* Licence details can be found in the file COPYING.
*
* \author Alejandro Aguilar Sierra
- * \author Lars Gullik Bjønnes
+ * \author Lars Gullik Bjønnes
* \author Jean-Marc Lasgouttes
* \author John Levon
*
#include <config.h>
#include "Lexer.h"
+#include "Format.h"
#include "support/convert.h"
#include "support/debug.h"
#include "support/FileName.h"
#include "support/filetools.h"
#include "support/gzstream.h"
+#include "support/lassert.h"
#include "support/lstrings.h"
#include "support/lyxalgo.h"
-#include "support/types.h"
+#include <algorithm> // sort, lower_bound
#include <functional>
+#include <fstream>
#include <istream>
#include <stack>
#include <vector>
int lineno;
///
string pushTok;
- ///
- char commentChar;
/// used for error messages
string context;
+ ///
+ char commentChar;
private:
/// non-copyable
Pimpl(Pimpl const &);
public:
///
PushedTable()
- : table_elem(0), table_siz(0) {}
+ : table_elem(nullptr), table_siz(0) {}
///
PushedTable(LexerKeyword * ki, int siz)
: table_elem(ki), table_siz(siz) {}
};
-
namespace {
-class CompareTags
- : public binary_function<LexerKeyword, LexerKeyword, bool> {
-public:
- // used by lower_bound, sort and sorted
- bool operator()(LexerKeyword const & a, LexerKeyword const & b) const
- {
- // we use the ascii version, because in turkish, 'i'
- // is not the lowercase version of 'I', and thus
- // turkish locale breaks parsing of tags.
- return compare_ascii_no_case(a.tag, b.tag) < 0;
- }
-};
+// used by lower_bound, sort and sorted
+bool compareTags(LexerKeyword const & a, LexerKeyword const & b)
+{
+ // we use the ascii version, because in turkish, 'i'
+ // is not the lowercase version of 'I', and thus
+ // turkish locale breaks parsing of tags.
+ return compare_ascii_no_case(a.tag, b.tag) < 0;
+}
+
+} // namespace
-} // end of anon namespace
Lexer::Pimpl::Pimpl(LexerKeyword * tab, int num)
{
string const tmpmsg = subst(message, "$$Token", getString());
lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
- << " of file " << to_utf8(makeDisplayPath(name)) << ']' << endl;
+ << " of file " << to_utf8(makeDisplayPath(name))
+ << " current token: '" << getString() << "'"
+ << " context: '" << context << "']" << endl;
}
{
// Check if the table is sorted and if not, sort it.
if (table
- && !lyx::sorted(table, table + no_items, CompareTags())) {
+ && !lyx::sorted(table, table + no_items, &compareTags)) {
lyxerr << "The table passed to Lexer is not sorted!\n"
<< "Tell the developers to fix it!" << endl;
// We sort it anyway to avoid problems.
lyxerr << "\nUnsorted:" << endl;
printTable(lyxerr);
- sort(table, table + no_items, CompareTags());
+ sort(table, table + no_items, &compareTags);
lyxerr << "\nSorted:" << endl;
printTable(lyxerr);
}
bool Lexer::Pimpl::setFile(FileName const & filename)
{
// Check the format of the file.
- string const format = filename.guessFormatFromContents();
-
- if (format == "gzip" || format == "zip" || format == "compress") {
+ if (theFormats().isZippedFile(filename)) {
LYXERR(Debug::LYXLEX, "lyxlex: compressed");
// The check only outputs a debug message, because it triggers
// a bug in compaq cxx 6.2, where is_open() returns 'true' for
"file or stream already set.");
gz_.open(filename.toFilesystemEncoding().c_str(), ios::in);
is.rdbuf(&gz_);
- name = filename.absFilename();
+ name = filename.absFileName();
lineno = 0;
- return gz_.is_open() && is.good();
+ if (!gz_.is_open() || !is.good())
+ return false;
} else {
LYXERR(Debug::LYXLEX, "lyxlex: UNcompressed");
LYXERR(Debug::LYXLEX, "Error in Lexer::setFile: "
"file or stream already set.");
}
- fb_.open(filename.toFilesystemEncoding().c_str(), ios::in);
+ fb_.open(filename.toSafeFilesystemEncoding().c_str(), ios::in);
is.rdbuf(&fb_);
- name = filename.absFilename();
+ name = filename.absFileName();
lineno = 0;
- return fb_.is_open() && is.good();
+ if (!fb_.is_open() || !is.good())
+ return false;
+ }
+
+ // Skip byte order mark.
+ if (is.peek() == 0xef) {
+ is.get();
+ if (is.peek() == 0xbb) {
+ is.get();
+ LASSERT(is.get() == 0xbf, /**/);
+ } else
+ is.unget();
}
+
+ return true;
}
}
- unsigned char c = 0; // getc() returns an int
char cc = 0;
status = 0;
while (is && !status) {
is.get(cc);
- c = cc;
-
- // skip ','s
- if (esc && c == ',')
- continue;
+ unsigned char c = cc;
-
if (c == commentChar) {
// Read rest of line (fast :-)
#if 1
if (esc) {
- bool escaped = false;
do {
- escaped = false;
+ bool escaped = false;
is.get(cc);
c = cc;
if (c == '\r') continue;
break;
}
- if (!esc && c == ',')
+ if (c == ',')
continue; /* Skip ','s */
- // using relational operators with chars other
- // than == and != is not safe. And if it is done
- // the type _have_ to be unsigned. It usually a
- // lot better to use the functions from cctype
+ // using relational operators with chars other
+ // than == and != is not safe. And if it is done
+ // the type _have_ to be unsigned. It usually a
+ // lot better to use the functions from cctype
if (c > ' ' && is) {
buff.clear();
status = LEX_TOKEN;
}
- if (!esc && c == '\r' && is) {
+ if (c == '\r' && is) {
// The Windows support has lead to the
// possibility of "\r\n" at the end of
// a line. This will stop LyX choking
LexerKeyword search_tag = { tag, 0 };
LexerKeyword * res =
lower_bound(table, table + no_items,
- search_tag, CompareTags());
+ search_tag, &compareTags);
// use the compare_ascii_no_case instead of compare_no_case,
// because in turkish, 'i' is not the lowercase version of 'I',
// and thus turkish locale breaks parsing of tags.
is.get(cc);
c = cc;
//LYXERR(Debug::LYXLEX, "Lexer::EatLine read char: `" << c << '\'');
- if (c != '\r')
+ if (c != '\r' && is)
buff.push_back(c);
}
char cc = 0;
is.get(cc);
c = cc;
- if (c >= ' ' && is) {
+ if ((c >= ' ' || c == '\t') && is) {
buff.clear();
if (c == '\\') { // first char == '\\'
buff.push_back(c);
is.get(cc);
c = cc;
- } while (c >= ' ' && c != '\\' && is);
+ } while ((c >= ' ' || c == '\t') && c != '\\' && is);
}
if (c == '\\')
//////////////////////////////////////////////////////////////////////
Lexer::Lexer()
- : pimpl_(new Pimpl(0, 0))
+ : pimpl_(new Pimpl(nullptr, 0)), lastReadOk_(false)
{}
pimpl_->setCommentChar(c);
}
+
int Lexer::lex()
{
return pimpl_->lex();
}
-string const Lexer::getString() const
+string const Lexer::getString(bool trim) const
{
lastReadOk_ = pimpl_->status == LEX_DATA || pimpl_->status == LEX_TOKEN;
if (lastReadOk_)
- return pimpl_->getString();
+ return trim ? support::trim(pimpl_->getString(), "\t ") : pimpl_->getString();
return string();
}
-docstring const Lexer::getDocString() const
+docstring const Lexer::getDocString(bool trim) const
{
lastReadOk_ = pimpl_->status == LEX_DATA || pimpl_->status == LEX_TOKEN;
if (lastReadOk_)
- return pimpl_->getDocString();
+ return trim ? support::trim(pimpl_->getDocString(), "\t ") : pimpl_->getDocString();
return docstring();
}
// I would prefer to give a tag number instead of an explicit token
// here, but it is not possible because Buffer::readDocument uses
// explicit tokens (JMarc)
-string const Lexer::getLongString(string const & endtoken)
+docstring Lexer::getLongString(docstring const & endtoken)
{
- string str;
- string prefix;
+ docstring str;
+ docstring prefix;
bool firstline = true;
while (pimpl_->is) { //< eatLine only reads from is, not from pushTok
if (!eatLine())
// blank line in the file being read
continue;
+ docstring tmpstr = getDocString();
+ docstring const token = trim(tmpstr, " \t");
- string const token = trim(getString(), " \t");
-
- LYXERR(Debug::PARSER, "LongString: `" << getString() << '\'');
+ LYXERR(Debug::PARSER, "LongString: `" << tmpstr << '\'');
// We do a case independent comparison, like searchKeyword does.
- if (compare_ascii_no_case(token, endtoken) == 0)
+ if (compare_no_case(token, endtoken) == 0)
break;
- string tmpstr = getString();
if (firstline) {
- size_t i = tmpstr.find_first_not_of(' ');
+ size_t i = tmpstr.find_first_not_of(from_ascii(" \t"));
if (i != string::npos)
prefix = tmpstr.substr(0, i);
firstline = false;
// further lines in long strings may have the same
// whitespace prefix as the first line. Remove it.
- if (prefix.length() && prefixIs(tmpstr, prefix))
- tmpstr.erase(0, prefix.length() - 1);
+ if (!prefix.empty() && prefixIs(tmpstr, prefix))
+ tmpstr.erase(0, prefix.length());
- str += ltrim(tmpstr, "\t") + '\n';
+ str += tmpstr + '\n';
}
if (!pimpl_->is)
- printError("Long string not ended by `" + endtoken + '\'');
+ printError("Long string not ended by `" + to_utf8(endtoken) + '\'');
return str;
}
bool Lexer::getBool() const
{
- string const s = pimpl_->getString();
+ string const s = pimpl_->getString();
if (s == "false" || s == "0") {
lastReadOk_ = true;
return false;
pimpl_->printError("Bad boolean `$$Token'. "
"Use \"false\" or \"true\"");
lastReadOk_ = false;
+ return false;
}
// use fail() here. However, our implementation of getString() et al.
// can cause the eof() and fail() bits to be set, even though we
// haven't tried to read 'em.
- return lastReadOk_? this : 0;
+ return lastReadOk_? this : nullptr;
}
}
+// same for docstring
+docstring Lexer::quoteString(docstring const & arg)
+{
+ docstring res;
+ res += '"';
+ res += subst(subst(arg, from_ascii("\\"), from_ascii("\\\\")),
+ from_ascii("\""), from_ascii("\\\""));
+ res += '"';
+ return res;
+}
+
+
Lexer & Lexer::operator>>(char const * required)
{
string token;
*this >> token;
if (token != required) {
- LYXERR0("Missing '" << required << "'-tag in " << pimpl_->context
+ LYXERR0("Missing '" << required << "'-tag in " << pimpl_->context
<< ". Got " << token << " instead. Line: " << lineNumber());
pushToken(token);
}
}
-void Lexer::setContext(std::string const & str)
+void Lexer::setContext(std::string const & functionName)
{
- pimpl_->context = str;
+ pimpl_->context = functionName;
}