-#include <config.h>
+/**
+ * \file lyxlex_pimpl.C
+ * This file is part of LyX, the document processor.
+ * Licence details can be found in the file COPYING.
+ *
+ * \author Lars Gullik Bjønnes
+ * \author Jean-Marc Lasgouttes
+ * \author Jürgen Vigna
+ *
+ * Full author contact details are available in file CREDITS.
+ */
-#ifdef __GNUG__
-#pragma implementation
-#endif
-#include <algorithm>
+#include <config.h>
#include "lyxlex_pimpl.h"
-#include "support/lyxalgo.h"
+
+#include "debug.h"
+
#include "support/filetools.h"
+#include "support/lyxalgo.h"
#include "support/lstrings.h"
-#include "debug.h"
+using lyx::support::compare_ascii_no_case;
+using lyx::support::getFormatFromContents;
+using lyx::support::MakeDisplayPath;
+using lyx::support::split;
+using lyx::support::subst;
+
+using std::endl;
+using std::getline;
+using std::lower_bound;
using std::sort;
-using std::ostream;
+using std::string;
using std::ios;
using std::istream;
-using std::endl;
-using std::lower_bound;
+using std::ostream;
+
+namespace {
-// namespace {
-struct compare_tags {
+struct compare_tags
+ : public std::binary_function<keyword_item, keyword_item, bool> {
// used by lower_bound, sort and sorted
- inline
- int operator()(keyword_item const & a, keyword_item const & b) const {
+ bool operator()(keyword_item const & a, keyword_item 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;
}
};
-// } // end of anon namespace
+
+} // end of anon namespace
-LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
- : is(&fb__), table(tab), no_items(num),
+LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
+ : is(&fb_), table(tab), no_items(num),
status(0), lineno(0), commentChar('#')
{
verifyTable();
string const LyXLex::Pimpl::getString() const
{
- return string(buff);
+ return string(buff.begin(), buff.end());
}
<< " of file " << MakeDisplayPath(name) << ']' << endl;
}
-
+
void LyXLex::Pimpl::printTable(ostream & os)
{
- os << "\nNumber of tags: " << no_items << '\n';
+ os << "\nNumber of tags: " << no_items << endl;
for (int i= 0; i < no_items; ++i)
os << "table[" << i
<< "]: tag: `" << table[i].tag
lyxerr << "The table passed to LyXLex is not sorted!\n"
<< "Tell the developers to fix it!" << endl;
// We sort it anyway to avoid problems.
- lyxerr << "\nUnsorted:\n";
+ lyxerr << "\nUnsorted:" << endl;
printTable(lyxerr);
sort(table, table + no_items, compare_tags());
- lyxerr << "\nSorted:\n";
+ lyxerr << "\nSorted:" << endl;
printTable(lyxerr);
}
}
verifyTable();
}
-
+
void LyXLex::Pimpl::popTable()
{
if (pushed.empty()) {
lyxerr << "LyXLex error: nothing to pop!" << endl;
return;
}
-
+
pushed_table tmp = pushed.top();
pushed.pop();
table = tmp.table_elem;
bool LyXLex::Pimpl::setFile(string const & filename)
{
- // 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() || is.tellg() > 0)
- lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
- "file or stream already set." << endl;
- fb__.open(filename.c_str(), ios::in);
- is.rdbuf(&fb__);
- name = filename;
- lineno = 0;
- return fb__.is_open() && is.good();
+ // Check the format of the file.
+ string const format = getFormatFromContents(filename);
+
+ if (format == "gzip" || format == "zip" || format == "compress") {
+ lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << endl;
+
+ // 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;
+ gz_.open(filename.c_str(), ios::in);
+ is.rdbuf(&gz_);
+ name = filename;
+ lineno = 0;
+ return gz_.is_open() && is.good();
+ } else {
+ lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
+
+ // 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 LyXLex::setFile: "
+ "file or stream already set." << endl;
+ fb_.open(filename.c_str(), ios::in);
+ is.rdbuf(&fb_);
+ name = filename;
+ lineno = 0;
+ return fb_.is_open() && is.good();
+ }
}
-
+
void LyXLex::Pimpl::setStream(istream & i)
{
- if (fb__.is_open() || is.tellg() > 0)
+ if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
lyxerr[Debug::LYXLEX] << "Error in LyXLex::setStream: "
"file or stream already set." << endl;
is.rdbuf(i.rdbuf());
lineno = 0;
}
+
void LyXLex::Pimpl::setCommentChar(char c)
{
commentChar = c;
// There can have been a whole line pushed so
// we extract the first word and leaves the rest
// in pushTok. (Lgb)
- if (pushTok.find(' ') != string::npos) {
+ if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
string tmp;
pushTok = split(pushTok, tmp, ' ');
- tmp.copy(buff, string::npos);
- buff[tmp.length()] = '\0';
+ buff.assign(tmp.begin(), tmp.end());
return true;
} else {
- pushTok.copy(buff, string::npos);
- buff[pushTok.length()] = '\0';
+ buff.assign(pushTok.begin(), pushTok.end());
pushTok.erase();
return true;
- }
+ }
}
if (!esc) {
unsigned char c = 0; // getc() returns an int
c = cc;
if (c == commentChar) {
// Read rest of line (fast :-)
- // That is not fast... (Lgb)
#if 1
- is.getline(buff, sizeof(buff));
+ // That is not fast... (Lgb)
+ string dummy;
+ getline(is, dummy);
+
lyxerr[Debug::LYXLEX] << "Comment read: `" << c
- << buff << "'" << endl;
+ << dummy << '\'' << endl;
#else
// unfortunately ignore is buggy (Lgb)
is.ignore(100, '\n');
++lineno;
continue;
}
-
+
if (c == '\"') {
- int i = -1;
+ buff.clear();
+
do {
is.get(cc);
c = cc;
if (c != '\r')
- buff[++i] = c;
- } while (c != '\"' && c != '\n' && is &&
- i != (LEX_MAX_BUFF - 2));
-
- if (i == (LEX_MAX_BUFF - 2)) {
- printError("Line too long");
- c = '\"'; // Pretend we got a "
- ++i;
- }
-
+ buff.push_back(c);
+ } while (c != '\"' && c != '\n' && is);
+
if (c != '\"') {
printError("Missing quote");
if (c == '\n')
++lineno;
}
-
- buff[i] = '\0';
+
+ buff.pop_back();
status = LEX_DATA;
- break;
+ 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) {
- int i = 0;
+ buff.clear();
+
do {
- buff[i++] = c;
+ buff.push_back(c);
is.get(cc);
c = cc;
- } while (c > ' ' && c != ',' && is
- && (i != LEX_MAX_BUFF - 1) );
- if (i == LEX_MAX_BUFF - 1) {
- printError("Line too long");
- }
- buff[i] = '\0';
+ } 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
is.get(cc);
c = cc;
}
-
+
if (c == '\n')
++lineno;
-
+
}
if (status) return true;
-
+
status = is.eof() ? LEX_FEOF: LEX_UNDEF;
- buff[0] = '\0';
+ 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;
-
+
if (c == '\\') {
// escape
- int i = 0;
+ buff.clear();
+
do {
if (c == '\\') {
// escape the next char
is.get(cc);
c = cc;
}
- buff[i++] = c;
+ buff.push_back(c);
is.get(cc);
c = cc;
- } while (c > ' ' && c != ',' && is
- && (i != LEX_MAX_BUFF - 1) );
- if (i == LEX_MAX_BUFF - 1) {
- printError("Line too long");
- }
- buff[i] = '\0';
+ } while (c > ' ' && c != ',' && is);
+
status = LEX_TOKEN;
continue;
}
-
+
if (c == commentChar) {
// Read rest of line (fast :-)
- // That is still not fast... (Lgb)
#if 1
- is.getline(buff, sizeof(buff));
+ // That is still not fast... (Lgb)
+ string dummy;
+ getline(is, dummy);
+
lyxerr[Debug::LYXLEX] << "Comment read: `" << c
- << buff << "'" << endl;
+ << dummy << '\'' << endl;
#else
// but ignore is also still buggy (Lgb)
// This is fast (Lgb)
++lineno;
continue;
}
-
+
// string
if (c == '\"') {
- int i = -1;
+ buff.clear();
+
bool escaped = false;
do {
escaped = false;
if (c == '\"' || c == '\\')
escaped = true;
else
- buff[++i] = '\\';
+ buff.push_back('\\');
}
- buff[++i] = c;
-
+ buff.push_back(c);
+
if (!escaped && c == '\"') break;
- } while (c != '\n' && is &&
- i != (LEX_MAX_BUFF - 2));
-
- if (i == (LEX_MAX_BUFF - 2)) {
- printError("Line too long");
- c = '\"'; // Pretend we got a "
- ++i;
- }
-
+ } while (c != '\n' && is);
+
if (c != '\"') {
printError("Missing quote");
if (c == '\n')
++lineno;
}
-
- buff[i] = '\0';
+
+ buff.pop_back();
status = LEX_DATA;
- break;
+ break;
}
-
+
if (c > ' ' && is) {
- int i = 0;
+ buff.clear();
+
do {
if (c == '\\') {
// escape the next char
c = cc;
//escaped = true;
}
- buff[i++] = c;
+ buff.push_back(c);
is.get(cc);
c = cc;
- } while (c > ' ' && c != ',' && is
- && (i != LEX_MAX_BUFF-1) );
- if (i == LEX_MAX_BUFF-1) {
- printError("Line too long");
- }
- buff[i] = '\0';
+ } while (c > ' ' && c != ',' && is);
+
status = LEX_TOKEN;
}
// new line
if (c == '\n')
++lineno;
}
-
+
if (status) return true;
-
+
status = is.eof() ? LEX_FEOF : LEX_UNDEF;
- buff[0] = '\0';
+ buff.clear();
return false;
}
}
keyword_item * res =
lower_bound(table, table + no_items,
search_tag, compare_tags());
+ // 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.
if (res != table + no_items
- && !compare_no_case(res->tag, tag))
+ && !compare_ascii_no_case(res->tag, tag))
return res->code;
return LEX_UNDEF;
}
int LyXLex::Pimpl::lex()
{
//NOTE: possible bug.
- if (next() && status == LEX_TOKEN)
- return search_kw(buff);
- else
+ if (next() && status == LEX_TOKEN) {
+ return search_kw(getString().c_str());
+ } else
return status;
}
-
+
bool LyXLex::Pimpl::eatLine()
{
- int i = 0;
+ buff.clear();
+
unsigned char c = '\0';
char cc = 0;
- while(is && c != '\n' && i != (LEX_MAX_BUFF - 1)) {
+ while (is && c != '\n') {
is.get(cc);
c = cc;
//lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
- // << c << "'" << endl;
+ // << c << '\'' << endl;
if (c != '\r')
- buff[i++] = c;
- }
- if (i == (LEX_MAX_BUFF - 1) && c != '\n') {
- printError("Line too long");
- c = '\n'; // Pretend we had an end of line
- --lineno; // but don't increase line counter (netto effect)
- ++i; // and preserve last character read.
+ buff.push_back(c);
}
+
if (c == '\n') {
++lineno;
- buff[--i] = '\0'; // i can never be 0 here, so no danger
+ buff.pop_back();
status = LEX_DATA;
return true;
} else {
- buff[i] = '\0';
return false;
}
}
// There can have been a whole line pushed so
// we extract the first word and leaves the rest
// in pushTok. (Lgb)
- if (pushTok.find(' ') != string::npos) {
+ if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
string tmp;
pushTok = split(pushTok, tmp, ' ');
- tmp.copy(buff, string::npos);
- buff[tmp.length()] = '\0';
+ buff.assign(tmp.begin(), tmp.end());
return true;
} else {
- pushTok.copy(buff, string::npos);
- buff[pushTok.length()] = '\0';
+ buff.assign(pushTok.begin(), pushTok.end());
pushTok.erase();
return true;
}
is.get(cc);
c = cc;
if (c >= ' ' && is) {
- int i = 0;
+ buff.clear();
+
if (c == '\\') { // first char == '\\'
do {
- buff[i++] = c;
+ buff.push_back(c);
is.get(cc);
c = cc;
- } while (c > ' ' && c != '\\' && is
- && i != (LEX_MAX_BUFF-1));
+ } while (c > ' ' && c != '\\' && is);
} else {
do {
- buff[i++] = c;
+ buff.push_back(c);
is.get(cc);
c = cc;
- } while (c >= ' ' && c != '\\' && is
- && i != (LEX_MAX_BUFF-1));
- }
-
- if (i == (LEX_MAX_BUFF - 1)) {
- printError("Line too long");
+ } while (c >= ' ' && c != '\\' && is);
}
-
+
if (c == '\\') is.putback(c); // put it back
- buff[i] = '\0';
status = LEX_TOKEN;
}
-
+
if (c == '\n')
++lineno;
-
+
}
if (status) return true;
-
+
status = is.eof() ? LEX_FEOF: LEX_UNDEF;
- buff[0] = '\0';
+ buff.clear();
return false;
}