src/insets/InsetIndex.cpp

   1 /**
   2  * \file InsetIndex.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Lars Gullik Bjønnes
   7  * \author Jürgen Spitzmüller
   8  *
   9  * Full author contact details are available in file CREDITS.
  10  */
  11 #include <config.h>
  12
  13 #include "InsetIndex.h"
  14 #include "InsetIndexMacro.h"
  15
  16 #include "Buffer.h"
  17 #include "BufferParams.h"
  18 #include "BufferView.h"
  19 #include "ColorSet.h"
  20 #include "Cursor.h"
  21 #include "DispatchResult.h"
  22 #include "Encoding.h"
  23 #include "ErrorList.h"
  24 #include "FuncRequest.h"
  25 #include "FuncStatus.h"
  26 #include "IndicesList.h"
  27 #include "InsetList.h"
  28 #include "Language.h"
  29 #include "LaTeX.h"
  30 #include "LaTeXFeatures.h"
  31 #include "Lexer.h"
  32 #include "output_latex.h"
  33 #include "output_xhtml.h"
  34 #include "xml.h"
  35 #include "texstream.h"
  36 #include "TextClass.h"
  37 #include "TocBackend.h"
  38
  39 #include "support/debug.h"
  40 #include "support/docstream.h"
  41 #include "support/FileName.h"
  42 #include "support/gettext.h"
  43 #include "support/lstrings.h"
  44 #include "support/Translator.h"
  45
  46 #include "frontends/alert.h"
  47
  48 #include <algorithm>
  49 #include <set>
  50 #include <iostream>
  51
  52 #include <QThreadStorage>
  53
  54 using namespace std;
  55 using namespace lyx::support;
  56
  57 namespace lyx {
  58
  59 namespace {
  60
  61 typedef Translator<string, InsetIndexParams::PageRange> PageRangeTranslator;
  62 typedef Translator<docstring, InsetIndexParams::PageRange> PageRangeTranslatorLoc;
  63
  64 PageRangeTranslator const init_insetindexpagerangetranslator()
  65 {
  66         PageRangeTranslator translator("none", InsetIndexParams::None);
  67         translator.addPair("start", InsetIndexParams::Start);
  68         translator.addPair("end", InsetIndexParams::End);
  69         return translator;
  70 }
  71
  72 PageRangeTranslator const init_insetindexpagerangetranslator_latex()
  73 {
  74         PageRangeTranslator translator("", InsetIndexParams::None);
  75         translator.addPair("(", InsetIndexParams::Start);
  76         translator.addPair(")", InsetIndexParams::End);
  77         return translator;
  78 }
  79
  80
  81 PageRangeTranslatorLoc const init_insetindexpagerangetranslator_loc()
  82 {
  83         PageRangeTranslatorLoc translator(docstring(), InsetIndexParams::None);
  84         translator.addPair(_("Starts page range"), InsetIndexParams::Start);
  85         translator.addPair(_("Ends page range"), InsetIndexParams::End);
  86         return translator;
  87 }
  88
  89
  90 PageRangeTranslator const & insetindexpagerangetranslator()
  91 {
  92         static PageRangeTranslator const prtranslator =
  93                         init_insetindexpagerangetranslator();
  94         return prtranslator;
  95 }
  96
  97
  98 PageRangeTranslatorLoc const & insetindexpagerangetranslator_loc()
  99 {
 100         static PageRangeTranslatorLoc const translator =
 101                         init_insetindexpagerangetranslator_loc();
 102         return translator;
 103 }
 104
 105
 106 PageRangeTranslator const & insetindexpagerangetranslator_latex()
 107 {
 108         static PageRangeTranslator const lttranslator =
 109                         init_insetindexpagerangetranslator_latex();
 110         return lttranslator;
 111 }
 112
 113 } // namespace anon
 114
 115 /////////////////////////////////////////////////////////////////////
 116 //
 117 // InsetIndex
 118 //
 119 ///////////////////////////////////////////////////////////////////////
 120
 121
 122 InsetIndex::InsetIndex(Buffer * buf, InsetIndexParams const & params)
 123         : InsetCollapsible(buf), params_(params)
 124 {}
 125
 126
 127 void InsetIndex::latex(otexstream & ios, OutputParams const & runparams_in) const
 128 {
 129         OutputParams runparams(runparams_in);
 130         runparams.inIndexEntry = true;
 131
 132         otexstringstream os;
 133
 134         if (buffer().masterBuffer()->params().use_indices && !params_.index.empty()
 135                 && params_.index != "idx") {
 136                 os << "\\sindex[";
 137                 os << escape(params_.index);
 138                 os << "]{";
 139         } else {
 140                 os << "\\index";
 141                 os << '{';
 142         }
 143
 144         // Get the LaTeX output from InsetText. We need to deconstruct this later
 145         // in order to check if we need to generate a sorting key
 146         odocstringstream ourlatex;
 147         otexstream ots(ourlatex);
 148         InsetText::latex(ots, runparams);
 149         if (runparams.for_search != OutputParams::NoSearch) {
 150                 // No need for special handling, if we are only searching for some patterns
 151                 os << ourlatex.str() << "}";
 152                 return;
 153         }
 154
 155         if (hasSortKey()) {
 156                 getSortkey(os, runparams);
 157                 os << "@";
 158                 os << ourlatex.str();
 159                 getSubentries(os, runparams);
 160                 if (hasSeeRef()) {
 161                         os << "|";
 162                         os << insetindexpagerangetranslator_latex().find(params_.range);
 163                         getSeeRefs(os, runparams);
 164                 }
 165         } else {
 166                 // We check whether we need a sort key.
 167                 // If so, we use the plaintext version
 168                 odocstringstream ourplain;
 169                 InsetText::plaintext(ourplain, runparams);
 170
 171                 // These are the LaTeX and plaintext representations
 172                 docstring latexstr = ourlatex.str();
 173                 docstring plainstr = ourplain.str();
 174
 175                 // This will get what follows | if anything does,
 176                 // the command (e.g., see, textbf) for pagination
 177                 // formatting
 178                 docstring cmd;
 179
 180                 if (hasSeeRef()) {
 181                         odocstringstream seeref;
 182                         otexstream otsee(seeref);
 183                         getSeeRefs(otsee, runparams);
 184                         cmd = seeref.str();
 185                 } else if (!params_.pagefmt.empty() && params_.pagefmt != "default") {
 186                         cmd = from_utf8(params_.pagefmt);
 187                 } else {
 188                         // Check for the | separator to strip the cmd.
 189                         // This goes wrong on an escaped "|", but as the escape
 190                         // character can be changed in style files, we cannot
 191                         // prevent that.
 192                         size_t pos = latexstr.find(from_ascii("|"));
 193                         if (pos != docstring::npos) {
 194                                 // Put the bit after "|" into cmd...
 195                                 cmd = latexstr.substr(pos + 1);
 196                                 // ...and erase that stuff from latexstr
 197                                 latexstr = latexstr.erase(pos);
 198                                 // ...as well as from plainstr
 199                                 size_t ppos = plainstr.find(from_ascii("|"));
 200                                 if (ppos < plainstr.size())
 201                                         plainstr.erase(ppos);
 202                                 else
 203                                         LYXERR0("The `|' separator was not found in the plaintext version!");
 204                         }
 205                 }
 206
 207                 odocstringstream subentries;
 208                 otexstream otsub(subentries);
 209                 getSubentries(otsub, runparams);
 210                 if (subentries.str().empty()) {
 211                         // Separate the entries and subentries, i.e., split on "!".
 212                         // This goes wrong on an escaped "!", but as the escape
 213                         // character can be changed in style files, we cannot
 214                         // prevent that.
 215                         std::vector<docstring> const levels =
 216                                         getVectorFromString(latexstr, from_ascii("!"), true);
 217                         std::vector<docstring> const levels_plain =
 218                                         getVectorFromString(plainstr, from_ascii("!"), true);
 219
 220                         vector<docstring>::const_iterator it = levels.begin();
 221                         vector<docstring>::const_iterator end = levels.end();
 222                         vector<docstring>::const_iterator it2 = levels_plain.begin();
 223                         bool first = true;
 224                         for (; it != end; ++it) {
 225                                 // The separator needs to be put back when
 226                                 // writing the levels, except for the first level
 227                                 if (!first)
 228                                         os << '!';
 229                                 else
 230                                         first = false;
 231
 232                                 // Now here comes the reason for this whole procedure:
 233                                 // We try to correctly sort macros and formatted strings.
 234                                 // If we find a command, prepend a plain text
 235                                 // version of the content to get sorting right,
 236                                 // e.g. \index{LyX@\LyX}, \index{text@\textbf{text}}.
 237                                 // We do this on all levels.
 238                                 // We don't do it if the level already contains a '@', though.
 239                                 // Plaintext might return nothing (e.g. for ERTs).
 240                                 // In that case, we use LaTeX.
 241                                 docstring const spart = (levels_plain.empty() || (*it2).empty()) ? *it : *it2;
 242                                 processLatexSorting(os, runparams, *it, spart);
 243                                 if (it2 < levels_plain.end())
 244                                         ++it2;
 245                         }
 246                 } else {
 247                         processLatexSorting(os, runparams, latexstr, plainstr);
 248                         os << subentries.str();
 249                 }
 250
 251                 // At last, re-insert the command, separated by "|"
 252                 if (!cmd.empty()) {
 253                         os << "|"
 254                            << insetindexpagerangetranslator_latex().find(params_.range)
 255                            << cmd;
 256                 }
 257         }
 258         os << '}';
 259
 260         // In macros with moving arguments, such as \section,
 261         // we store the index and output it after the macro (#2154)
 262         if (runparams_in.postpone_fragile_stuff)
 263                 runparams_in.post_macro += os.str();
 264         else
 265                 ios << os.release();
 266 }
 267
 268
 269 void InsetIndex::processLatexSorting(otexstream & os, OutputParams const & runparams,
 270                                 docstring const latex, docstring const spart) const
 271 {
 272         if (contains(latex, '\\') && !contains(latex, '@')) {
 273                 // Now we need to validate that all characters in
 274                 // the sorting part are representable in the current
 275                 // encoding. If not try the LaTeX macro which might
 276                 // or might not be a good choice, and issue a warning.
 277                 pair<docstring, docstring> spart_latexed =
 278                                 runparams.encoding->latexString(spart, runparams.dryrun);
 279                 if (!spart_latexed.second.empty())
 280                         LYXERR0("Uncodable character in index entry. Sorting might be wrong!");
 281                 if (spart != spart_latexed.first && !runparams.dryrun) {
 282                         TeXErrors terr;
 283                         ErrorList & errorList = buffer().errorList("Export");
 284                         docstring const s = bformat(_("LyX's automatic index sorting algorithm faced "
 285                                                       "problems with the entry '%1$s'.\n"
 286                                                       "Please specify the sorting of this entry manually, as "
 287                                                       "explained in the User Guide."), spart);
 288                         Paragraph const & par = buffer().paragraphs().front();
 289                         errorList.push_back(ErrorItem(_("Index sorting failed"), s,
 290                                                       {par.id(), 0}, {par.id(), -1}));
 291                         buffer().bufferErrors(terr, errorList);
 292                 }
 293                 // Remove remaining \'s from the sort key
 294                 docstring ppart = subst(spart_latexed.first, from_ascii("\\"), docstring());
 295                 // Plain quotes need to be escaped, however (#10649), as this
 296                 // is the default escape character
 297                 ppart = subst(ppart, from_ascii("\""), from_ascii("\\\""));
 298
 299                 // Now insert the sortkey, separated by '@'.
 300                 os << ppart;
 301                 os << '@';
 302         }
 303         // Insert the actual level text
 304         os << latex;
 305 }
 306
 307
 308 void InsetIndex::docbook(XMLStream & xs, OutputParams const & runparams) const
 309 {
 310         // Two ways of processing this inset are implemented:
 311         // - the legacy one, based on parsing the raw LaTeX (before LyX 2.4) -- unlikely to be deprecated
 312         // - the modern one, based on precise insets for indexing features
 313         // Like the LaTeX implementation, consider the user chooses either of those options.
 314
 315         // Get the content of the inset as LaTeX, as some things may be encoded as ERT (like {}).
 316         // TODO: if there is an ERT within the index term, its conversion should be tried, in case it becomes useful;
 317         //  otherwise, ERTs should become comments. For now, they are just copied as-is, which is barely satisfactory.
 318         odocstringstream odss;
 319         otexstream ots(odss);
 320         InsetText::latex(ots, runparams);
 321         docstring latexString = trim(odss.str());
 322
 323         // Check whether there are unsupported things. @ is supported, but only for sorting, without specific formatting.
 324         if (latexString.find(from_utf8("@\\")) != lyx::docstring::npos) {
 325                 docstring error = from_utf8("Unsupported feature: an index entry contains an @\\. "
 326                                                                         "Complete entry: \"") + latexString + from_utf8("\"");
 327                 LYXERR0(error);
 328                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + error + from_utf8(" -->\n"));
 329         }
 330
 331         // Handle several indices (indicated in the inset instead of the raw latexString).
 332         docstring indexType = from_utf8("");
 333         if (buffer().masterBuffer()->params().use_indices) {
 334                 indexType += " type=\"" + params_.index + "\"";
 335         }
 336
 337         // Split the string into its main constituents: terms, and command (see, see also, range).
 338         size_t positionVerticalBar = latexString.find(from_ascii("|")); // What comes before | is (sub)(sub)entries.
 339         docstring indexTerms = latexString.substr(0, positionVerticalBar);
 340         docstring command;
 341         if (positionVerticalBar != lyx::docstring::npos) {
 342                 command = latexString.substr(positionVerticalBar + 1);
 343         }
 344
 345         // Handle sorting issues, with @.
 346         docstring sortAs;
 347         if (hasSortKey()) {
 348                 sortAs = getSortkeyAsText(runparams);
 349                 // indexTerms may contain a sort key if the user has both the inset and the manual key.
 350         } else {
 351                 vector<docstring> sortingElements = getVectorFromString(indexTerms, from_ascii("@"), false);
 352                 if (sortingElements.size() == 2) {
 353                         sortAs = sortingElements[0];
 354                         indexTerms = sortingElements[1];
 355                 }
 356         }
 357
 358         // Handle primary, secondary, and tertiary terms (entries, subentries, and subsubentries, for LaTeX).
 359         vector<docstring> terms;
 360         if (const vector<docstring> potential_terms = getSubentriesAsText(runparams); !potential_terms.empty()) {
 361                 terms = potential_terms;
 362                 // The main term is not present in the vector, as it's not a subentry. The main index term is inserted raw in
 363                 // the index inset. Considering that the user either uses the new or the legacy mechanism, the main term is the
 364                 // full string within this inset (i.e. without the subinsets).
 365                 terms.insert(terms.begin(), latexString);
 366         } else {
 367                 terms = getVectorFromString(indexTerms, from_ascii("!"), false);
 368         }
 369
 370         // Handle ranges. Happily, in the raw LaTeX mode, (| and |) can only be at the end of the string!
 371         bool hasInsetRange = params_.range != InsetIndexParams::PageRange::None;
 372         bool hasStartRange = params_.range == InsetIndexParams::PageRange::Start ||
 373                         latexString.find(from_ascii("|(")) != lyx::docstring::npos;
 374         bool hasEndRange = params_.range == InsetIndexParams::PageRange::End ||
 375                         latexString.find(from_ascii("|)")) != lyx::docstring::npos;
 376
 377         if (hasInsetRange) {
 378                 // Remove the ranges from the command if they do not appear at the beginning.
 379                 size_t index = 0;
 380                 while ((index = command.find(from_utf8("|("), index)) != std::string::npos)
 381                         command.erase(index, 1);
 382                 index = 0;
 383                 while ((index = command.find(from_utf8("|)"), index)) != std::string::npos)
 384                         command.erase(index, 1);
 385
 386                 // Remove the ranges when they are the only vertical bar in the complete string.
 387                 if (command[0] == '(' || command[0] == ')')
 388                         command.erase(0, 1);
 389         }
 390
 391         // Handle see and seealso. As "see" is a prefix of "seealso", the order of the comparisons is important.
 392         // Both commands are mutually exclusive!
 393         docstring see = getSeeAsText(runparams);
 394         vector<docstring> seeAlsoes = getSeeAlsoesAsText(runparams);
 395
 396         if (see.empty() && seeAlsoes.empty() && command.substr(0, 3) == "see") {
 397                 // Unescape brackets.
 398                 size_t index = 0;
 399                 while ((index = command.find(from_utf8("\\{"), index)) != std::string::npos)
 400                         command.erase(index, 1);
 401                 index = 0;
 402                 while ((index = command.find(from_utf8("\\}"), index)) != std::string::npos)
 403                         command.erase(index, 1);
 404
 405                 // Retrieve the part between brackets, and remove the complete seealso.
 406                 size_t positionOpeningBracket = command.find(from_ascii("{"));
 407                 size_t positionClosingBracket = command.find(from_ascii("}"));
 408                 docstring list = command.substr(positionOpeningBracket + 1, positionClosingBracket - positionOpeningBracket - 1);
 409
 410                 // Parse the list of referenced entries (or a single one for see).
 411                 if (command.substr(0, 7) == "seealso") {
 412                         seeAlsoes = getVectorFromString(list, from_ascii(","), false);
 413                 } else {
 414                         see = list;
 415
 416                         if (see.find(from_ascii(",")) != std::string::npos) {
 417                                 docstring error = from_utf8("Several index terms found as \"see\"! Only one is acceptable. "
 418                                                                                         "Complete entry: \"") + latexString + from_utf8("\"");
 419                                 LYXERR0(error);
 420                                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + error + from_utf8(" -->\n"));
 421                         }
 422                 }
 423
 424                 // Remove the complete see/seealso from the commands, in case there is something else to parse.
 425                 command = command.substr(positionClosingBracket + 1);
 426         }
 427
 428         // Some parts of the strings are not parsed, as they do not have anything matching in DocBook: things like
 429         // formatting the entry or the page number, other strings for sorting. https://wiki.lyx.org/Tips/Indexing
 430         // If there are such things in the index entry, then this code may miserably fail. For example, for "Peter|(textbf",
 431         // no range will be detected.
 432         // TODO: Could handle formatting as significance="preferred"?
 433         if (!command.empty()) {
 434                 docstring error = from_utf8("Unsupported feature: an index entry contains a | with an unsupported command, ")
 435                                           + command + from_utf8(". ") + from_utf8("Complete entry: \"") + latexString + from_utf8("\"");
 436                 LYXERR0(error);
 437                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + error + from_utf8(" -->\n"));
 438         }
 439
 440     // Write all of this down.
 441         if (terms.empty() && !hasEndRange) {
 442                 docstring error = from_utf8("No index term found! Complete entry: \"") + latexString + from_utf8("\"");
 443                 LYXERR0(error);
 444                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + error + from_utf8(" -->\n"));
 445         } else {
 446                 // Generate the attributes for ranges. It is based on the terms that are indexed, but the ID must be unique
 447                 // to this indexing area (xml::cleanID does not guarantee this: for each call with the same arguments,
 448                 // the same legal ID is produced; here, as the input would be the same, the output must be, by design).
 449                 // Hence the thread-local storage, as the numbers must strictly be unique, and thus cannot be shared across
 450                 // a paragraph (making the solution used for HTML worthless). This solution is very similar to the one used in
 451                 // xml::cleanID.
 452                 // indexType can only be used for singular and startofrange types!
 453                 docstring attrs;
 454                 if (!hasStartRange && !hasEndRange) {
 455                         attrs = indexType;
 456                 } else {
 457                         // Append an ID if uniqueness is not guaranteed across the document.
 458                         static QThreadStorage<set<docstring>> tKnownTermLists;
 459                         static QThreadStorage<int> tID;
 460
 461                         set<docstring> &knownTermLists = tKnownTermLists.localData();
 462                         int &ID = tID.localData();
 463
 464                         if (!tID.hasLocalData()) {
 465                                 tID.localData() = 0;
 466                         }
 467
 468                         // Modify the index terms to add the unique ID if needed.
 469                         docstring newIndexTerms = indexTerms;
 470                         if (knownTermLists.find(indexTerms) != knownTermLists.end()) {
 471                                 newIndexTerms += from_ascii(string("-") + to_string(ID));
 472
 473                                 // Only increment for the end of range, so that the same number is used for the start of range.
 474                                 if (hasEndRange) {
 475                                         ID++;
 476                                 }
 477                         }
 478
 479                         // Term list not yet known: add it to the set AFTER the end of range. After
 480                         if (knownTermLists.find(indexTerms) == knownTermLists.end() && hasEndRange) {
 481                                 knownTermLists.insert(indexTerms);
 482                         }
 483
 484                         // Generate the attributes.
 485                         docstring id = xml::cleanID(newIndexTerms);
 486                         if (hasStartRange) {
 487                                 attrs = indexType + " class=\"startofrange\" xml:id=\"" + id + "\"";
 488                         } else {
 489                                 attrs = " class=\"endofrange\" startref=\"" + id + "\"";
 490                         }
 491                 }
 492
 493                 // Handle the index terms (including the specific index for this entry).
 494                 if (hasEndRange) {
 495                         xs << xml::CompTag("indexterm", attrs);
 496                 } else {
 497                         xs << xml::StartTag("indexterm", attrs);
 498                         if (!terms.empty()) { // hasEndRange has no content.
 499                                 docstring attr;
 500                                 if (!sortAs.empty()) {
 501                                         attr = from_utf8("sortas='") + sortAs + from_utf8("'");
 502                                 }
 503
 504                                 xs << xml::StartTag("primary", attr);
 505                                 xs << terms[0];
 506                                 xs << xml::EndTag("primary");
 507                         }
 508                         if (terms.size() > 1) {
 509                                 xs << xml::StartTag("secondary");
 510                                 xs << terms[1];
 511                                 xs << xml::EndTag("secondary");
 512                         }
 513                         if (terms.size() > 2) {
 514                                 xs << xml::StartTag("tertiary");
 515                                 xs << terms[2];
 516                                 xs << xml::EndTag("tertiary");
 517                         }
 518
 519                         // Handle see and see also.
 520                         if (!see.empty()) {
 521                                 xs << xml::StartTag("see");
 522                                 xs << see;
 523                                 xs << xml::EndTag("see");
 524                         }
 525
 526                         if (!seeAlsoes.empty()) {
 527                                 for (auto &entry : seeAlsoes) {
 528                                         xs << xml::StartTag("seealso");
 529                                         xs << entry;
 530                                         xs << xml::EndTag("seealso");
 531                                 }
 532                         }
 533
 534                         // Close the entry.
 535                         xs << xml::EndTag("indexterm");
 536                 }
 537         }
 538 }
 539
 540
 541 docstring InsetIndex::xhtml(XMLStream & xs, OutputParams const &) const
 542 {
 543         // we just print an anchor, taking the paragraph ID from
 544         // our own interior paragraph, which doesn't get printed
 545         std::string const magic = paragraphs().front().magicLabel();
 546         std::string const attr = "id='" + magic + "'";
 547         xs << xml::CompTag("a", attr);
 548         return docstring();
 549 }
 550
 551
 552 bool InsetIndex::showInsetDialog(BufferView * bv) const
 553 {
 554         bv->showDialog("index", params2string(params_),
 555                         const_cast<InsetIndex *>(this));
 556         return true;
 557 }
 558
 559
 560 void InsetIndex::doDispatch(Cursor & cur, FuncRequest & cmd)
 561 {
 562         switch (cmd.action()) {
 563
 564         case LFUN_INSET_MODIFY: {
 565                 if (cmd.getArg(0) == "changetype") {
 566                         cur.recordUndoInset(this);
 567                         params_.index = from_utf8(cmd.getArg(1));
 568                         break;
 569                 }
 570                 InsetIndexParams params;
 571                 InsetIndex::string2params(to_utf8(cmd.argument()), params);
 572                 cur.recordUndoInset(this);
 573                 params_.index = params.index;
 574                 params_.range = params.range;
 575                 params_.pagefmt = params.pagefmt;
 576                 // what we really want here is a TOC update, but that means
 577                 // a full buffer update
 578                 cur.forceBufferUpdate();
 579                 break;
 580         }
 581
 582         case LFUN_INSET_DIALOG_UPDATE:
 583                 cur.bv().updateDialog("index", params2string(params_));
 584                 break;
 585
 586         default:
 587                 InsetCollapsible::doDispatch(cur, cmd);
 588                 break;
 589         }
 590 }
 591
 592
 593 bool InsetIndex::getStatus(Cursor & cur, FuncRequest const & cmd,
 594                 FuncStatus & flag) const
 595 {
 596         switch (cmd.action()) {
 597
 598         case LFUN_INSET_MODIFY:
 599                 if (cmd.getArg(0) == "changetype") {
 600                         docstring const newtype = from_utf8(cmd.getArg(1));
 601                         Buffer const & realbuffer = *buffer().masterBuffer();
 602                         IndicesList const & indiceslist = realbuffer.params().indiceslist();
 603                         Index const * index = indiceslist.findShortcut(newtype);
 604                         flag.setEnabled(index != 0);
 605                         flag.setOnOff(
 606                                 from_utf8(cmd.getArg(1)) == params_.index);
 607                         return true;
 608                 }
 609                 return InsetCollapsible::getStatus(cur, cmd, flag);
 610
 611         case LFUN_INSET_DIALOG_UPDATE: {
 612                 Buffer const & realbuffer = *buffer().masterBuffer();
 613                 flag.setEnabled(realbuffer.params().use_indices);
 614                 return true;
 615         }
 616
 617         case LFUN_INDEXMACRO_INSERT:
 618                 return macrosPossible(cmd.getArg(0));
 619
 620         default:
 621                 return InsetCollapsible::getStatus(cur, cmd, flag);
 622         }
 623 }
 624
 625
 626 void InsetIndex::getSortkey(otexstream & os, OutputParams const & runparams) const
 627 {
 628         Paragraph const & par = paragraphs().front();
 629         InsetList::const_iterator it = par.insetList().begin();
 630         for (; it != par.insetList().end(); ++it) {
 631                 Inset & inset = *it->inset;
 632                 if (inset.lyxCode() == INDEXMACRO_SORTKEY_CODE) {
 633                         InsetIndexMacro const & iim =
 634                                 static_cast<InsetIndexMacro const &>(inset);
 635                         iim.getLatex(os, runparams);
 636                         return;
 637                 }
 638         }
 639 }
 640
 641
 642 docstring InsetIndex::getSortkeyAsText(OutputParams const & runparams) const
 643 {
 644         Paragraph const & par = paragraphs().front();
 645         InsetList::const_iterator it = par.insetList().begin();
 646         for (; it != par.insetList().end(); ++it) {
 647                 Inset & inset = *it->inset;
 648                 if (inset.lyxCode() == INDEXMACRO_SORTKEY_CODE) {
 649                         otexstringstream os;
 650                         InsetIndexMacro const & iim =
 651                                 static_cast<InsetIndexMacro const &>(inset);
 652                         iim.getLatex(os, runparams);
 653                         return os.str();
 654                 }
 655         }
 656         return from_ascii("");
 657 }
 658
 659
 660 void InsetIndex::getSubentries(otexstream & os, OutputParams const & runparams) const
 661 {
 662         Paragraph const & par = paragraphs().front();
 663         InsetList::const_iterator it = par.insetList().begin();
 664         int i = 0;
 665         for (; it != par.insetList().end(); ++it) {
 666                 Inset & inset = *it->inset;
 667                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 668                         InsetIndexMacro const & iim =
 669                                 static_cast<InsetIndexMacro const &>(inset);
 670                         if (iim.params().type == InsetIndexMacroParams::Subindex) {
 671                                 ++i;
 672                                 if (i > 2)
 673                                         return;
 674                                 os << "!";
 675                                 iim.getLatex(os, runparams);
 676                         }
 677                 }
 678         }
 679 }
 680
 681
 682 std::vector<docstring> InsetIndex::getSubentriesAsText(OutputParams const & runparams) const
 683 {
 684         std::vector<docstring> subentries;
 685
 686         Paragraph const & par = paragraphs().front();
 687         InsetList::const_iterator it = par.insetList().begin();
 688         int i = 0;
 689         for (; it != par.insetList().end(); ++it) {
 690                 Inset & inset = *it->inset;
 691                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 692                         InsetIndexMacro const & iim =
 693                                 static_cast<InsetIndexMacro const &>(inset);
 694                         if (iim.params().type == InsetIndexMacroParams::Subindex) {
 695                                 ++i;
 696                                 if (i > 2)
 697                                         break;
 698
 699                                 otexstringstream os;
 700                                 iim.getLatex(os, runparams);
 701                                 subentries.emplace_back(os.str());
 702                         }
 703                 }
 704         }
 705
 706         return subentries;
 707 }
 708
 709
 710 docstring InsetIndex::getMainSubentryAsText(OutputParams const & runparams) const
 711 {
 712         otexstringstream os;
 713         InsetText::latex(os, runparams);
 714         return os.str();
 715 }
 716
 717
 718 void InsetIndex::getSeeRefs(otexstream & os, OutputParams const & runparams) const
 719 {
 720         Paragraph const & par = paragraphs().front();
 721         InsetList::const_iterator it = par.insetList().begin();
 722         for (; it != par.insetList().end(); ++it) {
 723                 Inset & inset = *it->inset;
 724                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 725                         InsetIndexMacro const & iim =
 726                                 static_cast<InsetIndexMacro const &>(inset);
 727                         if (iim.params().type == InsetIndexMacroParams::See
 728                             || iim.params().type == InsetIndexMacroParams::Seealso) {
 729                                 iim.getLatex(os, runparams);
 730                                 return;
 731                         }
 732                 }
 733         }
 734 }
 735
 736
 737 docstring InsetIndex::getSeeAsText(OutputParams const & runparams) const
 738 {
 739         Paragraph const & par = paragraphs().front();
 740         InsetList::const_iterator it = par.insetList().begin();
 741         for (; it != par.insetList().end(); ++it) {
 742                 Inset & inset = *it->inset;
 743                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 744                         InsetIndexMacro const & iim =
 745                                 static_cast<InsetIndexMacro const &>(inset);
 746                         if (iim.params().type == InsetIndexMacroParams::See) {
 747                                 otexstringstream os;
 748                                 iim.getLatex(os, runparams);
 749                                 return os.str();
 750                         }
 751                 }
 752         }
 753         return from_ascii("");
 754 }
 755
 756
 757 std::vector<docstring> InsetIndex::getSeeAlsoesAsText(OutputParams const & runparams) const
 758 {
 759         std::vector<docstring> seeAlsoes;
 760
 761         Paragraph const & par = paragraphs().front();
 762         InsetList::const_iterator it = par.insetList().begin();
 763         for (; it != par.insetList().end(); ++it) {
 764                 Inset & inset = *it->inset;
 765                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 766                         InsetIndexMacro const & iim =
 767                                 static_cast<InsetIndexMacro const &>(inset);
 768                         if (iim.params().type == InsetIndexMacroParams::Seealso) {
 769                                 otexstringstream os;
 770                                 iim.getLatex(os, runparams);
 771                                 seeAlsoes.emplace_back(os.str());
 772                         }
 773                 }
 774         }
 775
 776         return seeAlsoes;
 777 }
 778
 779
 780 namespace {
 781
 782 bool hasInsetWithCode(const InsetIndex * const inset_index, const InsetCode code,
 783                                           const std::set<InsetIndexMacroParams::Type> types = {})
 784 {
 785         Paragraph const & par = inset_index->paragraphs().front();
 786         InsetList::const_iterator it = par.insetList().begin();
 787         for (; it != par.insetList().end(); ++it) {
 788                 Inset & inset = *it->inset;
 789                 if (inset.lyxCode() == code) {
 790                         if (types.empty())
 791                                 return true;
 792
 793                         LASSERT(code == INDEXMACRO_CODE, return false);
 794                         InsetIndexMacro const & iim =
 795                                         static_cast<InsetIndexMacro const &>(inset);
 796                         if (types.find(iim.params().type) != types.end())
 797                                 return true;
 798                 }
 799         }
 800         return false;
 801 }
 802
 803 } // namespace
 804
 805
 806 bool InsetIndex::hasSubentries() const
 807 {
 808         return hasInsetWithCode(this, INDEXMACRO_CODE, {InsetIndexMacroParams::Subindex});
 809 }
 810
 811
 812 bool InsetIndex::hasSeeRef() const
 813 {
 814         return hasInsetWithCode(this, INDEXMACRO_CODE, {InsetIndexMacroParams::See, InsetIndexMacroParams::Seealso});
 815 }
 816
 817
 818 bool InsetIndex::hasSortKey() const
 819 {
 820         return hasInsetWithCode(this, INDEXMACRO_SORTKEY_CODE);
 821 }
 822
 823
 824 bool InsetIndex::macrosPossible(string const type) const
 825 {
 826         if (type != "see" && type != "seealso"
 827             && type != "sortkey" && type != "subindex")
 828                 return false;
 829
 830         Paragraph const & par = paragraphs().front();
 831         InsetList::const_iterator it = par.insetList().begin();
 832         int subidxs = 0;
 833         for (; it != par.insetList().end(); ++it) {
 834                 Inset & inset = *it->inset;
 835                 if (type == "sortkey" && inset.lyxCode() == INDEXMACRO_SORTKEY_CODE)
 836                         return false;
 837                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 838                         InsetIndexMacro const & iim = static_cast<InsetIndexMacro const &>(inset);
 839                         if ((type == "see" || type == "seealso")
 840                              && (iim.params().type == InsetIndexMacroParams::See
 841                                  || iim.params().type == InsetIndexMacroParams::Seealso))
 842                                 return false;
 843                         if (type == "subindex"
 844                              && iim.params().type == InsetIndexMacroParams::Subindex) {
 845                                 ++subidxs;
 846                                 if (subidxs > 1)
 847                                         return false;
 848                         }
 849                 }
 850         }
 851         return true;
 852 }
 853
 854
 855 ColorCode InsetIndex::labelColor() const
 856 {
 857         if (params_.index.empty() || params_.index == from_ascii("idx"))
 858                 return InsetCollapsible::labelColor();
 859         // FIXME UNICODE
 860         ColorCode c = lcolor.getFromLyXName(to_utf8(params_.index)
 861                                             + "@" + buffer().fileName().absFileName());
 862         if (c == Color_none)
 863                 c = InsetCollapsible::labelColor();
 864         return c;
 865 }
 866
 867
 868 docstring InsetIndex::toolTip(BufferView const &, int, int) const
 869 {
 870         docstring tip = _("Index Entry");
 871         if (buffer().params().use_indices && !params_.index.empty()) {
 872                 Buffer const & realbuffer = *buffer().masterBuffer();
 873                 IndicesList const & indiceslist = realbuffer.params().indiceslist();
 874                 tip += " (";
 875                 Index const * index = indiceslist.findShortcut(params_.index);
 876                 if (!index)
 877                         tip += _("unknown type!");
 878                 else
 879                         tip += index->index();
 880                 tip += ")";
 881         }
 882         tip += ": ";
 883         docstring res = toolTipText(tip);
 884         if (!insetindexpagerangetranslator_loc().find(params_.range).empty())
 885                 res += "\n" + insetindexpagerangetranslator_loc().find(params_.range);
 886         if (!params_.pagefmt.empty() && params_.pagefmt != "default") {
 887                 res += "\n" + _("Pagination format:") + " ";
 888                 if (params_.pagefmt == "textbf")
 889                         res += _("bold");
 890                 else if (params_.pagefmt == "textit")
 891                         res += _("italic");
 892                 else if (params_.pagefmt == "emph")
 893                         res += _("emphasized");
 894                 else
 895                         res += from_utf8(params_.pagefmt);
 896         }
 897         return res;
 898 }
 899
 900
 901 docstring const InsetIndex::buttonLabel(BufferView const & bv) const
 902 {
 903         InsetLayout const & il = getLayout();
 904         docstring label = translateIfPossible(il.labelstring());
 905
 906         if (buffer().params().use_indices && !params_.index.empty()) {
 907                 Buffer const & realbuffer = *buffer().masterBuffer();
 908                 IndicesList const & indiceslist = realbuffer.params().indiceslist();
 909                 label += " (";
 910                 Index const * index = indiceslist.findShortcut(params_.index);
 911                 if (!index)
 912                         label += _("unknown type!");
 913                 else
 914                         label += index->index();
 915                 label += ")";
 916         }
 917
 918         docstring res;
 919         if (!il.contentaslabel() || geometry(bv) != ButtonOnly)
 920                 res = label;
 921         else
 922                 res = getNewLabel(label);
 923         if (!insetindexpagerangetranslator_latex().find(params_.range).empty())
 924                 res += " " + from_ascii(insetindexpagerangetranslator_latex().find(params_.range));
 925         return res;
 926 }
 927
 928
 929 void InsetIndex::write(ostream & os) const
 930 {
 931         os << to_utf8(layoutName());
 932         params_.write(os);
 933         InsetCollapsible::write(os);
 934 }
 935
 936
 937 void InsetIndex::read(Lexer & lex)
 938 {
 939         params_.read(lex);
 940         InsetCollapsible::read(lex);
 941 }
 942
 943
 944 string InsetIndex::params2string(InsetIndexParams const & params)
 945 {
 946         ostringstream data;
 947         data << "index";
 948         params.write(data);
 949         return data.str();
 950 }
 951
 952
 953 void InsetIndex::string2params(string const & in, InsetIndexParams & params)
 954 {
 955         params = InsetIndexParams();
 956         if (in.empty())
 957                 return;
 958
 959         istringstream data(in);
 960         Lexer lex;
 961         lex.setStream(data);
 962         lex.setContext("InsetIndex::string2params");
 963         lex >> "index";
 964         params.read(lex);
 965 }
 966
 967
 968 void InsetIndex::addToToc(DocIterator const & cpit, bool output_active,
 969                                                   UpdateType utype, TocBackend & backend) const
 970 {
 971         DocIterator pit = cpit;
 972         pit.push_back(CursorSlice(const_cast<InsetIndex &>(*this)));
 973         docstring str;
 974         string type = "index";
 975         if (buffer().masterBuffer()->params().use_indices)
 976                 type += ":" + to_utf8(params_.index);
 977         // this is unlikely to be terribly long
 978         text().forOutliner(str, INT_MAX);
 979         TocBuilder & b = backend.builder(type);
 980         b.pushItem(pit, str, output_active);
 981         // Proceed with the rest of the inset.
 982         InsetCollapsible::addToToc(cpit, output_active, utype, backend);
 983         b.pop();
 984 }
 985
 986
 987 void InsetIndex::validate(LaTeXFeatures & features) const
 988 {
 989         if (buffer().masterBuffer()->params().use_indices
 990             && !params_.index.empty()
 991             && params_.index != "idx")
 992                 features.require("splitidx");
 993         InsetCollapsible::validate(features);
 994 }
 995
 996
 997 string InsetIndex::contextMenuName() const
 998 {
 999         return "context-index";
1000 }
1001
1002
1003 string InsetIndex::contextMenu(BufferView const & bv, int x, int y) const
1004 {
1005         // We override the implementation of InsetCollapsible,
1006         // because we have eytra entries.
1007         string owncm = "context-edit-index;";
1008         return owncm + InsetCollapsible::contextMenu(bv, x, y);
1009 }
1010
1011
1012 bool InsetIndex::hasSettings() const
1013 {
1014         return true;
1015 }
1016
1017
1018 bool InsetIndex::insetAllowed(InsetCode code) const
1019 {
1020         switch (code) {
1021         case INDEXMACRO_CODE:
1022         case INDEXMACRO_SORTKEY_CODE:
1023                 return true;
1024         case INDEX_CODE:
1025                 return false;
1026         default:
1027                 return InsetCollapsible::insetAllowed(code);
1028         }
1029 }
1030
1031
1032 /////////////////////////////////////////////////////////////////////
1033 //
1034 // InsetIndexParams
1035 //
1036 ///////////////////////////////////////////////////////////////////////
1037
1038
1039 void InsetIndexParams::write(ostream & os) const
1040 {
1041         os << ' ';
1042         if (!index.empty())
1043                 os << to_utf8(index);
1044         else
1045                 os << "idx";
1046         os << '\n';
1047         os << "range "
1048            << insetindexpagerangetranslator().find(range)
1049            << '\n';
1050         os << "pageformat "
1051            << pagefmt
1052            << '\n';
1053 }
1054
1055
1056 void InsetIndexParams::read(Lexer & lex)
1057 {
1058         if (lex.eatLine())
1059                 index = lex.getDocString();
1060         else
1061                 index = from_ascii("idx");
1062         if (lex.checkFor("range")) {
1063                 string st = lex.getString();
1064                 if (lex.eatLine()) {
1065                         st = lex.getString();
1066                         range = insetindexpagerangetranslator().find(lex.getString());
1067                 }
1068         }
1069         if (lex.checkFor("pageformat") && lex.eatLine()) {
1070                 pagefmt = lex.getString();
1071         }
1072 }
1073
1074
1075 /////////////////////////////////////////////////////////////////////
1076 //
1077 // InsetPrintIndex
1078 //
1079 ///////////////////////////////////////////////////////////////////////
1080
1081 InsetPrintIndex::InsetPrintIndex(Buffer * buf, InsetCommandParams const & p)
1082         : InsetCommand(buf, p)
1083 {}
1084
1085
1086 ParamInfo const & InsetPrintIndex::findInfo(string const & /* cmdName */)
1087 {
1088         static ParamInfo param_info_;
1089         if (param_info_.empty()) {
1090                 param_info_.add("type", ParamInfo::LATEX_OPTIONAL,
1091                                 ParamInfo::HANDLING_ESCAPE);
1092                 param_info_.add("name", ParamInfo::LATEX_OPTIONAL,
1093                                 ParamInfo::HANDLING_LATEXIFY);
1094                 param_info_.add("literal", ParamInfo::LYX_INTERNAL);
1095         }
1096         return param_info_;
1097 }
1098
1099
1100 docstring InsetPrintIndex::screenLabel() const
1101 {
1102         bool const printall = suffixIs(getCmdName(), '*');
1103         bool const multind = buffer().masterBuffer()->params().use_indices;
1104         if ((!multind
1105              && getParam("type") == from_ascii("idx"))
1106             || (getParam("type").empty() && !printall))
1107                 return _("Index");
1108         Buffer const & realbuffer = *buffer().masterBuffer();
1109         IndicesList const & indiceslist = realbuffer.params().indiceslist();
1110         Index const * index = indiceslist.findShortcut(getParam("type"));
1111         if (!index && !printall)
1112                 return _("Unknown index type!");
1113         docstring res = printall ? _("All indexes") : index->index();
1114         if (!multind)
1115                 res += " (" + _("non-active") + ")";
1116         else if (contains(getCmdName(), "printsubindex"))
1117                 res += " (" + _("subindex") + ")";
1118         return res;
1119 }
1120
1121
1122 bool InsetPrintIndex::isCompatibleCommand(string const & s)
1123 {
1124         return s == "printindex" || s == "printsubindex"
1125                 || s == "printindex*" || s == "printsubindex*";
1126 }
1127
1128
1129 void InsetPrintIndex::doDispatch(Cursor & cur, FuncRequest & cmd)
1130 {
1131         switch (cmd.action()) {
1132
1133         case LFUN_INSET_MODIFY: {
1134                 if (cmd.argument() == from_ascii("toggle-subindex")) {
1135                         string scmd = getCmdName();
1136                         if (contains(scmd, "printindex"))
1137                                 scmd = subst(scmd, "printindex", "printsubindex");
1138                         else
1139                                 scmd = subst(scmd, "printsubindex", "printindex");
1140                         cur.recordUndo();
1141                         setCmdName(scmd);
1142                         break;
1143                 } else if (cmd.argument() == from_ascii("check-printindex*")) {
1144                         string scmd = getCmdName();
1145                         if (suffixIs(scmd, '*'))
1146                                 break;
1147                         scmd += '*';
1148                         cur.recordUndo();
1149                         setParam("type", docstring());
1150                         setCmdName(scmd);
1151                         break;
1152                 }
1153                 InsetCommandParams p(INDEX_PRINT_CODE);
1154                 // FIXME UNICODE
1155                 InsetCommand::string2params(to_utf8(cmd.argument()), p);
1156                 if (p.getCmdName().empty()) {
1157                         cur.noScreenUpdate();
1158                         break;
1159                 }
1160                 cur.recordUndo();
1161                 setParams(p);
1162                 break;
1163         }
1164
1165         default:
1166                 InsetCommand::doDispatch(cur, cmd);
1167                 break;
1168         }
1169 }
1170
1171
1172 bool InsetPrintIndex::getStatus(Cursor & cur, FuncRequest const & cmd,
1173         FuncStatus & status) const
1174 {
1175         switch (cmd.action()) {
1176
1177         case LFUN_INSET_MODIFY: {
1178                 if (cmd.argument() == from_ascii("toggle-subindex")) {
1179                         status.setEnabled(buffer().masterBuffer()->params().use_indices);
1180                         status.setOnOff(contains(getCmdName(), "printsubindex"));
1181                         return true;
1182                 } else if (cmd.argument() == from_ascii("check-printindex*")) {
1183                         status.setEnabled(buffer().masterBuffer()->params().use_indices);
1184                         status.setOnOff(suffixIs(getCmdName(), '*'));
1185                         return true;
1186                 } if (cmd.getArg(0) == "index_print"
1187                     && cmd.getArg(1) == "CommandInset") {
1188                         InsetCommandParams p(INDEX_PRINT_CODE);
1189                         InsetCommand::string2params(to_utf8(cmd.argument()), p);
1190                         if (suffixIs(p.getCmdName(), '*')) {
1191                                 status.setEnabled(true);
1192                                 status.setOnOff(false);
1193                                 return true;
1194                         }
1195                         Buffer const & realbuffer = *buffer().masterBuffer();
1196                         IndicesList const & indiceslist =
1197                                 realbuffer.params().indiceslist();
1198                         Index const * index = indiceslist.findShortcut(p["type"]);
1199                         status.setEnabled(index != 0);
1200                         status.setOnOff(p["type"] == getParam("type"));
1201                         return true;
1202                 } else
1203                         return InsetCommand::getStatus(cur, cmd, status);
1204         }
1205
1206         case LFUN_INSET_DIALOG_UPDATE: {
1207                 status.setEnabled(buffer().masterBuffer()->params().use_indices);
1208                 return true;
1209         }
1210
1211         default:
1212                 return InsetCommand::getStatus(cur, cmd, status);
1213         }
1214 }
1215
1216
1217 void InsetPrintIndex::updateBuffer(ParIterator const &, UpdateType, bool const /*deleted*/)
1218 {
1219         Index const * index =
1220                 buffer().masterParams().indiceslist().findShortcut(getParam("type"));
1221         if (index)
1222                 setParam("name", index->index());
1223 }
1224
1225
1226 void InsetPrintIndex::latex(otexstream & os, OutputParams const & runparams_in) const
1227 {
1228         if (!buffer().masterBuffer()->params().use_indices) {
1229                 if (getParam("type") == from_ascii("idx"))
1230                         os << "\\printindex" << termcmd;
1231                 return;
1232         }
1233         OutputParams runparams = runparams_in;
1234         os << getCommand(runparams);
1235 }
1236
1237
1238 void InsetPrintIndex::validate(LaTeXFeatures & features) const
1239 {
1240         features.require("makeidx");
1241         if (buffer().masterBuffer()->params().use_indices)
1242                 features.require("splitidx");
1243         InsetCommand::validate(features);
1244 }
1245
1246
1247 string InsetPrintIndex::contextMenuName() const
1248 {
1249         return buffer().masterBuffer()->params().use_indices ?
1250                 "context-indexprint" : string();
1251 }
1252
1253
1254 bool InsetPrintIndex::hasSettings() const
1255 {
1256         return buffer().masterBuffer()->params().use_indices;
1257 }
1258
1259
1260 class IndexEntry
1261 {
1262 public:
1263         /// Builds an entry for the index.
1264         IndexEntry(const InsetIndex * inset, OutputParams const * runparams) : inset_(inset), runparams_(runparams)
1265         {
1266                 LASSERT(runparams, return);
1267
1268                 // Convert the inset as text. The resulting text usually only contains an XHTML anchor (<a id='...'/>) and text.
1269                 odocstringstream entry;
1270                 OutputParams ours = *runparams;
1271                 ours.for_toc = false;
1272                 inset_->plaintext(entry, ours);
1273                 entry_ = entry.str();
1274
1275                 // Determine in which index this entry belongs to.
1276                 if (inset_->buffer().masterBuffer()->params().use_indices) {
1277                         index_ = inset_->params_.index;
1278                 }
1279
1280                 // Attempt parsing the inset.
1281                 if (isModern())
1282                         parseAsModern();
1283                 else
1284                         parseAsLegacy();
1285         }
1286
1287         /// When parsing this entry, some errors may be found; they are reported as a single string.
1288         // It is up to the caller to send this string to LYXERR and the output file, as needed.
1289         const docstring & output_error() const
1290         {
1291                 return output_error_;
1292         }
1293
1294         void output_error(XMLStream xs) const
1295         {
1296                 LYXERR0(output_error());
1297                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + output_error() + from_utf8(" -->\n"));
1298         }
1299
1300
1301 private:
1302         bool isModern()
1303         {
1304                 std::cout << to_utf8(entry_) << std::endl;
1305
1306                 // If a modern parameter is present, this is definitely a modern index inset. Similarly, if it contains the
1307                 // usual LaTeX symbols (!|@), then it is definitely a legacy index inset. Otherwise, if it has features of
1308                 // neither, it is both: consider this is a modern inset, to trigger the least complex code. Mixing both types
1309                 // is not allowed (i.e. behaviour is undefined).
1310                 const bool is_definitely_modern = inset_->hasSortKey() || inset_->hasSeeRef() || inset_->hasSubentries()
1311                                             || inset_->params_.range != InsetIndexParams::PageRange::None;
1312                 const bool is_definitely_legacy = entry_.find('@') != std::string::npos
1313                                 || entry_.find('|') != std::string::npos || entry_.find('!') != std::string::npos;
1314
1315                 if (is_definitely_legacy && is_definitely_modern)
1316                         output_error_ += from_utf8("Mix of index properties and raw LaTeX index commands is unsupported. ");
1317
1318                 // Truth table:
1319                 // - is_definitely_modern == true:
1320                 //   - is_definitely_legacy == true: error (return whatever)
1321                 //   - is_definitely_legacy == false: return modern
1322                 // - is_definitely_modern == false:
1323                 //   - is_definitely_legacy == true: return legacy
1324                 //   - is_definitely_legacy == false: return modern
1325                 return !is_definitely_legacy;
1326         }
1327
1328         void parseAsModern()
1329         {
1330                 LASSERT(runparams_, return);
1331
1332                 if (inset_->hasSortKey()) {
1333                         sort_as_ = inset_->getSortkeyAsText(*runparams_);
1334                 }
1335
1336                 terms_ = inset_->getSubentriesAsText(*runparams_);
1337                 // The main term is not present in the vector, as it's not a subentry. The main index term is inserted raw in
1338                 // the index inset. Considering that the user either uses the new or the legacy mechanism, the main term is the
1339                 // full string within this inset (i.e. without the subinsets).
1340                 terms_.insert(terms_.begin(), inset_->getMainSubentryAsText(*runparams_));
1341
1342                 has_start_range_ = inset_->params_.range == InsetIndexParams::PageRange::Start;
1343                 has_end_range_ = inset_->params_.range == InsetIndexParams::PageRange::End;
1344
1345                 see_ = inset_->getSeeAsText(*runparams_);
1346                 see_alsoes_ = inset_->getSeeAlsoesAsText(*runparams_);
1347         }
1348
1349         void parseAsLegacy() {
1350                 // Determine if some features are known not to be supported. For now, this is only formatting like
1351                 // \index{alpha@\textbf{alpha}} or \index{alpha@$\alpha$}.
1352                 // @ is supported, but only for sorting, without specific formatting.
1353                 if (entry_.find(from_utf8("@\\")) != lyx::docstring::npos) {
1354                         output_error_ += from_utf8("Unsupported feature: an index entry contains an @\\. "
1355                                                    "Complete entry: \"") + entry_ + from_utf8("\". ");
1356                 }
1357                 if (entry_.find(from_utf8("@$")) != lyx::docstring::npos) {
1358                         output_error_ += from_utf8("Unsupported feature: an index entry contains an @$. "
1359                                                    "Complete entry: \"") + entry_ + from_utf8("\". ");
1360                 }
1361
1362                 // Split the string into its main constituents: terms, and command (see, see also, range).
1363                 size_t positionVerticalBar = entry_.find(from_ascii("|")); // What comes before | is (sub)(sub)entries.
1364                 docstring indexTerms = entry_.substr(0, positionVerticalBar);
1365                 docstring command;
1366                 if (positionVerticalBar != lyx::docstring::npos) {
1367                         command = entry_.substr(positionVerticalBar + 1);
1368                 }
1369
1370                 // Handle sorting issues, with @.
1371                 vector<docstring> sortingElements = getVectorFromString(indexTerms, from_ascii("@"), false);
1372                 if (sortingElements.size() == 2) {
1373                         sort_as_ = sortingElements[0];
1374                         indexTerms = sortingElements[1];
1375                 }
1376
1377                 // Handle entries, subentries, and subsubentries.
1378                 terms_ = getVectorFromString(indexTerms, from_ascii("!"), false);
1379
1380                 // Handle ranges. Happily, (| and |) can only be at the end of the string!
1381                 has_start_range_ = entry_.find(from_ascii("|(")) != lyx::docstring::npos;
1382                 has_end_range_ = entry_.find(from_ascii("|)")) != lyx::docstring::npos;
1383
1384                 // - Remove the ranges from the command if they do not appear at the beginning.
1385                 size_t range_index = 0;
1386                 while ((range_index = command.find(from_utf8("|("), range_index)) != std::string::npos)
1387                         command.erase(range_index, 1);
1388                 range_index = 0;
1389                 while ((range_index = command.find(from_utf8("|)"), range_index)) != std::string::npos)
1390                         command.erase(range_index, 1);
1391
1392                 // - Remove the ranges when they are the only vertical bar in the complete string.
1393                 if (command[0] == '(' || command[0] == ')')
1394                         command.erase(0, 1);
1395
1396                 // Handle see and seealso. As "see" is a prefix of "seealso", the order of the comparisons is important.
1397                 // Both commands are mutually exclusive!
1398                 if (command.substr(0, 3) == "see") {
1399                         // Unescape brackets.
1400                         size_t index_argument_begin = 0;
1401                         while ((index_argument_begin = command.find(from_utf8("\\{"), index_argument_begin)) != std::string::npos)
1402                                 command.erase(index_argument_begin, 1);
1403                         size_t index_argument_end = 0;
1404                         while ((index_argument_end = command.find(from_utf8("\\}"), index_argument_end)) != std::string::npos)
1405                                 command.erase(index_argument_end, 1);
1406
1407                         // Retrieve the part between brackets, and remove the complete seealso.
1408                         size_t position_opening_bracket = command.find(from_ascii("{"));
1409                         size_t position_closing_bracket = command.find(from_ascii("}"));
1410                         docstring argument = command.substr(position_opening_bracket + 1,
1411                                                                                                 position_closing_bracket - position_opening_bracket - 1);
1412
1413                         // Parse the argument of referenced entries (or a single one for see).
1414                         if (command.substr(0, 7) == "seealso") {
1415                                 see_alsoes_ = getVectorFromString(argument, from_ascii(","), false);
1416                         } else {
1417                                 see_ = argument;
1418
1419                                 if (see_.find(from_ascii(",")) != std::string::npos) {
1420                                         output_error_ += from_utf8("Several index_argument_end terms found as \"see\"! Only one is "
1421                                                                    "acceptable. Complete entry: \"") + entry_ + from_utf8("\". ");
1422                                 }
1423                         }
1424
1425                         // Remove the complete see/seealso from the commands, in case there is something else to parse.
1426                         command = command.substr(position_closing_bracket + 1);
1427                 }
1428
1429                 // Some parts of the strings are not parsed, as they do not have anything matching in DocBook or XHTML:
1430                 // things like formatting the entry or the page number, other strings for sorting.
1431                 // https://wiki.lyx.org/Tips/Indexing
1432                 // If there are such things in the index entry, then this code may miserably fail. For example, for
1433                 // "Peter|(textbf", no range will be detected.
1434                 if (!command.empty()) {
1435                         output_error_ += from_utf8("Unsupported feature: an index entry contains a | with an unsupported command, ")
1436                                          + command + from_utf8(". Complete entry: \"") + entry_ + from_utf8("\". ");
1437                 }
1438         }
1439
1440 public:
1441         int level() const {
1442                 return terms_.size();
1443         }
1444
1445         const std::vector<docstring>& terms() const {
1446                 return terms_;
1447         }
1448
1449         std::vector<docstring>& terms() {
1450                 return terms_;
1451         }
1452
1453         const InsetIndex* inset() const {
1454                 return inset_;
1455         }
1456
1457 private:
1458         // Input inset. These should only be used when parsing the inset (either parseAsModern or parseAsLegacy, called in
1459         // the constructor).
1460         const InsetIndex * inset_;
1461         OutputParams const * runparams_;
1462         docstring entry_;
1463         docstring index_; // Useful when there are multiple indices in the same document.
1464
1465         // Errors, concatenated as a single string, available as soon as parsing is done, const afterwards (i.e. once
1466         // constructor is done).
1467         docstring output_error_;
1468
1469         // Parsed index entry.
1470         std::vector<docstring> terms_; // Up to three entries, in general.
1471         docstring sort_as_;
1472         docstring command_;
1473         bool has_start_range_;
1474         bool has_end_range_;
1475         docstring see_;
1476         vector<docstring> see_alsoes_;
1477
1478         // Operators used for sorting entries (alphabetical order).
1479         friend bool operator<(IndexEntry const & lhs, IndexEntry const & rhs);
1480 };
1481
1482 bool operator<(IndexEntry const & lhs, IndexEntry const & rhs)
1483 {
1484         if (lhs.terms_.empty())
1485                 return false;
1486
1487         for (int i = 0; i < min(rhs.terms_.size(), lhs.terms_.size()); ++i) {
1488                 int comp = compare_no_case(lhs.terms_[i], rhs.terms_[i]);
1489                 if (comp != 0)
1490                         return comp < 0;
1491         }
1492         return false;
1493 }
1494
1495
1496 namespace {
1497 std::string generateCssClassAtDepth(unsigned depth) {
1498         std::string css_class = "entry";
1499
1500         while (depth > 0) {
1501                 depth -= 1;
1502                 css_class.insert(0, "sub");
1503         }
1504
1505         return css_class;
1506 }
1507
1508 struct IndexNode {
1509         std::vector<IndexEntry> entries;
1510         std::vector<IndexNode*> children;
1511 };
1512
1513 docstring termAtLevel(const IndexNode* node, unsigned depth)
1514 {
1515         // The typical entry has a depth of 1 to 3: the call stack would then be at most 4 (due to the root node). This
1516         // function could be made constant time by copying the term in each node, but that would make data duplication that
1517         // may fall out of sync; the performance benefit would probably be negligible.
1518         if (!node->entries.empty()) {
1519                 LASSERT(node->entries.begin()->terms().size() >= depth + 1, return from_ascii(""));
1520                 return node->entries.begin()->terms()[depth];
1521         }
1522
1523         if (!node->children.empty()) {
1524                 return termAtLevel(*node->children.begin(), depth);
1525         }
1526
1527         LASSERT(false, return from_ascii(""));
1528 }
1529
1530 void insertIntoNode(const IndexEntry& entry, IndexNode* node, unsigned depth = 0)
1531 {
1532         // depth == 0 is for the root, not yet the index, hence the increase when going to vector size.
1533         for (IndexNode* child : node->children) {
1534                 if (entry.terms()[depth] == termAtLevel(child, depth)) {
1535                         if (depth + 1 == entry.terms().size()) { // == child.entries.begin()->terms().size()
1536                                 // All term entries match: it's an entry.
1537                                 child->entries.emplace_back(entry);
1538                                 return;
1539                         } else {
1540                                 insertIntoNode(entry, child, depth + 1);
1541                                 return;
1542                         }
1543                 }
1544         }
1545
1546         // Out of the loop: no matching child found, create a new (possibly nested) child for this entry. Due to the
1547         // possibility of nestedness, only insert the current entry when the right level is reached. This is needed if the
1548         // first entry for a word has several levels that never appeared.
1549         // In particular, this case is called for the first entry.
1550         IndexNode* new_node = node;
1551         do {
1552                 new_node->children.emplace_back(new IndexNode{{}, {}});
1553                 new_node = new_node->children.back();
1554                 depth += 1;
1555         } while (depth + 1 <= entry.terms().size()); // depth == 0: root node, no text associated.
1556         new_node->entries.emplace_back(entry);
1557 }
1558
1559 IndexNode* buildIndexTree(vector<IndexEntry>& entries)
1560 {
1561         // Sort the entries, first on the main entry, then the subentry, then the subsubentry,
1562         // thanks to the implementation of operator<.
1563         // If this operation is not performed, the algorithm below is no more correct (and ensuring that it works with
1564         // unsorted entries would make its complexity blow up).
1565         stable_sort(entries.begin(), entries.end());
1566
1567         // Cook the index into a nice tree data structure: entries at a given level in the index as a node, with subentries
1568         // as children.
1569         auto* index_root = new IndexNode{{}, {}};
1570         for (const IndexEntry& entry : entries) {
1571                 insertIntoNode(entry, index_root);
1572         }
1573
1574         return index_root;
1575 }
1576
1577 void outputIndexPage(XMLStream & xs, const IndexNode* root_node, unsigned depth = 0)
1578 {
1579         LASSERT(root_node->entries.size() + root_node->children.size() > 0, return);
1580
1581         xs << xml::StartTag("li", "class='" + generateCssClassAtDepth(depth) + "'");
1582         xs << xml::CR();
1583         xs << XMLStream::ESCAPE_NONE << termAtLevel(root_node, depth);
1584         // By tree assumption, all the entries at this node have the same set of terms.
1585
1586         if (!root_node->entries.empty()) {
1587                 xs << XMLStream::ESCAPE_NONE << " &#8212; ";
1588                 unsigned entry_number = 1;
1589
1590                 for (unsigned i = 0; i < root_node->entries.size(); ++i) {
1591                         const IndexEntry &entry = root_node->entries[i];
1592
1593                         std::string const link_attr = "href='#" + entry.inset()->paragraphs()[0].magicLabel() + "'";
1594                         xs << xml::StartTag("a", link_attr);
1595                         xs << from_ascii(std::to_string(entry_number));
1596                         xs << xml::EndTag("a");
1597
1598                         if (i < root_node->entries.size() - 1) {
1599                                 xs << ", ";
1600                         }
1601                         entry_number += 1;
1602                 }
1603         }
1604
1605         if (!root_node->entries.empty() && !root_node->children.empty()) {
1606                 xs << xml::CR();
1607         }
1608
1609         if (!root_node->children.empty()) {
1610                 xs << xml::StartTag("ul", "class='" + generateCssClassAtDepth(depth) + "'");
1611                 xs << xml::CR();
1612
1613                 for (const IndexNode* child : root_node->children) {
1614                         outputIndexPage(xs, child, depth + 1);
1615                 }
1616
1617                 xs << xml::EndTag("ul");
1618                 xs << xml::CR();
1619         }
1620
1621         xs << xml::EndTag("li");
1622         xs << xml::CR();
1623 }
1624
1625 // Only useful for debugging.
1626 void printTree(const IndexNode* root_node, unsigned depth = 0)
1627 {
1628         static const std::string pattern = "    ";
1629         std::string prefix;
1630         for (unsigned i = 0; i < depth; ++i) {
1631                 prefix += pattern;
1632         }
1633         const std::string prefix_long = prefix + pattern + pattern;
1634
1635         docstring term_at_level;
1636         if (depth == 0) {
1637                 // The root has no term.
1638                 std::cout << "<ROOT>" << std::endl;
1639         } else {
1640                 LASSERT(depth - 1 <= 10, return); // Check for overflows.
1641                 term_at_level = termAtLevel(root_node, depth - 1);
1642                 std::cout << prefix << to_utf8(term_at_level) << " (x " << std::to_string(root_node->entries.size()) << ")"
1643                           << std::endl;
1644         }
1645
1646         for (const IndexEntry& entry : root_node->entries) {
1647                 if (entry.terms().size() != depth) {
1648                         std::cout << prefix_long << "ERROR: an entry doesn't have the same number of terms" << std::endl;
1649                 }
1650                 if (depth > 0 && entry.terms()[depth - 1] != term_at_level) {
1651                         std::cout << prefix_long << "ERROR: an entry doesn't have the right term at depth " << std::to_string(depth)
1652                                 << std::endl;
1653                 }
1654         }
1655
1656         for (const IndexNode* node : root_node->children) {
1657                 printTree(node, depth + 1);
1658         }
1659 }
1660 }
1661
1662
1663 docstring InsetPrintIndex::xhtml(XMLStream &, OutputParams const & op) const
1664 {
1665         BufferParams const & bp = buffer().masterBuffer()->params();
1666
1667         // we do not presently support multiple indices, so we refuse to print
1668         // anything but the main index, so as not to generate multiple indices.
1669         // NOTE Multiple index support would require some work. The reason
1670         // is that the TOC does not know about multiple indices. Either it would
1671         // need to be told about them (not a bad idea), or else the index entries
1672         // would need to be collected differently, say, during validation.
1673         if (bp.use_indices && getParam("type") != from_ascii("idx"))
1674                 return docstring();
1675
1676         shared_ptr<Toc const> toc = buffer().tocBackend().toc("index");
1677         if (toc->empty())
1678                 return docstring();
1679
1680         // Collect the index entries in a form we can use them.
1681         vector<IndexEntry> entries;
1682         for (const TocItem& item : *toc) {
1683                 if (item.isOutput())
1684                         entries.emplace_back(IndexEntry{static_cast<const InsetIndex*>(&(item.dit().inset())), &op});
1685         }
1686
1687         // If all the index entries are in notes or not displayed, get out sooner.
1688         if (entries.empty())
1689                 return docstring();
1690
1691         const IndexNode* index_root = buildIndexTree(entries);
1692 #if 0
1693         printTree(index_root);
1694 #endif
1695
1696         // Start generating the XHTML index.
1697         Layout const & lay = bp.documentClass().htmlTOCLayout();
1698         string const & tocclass = lay.defaultCSSClass();
1699         string const tocattr = "class='index " + tocclass + "'";
1700
1701         // we'll use our own stream, because we are going to defer everything.
1702         // that's how we deal with the fact that we're probably inside a standard
1703         // paragraph, and we don't want to be.
1704         odocstringstream ods;
1705         XMLStream xs(ods);
1706
1707         xs << xml::StartTag("div", tocattr);
1708         xs << xml::CR();
1709         xs << xml::StartTag(lay.htmltag(), lay.htmlattr());
1710         xs << translateIfPossible(from_ascii("Index"), op.local_font->language()->lang());
1711         xs << xml::EndTag(lay.htmltag());
1712         xs << xml::CR();
1713         xs << xml::StartTag("ul", "class='main'");
1714         xs << xml::CR();
1715
1716         LASSERT(index_root->entries.empty(), return docstring()); // No index entry should have zero terms.
1717         for (const IndexNode* node : index_root->children) {
1718                 outputIndexPage(xs, node);
1719         }
1720
1721         xs << xml::EndTag("ul");
1722         xs << xml::CR();
1723         xs << xml::EndTag("div");
1724
1725         return ods.str();
1726 }
1727
1728 } // namespace lyx