]> git.lyx.org Git - lyx.git/blobdiff - src/Paragraph.cpp
Remove extra whitespace.
[lyx.git] / src / Paragraph.cpp
index 8845054b0c131f688ef2fd0e5a27d0158b0ca760..df8de4c39f09ced1bece8559b5f0405d4eb102b3 100644 (file)
@@ -43,7 +43,6 @@
 #include "TextClass.h"
 #include "TexRow.h"
 #include "Text.h"
-#include "VSpace.h"
 #include "WordLangTuple.h"
 #include "WordList.h"
 
@@ -72,7 +71,7 @@ namespace lyx {
 namespace {
 /// Inset identifier (above 0x10ffff, for ucs-4)
 char_type const META_INSET = 0x200001;
-};
+}
 
 
 /////////////////////////////////////////////////////////////////////
@@ -112,7 +111,7 @@ public:
                if (range_.first > pos) {
                        range_.first += offset;
                        range_.last += offset;
-               } else if (range_.last > pos) {
+               } else if (range_.last >= pos) {
                        range_.last += offset;
                }
        }
@@ -137,7 +136,30 @@ public:
 
        void setRange(FontSpan const fp, SpellChecker::Result state)
        {
-               eraseCoveredRanges(fp);
+               Ranges result;
+               RangesIterator et = ranges_.end();
+               RangesIterator it = ranges_.begin();
+               for (; it != et; ++it) {
+                       if (!it->covered(fp))
+                               result.push_back(SpellResultRange(it->range(), it->result()));
+                       else if (state == SpellChecker::WORD_OK) {
+                               // trim or split the current misspelled range
+                               // store misspelled ranges only
+                               FontSpan range = it->range();
+                               if (fp.first > range.first) {
+                                       // misspelled area in front of WORD_OK
+                                       range.last = fp.first - 1;
+                                       result.push_back(SpellResultRange(range, it->result()));
+                                       range = it->range();
+                               }
+                               if (fp.last < range.last) {
+                                       // misspelled area after WORD_OK range
+                                       range.first = fp.last + 1;
+                                       result.push_back(SpellResultRange(range, it->result()));
+                               }
+                       }
+               }
+               ranges_ = result;
                if (state != SpellChecker::WORD_OK)
                        ranges_.push_back(SpellResultRange(fp, state));
        }
@@ -173,6 +195,20 @@ public:
                return result;
        }
 
+       FontSpan const & getRange(pos_type pos) const
+       {
+               /// empty span to indicate mismatch
+               static FontSpan empty_;
+               RangesIterator et = ranges_.end();
+               RangesIterator it = ranges_.begin();
+               for (; it != et; ++it) {
+                       if(it->inside(pos)) {
+                               return it->range();
+                       }
+               }
+               return empty_;
+       }
+
        bool needsRefresh() const {
                return needs_refresh_;
        }
@@ -193,8 +229,10 @@ public:
                        if (pos > refresh_.last)
                                refresh_.last = pos;
                } else if (pos != -1) {
-                       refresh_.first = pos;
-                       refresh_.last = pos;
+                       // init request check for neighbour positions too
+                       refresh_.first = pos > 0 ? pos - 1 : 0;
+                       // no need for special end of paragraph check
+                       refresh_.last = pos + 1;
                }
                needs_refresh_ = pos != -1;
        }
@@ -217,18 +255,6 @@ private:
        SpellChecker::ChangeNumber current_change_number_;
 
 
-       void eraseCoveredRanges(FontSpan const fp)
-       {
-               Ranges result;
-               RangesIterator et = ranges_.end();
-               RangesIterator it = ranges_.begin();
-               for (; it != et; ++it) {
-                       if (!it->covered(fp))
-                               result.push_back(SpellResultRange(it->range(), it->result()));
-               }
-               ranges_ = result;
-       }
-
        void correctRangesAfterPos(pos_type pos, int offset)
        {
                RangesIterator et = ranges_.end();
@@ -261,14 +287,14 @@ 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,
+       int latexSurrogatePair(otexstream & os, char_type c, char_type next,
                               OutputParams const &);
 
        /// Output a space in appropriate formatting (or a surrogate pair
        /// if the next character is a combining character).
        /// \return whether a surrogate pair was output.
        bool simpleTeXBlanks(OutputParams const &,
-                            odocstream &, TexRow & texrow,
+                            otexstream &,
                             pos_type i,
                             unsigned int & column,
                             Font const & font,
@@ -277,21 +303,21 @@ public:
        /// 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(odocstream & os, docstring const & ltx,
+       int writeScriptChars(otexstream & os, docstring const & ltx,
                           Change const &, Encoding const &, pos_type & i);
 
        /// This could go to ParagraphParameters if we want to.
-       int startTeXParParams(BufferParams const &, odocstream &, TexRow &,
+       int startTeXParParams(BufferParams const &, otexstream &,
                              OutputParams const &) const;
 
        /// This could go to ParagraphParameters if we want to.
-       int endTeXParParams(BufferParams const &, odocstream &, TexRow &,
-                           OutputParams const &) const;
+       bool endTeXParParams(BufferParams const &, otexstream &,
+                            OutputParams const &) const;
 
        ///
        void latexInset(BufferParams const &,
-                                  odocstream &,
-                                  TexRow & texrow, OutputParams &,
+                                  otexstream &,
+                                  OutputParams &,
                                   Font & running_font,
                                   Font & basefont,
                                   Font const & outerfont,
@@ -303,30 +329,32 @@ public:
 
        ///
        void latexSpecialChar(
-                                  odocstream & os,
+                                  otexstream & os,
                                   OutputParams const & runparams,
                                   Font const & running_font,
                                   Change const & running_change,
                                   Layout const & style,
                                   pos_type & i,
+                                  pos_type end_pos,
                                   unsigned int & column);
 
        ///
        bool latexSpecialT1(
                char_type const c,
-               odocstream & os,
+               otexstream & os,
                pos_type i,
                unsigned int & column);
        ///
        bool latexSpecialTypewriter(
                char_type const c,
-               odocstream & os,
+               otexstream & os,
                pos_type i,
                unsigned int & column);
        ///
        bool latexSpecialPhrase(
-               odocstream & os,
+               otexstream & os,
                pos_type & i,
+               pos_type end_pos,
                unsigned int & column,
                OutputParams const & runparams);
 
@@ -358,21 +386,23 @@ public:
                return speller_change_number > speller_state_.currentChangeNumber();
        }
 
+       bool ignoreWord(docstring const & word) const ;
+       
        void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
        {
                pos_type textsize = owner_->size();
                // check for sane arguments
-               if (to < from || from >= textsize)
+               if (to <= from || from >= textsize)
                        return;
-               FontSpan fp = FontSpan(from, to);
-               // don't mark end of paragraph
-               if (fp.last >= textsize)
-                       fp.last = textsize - 1;
+               FontSpan fp = FontSpan(from, to - 1);
                speller_state_.setRange(fp, state);
        }
 
        void requestSpellCheck(pos_type pos) {
-               speller_state_.needsRefresh(pos);
+               if (pos == -1)
+                       speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
+               else
+                       speller_state_.needsRefresh(pos);
        }
 
        void readySpellCheck() {
@@ -811,7 +841,7 @@ int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
 }
 
 
-int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
+int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
                char_type next, OutputParams const & runparams)
 {
        // Writing next here may circumvent a possible font change between
@@ -840,7 +870,7 @@ int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
 
 
 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
-                                      odocstream & os, TexRow & texrow,
+                                      otexstream & os,
                                       pos_type i,
                                       unsigned int & column,
                                       Font const & font,
@@ -873,8 +903,7 @@ bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
                     || text_[i - 1] == ':'
                     || text_[i - 1] == '!'))) {
                os << '\n';
-               texrow.newline();
-               texrow.start(owner_->id(), i + 1);
+               os.texrow().start(owner_->id(), i + 1);
                column = 0;
        } else if (style.free_spacing) {
                os << '~';
@@ -885,7 +914,7 @@ bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
 }
 
 
-int Paragraph::Private::writeScriptChars(odocstream & os,
+int Paragraph::Private::writeScriptChars(otexstream & os,
                                         docstring const & ltx,
                                         Change const & runningChange,
                                         Encoding const & encoding,
@@ -979,8 +1008,7 @@ bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
 
 
 void Paragraph::Private::latexInset(BufferParams const & bparams,
-                                   odocstream & os,
-                                   TexRow & texrow,
+                                   otexstream & os,
                                    OutputParams & runparams,
                                    Font & running_font,
                                    Font & basefont,
@@ -995,7 +1023,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
        LASSERT(inset, /**/);
 
        if (style.pass_thru) {
-               inset->plaintext(os, runparams);
+               inset->plaintext(os.os(), runparams);
                return;
        }
 
@@ -1023,8 +1051,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
                                os << "\\protect ";
 
                }
-               texrow.newline();
-               texrow.start(owner_->id(), i + 1);
+               os.texrow().start(owner_->id(), i + 1);
                column = 0;
        }
 
@@ -1040,7 +1067,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
        }
 
        bool close = false;
-       odocstream::pos_type const len = os.tellp();
+       odocstream::pos_type const len = os.os().tellp();
 
        if (inset->forceLTR()
            && running_font.isRightToLeft()
@@ -1066,7 +1093,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
        // ArabTeX, though, cannot handle this special behavior, it seems.
        bool arabtex = basefont.language()->lang() == "arabic_arabtex"
                || running_font.language()->lang() == "arabic_arabtex";
-       if (open_font && inset->noFontChange()) {
+       if (open_font && !inset->inheritFont()) {
                bool closeLanguage = arabtex
                        || basefont.isRightToLeft() == running_font.isRightToLeft();
                unsigned int count = running_font.latexWriteEndChanges(os,
@@ -1089,10 +1116,12 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
                }
        }
 
-       int tmp;
+       int prev_rows = os.texrow().rows();
 
        try {
-               tmp = inset->latex(os, runparams);
+               runparams.lastid = id_;
+               runparams.lastpos = i;
+               inset->latex(os, runparams);
        } catch (EncodingException & e) {
                // add location information and throw again.
                e.par_id = id_;
@@ -1107,12 +1136,11 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
                                os << '}';
        }
 
-       if (tmp) {
-               texrow.newlines(tmp);
-               texrow.start(owner_->id(), i + 1);
+       if (os.texrow().rows() > prev_rows) {
+               os.texrow().start(owner_->id(), i + 1);
                column = 0;
        } else {
-               column += (unsigned int)(os.tellp() - len);
+               column += (unsigned int)(os.os().tellp() - len);
        }
 
        if (owner_->isDeleted(i))
@@ -1120,22 +1148,24 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
 }
 
 
-void Paragraph::Private::latexSpecialChar(
-                                            odocstream & os,
-                                            OutputParams const & runparams,
-                                            Font const & running_font,
-                                            Change const & running_change,
-                                            Layout const & style,
-                                            pos_type & i,
-                                            unsigned int & column)
+void Paragraph::Private::latexSpecialChar(otexstream & os,
+                                         OutputParams const & runparams,
+                                         Font const & running_font,
+                                         Change const & running_change,
+                                         Layout const & style,
+                                         pos_type & i,
+                                         pos_type end_pos,
+                                         unsigned int & column)
 {
        char_type const c = text_[i];
 
        if (style.pass_thru || runparams.pass_thru) {
-               if (c != '\0')
-                       // FIXME UNICODE: This can fail if c cannot
-                       // be encoded in the current encoding.
+               if (c != '\0') {
+                       Encoding const * const enc = runparams.encoding;
+                       if (enc && enc->latexChar(c, true).empty())
+                               throw EncodingException(c);
                        os.put(c);
+               }
                return;
        }
 
@@ -1216,7 +1246,7 @@ void Paragraph::Private::latexSpecialChar(
 
        default:
                // LyX, LaTeX etc.
-               if (latexSpecialPhrase(os, i, column, runparams))
+               if (latexSpecialPhrase(os, i, end_pos, column, runparams))
                        return;
 
                if (c == '\0')
@@ -1252,7 +1282,7 @@ void Paragraph::Private::latexSpecialChar(
 }
 
 
-bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
+bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
        pos_type i, unsigned int & column)
 {
        switch (c) {
@@ -1280,7 +1310,7 @@ bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
 }
 
 
-bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream & os,
+bool Paragraph::Private::latexSpecialTypewriter(char_type const c, otexstream & os,
        pos_type i, unsigned int & column)
 {
        switch (c) {
@@ -1302,7 +1332,10 @@ bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream &
 }
 
 
-bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
+/// \param end_pos
+///   If [start_pos, end_pos) does not include entirely the special phrase, then
+///   do not apply the macro transformation.
+bool Paragraph::Private::latexSpecialPhrase(otexstream & os, pos_type & i, pos_type end_pos,
        unsigned int & column, OutputParams const & runparams)
 {
        // FIXME: if we have "LaTeX" with a font
@@ -1312,7 +1345,8 @@ bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
        // "words" for some definition of word
 
        for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
-               if (!isTextAt(special_phrases[pnr].phrase, i))
+               if (!isTextAt(special_phrases[pnr].phrase, i)
+                   || (end_pos != -1 && i + int(special_phrases[pnr].phrase.size()) > end_pos))
                        continue;
                if (runparams.moving_arg)
                        os << "\\protect";
@@ -1328,20 +1362,36 @@ bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
 void Paragraph::Private::validate(LaTeXFeatures & features) const
 {
        if (layout_->inpreamble && inset_owner_) {
+               bool const is_command = layout_->latextype == LATEX_COMMAND;
                Buffer const & buf = inset_owner_->buffer();
                BufferParams const & bp = buf.params();
                Font f;
-               TexRow tr;
+               TexRow texrow;
+               // Using a string stream here circumvents the encoding
+               // switching machinery of odocstream. Therefore the
+               // output is wrong if this paragraph contains content
+               // that needs to switch encoding.
                odocstringstream ods;
-               // we have to provide all the optional arguments here, even though
-               // the last one is the only one we care about.
-               owner_->latex(bp, f, ods, tr, features.runparams(), 0, -1, true);
-               docstring const d = ods.str();
-               if (!d.empty()) {
-                       // this will have "{" at the beginning, but not at the end
-                       string const content = to_utf8(d);
-                       string const cmd = layout_->latexname();
-                       features.addPreambleSnippet("\\" + cmd + content + "}");
+               otexstream os(ods, texrow);
+               if (is_command) {
+                       os << '\\' << from_ascii(layout_->latexname());
+                       // we have to provide all the optional arguments here, even though
+                       // the last one is the only one we care about.
+                       // Separate handling of optional argument inset.
+                       if (layout_->optargs != 0 || layout_->reqargs != 0)
+                               latexArgInsets(*owner_, os, features.runparams(),
+                                       layout_->reqargs, layout_->optargs);
+                       else
+                               os << from_ascii(layout_->latexparam());
+               }
+               docstring::size_type const length = ods.str().length();
+               // this will output "{" at the beginning, but not at the end
+               owner_->latex(bp, f, os, features.runparams(), 0, -1, true);
+               if (ods.str().length() > length) {
+                       if (is_command)
+                               ods << '}';
+                       string const snippet = to_utf8(ods.str());
+                       features.addPreambleSnippet(snippet);
                }
        }
 
@@ -1587,6 +1637,7 @@ void Paragraph::appendChar(char_type c, Font const & font,
        // when appending characters, no need to update tables
        d->text_.push_back(c);
        setFont(d->text_.size() - 1, font);
+       d->requestSpellCheck(d->text_.size() - 1);
 }
 
 
@@ -1607,6 +1658,7 @@ void Paragraph::appendString(docstring const & s, Font const & font,
        for (size_t i = oldsize; i != newsize; ++i) {
                // track change
                d->changes_.insert(change, i);
+               d->requestSpellCheck(i);
        }
        d->fontlist_.set(oldsize, font);
        d->fontlist_.set(newsize - 1, font);
@@ -2026,6 +2078,11 @@ bool Paragraph::usePlainLayout() const
 }
 
 
+bool Paragraph::isPassThru() const
+{
+       return inInset().getLayout().isPassThru() || d->layout_->pass_thru;
+}
+
 namespace {
 
 // paragraphs inside floats need different alignment tags to avoid
@@ -2073,13 +2130,12 @@ string const corrected_env(string const & suffix, string const & env,
 }
 
 
-void adjust_row_column(string const & str, TexRow & texrow, int & column)
+void adjust_column(string const & str, int & column)
 {
        if (!contains(str, "\n"))
                column += str.size();
        else {
                string tmp;
-               texrow.newline();
                column = rsplit(str, tmp, '\n').size();
        }
 }
@@ -2088,12 +2144,11 @@ void adjust_row_column(string const & str, TexRow & texrow, int & column)
 
 
 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
-                                odocstream & os, TexRow & texrow,
-                                OutputParams const & runparams) const
+                       otexstream & os, OutputParams const & runparams) const
 {
        int column = 0;
 
-       if (params_.noindent()) {
+       if (params_.noindent() && !layout_->pass_thru) {
                os << "\\noindent ";
                column += 10;
        }
@@ -2138,7 +2193,7 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
                else
                        output = corrected_env(begin_tag, "flushright", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
+               adjust_column(output, column);
                break;
        } case LYX_ALIGN_RIGHT: {
                string output;
@@ -2147,13 +2202,13 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
                else
                        output = corrected_env(begin_tag, "flushleft", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
+               adjust_column(output, column);
                break;
        } case LYX_ALIGN_CENTER: {
                string output;
                output = corrected_env(begin_tag, "center", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
+               adjust_column(output, column);
                break;
        }
        }
@@ -2162,16 +2217,13 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
 }
 
 
-int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
-                              odocstream & os, TexRow & texrow,
-                              OutputParams const & runparams) const
+bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
+                       otexstream & os, OutputParams const & runparams) const
 {
-       int column = 0;
-
        LyXAlignment const curAlign = params_.align();
 
        if (curAlign == layout_->align)
-               return column;
+               return false;
 
        switch (curAlign) {
        case LYX_ALIGN_NONE:
@@ -2183,13 +2235,12 @@ int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
        case LYX_ALIGN_LEFT:
        case LYX_ALIGN_RIGHT:
        case LYX_ALIGN_CENTER:
-               if (runparams.moving_arg) {
+               if (runparams.moving_arg)
                        os << "\\protect";
-                       column = 8;
-               }
                break;
        }
 
+       string output;
        string const end_tag = "\n\\par\\end";
        InsetCode code = ownerCode();
        bool const lastpar = runparams.isLastPar;
@@ -2202,40 +2253,34 @@ int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
        case LYX_ALIGN_DECIMAL:
                break;
        case LYX_ALIGN_LEFT: {
-               string output;
                if (owner_->getParLanguage(bparams)->babel() != "hebrew")
                        output = corrected_env(end_tag, "flushleft", code, lastpar);
                else
                        output = corrected_env(end_tag, "flushright", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
                break;
        } case LYX_ALIGN_RIGHT: {
-               string output;
                if (owner_->getParLanguage(bparams)->babel() != "hebrew")
                        output = corrected_env(end_tag, "flushright", code, lastpar);
                else
                        output = corrected_env(end_tag, "flushleft", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
                break;
        } case LYX_ALIGN_CENTER: {
-               string output;
                output = corrected_env(end_tag, "center", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
                break;
        }
        }
 
-       return column;
+       return !output.empty() || lastpar;
 }
 
 
 // This one spits out the text of the paragraph
 void Paragraph::latex(BufferParams const & bparams,
        Font const & outerfont,
-       odocstream & os, TexRow & texrow,
+       otexstream & os,
        OutputParams const & runparams,
        int start_pos, int end_pos, bool force) const
 {
@@ -2266,6 +2311,10 @@ void Paragraph::latex(BufferParams const & bparams,
        if (body_pos > 0) {
                // the optional argument is kept in curly brackets in
                // case it contains a ']'
+               // This is not strictly needed, but if this is changed it
+               // would be a file format change, and tex2lyx would need
+               // to be adjusted, since it unconditionally removes the
+               // braces when it parses \item.
                os << "[{";
                column += 2;
                basefont = getLabelFont(bparams, outerfont);
@@ -2282,7 +2331,7 @@ void Paragraph::latex(BufferParams const & bparams,
 
        Encoding const * const prev_encoding = runparams.encoding;
 
-       texrow.start(id(), 0);
+       os.texrow().start(id(), 0);
 
        // if the paragraph is empty, the loop will not be entered at all
        if (empty()) {
@@ -2291,8 +2340,7 @@ void Paragraph::latex(BufferParams const & bparams,
                        ++column;
                }
                if (allowcust)
-                       column += d->startTeXParParams(bparams, os, texrow,
-                                                   runparams);
+                       column += d->startTeXParParams(bparams, os, runparams);
        }
 
        for (pos_type i = 0; i < size(); ++i) {
@@ -2323,12 +2371,11 @@ void Paragraph::latex(BufferParams const & bparams,
 
                        if (allowcust)
                                column += d->startTeXParParams(bparams, os,
-                                                           texrow,
                                                            runparams);
                }
 
-               Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
-                                                                : lookupChange(i);
+               Change const & change = runparams.inDeletedInset
+                       ? runparams.changeOfDeletedInset : lookupChange(i);
 
                if (bparams.outputChanges && runningChange != change) {
                        if (open_font) {
@@ -2387,8 +2434,9 @@ void Paragraph::latex(BufferParams const & bparams,
                if (!runparams.pass_thru && !style.pass_thru &&
                    runparams.encoding->package() != Encoding::none &&
                    font.language()->encoding()->package() != Encoding::none) {
-                       pair<bool, int> const enc_switch = switchEncoding(os, bparams,
-                                       runparams, *(font.language()->encoding()));
+                       pair<bool, int> const enc_switch =
+                               switchEncoding(os.os(), bparams, runparams,
+                                       *(font.language()->encoding()));
                        if (enc_switch.first) {
                                column += enc_switch.second;
                                runparams.encoding = font.language()->encoding();
@@ -2431,8 +2479,7 @@ void Paragraph::latex(BufferParams const & bparams,
                        // latexSpecialChar ignores spaces if
                        // style.pass_thru is false.
                        if (i != body_pos - 1) {
-                               if (d->simpleTeXBlanks(
-                                               runparams, os, texrow,
+                               if (d->simpleTeXBlanks(runparams, os,
                                                i, column, font, style)) {
                                        // A surrogate pair was output. We
                                        // must not call latexSpecialChar
@@ -2454,8 +2501,7 @@ void Paragraph::latex(BufferParams const & bparams,
                // and then split to handle the two modes separately.
                if (c == META_INSET) {
                        if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
-                               d->latexInset(bparams, os,
-                                               texrow, rp, running_font,
+                               d->latexInset(bparams, os, rp, running_font,
                                                basefont, outerfont, open_font,
                                                runningChange, style, i, column);
                        }
@@ -2463,7 +2509,7 @@ void Paragraph::latex(BufferParams const & bparams,
                        if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
                                try {
                                        d->latexSpecialChar(os, rp, running_font, runningChange,
-                                               style, i, column);
+                                                           style, i, end_pos, column);
                                } catch (EncodingException & e) {
                                if (runparams.dryrun) {
                                        os << "<" << _("LyX Warning: ")
@@ -2513,7 +2559,7 @@ void Paragraph::latex(BufferParams const & bparams,
                os << "}]~";
        }
 
-       if (allowcust && d->endTeXParParams(bparams, os, texrow, runparams)
+       if (allowcust && d->endTeXParParams(bparams, os, runparams)
            && runparams.encoding != prev_encoding) {
                runparams.encoding = prev_encoding;
                if (!runparams.isFullUnicode())
@@ -2684,10 +2730,11 @@ docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
 
        bool emph_flag = false;
        bool bold_flag = false;
-       string closing_tag;
 
        Layout const & style = *d->layout_;
 
+       xs.startParagraph(allowEmpty());
+
        if (!runparams.for_toc && runparams.html_make_pars) {
                // generate a magic label for this paragraph
                string const attr = "id='" + magicLabel() + "'";
@@ -2759,7 +2806,7 @@ docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
                                // 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;
+                               xs << XHTMLStream::ESCAPE_NONE << str;
                        } else
                                xs << c;
                }
@@ -2767,6 +2814,7 @@ docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
        }
 
        xs.closeFontTags();
+       xs.endParagraph();
        return retval;
 }
 
@@ -2798,13 +2846,47 @@ bool Paragraph::isLineSeparator(pos_type pos) const
 
 bool Paragraph::isWordSeparator(pos_type pos) const
 {
+       if (pos == size())
+               return true;
        if (Inset const * inset = getInset(pos))
                return !inset->isLetter();
+       // if we have a hard hyphen (no en- or emdash) or apostrophe
+       // we pass this to the spell checker
+       // FIXME: this method is subject to change, visit
+       // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
+       // to get an impression how complex this is.
+       if (isHardHyphenOrApostrophe(pos))
+               return false;
+       char_type const c = d->text_[pos];
+       // We want to pass the escape chars to the spellchecker
+       docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
+       return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
+}
+
+
+bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
+{
+       pos_type const psize = size();
+       if (pos >= psize)
+               return false;
        char_type const c = d->text_[pos];
-       // 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();
+       if (c != '-' && c != '\'')
+               return false;
+       int nextpos = pos + 1;
+       int prevpos = pos > 0 ? pos - 1 : 0;
+       if ((nextpos == psize || isSpace(nextpos))
+               && (pos == 0 || isSpace(prevpos)))
+               return false;
+       return c == '\''
+               || ((nextpos == psize || d->text_[nextpos] != '-')
+               && (pos == 0 || d->text_[prevpos] != '-'));
+}
+
+
+bool Paragraph::isSameSpellRange(pos_type pos1, pos_type pos2) const
+{
+       return pos1 == pos2
+               || d->speller_state_.getRange(pos1) == d->speller_state_.getRange(pos2);
 }
 
 
@@ -2813,7 +2895,7 @@ bool Paragraph::isChar(pos_type pos) const
        if (Inset const * inset = getInset(pos))
                return inset->isChar();
        char_type const c = d->text_[pos];
-       return !isLetterChar(c) && !isDigit(c) && !lyx::isSpace(c);
+       return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
 }
 
 
@@ -2853,9 +2935,9 @@ void Paragraph::changeLanguage(BufferParams const & bparams,
                if (font.language() == from) {
                        font.setLanguage(to);
                        setFont(i, font);
+                       d->requestSpellCheck(i);
                }
        }
-       d->requestSpellCheck(size());
 }
 
 
@@ -2911,7 +2993,7 @@ docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
                    || (c == '\n' && (options & AS_STR_NEWLINES)))
                        os.put(c);
                else if (c == META_INSET && (options & AS_STR_INSETS)) {
-                       getInset(i)->tocString(os);
+                       getInset(i)->toString(os);
                        if (getInset(i)->asInsetMath())
                                os << " ";
                }
@@ -2921,6 +3003,24 @@ docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
 }
 
 
+void Paragraph::forToc(docstring & os, size_t maxlen) const
+{
+       if (!d->params_.labelString().empty())
+               os += d->params_.labelString() + ' ';
+       for (pos_type i = 0; i < size() && os.length() < maxlen; ++i) {
+               if (isDeleted(i))
+                       continue;
+               char_type const c = d->text_[i];
+               if (isPrintable(c))
+                       os += c;
+               else if (c == '\t' || c == '\n')
+                       os += ' ';
+               else if (c == META_INSET)
+                       getInset(i)->forToc(os, maxlen);
+       }
+}
+
+
 docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputParams & runparams) const
 {
        odocstringstream os;
@@ -3075,7 +3175,7 @@ int Paragraph::checkBiblio(Buffer const & buffer)
 {
        // FIXME From JS:
        // This is getting more and more a mess. ...We really should clean
-       // up this bibitem issue for 1.6. See also bug 2743.
+       // up this bibitem issue for 1.6.
 
        // Add bibitem insets if necessary
        if (d->layout_->labeltype != LABEL_BIBLIO)
@@ -3271,12 +3371,12 @@ int Paragraph::find(docstring const & str, bool cs, bool mw,
        int i = 0;
        pos_type const parsize = d->text_.size();
        for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
-               // Ignore ligature break and hyphenation chars while searching
+               // Ignore "invisible" letters such as ligature breaks
+               // and hyphenation chars while searching
                while (pos < parsize - 1 && isInset(pos)) {
-                       const InsetSpecialChar *isc = dynamic_cast<const InsetSpecialChar*>(getInset(pos));
-                       if (isc == 0
-                           || (isc->kind() != InsetSpecialChar::HYPHENATION
-                               && isc->kind() != InsetSpecialChar::LIGATURE_BREAK))
+                       odocstringstream os;
+                       getInset(pos)->toString(os);
+                       if (!getInset(pos)->isLetter() || !os.str().empty())
                                break;
                        pos++;
                }
@@ -3514,7 +3614,7 @@ Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
 
 void Paragraph::requestSpellCheck(pos_type pos)
 {
-       d->requestSpellCheck(pos == -1 ? size() : pos);
+       d->requestSpellCheck(pos);
 }
 
 
@@ -3530,6 +3630,21 @@ bool Paragraph::needsSpellCheck() const
 }
 
 
+bool Paragraph::Private::ignoreWord(docstring const & word) const
+{
+       // Ignore words with digits
+       // FIXME: make this customizable
+       // (note that some checkers ignore words with digits by default)
+       docstring::const_iterator cit = word.begin();
+       docstring::const_iterator const end = word.end();
+       for (; cit != end; ++cit) {
+               if (isNumber((*cit)))
+                       return true;
+       }
+       return false;
+}
+
+
 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
        WordLangTuple & wl, docstring_list & suggestions,
        bool do_suggestion, bool check_learned) const
@@ -3546,7 +3661,7 @@ SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
        if (from == to || from >= size())
                return result;
 
-       docstring word = asString(from, to, AS_STR_INSETS + AS_STR_SKIPDELETE);
+       docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
        Language * lang = d->getSpellLanguage(from);
 
        wl = WordLangTuple(word, lang);
@@ -3555,10 +3670,8 @@ SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
                return result;
 
        if (needsSpellCheck() || check_learned) {
-               // Ignore words with digits
-               // FIXME: make this customizable
-               // (note that some checkers ignore words with digits by default)
-               if (!hasDigit(word)) {
+               pos_type end = to;
+               if (!d->ignoreWord(word)) {
                        bool const trailing_dot = to < size() && d->text_[to] == '.';
                        result = speller->check(wl);
                        if (SpellChecker::misspelled(result) && trailing_dot) {
@@ -3568,24 +3681,34 @@ SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
                                        LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
                                           word << "\" [" <<
                                           from << ".." << to << "]");
+                               } else {
+                                       // spell check with dot appended failed too
+                                       // restore original word/lang value
+                                       word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
+                                       wl = WordLangTuple(word, lang);
                                }
                        }
                }
-               d->setMisspelled(from, to, result);
+               if (!SpellChecker::misspelled(result)) {
+                       // area up to the begin of the next word is not misspelled
+                       while (end < size() && isWordSeparator(end))
+                               ++end;
+               }
+               d->setMisspelled(from, end, result);
        } else {
                result = d->speller_state_.getState(from);
        }
 
-       bool const misspelled_ = SpellChecker::misspelled(result) ;
-       if (misspelled_ && do_suggestion)
-               speller->suggest(wl, suggestions);
-       else if (misspelled_)
+       if (do_suggestion)
+               suggestions.clear();
+
+       if (SpellChecker::misspelled(result)) {
                LYXERR(Debug::GUI, "misspelled word: \"" <<
                           word << "\" [" <<
                           from << ".." << to << "]");
-       else
-               suggestions.clear();
-
+               if (do_suggestion)
+                       speller->suggest(wl, suggestions);
+       }
        return result;
 }
 
@@ -3674,9 +3797,14 @@ void Paragraph::spellCheck() const
 }
 
 
-bool Paragraph::isMisspelled(pos_type pos) const
+bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
 {
-       return SpellChecker::misspelled(d->speller_state_.getState(pos));
+       bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
+       if (result || pos <= 0 || pos > size())
+               return result;
+       if (check_boundary && (pos == size() || isWordSeparator(pos)))
+               result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
+       return result;
 }