* 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 "debug.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 "support/unicode.h"
-
-#include <boost/noncopyable.hpp>
#include <functional>
#include <istream>
#include <stack>
-#include <sstream>
#include <vector>
+using namespace std;
+using namespace lyx::support;
namespace lyx {
-using support::compare_ascii_no_case;
-using support::FileName;
-using support::isStrDbl;
-using support::isStrInt;
-using support::ltrim;
-using support::makeDisplayPath;
-using support::prefixIs;
-using support::split;
-using support::subst;
-using support::trim;
-
-using std::endl;
-using std::getline;
-using std::lower_bound;
-using std::sort;
-using std::string;
-using std::ios;
-using std::istream;
-using std::ostream;
-
-
//////////////////////////////////////////////////////////////////////
//
// Lexer::Pimpl
///
-class Lexer::Pimpl : boost::noncopyable {
+class Lexer::Pimpl {
public:
///
- Pimpl(keyword_item * tab, int num);
+ Pimpl(LexerKeyword * tab, int num);
///
- std::string const getString() const;
+ string const getString() const;
///
docstring const getDocString() const;
///
- void printError(std::string const & message) const;
+ void printError(string const & message) const;
///
- void printTable(std::ostream & os);
+ void printTable(ostream & os);
///
- void pushTable(keyword_item * tab, int num);
+ void pushTable(LexerKeyword * tab, int num);
///
void popTable();
///
- bool setFile(support::FileName const & filename);
+ bool setFile(FileName const & filename);
///
- void setStream(std::istream & i);
+ void setStream(istream & i);
///
void setCommentChar(char c);
///
bool next(bool esc = false);
///
- int search_kw(char const * const tag) const;
+ int searchKeyword(char const * const tag) const;
///
int lex();
///
/// test if there is a pushed token or the stream is ok
bool inputAvailable();
///
- void pushToken(std::string const &);
+ void pushToken(string const &);
/// fb_ is only used to open files, the stream is accessed through is.
- std::filebuf fb_;
+ filebuf fb_;
/// gz_ is only used to open files, the stream is accessed through is.
gz::gzstreambuf gz_;
/// the stream that we use.
- std::istream is;
+ istream is;
///
- std::string name;
+ string name;
///
- keyword_item * table;
+ LexerKeyword * table;
///
int no_items;
///
- std::string buff;
+ string buff;
///
int status;
///
int lineno;
///
- std::string pushTok;
+ string pushTok;
///
char commentChar;
+ /// used for error messages
+ string context;
private:
+ /// non-copyable
+ Pimpl(Pimpl const &);
+ void operator=(Pimpl const &);
+
///
void verifyTable();
///
- class pushed_table {
+ class PushedTable {
public:
///
- pushed_table()
+ PushedTable()
: table_elem(0), table_siz(0) {}
///
- pushed_table(keyword_item * ki, int siz)
+ PushedTable(LexerKeyword * ki, int siz)
: table_elem(ki), table_siz(siz) {}
///
- keyword_item * table_elem;
+ LexerKeyword * table_elem;
///
int table_siz;
};
///
- std::stack<pushed_table> pushed;
+ stack<PushedTable> pushed;
};
namespace {
-class compare_tags
- : public std::binary_function<keyword_item, keyword_item, bool> {
+class CompareTags
+ : public binary_function<LexerKeyword, LexerKeyword, bool> {
public:
// used by lower_bound, sort and sorted
- bool operator()(keyword_item const & a, keyword_item const & b) const
+ 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
} // end of anon namespace
-Lexer::Pimpl::Pimpl(keyword_item * tab, int num)
+Lexer::Pimpl::Pimpl(LexerKeyword * tab, int num)
: is(&fb_), table(tab), no_items(num),
status(0), lineno(0), commentChar('#')
{
{
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, compare_tags())) {
+ && !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, compare_tags());
+ sort(table, table + no_items, CompareTags());
lyxerr << "\nSorted:" << endl;
printTable(lyxerr);
}
}
-void Lexer::Pimpl::pushTable(keyword_item * tab, int num)
+void Lexer::Pimpl::pushTable(LexerKeyword * tab, int num)
{
- pushed_table tmppu(table, no_items);
+ PushedTable tmppu(table, no_items);
pushed.push(tmppu);
table = tab;
return;
}
- pushed_table tmp = pushed.top();
+ PushedTable tmp = pushed.top();
pushed.pop();
table = tmp.table_elem;
no_items = tmp.table_siz;
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") {
- LYXERR(Debug::LYXLEX) << "lyxlex: compressed" << endl;
+ 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
// a fresh new filebuf. (JMarc)
if (gz_.is_open() || istream::off_type(is.tellg()) > -1)
- lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
- "file or stream already set." << endl;
+ LYXERR(Debug::LYXLEX, "Error in LyXLex::setFile: "
+ "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" << endl;
+ LYXERR(Debug::LYXLEX, "lyxlex: UNcompressed");
// The check only outputs a debug message, because it triggers
// a bug in compaq cxx 6.2, where is_open() returns 'true' for
// a fresh new filebuf. (JMarc)
if (fb_.is_open() || istream::off_type(is.tellg()) > 0) {
- LYXERR(Debug::LYXLEX) << "Error in Lexer::setFile: "
- "file or stream already set." << endl;
+ 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;
}
void Lexer::Pimpl::setStream(istream & i)
{
if (fb_.is_open() || istream::off_type(is.tellg()) > 0) {
- LYXERR(Debug::LYXLEX) << "Error in Lexer::setStream: "
- "file or stream already set." << endl;
+ LYXERR(Debug::LYXLEX, "Error in Lexer::setStream: "
+ "file or stream already set.");
}
is.rdbuf(i.rdbuf());
lineno = 0;
status = LEX_TOKEN;
return true;
}
- if (!esc) {
- unsigned char c = 0; // getc() returns an int
- char cc = 0;
- status = 0;
- while (is && !status) {
- is.get(cc);
- c = cc;
- if (c == commentChar) {
- // Read rest of line (fast :-)
-#if 1
- // That is not fast... (Lgb)
- string dummy;
- getline(is, dummy);
-
- LYXERR(Debug::LYXLEX) << "Comment read: `" << c
- << dummy << '\'' << endl;
-#else
- // unfortunately ignore is buggy (Lgb)
- is.ignore(100, '\n');
-#endif
- ++lineno;
- continue;
- }
-
- if (c == '\"') {
- buff.clear();
-
- do {
- is.get(cc);
- c = cc;
- if (c != '\r')
- buff.push_back(c);
- } while (c != '\"' && c != '\n' && is);
-
- if (c != '\"') {
- printError("Missing quote");
- if (c == '\n')
- ++lineno;
- }
-
- buff.resize(buff.size()-1);
- status = LEX_DATA;
- break;
- }
-
- 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
- if (c > ' ' && is) {
- buff.clear();
-
- do {
- buff.push_back(c);
- is.get(cc);
- c = cc;
- } while (c > ' ' && c != ',' && is);
-
- status = LEX_TOKEN;
- }
-
- 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
- // when it expected to find a '\n'
- is.get(cc);
- c = cc;
- }
-
- if (c == '\n')
- ++lineno;
-
- }
- if (status)
- return true;
- status = is.eof() ? LEX_FEOF: LEX_UNDEF;
- buff.clear();
- return false;
- } else {
- unsigned char c = 0; // getc() returns an int
- char cc = 0;
- status = 0;
- while (is && !status) {
- is.get(cc);
- c = cc;
-
- // skip ','s
- if (c == ',')
- continue;
+ char cc = 0;
+ status = 0;
+ while (is && !status) {
+ is.get(cc);
+ unsigned char c = cc;
- if (c == commentChar) {
- // Read rest of line (fast :-)
+ if (c == commentChar) {
+ // Read rest of line (fast :-)
#if 1
- // That is still not fast... (Lgb)
- string dummy;
- getline(is, dummy);
+ // That is not fast... (Lgb)
+ string dummy;
+ getline(is, dummy);
- LYXERR(Debug::LYXLEX) << "Comment read: `" << c
- << dummy << '\'' << endl;
+ LYXERR(Debug::LYXLEX, "Comment read: `" << c << dummy << '\'');
#else
- // but ignore is also still buggy (Lgb)
- // This is fast (Lgb)
- is.ignore(100, '\n');
+ // unfortunately ignore is buggy (Lgb)
+ is.ignore(100, '\n');
#endif
- ++lineno;
- continue;
- }
+ ++lineno;
+ continue;
+ }
+
+ if (c == '\"') {
+ buff.clear();
- // string
- if (c == '\"') {
- buff.clear();
+ if (esc) {
- bool escaped = false;
do {
- escaped = false;
+ bool escaped = false;
is.get(cc);
c = cc;
if (c == '\r') continue;
break;
} while (c != '\n' && is);
- if (c != '\"') {
- printError("Missing quote");
- if (c == '\n')
- ++lineno;
- }
+ } else {
+
+ do {
+ is.get(cc);
+ c = cc;
+ if (c != '\r')
+ buff.push_back(c);
+ } while (c != '\"' && c != '\n' && is);
- buff.resize(buff.size() -1);
- status = LEX_DATA;
- break;
}
- if (c > ' ' && is) {
- buff.clear();
+ if (c != '\"') {
+ printError("Missing quote");
+ if (c == '\n')
+ ++lineno;
+ }
- do {
- if (c == '\\') {
- // escape the next char
- is.get(cc);
- c = cc;
- //escaped = true;
- }
- buff.push_back(c);
+ buff.resize(buff.size() - 1);
+ status = LEX_DATA;
+ break;
+ }
+
+ 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
+ if (c > ' ' && is) {
+ buff.clear();
+
+ do {
+ if (esc && c == '\\') {
+ // escape the next char
is.get(cc);
c = cc;
- } while (c > ' ' && c != ',' && is);
+ //escaped = true;
+ }
+ buff.push_back(c);
+ is.get(cc);
+ c = cc;
+ } while (c > ' ' && c != ',' && is);
+ status = LEX_TOKEN;
+ }
- status = LEX_TOKEN;
- }
- // new line
- if (c == '\n')
- ++lineno;
+ 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
+ // when it expected to find a '\n'
+ is.get(cc);
+ c = cc;
}
- if (status)
- return true;
+ if (c == '\n')
+ ++lineno;
- status = is.eof() ? LEX_FEOF : LEX_UNDEF;
- buff.clear();
- return false;
}
+ if (status)
+ return true;
+
+ status = is.eof() ? LEX_FEOF: LEX_UNDEF;
+ buff.clear();
+ return false;
}
-int Lexer::Pimpl::search_kw(char const * const tag) const
+int Lexer::Pimpl::searchKeyword(char const * const tag) const
{
- keyword_item search_tag = { tag, 0 };
- keyword_item * res =
+ LexerKeyword search_tag = { tag, 0 };
+ LexerKeyword * res =
lower_bound(table, table + no_items,
- search_tag, compare_tags());
+ 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.
{
//NOTE: possible bug.
if (next() && status == LEX_TOKEN)
- return search_kw(getString().c_str());
+ return searchKeyword(getString().c_str());
return status;
}
while (is && c != '\n') {
is.get(cc);
c = cc;
- //LYXERR(Debug::LYXLEX) << "Lexer::EatLine read char: `"
- // << c << '\'' << endl;
- if (c != '\r')
+ //LYXERR(Debug::LYXLEX, "Lexer::EatLine read char: `" << c << '\'');
+ 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 == '\\') is.putback(c); // put it back
+ if (c == '\\')
+ is.putback(c); // put it back
status = LEX_TOKEN;
}
//
//////////////////////////////////////////////////////////////////////
-Lexer::Lexer(keyword_item * tab, int num)
- : pimpl_(new Pimpl(tab, num))
+Lexer::Lexer()
+ : pimpl_(new Pimpl(0, 0)), lastReadOk_(false)
{}
+void Lexer::init(LexerKeyword * tab, int num)
+{
+ pimpl_ = new Pimpl(tab, num);
+}
+
+
Lexer::~Lexer()
{
delete pimpl_;
}
-void Lexer::setLineNo(int l)
+void Lexer::setLineNumber(int l)
{
pimpl_->lineno = l;
}
-int Lexer::getLineNo() const
+int Lexer::lineNumber() const
{
return pimpl_->lineno;
}
}
-void Lexer::pushTable(keyword_item * tab, int num)
+void Lexer::pushTable(LexerKeyword * tab, int num)
{
pimpl_->pushTable(tab, num);
}
}
-bool Lexer::setFile(support::FileName const & filename)
+bool Lexer::setFile(FileName const & filename)
{
return pimpl_->setFile(filename);
}
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, 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() << '\'' << endl;
+ LYXERR(Debug::PARSER, "LongString: `" << tmpstr << '\'');
- // We do a case independent comparison, like search_kw does.
- if (compare_ascii_no_case(token, endtoken) == 0)
+ // We do a case independent comparison, like searchKeyword does.
+ if (compare_no_case(token, endtoken) == 0)
break;
- string tmpstr = getString();
if (firstline) {
- string::size_type i(tmpstr.find_first_not_of(' '));
+ size_t i = tmpstr.find_first_not_of(char_type(' '));
if (i != string::npos)
prefix = tmpstr.substr(0, i);
firstline = false;
- LYXERR(Debug::PARSER)
- << "Prefix = `" << prefix << "\'" << endl;
+ LYXERR(Debug::PARSER, "Prefix = `" << prefix << "\'");
}
// further lines in long strings may have the same
// whitespace prefix as the first line. Remove it.
- if (prefix.length() && prefixIs(tmpstr, prefix)) {
+ if (prefix.length() && prefixIs(tmpstr, prefix))
tmpstr.erase(0, prefix.length() - 1);
- }
str += ltrim(tmpstr, "\t") + '\n';
}
- if (!pimpl_->is) {
- printError("Long string not ended by `" + endtoken + '\'');
- }
+ if (!pimpl_->is)
+ printError("Long string not ended by `" + to_utf8(endtoken) + '\'');
return str;
}
bool Lexer::getBool() const
{
- if (pimpl_->getString() == "true") {
+ string const s = pimpl_->getString();
+ if (s == "false" || s == "0") {
+ lastReadOk_ = true;
+ return false;
+ }
+ if (s == "true" || s == "1") {
lastReadOk_ = true;
return true;
- } else if (pimpl_->getString() != "false") {
- pimpl_->printError("Bad boolean `$$Token'. "
- "Use \"false\" or \"true\"");
- lastReadOk_ = false;
}
- lastReadOk_ = true;
+ pimpl_->printError("Bad boolean `$$Token'. "
+ "Use \"false\" or \"true\"");
+ lastReadOk_ = false;
return false;
}
Lexer::operator void const *() const
{
- // This behaviour is NOT the same as the std::streams which would
+ // This behaviour is NOT the same as the streams which would
// 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.
}
-Lexer & Lexer::operator>>(std::string & s)
+Lexer & Lexer::operator>>(string & s)
{
if (isOK()) {
next();
}
-/// quotes a string, e.g. for use in preferences files or as an argument of the "log" dialog
-string const Lexer::quoteString(string const & arg)
+Lexer & Lexer::operator>>(char & c)
+{
+ string s;
+ operator>>(s);
+ if (!s.empty())
+ c = s[0];
+ return *this;
+}
+
+
+// quotes a string, e.g. for use in preferences files or as an argument
+// of the "log" dialog
+string Lexer::quoteString(string const & arg)
+{
+ string res;
+ res += '"';
+ res += subst(subst(arg, "\\", "\\\\"), "\"", "\\\"");
+ res += '"';
+ return res;
+}
+
+
+// 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
+ << ". Got " << token << " instead. Line: " << lineNumber());
+ pushToken(token);
+ }
+ return *this;
+}
+
+
+bool Lexer::checkFor(char const * required)
+{
+ string token;
+ *this >> token;
+ if (token == required)
+ return true;
+ pushToken(token);
+ return false;
+}
+
+
+void Lexer::setContext(std::string const & str)
{
- std::ostringstream os;
- os << '"' << subst(subst(arg, "\\", "\\\\"), "\"", "\\\"") << '"';
- return os.str();
+ pimpl_->context = str;
}