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