]> git.lyx.org Git - lyx.git/blobdiff - src/Paragraph.cpp
This should do it for the delimiters.
[lyx.git] / src / Paragraph.cpp
index 61ca78ae1364bca4afcdfad468aa1f5990ad2858..0266c2e6ff9cf19a5b7875a12ebee99a7fceed7e 100644 (file)
@@ -4,13 +4,14 @@
  * Licence details can be found in the file COPYING.
  *
  * \author Asger Alstrup
- * \author Lars Gullik Bjønnes
+ * \author Lars Gullik Bjønnes
+ * \author Richard Heck (XHTML output)
  * \author Jean-Marc Lasgouttes
  * \author Angus Leeming
  * \author John Levon
- * \author André Pönitz
+ * \author André Pönitz
  * \author Dekel Tsur
- * \author Jürgen Vigna
+ * \author Jürgen Vigna
  *
  * Full author contact details are available in file CREDITS.
  */
 #include "LyXRC.h"
 #include "OutputParams.h"
 #include "output_latex.h"
-#include "paragraph_funcs.h"
+#include "output_xhtml.h"
 #include "ParagraphParameters.h"
+#include "SpellChecker.h"
 #include "sgml.h"
 #include "TextClass.h"
 #include "TexRow.h"
 #include "Text.h"
 #include "VSpace.h"
+#include "WordLangTuple.h"
 #include "WordList.h"
 
 #include "frontends/alert.h"
 #include "insets/InsetBibitem.h"
 #include "insets/InsetLabel.h"
 
-#include "support/lassert.h"
-#include "support/convert.h"
 #include "support/debug.h"
+#include "support/docstring_list.h"
 #include "support/ExceptionMessage.h"
 #include "support/gettext.h"
+#include "support/lassert.h"
 #include "support/lstrings.h"
-#include "support/Messages.h"
 #include "support/textutils.h"
 
 #include <sstream>
@@ -93,7 +95,7 @@ public:
        /// Output the surrogate pair formed by \p c and \p next to \p os.
        /// \return the number of characters written.
        int latexSurrogatePair(odocstream & os, char_type c, char_type next,
-                              Encoding const &);
+                              OutputParams const &);
 
        /// Output a space in appropriate formatting (or a surrogate pair
        /// if the next character is a combining character).
@@ -162,8 +164,7 @@ public:
                OutputParams const & runparams);
 
        ///
-       void validate(LaTeXFeatures & features,
-                     Layout const & layout) const;
+       void validate(LaTeXFeatures & features) const;
 
        /// Checks if the paragraph contains only text and no inset or font change.
        bool onlyText(Buffer const & buf, Font const & outerfont,
@@ -171,6 +172,12 @@ public:
 
        /// match a string against a particular point in the paragraph
        bool isTextAt(string const & str, pos_type pos) const;
+
+
+       InsetCode ownerCode() const
+       {
+               return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
+       }
        
        /// Which Paragraph owns us?
        Paragraph * owner_;
@@ -286,6 +293,15 @@ void Paragraph::addChangesToToc(DocIterator const & cdit,
 }
 
 
+bool Paragraph::isDeleted(pos_type start, pos_type end) const
+{
+       LASSERT(start >= 0 && start <= size(), /**/);
+       LASSERT(end > start && end <= size() + 1, /**/);
+
+       return d->changes_.isDeleted(start, end);
+}
+
+
 bool Paragraph::isChanged(pos_type start, pos_type end) const
 {
        LASSERT(start >= 0 && start <= size(), /**/);
@@ -302,7 +318,7 @@ bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
                return true;
 
        Change const change = d->changes_.lookup(size());
-       return change.type == Change::INSERTED && change.author == 0;
+       return change.inserted() && change.currentAuthor();
 }
 
 
@@ -323,7 +339,7 @@ void Paragraph::setChange(Change const & change)
         * Conclusion: An inset's content should remain untouched if you delete it
         */
 
-       if (change.type != Change::DELETED) {
+       if (!change.deleted()) {
                for (pos_type pos = 0; pos < size(); ++pos) {
                        if (Inset * inset = getInset(pos))
                                inset->setChange(change);
@@ -338,7 +354,7 @@ void Paragraph::setChange(pos_type pos, Change const & change)
        d->changes_.set(change, pos);
 
        // see comment in setChange(Change const &) above
-       if (change.type != Change::DELETED && pos < size())
+       if (!change.deleted() && pos < size())
                        if (Inset * inset = getInset(pos))
                                inset->setChange(change);
 }
@@ -351,8 +367,7 @@ Change const & Paragraph::lookupChange(pos_type pos) const
 }
 
 
-void Paragraph::acceptChanges(BufferParams const & bparams, pos_type start,
-               pos_type end)
+void Paragraph::acceptChanges(pos_type start, pos_type end)
 {
        LASSERT(start >= 0 && start <= size(), /**/);
        LASSERT(end > start && end <= size() + 1, /**/);
@@ -362,14 +377,14 @@ void Paragraph::acceptChanges(BufferParams const & bparams, pos_type start,
                        case Change::UNCHANGED:
                                // accept changes in nested inset
                                if (Inset * inset = getInset(pos))
-                                       inset->acceptChanges(bparams);
+                                       inset->acceptChanges();
                                break;
 
                        case Change::INSERTED:
                                d->changes_.set(Change(Change::UNCHANGED), pos);
                                // also accept changes in nested inset
                                if (Inset * inset = getInset(pos))
-                                       inset->acceptChanges(bparams);
+                                       inset->acceptChanges();
                                break;
 
                        case Change::DELETED:
@@ -387,8 +402,7 @@ void Paragraph::acceptChanges(BufferParams const & bparams, pos_type start,
 }
 
 
-void Paragraph::rejectChanges(BufferParams const & bparams,
-               pos_type start, pos_type end)
+void Paragraph::rejectChanges(pos_type start, pos_type end)
 {
        LASSERT(start >= 0 && start <= size(), /**/);
        LASSERT(end > start && end <= size() + 1, /**/);
@@ -398,7 +412,7 @@ void Paragraph::rejectChanges(BufferParams const & bparams,
                        case Change::UNCHANGED:
                                // reject changes in nested inset
                                if (Inset * inset = getInset(pos))
-                                               inset->rejectChanges(bparams);
+                                               inset->rejectChanges();
                                break;
 
                        case Change::INSERTED:
@@ -482,13 +496,13 @@ bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
                //  a) it was previously unchanged or
                //  b) it was inserted by a co-author
 
-               if (change.type == Change::UNCHANGED ||
-                   (change.type == Change::INSERTED && change.author != 0)) {
+               if (!change.changed() ||
+                     (change.inserted() && !change.currentAuthor())) {
                        setChange(pos, Change(Change::DELETED));
                        return false;
                }
 
-               if (change.type == Change::DELETED)
+               if (change.deleted())
                        return false;
        }
 
@@ -533,7 +547,7 @@ int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
 
 
 int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
-               char_type next, Encoding const & encoding)
+               char_type next, OutputParams const & runparams)
 {
        // Writing next here may circumvent a possible font change between
        // c and next. Since next is only output if it forms a surrogate pair
@@ -542,12 +556,18 @@ int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
        // hopefully impossible to input.
        // FIXME: change tracking
        // Is this correct WRT change tracking?
+       Encoding const & encoding = *(runparams.encoding);
        docstring const latex1 = encoding.latexChar(next);
        docstring const latex2 = encoding.latexChar(c);
        if (docstring(1, next) == latex1) {
                // the encoding supports the combination
                os << latex2 << latex1;
                return latex1.length() + latex2.length();
+       } else if (runparams.local_font &&
+                  runparams.local_font->language()->lang() == "polutonikogreek") {
+               // polutonikogreek only works without the brackets
+               os << latex1 << latex2;
+               return latex1.length() + latex2.length();
        } else
                os << latex1 << '{' << latex2 << '}';
        return latex1.length() + latex2.length() + 2;
@@ -567,15 +587,14 @@ bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
        if (i + 1 < int(text_.size())) {
                char_type next = text_[i + 1];
                if (Encodings::isCombiningChar(next)) {
-                       Encoding const & encoding = *(runparams.encoding);
                        // This space has an accent, so we must always output it.
-                       column += latexSurrogatePair(os, ' ', next, encoding) - 1;
+                       column += latexSurrogatePair(os, ' ', next, runparams) - 1;
                        return true;
                }
        }
 
-       if (lyxrc.plaintext_linelen > 0
-           && column > lyxrc.plaintext_linelen
+       if (runparams.linelen > 0
+           && column > runparams.linelen
            && i
            && text_[i - 1] != ' '
            && (i + 1 < int(text_.size()))
@@ -745,24 +764,25 @@ void Paragraph::Private::latexInset(
                column = 0;
        }
 
-       if (owner_->lookupChange(i).type == Change::DELETED) {
+       if (owner_->isDeleted(i)) {
                if( ++runparams.inDeletedInset == 1)
                        runparams.changeOfDeletedInset = owner_->lookupChange(i);
        }
 
        if (inset->canTrackChanges()) {
                column += Changes::latexMarkChange(os, bparams, running_change,
-                       Change(Change::UNCHANGED));
+                       Change(Change::UNCHANGED), runparams);
                running_change = Change(Change::UNCHANGED);
        }
 
        bool close = false;
        odocstream::pos_type const len = os.tellp();
 
-       if (inset->forceLTR() 
+       if (inset->forceLTR()
            && running_font.isRightToLeft()
-               // ERT is an exception, it should be output with no decorations at all
-               && inset->lyxCode() != ERT_CODE) {
+           // 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{}";
                else
@@ -817,10 +837,10 @@ void Paragraph::Private::latexInset(
        }
 
        if (close) {
-       if (running_font.language()->lang() == "farsi")
-                       os << "\\endL{}";
-               else
-                       os << '}';
+               if (running_font.language()->lang() == "farsi")
+                               os << "\\endL{}";
+                       else
+                               os << '}';
        }
 
        if (tmp) {
@@ -833,7 +853,7 @@ void Paragraph::Private::latexInset(
                column += os.tellp() - len;
        }
 
-       if (owner_->lookupChange(i).type == Change::DELETED)
+       if (owner_->isDeleted(i))
                --runparams.inDeletedInset;
 }
 
@@ -864,9 +884,15 @@ void Paragraph::Private::latexSpecialChar(
                return;
        }
 
-       if (lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
+       // If T1 font encoding is used, use the special
+       // characters it provides.
+       // NOTE: some languages reset the font encoding
+       // internally
+       if (!running_font.language()->internalFontEncoding()
+           && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
                return;
 
+       // \tt font needs special treatment
        if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
                && latexSpecialTypewriter(c, os, i, column))
                return;
@@ -915,7 +941,9 @@ void Paragraph::Private::latexSpecialChar(
                column += 17;
                break;
 
-       case '*': case '[':
+       case '*':
+       case '[':
+       case ']':
                // avoid being mistaken for optional arguments
                os << '{';
                os.put(c);
@@ -943,7 +971,7 @@ void Paragraph::Private::latexSpecialChar(
                if (i + 1 < int(text_.size())) {
                        char_type next = text_[i + 1];
                        if (Encodings::isCombiningChar(next)) {
-                               column += latexSurrogatePair(os, c, next, encoding) - 1;
+                               column += latexSurrogatePair(os, c, next, runparams) - 1;
                                ++i;
                                break;
                        }
@@ -1042,15 +1070,26 @@ bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
 }
 
 
-void Paragraph::Private::validate(LaTeXFeatures & features,
-                               Layout const & layout) const
+void Paragraph::Private::validate(LaTeXFeatures & features) const
 {
+       if (layout_->inpreamble && inset_owner_) {
+               Buffer const & buf = inset_owner_->buffer();
+               BufferParams const & bp = buf.params();
+               Font f;
+               TexRow tr;
+               odocstringstream ods;
+               owner_->latex(bp, f, ods, tr, features.runparams());
+               docstring d = ods.str();
+               if (!d.empty())
+                       features.addPreambleSnippet(to_utf8(d));
+       }
+       
        // check the params.
        if (!params_.spacing().isDefault())
                features.require("setspace");
 
        // then the layouts
-       features.useLayout(layout.name());
+       features.useLayout(layout_->name());
 
        // then the fonts
        fontlist_.validate(features);
@@ -1065,7 +1104,7 @@ void Paragraph::Private::validate(LaTeXFeatures & features,
        for (; icit != iend; ++icit) {
                if (icit->inset) {
                        icit->inset->validate(features);
-                       if (layout.needprotect &&
+                       if (layout_->needprotect &&
                            icit->inset->lyxCode() == FOOT_CODE)
                                features.require("NeedLyXFootnoteCode");
                }
@@ -1140,6 +1179,18 @@ Paragraph::~Paragraph()
 }
 
 
+namespace {
+
+// this shall be called just before every "os << ..." action.
+void flushString(ostream & os, docstring & s)
+{
+       os << to_utf8(s);
+       s.erase();
+}
+
+}
+
+
 void Paragraph::write(ostream & os, BufferParams const & bparams,
        depth_type & dth) const
 {
@@ -1167,19 +1218,27 @@ void Paragraph::write(ostream & os, BufferParams const & bparams,
 
        Change running_change = Change(Change::UNCHANGED);
 
+       // this string is used as a buffer to avoid repetitive calls
+       // to to_utf8(), which turn out to be expensive (JMarc)
+       docstring write_buffer;
+
        int column = 0;
        for (pos_type i = 0; i <= size(); ++i) {
 
-               Change change = lookupChange(i);
-               Changes::lyxMarkChange(os, column, running_change, change);
+               Change const change = lookupChange(i);
+               if (change != running_change)
+                       flushString(os, write_buffer);
+               Changes::lyxMarkChange(os, bparams, column, running_change, change);
                running_change = change;
 
                if (i == size())
                        break;
 
-               // Write font changes
+               // Write font changes (ignore spelling markers)
                Font font2 = getFontSettings(bparams, i);
+               font2.setMisspelled(false);
                if (font2 != font1) {
+                       flushString(os, write_buffer);
                        font2.lyxWriteChanges(font1, os);
                        column = 0;
                        font1 = font2;
@@ -1189,6 +1248,7 @@ void Paragraph::write(ostream & os, BufferParams const & bparams,
                switch (c) {
                case META_INSET:
                        if (Inset const * inset = getInset(i)) {
+                               flushString(os, write_buffer);
                                if (inset->directWrite()) {
                                        // international char, let it write
                                        // code directly so it's shorter in
@@ -1205,10 +1265,12 @@ void Paragraph::write(ostream & os, BufferParams const & bparams,
                        }
                        break;
                case '\\':
+                       flushString(os, write_buffer);
                        os << "\n\\backslash\n";
                        column = 0;
                        break;
                case '.':
+                       flushString(os, write_buffer);
                        if (i + 1 < size() && d->text_[i + 1] == ' ') {
                                os << ".\n";
                                column = 0;
@@ -1218,13 +1280,14 @@ void Paragraph::write(ostream & os, BufferParams const & bparams,
                default:
                        if ((column > 70 && c == ' ')
                            || column > 79) {
+                               flushString(os, write_buffer);
                                os << '\n';
                                column = 0;
                        }
                        // this check is to amend a bug. LyX sometimes
                        // inserts '\0' this could cause problems.
                        if (c != '\0')
-                               os << to_utf8(docstring(1, c));
+                               write_buffer.push_back(c);
                        else
                                LYXERR0("NUL char in structure.");
                        ++column;
@@ -1232,13 +1295,14 @@ void Paragraph::write(ostream & os, BufferParams const & bparams,
                }
        }
 
+       flushString(os, write_buffer);
        os << "\n\\end_layout\n";
 }
 
 
 void Paragraph::validate(LaTeXFeatures & features) const
 {
-       d->validate(features, *d->layout_);
+       d->validate(features);
 }
 
 
@@ -1275,7 +1339,7 @@ void Paragraph::appendString(docstring const & s, Font const & font,
        d->text_.append(s);
 
        // FIXME: Optimize this!
-       for (pos_type i = oldsize; i != newsize; ++i) {
+       for (size_t i = oldsize; i != newsize; ++i) {
                // track change
                d->changes_.insert(change, i);
        }
@@ -1413,13 +1477,14 @@ Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
        Font font = getFontSettings(bparams, pos);
 
        pos_type const body_pos = beginOfBody();
+       FontInfo & fi = font.fontInfo();
        if (pos < body_pos)
-               font.fontInfo().realize(d->layout_->labelfont);
+               fi.realize(d->layout_->labelfont);
        else
-               font.fontInfo().realize(d->layout_->font);
+               fi.realize(d->layout_->font);
 
-       font.fontInfo().realize(outerfont.fontInfo());
-       font.fontInfo().realize(bparams.getFont().fontInfo());
+       fi.realize(outerfont.fontInfo());
+       fi.realize(bparams.getFont().fontInfo());
 
        return font;
 }
@@ -1570,7 +1635,8 @@ docstring const & Paragraph::labelString() const
 // the next two functions are for the manual labels
 docstring const Paragraph::getLabelWidthString() const
 {
-       if (d->layout_->margintype == MARGIN_MANUAL)
+       if (d->layout_->margintype == MARGIN_MANUAL
+           || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
                return d->params_.labelWidthString();
        else
                return _("Senseless with this layout!");
@@ -1583,35 +1649,17 @@ void Paragraph::setLabelWidthString(docstring const & s)
 }
 
 
-docstring const Paragraph::translateIfPossible(docstring const & s,
-               BufferParams const & bparams) const
-{
-       if (!isAscii(s) || s.empty()) {
-               // This must be a user defined layout. We cannot translate
-               // this, since gettext accepts only ascii keys.
-               return s;
-       }
-       // Probably standard layout, try to translate
-       Messages & m = getMessages(getParLanguage(bparams)->code());
-       return m.get(to_ascii(s));
-}
-
-
 docstring Paragraph::expandLabel(Layout const & layout,
                BufferParams const & bparams, bool process_appendix) const
 {
        DocumentClass const & tclass = bparams.documentClass();
-
-       docstring fmt;
-       if (process_appendix && d->params_.appendix())
-               fmt = translateIfPossible(layout.labelstring_appendix(),
-                       bparams);
-       else
-               fmt = translateIfPossible(layout.labelstring(), bparams);
+       string const & lang = getParLanguage(bparams)->code();
+       bool const in_appendix = process_appendix && d->params_.appendix();
+       docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
 
        if (fmt.empty() && layout.labeltype == LABEL_COUNTER 
            && !layout.counter.empty())
-               fmt = "\\the" + layout.counter;
+               return tclass.counters().theCounter(layout.counter, lang);
 
        // handle 'inherited level parts' in 'fmt',
        // i.e. the stuff between '@' in   '@Section@.\arabic{subsection}'
@@ -1629,7 +1677,7 @@ docstring Paragraph::expandLabel(Layout const & layout,
                }
        }
 
-       return tclass.counters().counterLabel(fmt);
+       return tclass.counters().counterLabel(fmt, lang);
 }
 
 
@@ -1687,12 +1735,6 @@ void Paragraph::setBeginOfBody()
 }
 
 
-bool Paragraph::forcePlainLayout() const
-{
-       return inInset().forcePlainLayout();
-}
-
-
 bool Paragraph::allowParagraphCustomization() const
 {
        return inInset().allowParagraphCustomization();
@@ -1799,7 +1841,7 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
        }
 
        string const begin_tag = "\\begin";
-       InsetCode code = owner_->ownerCode();
+       InsetCode code = ownerCode();
        bool const lastpar = runparams.isLastPar;
 
        switch (curAlign) {
@@ -1845,7 +1887,12 @@ int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
 {
        int column = 0;
 
-       switch (params_.align()) {
+       LyXAlignment const curAlign = params_.align();
+
+       if (curAlign == layout_->align)
+               return column;
+
+       switch (curAlign) {
        case LYX_ALIGN_NONE:
        case LYX_ALIGN_BLOCK:
        case LYX_ALIGN_LAYOUT:
@@ -1862,10 +1909,10 @@ int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
        }
 
        string const end_tag = "\n\\par\\end";
-       InsetCode code = owner_->ownerCode();
+       InsetCode code = ownerCode();
        bool const lastpar = runparams.isLastPar;
 
-       switch (params_.align()) {
+       switch (curAlign) {
        case LYX_ALIGN_NONE:
        case LYX_ALIGN_BLOCK:
        case LYX_ALIGN_LAYOUT:
@@ -1904,19 +1951,24 @@ int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
 
 // This one spits out the text of the paragraph
 bool Paragraph::latex(BufferParams const & bparams,
-                               Font const & outerfont,
-                               odocstream & os, TexRow & texrow,
-                               OutputParams const & runparams) const
+       Font const & outerfont,
+       odocstream & os, TexRow & texrow,
+       OutputParams const & runparams,
+       int start_pos, int end_pos) const
 {
        LYXERR(Debug::LATEX, "Paragraph::latex...     " << this);
 
+       if (layout().inpreamble)
+               return true;
+
        bool return_value = false;
 
-       bool asdefault = forcePlainLayout();
+       bool const allowcust = allowParagraphCustomization();
 
-       Layout const & style = asdefault ?
-               bparams.documentClass().plainLayout() :
-               *d->layout_;
+       // FIXME This check should not be needed. Perhaps issue an
+       // error if it triggers.
+       Layout const & style = inInset().forcePlainLayout() ?
+               bparams.documentClass().plainLayout() : *d->layout_;
 
        // Current base font for all inherited font changes, without any
        // change caused by an individual character, except for the language:
@@ -1931,10 +1983,8 @@ bool Paragraph::latex(BufferParams const & bparams,
        unsigned int column = 0;
 
        if (body_pos > 0) {
-               // the optional argument is kept in curly brackets in
-               // case it contains a ']'
-               os << "[{";
-               column += 2;
+               os << '[';
+               column += 1;
                basefont = getLabelFont(bparams, outerfont);
        } else {
                basefont = getLayoutFont(bparams, outerfont);
@@ -1947,6 +1997,8 @@ bool Paragraph::latex(BufferParams const & bparams,
 
        Change runningChange = Change(Change::UNCHANGED);
 
+       Encoding const * const prev_encoding = runparams.encoding;
+
        texrow.start(id(), 0);
 
        // if the paragraph is empty, the loop will not be entered at all
@@ -1955,7 +2007,7 @@ bool Paragraph::latex(BufferParams const & bparams,
                        os << '{';
                        ++column;
                }
-               if (!asdefault)
+               if (allowcust)
                        column += d->startTeXParParams(bparams, os, texrow,
                                                    runparams);
        }
@@ -1974,18 +2026,19 @@ bool Paragraph::latex(BufferParams const & bparams,
                                running_font = basefont;
 
                                column += Changes::latexMarkChange(os, bparams,
-                                               runningChange, Change(Change::UNCHANGED));
+                                               runningChange, Change(Change::UNCHANGED),
+                                               runparams);
                                runningChange = Change(Change::UNCHANGED);
 
-                               os << "}] ";
-                               column +=3;
+                               os << "] ";
+                               column +=2;
                        }
                        if (style.isCommand()) {
                                os << '{';
                                ++column;
                        }
 
-                       if (!asdefault)
+                       if (allowcust)
                                column += d->startTeXParParams(bparams, os,
                                                            texrow,
                                                            runparams);
@@ -2003,13 +2056,14 @@ bool Paragraph::latex(BufferParams const & bparams,
                        basefont = getLayoutFont(bparams, outerfont);
                        running_font = basefont;
 
-                       column += Changes::latexMarkChange(os, bparams, runningChange, change);
+                       column += Changes::latexMarkChange(os, bparams, runningChange,
+                                                          change, runparams);
                        runningChange = change;
                }
 
                // do not output text which is marked deleted
                // if change tracking output is disabled
-               if (!bparams.outputChanges && change.type == Change::DELETED) {
+               if (!bparams.outputChanges && change.deleted()) {
                        continue;
                }
 
@@ -2068,16 +2122,22 @@ bool Paragraph::latex(BufferParams const & bparams,
                        running_font = font;
                        open_font = true;
                        docstring fontchange = ods.str();
+                       // check whether the fontchange ends with a \\textcolor
+                       // modifier and the text starts with a space (bug 4473)
+                       docstring const last_modifier = rsplit(fontchange, '\\');
+                       if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
+                               os << fontchange << from_ascii("{}");
                        // check if the fontchange ends with a trailing blank
                        // (like "\small " (see bug 3382)
-                       if (suffixIs(fontchange, ' ') && c == ' ')
+                       else if (suffixIs(fontchange, ' ') && c == ' ')
                                os << fontchange.substr(0, fontchange.size() - 1) 
                                   << from_ascii("{}");
                        else
                                os << fontchange;
                }
 
-               if (c == ' ') {
+               // FIXME: think about end_pos implementation...
+               if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
                        // FIXME: integrate this case in latexSpecialChar
                        // Do not print the separation of the optional argument
                        // if style.pass_thru is false. This works because
@@ -2105,16 +2165,19 @@ bool Paragraph::latex(BufferParams const & bparams,
                // Two major modes:  LaTeX or plain
                // Handle here those cases common to both modes
                // and then split to handle the two modes separately.
-               if (c == META_INSET)
-                       d->latexInset(bparams, os,
-                                       texrow, rp, running_font,
-                                       basefont, outerfont, open_font,
-                                       runningChange, style, i, column);
-               else {
-                       try {
-                               d->latexSpecialChar(os, rp, running_font, runningChange,
-                                       style, i, column);
-                       } catch (EncodingException & e) {
+               if (c == META_INSET) {
+                       if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
+                               d->latexInset(bparams, os,
+                                               texrow, rp, running_font,
+                                               basefont, outerfont, open_font,
+                                               runningChange, style, i, column);
+                       }
+               } else {
+                       if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
+                               try {
+                                       d->latexSpecialChar(os, rp, running_font, runningChange,
+                                               style, i, column);
+                               } catch (EncodingException & e) {
                                if (runparams.dryrun) {
                                        os << "<" << _("LyX Warning: ")
                                           << _("uncodable character") << " '";
@@ -2128,6 +2191,7 @@ bool Paragraph::latex(BufferParams const & bparams,
                                }
                        }
                }
+               }
 
                // Set the encoding to that returned from latexSpecialChar (see
                // comment for encoding member in OutputParams.h)
@@ -2155,17 +2219,20 @@ bool Paragraph::latex(BufferParams const & bparams,
 #endif
        }
 
-       column += Changes::latexMarkChange(os, bparams, runningChange, Change(Change::UNCHANGED));
+       column += Changes::latexMarkChange(os, bparams, runningChange,
+                                          Change(Change::UNCHANGED), runparams);
 
        // Needed if there is an optional argument but no contents.
        if (body_pos > 0 && body_pos == size()) {
-               os << "}]~";
+               os << "]~";
                return_value = false;
        }
 
-       if (!asdefault) {
-               column += d->endTeXParParams(bparams, os, texrow,
-                                         runparams);
+       if (allowcust && d->endTeXParParams(bparams, os, texrow, runparams)
+           && runparams.encoding != prev_encoding) {
+               runparams.encoding = prev_encoding;
+               if (!bparams.useXetex)
+                       os << setEncoding(prev_encoding->iconvName());
        }
 
        LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
@@ -2178,6 +2245,8 @@ bool Paragraph::emptyTag() const
        for (pos_type i = 0; i < size(); ++i) {
                if (Inset const * inset = getInset(i)) {
                        InsetCode lyx_code = inset->lyxCode();
+                       // FIXME testing like that is wrong. What is
+                       // the intent?
                        if (lyx_code != TOC_CODE &&
                            lyx_code != INCLUDE_CODE &&
                            lyx_code != GRAPHICS_CODE &&
@@ -2214,7 +2283,7 @@ string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
 }
 
 
-pos_type Paragraph::firstWord(odocstream & os, OutputParams const & runparams)
+pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
        const
 {
        pos_type i;
@@ -2232,6 +2301,24 @@ pos_type Paragraph::firstWord(odocstream & os, OutputParams const & runparams)
 }
 
 
+pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
+       const
+{
+       pos_type i;
+       for (i = 0; i < size(); ++i) {
+               if (Inset const * inset = getInset(i)) {
+                       inset->xhtml(xs, runparams);
+               } else {
+                       char_type c = d->text_[i];
+                       if (c == ' ')
+                               break;
+                       xs << c;
+               }
+       }
+       return i;
+}
+
+
 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
 {
        Font font_old;
@@ -2303,6 +2390,102 @@ void Paragraph::simpleDocBookOnePar(Buffer const & buf,
 }
 
 
+docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
+                                   XHTMLStream & xs,
+                                   OutputParams const & runparams,
+                                   Font const & outerfont,
+                                               bool fortoc,
+                                   pos_type initial) const
+{
+       docstring retval;
+
+       bool emph_flag = false;
+       bool bold_flag = false;
+       std::string closing_tag;
+
+       Layout const & style = *d->layout_;
+
+       if (!fortoc) {
+               // generate a magic label for this paragraph
+               string const attr = "id='" + magicLabel() + "'";
+               xs << CompTag("a", attr);
+       }
+
+       FontInfo font_old =
+               style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
+
+       // parsing main loop
+       for (pos_type i = initial; i < size(); ++i) {
+               Font font = getFont(buf.params(), i, outerfont);
+
+               // emphasis
+               if (font_old.emph() != font.fontInfo().emph()) {
+                       if (font.fontInfo().emph() == FONT_ON) {
+                               xs << StartTag("em");
+                               emph_flag = true;
+                       } else if (emph_flag && i != initial) {
+                               xs << EndTag("em");
+                               emph_flag = false;
+                       }
+               }
+               // bold
+               if (font_old.series() != font.fontInfo().series()) {
+                       if (font.fontInfo().series() == BOLD_SERIES) {
+                               xs << StartTag("strong");
+                               bold_flag = true;
+                       } else if (bold_flag && i != initial) {
+                               xs << EndTag("strong");
+                               bold_flag = false;
+                       }
+               }
+               // FIXME XHTML
+               // Other such tags? What about the other text ranges?
+
+               Inset const * inset = getInset(i);
+               if (inset) {
+                       InsetCommand const * ic = inset->asInsetCommand();
+                       InsetLayout const & il = inset->getLayout();
+                       if (!fortoc || il.isInToc() || (ic && ic->isInToc())) {
+                               OutputParams np = runparams;
+                               if (!il.htmlisblock())
+                                       np.html_in_par = true;
+                               retval += inset->xhtml(xs, np);
+                       }
+               } else {
+                       char_type c = d->text_[i];
+
+                       if (style.pass_thru)
+                               xs << c;
+                       else if (c == '-') {
+                               docstring str;
+                               int j = i + 1;
+                               if (j < size() && d->text_[j] == '-') {
+                                       j += 1;
+                                       if (j < size() && d->text_[j] == '-') {
+                                               str += from_ascii("&mdash;");
+                                               i += 2;
+                                       } else {
+                                               str += from_ascii("&ndash;");
+                                               i += 1;
+                                       }
+                               }
+                               else
+                                       str += c;
+                               // We don't want to escape the entities. Note that
+                               // it is safe to do this, since str can otherwise
+                               // only be "-". E.g., it can't be "<".
+                               xs << XHTMLStream::NextRaw() << str;
+                       } else
+                               xs << c;
+               }
+               font_old = font.fontInfo();
+       }
+
+       xs.closeFontTags();
+       return retval;
+}
+
+
 bool Paragraph::isHfill(pos_type pos) const
 {
        Inset const * inset = getInset(pos);
@@ -2328,13 +2511,15 @@ bool Paragraph::isLineSeparator(pos_type pos) const
 }
 
 
-/// Used by the spellchecker
-bool Paragraph::isLetter(pos_type pos) const
+bool Paragraph::isWordSeparator(pos_type pos) const
 {
        if (Inset const * inset = getInset(pos))
-               return inset->isLetter();
+               return !inset->isLetter();
        char_type const c = d->text_[pos];
-       return isLetterChar(c) || isDigit(c);
+       // We want to pass the ' and escape chars to the spellchecker
+       static docstring const quote = from_utf8(lyxrc.spellchecker_esc_chars + '\'');
+       return (!isLetterChar(c) && !isDigit(c) && !contains(quote, c))
+               || pos == size();
 }
 
 
@@ -2370,8 +2555,7 @@ bool Paragraph::isRTL(BufferParams const & bparams) const
 {
        return lyxrc.rtl_support
                && getParLanguage(bparams)->rightToLeft()
-               && ownerCode() != ERT_CODE
-               && ownerCode() != LISTINGS_CODE;
+               && !inInset().getLayout().forceLTR();
 }
 
 
@@ -2421,10 +2605,37 @@ docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
 
        for (pos_type i = beg; i < end; ++i) {
                char_type const c = d->text_[i];
-               if (isPrintable(c) || c == '\t')
+               if (isPrintable(c) || c == '\t'
+                   || (c == '\n' && (options & AS_STR_NEWLINES)))
                        os.put(c);
-               else if (c == META_INSET && options & AS_STR_INSETS)
-                       getInset(i)->textString(os);
+               else if (c == META_INSET && (options & AS_STR_INSETS)) {
+                       getInset(i)->tocString(os);
+                       if (getInset(i)->asInsetMath())
+                               os << " ";
+               }
+       }
+
+       return os.str();
+}
+
+
+docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputParams & runparams) const
+{
+       odocstringstream os;
+
+       if (beg == 0 
+               && options & AS_STR_LABEL
+               && !d->params_.labelString().empty())
+               os << d->params_.labelString() << ' ';
+
+       for (pos_type i = beg; i < end; ++i) {
+               char_type const c = d->text_[i];
+               if (isPrintable(c) || c == '\t'
+                   || (c == '\n' && (options & AS_STR_NEWLINES)))
+                       os.put(c);
+               else if (c == META_INSET && (options & AS_STR_INSETS)) {
+                       getInset(i)->plaintext(os, runparams);
+               }
        }
 
        return os.str();
@@ -2455,29 +2666,35 @@ void Paragraph::setLayout(Layout const & layout)
 }
 
 
+void Paragraph::setDefaultLayout(DocumentClass const & tc)
+{ 
+       setLayout(tc.defaultLayout()); 
+}
+
+
+void Paragraph::setPlainLayout(DocumentClass const & tc)
+{ 
+       setLayout(tc.plainLayout()); 
+}
+
+
 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
 {
        if (usePlainLayout())
-               setLayout(tclass.plainLayout());
+               setPlainLayout(tclass);
        else
-               setLayout(tclass.defaultLayout());
+               setDefaultLayout(tclass);
 }
 
 
 Inset const & Paragraph::inInset() const
 {
        LASSERT(d->inset_owner_, throw ExceptionMessage(BufferException,
-               _("Memory problem"), _("Paragraph not properly initiliazed")));
+               _("Memory problem"), _("Paragraph not properly initialized")));
        return *d->inset_owner_;
 }
 
 
-InsetCode Paragraph::ownerCode() const
-{
-       return d->inset_owner_ ? d->inset_owner_->lyxCode() : NO_CODE;
-}
-
-
 ParagraphParameters & Paragraph::params()
 {
        return d->params_;
@@ -2604,7 +2821,7 @@ int Paragraph::checkBiblio(Buffer const & buffer)
        // There was no inset at the beginning, so we need to create one with
        // the key and label of the one we erased.
        InsetBibitem * inset = 
-               new InsetBibitem(buffer, InsetCommandParams(BIBITEM_CODE));
+               new InsetBibitem(const_cast<Buffer *>(&buffer), InsetCommandParams(BIBITEM_CODE));
        // restore values of previously deleted item in this par.
        if (!oldkey.empty())
                inset->setParam("key", oldkey);
@@ -2622,21 +2839,21 @@ void Paragraph::checkAuthors(AuthorList const & authorList)
 }
 
 
-bool Paragraph::isUnchanged(pos_type pos) const
+bool Paragraph::isChanged(pos_type pos) const
 {
-       return lookupChange(pos).type == Change::UNCHANGED;
+       return lookupChange(pos).changed();
 }
 
 
 bool Paragraph::isInserted(pos_type pos) const
 {
-       return lookupChange(pos).type == Change::INSERTED;
+       return lookupChange(pos).inserted();
 }
 
 
 bool Paragraph::isDeleted(pos_type pos) const
 {
-       return lookupChange(pos).type == Change::DELETED;
+       return lookupChange(pos).deleted();
 }
 
 
@@ -2709,33 +2926,32 @@ void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
                        }
                }
 
-               if (!isLetter(pos) || isDeleted(pos)) {
+               if (isWordSeparator(pos) || isDeleted(pos)) {
                        // permit capitalization again
                        capitalize = true;
                }
 
-               if (oldChar != newChar)
+               if (oldChar != newChar) {
                        changes += newChar;
+                       if (pos != right - 1)
+                               continue;
+                       // step behind the changing area
+                       pos++;
+               }
 
-               if (oldChar == newChar || pos == right - 1) {
-                       if (oldChar != newChar) {
-                               // step behind the changing area
-                               pos++;
-                       }
-                       int erasePos = pos - changes.size();
-                       for (size_t i = 0; i < changes.size(); i++) {
-                               insertChar(pos, changes[i],
-                                       getFontSettings(bparams,
-                                       erasePos),
-                                       trackChanges);
-                               if (!eraseChar(erasePos, trackChanges)) {
-                                       ++erasePos;
-                                       ++pos; // advance
-                                       ++right; // expand selection
-                               }
+               int erasePos = pos - changes.size();
+               for (size_t i = 0; i < changes.size(); i++) {
+                       insertChar(pos, changes[i],
+                                  getFontSettings(bparams,
+                                                  erasePos),
+                                  trackChanges);
+                       if (!eraseChar(erasePos, trackChanges)) {
+                               ++erasePos;
+                               ++pos; // advance
+                               ++right; // expand selection
                        }
-                       changes.clear();
                }
+               changes.clear();
        }
 }
 
@@ -2762,10 +2978,10 @@ bool Paragraph::find(docstring const & str, bool cs, bool mw,
 
        // if necessary, check whether string matches word
        if (mw) {
-               if (pos > 0 && isLetter(pos - 1))
+               if (pos > 0 && !isWordSeparator(pos - 1))
                        return false;
                if (pos + strsize < parsize
-                       && isLetter(pos + strsize))
+                       && !isWordSeparator(pos + strsize))
                        return false;
        }
 
@@ -2814,38 +3030,56 @@ void Paragraph::deregisterWords()
 }
 
 
-void Paragraph::collectWords(CursorSlice const & sl)
+void Paragraph::locateWord(pos_type & from, pos_type & to,
+       word_location const loc) const
 {
-       // find new words
-       bool inword = false;
+       switch (loc) {
+       case WHOLE_WORD_STRICT:
+               if (from == 0 || from == size()
+                   || isWordSeparator(from)
+                   || isWordSeparator(from - 1)) {
+                       to = from;
+                       return;
+               }
+               // no break here, we go to the next
 
-       //lyxerr << "Words: ";
-       pos_type n = size();
-       for (pos_type pos = 0; pos != n; ++pos) {
-               if (isDeleted(pos))
-                       continue;
+       case WHOLE_WORD:
+               // If we are already at the beginning of a word, do nothing
+               if (!from || isWordSeparator(from - 1))
+                       break;
+               // no break here, we go to the next
 
-               if (!isLetter(pos)) {
-                       inword = false;
-                       continue;
-               }
+       case PREVIOUS_WORD:
+               // always move the cursor to the beginning of previous word
+               while (from && !isWordSeparator(from - 1))
+                       --from;
+               break;
+       case NEXT_WORD:
+               LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
+               break;
+       case PARTIAL_WORD:
+               // no need to move the 'from' cursor
+               break;
+       }
+       to = from;
+       while (to < size() && !isWordSeparator(to))
+               ++to;
+}
 
-               if (inword)
-                       continue;
 
-               inword = true;
-               CursorSlice from = sl;
-               CursorSlice to = sl;
-               from.pos() = pos;
-               to.pos() = pos;
-               from.text()->getWord(from, to, WHOLE_WORD);
-               if (to.pos() - from.pos() < 6)
+void Paragraph::collectWords()
+{
+       pos_type n = size();
+       for (pos_type pos = 0; pos < n; ++pos) {
+               if (isWordSeparator(pos))
                        continue;
-               docstring word = asString(from.pos(), to.pos(), false);
-               d->words_.insert(word);
-               //lyxerr << word << " ";
+               pos_type from = pos;
+               locateWord(from, pos, WHOLE_WORD);
+               if (pos - from >= 6) {
+                       docstring word = asString(from, pos, AS_STR_NONE);
+                       d->words_.insert(word);
+               }
        }
-       //lyxerr << std::endl;
 }
 
 
@@ -2858,12 +3092,68 @@ void Paragraph::registerWords()
 }
 
 
-void Paragraph::updateWords(CursorSlice const & sl)
+void Paragraph::updateWords()
 {
-       LASSERT(&sl.paragraph() == this, /**/);
        deregisterWords();
-       collectWords(sl);
+       collectWords();
        registerWords();
 }
 
+
+bool Paragraph::spellCheck(pos_type & from, pos_type & to, WordLangTuple & wl,
+       docstring_list & suggestions, bool do_suggestion) const
+{
+       SpellChecker * speller = theSpellChecker();
+       if (!speller)
+               return false;
+
+       locateWord(from, to, WHOLE_WORD);
+       if (from == to || from >= pos_type(d->text_.size()))
+               return false;
+
+       docstring word = asString(from, to, AS_STR_INSETS);
+       string const lang_code = lyxrc.spellchecker_alt_lang.empty()
+               ? getFontSettings(d->inset_owner_->buffer().params(), from).language()->code()
+               : lyxrc.spellchecker_alt_lang;
+       wl = WordLangTuple(word, lang_code);
+       SpellChecker::Result res = speller->check(wl);
+       // Just ignore any error that the spellchecker reports.
+       // FIXME: we should through out an exception and catch it in the GUI to
+       // display the error.
+       if (!speller->error().empty())
+               return false;
+
+       bool const misspelled = res != SpellChecker::OK
+               && res != SpellChecker::IGNORED_WORD;
+
+       if (lyxrc.spellcheck_continuously)
+               d->fontlist_.setMisspelled(from, to, misspelled);
+
+       if (misspelled && do_suggestion)
+               speller->suggest(wl, suggestions);
+       else
+               suggestions.clear();
+
+       return misspelled;
+}
+
+
+bool Paragraph::isMisspelled(pos_type pos) const
+{
+       pos_type from = pos;
+       pos_type to = pos;
+       WordLangTuple wl;
+       docstring_list suggestions;
+       return spellCheck(from, to, wl, suggestions, false);
+}
+
+
+string Paragraph::magicLabel() const
+{
+       stringstream ss;
+       ss << "magicparlabel-" << id();
+       return ss.str();
+}
+
+
 } // namespace lyx