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.
8 * This code is based upon output_docbook.cpp
10 * Full author contact details are available in file CREDITS.
15 #include "output_xhtml.h"
18 #include "buffer_funcs.h"
19 #include "BufferParams.h"
22 #include "OutputParams.h"
23 #include "Paragraph.h"
24 #include "ParagraphList.h"
25 #include "ParagraphParameters.h"
28 #include "TextClass.h"
30 #include "support/lassert.h"
31 #include "support/debug.h"
32 #include "support/lstrings.h"
37 using namespace lyx::support;
43 docstring escapeChar(char_type c)
67 // escape what needs escaping
68 docstring htmlize(docstring const & str) {
70 docstring::const_iterator it = str.begin();
71 docstring::const_iterator en = str.end();
72 for (; it != en; ++it)
78 bool isFontTag(string const & s)
80 return s == "em" || s == "strong"; // others?
85 docstring StartTag::asTag() const
87 string output = "<" + tag_;
89 output += " " + attr_;
91 return from_utf8(output);
95 docstring StartTag::asEndTag() const
97 string output = "</" + tag_ + ">";
98 return from_utf8(output);
102 docstring EndTag::asEndTag() const
104 string output = "</" + tag_ + ">";
105 return from_utf8(output);
109 docstring CompTag::asTag() const
111 string output = "<" + tag_;
113 output += " " + attr_;
115 return from_utf8(output);
119 ////////////////////////////////////////////////////////////////
123 ////////////////////////////////////////////////////////////////
125 XHTMLStream::XHTMLStream(odocstream & os)
130 void XHTMLStream::cr()
133 os_ << from_ascii("\n");
137 bool XHTMLStream::closeFontTags()
139 // first, we close any open font tags we can close
140 StartTag curtag = tag_stack_.back();
141 while (html::isFontTag(curtag.tag_)) {
142 os_ << curtag.asEndTag();
143 tag_stack_.pop_back();
144 if (tag_stack_.empty())
145 // this probably shouldn't happen, since then the
146 // font tags weren't in any other tag. but that
147 // problem will likely be caught elsewhere.
149 curtag = tag_stack_.back();
151 // so we've hit a non-font tag. let's see if any of the
152 // remaining tags are font tags.
153 TagStack::const_iterator it = tag_stack_.begin();
154 TagStack::const_iterator en = tag_stack_.end();
155 bool noFontTags = true;
156 for (; it != en; ++it) {
157 if (html::isFontTag(it->tag_)) {
158 LYXERR0("Font tag `" << it->tag_ << "' still open in closeFontTags().");
166 void XHTMLStream::clearTagDeque()
168 while (!pending_tags_.empty()) {
169 StartTag const & tag = pending_tags_.front();
172 tag_stack_.push_back(tag);
173 pending_tags_.pop_front();
177 XHTMLStream & XHTMLStream::operator<<(docstring const & d)
179 // I'm tempted to make sure here that there are no tags in the input
181 os_ << html::htmlize(d);
186 XHTMLStream & XHTMLStream::operator<<(char_type c)
189 os_ << html::escapeChar(c);
194 XHTMLStream & XHTMLStream::operator<<(StartTag const & tag)
196 pending_tags_.push_back(tag);
203 XHTMLStream & XHTMLStream::operator<<(CompTag const & tag)
212 bool XHTMLStream::isTagOpen(string const & stag)
214 TagStack::const_iterator sit = tag_stack_.begin();
215 TagStack::const_iterator const sen = tag_stack_.end();
216 for (; sit != sen; ++sit)
217 // we could check for the
218 if (sit->tag_ == stag)
224 // this is complicated, because we want to make sure that
225 // everything is properly nested. the code ought to make
226 // sure of that, but we won't assert (yet) if we run into
227 // a problem. we'll just output error messages and try our
228 // best to make things work.
229 XHTMLStream & XHTMLStream::operator<<(EndTag const & etag)
231 // first make sure we're not closing an empty tag
232 if (!pending_tags_.empty()) {
233 StartTag const & stag = pending_tags_.back();
234 if (etag.tag_ == stag.tag_) {
235 // we have <tag></tag>, so we discard it and remove it
236 // from the pending_tags_.
237 pending_tags_.pop_back();
240 // there is a pending tag that isn't the one we are trying
242 // is this tag itself pending?
243 // non-const iterators because we may call erase().
244 TagDeque::iterator dit = pending_tags_.begin();
245 TagDeque::iterator const den = pending_tags_.end();
246 for (; dit != den; ++dit) {
247 if (dit->tag_ == etag.tag_) {
248 // it was pending, so we just erase it
249 LYXERR0("Tried to close pending tag `" << etag.tag_
250 << "' when other tags were pending. Tag discarded.");
251 pending_tags_.erase(dit);
255 // so etag isn't itself pending. is it even open?
256 if (!isTagOpen(etag.tag_)) {
257 LYXERR0("Tried to close `" << etag.tag_
258 << "' when tag was not open. Tag discarded.");
261 // ok, so etag is open.
262 // our strategy will be as below: we will do what we need to
263 // do to close this tag.
264 LYXERR0("Closing tag `" << etag.tag_
265 << "' when other tags are pending. Discarded pending tags:");
266 for (dit = pending_tags_.begin(); dit != den; ++dit)
268 // clear the pending tags...
269 pending_tags_.clear();
270 // ...and then just fall through.
273 // is the tag we are closing the last one we opened?
274 if (etag.tag_ == tag_stack_.back().tag_) {
276 os_ << etag.asEndTag();
277 // ...and forget about it
278 tag_stack_.pop_back();
282 // we are trying to close a tag other than the one last opened.
283 // let's first see if this particular tag is still open somehow.
284 if (!isTagOpen(etag.tag_)) {
285 LYXERR0("Tried to close `" << etag.tag_
286 << "' when tag was not open. Tag discarded.");
290 // so the tag was opened, but other tags have been opened since
291 // and not yet closed.
292 // if it's a font tag, though...
293 if (html::isFontTag(etag.tag_)) {
294 // it won't be a problem if the other tags open since this one
295 // are also font tags.
296 TagStack::const_reverse_iterator rit = tag_stack_.rbegin();
297 TagStack::const_reverse_iterator ren = tag_stack_.rend();
298 for (; rit != ren; ++rit) {
299 if (rit->tag_ == etag.tag_)
301 if (!html::isFontTag(rit->tag_)) {
302 // we'll just leave it and, presumably, have to close it later.
303 LYXERR0("Unable to close font tag `" << etag.tag_
304 << "' due to open non-font tag `" << rit->tag_ << "'.");
310 // <em>this is <strong>bold
311 // and are being asked to closed em. we want:
312 // <em>this is <strong>bold</strong></em><strong>
313 // first, we close the intervening tags...
314 StartTag curtag = tag_stack_.back();
315 // ...remembering them in a stack.
317 while (curtag.tag_ != etag.tag_) {
318 os_ << curtag.asEndTag();
319 fontstack.push_back(curtag);
320 tag_stack_.pop_back();
321 curtag = tag_stack_.back();
323 // now close our tag...
324 os_ << etag.asEndTag();
325 // ...and restore the other tags.
326 rit = fontstack.rbegin();
327 ren = fontstack.rend();
328 for (; rit != ren; ++rit)
329 pending_tags_.push_back(*rit);
333 // it wasn't a font tag.
334 // so other tags were opened before this one and not properly closed.
335 // so we'll close them, too. that may cause other issues later, but it
336 // at least guarantees proper nesting.
337 LYXERR0("Closing tag `" << etag.tag_
338 << "' when other tags are open, namely:");
339 StartTag curtag = tag_stack_.back();
340 while (curtag.tag_ != etag.tag_) {
341 LYXERR0(curtag.tag_);
342 os_ << curtag.asEndTag();
343 tag_stack_.pop_back();
344 curtag = tag_stack_.back();
346 // curtag is now the one we actually want.
347 os_ << curtag.asEndTag();
348 tag_stack_.pop_back();
353 // End code for XHTMLStream
357 // convenience functions
359 inline void openTag(XHTMLStream & xs, Layout const & lay)
361 xs << StartTag(lay.htmltag(), lay.htmlattr());
365 inline void closeTag(XHTMLStream & xs, Layout const & lay)
367 xs << EndTag(lay.htmltag());
371 inline void openLabelTag(XHTMLStream & xs, Layout const & lay)
373 xs << StartTag(lay.htmllabeltag(), lay.htmllabelattr());
377 inline void closeLabelTag(XHTMLStream & xs, Layout const & lay)
379 xs << EndTag(lay.htmllabeltag());
383 inline void openItemTag(XHTMLStream & xs, Layout const & lay)
385 xs << StartTag(lay.htmlitemtag(), lay.htmlitemattr(), true);
389 inline void closeItemTag(XHTMLStream & xs, Layout const & lay)
391 xs << EndTag(lay.htmlitemtag());
394 // end of convenience functions
396 ParagraphList::const_iterator searchParagraphHtml(
397 ParagraphList::const_iterator p,
398 ParagraphList::const_iterator const & pend)
400 for (++p; p != pend && p->layout().latextype == LATEX_PARAGRAPH; ++p)
407 ParagraphList::const_iterator searchEnvironmentHtml(
408 ParagraphList::const_iterator const pstart,
409 ParagraphList::const_iterator const & pend)
411 ParagraphList::const_iterator p = pstart;
412 Layout const & bstyle = p->layout();
413 size_t const depth = p->params().depth();
414 for (++p; p != pend; ++p) {
415 Layout const & style = p->layout();
416 // It shouldn't happen that e.g. a section command occurs inside
417 // a quotation environment, at a higher depth, but as of 6/2009,
418 // it can happen. We pretend that it's just at lowest depth.
419 if (style.latextype == LATEX_COMMAND)
421 // If depth is down, we're done
422 if (p->params().depth() < depth)
424 // If depth is up, we're not done
425 if (p->params().depth() > depth)
427 // Now we know we are at the same depth
428 if (style.latextype == LATEX_PARAGRAPH
429 || style.latexname() != bstyle.latexname())
436 ParagraphList::const_iterator makeParagraphs(Buffer const & buf,
438 OutputParams const & runparams,
440 ParagraphList::const_iterator const & pbegin,
441 ParagraphList::const_iterator const & pend)
443 ParagraphList::const_iterator const begin = text.paragraphs().begin();
444 ParagraphList::const_iterator par = pbegin;
445 for (; par != pend; ++par) {
446 Layout const & lay = par->layout();
447 if (!lay.counter.empty())
448 buf.params().documentClass().counters().step(lay.counter);
449 // FIXME We should see if there's a label to be output and
450 // do something with it.
454 // FIXME Should we really allow anything other than 'p' here?
456 // If we are already in a paragraph, and this is the first one, then we
457 // do not want to open the paragraph tag.
459 (par == pbegin && runparams.html_in_par) ? false : true;
462 docstring const deferred =
463 par->simpleLyXHTMLOnePar(buf, xs, runparams, text.outerFont(distance(begin, par)));
465 // We want to issue the closing tag if either:
466 // (i) We opened it, and either html_in_par is false,
467 // or we're not in the last paragraph, anyway.
468 // (ii) We didn't open it and html_in_par is true,
469 // but we are in the first par, and there is a next par.
470 ParagraphList::const_iterator nextpar = par;
472 bool const needclose =
473 (opened && (!runparams.html_in_par || nextpar != pend))
474 || (!opened && runparams.html_in_par && par == pbegin && nextpar != pend);
479 if (!deferred.empty()) {
488 ParagraphList::const_iterator makeBibliography(Buffer const & buf,
490 OutputParams const & runparams,
492 ParagraphList::const_iterator const & pbegin,
493 ParagraphList::const_iterator const & pend)
495 xs << StartTag("h2", "class='bibliography'");
496 xs << pbegin->layout().labelstring(false);
499 xs << StartTag("div", "class='bibliography'");
501 makeParagraphs(buf, xs, runparams, text, pbegin, pend);
507 bool isNormalEnv(Layout const & lay)
509 return lay.latextype == LATEX_ENVIRONMENT;
513 ParagraphList::const_iterator makeEnvironmentHtml(Buffer const & buf,
515 OutputParams const & runparams,
517 ParagraphList::const_iterator const & pbegin,
518 ParagraphList::const_iterator const & pend)
520 ParagraphList::const_iterator const begin = text.paragraphs().begin();
521 ParagraphList::const_iterator par = pbegin;
522 Layout const & bstyle = par->layout();
523 depth_type const origdepth = pbegin->params().depth();
525 // open tag for this environment
529 // we will on occasion need to remember a layout from before.
530 Layout const * lastlay = 0;
532 while (par != pend) {
533 Layout const & style = par->layout();
534 // the counter only gets stepped if we're in some kind of list,
535 // or if it's the first time through.
536 if (!style.counter.empty() && (par == pbegin || !isNormalEnv(style)))
537 buf.params().documentClass().counters().step(style.counter);
538 ParagraphList::const_iterator send;
539 // this will be positive, if we want to skip the initial word
540 // (if it's been taken for the label).
543 switch (style.latextype) {
544 case LATEX_ENVIRONMENT:
545 case LATEX_LIST_ENVIRONMENT:
546 case LATEX_ITEM_ENVIRONMENT: {
547 // There are two possiblities in this case.
548 // One is that we are still in the environment in which we
549 // started---which we will be if the depth is the same.
550 if (par->params().depth() == origdepth) {
551 LASSERT(bstyle == style, /* */);
553 closeItemTag(xs, *lastlay);
556 bool const labelfirst = style.htmllabelfirst();
557 if (isNormalEnv(style)) {
558 // in this case, we print the label only for the first
559 // paragraph (as in a theorem).
560 openItemTag(xs, style);
561 if (par == pbegin && style.htmllabeltag() != "NONE") {
562 docstring const lbl =
563 pbegin->expandLabel(style, buf.params(), false);
565 openLabelTag(xs, style);
567 closeLabelTag(xs, style);
571 } else { // some kind of list
573 openItemTag(xs, style);
574 if (style.labeltype == LABEL_MANUAL
575 && style.htmllabeltag() != "NONE") {
576 openLabelTag(xs, style);
577 // sep = par->firstWordLyXHTML(xs, runparams);
578 closeLabelTag(xs, style);
581 else if (style.labeltype != LABEL_NO_LABEL
582 && style.htmllabeltag() != "NONE") {
583 openLabelTag(xs, style);
584 xs << par->expandLabel(style, buf.params(), false);
585 closeLabelTag(xs, style);
589 openItemTag(xs, style);
591 xs << StartTag("span", "class='" + to_utf8(style.name()) + " inneritem'>");
593 par->simpleLyXHTMLOnePar(buf, xs, runparams,
594 text.outerFont(distance(begin, par)), sep);
595 if (!isNormalEnv(style) && !labelfirst)
596 xs << EndTag("span");
598 // We may not want to close the tag yet, in particular,
599 // if we're not at the end...
601 // and are doing items...
602 && style.latextype == LATEX_ITEM_ENVIRONMENT
603 // and if the depth has changed...
604 && par->params().depth() != origdepth) {
605 // then we'll save this layout for later, and close it when
606 // we get another item.
609 closeItemTag(xs, style);
612 // The other possibility is that the depth has increased, in which
613 // case we need to recurse.
615 send = searchEnvironmentHtml(par, pend);
616 par = makeEnvironmentHtml(buf, xs, runparams, text, par, send);
620 case LATEX_PARAGRAPH:
621 send = searchParagraphHtml(par, pend);
622 par = makeParagraphs(buf, xs, runparams, text, par, send);
625 case LATEX_BIB_ENVIRONMENT:
628 par = makeParagraphs(buf, xs, runparams, text, par, send);
638 closeItemTag(xs, *lastlay);
639 closeTag(xs, bstyle);
645 void makeCommand(Buffer const & buf,
647 OutputParams const & runparams,
649 ParagraphList::const_iterator const & pbegin)
651 Layout const & style = pbegin->layout();
652 if (!style.counter.empty())
653 buf.params().documentClass().counters().step(style.counter);
657 // Label around sectioning number:
658 // FIXME Probably need to account for LABEL_MANUAL
659 if (style.labeltype != LABEL_NO_LABEL) {
660 openLabelTag(xs, style);
661 xs << pbegin->expandLabel(style, buf.params(), false);
662 closeLabelTag(xs, style);
663 // Otherwise the label might run together with the text
664 xs << from_ascii(" ");
667 ParagraphList::const_iterator const begin = text.paragraphs().begin();
668 pbegin->simpleLyXHTMLOnePar(buf, xs, runparams,
669 text.outerFont(distance(begin, pbegin)));
674 } // end anonymous namespace
677 void xhtmlParagraphs(Text const & text,
680 OutputParams const & runparams)
682 ParagraphList const & paragraphs = text.paragraphs();
683 ParagraphList::const_iterator par = paragraphs.begin();
684 ParagraphList::const_iterator pend = paragraphs.end();
686 OutputParams ourparams = runparams;
687 while (par != pend) {
688 Layout const & style = par->layout();
689 ParagraphList::const_iterator lastpar = par;
690 ParagraphList::const_iterator send;
692 switch (style.latextype) {
693 case LATEX_COMMAND: {
694 // The files with which we are working never have more than
695 // one paragraph in a command structure.
697 // if (ourparams.html_in_par)
698 // fix it so we don't get sections inside standard, e.g.
699 // note that we may then need to make runparams not const, so we
700 // can communicate that back.
701 // FIXME Maybe this fix should be in the routines themselves, in case
702 // they are called from elsewhere.
703 makeCommand(buf, xs, ourparams, text, par);
707 case LATEX_ENVIRONMENT:
708 case LATEX_LIST_ENVIRONMENT:
709 case LATEX_ITEM_ENVIRONMENT: {
710 // FIXME Same fix here.
711 send = searchEnvironmentHtml(par, pend);
712 par = makeEnvironmentHtml(buf, xs, ourparams, text, par, send);
715 case LATEX_BIB_ENVIRONMENT: {
716 // FIXME Same fix here.
717 send = searchEnvironmentHtml(par, pend);
718 par = makeBibliography(buf, xs, ourparams, text, par, send);
721 case LATEX_PARAGRAPH:
722 send = searchParagraphHtml(par, pend);
723 par = makeParagraphs(buf, xs, ourparams, text, par, send);
727 // makeEnvironment may process more than one paragraphs and bypass pend
728 if (distance(lastpar, par) >= distance(lastpar, pend))