Font const & font,
Layout const & style);
- /// Output consecutive unicode chars, belonging to the same script as
- /// specified by the latex macro \p ltx, to \p os starting from \p i.
- /// \return the number of characters written.
- int writeScriptChars(BufferParams const &, OutputParams const &,
- otexstream & os,
- docstring const & ltx,
- Change const &, Encoding const &,
- std::string const, pos_type & i);
-
/// This could go to ParagraphParameters if we want to.
int startTeXParParams(BufferParams const &, otexstream &,
OutputParams const &) const;
BufferParams const & bparams,
OutputParams const & runparams,
Font const & running_font,
- Change const & running_change,
+ string & alien_script,
Layout const & style,
pos_type & i,
pos_type end_pos,
return end - i;
}
-
+// Handle combining characters
int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
otexstream & os, char_type c, char_type next,
OutputParams const & runparams)
}
docstring latex2 = encoding.latexChar(c).first;
- if (docstring(1, next) == latex1) {
- // The encoding supports the combination:
+ if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
+ // Encoding supports the combination:
// output as is (combining char after base char).
os << latex2 << latex1;
return latex1.length() + latex2.length();
}
- // Handle combining characters in "script" context (i.e., \textgreek and \textcyrillic)
- docstring::size_type const brace1 = latex2.find_first_of(from_ascii("{"));
- docstring::size_type const brace2 = latex2.find_last_of(from_ascii("}"));
- string script = to_ascii(latex2.substr(1, brace1 - 1));
- // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding
- // whether they are encodable or not (it only depends on the font encoding)
- if (!runparams.isFullUnicode())
- // This will get us a script value to deal with below
- Encodings::isKnownScriptChar(c, script);
- int pos = 0;
- int length = brace2;
- string fontenc;
- if (runparams.local_font)
- fontenc = runparams.local_font->language()->fontenc(bparams);
- else
- fontenc = runparams.main_fontenc;
- docstring scriptmacro;
- docstring cb;
- if (script == "textgreek" || script == "textcyrillic") {
- // We separate the script macro (\text[greek|cyr]) from the rest,
- // since we need to include the combining char in it (#6463).
- // This is "the rest":
- pos = brace1 + 1;
- length -= pos;
- latex2 = latex2.substr(pos, length);
- // We only need the script macro with non-native font encodings
- if (Encodings::needsScriptWrapper(script, fontenc)) {
- scriptmacro = from_ascii("\\" + script + "{");
- cb = from_ascii("}");
- }
- }
-
- docstring lb;
- docstring rb;
- // polutonikogreek does not play nice with brackets
- if (!runparams.local_font
- || runparams.local_font->language()->lang() != "polutonikogreek") {
- lb = from_ascii("{");
- rb = from_ascii("}");
- }
-
- os << scriptmacro << latex1 << lb << latex2 << rb << cb;
- return latex1.length() + latex2.length() + lb.length() + rb.length() + cb.length();
+ os << latex1 << "{" << latex2 << "}";
+ return latex1.length() + latex2.length() + 2;
}
}
-int Paragraph::Private::writeScriptChars(BufferParams const & bparams,
- OutputParams const & runparams,
- otexstream & os,
- docstring const & ltx,
- Change const & runningChange,
- Encoding const & encoding,
- string const fontenc,
- pos_type & i)
-{
- // FIXME: modifying i here is not very nice...
-
- // We only arrive here when character text_[i] could not be translated
- // into the current latex encoding (or its latex translation has been forced,)
- // and it belongs to a known script.
- // Parameter ltx contains the latex translation of text_[i] as specified
- // in the unicodesymbols file and is something like "\textXXX{<spec>}".
- // The latex macro name "textXXX" specifies the script to which text_[i]
- // belongs and we use it in order to check whether characters from the
- // same script immediately follow, such that we can collect them in a
- // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
- // for the first char but only "<spec>" for all subsequent chars.
- docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
- docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
- string script = to_ascii(ltx.substr(1, brace1 - 1));
- int pos = 0;
- int length = brace2;
- bool closing_brace = true;
- if (!Encodings::needsScriptWrapper(script, fontenc)) {
- // Correct font encoding is being used, so we can avoid \text[greek|cyr].
- pos = brace1 + 1;
- length -= pos;
- closing_brace = false;
- }
- os << ltx.substr(pos, length);
- int size = text_.size();
- while (i + 1 < size) {
- char_type const next = text_[i + 1];
- // Stop here if next character belongs to another script
- // or there is a change in change tracking status.
- if (!Encodings::isKnownScriptChar(next, script) ||
- runningChange != owner_->lookupChange(i + 1))
- break;
- Font prev_font;
- bool found = false;
- FontList::const_iterator cit = fontlist_.begin();
- FontList::const_iterator end = fontlist_.end();
- for (; cit != end; ++cit) {
- if (cit->pos() >= i && !found) {
- prev_font = cit->font();
- found = true;
- }
- if (cit->pos() >= i + 1)
- break;
- }
- // Stop here if there is a font attribute or encoding change.
- if (found && cit != end && prev_font != cit->font())
- break;
-
- // Check whether we have a combining pair
- char_type next_next = '\0';
- if (i + 2 < size) {
- next_next = text_[i + 2];
- if (Encodings::isCombiningChar(next_next)) {
- length += latexSurrogatePair(bparams, os, next, next_next, runparams) - 1;
- i += 2;
- continue;
- }
- }
-
- docstring const latex = encoding.latexChar(next).first;
- docstring::size_type const b1 =
- latex.find_first_of(from_ascii("{"));
- docstring::size_type const b2 =
- latex.find_last_of(from_ascii("}"));
- int const len = b2 - b1 - 1;
- os << latex.substr(b1 + 1, len);
- length += len;
- ++i;
- }
- if (closing_brace) {
- os << '}';
- ++length;
- }
- return length;
-}
-
-
void Paragraph::Private::latexInset(BufferParams const & bparams,
otexstream & os,
OutputParams & runparams,
odocstream::pos_type const len = os.os().tellp();
if (inset->forceLTR()
- && !runparams.use_polyglossia
&& running_font.isRightToLeft()
// ERT is an exception, it should be output with no
// decorations at all
&& inset->lyxCode() != ERT_CODE) {
- if (running_font.language()->lang() == "farsi")
- os << "\\beginL" << termcmd;
+ if (runparams.use_polyglossia) {
+ os << "\\LRE{";
+ } else if (running_font.language()->lang() == "farsi"
+ || running_font.language()->lang() == "arabic_arabi")
+ os << "\\textLR{" << termcmd;
else
os << "\\L{";
close = true;
throw(e);
}
- if (close) {
- if (running_font.language()->lang() == "farsi")
- os << "\\endL" << termcmd;
- else
- os << '}';
- }
+ if (close)
+ os << '}';
if (os.texrow().rows() > previous_row_count) {
os.texrow().start(owner_->id(), i + 1);
BufferParams const & bparams,
OutputParams const & runparams,
Font const & running_font,
- Change const & running_change,
+ string & alien_script,
Layout const & style,
pos_type & i,
pos_type end_pos,
unsigned int & column)
{
- // With polyglossia, brackets and stuff need not be reversed
- // in RTL scripts (see bug #8251)
- char_type const c = (runparams.use_polyglossia) ?
- owner_->getUChar(bparams, i) : text_[i];
+ char_type const c = owner_->getUChar(bparams, runparams, i);
if (style.pass_thru || runparams.pass_thru
|| contains(style.pass_thru_chars, c)
&& !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
&& latexSpecialT1(c, os, i, column))
return;
- // NOTE: XeTeX and LuaTeX use EU1/2 (pre 2017) or TU (as of 2017) encoding
+ // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
&& runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
return;
break;
}
}
- string script;
pair<docstring, bool> latex = encoding.latexChar(c);
docstring nextlatex;
bool nexttipas = false;
string nexttipashortcut;
- if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
+ if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
nextlatex = encoding.latexChar(next).first;
if (runparams.inIPA) {
nexttipashortcut = Encodings::TIPAShortcut(next);
tipas = true;
}
}
- string fontenc;
- if (running_font.language()->lang() == bparams.language->lang())
- fontenc = runparams.main_fontenc;
- else
- fontenc = running_font.language()->fontenc(bparams);
- // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding
- // whether they are encodable or not (it only depends on the font encoding)
- if (!runparams.isFullUnicode() && Encodings::isKnownScriptChar(c, script)) {
- docstring const wrapper = from_ascii("\\" + script + "{");
- docstring ltx = latex.first;
- if (!prefixIs(ltx, wrapper))
- ltx = wrapper + latex.first + from_ascii("}");
- column += writeScriptChars(bparams, runparams, os, ltx, running_change,
- encoding, fontenc, i) - 1;
- } else if (latex.second
+ // eventually close "script wrapper" command (see `Paragraph::latex`)
+ if (!alien_script.empty()
+ && alien_script != Encodings::isKnownScriptChar(next)) {
+ column += latex.first.length();
+ alien_script.clear();
+ os << latex.first << "}";
+ break;
+ }
+ if (latex.second
&& ((!prefixIs(nextlatex, '\\')
&& !prefixIs(nextlatex, '{')
&& !prefixIs(nextlatex, '}'))
|| (nexttipas
&& !prefixIs(from_ascii(nexttipashortcut), '\\')))
&& !tipas) {
- // Prevent eating of a following
- // space or command corruption by
+ // Prevent eating of a following space or command corruption by
// following characters
if (next == ' ' || next == '\0') {
column += latex.first.length() + 1;
// then the contents
BufferParams const bp = features.runparams().is_child
? features.buffer().masterParams() : features.buffer().params();
- string bscript = "textbaltic";
for (pos_type i = 0; i < int(text_.size()) ; ++i) {
char_type c = text_[i];
+ CharInfo const & ci = Encodings::unicodeCharInfo(c);
if (c == 0x0022) {
if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
features.require("textquotedblp");
else if (bp.main_font_encoding() != "T1"
|| ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
features.require("textquotedbl");
- } else if (Encodings::isKnownScriptChar(c, bscript)){
+ } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
+ // features that depend on the font or input encoding
+ string feats = ci.textpreamble();
string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
if (fontenc.empty())
fontenc = features.runparams().main_fontenc;
- if (Encodings::needsScriptWrapper("textbaltic", fontenc))
- features.require("textbalticdefs");
+ while (!feats.empty()) {
+ string feat;
+ feats = split(feats, feat, ',');
+ if (contains(feat, "!=")) {
+ // a feature that is required except for the spcified
+ // font or input encodings
+ string realfeature;
+ string const contexts = ltrim(split(feat, realfeature, '!'), "=");
+ // multiple encodings are separated by semicolon
+ vector<string> context = getVectorFromString(contexts, ";");
+ // require feature if the context matches neither current font
+ // nor input encoding
+ if (std::find(context.begin(), context.end(), fontenc) == context.end()
+ && std::find(context.begin(), context.end(),
+ features.runparams().encoding->name()) == context.end())
+ features.require(realfeature);
+ } else if (contains(feat, '=')) {
+ // a feature that is required only for the spcified
+ // font or input encodings
+ string realfeature;
+ string const contexts = split(feat, realfeature, '=');
+ // multiple encodings are separated by semicolon
+ vector<string> context = getVectorFromString(contexts, ";");
+ // require feature if the context matches either current font
+ // or input encoding
+ if (std::find(context.begin(), context.end(), fontenc) != context.end()
+ || std::find(context.begin(), context.end(),
+ features.runparams().encoding->name()) != context.end())
+ features.require(realfeature);
+ }
+ }
} else if (!bp.use_dash_ligatures
&& (c == 0x2013 || c == 0x2014)
&& bp.useNonTeXFonts
void Paragraph::validate(LaTeXFeatures & features) const
{
d->validate(features);
- if (needsCProtection())
+ bool fragile = features.runparams().moving_arg;
+ fragile |= layout().needprotect;
+ if (needsCProtection(fragile))
features.require("cprotect");
}
}
-char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
+char_type Paragraph::getUChar(BufferParams const & bparams,
+ OutputParams const & rp,
+ pos_type pos) const
{
char_type c = d->text_[pos];
- if (!getFontSettings(bparams, pos).isRightToLeft())
+
+ // Return unchanged character in LTR languages
+ // or if we use poylglossia/bidi.
+ if (rp.use_polyglossia || !getFontSettings(bparams, pos).isRightToLeft())
return c;
- // FIXME: The arabic special casing is due to the difference of arabic
- // round brackets input introduced in r18599. Check if this should be
- // unified with Hebrew or at least if all bracket types should be
- // handled the same (file format change in either case).
+ // Without polyglossia/bidi, we need to account for some special cases.
+ // FIXME This needs to be audited!
+ // Check if:
+ // * The input is as expected for all delimiters
+ // => checked for Hebrew!
+ // * The output matches the display in the LyX workarea
+ // => checked for Hebrew!
+ // * The special cases below are really necessary
+ // => checked for Hebrew!
+ // * In arabic_arabi, brackets are transformed to Arabic
+ // Ornate Parentheses. Is this is really wanted?
+
string const & lang = getFontSettings(bparams, pos).language()->lang();
- bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
- || lang == "farsi";
char_type uc = c;
+
+ // 1. In the following languages, parentheses need to be reversed.
+ bool const reverseparens = lang == "hebrew";
+
+ // 2. In the following languages, brackets don't need to be reversed.
+ bool const reversebrackets = lang != "arabic_arabtex"
+ && lang != "arabic_arabi"
+ && lang != "farsi";
+
+ // Now swap delimiters if needed.
switch (c) {
case '(':
- uc = arabic ? c : ')';
+ if (reverseparens)
+ uc = ')';
break;
case ')':
- uc = arabic ? c : '(';
+ if (reverseparens)
+ uc = '(';
break;
case '[':
- uc = ']';
+ if (reversebrackets)
+ uc = ']';
break;
case ']':
- uc = '[';
+ if (reversebrackets)
+ uc = '[';
break;
case '{':
uc = '}';
string const begin_tag = "\\begin";
InsetCode code = ownerCode();
bool const lastpar = runparams.isLastPar;
+ // RTL in classic (PDF)LaTeX (without the Bidi package)
+ bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
+ && !runparams.use_polyglossia;
switch (curAlign) {
case LYX_ALIGN_NONE:
case LYX_ALIGN_DECIMAL:
break;
case LYX_ALIGN_LEFT: {
- if (owner_->getParLanguage(bparams)->babel() != "hebrew")
- corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
- else
+ if (rtl_classic)
+ // Classic (PDF)LaTeX switches the left/right logic in RTL mode
corrected_env(os, begin_tag, "flushright", code, lastpar, column);
+ else
+ corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
break;
} case LYX_ALIGN_RIGHT: {
- if (owner_->getParLanguage(bparams)->babel() != "hebrew")
- corrected_env(os, begin_tag, "flushright", code, lastpar, column);
- else
+ if (rtl_classic)
+ // Classic (PDF)LaTeX switches the left/right logic in RTL mode
corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
+ else
+ corrected_env(os, begin_tag, "flushright", code, lastpar, column);
break;
} case LYX_ALIGN_CENTER: {
corrected_env(os, begin_tag, "center", code, lastpar, column);
string const end_tag = "\\par\\end";
InsetCode code = ownerCode();
bool const lastpar = runparams.isLastPar;
+ // RTL in classic (PDF)LaTeX (without the Bidi package)
+ bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
+ && !runparams.use_polyglossia;
switch (curAlign) {
case LYX_ALIGN_NONE:
case LYX_ALIGN_DECIMAL:
break;
case LYX_ALIGN_LEFT: {
- if (owner_->getParLanguage(bparams)->babel() != "hebrew")
- output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
- else
+ if (rtl_classic)
+ // Classic (PDF)LaTeX switches the left/right logic in RTL mode
output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
+ else
+ output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
break;
} case LYX_ALIGN_RIGHT: {
- if (owner_->getParLanguage(bparams)->babel() != "hebrew")
- output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
- else
+ if (rtl_classic)
+ // Classic (PDF)LaTeX switches the left/right logic in RTL mode
output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
+ else
+ output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
break;
} case LYX_ALIGN_CENTER: {
corrected_env(os, end_tag, "center", code, lastpar, col);
// of the body.
Font basefont;
+ // If there is an open font-encoding changing command (script wrapper),
+ // alien_script is set to its name
+ string alien_script;
+ string script;
+
// Maybe we have to create a optional argument.
pos_type body_pos = beginOfBody();
unsigned int column = 0;
}
if (bparams.output_changes && runningChange != change) {
+ if (!alien_script.empty()) {
+ column += 1;
+ os << "}";
+ alien_script.clear();
+ }
if (open_font) {
bool needPar = false;
column += running_font.latexWriteEndChanges(
(current_font != running_font ||
current_font.language() != running_font.language()))
{
+ // ensure there is no open script-wrapper
+ if (!alien_script.empty()) {
+ column += 1;
+ os << "}";
+ alien_script.clear();
+ }
bool needPar = false;
column += running_font.latexWriteEndChanges(
os, bparams, runparams, basefont,
--parInline;
}
} else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
+ if (!bparams.useNonTeXFonts)
+ script = Encodings::isKnownScriptChar(c);
+ if (script != alien_script) {
+ if (!alien_script.empty()) {
+ os << "}";
+ alien_script.clear();
+ }
+ string fontenc = running_font.language()->fontenc(bparams);
+ if (!script.empty()
+ && !Encodings::fontencSupportsScript(fontenc, script)) {
+ column += script.length() + 2;
+ os << "\\" << script << "{";
+ alien_script = script;
+ }
+ }
try {
- d->latexSpecialChar(os, bparams, rp,
- running_font, runningChange,
- style, i, end_pos, column);
+ d->latexSpecialChar(os, bparams, rp, running_font,
+ alien_script, style, i, end_pos, column);
} catch (EncodingException & e) {
if (runparams.dryrun) {
os << "<" << _("LyX Warning: ")
// such as Note that do not produce any output, so that no
// command is ever executed but its opening was recorded.
runparams.inulemcmd = rp.inulemcmd;
+
+ // And finally, pass the post_macros upstream
+ runparams.post_macro = rp.post_macro;
}
+ // Close wrapper for alien script
+ if (!alien_script.empty()) {
+ os << "}";
+ alien_script.clear();
+ }
+
// If we have an open font definition, we have to close it
if (open_font) {
// Make sure that \\par is done with the font of the last
retval += inset->xhtml(xs, np);
}
} else {
- char_type c = getUChar(buf.masterBuffer()->params(), i);
- xs << c;
+ char_type c = getUChar(buf.masterBuffer()->params(),
+ runparams, i);
+ if (c == ' ' && (style.free_spacing || runparams.free_spacing))
+ xs << XHTMLStream::ESCAPE_NONE << " ";
+ else
+ xs << c;
}
font_old = font.fontInfo();
}
}
-bool Paragraph::needsCProtection() const
+bool Paragraph::needsCProtection(bool const fragile) const
{
// first check the layout of the paragraph, but only in insets
InsetText const * textinset = inInset().asInsetText();
// now check whether we have insets that need cprotection
pos_type size = d->text_.size();
for (pos_type i = 0; i < size; ++i)
- if (isInset(i) && getInset(i)->needsCProtection())
+ if (isInset(i) && getInset(i)->needsCProtection(maintext, fragile))
return true;
return false;
docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
Language * lang = d->getSpellLanguage(from);
+ if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
+ return result;
+
wl = WordLangTuple(word, lang);
if (word.empty())