*/
#include <config.h>
+#include <iterator>
#include "lyxfind.h"
//#define ResultsDebug
#define USE_QT_FOR_SEARCH
#if defined(USE_QT_FOR_SEARCH)
- #include <QtCore> // sets QT_VERSION
- #if (QT_VERSION >= 0x050000)
- #include <QRegularExpression>
- #define QTSEARCH 1
- #else
- #define QTSEARCH 0
- #endif
+ #include <QRegularExpression>
+ #define QTSEARCH 1
#else
#define QTSEARCH 0
#endif
if (match_len > 0)
bv->putSelectionAt(cur, match_len, !forward);
- else if (onlysel) {
+ else if (onlysel && bv->cursor().selection()) {
docstring q = _("The search string was not found within the selection.\n"
"Continue search outside?");
int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
// normal blanks
blanks++;
}
- else if ((tempx[i] == '\302' && tempx[i+1] == '\240')
- || (tempx[i] == '\342' && tempx[i+1] == '\200')) {
- // protected space
- // thin space
+ else if (tempx[i] == '\302' && tempx[i+1] == '\240') {
+ // Normal Space
blanks++;
i++;
}
+ else if (tempx[i] == '\342') {
+ if (tempx[i+1] == '\200') {
+ if ((tempx[i+2] == '\257')
+ || (tempx[i+2] == '\203')
+ || (tempx[i+2] == '\202')) {
+ // Non-breaking Thin (1/6 em)
+ // Quad(1 em), (Double quad counts as 2 blanks)
+ // Half Quad
+ blanks++;
+ i += 2;
+ }
+ else if (tempx[i+2] == '\213') {
+ // Ignoring parts of Medium and Thick
+ i += 2;
+ continue;
+ }
+ else if ((tempx[i+2] == '\204') || (tempx[i+2] == '\205')) {
+ // Thick
+ // Medium
+ blanks++;
+ i += 2;
+ }
+ }
+ else if (tempx[i+1] == '\201') {
+ if (tempx[i+2] == '\240') {
+ // Ignoring parts of half quad
+ i += 2;
+ continue;
+ }
+ }
+ else if ((tempx[i+1] == '\220') && (tempx[i+2] == '\243')) {
+ // Visible space
+ blanks++;
+ i += 2;
+ }
+ }
else {
if (blanks > 0) {
temp += getRegexSpaceCount(blanks);
}
else {
if (sub.str(4) == "backslash") {
- replace = "\\";
+ replace = string("\\");
{
// transforms '\backslash \{' into '\{'
string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
}
else if (sub.str(4) == "mathcircumflex")
replace = "^";
+ else if ((sub.str(4) == "negthinspace") || (sub.str(4) == "negmedspace") || (sub.str(4) == "negthickspace")) {
+ replace = accents[sub.str(4)+"{}"];
+ }
else if (backslashed) {
backslashed = false;
if (withformat) {
** @todo Normalization should also expand macros, if the corresponding
** search option was checked.
**/
- string normalize(docstring const & s, bool ignore_fomat) const;
+ string convertLF2Space(docstring const & s, bool ignore_fomat) const;
// normalized string to search
string par_as_string;
// regular expression to use for searching
// par_as_string after removal of lead_as_string
string par_as_string_nolead;
// unmatched open braces in the search string/regexp
- int open_braces;
+ int open_braces = 0;
// number of (.*?) subexpressions added at end of search regexp for closing
// environments, math mode, styles, etc...
- int close_wildcards;
+ int close_wildcards = 0;
public:
// Are we searching with regular expressions ?
bool use_regexp = false;
accents["guillemotleft"] = "«";
accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
- accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
+ accents["negthinspace{}"]= getutf8(0xf0003); // to omit backslashed latex macros
accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
- accents["negmedspace"] = getutf8(0xf0005);
+ accents["negmedspace{}"] = getutf8(0xf0005);
accents["thickspace"] = getutf8(0xf0006);
- accents["negthickspace"] = getutf8(0xf0007);
+ accents["negthickspace{}"]= getutf8(0xf0007);
accents["lyx"] = getutf8(0xf0010); // Used logos
accents["LyX"] = getutf8(0xf0010);
accents["tex"] = getutf8(0xf0011);
if (accents.empty())
buildAccentsMap();
static regex const accre("\\\\("
- "([\\S]|[A-Za-z]+)\\{[^\\\\\\{\\}]+\\}"
+ "([\\S]|[A-Za-z]+)\\{[^\\\\\\{\\}]*\\}"
"|([\\S]|[A-Za-z]+)\\{\\\\[ij](math)?\\}"
"|("
"(backslash ([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))"
}
// Ignore language if there is math somewhere in pattern-string
if (isPatternString) {
- for (auto s: usedText) {
+ for (auto const & s: usedText) {
// Remove entries created in previous search runs
keys.erase(s);
}
previous_single_replace = true;
}
// When using regexp, braces are hacked already by escape_for_regex()
- par_as_string = normalize(ds, opt.ignoreformat);
- open_braces = 0;
- close_wildcards = 0;
+ par_as_string = convertLF2Space(ds, opt.ignoreformat);
size_t lead_size = 0;
// correct the language settings
docstring docstr = stringifyFromForSearch(opt, cur, len);
string str;
- str = normalize(docstr, opt.ignoreformat);
+ str = convertLF2Space(docstr, opt.ignoreformat);
if (!opt.ignoreformat) {
str = correctlanguagesetting(str, false, !opt.ignoreformat);
// remove closing '}' and '\n' to allow for use of '$' in regex
}
#endif
-#if 1
-static string convertLF2Space(docstring const &s, bool ignore_format)
+string MatchStringAdv::convertLF2Space(docstring const &s, bool ignore_format) const
{
// Using original docstring to handle '\n'
return(t.str());
}
-#else
-static string convertLF2Space(docstring const & s, bool ignore_format)
-{
- // Using utf8-converted string to handle '\n'
-
- string t;
- t = lyx::to_utf8(s);
- // Remove \n at begin
- while (!t.empty() && t[0] == '\n')
- t = t.substr(1);
- // Remove \n* at end
- while (!t.empty() && t[t.size() - 1] == '\n') {
- t = t.substr(0, t.size() - 1);
- }
- size_t pos;
- // Handle all other '\n'
- while ((pos = t.find("\n")) != string::npos) {
- if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
- // Handle '\\\n'
- if (isPrintableNonspace(t[pos+1]) && ((pos < 3) || isPrintableNonspace(t[pos-3]))) {
- t.replace(pos-2, 3, " ");
- }
- else {
- // Already a space there
- t.replace(pos-2, 3, "");
- }
- }
- else {
- if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
- // '\n' adjacent to non-alpha-numerics, discard
- t.replace(pos, 1, "");
- }
- else {
- // Replace all other \n with spaces
- t.replace(pos, 1, " ");
- }
- if (!ignore_format) {
- size_t count = 0;
- while ((pos > count + 1) && (t[pos - 1 -count] == '%')) {
- count++;
- }
- if (count > 0) {
- t.replace(pos - count, count, "");
- }
- }
- }
- }
- return(t);
-
-}
-#endif
-
-string MatchStringAdv::normalize(docstring const & s, bool ignore_format) const
-{
- string t = convertLF2Space(s, ignore_format);
-
- // The following replaces are not appropriate in non-format-search mode
- if (!ignore_format) {
- // 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::FINDVERBOSE, "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::FINDVERBOSE, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
- while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
- LYXERR(Debug::FINDVERBOSE, " further removing stale empty \\section{}, \\part{}, \\paragraph{} macros from: " << t);
- while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
- }
- return t;
-}
-
docstring stringifyFromCursor(DocIterator const & cur, int len)
{
LYXERR(Debug::FINDVERBOSE, "Stringifying with len=" << len << " from cursor at pos: " << cur);
{
if (!cur)
return 0;
- bool repeat = false;
+ int repeat = 0;
DocIterator orig_cur; // to be used if repeat not successful
MatchResult orig_mres;
+ do {
+ orig_cur = cur;
+ cur.forwardPos();
+ } while (cur.depth() > orig_cur.depth());
+ cur = orig_cur;
while (!theApp()->longOperationCancelled() && cur) {
//(void) findAdvForwardInnermost(cur);
LYXERR(Debug::FINDVERBOSE, "findForwardAdv() cur: " << cur);
MatchResult mres = match(cur, -1, MatchStringAdv::MatchAnyPlace);
string msg = "Starting";
- if (repeat)
+ if (repeat > 0)
msg = "Repeated";
displayMres(mres, msg + " findForwardAdv", cur)
int match_len = mres.match_len;
match_len = 0;
}
if (match_len <= 0) {
- // This should exit nested insets, if any, or otherwise undefine the currsor.
- cur.pos() = cur.lastpos();
+ if (repeat > 0) {
+ repeat--;
+ }
+ else {
+ // This should exit nested insets, if any, or otherwise undefine the currsor.
+ cur.pos() = cur.lastpos();
+ }
LYXERR(Debug::FINDVERBOSE, "Advancing pos: cur=" << cur);
cur.forwardPos();
}
cur.pos() = cur.pos() + increment;
MatchResult mres2 = match(cur, -1, MatchStringAdv::MatchAnyPlace);
displayMres(mres2, "findForwardAdv loop", cur)
- switch (interpretMatch(mres, mres2)) {
+ switch (interpretMatch(mres, mres2)) {
case MatchResult::newIsTooFar:
// behind the expected match
firstInvalid = increment;
increment /= 2;
break;
case MatchResult::newIsBetter:
- // not reached yet, but cur.pos()+increment is bettert
+ // not reached yet, but cur.pos()+increment is better
mres = mres2;
firstInvalid -= increment;
if (increment > firstInvalid*3/4)
if (mres.match_len > 0) {
if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
// The match seems to indicate some deeper level
- repeat = true;
+ repeat = 2;
orig_cur = cur;
orig_mres = mres;
cur.forwardPos();
continue;
}
}
- else if (repeat) {
+ else if (repeat > 0) {
// should never be reached.
cur = orig_cur;
mres = orig_mres;
return 1;
}
+static bool isWordChar(char_type c)
+{
+ return isLetterChar(c) || isNumberChar(c);
+}
/// Perform a FindAdv operation.
bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
num_replaced += findAdvReplace(bv, opt, matchAdv);
cur = bv->cursor();
if (opt.forward) {
- if (opt.matchword) { // Skip word-characters if we are in the mid of a word
- Paragraph const & par = cur.paragraph();
- if ((cur.pos() > 0) && !par.isWordSeparator(cur.pos() -1, true)) {
- while (cur.pos() < par.size()) {
- if (par.isWordSeparator(cur.pos(), true))
+ if (opt.matchword && cur.pos() > 0) { // Skip word-characters if we are in the mid of a word
+ if (cur.inTexted()) {
+ Paragraph const & par = cur.paragraph();
+ int len_limit, new_pos;
+ if (cur.lastpos() < par.size())
+ len_limit = cur.lastpos();
+ else
+ len_limit = par.size();
+ for (new_pos = cur.pos() - 1; new_pos < len_limit; new_pos++) {
+ if (!isWordChar(par.getChar(new_pos)))
break;
- else
- cur.forwardPos();
+ }
+ if (new_pos > cur.pos())
+ cur.pos() = new_pos;
+ }
+ else if (cur.inMathed()) {
+ // Check if 'cur.pos()-1' and 'cur.pos()' both point to a letter,
+ // I am not sure, we should consider the selection
+ bool sel = bv->cursor().selection();
+ if (!sel && cur.pos() < cur.lastpos()) {
+ CursorSlice const & cs = cur.top();
+ MathData md = cs.cell();
+ int len = -1;
+ MathData::const_iterator it_end = md.end();
+ MathData md2;
+ // Start the check with one character before actual cursor position
+ for (MathData::const_iterator it = md.begin() + cs.pos() - 1;
+ it != it_end; ++it)
+ md2.push_back(*it);
+ docstring inp = asString(md2);
+ for (len = 0; (unsigned) len < inp.size() && len + cur.pos() <= cur.lastpos(); len++) {
+ if (!isWordChar(inp[len]))
+ break;
+ }
+ // len == 0 means previous char was a word separator
+ // len == 1 search starts with a word separator
+ // len == 2 ... we have to skip len -1 chars
+ if (len > 1)
+ cur.pos() = cur.pos() + len - 1;
}
}
opt.matchword = false;
projects / lyx.git / blobdiff