using std::fill;
using std::istream;
using std::istringstream;
+using std::ostringstream;
using std::ostream;
using std::string;
theCatcode[int('}')] = catEnd;
theCatcode[int('$')] = catMath;
theCatcode[int('&')] = catAlign;
- theCatcode[10] = catNewline;
+ theCatcode[int('\n')] = catNewline;
theCatcode[int('#')] = catParameter;
theCatcode[int('^')] = catSuper;
theCatcode[int('_')] = catSub;
- theCatcode[0x7f] = catIgnore;
+ theCatcode[0x7f] = catIgnore;
theCatcode[int(' ')] = catSpace;
theCatcode[int('\t')] = catSpace;
- theCatcode[13] = catIgnore;
+ theCatcode[int('\r')] = catNewline;
theCatcode[int('~')] = catActive;
theCatcode[int('%')] = catComment;
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 getNewline(istream & is, char c)
+{
+ // we have to handle 3 different line endings:
+ // - UNIX (\n)
+ // - MAC (\r)
+ // - DOS (\r\n)
+ if (c == '\r') {
+ // MAC or DOS
+ if (is.get(c) && c != '\n') {
+ // MAC
+ is.putback(c);
+ }
+ return '\n';
+ }
+ // UNIX
+ return c;
+}
+
}
// catcodes
//
-mode_type asMode(mode_type oldmode, string const & str)
-{
- if (str == "mathmode")
- return MATH_MODE;
- if (str == "textmode" || str == "forcetext")
- return TEXT_MODE;
- return oldmode;
-}
-
-
CatCode catcode(unsigned char c)
{
return theCatcode[c];
}
+bool Parser::isParagraph() const
+{
+ // A new paragraph in TeX ist started
+ // - either by a newline, following any amount of whitespace
+ // characters (including zero), and another newline
+ // - or the token \par
+ if (curr_token().cat() == catNewline &&
+ (curr_token().cs().size() > 1 ||
+ (next_token().cat() == catSpace &&
+ pos_ < tokens_.size() - 1 &&
+ tokens_[pos_ + 1].cat() == catNewline)))
+ return true;
+ if (curr_token().cat() == catEscape && curr_token().cs() == "par")
+ return true;
+ return false;
+}
+
+
void Parser::skip_spaces(bool skip_comments)
{
// We just silently return if we have no more tokens.
// skip_spaces() should be callable at any time,
// the caller must check p::good() anyway.
while (good()) {
- if ( next_token().cat() == catSpace ||
- (next_token().cat() == catNewline && next_token().cs().size() == 1) ||
- next_token().cat() == catComment && next_token().cs().empty())
- get_token();
- else if (skip_comments && next_token().cat() == catComment)
- cerr << " Ignoring comment: " << get_token().asInput();
- else
+ get_token();
+ if (isParagraph()) {
+ putback();
break;
+ }
+ if ( curr_token().cat() == catSpace ||
+ curr_token().cat() == catNewline ||
+ (curr_token().cat() == catComment && curr_token().cs().empty()))
+ continue;
+ if (skip_comments && curr_token().cat() == catComment)
+ cerr << " Ignoring comment: " << curr_token().asInput();
+ else {
+ putback();
+ break;
+ }
}
}
}
-string Parser::getArg(char left, char right)
+Parser::Arg Parser::getFullArg(char left, char right)
{
skip_spaces(true);
// This is needed if a partial file ends with a command without arguments,
// e. g. \medskip
if (! good())
- return string();
+ return std::make_pair(false, string());
string result;
char c = getChar();
- if (c != left)
+ if (c != left) {
putback();
- else
+ return std::make_pair(false, string());
+ } else
while ((c = getChar()) != right && good()) {
// Ignore comments
if (curr_token().cat() == catComment) {
if (!curr_token().cs().empty())
cerr << "Ignoring comment: " << curr_token().asInput();
}
- else if (curr_token().cat() == catSpace || curr_token().cat() == catNewline)
- result += curr_token().cs();
else
- result += c;
+ result += curr_token().asInput();
}
- return result;
+ return std::make_pair(true, result);
+}
+
+
+string Parser::getArg(char left, char right)
+{
+ return getFullArg(left, right).second;
+}
+
+
+string Parser::getFullOpt()
+{
+ Arg arg = getFullArg('[', ']');
+ if (arg.first)
+ return '[' + arg.second + ']';
+ return arg.second;
}
string Parser::getOpt()
{
string const res = getArg('[', ']');
- return res.size() ? '[' + res + ']' : string();
+ return res.empty() ? string() : '[' + res + ']';
+}
+
+
+string const Parser::verbatimEnvironment(string const & name)
+{
+ if (!good())
+ return string();
+
+ ostringstream os;
+ for (Token t = get_token(); good(); t = get_token()) {
+ if (t.cat() == catBegin) {
+ putback();
+ os << '{' << verbatim_item() << '}';
+ } else if (t.asInput() == "\\begin") {
+ string const env = getArg('{', '}');
+ os << "\\begin{" << env << '}'
+ << verbatimEnvironment(env)
+ << "\\end{" << env << '}';
+ } else if (t.asInput() == "\\end") {
+ string const end = getArg('{', '}');
+ if (end != name)
+ cerr << "\\end{" << end
+ << "} does not match \\begin{" << name
+ << "}." << endl;
+ return os.str();
+ } else
+ os << t.asInput();
+ }
+ cerr << "unexpected end of input" << endl;
+ return os.str();
}
case catNewline: {
++lineno_;
- string s(1, c);
+ string s(1, getNewline(is, c));
while (is.get(c) && catcode(c) == catNewline) {
++lineno_;
- s += c;
+ s += getNewline(is, c);
}
if (catcode(c) != catNewline)
is.putback(c);
string s;
while (is.get(c) && catcode(c) != catNewline)
s += c;
+ // handle possible DOS line ending
+ if (catcode(c) == catNewline)
+ c = getNewline(is, c);
// Note: The '%' at the beginning and the '\n' at the end
// of the comment are not stored.
++lineno_;
}
case catIgnore: {
- if (c != 13)
- cerr << "ignoring a char: " << int(c) << "\n";
+ cerr << "ignoring a char: " << int(c) << "\n";
break;
}
projects / lyx.git / blobdiff