X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2Ftex2lyx%2FParser.h;h=19f0c5fbe33a61ab5bbbb5af4a06fcef8dce8fb4;hb=70b0f8d953f14d450e817294204b4069cc16db8b;hp=558b5560bafcd7136ee8939b5a065722049a32f1;hpb=be42f1398db05353bdab6fa328a4e86d11ce6b97;p=lyx.git
diff --git a/src/tex2lyx/Parser.h b/src/tex2lyx/Parser.h
index 558b5560ba..19f0c5fbe3 100644
--- a/src/tex2lyx/Parser.h
+++ b/src/tex2lyx/Parser.h
@@ -117,6 +117,44 @@ std::ostream & operator<<(std::ostream & os, Token const & t);
extern void debugToken(std::ostream & os, Token const & t, unsigned int flags);
#endif
+/// A docstream version that supports putback even when not buffered
+class iparserdocstream
+{
+public:
+ typedef idocstream::int_type int_type;
+
+ iparserdocstream(idocstream & is) : is_(is) {}
+
+ /// Like std::istream::operator bool()
+ /// Do not convert is_ implicitly to bool, since that is forbidden in C++11.
+ explicit operator bool() const { return s_.empty() ? !is_.fail() : true; }
+
+ /// change the encoding of the input stream to \p e (iconv name)
+ void setEncoding(std::string const & e);
+
+ // add to the list of characters to read before actually reading
+ // the stream
+ void putback(char_type c);
+
+ // add to the list of characters to read before actually reading
+ // the stream
+ void putback(docstring s);
+
+ /// Like std::istream::get()
+ iparserdocstream & get(char_type &c);
+
+ /// Like std::istream::good()
+ bool good() const { return s_.empty() ? is_.good() : true; }
+
+ /// Like std::istream::peek()
+ int_type peek() const { return s_.empty() ? is_.peek() : s_[0]; }
+private:
+ ///
+ idocstream & is_;
+ /// characters to read before actually reading the stream
+ docstring s_;
+};
+
/*!
* Actual parser class
@@ -135,12 +173,25 @@ class Parser {
Parser & operator=(Parser const & p);
public:
///
- Parser(idocstream & is);
+ Parser(idocstream & is, std::string const & fixedenc);
///
Parser(std::string const & s);
///
~Parser();
+ /** forget already parsed next tokens and put the
+ * corresponding characters into the input stream for
+ * re-reading. Useful when changing catcodes. */
+ void deparse();
+
+ /// change the encoding of the input stream according to \p encoding
+ /// (latex name) and package \p package
+ bool setEncoding(std::string const & encoding, int const & package);
+ /// change the encoding of the input stream to \p encoding (iconv name)
+ bool setEncoding(std::string const & encoding);
+ /// get the current iconv encoding of the input stream
+ std::string getEncoding() const { return encoding_iconv_; }
+
///
CatCode catcode(char_type c) const;
///
@@ -148,14 +199,6 @@ public:
/// set parser to normal or verbatim mode
void setCatcodes(cat_type t);
- /// change the iconv encoding of the input stream
- /// according to the latex encoding and package
- void setEncoding(std::string const & encoding, int const & package);
- /// change the iconv encoding of the input stream
- void setEncoding(std::string const & encoding);
- /// get the current iconv encoding of the input stream
- std::string getEncoding() const { return encoding_iconv_; }
-
///
int lineno() const { return lineno_; }
///
@@ -164,6 +207,8 @@ public:
void pushPosition();
/// restore previous position
void popPosition();
+ /// forget last saved position
+ void dropPosition();
/// dump contents to screen
void dump() const;
@@ -236,27 +281,37 @@ public:
* stopped at string \p end_string. Contrary to the other
* methods, this uses proper catcode setting. This function is
* designed to parse verbatim environments and command. The
- * intention is to eventually replace all of its siblings.
+ * intention is to eventually replace all of its siblings. the
+ * member \p first of the result tells whether the arg was
+ * found and the member \p second is the value. If \p
+ * allow_linebreak is false, then the parsing is limited to one line
*/
- std::string const verbatimStuff(std::string const & end_string);
- /*!
- * Returns the character of the current token and increments
- * the token position.
+ Arg verbatimStuff(std::string const & end_string,
+ bool allow_linebreak = true);
+ /*
+ * \returns the contents of the environment \p name.
+ * \begin{name} must be parsed already,
+ * \end{name} is parsed but not returned. The string
+ * is parsed with proper verbatim catcodes and one newline is
+ * removed from head and tail of the string if applicable.
*/
- char getChar();
+ std::string const verbatimEnvironment(std::string const & end_string);
///
- void error(std::string const & msg);
- /// Parses one token from \p is
- void tokenize_one();
+ std::string verbatim_item();
///
- void push_back(Token const & t);
+ std::string verbatimOption();
+ ///
+ void error(std::string const & msg);
/// The previous token.
Token const prev_token() const;
/// The current token.
Token const curr_token() const;
- /// The next token.
+ /// The next token. Caution: If this is called, an encoding change is
+ /// only possible again after get_token() has been called.
Token const next_token();
- /// The next but one token.
+ /// The next but one token. Caution: If this is called, an encoding
+ /// change is only possible again after get_token() has been called
+ /// twice.
Token const next_next_token();
/// Make the next token current and return that.
Token const get_token();
@@ -267,40 +322,42 @@ public:
bool skip_spaces(bool skip_comments = false);
/// puts back spaces (and comments if \p skip_comments is true)
void unskip_spaces(bool skip_comments = false);
- ///
- void lex(std::string const & s);
- ///
+ /// Is any further input pending()? This is not like
+ /// std::istream::good(), which returns true if all available input
+ /// was read, and the next attempt to read would return EOF.
bool good();
- ///
- std::string verbatim_item();
- ///
- std::string verbatimOption();
/// resets the parser to initial state
void reset();
private:
/// Setup catcode table
void catInit();
+ /// Parses one token from \p is
+ void tokenize_one();
+ ///
+ void push_back(Token const & t);
///
int lineno_;
///
std::vector tokens_;
///
- unsigned pos_;
+ size_t pos_;
///
std::vector positions_;
///
idocstringstream * iss_;
///
- idocstream & is_;
+ iparserdocstream is_;
/// iconv name of the current encoding
std::string encoding_iconv_;
///
CatCode theCatcode_[256];
- //
+ ///
cat_type theCatcodesType_;
- //
+ ///
cat_type curr_cat_;
+ ///
+ bool fixed_enc_;
};