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) const
 295 {
 296         writeError(msg + ": Tag Stack");
 297         TagStack::const_reverse_iterator it = tag_stack_.rbegin();
 298         TagStack::const_reverse_iterator en = tag_stack_.rend();
 299         for (; it != en; ++it) {
 300                 writeError(it->tag_);
 301         }
 302         writeError("Pending Tags");
 303         it = pending_tags_.rbegin();
 304         en = pending_tags_.rend();
 305         for (; it != en; ++it) {
 306                 writeError(it->tag_);
 307         }
 308         writeError("End Tag Stack");
 309 }
 310 #endif
 311
 312
 313 void XHTMLStream::writeError(std::string const & s) const
 314 {
 315         LYXERR0(s);
 316         os_ << from_utf8("<!-- Output Error: " + s + " -->\n");
 317 }
 318
 319
 320 namespace {
 321         // an illegal tag for internal use
 322         static html::StartTag const parsep_tag("&LyX_parsep_tag&");
 323 }
 324
 325
 326 bool XHTMLStream::closeFontTags()
 327 {
 328         if (isTagPending(parsep_tag))
 329                 // we haven't had any content
 330                 return true;
 331
 332         // this may be a useless check, since we ought at least to have
 333         // the parsep_tag. but it can't hurt too much to be careful.
 334         if (tag_stack_.empty())
 335                 return true;
 336
 337         // first, we close any open font tags we can close
 338         TagPtr curtag = tag_stack_.back();
 339         while (curtag->asFontTag()) {
 340                 os_ << curtag->writeEndTag();
 341                 tag_stack_.pop_back();
 342                 // this shouldn't happen, since then the font tags
 343                 // weren't in any other tag.
 344                 LBUFERR(!tag_stack_.empty());
 345                 curtag = tag_stack_.back();
 346         }
 347
 348         if (*curtag == parsep_tag)
 349                 return true;
 350
 351         // so we've hit a non-font tag.
 352         writeError("Tags still open in closeFontTags(). Probably not a problem,\n"
 353                    "but you might want to check these tags:");
 354         TagDeque::const_reverse_iterator it = tag_stack_.rbegin();
 355         TagDeque::const_reverse_iterator const en = tag_stack_.rend();
 356         for (; it != en; ++it) {
 357                 if (**it == parsep_tag)
 358                         break;
 359                 writeError((*it)->tag_);
 360         }
 361         return false;
 362 }
 363
 364
 365 void XHTMLStream::startParagraph(bool keep_empty)
 366 {
 367         pending_tags_.push_back(makeTagPtr(html::StartTag(parsep_tag)));
 368         if (keep_empty)
 369                 clearTagDeque();
 370 }
 371
 372
 373 void XHTMLStream::endParagraph()
 374 {
 375         if (isTagPending(parsep_tag)) {
 376                 // this case is normal. it just means we didn't have content,
 377                 // so the parsep_tag never got moved onto the tag stack.
 378                 while (!pending_tags_.empty()) {
 379                         // clear all pending tags up to and including the parsep tag.
 380                         // note that we work from the back, because we want to get rid
 381                         // of everything that hasn't been used.
 382                         TagPtr const cur_tag = pending_tags_.back();
 383                         pending_tags_.pop_back();
 384                         if (*cur_tag == parsep_tag)
 385                                 break;
 386                 }
 387                 return;
 388         }
 389
 390         if (!isTagOpen(parsep_tag)) {
 391                 writeError("No paragraph separation tag found in endParagraph().");
 392                 return;
 393         }
 394
 395         // this case is also normal, if the parsep tag is the last one
 396         // on the stack. otherwise, it's an error.
 397         while (!tag_stack_.empty()) {
 398                 TagPtr const cur_tag = tag_stack_.back();
 399                 tag_stack_.pop_back();
 400                 if (*cur_tag == parsep_tag)
 401                         break;
 402                 writeError("Tag `" + cur_tag->tag_ + "' still open at end of paragraph. Closing.");
 403                 os_ << cur_tag->writeEndTag();
 404         }
 405 }
 406
 407
 408 void XHTMLStream::clearTagDeque()
 409 {
 410         while (!pending_tags_.empty()) {
 411                 TagPtr const tag = pending_tags_.front();
 412                 if (*tag != parsep_tag)
 413                         // tabs?
 414                         os_ << tag->writeTag();
 415                 tag_stack_.push_back(tag);
 416                 pending_tags_.pop_front();
 417         }
 418 }
 419
 420
 421 XHTMLStream & XHTMLStream::operator<<(docstring const & d)
 422 {
 423         clearTagDeque();
 424         os_ << html::htmlize(d, escape_);
 425         escape_ = ESCAPE_ALL;
 426         return *this;
 427 }
 428
 429
 430 XHTMLStream & XHTMLStream::operator<<(const char * s)
 431 {
 432         clearTagDeque();
 433         docstring const d = from_ascii(s);
 434         os_ << html::htmlize(d, escape_);
 435         escape_ = ESCAPE_ALL;
 436         return *this;
 437 }
 438
 439
 440 XHTMLStream & XHTMLStream::operator<<(char_type c)
 441 {
 442         clearTagDeque();
 443         os_ << html::escapeChar(c, escape_);
 444         escape_ = ESCAPE_ALL;
 445         return *this;
 446 }
 447
 448
 449 XHTMLStream & XHTMLStream::operator<<(char c)
 450 {
 451         clearTagDeque();
 452         os_ << html::escapeChar(c, escape_);
 453         escape_ = ESCAPE_ALL;
 454         return *this;
 455 }
 456
 457
 458 XHTMLStream & XHTMLStream::operator<<(int i)
 459 {
 460         clearTagDeque();
 461         os_ << i;
 462         escape_ = ESCAPE_ALL;
 463         return *this;
 464 }
 465
 466
 467 XHTMLStream & XHTMLStream::operator<<(EscapeSettings e)
 468 {
 469         escape_ = e;
 470         return *this;
 471 }
 472
 473
 474 XHTMLStream & XHTMLStream::operator<<(html::StartTag const & tag)
 475 {
 476         if (tag.tag_.empty())
 477                 return *this;
 478         pending_tags_.push_back(makeTagPtr(tag));
 479         if (tag.keepempty_)
 480                 clearTagDeque();
 481         return *this;
 482 }
 483
 484
 485 XHTMLStream & XHTMLStream::operator<<(html::ParTag const & tag)
 486 {
 487         if (tag.tag_.empty())
 488                 return *this;
 489         pending_tags_.push_back(makeTagPtr(tag));
 490         return *this;
 491 }
 492
 493
 494 XHTMLStream & XHTMLStream::operator<<(html::CompTag const & tag)
 495 {
 496         if (tag.tag_.empty())
 497                 return *this;
 498         clearTagDeque();
 499         os_ << tag.writeTag();
 500         *this << html::CR();
 501         return *this;
 502 }
 503
 504
 505 XHTMLStream & XHTMLStream::operator<<(html::FontTag const & tag)
 506 {
 507         if (tag.tag_.empty())
 508                 return *this;
 509         pending_tags_.push_back(makeTagPtr(tag));
 510         return *this;
 511 }
 512
 513
 514 XHTMLStream & XHTMLStream::operator<<(html::CR const &)
 515 {
 516         // tabs?
 517         os_ << from_ascii("\n");
 518         return *this;
 519 }
 520
 521
 522 bool XHTMLStream::isTagOpen(html::StartTag const & stag) const
 523 {
 524         TagDeque::const_iterator sit = tag_stack_.begin();
 525         TagDeque::const_iterator const sen = tag_stack_.end();
 526         for (; sit != sen; ++sit)
 527                 if (**sit == stag)
 528                         return true;
 529         return false;
 530 }
 531
 532
 533 bool XHTMLStream::isTagOpen(html::EndTag const & etag) const
 534 {
 535         TagDeque::const_iterator sit = tag_stack_.begin();
 536         TagDeque::const_iterator const sen = tag_stack_.end();
 537         for (; sit != sen; ++sit)
 538                 if (etag == **sit)
 539                         return true;
 540         return false;
 541 }
 542
 543
 544 bool XHTMLStream::isTagPending(html::StartTag const & stag) const
 545 {
 546         TagDeque::const_iterator sit = pending_tags_.begin();
 547         TagDeque::const_iterator const sen = pending_tags_.end();
 548         for (; sit != sen; ++sit)
 549                 if (**sit == stag)
 550                         return true;
 551         return false;
 552 }
 553
 554
 555 // this is complicated, because we want to make sure that
 556 // everything is properly nested. the code ought to make
 557 // sure of that, but we won't assert (yet) if we run into
 558 // a problem. we'll just output error messages and try our
 559 // best to make things work.
 560 XHTMLStream & XHTMLStream::operator<<(html::EndTag const & etag)
 561 {
 562         if (etag.tag_.empty())
 563                 return *this;
 564
 565         // if this tag is pending, we can simply discard it.
 566         if (!pending_tags_.empty()) {
 567
 568                 if (etag == *pending_tags_.back()) {
 569                         // we have <tag></tag>, so we discard it and remove it
 570                         // from the pending_tags_.
 571                         pending_tags_.pop_back();
 572                         return *this;
 573                 }
 574
 575                 // there is a pending tag that isn't the one we are trying
 576                 // to close.
 577
 578                 // is this tag itself pending?
 579                 // non-const iterators because we may call erase().
 580                 TagDeque::iterator dit = pending_tags_.begin();
 581                 TagDeque::iterator const den = pending_tags_.end();
 582                 for (; dit != den; ++dit) {
 583                         if (etag == **dit) {
 584                                 // it was pending, so we just erase it
 585                                 writeError("Tried to close pending tag `" + etag.tag_
 586                                         + "' when other tags were pending. Last pending tag is `"
 587                                         + to_utf8(pending_tags_.back()->writeTag())
 588                                         + "'. Tag discarded.");
 589                                 pending_tags_.erase(dit);
 590                                 return *this;
 591                         }
 592                 }
 593                 // so etag isn't itself pending. is it even open?
 594                 if (!isTagOpen(etag)) {
 595                         writeError("Tried to close `" + etag.tag_
 596                                  + "' when tag was not open. Tag discarded.");
 597                         return *this;
 598                 }
 599                 // ok, so etag is open.
 600                 // our strategy will be as below: we will do what we need to
 601                 // do to close this tag.
 602                 string estr = "Closing tag `" + etag.tag_
 603                         + "' when other tags are pending. Discarded pending tags:\n";
 604                 for (dit = pending_tags_.begin(); dit != den; ++dit)
 605                         estr += to_utf8(html::htmlize((*dit)->writeTag(), XHTMLStream::ESCAPE_ALL)) + "\n";
 606                 writeError(estr);
 607                 // clear the pending tags...
 608                 pending_tags_.clear();
 609                 // ...and then just fall through.
 610         }
 611
 612         // make sure there are tags to be closed
 613         if (tag_stack_.empty()) {
 614                 writeError("Tried to close `" + etag.tag_
 615                          + "' when no tags were open!");
 616                 return *this;
 617         }
 618
 619         // is the tag we are closing the last one we opened?
 620         if (etag == *tag_stack_.back()) {
 621                 // output it...
 622                 os_ << etag.writeEndTag();
 623                 // ...and forget about it
 624                 tag_stack_.pop_back();
 625                 return *this;
 626         }
 627
 628         // we are trying to close a tag other than the one last opened.
 629         // let's first see if this particular tag is still open somehow.
 630         if (!isTagOpen(etag)) {
 631                 writeError("Tried to close `" + etag.tag_
 632                         + "' when tag was not open. Tag discarded.");
 633                 return *this;
 634         }
 635
 636         // so the tag was opened, but other tags have been opened since
 637         // and not yet closed.
 638         // if it's a font tag, though...
 639         if (etag.asFontTag()) {
 640                 // it won't be a problem if the other tags open since this one
 641                 // are also font tags.
 642                 TagDeque::const_reverse_iterator rit = tag_stack_.rbegin();
 643                 TagDeque::const_reverse_iterator ren = tag_stack_.rend();
 644                 for (; rit != ren; ++rit) {
 645                         if (etag == **rit)
 646                                 break;
 647                         if (!(*rit)->asFontTag()) {
 648                                 // we'll just leave it and, presumably, have to close it later.
 649                                 writeError("Unable to close font tag `" + etag.tag_
 650                                         + "' due to open non-font tag `" + (*rit)->tag_ + "'.");
 651                                 return *this;
 652                         }
 653                 }
 654
 655                 // so we have e.g.:
 656                 //    <em>this is <strong>bold
 657                 // and are being asked to closed em. we want:
 658                 //    <em>this is <strong>bold</strong></em><strong>
 659                 // first, we close the intervening tags...
 660                 TagPtr curtag = tag_stack_.back();
 661                 // ...remembering them in a stack.
 662                 TagDeque fontstack;
 663                 while (etag != *curtag) {
 664                         os_ << curtag->writeEndTag();
 665                         fontstack.push_back(curtag);
 666                         tag_stack_.pop_back();
 667                         curtag = tag_stack_.back();
 668                 }
 669                 os_ << etag.writeEndTag();
 670                 tag_stack_.pop_back();
 671
 672                 // ...and restore the other tags.
 673                 rit = fontstack.rbegin();
 674                 ren = fontstack.rend();
 675                 for (; rit != ren; ++rit)
 676                         pending_tags_.push_back(*rit);
 677                 return *this;
 678         }
 679
 680         // it wasn't a font tag.
 681         // so other tags were opened before this one and not properly closed.
 682         // so we'll close them, too. that may cause other issues later, but it
 683         // at least guarantees proper nesting.
 684         writeError("Closing tag `" + etag.tag_
 685                 + "' when other tags are open, namely:");
 686         TagPtr curtag = tag_stack_.back();
 687         while (etag != *curtag) {
 688                 writeError(curtag->tag_);
 689                 if (*curtag != parsep_tag)
 690                         os_ << curtag->writeEndTag();
 691                 tag_stack_.pop_back();
 692                 curtag = tag_stack_.back();
 693         }
 694         // curtag is now the one we actually want.
 695         os_ << curtag->writeEndTag();
 696         tag_stack_.pop_back();
 697
 698         return *this;
 699 }
 700
 701 // End code for XHTMLStream
 702
 703 namespace {
 704
 705 // convenience functions
 706
 707 inline void openParTag(XHTMLStream & xs, Layout const & lay,
 708                        std::string parlabel)
 709 {
 710         xs << html::ParTag(lay.htmltag(), lay.htmlattr(), parlabel);
 711 }
 712
 713
 714 void openParTag(XHTMLStream & xs, Layout const & lay,
 715                 ParagraphParameters const & params,
 716                 std::string parlabel)
 717 {
 718         // FIXME Are there other things we should handle here?
 719         string const align = alignmentToCSS(params.align());
 720         if (align.empty()) {
 721                 openParTag(xs, lay, parlabel);
 722                 return;
 723         }
 724         string attrs = lay.htmlattr() + " style='text-align: " + align + ";'";
 725         xs << html::ParTag(lay.htmltag(), attrs, parlabel);
 726 }
 727
 728
 729 inline void closeTag(XHTMLStream & xs, Layout const & lay)
 730 {
 731         xs << html::EndTag(lay.htmltag());
 732 }
 733
 734
 735 inline void openLabelTag(XHTMLStream & xs, Layout const & lay)
 736 {
 737         xs << html::StartTag(lay.htmllabeltag(), lay.htmllabelattr());
 738 }
 739
 740
 741 inline void closeLabelTag(XHTMLStream & xs, Layout const & lay)
 742 {
 743         xs << html::EndTag(lay.htmllabeltag());
 744 }
 745
 746
 747 inline void openItemTag(XHTMLStream & xs, Layout const & lay)
 748 {
 749         xs << html::StartTag(lay.htmlitemtag(), lay.htmlitemattr(), true);
 750 }
 751
 752
 753 void openItemTag(XHTMLStream & xs, Layout const & lay,
 754              ParagraphParameters const & params)
 755 {
 756         // FIXME Are there other things we should handle here?
 757         string const align = alignmentToCSS(params.align());
 758         if (align.empty()) {
 759                 openItemTag(xs, lay);
 760                 return;
 761         }
 762         string attrs = lay.htmlattr() + " style='text-align: " + align + ";'";
 763         xs << html::StartTag(lay.htmlitemtag(), attrs);
 764 }
 765
 766
 767 inline void closeItemTag(XHTMLStream & xs, Layout const & lay)
 768 {
 769         xs << html::EndTag(lay.htmlitemtag());
 770 }
 771
 772 // end of convenience functions
 773
 774 ParagraphList::const_iterator findLastParagraph(
 775         ParagraphList::const_iterator p,
 776         ParagraphList::const_iterator const & pend)
 777 {
 778         for (++p; p != pend && p->layout().latextype == LATEX_PARAGRAPH; ++p)
 779                 ;
 780
 781         return p;
 782 }
 783
 784
 785 ParagraphList::const_iterator findEndOfEnvironment(
 786                 ParagraphList::const_iterator const & pstart,
 787                 ParagraphList::const_iterator const & pend)
 788 {
 789         ParagraphList::const_iterator p = pstart;
 790         Layout const & bstyle = p->layout();
 791         size_t const depth = p->params().depth();
 792         for (++p; p != pend; ++p) {
 793                 Layout const & style = p->layout();
 794                 // It shouldn't happen that e.g. a section command occurs inside
 795                 // a quotation environment, at a higher depth, but as of 6/2009,
 796                 // it can happen. We pretend that it's just at lowest depth.
 797                 if (style.latextype == LATEX_COMMAND)
 798                         return p;
 799
 800                 // If depth is down, we're done
 801                 if (p->params().depth() < depth)
 802                         return p;
 803
 804                 // If depth is up, we're not done
 805                 if (p->params().depth() > depth)
 806                         continue;
 807
 808                 // FIXME I am not sure about the first check.
 809                 // Surely we *could* have different layouts that count as
 810                 // LATEX_PARAGRAPH, right?
 811                 if (style.latextype == LATEX_PARAGRAPH || style != bstyle)
 812                         return p;
 813         }
 814         return pend;
 815 }
 816
 817
 818 ParagraphList::const_iterator makeParagraphs(Buffer const & buf,
 819                                             XHTMLStream & xs,
 820                                             OutputParams const & runparams,
 821                                             Text const & text,
 822                                             ParagraphList::const_iterator const & pbegin,
 823                                             ParagraphList::const_iterator const & pend)
 824 {
 825         ParagraphList::const_iterator const begin = text.paragraphs().begin();
 826         ParagraphList::const_iterator par = pbegin;
 827         for (; par != pend; ++par) {
 828                 Layout const & lay = par->layout();
 829                 if (!lay.counter.empty())
 830                         buf.masterBuffer()->params().
 831                             documentClass().counters().step(lay.counter, OutputUpdate);
 832                 // FIXME We should see if there's a label to be output and
 833                 // do something with it.
 834                 if (par != pbegin)
 835                         xs << html::CR();
 836
 837                 // If we are already in a paragraph, and this is the first one, then we
 838                 // do not want to open the paragraph tag.
 839                 // we also do not want to open it if the current layout does not permit
 840                 // multiple paragraphs.
 841                 bool const opened = runparams.html_make_pars &&
 842                         (par != pbegin || !runparams.html_in_par);
 843                 bool const make_parid = !runparams.for_toc && runparams.html_make_pars;
 844
 845                 if (opened)
 846                         openParTag(xs, lay, par->params(),
 847                                    make_parid ? par->magicLabel() : "");
 848
 849                 docstring const deferred =
 850                         par->simpleLyXHTMLOnePar(buf, xs, runparams, text.outerFont(distance(begin, par)));
 851
 852                 // We want to issue the closing tag if either:
 853                 //   (i)  We opened it, and either html_in_par is false,
 854                 //        or we're not in the last paragraph, anyway.
 855                 //   (ii) We didn't open it and html_in_par is true,
 856                 //        but we are in the first par, and there is a next par.
 857                 ParagraphList::const_iterator nextpar = par;
 858                 ++nextpar;
 859                 bool const needclose =
 860                         (opened && (!runparams.html_in_par || nextpar != pend))
 861                         || (!opened && runparams.html_in_par && par == pbegin && nextpar != pend);
 862                 if (needclose) {
 863                         closeTag(xs, lay);
 864                         xs << html::CR();
 865                 }
 866                 if (!deferred.empty()) {
 867                         xs << XHTMLStream::ESCAPE_NONE << deferred << html::CR();
 868                 }
 869         }
 870         return pend;
 871 }
 872
 873
 874 ParagraphList::const_iterator makeBibliography(Buffer const & buf,
 875                                 XHTMLStream & xs,
 876                                 OutputParams const & runparams,
 877                                 Text const & text,
 878                                 ParagraphList::const_iterator const & pbegin,
 879                                 ParagraphList::const_iterator const & pend)
 880 {
 881         // FIXME XHTML
 882         // Use TextClass::htmlTOCLayout() to figure out how we should look.
 883         xs << html::StartTag("h2", "class='bibliography'")
 884            << pbegin->layout().labelstring(false)
 885            << html::EndTag("h2")
 886            << html::CR()
 887            << html::StartTag("div", "class='bibliography'")
 888            << html::CR();
 889         makeParagraphs(buf, xs, runparams, text, pbegin, pend);
 890         xs << html::EndTag("div");
 891         return pend;
 892 }
 893
 894
 895 bool isNormalEnv(Layout const & lay)
 896 {
 897         return lay.latextype == LATEX_ENVIRONMENT
 898             || lay.latextype == LATEX_BIB_ENVIRONMENT;
 899 }
 900
 901
 902 ParagraphList::const_iterator makeEnvironment(Buffer const & buf,
 903                                               XHTMLStream & xs,
 904                                               OutputParams const & runparams,
 905                                               Text const & text,
 906                                               ParagraphList::const_iterator const & pbegin,
 907                                               ParagraphList::const_iterator const & pend)
 908 {
 909         ParagraphList::const_iterator const begin = text.paragraphs().begin();
 910         ParagraphList::const_iterator par = pbegin;
 911         Layout const & bstyle = par->layout();
 912         depth_type const origdepth = pbegin->params().depth();
 913
 914         // open tag for this environment
 915         openParTag(xs, bstyle, pbegin->magicLabel());
 916         xs << html::CR();
 917
 918         // we will on occasion need to remember a layout from before.
 919         Layout const * lastlay = 0;
 920
 921         while (par != pend) {
 922                 Layout const & style = par->layout();
 923                 // the counter only gets stepped if we're in some kind of list,
 924                 // or if it's the first time through.
 925                 // note that enum, etc, are handled automatically.
 926                 // FIXME There may be a bug here about user defined enumeration
 927                 // types. If so, then we'll need to take the counter and add "i",
 928                 // "ii", etc, as with enum.
 929                 Counters & cnts = buf.masterBuffer()->params().documentClass().counters();
 930                 docstring const & cntr = style.counter;
 931                 if (!style.counter.empty()
 932                     && (par == pbegin || !isNormalEnv(style))
 933                                 && cnts.hasCounter(cntr)
 934                 )
 935                         cnts.step(cntr, OutputUpdate);
 936                 ParagraphList::const_iterator send;
 937
 938                 switch (style.latextype) {
 939                 case LATEX_ENVIRONMENT:
 940                 case LATEX_LIST_ENVIRONMENT:
 941                 case LATEX_ITEM_ENVIRONMENT: {
 942                         // There are two possiblities in this case.
 943                         // One is that we are still in the environment in which we
 944                         // started---which we will be if the depth is the same.
 945                         if (par->params().depth() == origdepth) {
 946                                 LATTEST(bstyle == style);
 947                                 if (lastlay != 0) {
 948                                         closeItemTag(xs, *lastlay);
 949                                         lastlay = 0;
 950                                 }
 951
 952                                 // this will be positive, if we want to skip the
 953                                 // initial word (if it's been taken for the label).
 954                                 pos_type sep = 0;
 955                                 bool const labelfirst = style.htmllabelfirst();
 956                                 if (!labelfirst)
 957                                         openItemTag(xs, style, par->params());
 958
 959                                 // label output
 960                                 if (style.labeltype != LABEL_NO_LABEL &&
 961                                     style.htmllabeltag() != "NONE") {
 962                                         if (isNormalEnv(style)) {
 963                                                 // in this case, we print the label only for the first
 964                                                 // paragraph (as in a theorem).
 965                                                 if (par == pbegin) {
 966                                                         docstring const lbl =
 967                                                                         pbegin->params().labelString();
 968                                                         if (!lbl.empty()) {
 969                                                                 openLabelTag(xs, style);
 970                                                                 xs << lbl;
 971                                                                 closeLabelTag(xs, style);
 972                                                         }
 973                                                         xs << html::CR();
 974                                                 }
 975                                         } else { // some kind of list
 976                                                 if (style.labeltype == LABEL_MANUAL) {
 977                                                         openLabelTag(xs, style);
 978                                                         sep = par->firstWordLyXHTML(xs, runparams);
 979                                                         closeLabelTag(xs, style);
 980                                                         xs << html::CR();
 981                                                 }
 982                                                 else {
 983                                                         openLabelTag(xs, style);
 984                                                         xs << par->params().labelString();
 985                                                         closeLabelTag(xs, style);
 986                                                         xs << html::CR();
 987                                                 }
 988                                         }
 989                                 } // end label output
 990
 991                                 if (labelfirst)
 992                                         openItemTag(xs, style, par->params());
 993
 994                                 par->simpleLyXHTMLOnePar(buf, xs, runparams,
 995                                         text.outerFont(distance(begin, par)), sep);
 996                                 ++par;
 997
 998                                 // We may not want to close the tag yet, in particular:
 999                                 // If we're not at the end...
1000                                 if (par != pend
1001                                         //  and are doing items...
1002                                          && !isNormalEnv(style)
1003                                          // and if the depth has changed...
1004                                          && par->params().depth() != origdepth) {
1005                                          // then we'll save this layout for later, and close it when
1006                                          // we get another item.
1007                                         lastlay = &style;
1008                                 } else
1009                                         closeItemTag(xs, style);
1010                                 xs << html::CR();
1011                         }
1012                         // The other possibility is that the depth has increased, in which
1013                         // case we need to recurse.
1014                         else {
1015                                 send = findEndOfEnvironment(par, pend);
1016                                 par = makeEnvironment(buf, xs, runparams, text, par, send);
1017                         }
1018                         break;
1019                 }
1020                 case LATEX_PARAGRAPH:
1021                         send = findLastParagraph(par, pend);
1022                         par = makeParagraphs(buf, xs, runparams, text, par, send);
1023                         break;
1024                 // Shouldn't happen
1025                 case LATEX_BIB_ENVIRONMENT:
1026                         send = par;
1027                         ++send;
1028                         par = makeParagraphs(buf, xs, runparams, text, par, send);
1029                         break;
1030                 // Shouldn't happen
1031                 case LATEX_COMMAND:
1032                         ++par;
1033                         break;
1034                 }
1035         }
1036
1037         if (lastlay != 0)
1038                 closeItemTag(xs, *lastlay);
1039         closeTag(xs, bstyle);
1040         xs << html::CR();
1041         return pend;
1042 }
1043
1044
1045 void makeCommand(Buffer const & buf,
1046                  XHTMLStream & xs,
1047                  OutputParams const & runparams,
1048                  Text const & text,
1049                  ParagraphList::const_iterator const & pbegin)
1050 {
1051         Layout const & style = pbegin->layout();
1052         if (!style.counter.empty())
1053                 buf.masterBuffer()->params().
1054                     documentClass().counters().step(style.counter, OutputUpdate);
1055
1056         bool const make_parid = !runparams.for_toc && runparams.html_make_pars;
1057
1058         if (style.labeltype == LABEL_ABOVE)
1059                 xs << html::StartTag("div")
1060                    << pbegin->params().labelString()
1061                    << html::EndTag("div");
1062         else if (style.labeltype == LABEL_CENTERED)
1063                 xs << html::StartTag("div", "style = \"text-align: center;\"")
1064                    << pbegin->params().labelString()
1065                    << html::EndTag("div");
1066
1067         openParTag(xs, style, pbegin->params(),
1068                    make_parid ? pbegin->magicLabel() : "");
1069
1070         // Label around sectioning number:
1071         // FIXME Probably need to account for LABEL_MANUAL
1072         if (style.labeltype != LABEL_NO_LABEL &&
1073             style.labeltype != LABEL_ABOVE &&
1074             style.labeltype != LABEL_CENTERED ) {
1075                 openLabelTag(xs, style);
1076                 xs << pbegin->params().labelString();
1077                 closeLabelTag(xs, style);
1078                 // Otherwise the label might run together with the text
1079                 xs << from_ascii(" ");
1080         }
1081
1082         ParagraphList::const_iterator const begin = text.paragraphs().begin();
1083         pbegin->simpleLyXHTMLOnePar(buf, xs, runparams,
1084                         text.outerFont(distance(begin, pbegin)));
1085         closeTag(xs, style);
1086         xs << html::CR();
1087 }
1088
1089 } // end anonymous namespace
1090
1091
1092 void xhtmlParagraphs(Text const & text,
1093                        Buffer const & buf,
1094                        XHTMLStream & xs,
1095                        OutputParams const & runparams)
1096 {
1097         ParagraphList const & paragraphs = text.paragraphs();
1098         if (runparams.par_begin == runparams.par_end) {
1099                 runparams.par_begin = 0;
1100                 runparams.par_end = paragraphs.size();
1101         }
1102         pit_type bpit = runparams.par_begin;
1103         pit_type const epit = runparams.par_end;
1104         LASSERT(bpit < epit,
1105                 { xs << XHTMLStream::ESCAPE_NONE << "<!-- XHTML output error! -->\n"; return; });
1106
1107         OutputParams ourparams = runparams;
1108         ParagraphList::const_iterator const pend =
1109                 (epit == (int) paragraphs.size()) ?
1110                         paragraphs.end() : paragraphs.constIterator(epit);
1111         while (bpit < epit) {
1112                 ParagraphList::const_iterator par = paragraphs.constIterator(bpit);
1113                 if (par->params().startOfAppendix()) {
1114                         // We want to reset the counter corresponding to toplevel sectioning
1115                         Layout const & lay =
1116                                 buf.masterBuffer()->params().documentClass().getTOCLayout();
1117                         docstring const cnt = lay.counter;
1118                         if (!cnt.empty()) {
1119                                 Counters & cnts =
1120                                         buf.masterBuffer()->params().documentClass().counters();
1121                                 cnts.reset(cnt);
1122                         }
1123                 }
1124                 Layout const & style = par->layout();
1125                 ParagraphList::const_iterator const lastpar = par;
1126                 ParagraphList::const_iterator send;
1127
1128                 switch (style.latextype) {
1129                 case LATEX_COMMAND: {
1130                         // The files with which we are working never have more than
1131                         // one paragraph in a command structure.
1132                         // FIXME
1133                         // if (ourparams.html_in_par)
1134                         //   fix it so we don't get sections inside standard, e.g.
1135                         // note that we may then need to make runparams not const, so we
1136                         // can communicate that back.
1137                         // FIXME Maybe this fix should be in the routines themselves, in case
1138                         // they are called from elsewhere.
1139                         makeCommand(buf, xs, ourparams, text, par);
1140                         ++par;
1141                         break;
1142                 }
1143                 case LATEX_ENVIRONMENT:
1144                 case LATEX_LIST_ENVIRONMENT:
1145                 case LATEX_ITEM_ENVIRONMENT: {
1146                         // FIXME Same fix here.
1147                         send = findEndOfEnvironment(par, pend);
1148                         par = makeEnvironment(buf, xs, ourparams, text, par, send);
1149                         break;
1150                 }
1151                 case LATEX_BIB_ENVIRONMENT: {
1152                         // FIXME Same fix here.
1153                         send = findEndOfEnvironment(par, pend);
1154                         par = makeBibliography(buf, xs, ourparams, text, par, send);
1155                         break;
1156                 }
1157                 case LATEX_PARAGRAPH:
1158                         send = findLastParagraph(par, pend);
1159                         par = makeParagraphs(buf, xs, ourparams, text, par, send);
1160                         break;
1161                 }
1162                 bpit += distance(lastpar, par);
1163         }
1164 }
1165
1166
1167 string alignmentToCSS(LyXAlignment align)
1168 {
1169         switch (align) {
1170         case LYX_ALIGN_BLOCK:
1171                 // we are NOT going to use text-align: justify!!
1172         case LYX_ALIGN_LEFT:
1173                 return "left";
1174         case LYX_ALIGN_RIGHT:
1175                 return "right";
1176         case LYX_ALIGN_CENTER:
1177                 return "center";
1178         default:
1179                 break;
1180         }
1181         return "";
1182 }
1183
1184 } // namespace lyx