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         // this may be a useless check, since we ought at least to have
 339         // the parsep_tag. but it can't hurt too much to be careful.
 340         if (tag_stack_.empty())
 341                 return true;
 342
 343         // first, we close any open font tags we can close
 344         TagPtr curtag = tag_stack_.back();
 345         while (curtag->asFontTag()) {
 346                 os_ << curtag->writeEndTag();
 347                 tag_stack_.pop_back();
 348                 // this shouldn't happen, since then the font tags
 349                 // weren't in any other tag.
 350                 LBUFERR(!tag_stack_.empty());
 351                 curtag = tag_stack_.back();
 352         }
 353
 354         if (*curtag == parsep_tag)
 355                 return true;
 356
 357         // so we've hit a non-font tag.
 358         writeError("Tags still open in closeFontTags(). Probably not a problem,\n"
 359                    "but you might want to check these tags:");
 360         TagDeque::const_reverse_iterator it = tag_stack_.rbegin();
 361         TagDeque::const_reverse_iterator const en = tag_stack_.rend();
 362         for (; it != en; ++it) {
 363                 if (**it == parsep_tag)
 364                         break;
 365                 writeError((*it)->tag_);
 366         }
 367         return false;
 368 }
 369
 370
 371 void XHTMLStream::startDivision(bool keep_empty)
 372 {
 373         pending_tags_.push_back(makeTagPtr(html::StartTag(parsep_tag)));
 374         if (keep_empty)
 375                 clearTagDeque();
 376 }
 377
 378
 379 void XHTMLStream::endDivision()
 380 {
 381         if (isTagPending(parsep_tag)) {
 382                 // this case is normal. it just means we didn't have content,
 383                 // so the parsep_tag never got moved onto the tag stack.
 384                 while (!pending_tags_.empty()) {
 385                         // clear all pending tags up to and including the parsep tag.
 386                         // note that we work from the back, because we want to get rid
 387                         // of everything that hasn't been used.
 388                         TagPtr const cur_tag = pending_tags_.back();
 389                         pending_tags_.pop_back();
 390                         if (*cur_tag == parsep_tag)
 391                                 break;
 392                 }
 393                 return;
 394         }
 395
 396         if (!isTagOpen(parsep_tag)) {
 397                 writeError("No division separation tag found in endDivision().");
 398                 return;
 399         }
 400
 401         // this case is also normal, if the parsep tag is the last one
 402         // on the stack. otherwise, it's an error.
 403         while (!tag_stack_.empty()) {
 404                 TagPtr const cur_tag = tag_stack_.back();
 405                 tag_stack_.pop_back();
 406                 if (*cur_tag == parsep_tag)
 407                         break;
 408                 writeError("Tag `" + cur_tag->tag_ + "' still open at end of paragraph. Closing.");
 409                 os_ << cur_tag->writeEndTag();
 410         }
 411 }
 412
 413
 414 void XHTMLStream::clearTagDeque()
 415 {
 416         while (!pending_tags_.empty()) {
 417                 TagPtr const tag = pending_tags_.front();
 418                 if (*tag != parsep_tag)
 419                         // tabs?
 420                         os_ << tag->writeTag();
 421                 tag_stack_.push_back(tag);
 422                 pending_tags_.pop_front();
 423         }
 424 }
 425
 426
 427 XHTMLStream & XHTMLStream::operator<<(docstring const & d)
 428 {
 429         clearTagDeque();
 430         os_ << html::htmlize(d, escape_);
 431         escape_ = ESCAPE_ALL;
 432         return *this;
 433 }
 434
 435
 436 XHTMLStream & XHTMLStream::operator<<(const char * s)
 437 {
 438         clearTagDeque();
 439         docstring const d = from_ascii(s);
 440         os_ << html::htmlize(d, escape_);
 441         escape_ = ESCAPE_ALL;
 442         return *this;
 443 }
 444
 445
 446 XHTMLStream & XHTMLStream::operator<<(char_type c)
 447 {
 448         clearTagDeque();
 449         os_ << html::escapeChar(c, escape_);
 450         escape_ = ESCAPE_ALL;
 451         return *this;
 452 }
 453
 454
 455 XHTMLStream & XHTMLStream::operator<<(char c)
 456 {
 457         clearTagDeque();
 458         os_ << html::escapeChar(c, escape_);
 459         escape_ = ESCAPE_ALL;
 460         return *this;
 461 }
 462
 463
 464 XHTMLStream & XHTMLStream::operator<<(int i)
 465 {
 466         clearTagDeque();
 467         os_ << i;
 468         escape_ = ESCAPE_ALL;
 469         return *this;
 470 }
 471
 472
 473 XHTMLStream & XHTMLStream::operator<<(EscapeSettings e)
 474 {
 475         escape_ = e;
 476         return *this;
 477 }
 478
 479
 480 XHTMLStream & XHTMLStream::operator<<(html::StartTag const & tag)
 481 {
 482         if (tag.tag_.empty())
 483                 return *this;
 484         pending_tags_.push_back(makeTagPtr(tag));
 485         if (tag.keepempty_)
 486                 clearTagDeque();
 487         return *this;
 488 }
 489
 490
 491 XHTMLStream & XHTMLStream::operator<<(html::ParTag const & tag)
 492 {
 493         if (tag.tag_.empty())
 494                 return *this;
 495         pending_tags_.push_back(makeTagPtr(tag));
 496         return *this;
 497 }
 498
 499
 500 XHTMLStream & XHTMLStream::operator<<(html::CompTag const & tag)
 501 {
 502         if (tag.tag_.empty())
 503                 return *this;
 504         clearTagDeque();
 505         os_ << tag.writeTag();
 506         *this << html::CR();
 507         return *this;
 508 }
 509
 510
 511 XHTMLStream & XHTMLStream::operator<<(html::FontTag 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::CR const &)
 521 {
 522         // tabs?
 523         os_ << from_ascii("\n");
 524         return *this;
 525 }
 526
 527
 528 bool XHTMLStream::isTagOpen(html::StartTag const & stag) const
 529 {
 530         TagDeque::const_iterator sit = tag_stack_.begin();
 531         TagDeque::const_iterator const sen = tag_stack_.end();
 532         for (; sit != sen; ++sit)
 533                 if (**sit == stag)
 534                         return true;
 535         return false;
 536 }
 537
 538
 539 bool XHTMLStream::isTagOpen(html::EndTag const & etag) const
 540 {
 541         TagDeque::const_iterator sit = tag_stack_.begin();
 542         TagDeque::const_iterator const sen = tag_stack_.end();
 543         for (; sit != sen; ++sit)
 544                 if (etag == **sit)
 545                         return true;
 546         return false;
 547 }
 548
 549
 550 bool XHTMLStream::isTagPending(html::StartTag const & stag) const
 551 {
 552         TagDeque::const_iterator sit = pending_tags_.begin();
 553         TagDeque::const_iterator const sen = pending_tags_.end();
 554         for (; sit != sen; ++sit)
 555                 if (**sit == stag)
 556                         return true;
 557         return false;
 558 }
 559
 560
 561 // this is complicated, because we want to make sure that
 562 // everything is properly nested. the code ought to make
 563 // sure of that, but we won't assert (yet) if we run into
 564 // a problem. we'll just output error messages and try our
 565 // best to make things work.
 566 XHTMLStream & XHTMLStream::operator<<(html::EndTag const & etag)
 567 {
 568         if (etag.tag_.empty())
 569                 return *this;
 570
 571         // if this tag is pending, we can simply discard it.
 572         if (!pending_tags_.empty()) {
 573
 574                 if (etag == *pending_tags_.back()) {
 575                         // we have <tag></tag>, so we discard it and remove it
 576                         // from the pending_tags_.
 577                         pending_tags_.pop_back();
 578                         return *this;
 579                 }
 580
 581                 // there is a pending tag that isn't the one we are trying
 582                 // to close.
 583
 584                 // is this tag itself pending?
 585                 // non-const iterators because we may call erase().
 586                 TagDeque::iterator dit = pending_tags_.begin();
 587                 TagDeque::iterator const den = pending_tags_.end();
 588                 for (; dit != den; ++dit) {
 589                         if (etag == **dit) {
 590                                 // it was pending, so we just erase it
 591                                 writeError("Tried to close pending tag `" + etag.tag_
 592                                         + "' when other tags were pending. Last pending tag is `"
 593                                         + to_utf8(pending_tags_.back()->writeTag())
 594                                         + "'. Tag discarded.");
 595                                 pending_tags_.erase(dit);
 596                                 return *this;
 597                         }
 598                 }
 599                 // so etag isn't itself pending. is it even open?
 600                 if (!isTagOpen(etag)) {
 601                         writeError("Tried to close `" + etag.tag_
 602                                  + "' when tag was not open. Tag discarded.");
 603                         return *this;
 604                 }
 605                 // ok, so etag is open.
 606                 // our strategy will be as below: we will do what we need to
 607                 // do to close this tag.
 608                 string estr = "Closing tag `" + etag.tag_
 609                         + "' when other tags are pending. Discarded pending tags:\n";
 610                 for (dit = pending_tags_.begin(); dit != den; ++dit)
 611                         estr += to_utf8(html::htmlize((*dit)->writeTag(), XHTMLStream::ESCAPE_ALL)) + "\n";
 612                 writeError(estr);
 613                 // clear the pending tags...
 614                 pending_tags_.clear();
 615                 // ...and then just fall through.
 616         }
 617
 618         // make sure there are tags to be closed
 619         if (tag_stack_.empty()) {
 620                 writeError("Tried to close `" + etag.tag_
 621                          + "' when no tags were open!");
 622                 return *this;
 623         }
 624
 625         // is the tag we are closing the last one we opened?
 626         if (etag == *tag_stack_.back()) {
 627                 // output it...
 628                 os_ << etag.writeEndTag();
 629                 // ...and forget about it
 630                 tag_stack_.pop_back();
 631                 return *this;
 632         }
 633
 634         // we are trying to close a tag other than the one last opened.
 635         // let's first see if this particular tag is still open somehow.
 636         if (!isTagOpen(etag)) {
 637                 writeError("Tried to close `" + etag.tag_
 638                         + "' when tag was not open. Tag discarded.");
 639                 return *this;
 640         }
 641
 642         // so the tag was opened, but other tags have been opened since
 643         // and not yet closed.
 644         // if it's a font tag, though...
 645         if (etag.asFontTag()) {
 646                 // it won't be a problem if the other tags open since this one
 647                 // are also font tags.
 648                 TagDeque::const_reverse_iterator rit = tag_stack_.rbegin();
 649                 TagDeque::const_reverse_iterator ren = tag_stack_.rend();
 650                 for (; rit != ren; ++rit) {
 651                         if (etag == **rit)
 652                                 break;
 653                         if (!(*rit)->asFontTag()) {
 654                                 // we'll just leave it and, presumably, have to close it later.
 655                                 writeError("Unable to close font tag `" + etag.tag_
 656                                         + "' due to open non-font tag `" + (*rit)->tag_ + "'.");
 657                                 return *this;
 658                         }
 659                 }
 660
 661                 // so we have e.g.:
 662                 //    <em>this is <strong>bold
 663                 // and are being asked to closed em. we want:
 664                 //    <em>this is <strong>bold</strong></em><strong>
 665                 // first, we close the intervening tags...
 666                 TagPtr curtag = tag_stack_.back();
 667                 // ...remembering them in a stack.
 668                 TagDeque fontstack;
 669                 while (etag != *curtag) {
 670                         os_ << curtag->writeEndTag();
 671                         fontstack.push_back(curtag);
 672                         tag_stack_.pop_back();
 673                         curtag = tag_stack_.back();
 674                 }
 675                 os_ << etag.writeEndTag();
 676                 tag_stack_.pop_back();
 677
 678                 // ...and restore the other tags.
 679                 rit = fontstack.rbegin();
 680                 ren = fontstack.rend();
 681                 for (; rit != ren; ++rit)
 682                         pending_tags_.push_back(*rit);
 683                 return *this;
 684         }
 685
 686         // it wasn't a font tag.
 687         // so other tags were opened before this one and not properly closed.
 688         // so we'll close them, too. that may cause other issues later, but it
 689         // at least guarantees proper nesting.
 690         writeError("Closing tag `" + etag.tag_
 691                 + "' when other tags are open, namely:");
 692         TagPtr curtag = tag_stack_.back();
 693         while (etag != *curtag) {
 694                 writeError(curtag->tag_);
 695                 if (*curtag != parsep_tag)
 696                         os_ << curtag->writeEndTag();
 697                 tag_stack_.pop_back();
 698                 curtag = tag_stack_.back();
 699         }
 700         // curtag is now the one we actually want.
 701         os_ << curtag->writeEndTag();
 702         tag_stack_.pop_back();
 703
 704         return *this;
 705 }
 706
 707 // End code for XHTMLStream
 708
 709 namespace {
 710
 711 // convenience functions
 712
 713 inline void openParTag(XHTMLStream & xs, Layout const & lay,
 714                        std::string parlabel)
 715 {
 716         xs << html::ParTag(lay.htmltag(), lay.htmlattr(), parlabel);
 717 }
 718
 719
 720 void openParTag(XHTMLStream & xs, Layout const & lay,
 721                 ParagraphParameters const & params,
 722                 std::string parlabel)
 723 {
 724         // FIXME Are there other things we should handle here?
 725         string const align = alignmentToCSS(params.align());
 726         if (align.empty()) {
 727                 openParTag(xs, lay, parlabel);
 728                 return;
 729         }
 730         string attrs = lay.htmlattr() + " style='text-align: " + align + ";'";
 731         xs << html::ParTag(lay.htmltag(), attrs, parlabel);
 732 }
 733
 734
 735 inline void closeTag(XHTMLStream & xs, Layout const & lay)
 736 {
 737         xs << html::EndTag(lay.htmltag());
 738 }
 739
 740
 741 inline void openLabelTag(XHTMLStream & xs, Layout const & lay)
 742 {
 743         xs << html::StartTag(lay.htmllabeltag(), lay.htmllabelattr());
 744 }
 745
 746
 747 inline void closeLabelTag(XHTMLStream & xs, Layout const & lay)
 748 {
 749         xs << html::EndTag(lay.htmllabeltag());
 750 }
 751
 752
 753 inline void openItemTag(XHTMLStream & xs, Layout const & lay)
 754 {
 755         xs << html::StartTag(lay.htmlitemtag(), lay.htmlitemattr(), true);
 756 }
 757
 758
 759 void openItemTag(XHTMLStream & xs, Layout const & lay,
 760              ParagraphParameters const & params)
 761 {
 762         // FIXME Are there other things we should handle here?
 763         string const align = alignmentToCSS(params.align());
 764         if (align.empty()) {
 765                 openItemTag(xs, lay);
 766                 return;
 767         }
 768         string attrs = lay.htmlattr() + " style='text-align: " + align + ";'";
 769         xs << html::StartTag(lay.htmlitemtag(), attrs);
 770 }
 771
 772
 773 inline void closeItemTag(XHTMLStream & xs, Layout const & lay)
 774 {
 775         xs << html::EndTag(lay.htmlitemtag());
 776 }
 777
 778 // end of convenience functions
 779
 780 ParagraphList::const_iterator findLastParagraph(
 781         ParagraphList::const_iterator p,
 782         ParagraphList::const_iterator const & pend)
 783 {
 784         for (++p; p != pend && p->layout().latextype == LATEX_PARAGRAPH; ++p)
 785                 ;
 786
 787         return p;
 788 }
 789
 790
 791 ParagraphList::const_iterator findEndOfEnvironment(
 792                 ParagraphList::const_iterator const & pstart,
 793                 ParagraphList::const_iterator const & pend)
 794 {
 795         ParagraphList::const_iterator p = pstart;
 796         Layout const & bstyle = p->layout();
 797         size_t const depth = p->params().depth();
 798         for (++p; p != pend; ++p) {
 799                 Layout const & style = p->layout();
 800                 // It shouldn't happen that e.g. a section command occurs inside
 801                 // a quotation environment, at a higher depth, but as of 6/2009,
 802                 // it can happen. We pretend that it's just at lowest depth.
 803                 if (style.latextype == LATEX_COMMAND)
 804                         return p;
 805
 806                 // If depth is down, we're done
 807                 if (p->params().depth() < depth)
 808                         return p;
 809
 810                 // If depth is up, we're not done
 811                 if (p->params().depth() > depth)
 812                         continue;
 813
 814                 // FIXME I am not sure about the first check.
 815                 // Surely we *could* have different layouts that count as
 816                 // LATEX_PARAGRAPH, right?
 817                 if (style.latextype == LATEX_PARAGRAPH || style != bstyle)
 818                         return p;
 819         }
 820         return pend;
 821 }
 822
 823
 824 ParagraphList::const_iterator makeParagraphs(Buffer const & buf,
 825                                             XHTMLStream & xs,
 826                                             OutputParams const & runparams,
 827                                             Text const & text,
 828                                             ParagraphList::const_iterator const & pbegin,
 829                                             ParagraphList::const_iterator const & pend)
 830 {
 831         ParagraphList::const_iterator const begin = text.paragraphs().begin();
 832         ParagraphList::const_iterator par = pbegin;
 833         for (; par != pend; ++par) {
 834                 Layout const & lay = par->layout();
 835                 if (!lay.counter.empty())
 836                         buf.masterBuffer()->params().
 837                             documentClass().counters().step(lay.counter, OutputUpdate);
 838
 839                 // FIXME We should see if there's a label to be output and
 840                 // do something with it.
 841                 if (par != pbegin)
 842                         xs << html::CR();
 843
 844                 // We want to open the paragraph tag if:
 845                 //   (i) the current layout permits multiple paragraphs
 846                 //  (ii) we are either not already inside a paragraph (HTMLIsBlock) OR
 847                 //       we are, but this is not the first paragraph
 848                 // But we do not want to open the paragraph tag if this paragraph contains
 849                 // only one item, and that item is "inline", i.e., not HTMLIsBlock (such
 850                 // as a branch). That is the "special case" we handle first.
 851                 Inset const * specinset = par->size() == 1 ? par->getInset(0) : 0;
 852                 bool const special_case =
 853                         specinset && !specinset->getLayout().htmlisblock();
 854
 855                 bool const open_par = runparams.html_make_pars
 856                         && (!runparams.html_in_par || par != pbegin)
 857                         && !special_case;
 858
 859                 // We want to issue the closing tag if either:
 860                 //   (i)  We opened it, and either html_in_par is false,
 861                 //        or we're not in the last paragraph, anyway.
 862                 //   (ii) We didn't open it and html_in_par is true,
 863                 //        but we are in the first par, and there is a next par.
 864                 ParagraphList::const_iterator nextpar = par;
 865                 ++nextpar;
 866                 bool const close_par =
 867                         (open_par && (!runparams.html_in_par || nextpar != pend))
 868                         || (!open_par && runparams.html_in_par && par == pbegin && nextpar != pend);
 869
 870                 if (open_par) {
 871                         // We do not issue the paragraph id if we are doing
 872                         // this for the TOC (or some similar purpose)
 873                         openParTag(xs, lay, par->params(),
 874                                    runparams.for_toc ? "" : par->magicLabel());
 875                 }
 876
 877                 docstring const deferred = par->simpleLyXHTMLOnePar(buf, xs,
 878                         runparams, text.outerFont(distance(begin, par)),
 879                         open_par, close_par);
 880
 881                 if (!deferred.empty()) {
 882                         xs << XHTMLStream::ESCAPE_NONE << deferred << html::CR();
 883                 }
 884                 if (close_par) {
 885                         closeTag(xs, lay);
 886                         xs << html::CR();
 887                 }
 888         }
 889         return pend;
 890 }
 891
 892
 893 ParagraphList::const_iterator makeBibliography(Buffer const & buf,
 894                                 XHTMLStream & xs,
 895                                 OutputParams const & runparams,
 896                                 Text const & text,
 897                                 ParagraphList::const_iterator const & pbegin,
 898                                 ParagraphList::const_iterator const & pend)
 899 {
 900         // FIXME XHTML
 901         // Use TextClass::htmlTOCLayout() to figure out how we should look.
 902         xs << html::StartTag("h2", "class='bibliography'")
 903            << pbegin->layout().labelstring(false)
 904            << html::EndTag("h2")
 905            << html::CR()
 906            << html::StartTag("div", "class='bibliography'")
 907            << html::CR();
 908         makeParagraphs(buf, xs, runparams, text, pbegin, pend);
 909         xs << html::EndTag("div");
 910         return pend;
 911 }
 912
 913
 914 bool isNormalEnv(Layout const & lay)
 915 {
 916         return lay.latextype == LATEX_ENVIRONMENT
 917             || lay.latextype == LATEX_BIB_ENVIRONMENT;
 918 }
 919
 920
 921 ParagraphList::const_iterator makeEnvironment(Buffer const & buf,
 922                                               XHTMLStream & xs,
 923                                               OutputParams const & runparams,
 924                                               Text const & text,
 925                                               ParagraphList::const_iterator const & pbegin,
 926                                               ParagraphList::const_iterator const & pend)
 927 {
 928         ParagraphList::const_iterator const begin = text.paragraphs().begin();
 929         ParagraphList::const_iterator par = pbegin;
 930         Layout const & bstyle = par->layout();
 931         depth_type const origdepth = pbegin->params().depth();
 932
 933         // open tag for this environment
 934         openParTag(xs, bstyle, pbegin->magicLabel());
 935         xs << html::CR();
 936
 937         // we will on occasion need to remember a layout from before.
 938         Layout const * lastlay = 0;
 939
 940         while (par != pend) {
 941                 Layout const & style = par->layout();
 942                 // the counter only gets stepped if we're in some kind of list,
 943                 // or if it's the first time through.
 944                 // note that enum, etc, are handled automatically.
 945                 // FIXME There may be a bug here about user defined enumeration
 946                 // types. If so, then we'll need to take the counter and add "i",
 947                 // "ii", etc, as with enum.
 948                 Counters & cnts = buf.masterBuffer()->params().documentClass().counters();
 949                 docstring const & cntr = style.counter;
 950                 if (!style.counter.empty()
 951                     && (par == pbegin || !isNormalEnv(style))
 952                                 && cnts.hasCounter(cntr)
 953                 )
 954                         cnts.step(cntr, OutputUpdate);
 955                 ParagraphList::const_iterator send;
 956
 957                 switch (style.latextype) {
 958                 case LATEX_ENVIRONMENT:
 959                 case LATEX_LIST_ENVIRONMENT:
 960                 case LATEX_ITEM_ENVIRONMENT: {
 961                         // There are two possiblities in this case.
 962                         // One is that we are still in the environment in which we
 963                         // started---which we will be if the depth is the same.
 964                         if (par->params().depth() == origdepth) {
 965                                 LATTEST(bstyle == style);
 966                                 if (lastlay != 0) {
 967                                         closeItemTag(xs, *lastlay);
 968                                         lastlay = 0;
 969                                 }
 970
 971                                 // this will be positive, if we want to skip the
 972                                 // initial word (if it's been taken for the label).
 973                                 pos_type sep = 0;
 974                                 bool const labelfirst = style.htmllabelfirst();
 975                                 if (!labelfirst)
 976                                         openItemTag(xs, style, par->params());
 977
 978                                 // label output
 979                                 if (style.labeltype != LABEL_NO_LABEL &&
 980                                     style.htmllabeltag() != "NONE") {
 981                                         if (isNormalEnv(style)) {
 982                                                 // in this case, we print the label only for the first
 983                                                 // paragraph (as in a theorem).
 984                                                 if (par == pbegin) {
 985                                                         docstring const lbl =
 986                                                                         pbegin->params().labelString();
 987                                                         if (!lbl.empty()) {
 988                                                                 openLabelTag(xs, style);
 989                                                                 xs << lbl;
 990                                                                 closeLabelTag(xs, style);
 991                                                         }
 992                                                         xs << html::CR();
 993                                                 }
 994                                         } else { // some kind of list
 995                                                 if (style.labeltype == LABEL_MANUAL) {
 996                                                         openLabelTag(xs, style);
 997                                                         sep = par->firstWordLyXHTML(xs, runparams);
 998                                                         closeLabelTag(xs, style);
 999                                                         xs << html::CR();
1000                                                 }
1001                                                 else {
1002                                                         openLabelTag(xs, style);
1003                                                         xs << par->params().labelString();
1004                                                         closeLabelTag(xs, style);
1005                                                         xs << html::CR();
1006                                                 }
1007                                         }
1008                                 } // end label output
1009
1010                                 if (labelfirst)
1011                                         openItemTag(xs, style, par->params());
1012
1013                                 par->simpleLyXHTMLOnePar(buf, xs, runparams,
1014                                         text.outerFont(distance(begin, par)), true, true, sep);
1015                                 ++par;
1016
1017                                 // We may not want to close the tag yet, in particular:
1018                                 // If we're not at the end...
1019                                 if (par != pend
1020                                         //  and are doing items...
1021                                          && !isNormalEnv(style)
1022                                          // and if the depth has changed...
1023                                          && par->params().depth() != origdepth) {
1024                                          // then we'll save this layout for later, and close it when
1025                                          // we get another item.
1026                                         lastlay = &style;
1027                                 } else
1028                                         closeItemTag(xs, style);
1029                                 xs << html::CR();
1030                         }
1031                         // The other possibility is that the depth has increased, in which
1032                         // case we need to recurse.
1033                         else {
1034                                 send = findEndOfEnvironment(par, pend);
1035                                 par = makeEnvironment(buf, xs, runparams, text, par, send);
1036                         }
1037                         break;
1038                 }
1039                 case LATEX_PARAGRAPH:
1040                         send = findLastParagraph(par, pend);
1041                         par = makeParagraphs(buf, xs, runparams, text, par, send);
1042                         break;
1043                 // Shouldn't happen
1044                 case LATEX_BIB_ENVIRONMENT:
1045                         send = par;
1046                         ++send;
1047                         par = makeParagraphs(buf, xs, runparams, text, par, send);
1048                         break;
1049                 // Shouldn't happen
1050                 case LATEX_COMMAND:
1051                         ++par;
1052                         break;
1053                 }
1054         }
1055
1056         if (lastlay != 0)
1057                 closeItemTag(xs, *lastlay);
1058         closeTag(xs, bstyle);
1059         xs << html::CR();
1060         return pend;
1061 }
1062
1063
1064 void makeCommand(Buffer const & buf,
1065                  XHTMLStream & xs,
1066                  OutputParams const & runparams,
1067                  Text const & text,
1068                  ParagraphList::const_iterator const & pbegin)
1069 {
1070         Layout const & style = pbegin->layout();
1071         if (!style.counter.empty())
1072                 buf.masterBuffer()->params().
1073                     documentClass().counters().step(style.counter, OutputUpdate);
1074
1075         bool const make_parid = !runparams.for_toc && runparams.html_make_pars;
1076
1077         if (style.labeltype == LABEL_ABOVE)
1078                 xs << html::StartTag("div")
1079                    << pbegin->params().labelString()
1080                    << html::EndTag("div");
1081         else if (style.labeltype == LABEL_CENTERED)
1082                 xs << html::StartTag("div", "style = \"text-align: center;\"")
1083                    << pbegin->params().labelString()
1084                    << html::EndTag("div");
1085
1086         openParTag(xs, style, pbegin->params(),
1087                    make_parid ? pbegin->magicLabel() : "");
1088
1089         // Label around sectioning number:
1090         // FIXME Probably need to account for LABEL_MANUAL
1091         if (style.labeltype != LABEL_NO_LABEL &&
1092             style.labeltype != LABEL_ABOVE &&
1093             style.labeltype != LABEL_CENTERED ) {
1094                 openLabelTag(xs, style);
1095                 xs << pbegin->params().labelString();
1096                 closeLabelTag(xs, style);
1097                 // Otherwise the label might run together with the text
1098                 xs << from_ascii(" ");
1099         }
1100
1101         ParagraphList::const_iterator const begin = text.paragraphs().begin();
1102         pbegin->simpleLyXHTMLOnePar(buf, xs, runparams,
1103                         text.outerFont(distance(begin, pbegin)));
1104         closeTag(xs, style);
1105         xs << html::CR();
1106 }
1107
1108 } // end anonymous namespace
1109
1110
1111 void xhtmlParagraphs(Text const & text,
1112                        Buffer const & buf,
1113                        XHTMLStream & xs,
1114                        OutputParams const & runparams)
1115 {
1116         ParagraphList const & paragraphs = text.paragraphs();
1117         if (runparams.par_begin == runparams.par_end) {
1118                 runparams.par_begin = 0;
1119                 runparams.par_end = paragraphs.size();
1120         }
1121         pit_type bpit = runparams.par_begin;
1122         pit_type const epit = runparams.par_end;
1123         LASSERT(bpit < epit,
1124                 { xs << XHTMLStream::ESCAPE_NONE << "<!-- XHTML output error! -->\n"; return; });
1125
1126         OutputParams ourparams = runparams;
1127         ParagraphList::const_iterator const pend =
1128                 (epit == (int) paragraphs.size()) ?
1129                         paragraphs.end() : paragraphs.constIterator(epit);
1130         while (bpit < epit) {
1131                 ParagraphList::const_iterator par = paragraphs.constIterator(bpit);
1132                 if (par->params().startOfAppendix()) {
1133                         // We want to reset the counter corresponding to toplevel sectioning
1134                         Layout const & lay =
1135                                 buf.masterBuffer()->params().documentClass().getTOCLayout();
1136                         docstring const cnt = lay.counter;
1137                         if (!cnt.empty()) {
1138                                 Counters & cnts =
1139                                         buf.masterBuffer()->params().documentClass().counters();
1140                                 cnts.reset(cnt);
1141                         }
1142                 }
1143                 Layout const & style = par->layout();
1144                 ParagraphList::const_iterator const lastpar = par;
1145                 ParagraphList::const_iterator send;
1146
1147                 switch (style.latextype) {
1148                 case LATEX_COMMAND: {
1149                         // The files with which we are working never have more than
1150                         // one paragraph in a command structure.
1151                         // FIXME
1152                         // if (ourparams.html_in_par)
1153                         //   fix it so we don't get sections inside standard, e.g.
1154                         // note that we may then need to make runparams not const, so we
1155                         // can communicate that back.
1156                         // FIXME Maybe this fix should be in the routines themselves, in case
1157                         // they are called from elsewhere.
1158                         makeCommand(buf, xs, ourparams, text, par);
1159                         ++par;
1160                         break;
1161                 }
1162                 case LATEX_ENVIRONMENT:
1163                 case LATEX_LIST_ENVIRONMENT:
1164                 case LATEX_ITEM_ENVIRONMENT: {
1165                         // FIXME Same fix here.
1166                         send = findEndOfEnvironment(par, pend);
1167                         par = makeEnvironment(buf, xs, ourparams, text, par, send);
1168                         break;
1169                 }
1170                 case LATEX_BIB_ENVIRONMENT: {
1171                         // FIXME Same fix here.
1172                         send = findEndOfEnvironment(par, pend);
1173                         par = makeBibliography(buf, xs, ourparams, text, par, send);
1174                         break;
1175                 }
1176                 case LATEX_PARAGRAPH:
1177                         send = findLastParagraph(par, pend);
1178                         par = makeParagraphs(buf, xs, ourparams, text, par, send);
1179                         break;
1180                 }
1181                 bpit += distance(lastpar, par);
1182         }
1183 }
1184
1185
1186 string alignmentToCSS(LyXAlignment align)
1187 {
1188         switch (align) {
1189         case LYX_ALIGN_BLOCK:
1190                 // we are NOT going to use text-align: justify!!
1191         case LYX_ALIGN_LEFT:
1192                 return "left";
1193         case LYX_ALIGN_RIGHT:
1194                 return "right";
1195         case LYX_ALIGN_CENTER:
1196                 return "center";
1197         default:
1198                 break;
1199         }
1200         return "";
1201 }
1202
1203 } // namespace lyx