src/output_xhtml.cpp

   1 /**
   2  * \file output_xhtml.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Richard Heck
   7  *
   8  * This code is based upon output_docbook.cpp
   9  *
  10  * Full author contact details are available in file CREDITS.
  11  */
  12
  13 #include <config.h>
  14
  15 #include "output_xhtml.h"
  16
  17 #include "Buffer.h"
  18 #include "buffer_funcs.h"
  19 #include "BufferParams.h"
  20 #include "Counters.h"
  21 #include "Font.h"
  22 #include "Layout.h"
  23 #include "OutputParams.h"
  24 #include "Paragraph.h"
  25 #include "ParagraphList.h"
  26 #include "ParagraphParameters.h"
  27 #include "sgml.h"
  28 #include "Text.h"
  29 #include "TextClass.h"
  30
  31 #include "support/convert.h"
  32 #include "support/debug.h"
  33 #include "support/lassert.h"
  34 #include "support/lstrings.h"
  35 #include "support/textutils.h"
  36
  37 #include <vector>
  38
  39 // Uncomment to activate debugging code.
  40 // #define XHTML_DEBUG
  41
  42 using namespace std;
  43 using namespace lyx::support;
  44
  45 namespace lyx {
  46
  47 namespace html {
  48
  49 docstring escapeChar(char_type c, XHTMLStream::EscapeSettings e)
  50 {
  51         docstring str;
  52         switch (e) {
  53         case XHTMLStream::ESCAPE_NONE:
  54                 str += c;
  55                 break;
  56         case XHTMLStream::ESCAPE_ALL:
  57                 if (c == '<') {
  58                         str += "&lt;";
  59                         break;
  60                 } else if (c == '>') {
  61                         str += "&gt;";
  62                         break;
  63                 }
  64         // fall through
  65         case XHTMLStream::ESCAPE_AND:
  66                 if (c == '&')
  67                         str += "&amp;";
  68                 else
  69                         str     +=c ;
  70                 break;
  71         }
  72         return str;
  73 }
  74
  75
  76 // escape what needs escaping
  77 docstring htmlize(docstring const & str, XHTMLStream::EscapeSettings e)
  78 {
  79         odocstringstream d;
  80         docstring::const_iterator it = str.begin();
  81         docstring::const_iterator en = str.end();
  82         for (; it != en; ++it)
  83                 d << escapeChar(*it, e);
  84         return d.str();
  85 }
  86
  87
  88 docstring escapeChar(char c, XHTMLStream::EscapeSettings e)
  89 {
  90         LATTEST(static_cast<unsigned char>(c) < 0x80);
  91         return escapeChar(static_cast<char_type>(c), e);
  92 }
  93
  94
  95 docstring cleanAttr(docstring const & str)
  96 {
  97         docstring newname;
  98         docstring::const_iterator it = str.begin();
  99         docstring::const_iterator en = str.end();
 100         for (; it != en; ++it) {
 101                 char_type const c = *it;
 102                 newname += isAlnumASCII(c) ? c : char_type('_');
 103         }
 104         return newname;
 105 }
 106
 107
 108 docstring StartTag::writeTag() const
 109 {
 110         docstring output = '<' + from_utf8(tag_);
 111         if (!attr_.empty())
 112                 output += ' ' + html::htmlize(from_utf8(attr_), XHTMLStream::ESCAPE_NONE);
 113         output += ">";
 114         return output;
 115 }
 116
 117
 118 docstring StartTag::writeEndTag() const
 119 {
 120         string output = "</" + tag_ + ">";
 121         return from_utf8(output);
 122 }
 123
 124
 125 bool StartTag::operator==(FontTag const & rhs) const
 126 {
 127         return rhs == *this;
 128 }
 129
 130
 131 docstring EndTag::writeEndTag() const
 132 {
 133         string output = "</" + tag_ + ">";
 134         return from_utf8(output);
 135 }
 136
 137
 138 ParTag::ParTag(std::string const & tag, std::string attr,
 139        std::string const & parid)
 140   : StartTag(tag)
 141 {
 142         if (!parid.empty())
 143                 attr += " id='" + parid + "'";
 144         attr_ = attr;
 145 }
 146
 147
 148 docstring CompTag::writeTag() const
 149 {
 150         docstring output = '<' + from_utf8(tag_);
 151         if (!attr_.empty())
 152                 output += ' ' + html::htmlize(from_utf8(attr_), XHTMLStream::ESCAPE_NONE);
 153         output += " />";
 154         return output;
 155 }
 156
 157
 158
 159 namespace {
 160
 161 string fontToTag(html::FontTypes type)
 162 {
 163         switch(type) {
 164         case FT_EMPH:
 165                 return "em";
 166         case FT_BOLD:
 167                 return "b";
 168         case FT_NOUN:
 169                 return "dfn";
 170         case FT_UBAR:
 171         case FT_WAVE:
 172         case FT_DBAR:
 173                 return "u";
 174         case FT_SOUT:
 175                 return "del";
 176         case FT_ITALIC:
 177                 return "i";
 178         case FT_UPRIGHT:
 179         case FT_SLANTED:
 180         case FT_SMALLCAPS:
 181         case FT_ROMAN:
 182         case FT_SANS:
 183         case FT_TYPE:
 184         case FT_SIZE_TINY:
 185         case FT_SIZE_SCRIPT:
 186         case FT_SIZE_FOOTNOTE:
 187         case FT_SIZE_SMALL:
 188         case FT_SIZE_NORMAL:
 189         case FT_SIZE_LARGE:
 190         case FT_SIZE_LARGER:
 191         case FT_SIZE_LARGEST:
 192         case FT_SIZE_HUGE:
 193         case FT_SIZE_HUGER:
 194         case FT_SIZE_INCREASE:
 195         case FT_SIZE_DECREASE:
 196                 return "span";
 197         }
 198         // kill warning
 199         return "";
 200 }
 201
 202 string fontToAttribute(html::FontTypes type)
 203 {
 204         switch(type) {
 205         case FT_EMPH:
 206         case FT_BOLD:
 207                 return "";
 208         case FT_NOUN:
 209                 return "class='lyxnoun'";
 210         case FT_UBAR:
 211                 return "";
 212         case FT_DBAR:
 213                 return "class='dline'";
 214         case FT_SOUT:
 215                 return "class='strikeout'";
 216         case FT_WAVE:
 217                 return "class='wline'";
 218         case FT_ITALIC:
 219                 return "";
 220         case FT_UPRIGHT:
 221                 return "style='font-style:normal;'";
 222         case FT_SLANTED:
 223                 return "style='font-style:oblique;'";
 224         case FT_SMALLCAPS:
 225                 return "style='font-variant:small-caps;'";
 226         case FT_ROMAN:
 227                 return "style='font-family:serif;'";
 228         case FT_SANS:
 229                 return "style='font-family:sans-serif;'";
 230         case FT_TYPE:
 231                 return "style='font-family:monospace;'";
 232         case FT_SIZE_TINY:
 233         case FT_SIZE_SCRIPT:
 234         case FT_SIZE_FOOTNOTE:
 235                 return "style='font-size:x-small;'";
 236         case FT_SIZE_SMALL:
 237                 return "style='font-size:small;'";
 238         case FT_SIZE_NORMAL:
 239                 return "style='font-size:normal;'";
 240         case FT_SIZE_LARGE:
 241                 return "style='font-size:large;'";
 242         case FT_SIZE_LARGER:
 243         case FT_SIZE_LARGEST:
 244                 return "style='font-size:x-large;'";
 245         case FT_SIZE_HUGE:
 246         case FT_SIZE_HUGER:
 247                 return "style='font-size:xx-large;'";
 248         case FT_SIZE_INCREASE:
 249                 return "style='font-size:larger;'";
 250         case FT_SIZE_DECREASE:
 251                 return "style='font-size:smaller;'";
 252         }
 253         // kill warning
 254         return "";
 255 }
 256
 257 } // end anonymous namespace
 258
 259
 260 FontTag::FontTag(FontTypes type)
 261   : StartTag(fontToTag(type), fontToAttribute(type)), font_type_(type)
 262 {}
 263
 264
 265 bool FontTag::operator==(StartTag const & tag) const
 266 {
 267         FontTag const * const ftag = tag.asFontTag();
 268         if (!ftag)
 269                 return false;
 270         return (font_type_ == ftag->font_type_);
 271 }
 272
 273
 274 EndFontTag::EndFontTag(FontTypes type)
 275           : EndTag(fontToTag(type)), font_type_(type)
 276 {}
 277
 278 } // namespace html
 279
 280
 281
 282 ////////////////////////////////////////////////////////////////
 283 ///
 284 /// XHTMLStream
 285 ///
 286 ////////////////////////////////////////////////////////////////
 287
 288 XHTMLStream::XHTMLStream(odocstream & os)
 289   : os_(os), escape_(ESCAPE_ALL)
 290 {}
 291
 292
 293 #ifdef XHTML_DEBUG
 294 void XHTMLStream::dumpTagStack(string const & msg)
 295 {
 296         *this << html::CR();
 297         writeError(msg);
 298         *this << html::CR();
 299         writeError("Tag Stack");
 300         TagDeque::const_reverse_iterator it = tag_stack_.rbegin();
 301         TagDeque::const_reverse_iterator en = tag_stack_.rend();
 302         for (; it != en; ++it) {
 303                 writeError(it->get()->tag_);
 304         }
 305         writeError("End Tag Stack");
 306         *this << html::CR();
 307         writeError("Pending Tags");
 308         it = pending_tags_.rbegin();
 309         en = pending_tags_.rend();
 310         for (; it != en; ++it) {
 311                 writeError(it->get()->tag_);
 312         }
 313         writeError("End Pending Tags");
 314         *this << html::CR();
 315 }
 316 #endif
 317
 318
 319 void XHTMLStream::writeError(std::string const & s) const
 320 {
 321         LYXERR0(s);
 322         os_ << from_utf8("<!-- Output Error: " + s + " -->\n");
 323 }
 324
 325
 326 namespace {
 327         // an illegal tag for internal use
 328         static html::StartTag const parsep_tag("&LyX_parsep_tag&");
 329 }
 330
 331
 332 bool XHTMLStream::closeFontTags()
 333 {
 334         if (isTagPending(parsep_tag))
 335                 // we haven't had any content
 336                 return true;
 337
 338 #ifdef  XHTML_DEBUG
 339         dumpTagStack("Beging Close Font Tags");
 340 #endif
 341
 342         // this may be a useless check, since we ought at least to have
 343         // the parsep_tag. but it can't hurt too much to be careful.
 344         if (tag_stack_.empty())
 345                 return true;
 346
 347         // first, we close any open font tags we can close
 348         TagPtr curtag = tag_stack_.back();
 349         while (curtag->asFontTag()) {
 350                 os_ << curtag->writeEndTag();
 351                 tag_stack_.pop_back();
 352                 // this shouldn't happen, since then the font tags
 353                 // weren't in any other tag.
 354                 LASSERT(!tag_stack_.empty(), return true);
 355                 curtag = tag_stack_.back();
 356         }
 357
 358 #ifdef  XHTML_DEBUG
 359         dumpTagStack("End Close Font Tags");
 360 #endif
 361
 362         if (*curtag == parsep_tag)
 363                 return true;
 364
 365         // so we've hit a non-font tag.
 366         writeError("Tags still open in closeFontTags(). Probably not a problem,\n"
 367                    "but you might want to check these tags:");
 368         TagDeque::const_reverse_iterator it = tag_stack_.rbegin();
 369         TagDeque::const_reverse_iterator const en = tag_stack_.rend();
 370         for (; it != en; ++it) {
 371                 if (**it == parsep_tag)
 372                         break;
 373                 writeError((*it)->tag_);
 374         }
 375         return false;
 376 }
 377
 378
 379 void XHTMLStream::startDivision(bool keep_empty)
 380 {
 381         pending_tags_.push_back(makeTagPtr(html::StartTag(parsep_tag)));
 382         if (keep_empty)
 383                 clearTagDeque();
 384 #ifdef  XHTML_DEBUG
 385         dumpTagStack("StartDivision");
 386 #endif
 387 }
 388
 389
 390 void XHTMLStream::endDivision()
 391 {
 392         if (isTagPending(parsep_tag)) {
 393                 // this case is normal. it just means we didn't have content,
 394                 // so the parsep_tag never got moved onto the tag stack.
 395                 while (!pending_tags_.empty()) {
 396                         // clear all pending tags up to and including the parsep tag.
 397                         // note that we work from the back, because we want to get rid
 398                         // of everything that hasn't been used.
 399                         TagPtr const cur_tag = pending_tags_.back();
 400                         pending_tags_.pop_back();
 401                         if (*cur_tag == parsep_tag)
 402                                 break;
 403                 }
 404
 405 #ifdef  XHTML_DEBUG
 406                 dumpTagStack("EndDivision");
 407 #endif
 408
 409                 return;
 410         }
 411
 412         if (!isTagOpen(parsep_tag)) {
 413                 writeError("No division separation tag found in endDivision().");
 414                 return;
 415         }
 416
 417         // this case is also normal, if the parsep tag is the last one
 418         // on the stack. otherwise, it's an error.
 419         while (!tag_stack_.empty()) {
 420                 TagPtr const cur_tag = tag_stack_.back();
 421                 tag_stack_.pop_back();
 422                 if (*cur_tag == parsep_tag)
 423                         break;
 424                 writeError("Tag `" + cur_tag->tag_ + "' still open at end of paragraph. Closing.");
 425                 os_ << cur_tag->writeEndTag();
 426         }
 427
 428 #ifdef  XHTML_DEBUG
 429         dumpTagStack("EndDivision");
 430 #endif
 431 }
 432
 433
 434 void XHTMLStream::clearTagDeque()
 435 {
 436         while (!pending_tags_.empty()) {
 437                 TagPtr const tag = pending_tags_.front();
 438                 if (*tag != parsep_tag)
 439                         // tabs?
 440                         os_ << tag->writeTag();
 441                 tag_stack_.push_back(tag);
 442                 pending_tags_.pop_front();
 443         }
 444 }
 445
 446
 447 XHTMLStream & XHTMLStream::operator<<(docstring const & d)
 448 {
 449         clearTagDeque();
 450         os_ << html::htmlize(d, escape_);
 451         escape_ = ESCAPE_ALL;
 452         return *this;
 453 }
 454
 455
 456 XHTMLStream & XHTMLStream::operator<<(const char * s)
 457 {
 458         clearTagDeque();
 459         docstring const d = from_ascii(s);
 460         os_ << html::htmlize(d, escape_);
 461         escape_ = ESCAPE_ALL;
 462         return *this;
 463 }
 464
 465
 466 XHTMLStream & XHTMLStream::operator<<(char_type c)
 467 {
 468         clearTagDeque();
 469         os_ << html::escapeChar(c, escape_);
 470         escape_ = ESCAPE_ALL;
 471         return *this;
 472 }
 473
 474
 475 XHTMLStream & XHTMLStream::operator<<(char c)
 476 {
 477         clearTagDeque();
 478         os_ << html::escapeChar(c, escape_);
 479         escape_ = ESCAPE_ALL;
 480         return *this;
 481 }
 482
 483
 484 XHTMLStream & XHTMLStream::operator<<(int i)
 485 {
 486         clearTagDeque();
 487         os_ << i;
 488         escape_ = ESCAPE_ALL;
 489         return *this;
 490 }
 491
 492
 493 XHTMLStream & XHTMLStream::operator<<(EscapeSettings e)
 494 {
 495         escape_ = e;
 496         return *this;
 497 }
 498
 499
 500 XHTMLStream & XHTMLStream::operator<<(html::StartTag const & tag)
 501 {
 502         if (tag.tag_.empty())
 503                 return *this;
 504         pending_tags_.push_back(makeTagPtr(tag));
 505         if (tag.keepempty_)
 506                 clearTagDeque();
 507         return *this;
 508 }
 509
 510
 511 XHTMLStream & XHTMLStream::operator<<(html::ParTag const & tag)
 512 {
 513         if (tag.tag_.empty())
 514                 return *this;
 515         pending_tags_.push_back(makeTagPtr(tag));
 516         return *this;
 517 }
 518
 519
 520 XHTMLStream & XHTMLStream::operator<<(html::CompTag const & tag)
 521 {
 522         if (tag.tag_.empty())
 523                 return *this;
 524         clearTagDeque();
 525         os_ << tag.writeTag();
 526         *this << html::CR();
 527         return *this;
 528 }
 529
 530
 531 XHTMLStream & XHTMLStream::operator<<(html::FontTag const & tag)
 532 {
 533         if (tag.tag_.empty())
 534                 return *this;
 535         pending_tags_.push_back(makeTagPtr(tag));
 536         return *this;
 537 }
 538
 539
 540 XHTMLStream & XHTMLStream::operator<<(html::CR const &)
 541 {
 542         // tabs?
 543         os_ << from_ascii("\n");
 544         return *this;
 545 }
 546
 547
 548 bool XHTMLStream::isTagOpen(html::StartTag const & stag) const
 549 {
 550         TagDeque::const_iterator sit = tag_stack_.begin();
 551         TagDeque::const_iterator const sen = tag_stack_.end();
 552         for (; sit != sen; ++sit)
 553                 if (**sit == stag)
 554                         return true;
 555         return false;
 556 }
 557
 558
 559 bool XHTMLStream::isTagOpen(html::EndTag const & etag) const
 560 {
 561         TagDeque::const_iterator sit = tag_stack_.begin();
 562         TagDeque::const_iterator const sen = tag_stack_.end();
 563         for (; sit != sen; ++sit)
 564                 if (etag == **sit)
 565                         return true;
 566         return false;
 567 }
 568
 569
 570 bool XHTMLStream::isTagPending(html::StartTag const & stag) const
 571 {
 572         TagDeque::const_iterator sit = pending_tags_.begin();
 573         TagDeque::const_iterator const sen = pending_tags_.end();
 574         for (; sit != sen; ++sit)
 575                 if (**sit == stag)
 576                         return true;
 577         return false;
 578 }
 579
 580
 581 // this is complicated, because we want to make sure that
 582 // everything is properly nested. the code ought to make
 583 // sure of that, but we won't assert (yet) if we run into
 584 // a problem. we'll just output error messages and try our
 585 // best to make things work.
 586 XHTMLStream & XHTMLStream::operator<<(html::EndTag const & etag)
 587 {
 588         if (etag.tag_.empty())
 589                 return *this;
 590
 591         // if this tag is pending, we can simply discard it.
 592         if (!pending_tags_.empty()) {
 593
 594                 if (etag == *pending_tags_.back()) {
 595                         // we have <tag></tag>, so we discard it and remove it
 596                         // from the pending_tags_.
 597                         pending_tags_.pop_back();
 598                         return *this;
 599                 }
 600
 601                 // there is a pending tag that isn't the one we are trying
 602                 // to close.
 603
 604                 // is this tag itself pending?
 605                 // non-const iterators because we may call erase().
 606                 TagDeque::iterator dit = pending_tags_.begin();
 607                 TagDeque::iterator const den = pending_tags_.end();
 608                 for (; dit != den; ++dit) {
 609                         if (etag == **dit) {
 610                                 // it was pending, so we just erase it
 611                                 writeError("Tried to close pending tag `" + etag.tag_
 612                                         + "' when other tags were pending. Last pending tag is `"
 613                                         + to_utf8(pending_tags_.back()->writeTag())
 614                                         + "'. Tag discarded.");
 615                                 pending_tags_.erase(dit);
 616                                 return *this;
 617                         }
 618                 }
 619                 // so etag isn't itself pending. is it even open?
 620                 if (!isTagOpen(etag)) {
 621                         writeError("Tried to close `" + etag.tag_
 622                                  + "' when tag was not open. Tag discarded.");
 623                         return *this;
 624                 }
 625                 // ok, so etag is open.
 626                 // our strategy will be as below: we will do what we need to
 627                 // do to close this tag.
 628                 string estr = "Closing tag `" + etag.tag_
 629                         + "' when other tags are pending. Discarded pending tags:\n";
 630                 for (dit = pending_tags_.begin(); dit != den; ++dit)
 631                         estr += to_utf8(html::htmlize((*dit)->writeTag(), XHTMLStream::ESCAPE_ALL)) + "\n";
 632                 writeError(estr);
 633                 // clear the pending tags...
 634                 pending_tags_.clear();
 635                 // ...and then just fall through.
 636         }
 637
 638         // make sure there are tags to be closed
 639         if (tag_stack_.empty()) {
 640                 writeError("Tried to close `" + etag.tag_
 641                          + "' when no tags were open!");
 642                 return *this;
 643         }
 644
 645         // is the tag we are closing the last one we opened?
 646         if (etag == *tag_stack_.back()) {
 647                 // output it...
 648                 os_ << etag.writeEndTag();
 649                 // ...and forget about it
 650                 tag_stack_.pop_back();
 651                 return *this;
 652         }
 653
 654         // we are trying to close a tag other than the one last opened.
 655         // let's first see if this particular tag is still open somehow.
 656         if (!isTagOpen(etag)) {
 657                 writeError("Tried to close `" + etag.tag_
 658                         + "' when tag was not open. Tag discarded.");
 659                 return *this;
 660         }
 661
 662         // so the tag was opened, but other tags have been opened since
 663         // and not yet closed.
 664         // if it's a font tag, though...
 665         if (etag.asFontTag()) {
 666                 // it won't be a problem if the other tags open since this one
 667                 // are also font tags.
 668                 TagDeque::const_reverse_iterator rit = tag_stack_.rbegin();
 669                 TagDeque::const_reverse_iterator ren = tag_stack_.rend();
 670                 for (; rit != ren; ++rit) {
 671                         if (etag == **rit)
 672                                 break;
 673                         if (!(*rit)->asFontTag()) {
 674                                 // we'll just leave it and, presumably, have to close it later.
 675                                 writeError("Unable to close font tag `" + etag.tag_
 676                                         + "' due to open non-font tag `" + (*rit)->tag_ + "'.");
 677                                 return *this;
 678                         }
 679                 }
 680
 681                 // so we have e.g.:
 682                 //    <em>this is <strong>bold
 683                 // and are being asked to closed em. we want:
 684                 //    <em>this is <strong>bold</strong></em><strong>
 685                 // first, we close the intervening tags...
 686                 TagPtr curtag = tag_stack_.back();
 687                 // ...remembering them in a stack.
 688                 TagDeque fontstack;
 689                 while (etag != *curtag) {
 690                         os_ << curtag->writeEndTag();
 691                         fontstack.push_back(curtag);
 692                         tag_stack_.pop_back();
 693                         curtag = tag_stack_.back();
 694                 }
 695                 os_ << etag.writeEndTag();
 696                 tag_stack_.pop_back();
 697
 698                 // ...and restore the other tags.
 699                 rit = fontstack.rbegin();
 700                 ren = fontstack.rend();
 701                 for (; rit != ren; ++rit)
 702                         pending_tags_.push_back(*rit);
 703                 return *this;
 704         }
 705
 706         // it wasn't a font tag.
 707         // so other tags were opened before this one and not properly closed.
 708         // so we'll close them, too. that may cause other issues later, but it
 709         // at least guarantees proper nesting.
 710         writeError("Closing tag `" + etag.tag_
 711                 + "' when other tags are open, namely:");
 712         TagPtr curtag = tag_stack_.back();
 713         while (etag != *curtag) {
 714                 writeError(curtag->tag_);
 715                 if (*curtag != parsep_tag)
 716                         os_ << curtag->writeEndTag();
 717                 tag_stack_.pop_back();
 718                 curtag = tag_stack_.back();
 719         }
 720         // curtag is now the one we actually want.
 721         os_ << curtag->writeEndTag();
 722         tag_stack_.pop_back();
 723
 724         return *this;
 725 }
 726
 727 // End code for XHTMLStream
 728
 729 namespace {
 730
 731 // convenience functions
 732
 733 inline void openParTag(XHTMLStream & xs, Layout const & lay,
 734                        std::string parlabel)
 735 {
 736         xs << html::ParTag(lay.htmltag(), lay.htmlattr(), parlabel);
 737 }
 738
 739
 740 void openParTag(XHTMLStream & xs, Layout const & lay,
 741                 ParagraphParameters const & params,
 742                 std::string parlabel)
 743 {
 744         // FIXME Are there other things we should handle here?
 745         string const align = alignmentToCSS(params.align());
 746         if (align.empty()) {
 747                 openParTag(xs, lay, parlabel);
 748                 return;
 749         }
 750         string attrs = lay.htmlattr() + " style='text-align: " + align + ";'";
 751         xs << html::ParTag(lay.htmltag(), attrs, parlabel);
 752 }
 753
 754
 755 inline void closeTag(XHTMLStream & xs, Layout const & lay)
 756 {
 757         xs << html::EndTag(lay.htmltag());
 758 }
 759
 760
 761 inline void openLabelTag(XHTMLStream & xs, Layout const & lay)
 762 {
 763         xs << html::StartTag(lay.htmllabeltag(), lay.htmllabelattr());
 764 }
 765
 766
 767 inline void closeLabelTag(XHTMLStream & xs, Layout const & lay)
 768 {
 769         xs << html::EndTag(lay.htmllabeltag());
 770 }
 771
 772
 773 inline void openItemTag(XHTMLStream & xs, Layout const & lay)
 774 {
 775         xs << html::StartTag(lay.htmlitemtag(), lay.htmlitemattr(), true);
 776 }
 777
 778
 779 void openItemTag(XHTMLStream & xs, Layout const & lay,
 780              ParagraphParameters const & params)
 781 {
 782         // FIXME Are there other things we should handle here?
 783         string const align = alignmentToCSS(params.align());
 784         if (align.empty()) {
 785                 openItemTag(xs, lay);
 786                 return;
 787         }
 788         string attrs = lay.htmlattr() + " style='text-align: " + align + ";'";
 789         xs << html::StartTag(lay.htmlitemtag(), attrs);
 790 }
 791
 792
 793 inline void closeItemTag(XHTMLStream & xs, Layout const & lay)
 794 {
 795         xs << html::EndTag(lay.htmlitemtag());
 796 }
 797
 798 // end of convenience functions
 799
 800 ParagraphList::const_iterator findLastParagraph(
 801         ParagraphList::const_iterator p,
 802         ParagraphList::const_iterator const & pend)
 803 {
 804         for (++p; p != pend && p->layout().latextype == LATEX_PARAGRAPH; ++p)
 805                 ;
 806
 807         return p;
 808 }
 809
 810
 811 ParagraphList::const_iterator findEndOfEnvironment(
 812                 ParagraphList::const_iterator const & pstart,
 813                 ParagraphList::const_iterator const & pend)
 814 {
 815         ParagraphList::const_iterator p = pstart;
 816         Layout const & bstyle = p->layout();
 817         size_t const depth = p->params().depth();
 818         for (++p; p != pend; ++p) {
 819                 Layout const & style = p->layout();
 820                 // It shouldn't happen that e.g. a section command occurs inside
 821                 // a quotation environment, at a higher depth, but as of 6/2009,
 822                 // it can happen. We pretend that it's just at lowest depth.
 823                 if (style.latextype == LATEX_COMMAND)
 824                         return p;
 825
 826                 // If depth is down, we're done
 827                 if (p->params().depth() < depth)
 828                         return p;
 829
 830                 // If depth is up, we're not done
 831                 if (p->params().depth() > depth)
 832                         continue;
 833
 834                 // FIXME I am not sure about the first check.
 835                 // Surely we *could* have different layouts that count as
 836                 // LATEX_PARAGRAPH, right?
 837                 if (style.latextype == LATEX_PARAGRAPH || style != bstyle)
 838                         return p;
 839         }
 840         return pend;
 841 }
 842
 843
 844 ParagraphList::const_iterator makeParagraphs(Buffer const & buf,
 845                                             XHTMLStream & xs,
 846                                             OutputParams const & runparams,
 847                                             Text const & text,
 848                                             ParagraphList::const_iterator const & pbegin,
 849                                             ParagraphList::const_iterator const & pend)
 850 {
 851         ParagraphList::const_iterator const begin = text.paragraphs().begin();
 852         ParagraphList::const_iterator par = pbegin;
 853         for (; par != pend; ++par) {
 854                 Layout const & lay = par->layout();
 855                 if (!lay.counter.empty())
 856                         buf.masterBuffer()->params().
 857                             documentClass().counters().step(lay.counter, OutputUpdate);
 858
 859                 // FIXME We should see if there's a label to be output and
 860                 // do something with it.
 861                 if (par != pbegin)
 862                         xs << html::CR();
 863
 864                 // We want to open the paragraph tag if:
 865                 //   (i) the current layout permits multiple paragraphs
 866                 //  (ii) we are either not already inside a paragraph (HTMLIsBlock) OR
 867                 //       we are, but this is not the first paragraph
 868                 //
 869                 // But there is also a special case, and we first see whether we are in it.
 870                 // We do not want to open the paragraph tag if this paragraph contains
 871                 // only one item, and that item is "inline", i.e., not HTMLIsBlock (such
 872                 // as a branch). On the other hand, if that single item has a font change
 873                 // applied to it, then we still do need to open the paragraph.
 874                 //
 875                 // Obviously, this is very fragile. The main reason we need to do this is
 876                 // because of branches, e.g., a branch that contains an entire new section.
 877                 // We do not really want to wrap that whole thing in a <div>...</div>.
 878                 bool special_case = false;
 879                 Inset const * specinset = par->size() == 1 ? par->getInset(0) : 0;
 880                 if (specinset && !specinset->getLayout().htmlisblock()) {
 881                         Layout const & style = par->layout();
 882                         FontInfo const first_font = style.labeltype == LABEL_MANUAL ?
 883                                                 style.labelfont : style.font;
 884                         FontInfo const our_font =
 885                                 par->getFont(buf.masterBuffer()->params(), 0,
 886                                        text.outerFont(distance(begin, par))).fontInfo();
 887                         if (first_font == our_font)
 888                                 special_case = true;
 889                 }
 890
 891                 bool const open_par = runparams.html_make_pars
 892                         && (!runparams.html_in_par || par != pbegin)
 893                         && !special_case;
 894
 895                 // We want to issue the closing tag if either:
 896                 //   (i)  We opened it, and either html_in_par is false,
 897                 //        or we're not in the last paragraph, anyway.
 898                 //   (ii) We didn't open it and html_in_par is true,
 899                 //        but we are in the first par, and there is a next par.
 900                 ParagraphList::const_iterator nextpar = par;
 901                 ++nextpar;
 902                 bool const close_par =
 903                         (open_par && (!runparams.html_in_par || nextpar != pend))
 904                         || (!open_par && runparams.html_in_par && par == pbegin && nextpar != pend);
 905
 906                 if (open_par) {
 907                         // We do not issue the paragraph id if we are doing
 908                         // this for the TOC (or some similar purpose)
 909                         openParTag(xs, lay, par->params(),
 910                                    runparams.for_toc ? "" : par->magicLabel());
 911                 }
 912
 913                 docstring const deferred = par->simpleLyXHTMLOnePar(buf, xs,
 914                         runparams, text.outerFont(distance(begin, par)),
 915                         open_par, close_par);
 916
 917                 if (close_par) {
 918                         closeTag(xs, lay);
 919                         xs << html::CR();
 920                 }
 921
 922                 if (!deferred.empty()) {
 923                         xs << XHTMLStream::ESCAPE_NONE << deferred << html::CR();
 924                 }
 925         }
 926         return pend;
 927 }
 928
 929
 930 ParagraphList::const_iterator makeBibliography(Buffer const & buf,
 931                                 XHTMLStream & xs,
 932                                 OutputParams const & runparams,
 933                                 Text const & text,
 934                                 ParagraphList::const_iterator const & pbegin,
 935                                 ParagraphList::const_iterator const & pend)
 936 {
 937         // FIXME XHTML
 938         // Use TextClass::htmlTOCLayout() to figure out how we should look.
 939         xs << html::StartTag("h2", "class='bibliography'")
 940            << pbegin->layout().labelstring(false)
 941            << html::EndTag("h2")
 942            << html::CR()
 943            << html::StartTag("div", "class='bibliography'")
 944            << html::CR();
 945         makeParagraphs(buf, xs, runparams, text, pbegin, pend);
 946         xs << html::EndTag("div");
 947         return pend;
 948 }
 949
 950
 951 bool isNormalEnv(Layout const & lay)
 952 {
 953         return lay.latextype == LATEX_ENVIRONMENT
 954             || lay.latextype == LATEX_BIB_ENVIRONMENT;
 955 }
 956
 957
 958 ParagraphList::const_iterator makeEnvironment(Buffer const & buf,
 959                                               XHTMLStream & xs,
 960                                               OutputParams const & runparams,
 961                                               Text const & text,
 962                                               ParagraphList::const_iterator const & pbegin,
 963                                               ParagraphList::const_iterator const & pend)
 964 {
 965         ParagraphList::const_iterator const begin = text.paragraphs().begin();
 966         ParagraphList::const_iterator par = pbegin;
 967         Layout const & bstyle = par->layout();
 968         depth_type const origdepth = pbegin->params().depth();
 969
 970         // open tag for this environment
 971         openParTag(xs, bstyle, pbegin->magicLabel());
 972         xs << html::CR();
 973
 974         // we will on occasion need to remember a layout from before.
 975         Layout const * lastlay = 0;
 976
 977         while (par != pend) {
 978                 Layout const & style = par->layout();
 979                 // the counter only gets stepped if we're in some kind of list,
 980                 // or if it's the first time through.
 981                 // note that enum, etc, are handled automatically.
 982                 // FIXME There may be a bug here about user defined enumeration
 983                 // types. If so, then we'll need to take the counter and add "i",
 984                 // "ii", etc, as with enum.
 985                 Counters & cnts = buf.masterBuffer()->params().documentClass().counters();
 986                 docstring const & cntr = style.counter;
 987                 if (!style.counter.empty()
 988                     && (par == pbegin || !isNormalEnv(style))
 989                                 && cnts.hasCounter(cntr)
 990                 )
 991                         cnts.step(cntr, OutputUpdate);
 992                 ParagraphList::const_iterator send;
 993
 994                 switch (style.latextype) {
 995                 case LATEX_ENVIRONMENT:
 996                 case LATEX_LIST_ENVIRONMENT:
 997                 case LATEX_ITEM_ENVIRONMENT: {
 998                         // There are two possiblities in this case.
 999                         // One is that we are still in the environment in which we
1000                         // started---which we will be if the depth is the same.
1001                         if (par->params().depth() == origdepth) {
1002                                 LATTEST(bstyle == style);
1003                                 if (lastlay != 0) {
1004                                         closeItemTag(xs, *lastlay);
1005                                         lastlay = 0;
1006                                 }
1007
1008                                 // this will be positive, if we want to skip the
1009                                 // initial word (if it's been taken for the label).
1010                                 pos_type sep = 0;
1011                                 bool const labelfirst = style.htmllabelfirst();
1012                                 if (!labelfirst)
1013                                         openItemTag(xs, style, par->params());
1014
1015                                 // label output
1016                                 if (style.labeltype != LABEL_NO_LABEL &&
1017                                     style.htmllabeltag() != "NONE") {
1018                                         if (isNormalEnv(style)) {
1019                                                 // in this case, we print the label only for the first
1020                                                 // paragraph (as in a theorem).
1021                                                 if (par == pbegin) {
1022                                                         docstring const lbl =
1023                                                                         pbegin->params().labelString();
1024                                                         if (!lbl.empty()) {
1025                                                                 openLabelTag(xs, style);
1026                                                                 xs << lbl;
1027                                                                 closeLabelTag(xs, style);
1028                                                         }
1029                                                         xs << html::CR();
1030                                                 }
1031                                         } else { // some kind of list
1032                                                 if (style.labeltype == LABEL_MANUAL) {
1033                                                         openLabelTag(xs, style);
1034                                                         sep = par->firstWordLyXHTML(xs, runparams);
1035                                                         closeLabelTag(xs, style);
1036                                                         xs << html::CR();
1037                                                 }
1038                                                 else {
1039                                                         openLabelTag(xs, style);
1040                                                         xs << par->params().labelString();
1041                                                         closeLabelTag(xs, style);
1042                                                         xs << html::CR();
1043                                                 }
1044                                         }
1045                                 } // end label output
1046
1047                                 if (labelfirst)
1048                                         openItemTag(xs, style, par->params());
1049
1050                                 docstring deferred = par->simpleLyXHTMLOnePar(buf, xs, runparams,
1051                                         text.outerFont(distance(begin, par)), true, true, sep);
1052                                 xs << XHTMLStream::ESCAPE_NONE << deferred;
1053                                 ++par;
1054
1055                                 // We may not want to close the tag yet, in particular:
1056                                 // If we're not at the end...
1057                                 if (par != pend
1058                                         //  and are doing items...
1059                                          && !isNormalEnv(style)
1060                                          // and if the depth has changed...
1061                                          && par->params().depth() != origdepth) {
1062                                          // then we'll save this layout for later, and close it when
1063                                          // we get another item.
1064                                         lastlay = &style;
1065                                 } else
1066                                         closeItemTag(xs, style);
1067                                 xs << html::CR();
1068                         }
1069                         // The other possibility is that the depth has increased, in which
1070                         // case we need to recurse.
1071                         else {
1072                                 send = findEndOfEnvironment(par, pend);
1073                                 par = makeEnvironment(buf, xs, runparams, text, par, send);
1074                         }
1075                         break;
1076                 }
1077                 case LATEX_PARAGRAPH:
1078                         send = findLastParagraph(par, pend);
1079                         par = makeParagraphs(buf, xs, runparams, text, par, send);
1080                         break;
1081                 // Shouldn't happen
1082                 case LATEX_BIB_ENVIRONMENT:
1083                         send = par;
1084                         ++send;
1085                         par = makeParagraphs(buf, xs, runparams, text, par, send);
1086                         break;
1087                 // Shouldn't happen
1088                 case LATEX_COMMAND:
1089                         ++par;
1090                         break;
1091                 }
1092         }
1093
1094         if (lastlay != 0)
1095                 closeItemTag(xs, *lastlay);
1096         closeTag(xs, bstyle);
1097         xs << html::CR();
1098         return pend;
1099 }
1100
1101
1102 void makeCommand(Buffer const & buf,
1103                  XHTMLStream & xs,
1104                  OutputParams const & runparams,
1105                  Text const & text,
1106                  ParagraphList::const_iterator const & pbegin)
1107 {
1108         Layout const & style = pbegin->layout();
1109         if (!style.counter.empty())
1110                 buf.masterBuffer()->params().
1111                     documentClass().counters().step(style.counter, OutputUpdate);
1112
1113         bool const make_parid = !runparams.for_toc && runparams.html_make_pars;
1114
1115         if (style.labeltype == LABEL_ABOVE)
1116                 xs << html::StartTag("div")
1117                    << pbegin->params().labelString()
1118                    << html::EndTag("div");
1119         else if (style.labeltype == LABEL_CENTERED)
1120                 xs << html::StartTag("div", "style = \"text-align: center;\"")
1121                    << pbegin->params().labelString()
1122                    << html::EndTag("div");
1123
1124         openParTag(xs, style, pbegin->params(),
1125                    make_parid ? pbegin->magicLabel() : "");
1126
1127         // Label around sectioning number:
1128         // FIXME Probably need to account for LABEL_MANUAL
1129         if (style.labeltype != LABEL_NO_LABEL &&
1130             style.labeltype != LABEL_ABOVE &&
1131             style.labeltype != LABEL_CENTERED ) {
1132                 openLabelTag(xs, style);
1133                 xs << pbegin->params().labelString();
1134                 closeLabelTag(xs, style);
1135                 // Otherwise the label might run together with the text
1136                 xs << from_ascii(" ");
1137         }
1138
1139         ParagraphList::const_iterator const begin = text.paragraphs().begin();
1140         pbegin->simpleLyXHTMLOnePar(buf, xs, runparams,
1141                         text.outerFont(distance(begin, pbegin)));
1142         closeTag(xs, style);
1143         xs << html::CR();
1144 }
1145
1146 } // end anonymous namespace
1147
1148
1149 void xhtmlParagraphs(Text const & text,
1150                        Buffer const & buf,
1151                        XHTMLStream & xs,
1152                        OutputParams const & runparams)
1153 {
1154         ParagraphList const & paragraphs = text.paragraphs();
1155         if (runparams.par_begin == runparams.par_end) {
1156                 runparams.par_begin = 0;
1157                 runparams.par_end = paragraphs.size();
1158         }
1159         pit_type bpit = runparams.par_begin;
1160         pit_type const epit = runparams.par_end;
1161         LASSERT(bpit < epit,
1162                 { xs << XHTMLStream::ESCAPE_NONE << "<!-- XHTML output error! -->\n"; return; });
1163
1164         OutputParams ourparams = runparams;
1165         ParagraphList::const_iterator const pend =
1166                 (epit == (int) paragraphs.size()) ?
1167                         paragraphs.end() : paragraphs.constIterator(epit);
1168         while (bpit < epit) {
1169                 ParagraphList::const_iterator par = paragraphs.constIterator(bpit);
1170                 if (par->params().startOfAppendix()) {
1171                         // We want to reset the counter corresponding to toplevel sectioning
1172                         Layout const & lay =
1173                                 buf.masterBuffer()->params().documentClass().getTOCLayout();
1174                         docstring const cnt = lay.counter;
1175                         if (!cnt.empty()) {
1176                                 Counters & cnts =
1177                                         buf.masterBuffer()->params().documentClass().counters();
1178                                 cnts.reset(cnt);
1179                         }
1180                 }
1181                 Layout const & style = par->layout();
1182                 ParagraphList::const_iterator const lastpar = par;
1183                 ParagraphList::const_iterator send;
1184
1185                 switch (style.latextype) {
1186                 case LATEX_COMMAND: {
1187                         // The files with which we are working never have more than
1188                         // one paragraph in a command structure.
1189                         // FIXME
1190                         // if (ourparams.html_in_par)
1191                         //   fix it so we don't get sections inside standard, e.g.
1192                         // note that we may then need to make runparams not const, so we
1193                         // can communicate that back.
1194                         // FIXME Maybe this fix should be in the routines themselves, in case
1195                         // they are called from elsewhere.
1196                         makeCommand(buf, xs, ourparams, text, par);
1197                         ++par;
1198                         break;
1199                 }
1200                 case LATEX_ENVIRONMENT:
1201                 case LATEX_LIST_ENVIRONMENT:
1202                 case LATEX_ITEM_ENVIRONMENT: {
1203                         // FIXME Same fix here.
1204                         send = findEndOfEnvironment(par, pend);
1205                         par = makeEnvironment(buf, xs, ourparams, text, par, send);
1206                         break;
1207                 }
1208                 case LATEX_BIB_ENVIRONMENT: {
1209                         // FIXME Same fix here.
1210                         send = findEndOfEnvironment(par, pend);
1211                         par = makeBibliography(buf, xs, ourparams, text, par, send);
1212                         break;
1213                 }
1214                 case LATEX_PARAGRAPH:
1215                         send = findLastParagraph(par, pend);
1216                         par = makeParagraphs(buf, xs, ourparams, text, par, send);
1217                         break;
1218                 }
1219                 bpit += distance(lastpar, par);
1220         }
1221 }
1222
1223
1224 string alignmentToCSS(LyXAlignment align)
1225 {
1226         switch (align) {
1227         case LYX_ALIGN_BLOCK:
1228                 // we are NOT going to use text-align: justify!!
1229         case LYX_ALIGN_LEFT:
1230                 return "left";
1231         case LYX_ALIGN_RIGHT:
1232                 return "right";
1233         case LYX_ALIGN_CENTER:
1234                 return "center";
1235         default:
1236                 break;
1237         }
1238         return "";
1239 }
1240
1241 } // namespace lyx