               else if (contains(str, " EMF"))
+                       format = "emf";
+
                else if (contains(str, "PNG"))
                        format = "png";
 
@@ -358,16 +348,18 @@ string guessFormatFromContents(FileName const & fn)
                         contains(str, "$$") ||
                         contains(str, "\\[") ||
                         contains(str, "\\]"))
-                       format = "latex";
+                       maybelatex = true;
                else {
                        if (contains(str, '\\'))
                                backslash = true;
                        dollars += count_char(str, '$');
+                       if (backslash && dollars > 1)
+                               // inline equation
+                               maybelatex = true;
                }
        }
 
-       if (format.empty() && backslash && dollars > 1)
-               // inline equation
+       if (format.empty() && maybelatex && !isBinaryFile(fn))
                format = "latex";
 
        if (format.empty()) {
@@ -384,7 +376,7 @@ string guessFormatFromContents(FileName const & fn)
        return string();
 }
 
-}
+} // namespace
 
 
 string Formats::getFormatFromFile(FileName const & filename) const
@@ -392,58 +384,74 @@ string Formats::getFormatFromFile(FileName const & filename) const
        if (filename.empty())
                return string();
 
-#ifdef HAVE_MAGIC_H
+       string psformat;
+       string format;
        if (filename.exists()) {
-               magic_t magic_cookie = magic_open(MAGIC_MIME);
-               if (magic_cookie) {
-                       string format;
-                       if (magic_load(magic_cookie, NULL) != 0) {
-                               LYXERR(Debug::GRAPHICS, "Formats::getFormatFromFile\n"
-                                       << "\tCouldn't load magic database - "
-                                       << magic_error(magic_cookie));
-                       } else {
-                               char const * result = magic_file(magic_cookie,
-                                       filename.toFilesystemEncoding().c_str());
-                               string mime;
-                               if (result)
-                                       mime = token(result, ';', 0);
-                               else {
-                                       LYXERR(Debug::GRAPHICS, "Formats::getFormatFromFile\n"
-                                               << "\tCouldn't query magic database - "
-                                               << magic_error(magic_cookie));
-                               }
-                               // we need our own ps/eps detection
-                               if (!mime.empty() && mime != "application/postscript" &&
-                                   mime != "text/plain") {
-                                       Formats::const_iterator cit =
-                                               find_if(formatlist.begin(), formatlist.end(),
-                                                       FormatMimeEqual(mime));
-                                       if (cit != formats.end()) {
-                                               LYXERR(Debug::GRAPHICS, "\tgot format from MIME type: "
-                                                       << mime << " -> " << cit->name());
-                                               format = cit->name();
-                                       }
-                               }
+               // one instance of Magic that will be reused for next calls
+               // This avoids to read the magic file everytime
+               // If libmagic is not available, Magic::file returns an empty string.
+               static Magic magic;
+               string const result = magic.file(filename.toFilesystemEncoding());
+               string const mime = token(result, ';', 0);
+               // our own detection is better for binary files (can be anything)
+               // and different plain text formats
+               if (!mime.empty() && mime != "application/octet-stream" &&
+                       mime != "text/plain") {
+                       Formats::const_iterator cit =
+                               find_if(formatlist_.begin(), formatlist_.end(),
+                                               [mime](Format const & f){ return f.mime() == mime; });
+                       if (cit != formatlist_.end()) {
+                               LYXERR(Debug::GRAPHICS, "\tgot format from MIME type: "
+                                          << mime << " -> " << cit->name());
+                               // See special eps/ps handling below
+                               if (mime == "application/postscript")
+                                       psformat = cit->name();
+                               else
+                                       format = cit->name();
                        }
-                       magic_close(magic_cookie);
-                       if (!format.empty())
-                               return format;
                }
+
+               // libmagic recognizes as latex also some formats of ours
+               // such as pstex and pdftex. Therefore we have to perform
+               // additional checks in this case (bug 9244).
+               if (!format.empty() && format != "latex")
+                       return format;
        }
-#endif
 
-       string const format = guessFormatFromContents(filename);
        string const ext = getExtension(filename.absFileName());
-       if (isZippedFileFormat(format) && !ext.empty()) {
-               string const & fmt_name = formats.getFormatFromExtension(ext);
-               if (!fmt_name.empty()) {
-                       Format const * p_format = formats.getFormat(fmt_name);
-                       if (p_format && p_format->zippedNative())
-                               return p_format->name();
+       if (format.empty()) {
+               // libmagic does not distinguish eps and ps.
+               // Therefore we need to use our own detection here, but only if it
+               // recognizes either ps or eps. Otherwise the libmagic guess will
+               // be better (bug 9146).
+               format = guessFormatFromContents(filename);
+               if (!psformat.empty()) {
+                       if (isPostScriptFileFormat(format))
+                               return format;
+                       else
+                               return psformat;
                }
+
+               if (isZippedFileFormat(format) && !ext.empty()) {
+                       string const & fmt_name = getFormatFromExtension(ext);
+                       if (!fmt_name.empty()) {
+                               Format const * p_format = getFormat(fmt_name);
+                               if (p_format && p_format->zippedNative())
+                                       return p_format->name();
+                       }
+               }
+               // Don't simply return latex (bug 9244).
+               if (!format.empty() && format != "latex")
+                       return format;
+       }
+
+       // Both libmagic and our guessing from contents may return as latex
+       // also lyx files and our pstex and pdftex formats. In this case we
+       // give precedence to the format determined by the extension.
+       if (format == "latex") {
+               format = getFormatFromExtension(ext);
+               return format.empty() ? "latex" : format;
        }
-       if (!format.empty())
-               return format;
 
        // try to find a format from the file extension.
        return getFormatFromExtension(ext);
@@ -456,9 +464,9 @@ string Formats::getFormatFromExtension(string const & ext) const
                // this is ambigous if two formats have the same extension,
                // but better than nothing
                Formats::const_iterator cit =
-                       find_if(formatlist.begin(), formatlist.end(),
-                               FormatExtensionsEqual(ext));
-               if (cit != formats.end()) {
+                       find_if(formatlist_.begin(), formatlist_.end(),
+                               [ext](Format const & f){ return f.hasExtension(ext); });
+               if (cit != formatlist_.end()) {
                        LYXERR(Debug::GRAPHICS, "\twill guess format from file extension: "
                                << ext << " -> " << cit->name());
                        return cit->name();
@@ -479,14 +487,20 @@ struct ZippedInfo {
 
 /// Mapping absolute pathnames of files to their ZippedInfo metadata.
 static std::map<std::string, ZippedInfo> zipped_;
+static Mutex zipped_mutex;
 
 
 bool Formats::isZippedFile(support::FileName const & filename) const {
        string const & fname = filename.absFileName();
        time_t timestamp = filename.lastModified();
+       Mutex::Locker lock(&zipped_mutex);
        map<string, ZippedInfo>::iterator it = zipped_.find(fname);
        if (it != zipped_.end() && it->second.timestamp == timestamp)
                return it->second.zipped;
+       // FIXME perf: This very expensive function is called on startup on each
+       // file whic is going to be parsed, and also on svgz icons. Maybe there is a
+       // quicker way to check whether a file is zipped?  I.e. for icons we
+       // probably just need to check the extension (svgz vs svg).
        string const & format = getFormatFromFile(filename);
        bool zipped = (format == "gzip" || format == "zip");
        zipped_.insert(make_pair(fname, ZippedInfo(zipped, timestamp)));
@@ -527,8 +541,8 @@ static string fixCommand(string const & cmd, string const & ext,
 
 void Formats::setAutoOpen()
 {
-       FormatList::iterator fit = formatlist.begin();
-       FormatList::iterator const fend = formatlist.end();
+       FormatList::iterator fit = formatlist_.begin();
+       FormatList::iterator const fend = formatlist_.end();
        for ( ; fit != fend ; ++fit) {
                fit->setViewer(fixCommand(fit->viewer(), fit->extension(), os::VIEW));
                fit->setEditor(fixCommand(fit->editor(), fit->extension(), os::EDIT));
@@ -539,75 +553,73 @@ void Formats::setAutoOpen()
 int Formats::getNumber(string const & name) const
 {
        FormatList::const_iterator cit =
-               find_if(formatlist.begin(), formatlist.end(),
-                       FormatNamesEqual(name));
-       if (cit != formatlist.end())
-               return distance(formatlist.begin(), cit);
-       else
+               find_if(formatlist_.begin(), formatlist_.end(),
+                       FormatNameIs(name));
+       if (cit == formatlist_.end())
                return -1;
+
+       return distance(formatlist_.begin(), cit);
 }
 
 
 void Formats::add(string const & name)
 {
        if (!getFormat(name))
-               add(name, name, name, string(), string(), string(),
+               add(name, name, from_utf8(name), string(), string(), string(),
                    string(), Format::document);
 }
 
 
 void Formats::add(string const & name, string const & extensions,
-                 string const & prettyname, string const & shortcut,
+                 docstring const & prettyname, string const & shortcut,
                  string const & viewer, string const & editor,
                  string const & mime, int flags)
 {
-       FormatList::iterator it =
-               find_if(formatlist.begin(), formatlist.end(),
-                       FormatNamesEqual(name));
-       if (it == formatlist.end())
-               formatlist.push_back(Format(name, extensions, prettyname,
-                                           shortcut, viewer, editor, mime, flags));
+       Format * format = getFormat(name);
+       if (format)
+               *format = Format(name, extensions, prettyname, shortcut, viewer,
+                                editor, mime, flags);
        else
-               *it = Format(name, extensions, prettyname, shortcut, viewer,
-                            editor, mime, flags);
+               formatlist_.push_back(Format(name, extensions, prettyname,
+                                               shortcut, viewer, editor, mime, flags));
 }
 
 
 void Formats::erase(string const & name)
 {
        FormatList::iterator it =
-               find_if(formatlist.begin(), formatlist.end(),
-                       FormatNamesEqual(name));
-       if (it != formatlist.end())
-               formatlist.erase(it);
+               find_if(formatlist_.begin(), formatlist_.end(),
+                       FormatNameIs(name));
+       if (it != formatlist_.end())
+               formatlist_.erase(it);
 }
 
 
 void Formats::sort()
 {
-       std::sort(formatlist.begin(), formatlist.end());
+       std::sort(formatlist_.begin(), formatlist_.end());
 }
 
 
 void Formats::setViewer(string const & name, string const & command)
 {
        add(name);
-       FormatList::iterator it =
-               find_if(formatlist.begin(), formatlist.end(),
-                       FormatNamesEqual(name));
-       if (it != formatlist.end())
-               it->setViewer(command);
+       Format * format = getFormat(name);
+       if (format)
+               format->setViewer(command);
+       else
+               LYXERR0("Unable to set viewer for non-existent format: " << name);
 }
 
 
 void Formats::setEditor(string const & name, string const & command)
 {
        add(name);
-       FormatList::iterator it =
-               find_if(formatlist.begin(), formatlist.end(),
-                       FormatNamesEqual(name));
-       if (it != formatlist.end())
-               it->setEditor(command);
+       Format * format = getFormat(name);
+       if (format)
+               format->setEditor(command);
+       else
+               LYXERR0("Unable to set editor for non-existent format: " << name);
 }
 
 
@@ -645,7 +657,29 @@ bool Formats::view(Buffer const & buffer, FileName const & filename,
                }
        }
 
-       string command = libScriptSearch(format->viewer());
+       string command = format->viewer();
+
+       // Escape backslashes if not already in double or single quotes.
+       // We cannot simply quote the whole command as there may be arguments.
+       if (contains(command, '\\')) {
+               bool inquote1 = false;
+               bool inquote2 = false;
+               string::iterator cit = command.begin();
+               for (; cit != command.end(); ++cit) {
+                       switch (*cit) {
+                       case '"':
+                               inquote1 = !inquote1;
+                               break;
+                       case '\'':
+                               inquote2 = !inquote2;
+                               break;
+                       case '\\':
+                               if (!inquote1 && !inquote2)
+                                       cit = ++command.insert(cit, '\\');
+                               break;
+                       }
+               }
+       }
 
        if (format_name == "dvi" &&
            !lyxrc.view_dvi_paper_option.empty()) {
@@ -662,8 +696,10 @@ bool Formats::view(Buffer const & buffer, FileName const & filename,
        if (!contains(command, token_from_format))
                command += ' ' + token_from_format;
 
-       command = subst(command, token_from_format, quoteName(onlyFileName(filename.toFilesystemEncoding())));
-       command = subst(command, token_path_format, quoteName(onlyPath(filename.toFilesystemEncoding())));
+       command = subst(command, token_from_format,
+                       quoteName(onlyFileName(filename.toFilesystemEncoding()), quote_shell_filename));
+       command = subst(command, token_path_format,
+                       quoteName(onlyPath(filename.toFilesystemEncoding()), quote_shell_filename));
        command = subst(command, token_socket_format, quoteName(theServerSocket().address()));
        LYXERR(Debug::FILES, "Executing command: " << command);
        // FIXME UNICODE utf8 can be wrong for files
@@ -671,7 +707,8 @@ bool Formats::view(Buffer const & buffer, FileName const & filename,
 
        PathChanger p(filename.onlyPath());
        Systemcall one;
-       one.startscript(Systemcall::DontWait, command, buffer.filePath());
+       one.startscript(Systemcall::DontWait, command,
+                       buffer.filePath(), buffer.layoutPos());
 
        // we can't report any sort of error, since we aren't waiting
        return true;
@@ -730,15 +767,18 @@ bool Formats::edit(Buffer const & buffer, FileName const & filename,
        if (!contains(command, token_from_format))
                command += ' ' + token_from_format;
 
-       command = subst(command, token_from_format, quoteName(filename.toFilesystemEncoding()));
-       command = subst(command, token_path_format, quoteName(onlyPath(filename.toFilesystemEncoding())));
+       command = subst(command, token_from_format,
+                       quoteName(filename.toFilesystemEncoding(), quote_shell_filename));
+       command = subst(command, token_path_format,
+                       quoteName(onlyPath(filename.toFilesystemEncoding()), quote_shell_filename));
        command = subst(command, token_socket_format, quoteName(theServerSocket().address()));
        LYXERR(Debug::FILES, "Executing command: " << command);
        // FIXME UNICODE utf8 can be wrong for files
        buffer.message(_("Executing command: ") + from_utf8(command));
 
        Systemcall one;
-       one.startscript(Systemcall::DontWait, command, buffer.filePath());
+       one.startscript(Systemcall::DontWait, command,
+                       buffer.filePath(), buffer.layoutPos());
 
        // we can't report any sort of error, since we aren't waiting
        return true;
@@ -749,7 +789,7 @@ docstring const Formats::prettyName(string const & name) const
 {
        Format const * format = getFormat(name);
        if (format)
-               return from_utf8(format->prettyname());
+               return format->prettyname();
        else
                return from_utf8(name);
 }
@@ -790,17 +830,18 @@ FlavorTranslator initFlavorTranslator()
        f.addPair(OutputParams::XML, "docbook-xml");
        f.addPair(OutputParams::HTML, "xhtml");
        f.addPair(OutputParams::TEXT, "text");
+       f.addPair(OutputParams::LYX, "lyx");
        return f;
 }
 
 
 FlavorTranslator const & flavorTranslator()
 {
-       static FlavorTranslator translator = initFlavorTranslator();
+       static FlavorTranslator const translator = initFlavorTranslator();
        return translator;
 }
 
-}
+} // namespace
 
 
 std::string flavor2format(OutputParams::FLAVOR flavor)
@@ -815,9 +856,4 @@ OutputParams::FLAVOR format2flavor(std::string fmt)
        return flavorTranslator().find(fmt);
 } */
 
-Formats formats;
-
-Formats system_formats;
-
-
 } // namespace lyx