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