Really fix bug #4468.

The old fix was incomplete (\verb~\~ was translated to \verb~~ in roundtrip).
The real cause for this bug (and also the mistranslation of \href{...}{\}})
was the misbehaviour of Token::character() (see comment in Parser.h): This
method even returns a character if the category is catEscape, and this is not
wanted in most (all?) cases.

diff --git a/src/tex2lyx/Parser.cpp b/src/tex2lyx/Parser.cpp
index 75b8ba4bdd3be80e6e7187cc13875f5fadf7eb91..530e73aa474530cbedcc8e98e3af4984701dc112 100644 (file)
--- a/src/tex2lyx/Parser.cpp
+++ b/src/tex2lyx/Parser.cpp
@@ -383,7 +383,7 @@ bool Parser::hasOpt()
 }
 
 
-Parser::Arg Parser::getFullArg(char left, char right)
+Parser::Arg Parser::getFullArg(char left, char right, bool allow_escaping)
 {
        skip_spaces(true);
 
@@ -393,36 +393,40 @@ Parser::Arg Parser::getFullArg(char left, char right)
                return make_pair(false, string());
 
        string result;
-       char c = getChar();
+       Token t = get_token();
 
-       if (c != left) {
+       if (t.cat() == catComment || t.cat() == catEscape ||
+           t.character() != left) {
                putback();
                return make_pair(false, string());
        } else {
-               // a single '\' is only allowed within \verb, no matter what the delimiter is,
-               // for example "\verb+\+" (reported as bug #4468)
-               // To support this, we allow single '\' if it is the only character
-               // within equal delimiters
-               if (next_token().cat() == catEscape)
-                       if (next_token().character() == right && right == left)
-                               result += '\\';
-               while ((c = getChar()) != right && good()) {
+               for (t = get_token(); good(); t = get_token()) {
                        // Ignore comments
-                       if (curr_token().cat() == catComment) {
-                               if (!curr_token().cs().empty())
-                                       cerr << "Ignoring comment: " << curr_token().asInput();
+                       if (t.cat() == catComment) {
+                               if (!t.cs().empty())
+                                       cerr << "Ignoring comment: " << t.asInput();
+                               continue;
                        }
-                       else
-                               result += curr_token().asInput();
+                       if (allow_escaping) {
+                               if (t.cat() != catEscape && t.character() == right)
+                                       break;
+                       } else {
+                               if (t.character() == right) {
+                                       if (t.cat() == catEscape)
+                                               result += '\\';
+                                       break;
+                               }
+                       }
+                       result += t.asInput();
                }
        }
        return make_pair(true, result);
 }
 
 
-string Parser::getArg(char left, char right)
+string Parser::getArg(char left, char right, bool allow_escaping)
 {
-       return getFullArg(left, right).second;
+       return getFullArg(left, right, allow_escaping).second;
 }
 
 

diff --git a/src/tex2lyx/Parser.h b/src/tex2lyx/Parser.h
index c0c5685bb83fc449cf2567c39f00206b38fb3629..59a89d5a75c901f2e9726fad0facc74efeb4bc5e 100644 (file)
--- a/src/tex2lyx/Parser.h
+++ b/src/tex2lyx/Parser.h
@@ -87,8 +87,8 @@ public:
         *        ../mathed/MathParser.cpp (which is the anchestor of this
         *        class) uses a separate char member for this method. I
         *        believe that the intended usage is to not cover tokens with
-        *        catEscape, e.g. \code
-        *        return (cs_.empty() || cat_ == catEscape) ? 0 : cs_[0];
+        *        catEscape or catComment, e.g. \code
+        *        return (cs_.empty() || cat_ == catEscape || cat_ == catComment) ? 0 : cs_[0];
         *        \endcode
         *        All usages of this method should be checked. gb 2011-01-05
         */
@@ -157,18 +157,24 @@ public:
        typedef std::pair<bool, std::string> Arg;
        /*!
         * Get an argument enclosed by \p left and \p right.
+        * If \p allow_escaping is true, a right delimiter escaped by a
+        * backslash does not count as delimiter, but is included in the
+        * argument.
         * \returns wether an argument was found in \p Arg.first and the
         * argument in \p Arg.second. \see getArg().
         */
-       Arg getFullArg(char left, char right);
+       Arg getFullArg(char left, char right, bool allow_escaping = true);
        /*!
         * Get an argument enclosed by \p left and \p right.
+        * If \p allow_escaping is true, a right delimiter escaped by a
+        * backslash does not count as delimiter, but is included in the
+        * argument.
         * \returns the argument (without \p left and \p right) or the empty
         * string if the next non-space token is not \p left. Use
         * getFullArg() if you need to know wether there was an empty
         * argument or no argument at all.
         */
-       std::string getArg(char left, char right);
+       std::string getArg(char left, char right, bool allow_escaping = true);
        /*!
         * Like getOpt(), but distinguishes between a missing argument ""
         * and an empty argument "[]".

diff --git a/src/tex2lyx/test/test-insets.lyx.lyx b/src/tex2lyx/test/test-insets.lyx.lyx
index a854dacbac4a05dbefcf174f93f6f9c6b30fe7f4..f5484ad87cb5f6dd083cb5ee2c8d5122611e14e4 100644 (file)
--- a/src/tex2lyx/test/test-insets.lyx.lyx
+++ b/src/tex2lyx/test/test-insets.lyx.lyx
@@ -968,6 +968,19 @@ target "http://www.test.test"
 \end_inset
 
 
+\end_layout
+
+\begin_layout Standard
+
+parser test (escaped):
+\begin_inset CommandInset href
+LatexCommand href
+name "a brace } and another one { and something"
+target "http://www.test.test"
+
+\end_inset
+
+
 \end_layout
 
 \begin_layout Section

diff --git a/src/tex2lyx/test/test-insets.tex b/src/tex2lyx/test/test-insets.tex
index 16354ba807e0231998822f90d62a9494a795e54e..a74a542289575e566c725509300075dd9a8ea8c5 100644 (file)
--- a/src/tex2lyx/test/test-insets.tex
+++ b/src/tex2lyx/test/test-insets.tex
@@ -160,6 +160,8 @@ ftp2:\href{ftp://www.test.test}{www.test.test}
 
 parser test (stupid, but valid):\href{http://www.test.test}{\}}
 
+parser test (escaped):\href{http://www.test.test}{a brace \} and another one \{ and something}
+
 
 \section{Lists\index{Lists}}
 

diff --git a/src/tex2lyx/test/test-structure.lyx.lyx b/src/tex2lyx/test/test-structure.lyx.lyx
index 27761e86f5146f0cf243f38379e86c1cb1c20a2d..8fc60c923fec270d51598aacc055178770e65e06 100644 (file)
--- a/src/tex2lyx/test/test-structure.lyx.lyx
+++ b/src/tex2lyx/test/test-structure.lyx.lyx
@@ -1119,7 +1119,9 @@ status collapsed
 
 
 \backslash
-verb~~
+verb~
+\backslash
+~
 \end_layout
 
 \end_inset
@@ -1139,7 +1141,22 @@ item[ABC] first item+
 
 \end_inset
 
+ 
+\begin_inset ERT
+status collapsed
+
+\begin_layout Standard
+
+
+\backslash
+verb+something
+\backslash
++
+\end_layout
+
+\end_inset
 
+ bug 4468
 \end_layout
 
 \begin_layout Standard

diff --git a/src/tex2lyx/test/test-structure.tex b/src/tex2lyx/test/test-structure.tex
index 89fa58915c1d362156943d8aee572b1d54e153ee..79989fc443537594ebb8f0bdba9196453a2cc0b6 100644 (file)
--- a/src/tex2lyx/test/test-structure.tex
+++ b/src/tex2lyx/test/test-structure.tex
@@ -321,6 +321,7 @@ zzz \section{
 \end{verbatim}
 \verb~\~
 \verb+\item[ABC] first item+
+\verb+something\+ bug 4468
 
 and bibliography:
 \begin{thebibliography}{9}

diff --git a/src/tex2lyx/text.cpp b/src/tex2lyx/text.cpp
index db9f9b5adc7be4b10020a6b7453a2bbdcc40c6d7..2ee3afc83249b86bebabc4926fe4e155598db4b6 100644 (file)
--- a/src/tex2lyx/text.cpp
+++ b/src/tex2lyx/text.cpp
@@ -3210,8 +3210,8 @@ 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) {
@@ -3729,7 +3729,11 @@ 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);