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 "LyX.h"
  33 #include "output_latex.h"
  34 #include "output_xhtml.h"
  35 #include "xml.h"
  36 #include "texstream.h"
  37 #include "TextClass.h"
  38 #include "TocBackend.h"
  39
  40 #include "support/debug.h"
  41 #include "support/docstream.h"
  42 #include "support/FileName.h"
  43 #include "support/gettext.h"
  44 #include "support/lstrings.h"
  45 #include "support/Translator.h"
  46
  47 #include "frontends/alert.h"
  48
  49 #include <algorithm>
  50 #include <set>
  51 #include <iostream>
  52
  53 #include <QThreadStorage>
  54
  55 using namespace std;
  56 using namespace lyx::support;
  57
  58 // Uncomment to enable InsetIndex-specific debugging mode: the tree for the index will be printed to std::cout.
  59 // #define LYX_INSET_INDEX_DEBUG
  60
  61 namespace lyx {
  62
  63 namespace {
  64
  65 typedef Translator<string, InsetIndexParams::PageRange> PageRangeTranslator;
  66 typedef Translator<docstring, InsetIndexParams::PageRange> PageRangeTranslatorLoc;
  67
  68 PageRangeTranslator const init_insetindexpagerangetranslator()
  69 {
  70         PageRangeTranslator translator("none", InsetIndexParams::None);
  71         translator.addPair("start", InsetIndexParams::Start);
  72         translator.addPair("end", InsetIndexParams::End);
  73         return translator;
  74 }
  75
  76 PageRangeTranslator const init_insetindexpagerangetranslator_latex()
  77 {
  78         PageRangeTranslator translator("", InsetIndexParams::None);
  79         translator.addPair("(", InsetIndexParams::Start);
  80         translator.addPair(")", InsetIndexParams::End);
  81         return translator;
  82 }
  83
  84
  85 PageRangeTranslatorLoc const init_insetindexpagerangetranslator_loc()
  86 {
  87         PageRangeTranslatorLoc translator(docstring(), InsetIndexParams::None);
  88         translator.addPair(_("Starts page range"), InsetIndexParams::Start);
  89         translator.addPair(_("Ends page range"), InsetIndexParams::End);
  90         return translator;
  91 }
  92
  93
  94 PageRangeTranslator const & insetindexpagerangetranslator()
  95 {
  96         static PageRangeTranslator const prtranslator =
  97                         init_insetindexpagerangetranslator();
  98         return prtranslator;
  99 }
 100
 101
 102 PageRangeTranslatorLoc const & insetindexpagerangetranslator_loc()
 103 {
 104         static PageRangeTranslatorLoc const translator =
 105                         init_insetindexpagerangetranslator_loc();
 106         return translator;
 107 }
 108
 109
 110 PageRangeTranslator const & insetindexpagerangetranslator_latex()
 111 {
 112         static PageRangeTranslator const lttranslator =
 113                         init_insetindexpagerangetranslator_latex();
 114         return lttranslator;
 115 }
 116
 117 } // namespace anon
 118
 119 /////////////////////////////////////////////////////////////////////
 120 //
 121 // InsetIndex
 122 //
 123 ///////////////////////////////////////////////////////////////////////
 124
 125
 126 InsetIndex::InsetIndex(Buffer * buf, InsetIndexParams const & params)
 127         : InsetCollapsible(buf), params_(params)
 128 {}
 129
 130
 131 void InsetIndex::latex(otexstream & ios, OutputParams const & runparams_in) const
 132 {
 133         OutputParams runparams(runparams_in);
 134         runparams.inIndexEntry = true;
 135
 136         otexstringstream os;
 137
 138         if (buffer().masterBuffer()->params().use_indices && !params_.index.empty()
 139                 && params_.index != "idx") {
 140                 os << "\\sindex[";
 141                 os << escape(params_.index);
 142                 os << "]{";
 143         } else {
 144                 os << "\\index";
 145                 os << '{';
 146         }
 147
 148         // Get the LaTeX output from InsetText. We need to deconstruct this later
 149         // in order to check if we need to generate a sorting key
 150         odocstringstream ourlatex;
 151         otexstream ots(ourlatex);
 152         InsetText::latex(ots, runparams);
 153         if (runparams.find_effective()) {
 154                 // No need for special handling, if we are only searching for some patterns
 155                 os << ourlatex.str() << "}";
 156                 return;
 157         }
 158
 159         if (hasSortKey()) {
 160                 getSortkey(os, runparams);
 161                 os << "@";
 162                 os << ourlatex.str();
 163                 getSubentries(os, runparams, ourlatex.str());
 164                 if (hasSeeRef()) {
 165                         os << "|";
 166                         os << insetindexpagerangetranslator_latex().find(params_.range);
 167                         getSeeRefs(os, runparams);
 168                 } else if (!params_.pagefmt.empty() && params_.pagefmt != "default") {
 169                         os << "|";
 170                         os << insetindexpagerangetranslator_latex().find(params_.range);
 171                         os << from_utf8(params_.pagefmt);
 172                 } else if (params_.range != InsetIndexParams::PageRange::None) {
 173                         os << "|";
 174                         os << insetindexpagerangetranslator_latex().find(params_.range);
 175                 }
 176         } else {
 177                 // We check whether we need a sort key.
 178                 // If so, we use the plaintext version
 179                 odocstringstream ourplain;
 180                 InsetText::plaintext(ourplain, runparams);
 181
 182                 // These are the LaTeX and plaintext representations
 183                 docstring latexstr = ourlatex.str();
 184                 docstring plainstr = ourplain.str();
 185
 186                 // This will get what follows | if anything does,
 187                 // the command (e.g., see, textbf) for pagination
 188                 // formatting
 189                 docstring cmd;
 190
 191                 if (hasSeeRef()) {
 192                         odocstringstream seeref;
 193                         otexstream otsee(seeref);
 194                         getSeeRefs(otsee, runparams);
 195                         cmd = seeref.str();
 196                 } else if (!params_.pagefmt.empty() && params_.pagefmt != "default") {
 197                         cmd = from_utf8(params_.pagefmt);
 198                 } else {
 199                         // Check for the | separator to strip the cmd.
 200                         // This goes wrong on an escaped "|", but as the escape
 201                         // character can be changed in style files, we cannot
 202                         // prevent that.
 203                         size_t pos = latexstr.find(from_ascii("|"));
 204                         if (pos != docstring::npos) {
 205                                 // Put the bit after "|" into cmd...
 206                                 cmd = latexstr.substr(pos + 1);
 207                                 // ...and erase that stuff from latexstr
 208                                 latexstr = latexstr.erase(pos);
 209                                 // ...as well as from plainstr
 210                                 size_t ppos = plainstr.find(from_ascii("|"));
 211                                 if (ppos < plainstr.size())
 212                                         plainstr.erase(ppos);
 213                                 else
 214                                         LYXERR0("The `|' separator was not found in the plaintext version!");
 215                         }
 216                 }
 217
 218                 odocstringstream subentries;
 219                 otexstream otsub(subentries);
 220                 getSubentries(otsub, runparams, ourlatex.str());
 221                 if (subentries.str().empty()) {
 222                         // Separate the entries and subentries, i.e., split on "!".
 223                         // This goes wrong on an escaped "!", but as the escape
 224                         // character can be changed in style files, we cannot
 225                         // prevent that.
 226                         std::vector<docstring> const levels =
 227                                         getVectorFromString(latexstr, from_ascii("!"), true);
 228                         std::vector<docstring> const levels_plain =
 229                                         getVectorFromString(plainstr, from_ascii("!"), true);
 230
 231                         vector<docstring>::const_iterator it = levels.begin();
 232                         vector<docstring>::const_iterator end = levels.end();
 233                         vector<docstring>::const_iterator it2 = levels_plain.begin();
 234                         bool first = true;
 235                         for (; it != end; ++it) {
 236                                 if ((*it).empty()) {
 237                                         emptySubentriesWarning(ourlatex.str());
 238                                         if (it2 < levels_plain.end())
 239                                                 ++it2;
 240                                         continue;
 241                                 }
 242                                 // The separator needs to be put back when
 243                                 // writing the levels, except for the first level
 244                                 if (!first)
 245                                         os << '!';
 246                                 else
 247                                         first = false;
 248
 249                                 // Now here comes the reason for this whole procedure:
 250                                 // We try to correctly sort macros and formatted strings.
 251                                 // If we find a command, prepend a plain text
 252                                 // version of the content to get sorting right,
 253                                 // e.g. \index{LyX@\LyX}, \index{text@\textbf{text}}.
 254                                 // We do this on all levels.
 255                                 // We don't do it if the level already contains a '@', though.
 256                                 // Plaintext might return nothing (e.g. for ERTs).
 257                                 // In that case, we use LaTeX.
 258                                 docstring const spart = (levels_plain.empty() || (*it2).empty()) ? *it : *it2;
 259                                 processLatexSorting(os, runparams, *it, spart);
 260                                 if (it2 < levels_plain.end())
 261                                         ++it2;
 262                         }
 263                 } else {
 264                         processLatexSorting(os, runparams, latexstr, plainstr);
 265                         os << subentries.str();
 266                 }
 267
 268                 // At last, re-insert the command, separated by "|"
 269                 if (!cmd.empty()) {
 270                         os << "|"
 271                            << insetindexpagerangetranslator_latex().find(params_.range)
 272                            << cmd;
 273                 } else if (params_.range != InsetIndexParams::PageRange::None) {
 274                         os << "|";
 275                         os << insetindexpagerangetranslator_latex().find(params_.range);
 276                 }
 277         }
 278         os << '}';
 279
 280         // In macros with moving arguments, such as \section,
 281         // we store the index and output it after the macro (#2154)
 282         if (runparams_in.postpone_fragile_stuff)
 283                 runparams_in.post_macro += os.str();
 284         else
 285                 ios << os.release();
 286 }
 287
 288
 289 void InsetIndex::processLatexSorting(otexstream & os, OutputParams const & runparams,
 290                                 docstring const latex, docstring const spart) const
 291 {
 292         if (contains(latex, '\\') && !contains(latex, '@')) {
 293                 // Now we need to validate that all characters in
 294                 // the sorting part are representable in the current
 295                 // encoding. If not try the LaTeX macro which might
 296                 // or might not be a good choice, and issue a warning.
 297                 pair<docstring, docstring> spart_latexed =
 298                                 runparams.encoding->latexString(spart, runparams.dryrun);
 299                 if (!spart_latexed.second.empty())
 300                         LYXERR0("Uncodable character in index entry. Sorting might be wrong!");
 301                 if (spart != spart_latexed.first && !runparams.dryrun) {
 302                         TeXErrors terr;
 303                         ErrorList & errorList = buffer().errorList("Export");
 304                         docstring const s = bformat(_("LyX's automatic index sorting algorithm faced "
 305                                                       "problems with the entry '%1$s'.\n"
 306                                                       "Please specify the sorting of this entry manually, as "
 307                                                       "explained in the User Guide."), spart);
 308                         Paragraph const & par = buffer().paragraphs().front();
 309                         errorList.push_back(ErrorItem(_("Index sorting failed"), s,
 310                                                       {par.id(), 0}, {par.id(), -1}));
 311                         buffer().bufferErrors(terr, errorList);
 312                 }
 313                 // Remove remaining \'s from the sort key
 314                 docstring ppart = subst(spart_latexed.first, from_ascii("\\"), docstring());
 315                 // Plain quotes need to be escaped, however (#10649), as this
 316                 // is the default escape character
 317                 ppart = subst(ppart, from_ascii("\""), from_ascii("\\\""));
 318
 319                 // Now insert the sortkey, separated by '@'.
 320                 os << ppart;
 321                 os << '@';
 322         }
 323         // Insert the actual level text
 324         os << latex;
 325 }
 326
 327
 328 void InsetIndex::docbook(XMLStream & xs, OutputParams const & runparams) const
 329 {
 330         // Two ways of processing this inset are implemented:
 331         // - the legacy one, based on parsing the raw LaTeX (before LyX 2.4) -- unlikely to be deprecated
 332         // - the modern one, based on precise insets for indexing features
 333         // Like the LaTeX implementation, consider the user chooses either of those options.
 334
 335         // Get the content of the inset as LaTeX, as some things may be encoded as ERT (like {}).
 336         // TODO: if there is an ERT within the index term, its conversion should be tried, in case it becomes useful;
 337         //  otherwise, ERTs should become comments. For now, they are just copied as-is, which is barely satisfactory.
 338         odocstringstream odss;
 339         otexstream ots(odss);
 340         InsetText::latex(ots, runparams);
 341         docstring latexString = trim(odss.str());
 342
 343         // Handle several indices (indicated in the inset instead of the raw latexString).
 344         docstring indexType = from_utf8("");
 345         if (buffer().masterBuffer()->params().use_indices) {
 346                 indexType += " type=\"" + params_.index + "\"";
 347         }
 348
 349         // Split the string into its main constituents: terms, and command (see, see also, range).
 350         size_t positionVerticalBar = latexString.find(from_ascii("|")); // What comes before | is (sub)(sub)entries.
 351         docstring indexTerms = latexString.substr(0, positionVerticalBar);
 352         docstring command;
 353         if (positionVerticalBar != lyx::docstring::npos) {
 354                 command = latexString.substr(positionVerticalBar + 1);
 355         }
 356
 357         // Handle sorting issues, with @.
 358         docstring sortAs;
 359         if (hasSortKey()) {
 360                 sortAs = getSortkeyAsText(runparams);
 361                 // indexTerms may contain a sort key if the user has both the inset and the manual key.
 362         } else {
 363                 vector<docstring> sortingElements = getVectorFromString(indexTerms, from_ascii("@"), false);
 364                 if (sortingElements.size() == 2) {
 365                         sortAs = sortingElements[0];
 366                         indexTerms = sortingElements[1];
 367                 }
 368         }
 369
 370         // Handle primary, secondary, and tertiary terms (entries, subentries, and subsubentries, for LaTeX).
 371         vector<docstring> terms;
 372         if (const vector<docstring> potential_terms = getSubentriesAsText(runparams); !potential_terms.empty()) {
 373                 terms = potential_terms;
 374                 // The main term is not present in the vector, as it's not a subentry. The main index term is inserted raw in
 375                 // the index inset. Considering that the user either uses the new or the legacy mechanism, the main term is the
 376                 // full string within this inset (i.e. without the subinsets).
 377                 terms.insert(terms.begin(), latexString);
 378         } else {
 379                 terms = getVectorFromString(indexTerms, from_ascii("!"), false);
 380         }
 381
 382         // Handle ranges. Happily, in the raw LaTeX mode, (| and |) can only be at the end of the string!
 383         const bool hasInsetRange = params_.range != InsetIndexParams::PageRange::None;
 384         const bool hasStartRange = params_.range == InsetIndexParams::PageRange::Start ||
 385                         latexString.find(from_ascii("|(")) != lyx::docstring::npos;
 386         const bool hasEndRange = params_.range == InsetIndexParams::PageRange::End ||
 387                         latexString.find(from_ascii("|)")) != lyx::docstring::npos;
 388
 389         if (hasInsetRange) {
 390                 // Remove the ranges from the command if they do not appear at the beginning.
 391                 size_t index = 0;
 392                 while ((index = command.find(from_utf8("|("), index)) != std::string::npos)
 393                         command.erase(index, 1);
 394                 index = 0;
 395                 while ((index = command.find(from_utf8("|)"), index)) != std::string::npos)
 396                         command.erase(index, 1);
 397
 398                 // Remove the ranges when they are the only vertical bar in the complete string.
 399                 if (command[0] == '(' || command[0] == ')')
 400                         command.erase(0, 1);
 401         }
 402
 403         // Handle see and seealso. As "see" is a prefix of "seealso", the order of the comparisons is important.
 404         // Both commands are mutually exclusive!
 405         docstring see = getSeeAsText(runparams);
 406         vector<docstring> seeAlsoes = getSeeAlsoesAsText(runparams);
 407
 408         if (see.empty() && seeAlsoes.empty() && command.substr(0, 3) == "see") {
 409                 // Unescape brackets.
 410                 size_t index = 0;
 411                 while ((index = command.find(from_utf8("\\{"), index)) != std::string::npos)
 412                         command.erase(index, 1);
 413                 index = 0;
 414                 while ((index = command.find(from_utf8("\\}"), index)) != std::string::npos)
 415                         command.erase(index, 1);
 416
 417                 // Retrieve the part between brackets, and remove the complete seealso.
 418                 size_t positionOpeningBracket = command.find(from_ascii("{"));
 419                 size_t positionClosingBracket = command.find(from_ascii("}"));
 420                 docstring list = command.substr(positionOpeningBracket + 1, positionClosingBracket - positionOpeningBracket - 1);
 421
 422                 // Parse the list of referenced entries (or a single one for see).
 423                 if (command.substr(0, 7) == "seealso") {
 424                         seeAlsoes = getVectorFromString(list, from_ascii(","), false);
 425                 } else {
 426                         see = list;
 427
 428                         if (see.find(from_ascii(",")) != std::string::npos) {
 429                                 docstring error = from_utf8("Several index terms found as \"see\"! Only one is acceptable. "
 430                                                                                         "Complete entry: \"") + latexString + from_utf8("\"");
 431                                 LYXERR0(error);
 432                                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + error + from_utf8(" -->\n"));
 433                         }
 434                 }
 435
 436                 // Remove the complete see/seealso from the commands, in case there is something else to parse.
 437                 command = command.substr(positionClosingBracket + 1);
 438         }
 439
 440         // Some parts of the strings are not parsed, as they do not have anything matching in DocBook: things like
 441         // formatting the entry or the page number, other strings for sorting. https://wiki.lyx.org/Tips/Indexing
 442         // If there are such things in the index entry, then this code may miserably fail. For example, for "Peter|(textbf",
 443         // no range will be detected.
 444         // TODO: Could handle formatting as significance="preferred"?
 445         if (!command.empty()) {
 446                 docstring error = from_utf8("Unsupported feature: an index entry contains a | with an unsupported command, ")
 447                                           + command + from_utf8(". ") + from_utf8("Complete entry: \"") + latexString + from_utf8("\"");
 448                 LYXERR0(error);
 449                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + error + from_utf8(" -->\n"));
 450         }
 451
 452         // Write all of this down.
 453         if (terms.empty() && !hasEndRange) {
 454                 docstring error = from_utf8("No index term found! Complete entry: \"") + latexString + from_utf8("\"");
 455                 LYXERR0(error);
 456                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + error + from_utf8(" -->\n"));
 457         } else {
 458                 // Generate the attributes for ranges. It is based on the terms that are indexed, but the ID must be unique
 459                 // to this indexing area (xml::cleanID does not guarantee this: for each call with the same arguments,
 460                 // the same legal ID is produced; here, as the input would be the same, the output must be, by design).
 461                 // Hence the thread-local storage, as the numbers must strictly be unique, and thus cannot be shared across
 462                 // a paragraph (making the solution used for HTML worthless). This solution is very similar to the one used in
 463                 // xml::cleanID.
 464                 // indexType can only be used for singular and startofrange types!
 465                 docstring attrs;
 466                 if (!hasStartRange && !hasEndRange) {
 467                         attrs = indexType;
 468                 } else {
 469                         // Append an ID if uniqueness is not guaranteed across the document.
 470                         static QThreadStorage<set<docstring>> tKnownTermLists;
 471                         static QThreadStorage<int> tID;
 472
 473                         set<docstring> &knownTermLists = tKnownTermLists.localData();
 474                         int &ID = tID.localData();
 475
 476                         if (!tID.hasLocalData()) {
 477                                 tID.localData() = 0;
 478                         }
 479
 480                         // Modify the index terms to add the unique ID if needed.
 481                         docstring newIndexTerms = indexTerms;
 482                         if (knownTermLists.find(indexTerms) != knownTermLists.end()) {
 483                                 newIndexTerms += from_ascii(string("-") + to_string(ID));
 484
 485                                 // Only increment for the end of range, so that the same number is used for the start of range.
 486                                 if (hasEndRange) {
 487                                         ID++;
 488                                 }
 489                         }
 490
 491                         // Term list not yet known: add it to the set AFTER the end of range. After
 492                         if (knownTermLists.find(indexTerms) == knownTermLists.end() && hasEndRange) {
 493                                 knownTermLists.insert(indexTerms);
 494                         }
 495
 496                         // Generate the attributes.
 497                         docstring id = xml::cleanID(newIndexTerms);
 498                         if (hasStartRange) {
 499                                 attrs = indexType + " class=\"startofrange\" xml:id=\"" + id + "\"";
 500                         } else {
 501                                 attrs = " class=\"endofrange\" startref=\"" + id + "\"";
 502                         }
 503                 }
 504
 505                 // Handle the index terms (including the specific index for this entry).
 506                 if (hasEndRange) {
 507                         xs << xml::CompTag("indexterm", attrs);
 508                 } else {
 509                         xs << xml::StartTag("indexterm", attrs);
 510                         if (!terms.empty()) { // hasEndRange has no content.
 511                                 docstring attr;
 512                                 if (!sortAs.empty()) {
 513                                         attr = from_utf8("sortas='") + sortAs + from_utf8("'");
 514                                 }
 515
 516                                 xs << xml::StartTag("primary", attr);
 517                                 xs << terms[0];
 518                                 xs << xml::EndTag("primary");
 519                         }
 520                         if (terms.size() > 1) {
 521                                 xs << xml::StartTag("secondary");
 522                                 xs << terms[1];
 523                                 xs << xml::EndTag("secondary");
 524                         }
 525                         if (terms.size() > 2) {
 526                                 xs << xml::StartTag("tertiary");
 527                                 xs << terms[2];
 528                                 xs << xml::EndTag("tertiary");
 529                         }
 530
 531                         // Handle see and see also.
 532                         if (!see.empty()) {
 533                                 xs << xml::StartTag("see");
 534                                 xs << see;
 535                                 xs << xml::EndTag("see");
 536                         }
 537
 538                         if (!seeAlsoes.empty()) {
 539                                 for (auto &entry : seeAlsoes) {
 540                                         xs << xml::StartTag("seealso");
 541                                         xs << entry;
 542                                         xs << xml::EndTag("seealso");
 543                                 }
 544                         }
 545
 546                         // Close the entry.
 547                         xs << xml::EndTag("indexterm");
 548                 }
 549         }
 550 }
 551
 552
 553 docstring InsetIndex::xhtml(XMLStream & xs, OutputParams const &) const
 554 {
 555         // we just print an anchor, taking the paragraph ID from
 556         // our own interior paragraph, which doesn't get printed
 557         std::string const magic = paragraphs().front().magicLabel();
 558         std::string const attr = "id='" + magic + "'";
 559         xs << xml::CompTag("a", attr);
 560         return docstring();
 561 }
 562
 563
 564 bool InsetIndex::showInsetDialog(BufferView * bv) const
 565 {
 566         bv->showDialog("index", params2string(params_),
 567                         const_cast<InsetIndex *>(this));
 568         return true;
 569 }
 570
 571
 572 void InsetIndex::doDispatch(Cursor & cur, FuncRequest & cmd)
 573 {
 574         switch (cmd.action()) {
 575
 576         case LFUN_INSET_MODIFY: {
 577                 if (cmd.getArg(0) == "changetype") {
 578                         cur.recordUndoInset(this);
 579                         params_.index = from_utf8(cmd.getArg(1));
 580                         break;
 581                 }
 582                 InsetIndexParams params;
 583                 InsetIndex::string2params(to_utf8(cmd.argument()), params);
 584                 cur.recordUndoInset(this);
 585                 params_.index = params.index;
 586                 params_.range = params.range;
 587                 params_.pagefmt = params.pagefmt;
 588                 // what we really want here is a TOC update, but that means
 589                 // a full buffer update
 590                 cur.forceBufferUpdate();
 591                 break;
 592         }
 593
 594         case LFUN_INSET_DIALOG_UPDATE:
 595                 cur.bv().updateDialog("index", params2string(params_));
 596                 break;
 597
 598         case LFUN_PARAGRAPH_BREAK: {
 599                 // Since this inset in single-par anyway, let's use
 600                 // return to enter subentries
 601                 FuncRequest fr(LFUN_INDEXMACRO_INSERT, "subentry");
 602                 lyx::dispatch(fr);
 603                 break;
 604         }
 605
 606         default:
 607                 InsetCollapsible::doDispatch(cur, cmd);
 608                 break;
 609         }
 610 }
 611
 612
 613 bool InsetIndex::getStatus(Cursor & cur, FuncRequest const & cmd,
 614                 FuncStatus & flag) const
 615 {
 616         switch (cmd.action()) {
 617
 618         case LFUN_INSET_MODIFY:
 619                 if (cmd.getArg(0) == "changetype") {
 620                         docstring const newtype = from_utf8(cmd.getArg(1));
 621                         Buffer const & realbuffer = *buffer().masterBuffer();
 622                         IndicesList const & indiceslist = realbuffer.params().indiceslist();
 623                         Index const * index = indiceslist.findShortcut(newtype);
 624                         flag.setEnabled(index != 0);
 625                         flag.setOnOff(
 626                                 from_utf8(cmd.getArg(1)) == params_.index);
 627                         return true;
 628                 }
 629                 return InsetCollapsible::getStatus(cur, cmd, flag);
 630
 631         case LFUN_INSET_DIALOG_UPDATE: {
 632                 Buffer const & realbuffer = *buffer().masterBuffer();
 633                 flag.setEnabled(realbuffer.params().use_indices);
 634                 return true;
 635         }
 636
 637         case LFUN_PARAGRAPH_BREAK:
 638                 return macrosPossible("subentry");
 639
 640         case LFUN_INDEXMACRO_INSERT:
 641                 return macrosPossible(cmd.getArg(0));
 642
 643         default:
 644                 return InsetCollapsible::getStatus(cur, cmd, flag);
 645         }
 646 }
 647
 648
 649 void InsetIndex::getSortkey(otexstream & os, OutputParams const & runparams) const
 650 {
 651         Paragraph const & par = paragraphs().front();
 652         InsetList::const_iterator it = par.insetList().begin();
 653         for (; it != par.insetList().end(); ++it) {
 654                 Inset & inset = *it->inset;
 655                 if (inset.lyxCode() == INDEXMACRO_SORTKEY_CODE) {
 656                         InsetIndexMacro const & iim =
 657                                 static_cast<InsetIndexMacro const &>(inset);
 658                         iim.getLatex(os, runparams);
 659                         return;
 660                 }
 661         }
 662 }
 663
 664
 665 docstring InsetIndex::getSortkeyAsText(OutputParams const & runparams) const
 666 {
 667         Paragraph const & par = paragraphs().front();
 668         InsetList::const_iterator it = par.insetList().begin();
 669         for (; it != par.insetList().end(); ++it) {
 670                 Inset & inset = *it->inset;
 671                 if (inset.lyxCode() == INDEXMACRO_SORTKEY_CODE) {
 672                         otexstringstream os;
 673                         InsetIndexMacro const & iim =
 674                                 static_cast<InsetIndexMacro const &>(inset);
 675                         iim.getLatex(os, runparams);
 676                         return os.str();
 677                 }
 678         }
 679         return from_ascii("");
 680 }
 681
 682
 683 void InsetIndex::emptySubentriesWarning(docstring const & mainentry) const
 684 {
 685         // Empty subentries crash makeindex. So warn and ignore this.
 686         TeXErrors terr;
 687         ErrorList & errorList = buffer().errorList("Export");
 688         docstring const s = bformat(_("There is an empty index subentry in the entry '%1$s'.\n"
 689                                       "It will be ignored in the output."), mainentry);
 690         Paragraph const & par = buffer().paragraphs().front();
 691         errorList.push_back(ErrorItem(_("Empty index subentry!"), s,
 692                                       {par.id(), 0}, {par.id(), -1}));
 693         buffer().bufferErrors(terr, errorList);
 694 }
 695
 696
 697 void InsetIndex::getSubentries(otexstream & os, OutputParams const & runparams,
 698                                docstring const & mainentry) const
 699 {
 700         Paragraph const & par = paragraphs().front();
 701         InsetList::const_iterator it = par.insetList().begin();
 702         int i = 0;
 703         for (; it != par.insetList().end(); ++it) {
 704                 Inset & inset = *it->inset;
 705                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 706                         InsetIndexMacro const & iim =
 707                                 static_cast<InsetIndexMacro const &>(inset);
 708                         if (iim.params().type == InsetIndexMacroParams::Subentry) {
 709                                 if (iim.hasNoContent()) {
 710                                         emptySubentriesWarning(mainentry);
 711                                         continue;
 712                                 }
 713                                 ++i;
 714                                 if (i > 2)
 715                                         return;
 716                                 os << "!";
 717                                 iim.getLatex(os, runparams);
 718                         }
 719                 }
 720         }
 721 }
 722
 723
 724 std::vector<docstring> InsetIndex::getSubentriesAsText(OutputParams const & runparams,
 725                                                        bool const asLabel) const
 726 {
 727         std::vector<docstring> subentries;
 728
 729         Paragraph const & par = paragraphs().front();
 730         InsetList::const_iterator it = par.insetList().begin();
 731         int i = 0;
 732         for (; it != par.insetList().end(); ++it) {
 733                 Inset & inset = *it->inset;
 734                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 735                         InsetIndexMacro const & iim =
 736                                 static_cast<InsetIndexMacro const &>(inset);
 737                         if (iim.params().type == InsetIndexMacroParams::Subentry) {
 738                                 ++i;
 739                                 if (i > 2)
 740                                         break;
 741                                 if (asLabel) {
 742                                         docstring const l;
 743                                         docstring const sl = iim.getNewLabel(l);
 744                                         subentries.emplace_back(sl);
 745                                 } else {
 746                                         otexstringstream os;
 747                                         iim.getLatex(os, runparams);
 748                                         subentries.emplace_back(os.str());
 749                                 }
 750                         }
 751                 }
 752         }
 753
 754         return subentries;
 755 }
 756
 757
 758 docstring InsetIndex::getMainSubentryAsText(OutputParams const & runparams) const
 759 {
 760         otexstringstream os;
 761         InsetText::latex(os, runparams);
 762         return os.str();
 763 }
 764
 765
 766 void InsetIndex::getSeeRefs(otexstream & os, OutputParams const & runparams) const
 767 {
 768         Paragraph const & par = paragraphs().front();
 769         InsetList::const_iterator it = par.insetList().begin();
 770         for (; it != par.insetList().end(); ++it) {
 771                 Inset & inset = *it->inset;
 772                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 773                         InsetIndexMacro const & iim =
 774                                 static_cast<InsetIndexMacro const &>(inset);
 775                         if (iim.params().type == InsetIndexMacroParams::See
 776                             || iim.params().type == InsetIndexMacroParams::Seealso) {
 777                                 iim.getLatex(os, runparams);
 778                                 return;
 779                         }
 780                 }
 781         }
 782 }
 783
 784
 785 docstring InsetIndex::getSeeAsText(OutputParams const & runparams,
 786                                    bool const asLabel) const
 787 {
 788         Paragraph const & par = paragraphs().front();
 789         InsetList::const_iterator it = par.insetList().begin();
 790         for (; it != par.insetList().end(); ++it) {
 791                 Inset & inset = *it->inset;
 792                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 793                         InsetIndexMacro const & iim =
 794                                 static_cast<InsetIndexMacro const &>(inset);
 795                         if (iim.params().type == InsetIndexMacroParams::See) {
 796                                 if (asLabel) {
 797                                         docstring const l;
 798                                         return iim.getNewLabel(l);
 799                                 } else {
 800                                         otexstringstream os;
 801                                         iim.getLatex(os, runparams);
 802                                         return os.str();
 803                                 }
 804                         }
 805                 }
 806         }
 807         return from_ascii("");
 808 }
 809
 810
 811 std::vector<docstring> InsetIndex::getSeeAlsoesAsText(OutputParams const & runparams,
 812                                                       bool const asLabel) const
 813 {
 814         std::vector<docstring> seeAlsoes;
 815
 816         Paragraph const & par = paragraphs().front();
 817         InsetList::const_iterator it = par.insetList().begin();
 818         for (; it != par.insetList().end(); ++it) {
 819                 Inset & inset = *it->inset;
 820                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 821                         InsetIndexMacro const & iim =
 822                                 static_cast<InsetIndexMacro const &>(inset);
 823                         if (iim.params().type == InsetIndexMacroParams::Seealso) {
 824                                 if (asLabel) {
 825                                         docstring const l;
 826                                         seeAlsoes.emplace_back(iim.getNewLabel(l));
 827                                 } else {
 828                                         otexstringstream os;
 829                                         iim.getLatex(os, runparams);
 830                                         seeAlsoes.emplace_back(os.str());
 831                                 }
 832                         }
 833                 }
 834         }
 835
 836         return seeAlsoes;
 837 }
 838
 839
 840 namespace {
 841
 842 bool hasInsetWithCode(const InsetIndex * const inset_index, const InsetCode code,
 843                                           const std::set<InsetIndexMacroParams::Type> types = {})
 844 {
 845         Paragraph const & par = inset_index->paragraphs().front();
 846         InsetList::const_iterator it = par.insetList().begin();
 847         for (; it != par.insetList().end(); ++it) {
 848                 Inset & inset = *it->inset;
 849                 if (inset.lyxCode() == code) {
 850                         if (types.empty())
 851                                 return true;
 852
 853                         LASSERT(code == INDEXMACRO_CODE, return false);
 854                         InsetIndexMacro const & iim =
 855                                         static_cast<InsetIndexMacro const &>(inset);
 856                         if (types.find(iim.params().type) != types.end())
 857                                 return true;
 858                 }
 859         }
 860         return false;
 861 }
 862
 863 } // namespace
 864
 865
 866 bool InsetIndex::hasSubentries() const
 867 {
 868         return hasInsetWithCode(this, INDEXMACRO_CODE, {InsetIndexMacroParams::Subentry});
 869 }
 870
 871
 872 bool InsetIndex::hasSeeRef() const
 873 {
 874         return hasInsetWithCode(this, INDEXMACRO_CODE, {InsetIndexMacroParams::See, InsetIndexMacroParams::Seealso});
 875 }
 876
 877
 878 bool InsetIndex::hasSortKey() const
 879 {
 880         return hasInsetWithCode(this, INDEXMACRO_SORTKEY_CODE);
 881 }
 882
 883
 884 bool InsetIndex::macrosPossible(string const type) const
 885 {
 886         if (type != "see" && type != "seealso"
 887             && type != "sortkey" && type != "subentry")
 888                 return false;
 889
 890         Paragraph const & par = paragraphs().front();
 891         InsetList::const_iterator it = par.insetList().begin();
 892         int subidxs = 0;
 893         for (; it != par.insetList().end(); ++it) {
 894                 Inset & inset = *it->inset;
 895                 if (type == "sortkey" && inset.lyxCode() == INDEXMACRO_SORTKEY_CODE)
 896                         return false;
 897                 if (inset.lyxCode() == INDEXMACRO_CODE) {
 898                         InsetIndexMacro const & iim = static_cast<InsetIndexMacro const &>(inset);
 899                         if ((type == "see" || type == "seealso")
 900                              && (iim.params().type == InsetIndexMacroParams::See
 901                                  || iim.params().type == InsetIndexMacroParams::Seealso))
 902                                 return false;
 903                         if (type == "subentry"
 904                              && iim.params().type == InsetIndexMacroParams::Subentry) {
 905                                 ++subidxs;
 906                                 if (subidxs > 1)
 907                                         return false;
 908                         }
 909                 }
 910         }
 911         return true;
 912 }
 913
 914
 915 ColorCode InsetIndex::labelColor() const
 916 {
 917         if (params_.index.empty() || params_.index == from_ascii("idx"))
 918                 return InsetCollapsible::labelColor();
 919         // FIXME UNICODE
 920         ColorCode c = lcolor.getFromLyXName(to_utf8(params_.index)
 921                                             + "@" + buffer().fileName().absFileName());
 922         if (c == Color_none)
 923                 c = InsetCollapsible::labelColor();
 924         return c;
 925 }
 926
 927
 928 docstring InsetIndex::toolTip(BufferView const &, int, int) const
 929 {
 930         docstring tip = _("Index Entry");
 931         if (buffer().params().use_indices && !params_.index.empty()) {
 932                 Buffer const & realbuffer = *buffer().masterBuffer();
 933                 IndicesList const & indiceslist = realbuffer.params().indiceslist();
 934                 tip += " (";
 935                 Index const * index = indiceslist.findShortcut(params_.index);
 936                 if (!index)
 937                         tip += _("unknown type!");
 938                 else
 939                         tip += index->index();
 940                 tip += ")";
 941         }
 942         tip += ": ";
 943         docstring res = toolTipText(tip);
 944         if (!insetindexpagerangetranslator_loc().find(params_.range).empty())
 945                 res += "\n" + insetindexpagerangetranslator_loc().find(params_.range);
 946         if (!params_.pagefmt.empty() && params_.pagefmt != "default") {
 947                 res += "\n" + _("Pagination format:") + " ";
 948                 if (params_.pagefmt == "textbf")
 949                         res += _("bold");
 950                 else if (params_.pagefmt == "textit")
 951                         res += _("italic");
 952                 else if (params_.pagefmt == "emph")
 953                         res += _("emphasized");
 954                 else
 955                         res += from_utf8(params_.pagefmt);
 956         }
 957         return res;
 958 }
 959
 960
 961 docstring const InsetIndex::buttonLabel(BufferView const & bv) const
 962 {
 963         InsetLayout const & il = getLayout();
 964         docstring label = translateIfPossible(il.labelstring());
 965
 966         if (buffer().params().use_indices && !params_.index.empty()) {
 967                 Buffer const & realbuffer = *buffer().masterBuffer();
 968                 IndicesList const & indiceslist = realbuffer.params().indiceslist();
 969                 label += " (";
 970                 Index const * index = indiceslist.findShortcut(params_.index);
 971                 if (!index)
 972                         label += _("unknown type!");
 973                 else
 974                         label += index->index();
 975                 label += ")";
 976         }
 977
 978         docstring res;
 979         if (!il.contentaslabel() || geometry(bv) != ButtonOnly)
 980                 res = label;
 981         else {
 982                 res = getNewLabel(label);
 983                 OutputParams const rp(0);
 984                 vector<docstring> sublbls = getSubentriesAsText(rp, true);
 985                 for (auto const & sublbl : sublbls) {
 986                         res += " " + docstring(1, char_type(0x2023));// TRIANGULAR BULLET
 987                         res += " " + sublbl;
 988                 }
 989                 docstring see = getSeeAsText(rp, true);
 990                 if (see.empty() && !getSeeAlsoesAsText(rp, true).empty())
 991                         see = getSeeAlsoesAsText(rp, true).front();
 992                 if (!see.empty()) {
 993                         res += " " + docstring(1, char_type(0x261e));// WHITE RIGHT POINTING INDEX
 994                         res += " " + see;
 995                 }
 996         }
 997         if (!insetindexpagerangetranslator_latex().find(params_.range).empty())
 998                 res += " " + from_ascii(insetindexpagerangetranslator_latex().find(params_.range));
 999         return res;
1000 }
1001
1002
1003 void InsetIndex::write(ostream & os) const
1004 {
1005         os << to_utf8(layoutName());
1006         params_.write(os);
1007         InsetCollapsible::write(os);
1008 }
1009
1010
1011 void InsetIndex::read(Lexer & lex)
1012 {
1013         params_.read(lex);
1014         InsetCollapsible::read(lex);
1015 }
1016
1017
1018 string InsetIndex::params2string(InsetIndexParams const & params)
1019 {
1020         ostringstream data;
1021         data << "index";
1022         params.write(data);
1023         return data.str();
1024 }
1025
1026
1027 void InsetIndex::string2params(string const & in, InsetIndexParams & params)
1028 {
1029         params = InsetIndexParams();
1030         if (in.empty())
1031                 return;
1032
1033         istringstream data(in);
1034         Lexer lex;
1035         lex.setStream(data);
1036         lex.setContext("InsetIndex::string2params");
1037         lex >> "index";
1038         params.read(lex);
1039 }
1040
1041
1042 void InsetIndex::addToToc(DocIterator const & cpit, bool output_active,
1043                                                   UpdateType utype, TocBackend & backend) const
1044 {
1045         DocIterator pit = cpit;
1046         pit.push_back(CursorSlice(const_cast<InsetIndex &>(*this)));
1047         docstring str;
1048         InsetLayout const & il = getLayout();
1049         docstring label = translateIfPossible(il.labelstring());
1050         if (!il.contentaslabel())
1051                 str = label;
1052         else {
1053                 str = getNewLabel(label);
1054                 OutputParams const rp(0);
1055                 vector<docstring> sublbls = getSubentriesAsText(rp, true);
1056                 for (auto const & sublbl : sublbls) {
1057                         str += " " + docstring(1, char_type(0x2023));// TRIANGULAR BULLET
1058                         str += " " + sublbl;
1059                 }
1060                 docstring see = getSeeAsText(rp, true);
1061                 if (see.empty() && !getSeeAlsoesAsText(rp, true).empty())
1062                         see = getSeeAlsoesAsText(rp, true).front();
1063                 if (!see.empty()) {
1064                         str += " " + docstring(1, char_type(0x261e));// WHITE RIGHT POINTING INDEX
1065                         str += " " + see;
1066                 }
1067         }
1068         string type = "index";
1069         if (buffer().masterBuffer()->params().use_indices)
1070                 type += ":" + to_utf8(params_.index);
1071         TocBuilder & b = backend.builder(type);
1072         b.pushItem(pit, str, output_active);
1073         // Proceed with the rest of the inset.
1074         InsetCollapsible::addToToc(cpit, output_active, utype, backend);
1075         b.pop();
1076 }
1077
1078
1079 void InsetIndex::validate(LaTeXFeatures & features) const
1080 {
1081         if (buffer().masterBuffer()->params().use_indices
1082             && !params_.index.empty()
1083             && params_.index != "idx")
1084                 features.require("splitidx");
1085         InsetCollapsible::validate(features);
1086 }
1087
1088
1089 string InsetIndex::contextMenuName() const
1090 {
1091         return "context-index";
1092 }
1093
1094
1095 string InsetIndex::contextMenu(BufferView const & bv, int x, int y) const
1096 {
1097         // We override the implementation of InsetCollapsible,
1098         // because we have eytra entries.
1099         string owncm = "context-edit-index;";
1100         return owncm + InsetCollapsible::contextMenu(bv, x, y);
1101 }
1102
1103
1104 bool InsetIndex::hasSettings() const
1105 {
1106         return true;
1107 }
1108
1109
1110 bool InsetIndex::insetAllowed(InsetCode code) const
1111 {
1112         switch (code) {
1113         case INDEXMACRO_CODE:
1114         case INDEXMACRO_SORTKEY_CODE:
1115                 return true;
1116         case INDEX_CODE:
1117                 return false;
1118         default:
1119                 return InsetCollapsible::insetAllowed(code);
1120         }
1121 }
1122
1123
1124 /////////////////////////////////////////////////////////////////////
1125 //
1126 // InsetIndexParams
1127 //
1128 ///////////////////////////////////////////////////////////////////////
1129
1130
1131 void InsetIndexParams::write(ostream & os) const
1132 {
1133         os << ' ';
1134         if (!index.empty())
1135                 os << to_utf8(index);
1136         else
1137                 os << "idx";
1138         os << '\n';
1139         os << "range "
1140            << insetindexpagerangetranslator().find(range)
1141            << '\n';
1142         os << "pageformat "
1143            << pagefmt
1144            << '\n';
1145 }
1146
1147
1148 void InsetIndexParams::read(Lexer & lex)
1149 {
1150         if (lex.eatLine())
1151                 index = lex.getDocString();
1152         else
1153                 index = from_ascii("idx");
1154         if (lex.checkFor("range")) {
1155                 string st = lex.getString();
1156                 if (lex.eatLine()) {
1157                         st = lex.getString();
1158                         range = insetindexpagerangetranslator().find(lex.getString());
1159                 }
1160         }
1161         if (lex.checkFor("pageformat") && lex.eatLine()) {
1162                 pagefmt = lex.getString();
1163         }
1164 }
1165
1166
1167 /////////////////////////////////////////////////////////////////////
1168 //
1169 // InsetPrintIndex
1170 //
1171 ///////////////////////////////////////////////////////////////////////
1172
1173 InsetPrintIndex::InsetPrintIndex(Buffer * buf, InsetCommandParams const & p)
1174         : InsetCommand(buf, p)
1175 {}
1176
1177
1178 ParamInfo const & InsetPrintIndex::findInfo(string const & /* cmdName */)
1179 {
1180         static ParamInfo param_info_;
1181         if (param_info_.empty()) {
1182                 param_info_.add("type", ParamInfo::LATEX_OPTIONAL,
1183                                 ParamInfo::HANDLING_ESCAPE);
1184                 param_info_.add("name", ParamInfo::LATEX_OPTIONAL,
1185                                 ParamInfo::HANDLING_LATEXIFY);
1186                 param_info_.add("literal", ParamInfo::LYX_INTERNAL);
1187         }
1188         return param_info_;
1189 }
1190
1191
1192 docstring InsetPrintIndex::screenLabel() const
1193 {
1194         bool const printall = suffixIs(getCmdName(), '*');
1195         bool const multind = buffer().masterBuffer()->params().use_indices;
1196         if ((!multind
1197              && getParam("type") == from_ascii("idx"))
1198             || (getParam("type").empty() && !printall))
1199                 return _("Index");
1200         Buffer const & realbuffer = *buffer().masterBuffer();
1201         IndicesList const & indiceslist = realbuffer.params().indiceslist();
1202         Index const * index = indiceslist.findShortcut(getParam("type"));
1203         if (!index && !printall)
1204                 return _("Unknown index type!");
1205         docstring res = printall ? _("All indexes") : index->index();
1206         if (!multind)
1207                 res += " (" + _("non-active") + ")";
1208         else if (contains(getCmdName(), "printsubindex"))
1209                 res += " (" + _("subindex") + ")";
1210         return res;
1211 }
1212
1213
1214 bool InsetPrintIndex::isCompatibleCommand(string const & s)
1215 {
1216         return s == "printindex" || s == "printsubindex"
1217                 || s == "printindex*" || s == "printsubindex*";
1218 }
1219
1220
1221 void InsetPrintIndex::doDispatch(Cursor & cur, FuncRequest & cmd)
1222 {
1223         switch (cmd.action()) {
1224
1225         case LFUN_INSET_MODIFY: {
1226                 if (cmd.argument() == from_ascii("toggle-subindex")) {
1227                         string scmd = getCmdName();
1228                         if (contains(scmd, "printindex"))
1229                                 scmd = subst(scmd, "printindex", "printsubindex");
1230                         else
1231                                 scmd = subst(scmd, "printsubindex", "printindex");
1232                         cur.recordUndo();
1233                         setCmdName(scmd);
1234                         break;
1235                 } else if (cmd.argument() == from_ascii("check-printindex*")) {
1236                         string scmd = getCmdName();
1237                         if (suffixIs(scmd, '*'))
1238                                 break;
1239                         scmd += '*';
1240                         cur.recordUndo();
1241                         setParam("type", docstring());
1242                         setCmdName(scmd);
1243                         break;
1244                 }
1245                 InsetCommandParams p(INDEX_PRINT_CODE);
1246                 // FIXME UNICODE
1247                 InsetCommand::string2params(to_utf8(cmd.argument()), p);
1248                 if (p.getCmdName().empty()) {
1249                         cur.noScreenUpdate();
1250                         break;
1251                 }
1252                 cur.recordUndo();
1253                 setParams(p);
1254                 break;
1255         }
1256
1257         default:
1258                 InsetCommand::doDispatch(cur, cmd);
1259                 break;
1260         }
1261 }
1262
1263
1264 bool InsetPrintIndex::getStatus(Cursor & cur, FuncRequest const & cmd,
1265         FuncStatus & status) const
1266 {
1267         switch (cmd.action()) {
1268
1269         case LFUN_INSET_MODIFY: {
1270                 if (cmd.argument() == from_ascii("toggle-subindex")) {
1271                         status.setEnabled(buffer().masterBuffer()->params().use_indices);
1272                         status.setOnOff(contains(getCmdName(), "printsubindex"));
1273                         return true;
1274                 } else if (cmd.argument() == from_ascii("check-printindex*")) {
1275                         status.setEnabled(buffer().masterBuffer()->params().use_indices);
1276                         status.setOnOff(suffixIs(getCmdName(), '*'));
1277                         return true;
1278                 } if (cmd.getArg(0) == "index_print"
1279                     && cmd.getArg(1) == "CommandInset") {
1280                         InsetCommandParams p(INDEX_PRINT_CODE);
1281                         InsetCommand::string2params(to_utf8(cmd.argument()), p);
1282                         if (suffixIs(p.getCmdName(), '*')) {
1283                                 status.setEnabled(true);
1284                                 status.setOnOff(false);
1285                                 return true;
1286                         }
1287                         Buffer const & realbuffer = *buffer().masterBuffer();
1288                         IndicesList const & indiceslist =
1289                                 realbuffer.params().indiceslist();
1290                         Index const * index = indiceslist.findShortcut(p["type"]);
1291                         status.setEnabled(index != 0);
1292                         status.setOnOff(p["type"] == getParam("type"));
1293                         return true;
1294                 } else
1295                         return InsetCommand::getStatus(cur, cmd, status);
1296         }
1297
1298         case LFUN_INSET_DIALOG_UPDATE: {
1299                 status.setEnabled(buffer().masterBuffer()->params().use_indices);
1300                 return true;
1301         }
1302
1303         default:
1304                 return InsetCommand::getStatus(cur, cmd, status);
1305         }
1306 }
1307
1308
1309 void InsetPrintIndex::updateBuffer(ParIterator const &, UpdateType, bool const /*deleted*/)
1310 {
1311         Index const * index =
1312                 buffer().masterParams().indiceslist().findShortcut(getParam("type"));
1313         if (index)
1314                 setParam("name", index->index());
1315 }
1316
1317
1318 void InsetPrintIndex::latex(otexstream & os, OutputParams const & runparams_in) const
1319 {
1320         if (!buffer().masterBuffer()->params().use_indices) {
1321                 if (getParam("type") == from_ascii("idx"))
1322                         os << "\\printindex" << termcmd;
1323                 return;
1324         }
1325         OutputParams runparams = runparams_in;
1326         os << getCommand(runparams);
1327 }
1328
1329
1330 void InsetPrintIndex::validate(LaTeXFeatures & features) const
1331 {
1332         features.require("makeidx");
1333         if (buffer().masterBuffer()->params().use_indices)
1334                 features.require("splitidx");
1335         InsetCommand::validate(features);
1336 }
1337
1338
1339 string InsetPrintIndex::contextMenuName() const
1340 {
1341         return buffer().masterBuffer()->params().use_indices ?
1342                 "context-indexprint" : string();
1343 }
1344
1345
1346 bool InsetPrintIndex::hasSettings() const
1347 {
1348         return buffer().masterBuffer()->params().use_indices;
1349 }
1350
1351
1352 class IndexEntry
1353 {
1354 public:
1355         /// Builds an entry for the index.
1356         IndexEntry(const InsetIndex * inset, OutputParams const * runparams) : inset_(inset), runparams_(runparams)
1357         {
1358                 LASSERT(runparams, return);
1359
1360                 // Convert the inset as text. The resulting text usually only contains an XHTML anchor (<a id='...'/>) and text.
1361                 odocstringstream entry;
1362                 OutputParams ours = *runparams;
1363                 ours.for_toc = false;
1364                 inset_->plaintext(entry, ours);
1365                 entry_ = entry.str();
1366
1367                 // Determine in which index this entry belongs to.
1368                 if (inset_->buffer().masterBuffer()->params().use_indices) {
1369                         index_ = inset_->params_.index;
1370                 }
1371
1372                 // Attempt parsing the inset.
1373                 if (isModern())
1374                         parseAsModern();
1375                 else
1376                         parseAsLegacy();
1377         }
1378
1379         /// When parsing this entry, some errors may be found; they are reported as a single string.
1380         // It is up to the caller to send this string to LYXERR and the output file, as needed.
1381         const docstring & output_error() const
1382         {
1383                 return output_error_;
1384         }
1385
1386         void output_error(XMLStream xs) const
1387         {
1388                 LYXERR0(output_error());
1389                 xs << XMLStream::ESCAPE_NONE << (from_utf8("<!-- Output Error: ") + output_error() + from_utf8(" -->\n"));
1390         }
1391
1392
1393 private:
1394         bool isModern()
1395         {
1396                 std::cout << to_utf8(entry_) << std::endl;
1397
1398                 // If a modern parameter is present, this is definitely a modern index inset. Similarly, if it contains the
1399                 // usual LaTeX symbols (!|@), then it is definitely a legacy index inset. Otherwise, if it has features of
1400                 // neither, it is both: consider this is a modern inset, to trigger the least complex code. Mixing both types
1401                 // is not allowed (i.e. behaviour is undefined).
1402                 const bool is_definitely_modern = inset_->hasSortKey() || inset_->hasSeeRef() || inset_->hasSubentries()
1403                                             || inset_->params_.range != InsetIndexParams::PageRange::None;
1404                 const bool is_definitely_legacy = entry_.find('@') != std::string::npos
1405                                 || entry_.find('|') != std::string::npos || entry_.find('!') != std::string::npos;
1406
1407                 if (is_definitely_legacy && is_definitely_modern)
1408                         output_error_ += from_utf8("Mix of index properties and raw LaTeX index commands is unsupported. ");
1409
1410                 // Truth table:
1411                 // - is_definitely_modern == true:
1412                 //   - is_definitely_legacy == true: error (return whatever)
1413                 //   - is_definitely_legacy == false: return modern
1414                 // - is_definitely_modern == false:
1415                 //   - is_definitely_legacy == true: return legacy
1416                 //   - is_definitely_legacy == false: return modern
1417                 return !is_definitely_legacy;
1418         }
1419
1420         void parseAsModern()
1421         {
1422                 LASSERT(runparams_, return);
1423
1424                 if (inset_->hasSortKey()) {
1425                         sort_as_ = inset_->getSortkeyAsText(*runparams_);
1426                 }
1427
1428                 terms_ = inset_->getSubentriesAsText(*runparams_);
1429                 // The main term is not present in the vector, as it's not a subentry. The main index term is inserted raw in
1430                 // the index inset. Considering that the user either uses the new or the legacy mechanism, the main term is the
1431                 // full string within this inset (i.e. without the subinsets).
1432                 terms_.insert(terms_.begin(), inset_->getMainSubentryAsText(*runparams_));
1433
1434                 has_start_range_ = inset_->params_.range == InsetIndexParams::PageRange::Start;
1435                 has_end_range_ = inset_->params_.range == InsetIndexParams::PageRange::End;
1436
1437                 see_ = inset_->getSeeAsText(*runparams_);
1438                 see_alsoes_ = inset_->getSeeAlsoesAsText(*runparams_);
1439         }
1440
1441         void parseAsLegacy() {
1442                 // Determine if some features are known not to be supported. For now, this is only formatting like
1443                 // \index{alpha@\textbf{alpha}} or \index{alpha@$\alpha$}.
1444                 // @ is supported, but only for sorting, without specific formatting.
1445                 if (entry_.find(from_utf8("@\\")) != lyx::docstring::npos) {
1446                         output_error_ += from_utf8("Unsupported feature: an index entry contains an @\\. "
1447                                                    "Complete entry: \"") + entry_ + from_utf8("\". ");
1448                 }
1449                 if (entry_.find(from_utf8("@$")) != lyx::docstring::npos) {
1450                         output_error_ += from_utf8("Unsupported feature: an index entry contains an @$. "
1451                                                    "Complete entry: \"") + entry_ + from_utf8("\". ");
1452                 }
1453
1454                 // Split the string into its main constituents: terms, and command (see, see also, range).
1455                 size_t positionVerticalBar = entry_.find(from_ascii("|")); // What comes before | is (sub)(sub)entries.
1456                 docstring indexTerms = entry_.substr(0, positionVerticalBar);
1457                 docstring command;
1458                 if (positionVerticalBar != lyx::docstring::npos) {
1459                         command = entry_.substr(positionVerticalBar + 1);
1460                 }
1461
1462                 // Handle sorting issues, with @.
1463                 vector<docstring> sortingElements = getVectorFromString(indexTerms, from_ascii("@"), false);
1464                 if (sortingElements.size() == 2) {
1465                         sort_as_ = sortingElements[0];
1466                         indexTerms = sortingElements[1];
1467                 }
1468
1469                 // Handle entries, subentries, and subsubentries.
1470                 terms_ = getVectorFromString(indexTerms, from_ascii("!"), false);
1471
1472                 // Handle ranges. Happily, (| and |) can only be at the end of the string!
1473                 has_start_range_ = entry_.find(from_ascii("|(")) != lyx::docstring::npos;
1474                 has_end_range_ = entry_.find(from_ascii("|)")) != lyx::docstring::npos;
1475
1476                 // - Remove the ranges from the command if they do not appear at the beginning.
1477                 size_t range_index = 0;
1478                 while ((range_index = command.find(from_utf8("|("), range_index)) != std::string::npos)
1479                         command.erase(range_index, 1);
1480                 range_index = 0;
1481                 while ((range_index = command.find(from_utf8("|)"), range_index)) != std::string::npos)
1482                         command.erase(range_index, 1);
1483
1484                 // - Remove the ranges when they are the only vertical bar in the complete string.
1485                 if (command[0] == '(' || command[0] == ')')
1486                         command.erase(0, 1);
1487
1488                 // Handle see and seealso. As "see" is a prefix of "seealso", the order of the comparisons is important.
1489                 // Both commands are mutually exclusive!
1490                 if (command.substr(0, 3) == "see") {
1491                         // Unescape brackets.
1492                         size_t index_argument_begin = 0;
1493                         while ((index_argument_begin = command.find(from_utf8("\\{"), index_argument_begin)) != std::string::npos)
1494                                 command.erase(index_argument_begin, 1);
1495                         size_t index_argument_end = 0;
1496                         while ((index_argument_end = command.find(from_utf8("\\}"), index_argument_end)) != std::string::npos)
1497                                 command.erase(index_argument_end, 1);
1498
1499                         // Retrieve the part between brackets, and remove the complete seealso.
1500                         size_t position_opening_bracket = command.find(from_ascii("{"));
1501                         size_t position_closing_bracket = command.find(from_ascii("}"));
1502                         docstring argument = command.substr(position_opening_bracket + 1,
1503                                                                                                 position_closing_bracket - position_opening_bracket - 1);
1504
1505                         // Parse the argument of referenced entries (or a single one for see).
1506                         if (command.substr(0, 7) == "seealso") {
1507                                 see_alsoes_ = getVectorFromString(argument, from_ascii(","), false);
1508                         } else {
1509                                 see_ = argument;
1510
1511                                 if (see_.find(from_ascii(",")) != std::string::npos) {
1512                                         output_error_ += from_utf8("Several index_argument_end terms found as \"see\"! Only one is "
1513                                                                    "acceptable. Complete entry: \"") + entry_ + from_utf8("\". ");
1514                                 }
1515                         }
1516
1517                         // Remove the complete see/seealso from the commands, in case there is something else to parse.
1518                         command = command.substr(position_closing_bracket + 1);
1519                 }
1520
1521                 // Some parts of the strings are not parsed, as they do not have anything matching in DocBook or XHTML:
1522                 // things like formatting the entry or the page number, other strings for sorting.
1523                 // https://wiki.lyx.org/Tips/Indexing
1524                 // If there are such things in the index entry, then this code may miserably fail. For example, for
1525                 // "Peter|(textbf", no range will be detected.
1526                 if (!command.empty()) {
1527                         output_error_ += from_utf8("Unsupported feature: an index entry contains a | with an unsupported command, ")
1528                                          + command + from_utf8(". Complete entry: \"") + entry_ + from_utf8("\". ");
1529                 }
1530         }
1531
1532 public:
1533         int level() const {
1534                 return terms_.size();
1535         }
1536
1537         const std::vector<docstring>& terms() const {
1538                 return terms_;
1539         }
1540
1541         std::vector<docstring>& terms() {
1542                 return terms_;
1543         }
1544
1545         const InsetIndex* inset() const {
1546                 return inset_;
1547         }
1548
1549 private:
1550         // Input inset. These should only be used when parsing the inset (either parseAsModern or parseAsLegacy, called in
1551         // the constructor).
1552         const InsetIndex * inset_;
1553         OutputParams const * runparams_;
1554         docstring entry_;
1555         docstring index_; // Useful when there are multiple indices in the same document.
1556
1557         // Errors, concatenated as a single string, available as soon as parsing is done, const afterwards (i.e. once
1558         // constructor is done).
1559         docstring output_error_;
1560
1561         // Parsed index entry.
1562         std::vector<docstring> terms_; // Up to three entries, in general.
1563         docstring sort_as_;
1564         docstring command_;
1565         bool has_start_range_;
1566         bool has_end_range_;
1567         docstring see_;
1568         vector<docstring> see_alsoes_;
1569
1570         // Operators used for sorting entries (alphabetical order).
1571         friend bool operator<(IndexEntry const & lhs, IndexEntry const & rhs);
1572 };
1573
1574 bool operator<(IndexEntry const & lhs, IndexEntry const & rhs)
1575 {
1576         if (lhs.terms_.empty())
1577                 return false;
1578
1579         for (unsigned i = 0; i < min(rhs.terms_.size(), lhs.terms_.size()); ++i) {
1580                 int comp = compare_no_case(lhs.terms_[i], rhs.terms_[i]);
1581                 if (comp != 0)
1582                         return comp < 0;
1583         }
1584         return false;
1585 }
1586
1587
1588 namespace {
1589 std::string generateCssClassAtDepth(unsigned depth) {
1590         std::string css_class = "entry";
1591
1592         while (depth > 0) {
1593                 depth -= 1;
1594                 css_class.insert(0, "sub");
1595         }
1596
1597         return css_class;
1598 }
1599
1600 struct IndexNode {
1601         std::vector<IndexEntry> entries;
1602         std::vector<IndexNode*> children;
1603 };
1604
1605 docstring termAtLevel(const IndexNode* node, unsigned depth)
1606 {
1607         // 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
1608         // function could be made constant time by copying the term in each node, but that would make data duplication that
1609         // may fall out of sync; the performance benefit would probably be negligible.
1610         if (!node->entries.empty()) {
1611                 LASSERT(node->entries.begin()->terms().size() >= depth + 1, return from_ascii(""));
1612                 return node->entries.begin()->terms()[depth];
1613         }
1614
1615         if (!node->children.empty()) {
1616                 return termAtLevel(*node->children.begin(), depth);
1617         }
1618
1619         LASSERT(false, return from_ascii(""));
1620 }
1621
1622 void insertIntoNode(const IndexEntry& entry, IndexNode* node, unsigned depth = 0)
1623 {
1624         // depth == 0 is for the root, not yet the index, hence the increase when going to vector size.
1625         for (IndexNode* child : node->children) {
1626                 if (entry.terms()[depth] == termAtLevel(child, depth)) {
1627                         if (depth + 1 == entry.terms().size()) { // == child.entries.begin()->terms().size()
1628                                 // All term entries match: it's an entry.
1629                                 child->entries.emplace_back(entry);
1630                                 return;
1631                         } else {
1632                                 insertIntoNode(entry, child, depth + 1);
1633                                 return;
1634                         }
1635                 }
1636         }
1637
1638         // Out of the loop: no matching child found, create a new (possibly nested) child for this entry. Due to the
1639         // possibility of nestedness, only insert the current entry when the right level is reached. This is needed if the
1640         // first entry for a word has several levels that never appeared.
1641         // In particular, this case is called for the first entry.
1642         IndexNode* new_node = node;
1643         do {
1644                 new_node->children.emplace_back(new IndexNode{{}, {}});
1645                 new_node = new_node->children.back();
1646                 depth += 1;
1647         } while (depth + 1 <= entry.terms().size()); // depth == 0: root node, no text associated.
1648         new_node->entries.emplace_back(entry);
1649 }
1650
1651 IndexNode* buildIndexTree(vector<IndexEntry>& entries)
1652 {
1653         // Sort the entries, first on the main entry, then the subentry, then the subsubentry,
1654         // thanks to the implementation of operator<.
1655         // If this operation is not performed, the algorithm below is no more correct (and ensuring that it works with
1656         // unsorted entries would make its complexity blow up).
1657         stable_sort(entries.begin(), entries.end());
1658
1659         // Cook the index into a nice tree data structure: entries at a given level in the index as a node, with subentries
1660         // as children.
1661         auto* index_root = new IndexNode{{}, {}};
1662         for (const IndexEntry& entry : entries) {
1663                 insertIntoNode(entry, index_root);
1664         }
1665
1666         return index_root;
1667 }
1668
1669 void outputIndexPage(XMLStream & xs, const IndexNode* root_node, unsigned depth = 0) // NOLINT(misc-no-recursion)
1670 {
1671         LASSERT(root_node->entries.size() + root_node->children.size() > 0, return);
1672
1673         xs << xml::StartTag("li", "class='" + generateCssClassAtDepth(depth) + "'");
1674         xs << xml::CR();
1675         xs << XMLStream::ESCAPE_NONE << termAtLevel(root_node, depth);
1676         // By tree assumption, all the entries at this node have the same set of terms.
1677
1678         if (!root_node->entries.empty()) {
1679                 xs << XMLStream::ESCAPE_NONE << " &#8212; "; // Em dash, i.e. long (---).
1680                 unsigned entry_number = 1;
1681
1682                 auto writeLinkToEntry = [&xs](const IndexEntry &entry, unsigned entry_number) {
1683                         std::string const link_attr = "href='#" + entry.inset()->paragraphs()[0].magicLabel() + "'";
1684                         xs << xml::StartTag("a", link_attr);
1685                         xs << from_ascii(std::to_string(entry_number));
1686                         xs << xml::EndTag("a");
1687                 };
1688
1689                 for (unsigned i = 0; i < root_node->entries.size(); ++i) {
1690                         const IndexEntry &entry = root_node->entries[i];
1691
1692                         switch (entry.inset()->params().range) {
1693                                 case InsetIndexParams::PageRange::None:
1694                                         writeLinkToEntry(entry, entry_number);
1695                                         break;
1696                                 case InsetIndexParams::PageRange::Start: {
1697                                         // Try to find the end of the range, if it is just after. Otherwise, the output will be slightly
1698                                         // scrambled, but understandable. Doing better would mean implementing more of the indexing logic here
1699                                         // and more complex indexing here (skipping the end is not just incrementing i). Worst case output:
1700                                         //     1--, 2, --3
1701                                         const bool nextEntryIsEnd = i + 1 < root_node->entries.size() &&
1702                                                                     root_node->entries[i + 1].inset()->params().range ==
1703                                                                     InsetIndexParams::PageRange::End;
1704                                         // No need to check if both entries are for the same terms: they are in the same IndexNode.
1705
1706                                         writeLinkToEntry(entry, entry_number);
1707                                         xs << XMLStream::ESCAPE_NONE << " &#8211; "; // En dash, i.e. semi-long (--).
1708
1709                                         if (nextEntryIsEnd) {
1710                                                 // Skip the next entry in the loop, write it right now, after the dash.
1711                                                 entry_number += 1;
1712                                                 i += 1;
1713                                                 writeLinkToEntry(root_node->entries[i], entry_number);
1714                                         }
1715                                 }
1716                                         break;
1717                                 case InsetIndexParams::PageRange::End:
1718                                         // This range end was not caught by the range start, do it now to avoid losing content.
1719                                         xs << XMLStream::ESCAPE_NONE << " &#8211; "; // En dash, i.e. semi-long (--).
1720                                         writeLinkToEntry(root_node->entries[i], entry_number);
1721                         }
1722
1723                         if (i < root_node->entries.size() - 1) {
1724                                 xs << ", ";
1725                         }
1726                         entry_number += 1;
1727                 }
1728         }
1729
1730         if (!root_node->entries.empty() && !root_node->children.empty()) {
1731                 xs << xml::CR();
1732         }
1733
1734         if (!root_node->children.empty()) {
1735                 xs << xml::StartTag("ul", "class='" + generateCssClassAtDepth(depth) + "'");
1736                 xs << xml::CR();
1737
1738                 for (const IndexNode* child : root_node->children) {
1739                         outputIndexPage(xs, child, depth + 1);
1740                 }
1741
1742                 xs << xml::EndTag("ul");
1743                 xs << xml::CR();
1744         }
1745
1746         xs << xml::EndTag("li");
1747         xs << xml::CR();
1748 }
1749
1750 #ifdef LYX_INSET_INDEX_DEBUG
1751 void printTree(const IndexNode* root_node, unsigned depth = 0)
1752 {
1753         static const std::string pattern = "    ";
1754         std::string prefix;
1755         for (unsigned i = 0; i < depth; ++i) {
1756                 prefix += pattern;
1757         }
1758         const std::string prefix_long = prefix + pattern + pattern;
1759
1760         docstring term_at_level;
1761         if (depth == 0) {
1762                 // The root has no term.
1763                 std::cout << "<ROOT>" << std::endl;
1764         } else {
1765                 LASSERT(depth - 1 <= 10, return); // Check for overflows.
1766                 term_at_level = termAtLevel(root_node, depth - 1);
1767                 std::cout << prefix << to_utf8(term_at_level) << " (x " << std::to_string(root_node->entries.size()) << ")"
1768                           << std::endl;
1769         }
1770
1771         for (const IndexEntry& entry : root_node->entries) {
1772                 if (entry.terms().size() != depth) {
1773                         std::cout << prefix_long << "ERROR: an entry doesn't have the same number of terms" << std::endl;
1774                 }
1775                 if (depth > 0 && entry.terms()[depth - 1] != term_at_level) {
1776                         std::cout << prefix_long << "ERROR: an entry doesn't have the right term at depth " << std::to_string(depth)
1777                                 << std::endl;
1778                 }
1779         }
1780
1781         for (const IndexNode* node : root_node->children) {
1782                 printTree(node, depth + 1);
1783         }
1784 }
1785 #endif // LYX_INSET_INDEX_DEBUG
1786 }
1787
1788
1789 docstring InsetPrintIndex::xhtml(XMLStream &, OutputParams const & op) const
1790 {
1791         BufferParams const & bp = buffer().masterBuffer()->params();
1792
1793         shared_ptr<Toc const> toc = buffer().tocBackend().toc("index");
1794         if (toc->empty())
1795                 return docstring();
1796
1797         // Collect the index entries in a form we can use them.
1798         vector<IndexEntry> entries;
1799         const docstring & indexType = params().getParamOr("type", from_ascii("idx"));
1800         for (const TocItem& item : *toc) {
1801                 const auto* inset = static_cast<const InsetIndex*>(&(item.dit().inset()));
1802                 if (item.isOutput() && inset->params().index == indexType)
1803                         entries.emplace_back(IndexEntry{inset, &op});
1804         }
1805
1806         // If all the index entries are in notes or not displayed, get out sooner.
1807         if (entries.empty())
1808                 return docstring();
1809
1810         const IndexNode* index_root = buildIndexTree(entries);
1811 #ifdef LYX_INSET_INDEX_DEBUG
1812         printTree(index_root);
1813 #endif
1814
1815         // Start generating the XHTML index.
1816         Layout const & lay = bp.documentClass().htmlTOCLayout();
1817         string const & tocclass = lay.defaultCSSClass();
1818         string const tocattr = "class='index " + tocclass + "'";
1819         docstring const indexName = params().getParamOr("name", from_ascii("Index"));
1820
1821         // we'll use our own stream, because we are going to defer everything.
1822         // that's how we deal with the fact that we're probably inside a standard
1823         // paragraph, and we don't want to be.
1824         odocstringstream ods;
1825         XMLStream xs(ods);
1826
1827         xs << xml::StartTag("div", tocattr);
1828         xs << xml::CR();
1829         xs << xml::StartTag(lay.htmltag(), lay.htmlattr());
1830         xs << translateIfPossible(indexName, op.local_font->language()->lang());
1831         xs << xml::EndTag(lay.htmltag());
1832         xs << xml::CR();
1833         xs << xml::StartTag("ul", "class='main'");
1834         xs << xml::CR();
1835
1836         LASSERT(index_root->entries.empty(), return docstring()); // No index entry should have zero terms.
1837         for (const IndexNode* node : index_root->children) {
1838                 outputIndexPage(xs, node);
1839         }
1840
1841         xs << xml::EndTag("ul");
1842         xs << xml::CR();
1843         xs << xml::EndTag("div");
1844
1845         return ods.str();
1846 }
1847
1848 } // namespace lyx