* This file is part of LyX, the document processor.
* Licence details can be found in the file COPYING.
*
- * \author André Pönitz
+ * \author André Pönitz
*
* Full author contact details are available in file CREDITS.
*/
#include <config.h>
-#include "Encoding.h"
#include "Parser.h"
+
+#include "Encoding.h"
+#include "support/lstrings.h"
#include "support/textutils.h"
#include <iostream>
using namespace std;
+using namespace lyx::support;
namespace lyx {
namespace {
-CatCode theCatcode[256];
-
-void catInit()
-{
- static bool init_done = false;
- if (init_done)
- return;
- init_done = true;
-
- fill(theCatcode, theCatcode + 256, catOther);
- fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
- fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
-
- theCatcode[int('\\')] = catEscape;
- theCatcode[int('{')] = catBegin;
- theCatcode[int('}')] = catEnd;
- theCatcode[int('$')] = catMath;
- theCatcode[int('&')] = catAlign;
- theCatcode[int('\n')] = catNewline;
- theCatcode[int('#')] = catParameter;
- theCatcode[int('^')] = catSuper;
- theCatcode[int('_')] = catSub;
- theCatcode[0x7f] = catIgnore;
- theCatcode[int(' ')] = catSpace;
- theCatcode[int('\t')] = catSpace;
- theCatcode[int('\r')] = catNewline;
- theCatcode[int('~')] = catActive;
- theCatcode[int('%')] = catComment;
-
- // This is wrong!
- theCatcode[int('@')] = catLetter;
-}
-
/*!
* Translate a line ending to '\n'.
* \p c must have catcode catNewline, and it must be the last character read
* from \p is.
*/
-char_type getNewline(idocstream & is, char_type c)
+char_type getNewline(iparserdocstream & is, char_type c)
{
// we have to handle 3 different line endings:
// - UNIX (\n)
return c;
}
-CatCode catcode(char_type c)
-{
- if (c < 256)
- return theCatcode[(unsigned char)c];
- return catOther;
-}
-
-}
-
+} // namespace
//
// Token
#endif
+//
+// Wrapper
+//
+
+void iparserdocstream::setEncoding(std::string const & e)
+{
+ is_ << lyx::setEncoding(e);
+}
+
+
+void iparserdocstream::putback(char_type c)
+{
+ s_ = c + s_;
+}
+
+
+void iparserdocstream::putback(docstring const & s)
+{
+ s_ = s + s_;
+}
+
+
+iparserdocstream & iparserdocstream::get(char_type &c)
+{
+ if (s_.empty())
+ is_.get(c);
+ else {
+ //cerr << "unparsed: " << to_utf8(s_) <<endl;
+ c = s_[0];
+ s_.erase(0,1);
+ }
+ return *this;
+}
+
+
//
// Parser
//
-Parser::Parser(idocstream & is)
- : lineno_(0), pos_(0), iss_(0), is_(is), encoding_latex_("utf8")
+Parser::Parser(idocstream & is, std::string const & fixedenc)
+ : lineno_(0), pos_(0), iss_(nullptr), is_(is),
+ encoding_iconv_(fixedenc.empty() ? "UTF-8" : fixedenc),
+ theCatcodesType_(NORMAL_CATCODES), curr_cat_(UNDECIDED_CATCODES),
+ fixed_enc_(!fixedenc.empty())
{
+ if (fixed_enc_)
+ is_.setEncoding(fixedenc);
+ catInit();
}
Parser::Parser(string const & s)
- : lineno_(0), pos_(0),
- iss_(new idocstringstream(from_utf8(s))), is_(*iss_),
- encoding_latex_("utf8")
+ : lineno_(0), pos_(0),
+ iss_(new idocstringstream(from_utf8(s))), is_(*iss_),
+ encoding_iconv_("UTF-8"),
+ theCatcodesType_(NORMAL_CATCODES), curr_cat_(UNDECIDED_CATCODES),
+ // An idocstringstream can not change the encoding
+ fixed_enc_(true)
{
+ catInit();
}
}
-void Parser::setEncoding(std::string const & e)
+void Parser::deparse()
+{
+ string s;
+ for(size_type i = pos_ ; i < tokens_.size() ; ++i) {
+ s += tokens_[i].asInput();
+ }
+ is_.putback(from_utf8(s));
+ tokens_.erase(tokens_.begin() + pos_, tokens_.end());
+ // make sure that next token is read
+ tokenize_one();
+}
+
+
+bool Parser::setEncoding(std::string const & e, int p)
{
- Encoding const * enc = encodings.fromLaTeXName(e);
+ // We may (and need to) use unsafe encodings here: Since the text is
+ // converted to unicode while reading from is_, we never see text in
+ // the original encoding of the parser, but operate on utf8 strings
+ // instead. Therefore, we cannot misparse high bytes as {, } or \\.
+ Encoding const * const enc = encodings.fromLaTeXName(e, p, true);
if (!enc) {
cerr << "Unknown encoding " << e << ". Ignoring." << std::endl;
+ return false;
+ }
+ return setEncoding(enc->iconvName());
+}
+
+
+void Parser::catInit()
+{
+ if (curr_cat_ == theCatcodesType_)
return;
+ curr_cat_ = theCatcodesType_;
+
+ fill(theCatcode_, theCatcode_ + 256, catOther);
+ fill(theCatcode_ + 'a', theCatcode_ + 'z' + 1, catLetter);
+ fill(theCatcode_ + 'A', theCatcode_ + 'Z' + 1, catLetter);
+ // This is wrong!
+ theCatcode_[int('@')] = catLetter;
+
+ if (theCatcodesType_ == NORMAL_CATCODES) {
+ theCatcode_[int('\\')] = catEscape;
+ theCatcode_[int('{')] = catBegin;
+ theCatcode_[int('}')] = catEnd;
+ theCatcode_[int('$')] = catMath;
+ theCatcode_[int('&')] = catAlign;
+ theCatcode_[int('\n')] = catNewline;
+ theCatcode_[int('#')] = catParameter;
+ theCatcode_[int('^')] = catSuper;
+ theCatcode_[int('_')] = catSub;
+ theCatcode_[0x7f] = catIgnore;
+ theCatcode_[int(' ')] = catSpace;
+ theCatcode_[int('\t')] = catSpace;
+ theCatcode_[int('\r')] = catNewline;
+ theCatcode_[int('~')] = catActive;
+ theCatcode_[int('%')] = catComment;
}
- //cerr << "setting encoding to " << enc->iconvName() << std::endl;
- is_ << lyx::setEncoding(enc->iconvName());
- encoding_latex_ = e;
+}
+
+CatCode Parser::catcode(char_type c) const
+{
+ if (c < 256)
+ return theCatcode_[(unsigned char)c];
+ return catOther;
+}
+
+
+void Parser::setCatcode(char c, CatCode cat)
+{
+ theCatcode_[(unsigned char)c] = cat;
+ deparse();
+}
+
+
+void Parser::setCatcodes(cat_type t)
+{
+ theCatcodesType_ = t;
+ deparse();
+}
+
+
+bool Parser::setEncoding(std::string const & e)
+{
+ //cerr << "setting encoding to " << e << std::endl;
+ encoding_iconv_ = e;
+ // If the encoding is fixed, we must not change the stream encoding
+ // (because the whole input uses that encoding, e.g. if it comes from
+ // the clipboard). We still need to track the original encoding in
+ // encoding_iconv_, so that the generated output is correct.
+ if (!fixed_enc_)
+ is_.setEncoding(e);
+ return true;
}
Token const Parser::next_token()
{
static const Token dummy;
- return good() ? tokens_[pos_] : dummy;
+ if (!good())
+ return dummy;
+ if (pos_ >= tokens_.size())
+ tokenize_one();
+ return pos_ < tokens_.size() ? tokens_[pos_] : dummy;
}
Token const Parser::next_next_token()
{
static const Token dummy;
- // If good() has not been called after the last get_token() we need
- // to tokenize two more tokens.
- if (pos_ + 1 >= tokens_.size()) {
+ if (!good())
+ return dummy;
+ // If tokenize_one() has not been called after the last get_token() we
+ // need to tokenize two more tokens.
+ if (pos_ >= tokens_.size())
tokenize_one();
+ if (pos_ + 1 >= tokens_.size())
tokenize_one();
- }
return pos_ + 1 < tokens_.size() ? tokens_[pos_ + 1] : dummy;
}
Token const Parser::get_token()
{
static const Token dummy;
- //cerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << '\n';
- return good() ? tokens_[pos_++] : dummy;
+ if (!good())
+ return dummy;
+ if (pos_ >= tokens_.size()) {
+ tokenize_one();
+ if (pos_ >= tokens_.size())
+ return dummy;
+ }
+ // cerr << "looking at token " << tokens_[pos_]
+ // << " pos: " << pos_ << '\n';
+ return tokens_[pos_++];
}
bool Parser::isParagraph()
{
- // A new paragraph in TeX ist started
+ // A new paragraph in TeX is started
// - either by a newline, following any amount of whitespace
// characters (including zero), and another newline
// - or the token \par
{
pos_ = positions_.back();
positions_.pop_back();
+ deparse();
}
-bool Parser::good()
+void Parser::dropPosition()
{
- if (pos_ < tokens_.size())
- return true;
- tokenize_one();
- return pos_ < tokens_.size();
+ positions_.pop_back();
}
-char Parser::getChar()
+bool Parser::good() const
{
- if (!good())
- error("The input stream is not well...");
- return get_token().character();
+ if (pos_ < tokens_.size())
+ return true;
+ if (!is_.good())
+ return false;
+ return is_.peek() != idocstream::traits_type::eof();
}
-bool Parser::hasOpt()
+bool Parser::hasOpt(string const & l)
{
// An optional argument can occur in any of the following forms:
// - \foo[bar]
putback();
break;
}
- bool const retval = (next_token().asInput() == "[");
+ bool const retval = (next_token().asInput() == l);
pos_ = oldpos;
return retval;
}
-Parser::Arg Parser::getFullArg(char left, char right)
+Parser::Arg Parser::getFullArg(char left, char right, bool allow_escaping)
{
skip_spaces(true);
if (! good())
return make_pair(false, string());
+ int group_level = 0;
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
- while ((c = getChar()) != right && good()) {
+ } else {
+ while (good()) {
+ t = get_token();
+ // honor grouping
+ if (left != '{' && t.cat() == catBegin) {
+ ++group_level;
+ continue;
+ }
+ if (left != '{' && t.cat() == catEnd) {
+ --group_level;
+ continue;
+ }
// 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
+ && group_level == 0)
+ break;
+ } else {
+ if (t.character() == right) {
+ if (t.cat() == catEscape)
+ result += '\\';
+ if (group_level == 0)
+ 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;
}
-string Parser::getFullOpt(bool keepws)
+string Parser::getFullOpt(bool keepws, char left, char right)
{
- Arg arg = getFullArg('[', ']');
+ Arg arg = getFullArg(left, right);
if (arg.first)
- return '[' + arg.second + ']';
+ return left + arg.second + right;
if (keepws)
unskip_spaces(true);
return string();
}
-string const Parser::verbatimEnvironment(string const & name)
+bool Parser::hasListPreamble(string const & itemcmd)
+{
+ // remember current position
+ unsigned int oldpos = pos_;
+ // jump over arguments
+ if (hasOpt())
+ getOpt();
+ if (hasOpt("{"))
+ getArg('{', '}');
+ // and swallow spaces and comments
+ skip_spaces(true);
+ // we have a list preamble if the next thing
+ // that follows is not the \item command
+ bool res = next_token().cs() != itemcmd;
+ // back to orig position
+ pos_ = oldpos;
+ return res;
+}
+
+
+string const Parser::ertEnvironment(string const & name)
{
if (!good())
return string();
} else if (t.asInput() == "\\begin") {
string const env = getArg('{', '}');
os << "\\begin{" << env << '}'
- << verbatimEnvironment(env)
+ << ertEnvironment(env)
<< "\\end{" << env << '}';
} else if (t.asInput() == "\\end") {
string const end = getArg('{', '}');
ostringstream os;
for (Token t = get_token(); good(); t = get_token()) {
- if (t.cat() == catBegin) {
- putback();
- os << '{' << verbatim_item() << '}';
- } else if (t.asInput() == "\\end") {
+ if (t.asInput() == "\\end") {
string const end = getArg('{', '}');
if (end == name)
return os.str();
}
+string const Parser::plainCommand(char left, char right, string const & name)
+{
+ if (!good())
+ return string();
+ // check if first token is really the start character
+ Token tok = get_token();
+ if (tok.character() != left) {
+ cerr << "first character does not match start character of command \\" << name << endl;
+ return string();
+ }
+ ostringstream os;
+ for (Token t = get_token(); good(); t = get_token()) {
+ if (t.character() == right) {
+ return os.str();
+ } else
+ os << t.asInput();
+ }
+ cerr << "unexpected end of input" << endl;
+ return os.str();
+}
+
+
+string const Parser::getCommandLatexParam()
+{
+ if (!good())
+ return string();
+ string res;
+ size_t offset = 0;
+ while (true) {
+ if (pos_ + offset >= tokens_.size())
+ tokenize_one();
+ if (pos_ + offset >= tokens_.size())
+ break;
+ Token t = tokens_[pos_ + offset];
+ if (t.cat() == catBegin)
+ break;
+ res += t.asInput();
+ ++offset;
+ }
+ return res;
+}
+
+
+Parser::Arg Parser::verbatimStuff(string const & end_string, bool const allow_linebreak)
+{
+ if (!good())
+ return Arg(false, string());
+
+ pushPosition();
+ ostringstream oss;
+ size_t match_index = 0;
+ setCatcodes(VERBATIM_CATCODES);
+ for (Token t = get_token(); good(); t = get_token()) {
+ // FIXME t.asInput() might be longer than we need ?
+ if (t.asInput() == end_string.substr(match_index,
+ t.asInput().length())) {
+ match_index += t.asInput().length();
+ if (match_index >= end_string.length())
+ break;
+ } else {
+ if (!allow_linebreak && t.asInput() == "\n") {
+ cerr << "unexpected end of input" << endl;
+ popPosition();
+ setCatcodes(NORMAL_CATCODES);
+ return Arg(false, string());
+ }
+ if (match_index) {
+ oss << end_string.substr(0, match_index)
+ << t.asInput();
+ match_index = 0;
+ } else
+ oss << t.asInput();
+ }
+ }
+
+ if (!good()) {
+ cerr << "unexpected end of input" << endl;
+ popPosition();
+ setCatcodes(NORMAL_CATCODES);
+ return Arg(false, string());
+ }
+ setCatcodes(NORMAL_CATCODES);
+ dropPosition();
+ return Arg(true, oss.str());
+}
+
+
+string const Parser::verbatimEnvironment(string const & name)
+{
+ //FIXME: do something if endstring is not found
+ string s = verbatimStuff("\\end{" + name + "}").second;
+ // ignore one newline at beginning or end of string
+ if (prefixIs(s, "\n"))
+ s.erase(0,1);
+ if (suffixIs(s, "\n"))
+ s.erase(s.length() - 1,1);
+ return s;
+}
+
+
+string Parser::verbatimOption()
+{
+ string res;
+ if (next_token().character() == '[') {
+ Token t = get_token();
+ for (t = get_token(); t.character() != ']' && good(); t = get_token()) {
+ if (t.cat() == catBegin) {
+ putback();
+ res += '{' + verbatim_item() + '}';
+ } else
+ res += t.asInput();
+ }
+ }
+ return res;
+}
+
+
+string Parser::verbatim_item()
+{
+ if (!good())
+ error("stream bad");
+ skip_spaces();
+ if (next_token().cat() == catBegin) {
+ Token t = get_token(); // skip brace
+ string res;
+ for (t = get_token(); t.cat() != catEnd && good(); t = get_token()) {
+ if (t.cat() == catBegin) {
+ putback();
+ res += '{' + verbatim_item() + '}';
+ }
+ else
+ res += t.asInput();
+ }
+ return res;
+ }
+ return get_token().asInput();
+}
+
+
void Parser::tokenize_one()
{
catInit();
char_type c;
- if (!is_.get(c))
+ if (!is_.get(c))
return;
switch (catcode(c)) {
push_back(Token(s, catSpace));
break;
}
-
+
case catNewline: {
++lineno_;
docstring s(1, getNewline(is_, c));
push_back(Token(s, catNewline));
break;
}
-
+
case catComment: {
// We don't treat "%\n" combinations here specially because
// we want to preserve them in the preamble
push_back(Token(s, catComment));
break;
}
-
+
case catEscape: {
is_.get(c);
if (!is_) {
}
break;
}
-
+
case catIgnore: {
- cerr << "ignoring a char: " << c << "\n";
+ cerr << "ignoring a char: " << static_cast<uint32_t>(c) << "\n";
break;
}
-
+
default:
push_back(Token(docstring(1, c), catcode(c)));
}
}
-void Parser::error(string const & msg)
+void Parser::error(string const & msg) const
{
cerr << "Line ~" << lineno_ << ": parse error: " << msg << endl;
dump();
}
-string Parser::verbatimOption()
-{
- string res;
- if (next_token().character() == '[') {
- Token t = get_token();
- for (t = get_token(); t.character() != ']' && good(); t = get_token()) {
- if (t.cat() == catBegin) {
- putback();
- res += '{' + verbatim_item() + '}';
- } else
- res += t.cs();
- }
- }
- return res;
-}
-
-
-string Parser::verbatim_item()
-{
- if (!good())
- error("stream bad");
- skip_spaces();
- if (next_token().cat() == catBegin) {
- Token t = get_token(); // skip brace
- string res;
- for (Token t = get_token(); t.cat() != catEnd && good(); t = get_token()) {
- if (t.cat() == catBegin) {
- putback();
- res += '{' + verbatim_item() + '}';
- }
- else
- res += t.asInput();
- }
- return res;
- }
- return get_token().asInput();
-}
-
-
void Parser::reset()
{
pos_ = 0;
}
-void Parser::setCatCode(char c, CatCode cat)
-{
- theCatcode[(unsigned char)c] = cat;
-}
-
-
-CatCode Parser::getCatCode(char c) const
-{
- return theCatcode[(unsigned char)c];
-}
-
-
} // namespace lyx