#include <map>
#include <regex>
-//#define ResultsDebug
+#define ResultsDebug
#define USE_QT_FOR_SEARCH
#if defined(USE_QT_FOR_SEARCH)
#include <QtCore> // sets QT_VERSION
namespace lyx {
+typedef map<string, string> AccentsMap;
+static AccentsMap accents = map<string, string>();
// Helper class for deciding what should be ignored
class IgnoreFormats {
typedef vector<pair<string, string> > Escapes;
-/// A map of symbols and their escaped equivalent needed within a regex.
-/// @note Beware of order
-/*
-Escapes const & get_regexp_escapes()
-{
- typedef std::pair<std::string, std::string> P;
-
- static Escapes escape_map;
- if (escape_map.empty()) {
- escape_map.push_back(P("$", "_x_$"));
- escape_map.push_back(P("{", "_x_{"));
- escape_map.push_back(P("}", "_x_}"));
- escape_map.push_back(P("[", "_x_["));
- escape_map.push_back(P("]", "_x_]"));
- escape_map.push_back(P("(", "_x_("));
- escape_map.push_back(P(")", "_x_)"));
- escape_map.push_back(P("+", "_x_+"));
- escape_map.push_back(P("*", "_x_*"));
- escape_map.push_back(P(".", "_x_."));
- escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
- escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
- escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
- escape_map.push_back(P("_x_", "\\"));
- }
- return escape_map;
-}
-*/
-
-/// A map of lyx escaped strings and their unescaped equivalent.
-/*
-Escapes const & get_lyx_unescapes()
-{
- typedef std::pair<std::string, std::string> P;
-
- static Escapes escape_map;
- if (escape_map.empty()) {
- escape_map.push_back(P("\\%", "%"));
- escape_map.push_back(P("\\{", "{"));
- escape_map.push_back(P("\\}", "}"));
- escape_map.push_back(P("\\mathcircumflex ", "^"));
- escape_map.push_back(P("\\mathcircumflex", "^"));
- escape_map.push_back(P("\\backslash ", "\\"));
- escape_map.push_back(P("\\backslash", "\\"));
- escape_map.push_back(P("\\sim ", "~"));
- escape_map.push_back(P("\\sim", "~"));
- }
- return escape_map;
-}
-*/
-
-/// A map of escapes turning a regexp matching text to one matching latex.
-/*
-Escapes const & get_regexp_latex_escapes()
-{
- typedef std::pair<std::string, std::string> P;
-
- static Escapes escape_map;
- if (escape_map.empty()) {
- escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
- escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
- escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
- escape_map.push_back(P("\\[", "\\{\\[\\}"));
- escape_map.push_back(P("\\]", "\\{\\]\\}"));
- escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
- escape_map.push_back(P("%", "\\\\\\%"));
- escape_map.push_back(P("#", "\\\\#"));
- }
- return escape_map;
-}
-*/
-
-/** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
- ** the found occurrence were escaped.
- **/
-/*
-string apply_escapes(string s, Escapes const & escape_map)
-{
- LYXERR(Debug::FIND, "Escaping: '" << s << "'");
- Escapes::const_iterator it;
- for (it = escape_map.begin(); it != escape_map.end(); ++it) {
-// LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
- unsigned int pos = 0;
- while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
- s.replace(pos, it->first.length(), it->second);
- LYXERR(Debug::FIND, "After escape: " << s);
- pos += it->second.length();
-// LYXERR(Debug::FIND, "pos: " << pos);
- }
- }
- LYXERR(Debug::FIND, "Escaped : '" << s << "'");
- return s;
-}
-*/
-
-
string string2regex(string in)
{
static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\$\\])" };
return temp2;
}
-string correctRegex(string t)
+string correctRegex(string t, bool withformat)
{
/* Convert \backslash => \
* and \{, \}, \[, \] => {, }, [, ]
regex wordre("(\\\\)*(\\\\((backslash|mathcircumflex) ?|[\\[\\]\\{\\}]))");
size_t lastpos = 0;
smatch sub;
+ bool backslashed = false;
for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
sub = *it;
string replace;
continue;
}
else {
- if (sub.str(4) == "backslash")
+ if (sub.str(4) == "backslash") {
replace = "\\";
+ if (withformat) {
+ // transforms '\backslash \{' into '\{'
+ // and '\{' into '{'
+ string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
+ if ((next == "\\{") || (next == "\\}")) {
+ replace = "";
+ backslashed = true;
+ }
+ }
+ }
else if (sub.str(4) == "mathcircumflex")
replace = "^";
+ else if (backslashed) {
+ backslashed = false;
+ if (withformat && (sub.str(3) == "{"))
+ replace = accents["braceleft"];
+ else if (withformat && (sub.str(3) == "}"))
+ replace = accents["braceright"];
+ else {
+ // else part should not exist
+ LASSERT(1, /**/);
+ }
+ }
else
replace = sub.str(3);
}
/// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
/// while outside apply get_lyx_unescapes()+get_regexp_escapes().
/// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
-string escape_for_regex(string s)
+string escape_for_regex(string s, bool withformat)
{
size_t lastpos = 0;
string result = "";
break;
}
size_t end_pos = s.find("\\endregexp{}}", regex_pos + 8);
- result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)));
+ result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
lastpos = end_pos + 13;
}
return result;
return rv;
}
-#if 0
-/** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
- **
- ** Verify that closed braces exactly match open braces. This avoids that, for example,
- ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
- **
- ** @param unmatched
- ** Number of open braces that must remain open at the end for the verification to succeed.
- **/
-#if QTSEARCH
-bool braces_match(QString const & beg,
- int unmatched = 0)
-#else
-bool braces_match(string const & beg,
- int unmatched = 0)
-#endif
-{
- int open_pars = 0;
-#if QTSEARCH
- LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg.toStdString() << "'");
-#else
- LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg << "'");
-#endif
- int lastidx = beg.size();
- for (int i=0; i < lastidx; ++i) {
- // Skip escaped braces in the count
-#if QTSEARCH
- QChar c = beg.at(i);
-#else
- char c = beg.at(i);
-#endif
- if (c == '\\') {
- ++i;
- if (i >= lastidx)
- break;
- } else if (c == '{') {
- ++open_pars;
- } else if (c == '}') {
- if (open_pars == 0) {
- LYXERR(Debug::FIND, "Found unmatched closed brace");
- return false;
- } else
- --open_pars;
- }
- }
- if (open_pars != unmatched) {
- LYXERR(Debug::FIND, "Found " << open_pars
- << " instead of " << unmatched
- << " unmatched open braces at the end of count");
- return false;
- }
- LYXERR(Debug::FIND, "Braces match as expected");
- return true;
-}
-#endif
-
class MatchResult {
public:
enum range {
int pos;
int leadsize;
int pos_len;
+ int searched_size;
vector <string> result = vector <string>();
- MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1) {};
+ MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {};
};
static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
{
- int range = oldres.match_len;
- if (range > 0) range--;
- if (newres.match2end < oldres.match2end - oldres.match_len)
+ if (newres.match2end < oldres.match2end)
return MatchResult::newIsTooFar;
if (newres.match_len < oldres.match_len)
return MatchResult::newIsTooFar;
- if ((newres.match_len == oldres.match_len) &&
- (newres.match2end < oldres.match2end + range) &&
- (newres.match2end > oldres.match2end - range)) {
- return MatchResult::newIsBetter;
+
+ if (newres.match_len == oldres.match_len) {
+ if (newres.match2end == oldres.match2end)
+ return MatchResult::newIsBetter;
}
return MatchResult::newIsInvalid;
}
}
}
-typedef map<string, string> AccentsMap;
-static AccentsMap accents = map<string, string>();
-
static void buildaccent(string n, string param, string values)
{
stringstream s(n);
accents["latexe"] = getutf8(0xf0013);
accents["LaTeXe"] = getutf8(0xf0013);
accents["lyxarrow"] = getutf8(0xf0020);
+ accents["braceleft"] = getutf8(0xf0030);
+ accents["braceright"] = getutf8(0xf0031);
accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
accents["backslash LyX"] = getutf8(0xf0010);
accents["backslash tex"] = getutf8(0xf0011);
buildAccentsMap();
static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
"cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
- "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))|guillemot(left|right)|textasciicircum|mathcircumflex|sim)(?![a-zA-Z]))");
+ "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))|(brace|guillemot)(left|right)|textasciicircum|mathcircumflex|sim)(?![a-zA-Z]))");
smatch sub;
for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
sub = *itacc;
case KeyInfo::doRemove: {
// Remove the key with all parameters and following spaces
size_t pos;
- for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
+ size_t start;
+ if (interval_.par[actual._dataEnd-1] == ' ')
+ start = actual._dataEnd;
+ else
+ start = actual._dataEnd+1;
+ for (pos = start; pos < interval_.par.length(); pos++) {
if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
break;
}
// can be digested by our search engine
LYXERR(Debug::FIND, "input: \"" << par << "\"");
result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
- LYXERR(Debug::FIND, "After split: \"" << result << "\"");
+ LYXERR(Debug::FIND, "After splitOnKnownMacros:\n\"" << result << "\"");
}
else
result = par.substr(0, parlen);
return "";
}
}
+
}
else {
// LYXERR(Debug::INFO, "No regex formats");
s += "\\S";
lastpos = sub.position(2) + sub.length(2);
}
- if (lastpos == 0)
+ if (lastpos == 0) {
+ s = "\\b" + t + "\\b";
+ t = s;
return;
+ }
else if (lastpos < t.length())
s += t.substr(lastpos, t.length() - lastpos);
t = "\\b" + s + "\\b";
LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
}
- LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
- par_as_string = escape_for_regex(par_as_string);
+ // LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
+ par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
// Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
- LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
- LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
+ // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
++close_wildcards;
- /*
- if (
- // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
- regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
- // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
- || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
- // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
- || regex_replace(par_as_string, par_as_string,
- "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
- // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
- || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
- ) {
- ++close_wildcards;
- }
- */
size_t lng = par_as_string.size();
if (!opt.ignoreformat) {
// Remove extra '\}' at end if not part of \{\.\}
if (lng < par_as_string.size())
par_as_string = par_as_string.substr(0,lng);
}
+ LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
if ((lng > 0) && (par_as_string[0] == '^')) {
par_as_string = par_as_string.substr(1);
--lng;
opt.matchstart = true;
}
- LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
- LYXERR(Debug::FIND, "Open braces: " << open_braces);
- LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
+ // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
+ // LYXERR(Debug::FIND, "Open braces: " << open_braces);
+ // LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
// If entered regexp must match at begin of searched string buffer
// Kornel: Added parentheses to use $1 for size of the leading string
string dest = "\\" + std::to_string(i+2);
while (regex_replace(par_as_string, par_as_string, orig, dest));
}
- /* opt.matchword is ignored if using regex
- so expanding par_as_string with "\\b" seems appropriate here
- if regex contains for instance '.*' or '.+'
- 1.) Nothing to do, if 'par_as_string' contains "\\b" already.
- (Means, that the user knows how to handle whole words
- 2.) else replace '.' with "\\S" and wrap the regex with "\\b"
- */
if (opt.matchword) {
modifyRegexForMatchWord(par_as_string);
opt.matchword = false;
}
}
-#if 0
-// Count number of characters in string
-// {]} ==> 1
-// \& ==> 1
-// --- ==> 1
-// \\[a-zA-Z]+ ==> 1
-#if QTSEARCH
-static int computeSize(QStringRef s, int len)
-#define isLyxAlpha(arg) arg.isLetter()
-#else
-static int computeSize(string s, int len)
-#define isLyxAlpha(arg) isalpha(arg)
-#endif
-{
- if (len == 0)
- return 0;
- int skip = 1;
- int count = 0;
- for (int i = 0; i < len; i += skip, count++) {
- if (s.at(i) == '\\') {
- skip = 2;
- if (i + 1 < len && isLyxAlpha(s.at(i+1))) {
- for (int j = 2; i+j < len; j++) {
- if (! isLyxAlpha(s.at(i+j))) {
- if (s.at(i+j) == ' ')
- skip++;
- else if (s.at(i+j) == '{') {
- if (i+j+1 < len && s.at(i+j+1) == '}')
- skip += 2;
- else if (i + j + 1 >= len)
- skip++;
- }
- break;
- }
- skip++;
- }
- }
- }
- else if (s.at(i) == '{') {
- if (i + 1 < len && s.at(i+1) == '}')
- skip = 2;
- else
- skip = 3;
- }
- else if (s.at(i) == '-') {
- if (i+1 < len && s.at(i+1) == '-') {
- if (i + 2 < len && s.at(i+2) == '-')
- skip = 3;
- else
- skip = 2;
- }
- else
- skip = 1;
- }
- else {
- skip = 1;
- }
- }
- return count;
-}
-#endif
-
MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
{
MatchResult mres;
+ mres.searched_size = len;
if (at_begin &&
(opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
return mres;
docstring docstr = stringifyFromForSearch(opt, cur, len);
string str;
- if (use_regexp || opt.casesensitive)
- str = normalize(docstr);
- else
- str = normalize(lowercase(docstr));
+ str = normalize(docstr);
if (!opt.ignoreformat) {
str = correctlanguagesetting(str, false, !opt.ignoreformat);
+ // remove closing '}' and '\n' to allow for use of '$' in regex
+ size_t lng = str.size();
+ while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
+ lng--;
+ if (lng != str.size()) {
+ str = str.substr(0, lng);
+ }
}
if (str.empty()) {
mres.match_len = -1;
return mres;
}
- LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
- LYXERR(Debug::FIND, "After normalization: '" << str << "'");
+ LYXERR(Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
- if (use_regexp) {
+ LASSERT(use_regexp, /**/);
+ {
+ // use_regexp always true
LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
#if QTSEARCH
QString qstr = QString::fromStdString(str);
#endif
return mres;
}
-
- // else !use_regexp: but all code paths above return
- LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
- << par_as_string << "', str='" << str << "'");
- LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
- << lead_as_string << "', par_as_string_nolead='"
- << par_as_string_nolead << "'");
-
- if (at_begin) {
- LYXERR(Debug::FIND, "size=" << par_as_string.size()
- << ", substr='" << str.substr(0, par_as_string.size()) << "'");
- if (str.substr(0, par_as_string.size()) == par_as_string) {
- mres.match_len = par_as_string.size();
- mres.match2end = str.size();
- mres.pos = 0;
- return mres;
- }
- } else {
- // Start the search _after_ the leading part
- size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
- if (pos != string::npos) {
- mres.match_len = par_as_string.size();
- mres.match2end = str.size() - pos;
- mres.pos = pos;
- return mres;
- }
- }
- return mres;
}
}
else
return mres;
- /* DEAD CODE follows
- if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
- return mres;
- if ((len > 0) && (res < len)) {
- mres.match_len = 0;
- return mres;
- }
- Paragraph const & par = cur.paragraph();
- bool ws_left = (cur.pos() > 0)
- ? par.isWordSeparator(cur.pos() - 1)
- : true;
- bool ws_right;
- if (len < 0)
- ws_right = true;
- else {
- ws_right = (cur.pos() + len < par.size())
- ? par.isWordSeparator(cur.pos() + len)
- : true;
- }
- LYXERR(Debug::FIND,
- "cur.pos()=" << cur.pos() << ", res=" << res
- << ", separ: " << ws_left << ", " << ws_right
- << ", len: " << len
- << endl);
- if (ws_left && ws_right) {
- // Check for word separators inside the found 'word'
- for (int i = 0; i < len; i++) {
- if (par.isWordSeparator(cur.pos() + i)) {
- mres.match_len = 0;
- return mres;
- }
- }
- return mres;
- }
- mres.match_len = 0;
- return mres;
- */
}
#if 0
// Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
// Kornel: Added textsl, textsf, textit, texttt and noun
// + allow to seach for colored text too
- LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
+ LYXERR(Debug::FIND, "Removing stale empty macros from: " << t);
while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
*/
docstring latexifyFromCursor(DocIterator const & cur, int len)
{
+ /*
LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
<< cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
+ */
Buffer const & buf = *cur.buffer();
odocstringstream ods;
#if defined(ResultsDebug)
// Debugging output
-static void displayMResult(MatchResult &mres, int increment, string from)
+static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
{
- LYXERR0( "from:\t\t\t" << from);
- LYXERR0( "pos: " << mres.pos << " increment " << increment);
- LYXERR0( "leadsize: " << mres.leadsize);
- LYXERR0( "match_len: " << mres.match_len);
- LYXERR0( "match_prefix: " << mres.match_prefix);
- LYXERR0( "match2end: " << mres.match2end);
- LYXERR0( "pos_len: " << mres.pos_len); // Set in finalize
- for (size_t i = 0; i < mres.result.size(); i++)
- LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
-}
- #define displayMres(s,i, txt) displayMResult(s,i, xtx);
-#else
- #define displayMres(s,i, txt)
-#endif
-
-/*
- * Not good, we miss possible matches containing also characters not found in
- * the innermost depth.
-static bool findAdvForwardInnermost(DocIterator & cur)
-{
- size_t d;
- DocIterator old_cur = cur;
- int forwardCount = 0;
- do {
- d = cur.depth();
- old_cur = cur;
- cur.forwardPos();
- if (!cur) {
- break;
- }
- if (cur.depth() > d) {
- forwardCount++;
- continue;
+ LYXERR0( "from:\t\t\t" << from);
+ string status;
+ if (mres.pos_len > 0) {
+ // Set in finalize
+ status = "FINALSEARCH";
+ }
+ else {
+ if (mres.match_len > 0) {
+ if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
+ status = "Good Match";
+ else
+ status = "Matched in";
}
- if (cur.depth() == d)
- break;
- } while(1);
- cur = old_cur;
- if (forwardCount > 0) {
- LYXERR(Debug::FIND, "Forwarded " << forwardCount << " step(s) (searching for innermost match)");
- return true;;
+ else
+ status = "MissedSearch";
}
- else
- return false;
+
+ LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
+ if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
+ LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
+ if ((mres.pos > 0) || (mres.match_prefix > 0))
+ LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
+ for (size_t i = 0; i < mres.result.size(); i++)
+ LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
}
-*/
+ #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
+#else
+ #define displayMres(s, txt, cur)
+#endif
/** Finalize an advanced find operation, advancing the cursor to the innermost
** position that matches, plus computing the length of the matching text to
// so the search for "www" gives prefix_len = 7 (== sizeof("http://")
// and although we search for only 3 chars, we find the whole hyperlink inset
bool at_begin = (expected.match_prefix == 0);
- //if (findAdvForwardInnermost(cur)) {
- if (expected.match_len > 0) {
+ LASSERT(at_begin, /**/);
+ if (expected.match_len > 0 && at_begin) {
+ // Search for deepest match
old_cur = cur;
max_match = expected;
do {
break;
if (cur.depth() < d)
break;
- if ((cur.depth() == d) && at_begin)
+ if (cur.depth() == d)
break;
- mres = match(cur, -1, at_begin);
- displayMres(mres, -1, "Checking innermost");
+ size_t lastd = d;
+ while (cur && cur.depth() > lastd) {
+ lastd = cur.depth();
+ mres = match(cur, -1, at_begin);
+ displayMres(mres, "Checking innermost", cur);
+ if (mres.match_len > 0)
+ break;
+ // maybe deeper?
+ cur.forwardPos();
+ }
if (mres.match_len < expected.match_len)
break;
- if (!at_begin && (mres.match_prefix > max_match.match_prefix))
- break;
max_match = mres;
- at_begin = (max_match.match_prefix == 0);
old_cur = cur;;
} while(1);
cur = old_cur;
else {
// (expected.match_len <= 0)
mres = match(cur); /* match valid only if not searching whole words */
- displayMres(mres, 0, "Start with negative match");
+ displayMres(mres, "Start with negative match", cur);
max_match = mres;
}
if (max_match.match_len <= 0) return fail;
int len = 1;
if (cur.pos() + len > cur.lastpos())
return fail;
- // regexp should use \w+, \S+, or \b(some string)\b
- // to search for whole words
- if (match.opt.matchword && !match.use_regexp) {
- LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
- while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
- ++len;
- LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
- }
- // Length of matched text (different from len param)
- static MatchResult old_match = match(cur, len, at_begin);
- if (old_match.match_len < 0)
- old_match = fail;
- MatchResult new_match;
- // Greedy behaviour while matching regexps
- while ((new_match = match(cur, len + 1, at_begin)).match_len > old_match.match_len) {
- ++len;
- old_match = new_match;
- LYXERR(Debug::FIND, "verifying match with len = " << len);
- }
- return old_match;
- }
- else {
+
+ LASSERT(match.use_regexp, /**/);
+ {
int minl = 1;
int maxl = cur.lastpos() - cur.pos();
// Greedy behaviour while matching regexps
while (maxl > minl) {
MatchResult mres2;
mres2 = match(cur, len, at_begin);
- displayMres(mres2, len, "Finalize loop");
- int actual_match = mres2.match_len;
- if (actual_match >= max_match.match_len) {
- // actual_match > max_match _can_ happen,
+ displayMres(mres2, "Finalize loop", cur);
+ int actual_match_len = mres2.match_len;
+ if (actual_match_len >= max_match.match_len) {
+ // actual_match_len > max_match _can_ happen,
// if the search area splits
// some following word so that the regex
// (e.g. 'r.*r\b' matches 'r' from the middle of the
// splitted word)
// This means, the len value is too big
+ actual_match_len = max_match.match_len;
+ max_match = mres2;
+ max_match.match_len = actual_match_len;
maxl = len;
if (maxl - minl < 4)
len = (int)((maxl + minl)/2);
len = (int)(minl + (maxl - minl + 3)/4);
}
else {
- // (actual_match < max_match)
+ // (actual_match_len < max_match.match_len)
minl = len + 1;
len = (int)((maxl + minl)/2);
}
}
+ len = minl;
old_cur = cur;
// Search for real start of matched characters
while (len > 1) {
// Ha, got it! The shorter selection has the same match length
len--;
old_cur = cur;
+ max_match = actual_match;
}
else {
// OK, the shorter selection matches less chars, revert to previous value
return fail;
else {
max_match.pos_len = len;
+ displayMres(max_match, "SEARCH RESULT", cur)
return max_match;
}
}
{
if (!cur)
return 0;
+ bool repeat = false;
+ DocIterator orig_cur; // to be used if repeat not successful
+ MatchResult orig_mres;
while (!theApp()->longOperationCancelled() && cur) {
//(void) findAdvForwardInnermost(cur);
LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
MatchResult mres = match(cur, -1, false);
- displayMres(mres,-1, "Starting findForwardAdv")
+ string msg = "Starting";
+ if (repeat)
+ msg = "Repeated";
+ displayMres(mres, msg + " findForwardAdv", cur)
int match_len = mres.match_len;
if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
match_len = 0;
}
- if (match_len > 0) {
+ if (match_len <= 0) {
+ // This should exit nested insets, if any, or otherwise undefine the currsor.
+ cur.pos() = cur.lastpos();
+ LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
+ cur.forwardPos();
+ }
+ else { // match_len > 0
// Try to find the begin of searched string
int increment;
int firstInvalid = 100000;
- if (mres.match_prefix + mres.pos - mres.leadsize > 1)
- increment = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
- else
- increment = 1;
+ {
+ int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
+ int incrcur = (cur.lastpos() - cur.pos() + 1 )*3/4;
+ if (incrcur < incrmatch)
+ increment = incrcur;
+ else
+ increment = incrmatch;
+ if (increment < 1)
+ increment = 1;
+ }
LYXERR(Debug::FIND, "Set increment to " << increment);
while (increment > 0) {
DocIterator old_cur = cur;
- for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
+ size_t skipping = cur.depth();
+ for (int i = 0; i < increment && cur; i++) {
+ cur.forwardPos();
+ while (cur && cur.depth() > skipping) {
+ cur.pos() = cur.lastpos();
+ cur.forwardPos();
+ }
}
if (! cur || (cur.pit() > old_cur.pit())) {
// Are we outside of the paragraph?
}
else {
MatchResult mres2 = match(cur, -1, false);
- displayMres(mres2,increment, "findForwardAdv loop")
+ displayMres(mres2, "findForwardAdv loop", cur)
switch (interpretMatch(mres, mres2)) {
case MatchResult::newIsTooFar:
// behind the expected match
increment /= 2;
break;
case MatchResult::newIsBetter:
- // not reached ye, but cur.pos()+increment is bettert
+ // not reached yet, but cur.pos()+increment is bettert
mres = mres2;
firstInvalid -= increment;
if (increment > firstInvalid*3/4)
default:
// Todo@
// Handle not like MatchResult::newIsTooFar
- // LYXERR0( "Something is wrong: Increment = " << increment << " match_prefix = " << mres.match_prefix);
+ LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
firstInvalid--;
increment = increment*3/4;
cur = old_cur;
}
}
}
- // LYXERR0("Leaving first loop");
- {
- LYXERR(Debug::FIND, "Finalizing 1");
- MatchResult found_match = findAdvFinalize(cur, match, mres);
- if (found_match.match_len > 0) {
- LASSERT(found_match.pos_len > 0, /**/);
- match.FillResults(found_match);
- return found_match.pos_len;
- }
- else {
- // try next possible match
- cur.forwardPos();
- continue;
- }
- }
- // The following code is newer reached
- // but parts of it may be needed in future
- int match_len_zero_count = 0;
- MatchResult mres3;
- for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
- if (i++ > 3) {
- mres3 = match(cur, -1, false);
- displayMres(mres3, 1, 1 "Prepare finalize in findForwardAdv")
- int remaining_len = mres3.match_len;
- if (remaining_len <= 0) {
- // Apparently the searched string is not in the remaining part
- break;
- }
- else {
- i = 0;
- }
- }
- LYXERR(Debug::FIND, "Advancing cur: " << cur);
- mres3 = match(cur, 1);
- displayMres(mres3, 1, "Prepare 2 finalize in findForwardAdv")
- int match_len3 = mres3.match_len;
- if (match_len3 < 0)
+ if (mres.match_len > 0) {
+ if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
+ repeat = true;
+ orig_cur = cur;
+ orig_mres = mres;
+ cur.forwardPos();
continue;
- mres3 = match(cur);
- displayMres(mres3, 1, "Prepare 3 finalize in findForwardAdv")
- int match_len2 = mres3.match_len;
- LYXERR(Debug::FIND, "match_len2: " << match_len2);
- if (match_len2 > 0) {
- // Sometimes in finalize we understand it wasn't a match
- // and we need to continue the outest loop
- LYXERR(Debug::FIND, "Finalizing 2");
- MatchResult mres4 = findAdvFinalize(cur, match, mres.match_len);
- if (mres4.match_len > 0) {
- match.FillResults(mres4);
- LASSERT(mres4.pos_len > 0, /**/);
- return mres4.pos_len;
- }
- }
- if (match_len2 > 0)
- match_len_zero_count = 0;
- else if (match_len2 == 0)
- match_len_zero_count++;
- if (match_len2 < 0) {
- if (++match_len_zero_count > 3) {
- LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
- }
- break;
}
}
- if (!cur)
- return 0;
- }
- if (match_len >= 0 && cur.pit() < cur.lastpit()) {
- LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
- cur.forwardPar();
- } else {
- // This should exit nested insets, if any, or otherwise undefine the currsor.
- cur.pos() = cur.lastpos();
- LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
- cur.forwardPos();
+ else if (repeat) {
+ // seems to never be reached.
+ cur = orig_cur;
+ mres = orig_mres;
+ }
+ // LYXERR0("Leaving first loop");
+ LYXERR(Debug::FIND, "Finalizing 1");
+ MatchResult found_match = findAdvFinalize(cur, match, mres);
+ if (found_match.match_len > 0) {
+ LASSERT(found_match.pos_len > 0, /**/);
+ match.FillResults(found_match);
+ return found_match.pos_len;
+ }
+ else {
+ // try next possible match
+ cur.forwardPos();
+ repeat = false;
+ continue;
+ }
}
}
return 0;
}
} // namespace
-#if 1
static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
{
// Should replace the string "$" + std::to_string(matchnum) with replacement
t = s;
return true;
}
-#endif
///
static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
{
- LYXERR(Debug::FIND, "parsing");
+ // LYXERR(Debug::FIND, "parsing");
string s;
string line;
getline(is, line);
break;
getline(is, line);
}
- LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
+ // LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
opt.find_buf_name = from_utf8(s);
is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
is.get(); // Waste space before replace string
break;
getline(is, line);
}
- LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
+ // LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
opt.repl_buf_name = from_utf8(s);
is >> opt.keep_case;
int i;
is >> i;
opt.restr = FindAndReplaceOptions::SearchRestriction(i);
+ /*
LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
<< opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
<< opt.scope << ' ' << opt.restr);
+ */
return is;
}