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:
+ // 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));
+
+ // Greek and Cyrillic letters need to be wrapped in \textcyrillic and \textgreek if they
+ // are not encodable in the current font encoding (regardless of the input encoding).
+ bool scriptchar = false;
+ if (!bparams.useNonTeXFonts) // With non-TeX fonts the font encoding is Unicode.
+ scriptchar = Encodings::isKnownScriptChar(c, script);
+
+ if (!scriptchar && docstring(1, next) == latex1) {
+ // Font and input encoding support 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;
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":
+ // Strip the \text(greek|cyrillic) script macro ...
pos = brace1 + 1;
length -= pos;
latex2 = latex2.substr(pos, length);
- // We only need the script macro with non-native font encodings
+ // and place it before the accent macro if required (#6463)
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 << scriptmacro << latex1 << "{" << latex2 << "}" << cb;
+ return latex1.length() + 1 + latex2.length() + 1 + cb.length();
}
int pos = 0;
int length = brace2;
bool closing_brace = true;
+ // We only need the script macro with non-native font encodings
if (!Encodings::needsScriptWrapper(script, fontenc)) {
- // Correct font encoding is being used, so we can avoid \text[greek|cyr].
+ // Correct font encoding is being used, so we can avoid \text(greek|cyrrillic).
pos = brace1 + 1;
length -= pos;
closing_brace = false;
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);
string fontenc;
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)) {
+ // whether they are encodable or not (it only depends on the font encoding),
+ // except if we are using fontspec.
+ if (!bparams.useNonTeXFonts && Encodings::isKnownScriptChar(c, script)) {
docstring const wrapper = from_ascii("\\" + script + "{");
docstring ltx = latex.first;
if (!prefixIs(ltx, wrapper))
{
char_type c = d->text_[pos];
- // Return unchanged character in LTR languages.
- 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 This is a complete mess due to all the language-specific
- // special cases. We need to unify this eventually, but this
- // requires a file format change and some thought.
- // We also need to unify the input of parentheses in different RTL
- // languages. Currently, some have their own methods (Arabic:
- // 18599/lyxsvn, Hebrew: e5f42f67d/lyxgit), some don't (Urdu, Syriac).
- // Also note that the representation in the LyX file is probably wrong
- // (see FIXME in TextMetrics::breakRow).
- // Most likely, we should simply rely on Qt's unicode handling here.
- string const & lang = getFontSettings(bparams, pos).language()->lang();
+ // 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?
- // With polyglossia, brackets and stuff need not be reversed in RTL scripts
- // FIXME: The special casing for Hebrew parens is due to the special
- // handling on input (for Hebrew in e5f42f67d/lyxgit); see #8251.
+ string const & lang = getFontSettings(bparams, pos).language()->lang();
char_type uc = c;
- if (rp.use_polyglossia) {
- switch (c) {
- case '(':
- if (lang == "hebrew")
- uc = ')';
- break;
- case ')':
- if (lang == "hebrew")
- uc = '(';
- break;
- }
- return uc;
- }
- // In the following languages, brackets don't need to be reversed.
- // Furthermore, in arabic_arabi, they are transformed to Arabic
- // Ornate Parentheses (dunno if this is really wanted)
+ // 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";
+ && lang != "farsi";
+ // Now swap delimiters if needed.
switch (c) {
+ case '(':
+ if (reverseparens)
+ uc = ')';
+ break;
+ case ')':
+ if (reverseparens)
+ uc = '(';
+ break;
case '[':
if (reversebrackets)
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);
} else {
char_type c = getUChar(buf.masterBuffer()->params(),
runparams, i);
- xs << c;
+ if (c == ' ' && (style.free_spacing || runparams.free_spacing))
+ xs << XHTMLStream::ESCAPE_NONE << " ";
+ else
+ xs << c;
}
font_old = font.fontInfo();
}