Am Donnerstag, 4. Oktober 2018 10:00:42 CEST schrieb Scott Kostyshak <skost...@lyx.org>: > On Thu, Oct 04, 2018 at 10:27:17AM +0200, Kornel Benko wrote: > > Am Donnerstag, 4. Oktober 2018 10:17:43 CEST schrieb Kornel Benko > > <kor...@lyx.org>: > > > Both, lyx2.4 and lyx2.3 > > > To reproduce: > > > > > > 1.) Open (or create) a small lyx-file. (Say 3 lines) > > > 2.) select the whole buffer > > > 3.) in findadv search for a string (ignore format or not) > > > ==> crash > > > > > > LASSERT() is called at lyxfind.cpp in stringifyFromForSearch() > > > LASSERT(cur.pos() >= 0 && cur.pos() <= cur.lastpos(), > > > return docstring()); > > > > > > In my case > > > cur.pos() == 12 > > > cur.lastpos() == 11 > > > len == -1 > > > > > > > OK, selecting the lines so, that the last selected line is longer than the > > first selected, > > then we get the crash. Looks like cur.pos() and cur.lastpos() are referring > > to different lines. > > I can reproduce. It seems I can reproduce even on 2.1.0, which is > surprising since I would have guessed this bug would be more common. > Well, good find! Probably a lot of people have run into this and it > hasn't been reported (?) [1]. > > Scott
I am fighting with regex and language distinction. Our latex output does not
specify
enough info for the find routines. The playing with this led to crash.
Now, I have a working solution. Would you like to test?
If setting for 'not ignore format', you should be able to find for instance a
string with specified language.
Unfortunately there is there are many possible 'features' which have to be the
same as in the searched string.
So searching as regex [a-z]* only words in document language, not
bold/emphasized or whatever.
Kornel
diff --git a/src/lyxfind.cpp b/src/lyxfind.cpp
index 936ea24..1452f6c 100644
--- a/src/lyxfind.cpp
+++ b/src/lyxfind.cpp
@@ -797,10 +797,11 @@ static docstring buffer_to_latex(Buffer & buffer)
runparams.nice = true;
runparams.flavor = OutputParams::LATEX;
runparams.linelen = 80; //lyxrc.plaintext_linelen;
// No side effect of file copying and image conversion
runparams.dryrun = true;
+ runparams.for_search = true;
pit_type const endpit = buffer.paragraphs().size();
for (pit_type pit = 0; pit != endpit; ++pit) {
TeXOnePar(buffer, buffer.text(), pit, os, runparams);
LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
}
@@ -841,19 +842,105 @@ static size_t identifyLeading(string const & s)
{
string t = s;
// @TODO Support \item[text]
// Kornel: Added textsl, textsf, textit, texttt and noun
// + allow to seach for colored text too
- while (regex_replace(t, t, REGEX_BOS "\\\\(emph|noun|text(bf|sl|sf|it|tt|color\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave|subsubsection|subsection|section|subparagraph|paragraph|part)\\*?\\{", "")
+ while (regex_replace(t, t, REGEX_BOS "\\\\foreignlanguage\\{[a-z]+\\}\\{", ""))
+ ;
+ while (regex_replace(t, t, REGEX_BOS "\\\\((emph|noun|text(bf|sl|sf|it|tt|color\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part)\\*?)\\{", "")
|| regex_replace(t, t, REGEX_BOS "\\$", "")
|| regex_replace(t, t, REGEX_BOS "\\\\\\[ ", "")
|| regex_replace(t, t, REGEX_BOS "\\\\item ", "")
|| regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
- LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
+ ;
+ LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
return s.find(t);
}
+static int findclosing(string p, int start, int end)
+{
+ int skip = 0;
+ int depth = 0;
+ for (int i = start; i < end; i += 1 + skip) {
+ char c;
+ c = p[i];
+ skip = 0;
+ if (c == '\\') skip = 1;
+ else if (c == '{') depth++;
+ else if (c == '}') {
+ if (depth == 0) return(i);
+ --depth;
+ }
+ }
+ return(-1);
+}
+
+
+static string correctlanguagesetting(string par, bool from_regex)
+{
+ static string langstart = "\\foreignlanguage{";
+ static int llen = langstart.length();
+ static bool removefirstlang = false;
+
+ int parlen = par.length();
+ string result = par;
+
+ while ((parlen > 0) && (par[parlen-1] == '\n')) {
+ parlen--;
+ }
+ if (par.compare(0, llen, langstart) == 0) {
+ if (from_regex) {
+ removefirstlang = false;
+ }
+ int i = findclosing(par, llen, par.length());
+ if (removefirstlang) {
+ if (i < 0)
+ result = "";
+ else {
+ int closepos = findclosing(par, i+2, par.length());
+ if (closepos > 0) {
+ result = par.substr(i+2, closepos-i-2) + par.substr(closepos+1, parlen - closepos-1);
+ }
+ else {
+ result = par.substr(i+2, parlen-i-2);
+ }
+ }
+ }
+ else if (i > 0) {
+ // skip '}{' after the language spec
+ int closepos = findclosing(par, i+2, par.length());
+ size_t insertpos = par.find(langstart, i+2);
+ if (closepos < 0) {
+ if (insertpos == string::npos) {
+ // there are no closing in par, and no next lang spec
+ result = par.substr(0, parlen) + "}";
+ }
+ else {
+ // Add '}' at insertpos only, because closing is missing
+ result = par.substr(0,insertpos) + "}" + par.substr(insertpos, parlen-insertpos);
+ }
+ }
+ else if ((size_t) closepos > insertpos) {
+ // Add '}' at insertpos and remove from closepos if closepos > insertpos
+ result = par.substr(0,insertpos) + "}" + par.substr(insertpos, closepos - insertpos) + par.substr(closepos+1, parlen -closepos-1);
+ }
+ }
+ else {
+ result = par;
+ // For i == 0, it is empty language spec
+ // and for i < 0 it is Error
+ }
+ }
+ else {
+ if (from_regex) {
+ removefirstlang = true;
+ }
+ }
+ // Either not found language spec,or is single and closed spec or empty
+ return(result);
+}
+
// Remove trailing closure of math, macros and environments, so to catch parts of them.
static int identifyClosing(string & t)
{
int open_braces = 0;
@@ -885,20 +972,23 @@ MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const &
par_as_string = normalize(ds, !use_regexp);
open_braces = 0;
close_wildcards = 0;
size_t lead_size = 0;
+ // correct the language settings
+ par_as_string = correctlanguagesetting(par_as_string, true);
if (opt.ignoreformat) {
if (!use_regexp) {
// if par_as_string_nolead were emty,
// the following call to findAux will always *find* the string
// in the checked data, and thus always using the slow
// examining of the current text part.
par_as_string_nolead = par_as_string;
}
} else {
lead_size = identifyLeading(par_as_string);
+ LYXERR(Debug::FIND, "Lead_size: " << lead_size);
lead_as_string = par_as_string.substr(0, lead_size);
par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
}
if (!use_regexp) {
@@ -1106,13 +1196,14 @@ string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
t.replace(pos, 1, " ");
// 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);
- while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt|color\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave|subsubsection|subsection|section|subparagraph|paragraph|part)(\\{\\})+", ""))
+ while (regex_replace(t, t, "\\\\((emph|noun|text(bf|sl|sf|it|tt|color\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part)\\*?)(\\{\\})+", ""))
LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
+ while (regex_replace(t, t, "\\\\foreignlanguage\\{[a-z]+\\}(\\{(\\\\item )?\\})+", ""));
// FIXME - check what preceeds the brace
if (hack_braces) {
if (opt.ignoreformat)
while (regex_replace(t, t, "\\{", "_x_<")
|| regex_replace(t, t, "\\}", "_x_>"))
@@ -1183,19 +1274,23 @@ docstring latexifyFromCursor(DocIterator const & cur, int len)
runparams.nice = false;
runparams.flavor = OutputParams::LATEX;
runparams.linelen = 8000; //lyxrc.plaintext_linelen;
// No side effect of file copying and image conversion
runparams.dryrun = true;
+ runparams.for_search = true;
if (cur.inTexted()) {
// @TODO what about searching beyond/across paragraph breaks ?
pos_type endpos = cur.paragraph().size();
if (len != -1 && endpos > cur.pos() + len)
endpos = cur.pos() + len;
TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
string(), cur.pos(), endpos);
LYXERR(Debug::FIND, "Latexified text: '" << lyx::to_utf8(ods.str()) << "'");
+ string s = correctlanguagesetting(lyx::to_utf8(ods.str()), false);
+ LYXERR(Debug::FIND, "Latexified text: '" << s << "'");
+ return(lyx::from_utf8(s));
} else if (cur.inMathed()) {
// Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
for (int s = cur.depth() - 1; s >= 0; --s) {
CursorSlice const & cs = cur[s];
if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
@@ -1391,12 +1486,12 @@ int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
DocIterator const & cur, int len)
{
- LASSERT(cur.pos() >= 0 && cur.pos() <= cur.lastpos(),
- return docstring());
+ if (cur.pos() < 0 || cur.pos() > cur.lastpos())
+ return docstring();
if (!opt.ignoreformat)
return latexifyFromCursor(cur, len);
else
return stringifyFromCursor(cur, len);
}
diff --git a/src/output_latex.cpp b/src/output_latex.cpp
index f73990d..96f51df 100644
--- a/src/output_latex.cpp
+++ b/src/output_latex.cpp
@@ -812,14 +812,16 @@ void TeXOnePar(Buffer const & buf,
&& priorpar->layout().isEnvironment()
&& (priorpar->getDepth() > par.getDepth()
|| (priorpar->getDepth() == par.getDepth()
&& priorpar->layout() != par.layout()));
Language const * const prev_language =
- (priorpar && !priorpar->isPassThru())
- ? (use_prev_env_language ? state->prev_env_language_
- : priorpar->getParLanguage(bparams))
- : outer_language;
+ runparams_in.for_search ?
+ languages.getLanguage("ignore")
+ :(priorpar && !priorpar->isPassThru())
+ ? (use_prev_env_language ? state->prev_env_language_
+ : priorpar->getParLanguage(bparams))
+ : outer_language;
bool const use_polyglossia = runparams.use_polyglossia;
string const par_lang = use_polyglossia ?
getPolyglossiaEnvName(par_language): par_language->babel();
string const prev_lang = use_polyglossia ?
@@ -852,11 +854,12 @@ void TeXOnePar(Buffer const & buf,
bool const in_polyglossia_rtl_env =
use_polyglossia
&& runparams.local_font != 0
&& outer_language->rightToLeft()
&& !par_language->rightToLeft();
- bool const localswitch = text.inset().forceLocalFontSwitch()
+ bool const localswitch = runparams_in.for_search
+ || text.inset().forceLocalFontSwitch()
|| (using_begin_end && text.inset().forcePlainLayout())
|| in_polyglossia_rtl_env;
if (localswitch) {
lang_begin_command = use_polyglossia ?
"\\text$$lang$$opts{" : lyxrc.language_command_local;
signature.asc
Description: This is a digitally signed message part.
