/* * Copyright 2018 Google Inc. * * Use of this source code is governed by a BSD-style license that can be * found in the LICENSE file. */ #include "bmhParser.h" const string kSpellingFileName("spelling.txt"); #define M(mt) (1LL << (int) MarkType::k##mt) #define M_D M(Description) #define M_CS M(Class) | M(Struct) #define M_MD M(Method) | M(Define) #define M_MDCM M_MD | M(Const) | M(Member) #define M_ST M(Subtopic) | M(Topic) #define M_CSST M_CS | M_ST #ifdef M_E #undef M_E #endif #define M_E M(Enum) | M(EnumClass) #define R_Y Resolvable::kYes #define R_N Resolvable::kNo #define R_O Resolvable::kOut #define R_K Resolvable::kCode #define R_F Resolvable::kFormula #define R_C Resolvable::kClone #define E_Y Exemplary::kYes #define E_N Exemplary::kNo #define E_O Exemplary::kOptional // ToDo: add column to denote which marks are one-liners BmhParser::MarkProps BmhParser::kMarkProps[] = { // names without formal definitions (e.g. Column) aren't included { "", MarkType::kNone, R_Y, E_N, 0 } , { "A", MarkType::kAnchor, R_N, E_N, 0 } , { "Alias", MarkType::kAlias, R_N, E_N, M_ST | M(Const) } , { "Bug", MarkType::kBug, R_N, E_N, M_CSST | M_MDCM | M_E | M(Example) | M(NoExample) } , { "Class", MarkType::kClass, R_Y, E_O, M_CSST } , { "Code", MarkType::kCode, R_K, E_N, M_CSST | M_E | M_MD | M(Typedef) } , { "", MarkType::kColumn, R_Y, E_N, M(Row) } , { "", MarkType::kComment, R_N, E_N, 0 } , { "Const", MarkType::kConst, R_Y, E_O, M_E | M_CSST } , { "Define", MarkType::kDefine, R_O, E_Y, M_ST } , { "Description", MarkType::kDescription, R_Y, E_N, M(Example) | M(NoExample) } , { "Details", MarkType::kDetails, R_N, E_N, M(Const) } , { "Duration", MarkType::kDuration, R_N, E_N, M(Example) | M(NoExample) } , { "Enum", MarkType::kEnum, R_Y, E_O, M_CSST } , { "EnumClass", MarkType::kEnumClass, R_Y, E_O, M_CSST } , { "Example", MarkType::kExample, R_O, E_N, M_CSST | M_E | M_MD | M(Const) } , { "External", MarkType::kExternal, R_Y, E_N, 0 } , { "File", MarkType::kFile, R_Y, E_N, M(Topic) } , { "Filter", MarkType::kFilter, R_N, E_N, M(Subtopic) | M(Code) } , { "Formula", MarkType::kFormula, R_F, E_N, M(Column) | M(Description) | M_E | M_ST | M_MDCM } , { "Function", MarkType::kFunction, R_O, E_N, M(Example) | M(NoExample) } , { "Height", MarkType::kHeight, R_N, E_N, M(Example) | M(NoExample) } , { "Illustration", MarkType::kIllustration, R_N, E_N, M_CSST | M_MD } , { "Image", MarkType::kImage, R_N, E_N, M(Example) | M(NoExample) } , { "In", MarkType::kIn, R_N, E_N, M_CSST | M_E | M(Method) | M(Typedef) | M(Code) } , { "Legend", MarkType::kLegend, R_Y, E_N, M(Table) } , { "Line", MarkType::kLine, R_N, E_N, M_CSST | M_E | M(Method) | M(Typedef) } , { "", MarkType::kLink, R_N, E_N, M(Anchor) } , { "List", MarkType::kList, R_Y, E_N, M(Method) | M_CSST | M_E | M_D } , { "Literal", MarkType::kLiteral, R_N, E_N, M(Code) } , { "", MarkType::kMarkChar, R_N, E_N, 0 } , { "Member", MarkType::kMember, R_Y, E_O, M_CSST } , { "Method", MarkType::kMethod, R_Y, E_Y, M_CSST } , { "NoExample", MarkType::kNoExample, R_N, E_N, M_CSST | M_E | M_MD } , { "NoJustify", MarkType::kNoJustify, R_N, E_N, M(Const) | M(Member) } , { "Outdent", MarkType::kOutdent, R_N, E_N, M(Code) } , { "Param", MarkType::kParam, R_Y, E_N, M(Method) | M(Define) } , { "PhraseDef", MarkType::kPhraseDef, R_Y, E_N, M_ST } , { "", MarkType::kPhraseParam, R_Y, E_N, 0 } , { "", MarkType::kPhraseRef, R_N, E_N, 0 } , { "Platform", MarkType::kPlatform, R_N, E_N, M(Example) | M(NoExample) } , { "Populate", MarkType::kPopulate, R_N, E_N, M(Code) | M(Method) } , { "Return", MarkType::kReturn, R_Y, E_N, M(Method) } , { "", MarkType::kRow, R_Y, E_N, M(Table) | M(List) } , { "SeeAlso", MarkType::kSeeAlso, R_C, E_N, M_CSST | M_E | M_MD | M(Typedef) } , { "Set", MarkType::kSet, R_N, E_N, M(Example) | M(NoExample) } , { "StdOut", MarkType::kStdOut, R_N, E_N, M(Example) | M(NoExample) } , { "Struct", MarkType::kStruct, R_Y, E_O, M(Class) | M_ST } , { "Substitute", MarkType::kSubstitute, R_N, E_N, M(Alias) | M_ST } , { "Subtopic", MarkType::kSubtopic, R_Y, E_Y, M_CSST | M_E } , { "Table", MarkType::kTable, R_Y, E_N, M(Method) | M_CSST | M_E } , { "Template", MarkType::kTemplate, R_Y, E_N, M_CSST } , { "", MarkType::kText, R_N, E_N, 0 } , { "ToDo", MarkType::kToDo, R_N, E_N, 0 } , { "Topic", MarkType::kTopic, R_Y, E_Y, 0 } , { "Typedef", MarkType::kTypedef, R_Y, E_O, M_CSST | M_E } , { "Union", MarkType::kUnion, R_Y, E_N, M_CSST } , { "Using", MarkType::kUsing, R_Y, E_O, M_CSST } , { "Volatile", MarkType::kVolatile, R_N, E_N, M(StdOut) } , { "Width", MarkType::kWidth, R_N, E_N, M(Example) | M(NoExample) } }; #undef R_O #undef R_N #undef R_Y #undef R_K #undef R_F #undef R_C #undef M_E #undef M_CSST #undef M_ST #undef M_CS #undef M_MCD #undef M_D #undef M #undef E_Y #undef E_N #undef E_O bool BmhParser::addDefinition(const char* defStart, bool hasEnd, MarkType markType, const vector& typeNameBuilder, HasTag hasTag) { Definition* definition = nullptr; switch (markType) { case MarkType::kComment: if (!this->skipToDefinitionEnd(markType)) { return false; } return true; // these types may be referred to by name case MarkType::kClass: case MarkType::kStruct: case MarkType::kConst: case MarkType::kDefine: case MarkType::kEnum: case MarkType::kEnumClass: case MarkType::kMember: case MarkType::kMethod: case MarkType::kTemplate: case MarkType::kTypedef: { if (!typeNameBuilder.size()) { return this->reportError("unnamed markup"); } if (typeNameBuilder.size() > 1) { return this->reportError("expected one name only"); } string name = typeNameBuilder[0]; if (nullptr == fRoot) { fRoot = this->findBmhObject(markType, name); fRoot->fFileName = fFileName; fRoot->fName = name; fRoot->fNames.fName = name; fRoot->fNames.fParent = &fGlobalNames; definition = fRoot; } else { if (nullptr == fParent) { return this->reportError("expected parent"); } if (fParent == fRoot && hasEnd) { RootDefinition* rootParent = fRoot->rootParent(); if (rootParent) { fRoot = rootParent; } definition = fParent; } else { if (!hasEnd && fRoot->find(name, RootDefinition::AllowParens::kNo)) { return this->reportError("duplicate symbol"); } if (MarkType::kStruct == markType || MarkType::kClass == markType || MarkType::kEnumClass == markType) { // if class or struct, build fRoot hierarchy // and change isDefined to search all parents of fRoot SkASSERT(!hasEnd); RootDefinition* childRoot = new RootDefinition; (fRoot->fBranches)[name] = childRoot; childRoot->setRootParent(fRoot); childRoot->fFileName = fFileName; SkASSERT(MarkType::kSubtopic != fRoot->fMarkType && MarkType::kTopic != fRoot->fMarkType); childRoot->fNames.fName = name; childRoot->fNames.fParent = &fRoot->fNames; fRoot = childRoot; definition = fRoot; } else { definition = &fRoot->fLeaves[name]; } } } if (hasEnd) { Exemplary hasExample = Exemplary::kNo; bool hasExcluder = false; for (auto child : definition->fChildren) { if (MarkType::kExample == child->fMarkType) { hasExample = Exemplary::kYes; } hasExcluder |= MarkType::kNoExample == child->fMarkType; } if (kMarkProps[(int) markType].fExemplary != hasExample && kMarkProps[(int) markType].fExemplary != Exemplary::kOptional) { if (string::npos == fFileName.find("undocumented") && !hasExcluder) { hasExample == Exemplary::kNo ? this->reportWarning("missing example") : this->reportWarning("unexpected example"); } } if (MarkType::kMethod == markType) { if (fCheckMethods && !definition->checkMethod()) { return false; } } if (HasTag::kYes == hasTag) { if (!this->checkEndMarker(markType, definition->fName)) { return false; } } if (!this->popParentStack(definition)) { return false; } if (fRoot == definition) { fRoot = nullptr; } } else { definition->fStart = defStart; this->skipSpace(); definition->fFileName = fFileName; definition->fContentStart = fChar; definition->fLineCount = fLineCount; definition->fClone = fCloned; if (MarkType::kConst == markType) { // todo: require that fChar points to def on same line as markup // additionally add definition to class children if it is not already there if (definition->fParent != fRoot) { // fRoot->fChildren.push_back(definition); } } SkASSERT(string::npos == name.find('\n')); definition->fName = name; if (MarkType::kMethod == markType) { if (string::npos != name.find(':', 0)) { definition->setCanonicalFiddle(); } else { definition->fFiddle = name; } } else { definition->fFiddle = Definition::NormalizedName(name); } definition->fMarkType = markType; definition->fAnonymous = fAnonymous; this->setAsParent(definition); } } break; case MarkType::kTopic: case MarkType::kSubtopic: SkASSERT(1 == typeNameBuilder.size()); if (!hasEnd) { if (!typeNameBuilder.size()) { return this->reportError("unnamed topic"); } fTopics.emplace_front(markType, defStart, fLineCount, fParent, fMC); RootDefinition* rootDefinition = &fTopics.front(); definition = rootDefinition; definition->fFileName = fFileName; definition->fContentStart = fChar; if (MarkType::kTopic == markType) { if (fParent) { return this->reportError("#Topic must be root"); } // topic name is unappended definition->fName = typeNameBuilder[0]; } else { if (!fParent) { return this->reportError("#Subtopic may not be root"); } Definition* parent = fParent; while (MarkType::kTopic != parent->fMarkType && MarkType::kSubtopic != parent->fMarkType) { parent = parent->fParent; if (!parent) { // subtopic must have subtopic or topic in parent chain return this->reportError("#Subtopic missing parent"); } } if (MarkType::kSubtopic == parent->fMarkType) { // subtopic prepends parent subtopic name, but not parent topic name definition->fName = parent->fName + '_'; } definition->fName += typeNameBuilder[0]; definition->fFiddle = parent->fFiddle + '_'; } rootDefinition->fNames.fName = rootDefinition->fName; rootDefinition->fNames.fParent = &fGlobalNames; definition->fFiddle += Definition::NormalizedName(typeNameBuilder[0]); this->setAsParent(definition); } { SkASSERT(hasEnd ? fParent : definition); string fullTopic = hasEnd ? fParent->fFiddle : definition->fFiddle; Definition* defPtr = fTopicMap[fullTopic]; if (hasEnd) { if (HasTag::kYes == hasTag && !this->checkEndMarker(markType, fullTopic)) { return false; } if (!definition) { definition = defPtr; } else if (definition != defPtr) { return this->reportError("mismatched topic"); } } else { if (nullptr != defPtr) { return this->reportError("already declared topic"); } fTopicMap[fullTopic] = definition; } } if (hasEnd) { if (!this->popParentStack(definition)) { return false; } } break; case MarkType::kFormula: // hasEnd : single line / multiple line if (!fParent || MarkType::kFormula != fParent->fMarkType) { SkASSERT(!definition || MarkType::kFormula == definition->fMarkType); fMarkup.emplace_front(markType, defStart, fLineCount, fParent, fMC); definition = &fMarkup.front(); definition->fContentStart = fChar; definition->fName = typeNameBuilder[0]; definition->fFiddle = fParent->fFiddle; fParent = definition; } else { SkASSERT(fParent && MarkType::kFormula == fParent->fMarkType); SkASSERT(fMC == defStart[0]); SkASSERT(fMC == defStart[-1]); definition = fParent; definition->fTerminator = fChar; if (!this->popParentStack(definition)) { return false; } this->parseHashFormula(definition); fParent->fChildren.push_back(definition); } break; // these types are children of parents, but are not in named maps case MarkType::kDescription: case MarkType::kStdOut: // may be one-liner case MarkType::kAlias: case MarkType::kNoExample: case MarkType::kParam: case MarkType::kPhraseDef: case MarkType::kReturn: case MarkType::kToDo: if (hasEnd) { if (markType == fParent->fMarkType) { definition = fParent; if (MarkType::kBug == markType || MarkType::kReturn == markType || MarkType::kToDo == markType) { this->skipNoName(); } if (!this->popParentStack(fParent)) { // if not one liner, pop return false; } if (MarkType::kParam == markType || MarkType::kReturn == markType || MarkType::kPhraseDef == markType) { if (!this->checkParamReturn(definition)) { return false; } } if (MarkType::kPhraseDef == markType) { string key = definition->fName; if (fPhraseMap.end() != fPhraseMap.find(key)) { this->reportError("duplicate phrase key"); } fPhraseMap[key] = definition; } } else { fMarkup.emplace_front(markType, defStart, fLineCount, fParent, fMC); definition = &fMarkup.front(); definition->fName = typeNameBuilder[0]; definition->fFiddle = fParent->fFiddle; definition->fContentStart = fChar; string endBracket; endBracket += fMC; endBracket += fMC; definition->fContentEnd = this->trimmedBracketEnd(endBracket); this->skipToEndBracket(endBracket.c_str()); SkAssertResult(fMC == this->next()); SkAssertResult(fMC == this->next()); definition->fTerminator = fChar; TextParser checkForChildren(definition); if (checkForChildren.strnchr(fMC, definition->fContentEnd)) { this->reportError("put ## on separate line"); } fParent->fChildren.push_back(definition); } if (MarkType::kAlias == markType) { const char* end = definition->fChildren.size() > 0 ? definition->fChildren[0]->fStart : definition->fContentEnd; TextParser parser(definition->fFileName, definition->fContentStart, end, definition->fLineCount); parser.trimEnd(); string key = string(parser.fStart, parser.lineLength()); if (fAliasMap.end() != fAliasMap.find(key)) { return this->reportError("duplicate alias"); } fAliasMap[key] = definition; definition->fFiddle = definition->fParent->fFiddle; } break; } else if (MarkType::kPhraseDef == markType) { bool hasParams = '(' == this->next(); fMarkup.emplace_front(markType, defStart, fLineCount, fParent, fMC); definition = &fMarkup.front(); definition->fName = typeNameBuilder[0]; definition->fFiddle = fParent->fFiddle; definition->fContentStart = fChar; if (hasParams) { char lastChar; do { const char* subEnd = this->anyOf(",)\n"); if (!subEnd || '\n' == *subEnd) { return this->reportError("unexpected phrase list end"); } fMarkup.emplace_front(MarkType::kPhraseParam, fChar, fLineCount, fParent, fMC); Definition* phraseParam = &fMarkup.front(); phraseParam->fContentStart = fChar; phraseParam->fContentEnd = subEnd; phraseParam->fName = string(fChar, subEnd - fChar); definition->fChildren.push_back(phraseParam); this->skipTo(subEnd); lastChar = this->next(); phraseParam->fTerminator = fChar; } while (')' != lastChar); this->skipWhiteSpace(); definition->fContentStart = fChar; } this->setAsParent(definition); break; } // not one-liners case MarkType::kCode: case MarkType::kExample: case MarkType::kFile: case MarkType::kFunction: case MarkType::kLegend: case MarkType::kList: case MarkType::kTable: if (hasEnd) { definition = fParent; if (markType != fParent->fMarkType) { return this->reportError("end element mismatch"); } else if (!this->popParentStack(fParent)) { return false; } if (MarkType::kExample == markType) { if (definition->fChildren.size() == 0) { TextParser emptyCheck(definition); if (emptyCheck.eof() || !emptyCheck.skipWhiteSpace()) { return this->reportError("missing example body"); } } // can't do this here; phrase refs may not have been defined yet // this->setWrapper(definition); } } else { fMarkup.emplace_front(markType, defStart, fLineCount, fParent, fMC); definition = &fMarkup.front(); definition->fContentStart = fChar; definition->fName = typeNameBuilder[0]; definition->fFiddle = fParent->fFiddle; char suffix = '\0'; bool tryAgain; do { tryAgain = false; for (const auto& child : fParent->fChildren) { if (child->fFiddle == definition->fFiddle) { if (MarkType::kExample != child->fMarkType) { continue; } if ('\0' == suffix) { suffix = 'a'; } else if (++suffix > 'z') { return reportError("too many examples"); } definition->fFiddle = fParent->fFiddle + '_'; definition->fFiddle += suffix; tryAgain = true; break; } } } while (tryAgain); this->setAsParent(definition); } break; // always treated as one-liners (can't detect misuse easily) case MarkType::kAnchor: case MarkType::kBug: case MarkType::kDetails: case MarkType::kDuration: case MarkType::kFilter: case MarkType::kHeight: case MarkType::kIllustration: case MarkType::kImage: case MarkType::kIn: case MarkType::kLine: case MarkType::kLiteral: case MarkType::kNoJustify: case MarkType::kOutdent: case MarkType::kPlatform: case MarkType::kPopulate: case MarkType::kSeeAlso: case MarkType::kSet: case MarkType::kSubstitute: case MarkType::kVolatile: case MarkType::kWidth: // todo : add check disallowing children? if (hasEnd && MarkType::kAnchor != markType && MarkType::kLine != markType) { return this->reportError("one liners omit end element"); } else if (!hasEnd && MarkType::kAnchor == markType) { return this->reportError("anchor line must have end element last"); } fMarkup.emplace_front(markType, defStart, fLineCount, fParent, fMC); definition = &fMarkup.front(); definition->fName = typeNameBuilder[0]; definition->fFiddle = Definition::NormalizedName(typeNameBuilder[0]); definition->fContentStart = fChar; definition->fContentEnd = this->trimmedBracketEnd('\n'); definition->fTerminator = this->lineEnd() - 1; fParent->fChildren.push_back(definition); if (MarkType::kAnchor == markType) { this->parseHashAnchor(definition); } else if (MarkType::kLine == markType) { this->parseHashLine(definition); } break; case MarkType::kExternal: (void) this->collectExternals(); // FIXME: detect errors in external defs? break; default: SkASSERT(0); // fixme : don't let any types be invisible return true; } if (fParent) { SkASSERT(definition); SkASSERT(definition->fName.length() > 0); } return true; } void BmhParser::reportDuplicates(const Definition& def, string dup) const { if (MarkType::kExample == def.fMarkType && dup == def.fFiddle) { TextParser reporter(&def); reporter.reportError("duplicate example name"); } for (auto& child : def.fChildren ) { reportDuplicates(*child, dup); } } static Definition* find_fiddle(Definition* def, string name) { if (MarkType::kExample == def->fMarkType && name == def->fFiddle) { return def; } for (auto& child : def->fChildren) { Definition* result = find_fiddle(child, name); if (result) { return result; } } return nullptr; } Definition* BmhParser::findExample(string name) const { for (const auto& topic : fTopicMap) { if (topic.second->fParent) { continue; } Definition* def = find_fiddle(topic.second, name); if (def) { return def; } } return nullptr; } static bool check_example_hashes(Definition* def) { if (MarkType::kExample == def->fMarkType) { if (def->fHash.length()) { return true; } for (auto child : def->fChildren) { if (MarkType::kPlatform == child->fMarkType) { if (string::npos != string(child->fContentStart, child->length()).find("!fiddle")) { return true; } } } return def->reportError("missing hash"); } for (auto& child : def->fChildren) { if (!check_example_hashes(child)) { return false; } } return true; } bool BmhParser::checkExampleHashes() const { for (const auto& topic : fTopicMap) { if (!topic.second->fParent && !check_example_hashes(topic.second)) { return false; } } return true; } static void reset_example_hashes(Definition* def) { if (MarkType::kExample == def->fMarkType) { def->fHash.clear(); return; } for (auto& child : def->fChildren) { reset_example_hashes(child); } } void BmhParser::resetExampleHashes() { for (const auto& topic : fTopicMap) { if (!topic.second->fParent) { reset_example_hashes(topic.second); } } } static void find_examples(const Definition& def, vector* exampleNames) { if (MarkType::kExample == def.fMarkType) { exampleNames->push_back(def.fFiddle); } for (auto& child : def.fChildren ) { find_examples(*child, exampleNames); } } bool BmhParser::checkEndMarker(MarkType markType, string match) const { TextParser tp(fFileName, fLine, fChar, fLineCount); tp.skipSpace(); if (fMC != tp.next()) { return this->reportError("mismatched end marker expect #"); } const char* nameStart = tp.fChar; tp.skipToNonName(); string markName(nameStart, tp.fChar - nameStart); if (kMarkProps[(int) markType].fName != markName) { return this->reportError("expected #XXX ## to match"); } tp.skipSpace(); nameStart = tp.fChar; tp.skipToNonName(); markName = string(nameStart, tp.fChar - nameStart); if ("" == markName) { if (fMC != tp.next() || fMC != tp.next()) { return this->reportError("expected ##"); } return true; } std::replace(markName.begin(), markName.end(), '-', '_'); auto defPos = match.rfind(markName); if (string::npos == defPos) { return this->reportError("mismatched end marker v1"); } if (markName.size() != match.size() - defPos) { return this->reportError("mismatched end marker v2"); } return true; } bool BmhParser::checkExamples() const { vector exampleNames; for (const auto& topic : fTopicMap) { if (topic.second->fParent) { continue; } find_examples(*topic.second, &exampleNames); } std::sort(exampleNames.begin(), exampleNames.end()); string* last = nullptr; string reported; bool checkOK = true; for (auto& nameIter : exampleNames) { if (last && *last == nameIter && reported != *last) { reported = *last; SkDebugf("%s\n", reported.c_str()); for (const auto& topic : fTopicMap) { if (topic.second->fParent) { continue; } this->reportDuplicates(*topic.second, reported); } checkOK = false; } last = &nameIter; } return checkOK; } bool BmhParser::checkParamReturn(const Definition* definition) const { const char* parmEndCheck = definition->fContentEnd; while (parmEndCheck < definition->fTerminator) { if (fMC == parmEndCheck[0]) { break; } if (' ' < parmEndCheck[0]) { this->reportError( "use full end marker on multiline #Param and #Return"); } ++parmEndCheck; } return true; } bool BmhParser::childOf(MarkType markType) const { auto childError = [this](MarkType markType) -> bool { string errStr = "expected "; errStr += kMarkProps[(int) markType].fName; errStr += " parent"; return this->reportError(errStr.c_str()); }; if (markType == fParent->fMarkType) { return true; } if (this->hasEndToken()) { if (!fParent->fParent) { return this->reportError("expected grandparent"); } if (markType == fParent->fParent->fMarkType) { return true; } } return childError(markType); } string BmhParser::className(MarkType markType) { const char* end = this->lineEnd(); const char* mc = this->strnchr(fMC, end); string classID; TextParserSave savePlace(this); this->skipSpace(); const char* wordStart = fChar; this->skipToNonName(); const char* wordEnd = fChar; classID = string(wordStart, wordEnd - wordStart); if (!mc) { savePlace.restore(); } string builder; const Definition* parent = this->parentSpace(); if (parent && parent->fName != classID) { builder += parent->fName; } if (mc) { if (mc + 1 < fEnd && fMC == mc[1]) { // if ## if (markType != fParent->fMarkType) { return this->reportError("unbalanced method"); } if (builder.length() > 0 && classID.size() > 0) { if (builder != fParent->fName) { builder += "::"; builder += classID; if (builder != fParent->fName) { return this->reportError("name mismatch"); } } } this->skipLine(); return fParent->fName; } else if (' ' == mc[1] && MarkType::kConst == markType && fParent && (MarkType::kEnum == fParent->fMarkType || MarkType::kEnumClass == fParent->fMarkType)) { this->skipToEndBracket('\n'); return builder + "::" + string(wordStart, wordEnd - wordStart); } fChar = mc; this->next(); } this->skipWhiteSpace(); if (MarkType::kEnum == markType && fChar >= end) { fAnonymous = true; builder += "::_anonymous"; return uniqueRootName(builder, markType); } builder = this->word(builder, "::"); return builder; } bool BmhParser::collectExternals() { do { this->skipWhiteSpace(); if (this->eof()) { break; } if (fMC == this->peek()) { this->next(); if (this->eof()) { break; } if (fMC == this->peek()) { this->skipLine(); break; } if (' ' >= this->peek()) { this->skipLine(); continue; } if (this->startsWith(kMarkProps[(int) MarkType::kExternal].fName)) { this->skipToNonName(); continue; } } this->skipToAlpha(); const char* wordStart = fChar; this->skipToWhiteSpace(); if (fChar - wordStart > 0) { fExternals.emplace_front(MarkType::kExternal, wordStart, fChar, fLineCount, fParent, fMC); RootDefinition* definition = &fExternals.front(); definition->fFileName = fFileName; definition->fName = string(wordStart ,fChar - wordStart); definition->fFiddle = Definition::NormalizedName(definition->fName); } } while (!this->eof()); return true; } bool BmhParser::dumpExamples(FILE* fiddleOut, Definition& def, bool* continuation) const { if (MarkType::kExample == def.fMarkType) { string result; if (!this->exampleToScript(&def, BmhParser::ExampleOptions::kAll, &result)) { return false; } if (result.length() > 0) { result += "\n"; result += "}"; if (*continuation) { fprintf(fiddleOut, ",\n"); } else { *continuation = true; } fprintf(fiddleOut, "%s", result.c_str()); } return true; } for (auto& child : def.fChildren ) { if (!this->dumpExamples(fiddleOut, *child, continuation)) { return false; } } return true; } bool BmhParser::dumpExamples(const char* fiddleJsonFileName) const { string oldFiddle(fiddleJsonFileName); string newFiddle(fiddleJsonFileName); newFiddle += "_new"; FILE* fiddleOut = fopen(newFiddle.c_str(), "wb"); if (!fiddleOut) { SkDebugf("could not open output file %s\n", newFiddle.c_str()); return false; } fprintf(fiddleOut, "{\n"); bool continuation = false; for (const auto& topic : fTopicMap) { if (topic.second->fParent) { continue; } this->dumpExamples(fiddleOut, *topic.second, &continuation); } fprintf(fiddleOut, "\n}\n"); fclose(fiddleOut); if (ParserCommon::WrittenFileDiffers(oldFiddle, newFiddle)) { ParserCommon::CopyToFile(oldFiddle, newFiddle); SkDebugf("wrote %s\n", fiddleJsonFileName); } else { remove(newFiddle.c_str()); } return true; } int BmhParser::endHashCount() const { const char* end = fLine + this->lineLength(); int count = 0; while (fLine < end && fMC == *--end) { count++; } return count; } bool BmhParser::endTableColumn(const char* end, const char* terminator) { if (!this->popParentStack(fParent)) { return false; } fWorkingColumn->fContentEnd = end; fWorkingColumn->fTerminator = terminator; fColStart = fChar - 1; this->skipSpace(); fTableState = TableState::kColumnStart; return true; } static size_t count_indent(string text, size_t test, size_t end) { size_t result = test; while (test < end) { if (' ' != text[test]) { break; } ++test; } return test - result; } static void add_code(string text, int pos, int end, size_t outIndent, size_t textIndent, string& example) { do { // fix this to move whole paragraph in, out, but preserve doc indent int nextIndent = count_indent(text, pos, end); size_t len = text.find('\n', pos); if (string::npos == len) { len = end; } if ((size_t) (pos + nextIndent) < len) { size_t indent = outIndent + nextIndent; SkASSERT(indent >= textIndent); indent -= textIndent; for (size_t index = 0; index < indent; ++index) { example += ' '; } pos += nextIndent; while ((size_t) pos < len) { example += '"' == text[pos] ? "\\\"" : '\\' == text[pos] ? "\\\\" : text.substr(pos, 1); ++pos; } example += "\\n"; } else { pos += nextIndent; } if ('\n' == text[pos]) { ++pos; } } while (pos < end); } bool BmhParser::IsExemplary(const Definition* def) { return kMarkProps[(int) def->fMarkType].fExemplary != Exemplary::kNo; } bool BmhParser::exampleToScript(Definition* def, ExampleOptions exampleOptions, string* result) const { bool hasFiddle = true; const Definition* platform = def->hasChild(MarkType::kPlatform); if (platform) { TextParser platParse(platform); hasFiddle = !platParse.strnstr("!fiddle", platParse.fEnd); } if (!hasFiddle) { *result = ""; return true; } string text = this->extractText(def, TrimExtract::kNo); bool textOut = string::npos != text.find("SkDebugf(") || string::npos != text.find("dump(") || string::npos != text.find("dumpHex("); string heightStr = "256"; string widthStr = "256"; string normalizedName(def->fFiddle); string code; string imageStr = "0"; string srgbStr = "false"; string durationStr = "0"; for (auto iter : def->fChildren) { switch (iter->fMarkType) { case MarkType::kDuration: durationStr = string(iter->fContentStart, iter->fContentEnd - iter->fContentStart); break; case MarkType::kHeight: heightStr = string(iter->fContentStart, iter->fContentEnd - iter->fContentStart); break; case MarkType::kWidth: widthStr = string(iter->fContentStart, iter->fContentEnd - iter->fContentStart); break; case MarkType::kDescription: // ignore for now break; case MarkType::kFunction: { // emit this, but don't wrap this in draw() string funcText = this->extractText(&*iter, TrimExtract::kNo); size_t pos = 0; while (pos < funcText.length() && ' ' > funcText[pos]) { ++pos; } size_t indent = count_indent(funcText, pos, funcText.length()); add_code(funcText, pos, funcText.length(), 0, indent, code); code += "\\n"; } break; case MarkType::kComment: break; case MarkType::kImage: imageStr = string(iter->fContentStart, iter->fContentEnd - iter->fContentStart); break; case MarkType::kToDo: break; case MarkType::kBug: case MarkType::kMarkChar: case MarkType::kPlatform: case MarkType::kPhraseRef: // ignore for now break; case MarkType::kSet: if ("sRGB" == string(iter->fContentStart, iter->fContentEnd - iter->fContentStart)) { srgbStr = "true"; } else { SkASSERT(0); // more work to do return false; } break; case MarkType::kStdOut: textOut = true; break; default: SkASSERT(0); // more coding to do } } string animatedStr = "0" != durationStr ? "true" : "false"; string textOutStr = textOut ? "true" : "false"; size_t pos = 0; while (pos < text.length() && ' ' > text[pos]) { ++pos; } size_t end = text.length(); size_t outIndent = 0; size_t textIndent = count_indent(text, pos, end); if ("" == def->fWrapper) { this->setWrapper(def); } if (def->fWrapper.length() > 0) { code += def->fWrapper; code += "\\n"; outIndent = 4; } add_code(text, pos, end, outIndent, textIndent, code); if (def->fWrapper.length() > 0) { code += "}"; } string example = "\"" + normalizedName + "\": {\n"; string filename = def->fileName(); string baseFile = filename.substr(0, filename.length() - 4); if (ExampleOptions::kText == exampleOptions) { example += " \"code\": \"" + code + "\",\n"; example += " \"hash\": \"" + def->fHash + "\",\n"; example += " \"file\": \"" + baseFile + "\",\n"; example += " \"name\": \"" + def->fName + "\","; } else { example += " \"code\": \"" + code + "\",\n"; if (ExampleOptions::kPng == exampleOptions) { example += " \"width\": " + widthStr + ",\n"; example += " \"height\": " + heightStr + ",\n"; example += " \"hash\": \"" + def->fHash + "\",\n"; example += " \"file\": \"" + baseFile + "\",\n"; example += " \"name\": \"" + def->fName + "\"\n"; example += "}"; } else { example += " \"options\": {\n"; example += " \"width\": " + widthStr + ",\n"; example += " \"height\": " + heightStr + ",\n"; example += " \"source\": " + imageStr + ",\n"; example += " \"srgb\": " + srgbStr + ",\n"; example += " \"f16\": false,\n"; example += " \"textOnly\": " + textOutStr + ",\n"; example += " \"animated\": " + animatedStr + ",\n"; example += " \"duration\": " + durationStr + "\n"; example += " },\n"; example += " \"fast\": true"; } } *result = example; return true; } string BmhParser::extractText(const Definition* def, TrimExtract trimExtract) const { string result; TextParser parser(def); auto childIter = def->fChildren.begin(); while (!parser.eof()) { const char* end = def->fChildren.end() == childIter ? parser.fEnd : (*childIter)->fStart; string fragment(parser.fChar, end - parser.fChar); trim_end(fragment); if (TrimExtract::kYes == trimExtract) { trim_start(fragment); if (result.length()) { result += '\n'; result += '\n'; } } if (TrimExtract::kYes == trimExtract || has_nonwhitespace(fragment)) { result += fragment; } parser.skipTo(end); if (def->fChildren.end() != childIter) { Definition* child = *childIter; if (MarkType::kPhraseRef == child->fMarkType) { auto phraseIter = fPhraseMap.find(child->fName); if (fPhraseMap.end() == phraseIter) { return def->reportError("missing phrase definition"); } Definition* phrase = phraseIter->second; // count indent of last line in result size_t lastLF = result.rfind('\n'); size_t startPos = string::npos == lastLF ? 0 : lastLF; size_t lastLen = result.length() - startPos; size_t indent = count_indent(result, startPos, result.length()) + 4; string phraseStr = this->extractText(phrase, TrimExtract::kNo); startPos = 0; bool firstTime = true; size_t endPos; do { endPos = phraseStr.find('\n', startPos); size_t len = (string::npos != endPos ? endPos : phraseStr.length()) - startPos; if (firstTime && lastLen + len + 1 < 100) { // FIXME: make 100 global const or something result += ' '; } else { result += '\n'; result += string(indent, ' '); } firstTime = false; string tmp = phraseStr.substr(startPos, len); result += tmp; startPos = endPos + 1; } while (string::npos != endPos); result += '\n'; } parser.skipTo(child->fTerminator); std::advance(childIter, 1); } } return result; } string BmhParser::loweredTopic(string name, Definition* def) { string lowered; SkASSERT('_' != name[0]); char last = '_'; for (char c : name) { SkASSERT(' ' != c); if (isupper(last)) { lowered += islower(c) ? tolower(last) : last; last = '\0'; } if ('_' == c) { last = c; c = ' '; } else if ('_' == last && isupper(c)) { last = c; continue; } lowered += c; if (' ' == c) { this->setUpPartialSubstitute(lowered); } } if (isupper(last)) { lowered += tolower(last); } return lowered; } void BmhParser::setUpGlobalSubstitutes() { for (auto& entry : fExternals) { string externalName = entry.fName; SkASSERT(fGlobalNames.fRefMap.end() == fGlobalNames.fRefMap.find(externalName)); fGlobalNames.fRefMap[externalName] = nullptr; } for (auto bMap : { &fClassMap, &fConstMap, &fDefineMap, &fEnumMap, &fMethodMap, &fTypedefMap } ) { for (auto& entry : *bMap) { Definition* parent = &entry.second; string name = parent->fName; SkASSERT(fGlobalNames.fLinkMap.end() == fGlobalNames.fLinkMap.find(name)); string ref = ParserCommon::HtmlFileName(parent->fFileName) + '#' + parent->fFiddle; fGlobalNames.fLinkMap[name] = ref; SkASSERT(fGlobalNames.fRefMap.end() == fGlobalNames.fRefMap.find(name)); fGlobalNames.fRefMap[name] = parent; NameMap* names = MarkType::kClass == parent->fMarkType || MarkType::kStruct == parent->fMarkType || MarkType::kEnumClass == parent->fMarkType ? &parent->asRoot()->fNames : &fGlobalNames; this->setUpSubstitutes(parent, names); if (names != &fGlobalNames) { names->copyToParent(&fGlobalNames); } } } for (auto& tEntry : fTypedefMap) { Definition* typeDef = &tEntry.second; string defName = typeDef->fName; TextParser parser(typeDef->fFileName, typeDef->fStart, typeDef->fContentStart, typeDef->fLineCount); parser.skipExact("#Typedef"); parser.skipWhiteSpace(); const char* refStart = parser.fChar; parser.skipToWhiteSpace(); string refName(refStart, parser.fChar - refStart); auto structIter = fClassMap.find(refName); if (fClassMap.end() == structIter) { continue; } for (auto& sEntry : structIter->second.fLeaves) { Definition* def = &sEntry.second; string name = def->fName; size_t colonPos = name.find("::"); SkASSERT(string::npos != colonPos); string typedName = defName + "::" + name.substr(colonPos + 2); string ref = ParserCommon::HtmlFileName(def->fFileName) + '#' + def->fFiddle; SkASSERT(fGlobalNames.fLinkMap.end() == fGlobalNames.fLinkMap.find(typedName)); fGlobalNames.fLinkMap[typedName] = ref; SkASSERT(fGlobalNames.fRefMap.end() == fGlobalNames.fRefMap.find(typedName)); fGlobalNames.fRefMap[typedName] = def; } } for (auto& topic : fTopicMap) { bool hasSubstitute = false; for (auto& child : topic.second->fChildren) { bool isAlias = MarkType::kAlias == child->fMarkType; bool isSubstitute = MarkType::kSubstitute == child->fMarkType; if (!isAlias && !isSubstitute) { continue; } hasSubstitute |= isSubstitute; string name(child->fContentStart, child->length()); if (isAlias) { name = ParserCommon::ConvertRef(name, false); for (auto aliasChild : child->fChildren) { if (MarkType::kSubstitute == aliasChild->fMarkType) { string sub(aliasChild->fContentStart, aliasChild->length()); this->setUpSubstitute(sub, topic.second); } } } this->setUpSubstitute(name, topic.second); } if (hasSubstitute) { continue; } string lowered = this->loweredTopic(topic.first, topic.second); SkDEBUGCODE(auto globalIter = fGlobalNames.fLinkMap.find(lowered)); SkASSERT(fGlobalNames.fLinkMap.end() == globalIter); fGlobalNames.fLinkMap[lowered] = ParserCommon::HtmlFileName(topic.second->fFileName) + '#' + topic.first; SkASSERT(fGlobalNames.fRefMap.end() == fGlobalNames.fRefMap.find(lowered)); fGlobalNames.fRefMap[lowered] = topic.second; } size_t slash = fRawFilePathDir.rfind('/'); size_t bslash = fRawFilePathDir.rfind('\\'); string spellFile; if (string::npos == slash && string::npos == bslash) { spellFile = fRawFilePathDir; } else { if (string::npos != bslash && bslash > slash) { slash = bslash; } spellFile = fRawFilePathDir.substr(0, slash); } spellFile += '/'; spellFile += kSpellingFileName; FILE* file = fopen(spellFile.c_str(), "r"); if (!file) { SkDebugf("missing %s\n", spellFile.c_str()); return; } fseek(file, 0L, SEEK_END); int sz = (int) ftell(file); rewind(file); char* buffer = new char[sz]; memset(buffer, ' ', sz); int read = (int)fread(buffer, 1, sz, file); SkAssertResult(read > 0); sz = read; // FIXME: ? why are sz and read different? fclose(file); int i = 0; int start = i; string last = " "; string word; do { if (' ' < buffer[i]) { ++i; continue; } last = word; word = string(&buffer[start], i - start); #ifdef SK_DEBUG SkASSERT(last.compare(word) < 0); for (char c : word) { SkASSERT(islower(c) || '-' == c); } #endif if (fGlobalNames.fRefMap.end() == fGlobalNames.fRefMap.find(word)) { fGlobalNames.fRefMap[word] = nullptr; } else { SkDebugf("%s ", word.c_str()); // debugging: word missing from spelling list } do { ++i; } while (i < sz && ' ' >= buffer[i]); start = i; } while (i < sz); delete[] buffer; } void BmhParser::setUpSubstitutes(const Definition* parent, NameMap* names) { for (const auto& child : parent->fChildren) { MarkType markType = child->fMarkType; if (MarkType::kAlias == markType) { continue; } if (MarkType::kSubstitute == markType) { continue; } string name(child->fName); if (&fGlobalNames != names) { size_t lastDC = name.rfind("::"); if (string::npos != lastDC) { name = name.substr(lastDC + 2); } if ("" == name) { continue; } } string ref; if (&fGlobalNames == names) { ref = ParserCommon::HtmlFileName(child->fFileName); } ref += '#' + child->fFiddle; if (MarkType::kClass == markType || MarkType::kStruct == markType || (MarkType::kMethod == markType && !child->fClone) || MarkType::kEnum == markType || MarkType::kEnumClass == markType || MarkType::kConst == markType || MarkType::kMember == markType || MarkType::kDefine == markType || MarkType::kTypedef == markType) { SkASSERT(names->fLinkMap.end() == names->fLinkMap.find(name)); names->fLinkMap[name] = ref; SkASSERT(names->fRefMap.end() == names->fRefMap.find(name)); names->fRefMap[name] = child; } if (MarkType::kClass == markType || MarkType::kStruct == markType || MarkType::kEnumClass == markType) { RootDefinition* rootDef = child->asRoot(); NameMap* nameMap = &rootDef->fNames; this->setUpSubstitutes(child, nameMap); nameMap->copyToParent(names); } if (MarkType::kEnum == markType) { this->setUpSubstitutes(child, names); } if (MarkType::kMethod == markType) { if (child->fClone || child->fCloned) { TextParser parser(child->fFileName, child->fStart, child->fContentStart, child->fLineCount); parser.skipExact("#Method"); parser.skipSpace(); string name = child->methodName(); const char* nameInParser = parser.strnstr(name.c_str(), parser.fEnd); parser.skipTo(nameInParser); const char* paren = parser.strnchr('(', parser.fEnd); parser.skipTo(paren); parser.skipToBalancedEndBracket('(', ')'); if ("()" != string(paren, parser.fChar - paren)) { string fullName = trim_inline_spaces(string(nameInParser, parser.fChar - nameInParser)); SkASSERT(names->fLinkMap.end() == names->fLinkMap.find(fullName)); names->fLinkMap[fullName] = ref; SkASSERT(names->fRefMap.end() == names->fRefMap.find(fullName)); names->fRefMap[fullName] = child; } } } if (MarkType::kSubtopic == markType) { if (&fGlobalNames != names && string::npos != child->fName.find('_')) { string lowered = this->loweredTopic(child->fName, child); SkDEBUGCODE(auto refIter = names->fRefMap.find(lowered)); SkDEBUGCODE(auto iter = names->fLinkMap.find(lowered)); SkASSERT(names->fLinkMap.end() == iter); names->fLinkMap[lowered] = '#' + child->fName; SkASSERT(names->fRefMap.end() == refIter); names->fRefMap[lowered] = child; } this->setUpSubstitutes(child, names); } } } void BmhParser::setUpPartialSubstitute(string name) { auto iter = fGlobalNames.fRefMap.find(name); if (fGlobalNames.fRefMap.end() != iter) { SkASSERT(nullptr == iter->second); return; } fGlobalNames.fRefMap[name] = nullptr; } void BmhParser::setUpSubstitute(string name, Definition* def) { SkASSERT(fGlobalNames.fRefMap.end() == fGlobalNames.fRefMap.find(name)); fGlobalNames.fRefMap[name] = def; SkASSERT(fGlobalNames.fLinkMap.end() == fGlobalNames.fLinkMap.find(name)); string str = ParserCommon::HtmlFileName(def->fFileName) + '#' + def->fName; fGlobalNames.fLinkMap[name] = str; size_t stop = name.length(); do { size_t space = name.rfind(' ', stop); if (string::npos == space) { break; } string partial = name.substr(0, space + 1); stop = space - 1; this->setUpPartialSubstitute(partial); } while (true); } void BmhParser::setWrapper(Definition* def) const { const char drawWrapper[] = "void draw(SkCanvas* canvas) {"; const char drawNoCanvas[] = "void draw(SkCanvas* ) {"; string text = this->extractText(def, TrimExtract::kNo); size_t nonSpace = 0; while (nonSpace < text.length() && ' ' >= text[nonSpace]) { ++nonSpace; } bool hasFunc = !text.compare(nonSpace, sizeof(drawWrapper) - 1, drawWrapper); bool noCanvas = !text.compare(nonSpace, sizeof(drawNoCanvas) - 1, drawNoCanvas); bool hasCanvas = string::npos != text.find("SkCanvas canvas"); SkASSERT(!hasFunc || !noCanvas); bool preprocessor = text[0] == '#'; bool wrapCode = !hasFunc && !noCanvas && !preprocessor; if (wrapCode) { def->fWrapper = hasCanvas ? string(drawNoCanvas) : string(drawWrapper); } } RootDefinition* BmhParser::findBmhObject(MarkType markType, string typeName) { const auto& mapIter = std::find_if(fMaps.begin(), fMaps.end(), [markType](DefinitionMap& defMap){ return markType == defMap.fMarkType; } ); if (mapIter == fMaps.end()) { return nullptr; } return &(*mapIter->fMap)[typeName]; } // FIXME: some examples may produce different output on different platforms // if the text output can be different, think of how to author that bool BmhParser::findDefinitions() { bool lineStart = true; const char* lastChar = nullptr; const char* lastMC = nullptr; fParent = nullptr; while (!this->eof()) { if (this->peek() == fMC) { lastMC = fChar; this->next(); if (this->peek() == fMC) { this->next(); if (!lineStart && ' ' < this->peek()) { if (!fParent || MarkType::kFormula != fParent->fMarkType) { return this->reportError("expected definition"); } } if (this->peek() != fMC) { if (MarkType::kColumn == fParent->fMarkType) { SkASSERT(TableState::kColumnEnd == fTableState); if (!this->endTableColumn(lastChar, lastMC)) { return false; } SkASSERT(fRow); if (!this->popParentStack(fParent)) { return false; } fRow->fContentEnd = fWorkingColumn->fContentEnd; fWorkingColumn = nullptr; fRow = nullptr; fTableState = TableState::kNone; } else { vector parentName; parentName.push_back(fParent->fName); if (!this->addDefinition(fChar - 1, true, fParent->fMarkType, parentName, HasTag::kNo)) { return false; } } } else { SkAssertResult(this->next() == fMC); fMC = this->next(); // change markup character if (' ' >= fMC) { return this->reportError("illegal markup character"); } fMarkup.emplace_front(MarkType::kMarkChar, fChar - 4, fLineCount, fParent, fMC); Definition* markChar = &fMarkup.front(); markChar->fContentStart = fChar - 1; this->skipToEndBracket('\n'); markChar->fContentEnd = fChar; markChar->fTerminator = fChar; fParent->fChildren.push_back(markChar); } } else if (this->peek() >= 'A' && this->peek() <= 'Z') { const char* defStart = fChar - 1; MarkType markType = this->getMarkType(MarkLookup::kRequire); bool hasEnd = this->hasEndToken(); if (!hasEnd && fParent) { MarkType parentType = fParent->fMarkType; uint64_t parentMask = kMarkProps[(int) markType].fParentMask; if (parentMask && !(parentMask & (1LL << (int) parentType))) { return this->reportError("invalid parent"); } } if (!this->skipName(kMarkProps[(int) markType].fName)) { return this->reportError("illegal markup character"); } if (!this->skipSpace()) { return this->reportError("unexpected end"); } lineStart = '\n' == this->peek(); bool expectEnd = true; vector typeNameBuilder = this->typeName(markType, &expectEnd); if (fCloned && MarkType::kMethod != markType && MarkType::kExample != markType && !fAnonymous) { return this->reportError("duplicate name"); } if (hasEnd && expectEnd) { if (fMC == this->peek()) { return this->reportError("missing body"); } } if (!this->addDefinition(defStart, hasEnd, markType, typeNameBuilder, HasTag::kYes)) { return false; } continue; } else if (this->peek() == ' ') { if (!fParent || (MarkType::kFormula != fParent->fMarkType && MarkType::kLegend != fParent->fMarkType && MarkType::kList != fParent->fMarkType && MarkType::kLine != fParent->fMarkType && MarkType::kTable != fParent->fMarkType)) { int endHashes = this->endHashCount(); if (endHashes <= 1) { if (fParent) { if (TableState::kColumnEnd == fTableState) { if (!this->endTableColumn(lastChar, lastMC)) { return false; } } else { // one line comment fMarkup.emplace_front(MarkType::kComment, fChar - 1, fLineCount, fParent, fMC); Definition* comment = &fMarkup.front(); comment->fContentStart = fChar - 1; this->skipToEndBracket('\n'); comment->fContentEnd = fChar; comment->fTerminator = fChar; fParent->fChildren.push_back(comment); } } else { fChar = fLine + this->lineLength() - 1; } } else { // table row if (2 != endHashes) { string errorStr = "expect "; errorStr += fMC; errorStr += fMC; return this->reportError(errorStr.c_str()); } if (!fParent || MarkType::kTable != fParent->fMarkType) { return this->reportError("missing table"); } } } else if (TableState::kNone == fTableState) { // fixme? no nested tables for now fColStart = fChar - 1; fMarkup.emplace_front(MarkType::kRow, fColStart, fLineCount, fParent, fMC); fRow = &fMarkup.front(); fRow->fName = fParent->fName; this->skipWhiteSpace(); fRow->fContentStart = fChar; this->setAsParent(fRow); fTableState = TableState::kColumnStart; } if (TableState::kColumnStart == fTableState) { fMarkup.emplace_front(MarkType::kColumn, fColStart, fLineCount, fParent, fMC); fWorkingColumn = &fMarkup.front(); fWorkingColumn->fName = fParent->fName; fWorkingColumn->fContentStart = fChar; this->setAsParent(fWorkingColumn); fTableState = TableState::kColumnEnd; continue; } } else if (this->peek() >= 'a' && this->peek() <= 'z') { // expect zero or more letters and underscores (no spaces) then hash const char* phraseNameStart = fChar; this->skipPhraseName(); string phraseKey = string(phraseNameStart, fChar - phraseNameStart); char delimiter = this->next(); vector params; vector paramsLoc; if (fMC != delimiter) { if ('(' != delimiter) { return this->reportError("expect # after phrase name"); } // phrase may take comma delimited parameter list do { const char* subEnd = this->anyOf(",)\n"); if (!subEnd || '\n' == *subEnd) { return this->reportError("unexpected phrase list end"); } params.push_back(string(fChar, subEnd - fChar)); paramsLoc.push_back(fChar); this->skipTo(subEnd); } while (')' != this->next()); } const char* start = phraseNameStart; SkASSERT('#' == start[-1]); --start; if (start > fStart && ' ' >= start[-1]) { --start; // preserve whether to add whitespace before substitution } fMarkup.emplace_front(MarkType::kPhraseRef, start, fLineCount, fParent, fMC); Definition* markChar = &fMarkup.front(); this->skipExact("#"); markChar->fContentStart = fChar; markChar->fContentEnd = fChar; markChar->fTerminator = fChar; markChar->fName = phraseKey; fParent->fChildren.push_back(markChar); int paramLocIndex = 0; for (auto param : params) { const char* paramLoc = paramsLoc[paramLocIndex++]; fMarkup.emplace_front(MarkType::kPhraseParam, paramLoc, fLineCount, fParent, fMC); Definition* phraseParam = &fMarkup.front(); phraseParam->fContentStart = paramLoc; phraseParam->fContentEnd = paramLoc + param.length(); phraseParam->fTerminator = paramLoc + param.length(); phraseParam->fName = param; markChar->fChildren.push_back(phraseParam); } } } char nextChar = this->next(); if (' ' < nextChar) { lastChar = fChar; lineStart = false; } else if (nextChar == '\n') { lineStart = true; } } if (fParent) { return fParent->reportError("mismatched end"); } return true; } MarkType BmhParser::getMarkType(MarkLookup lookup) const { for (int index = 0; index <= Last_MarkType; ++index) { int typeLen = strlen(kMarkProps[index].fName); if (typeLen == 0) { continue; } if (fChar + typeLen >= fEnd || fChar[typeLen] > ' ') { continue; } int chCompare = strncmp(fChar, kMarkProps[index].fName, typeLen); if (chCompare < 0) { goto fail; } if (chCompare == 0) { return (MarkType) index; } } fail: if (MarkLookup::kRequire == lookup) { return this->reportError("unknown mark type"); } return MarkType::kNone; } bool BmhParser::hasEndToken() const { const char* ptr = fLine; char test; do { if (ptr >= fEnd) { return false; } test = *ptr++; if ('\n' == test) { return false; } } while (fMC != test || fMC != *ptr); return true; } string BmhParser::memberName() { const char* wordStart; const char* prefixes[] = { "static", "const" }; do { this->skipSpace(); wordStart = fChar; this->skipToNonName(); } while (this->anyOf(wordStart, prefixes, SK_ARRAY_COUNT(prefixes))); if ('*' == this->peek()) { this->next(); } return this->className(MarkType::kMember); } string BmhParser::methodName() { if (this->hasEndToken()) { if (!fParent || !fParent->fName.length()) { return this->reportError("missing parent method name"); } SkASSERT(fMC == this->peek()); this->next(); SkASSERT(fMC == this->peek()); this->next(); SkASSERT(fMC != this->peek()); return fParent->fName; } string builder; const char* end = this->lineEnd(); const char* paren = this->strnchr('(', end); if (!paren) { return this->reportError("missing method name and reference"); } { TextParserSave endCheck(this); while (end < fEnd && !this->strnchr(')', end)) { fChar = end + 1; end = this->lineEnd(); } if (end >= fEnd) { return this->reportError("missing method end paren"); } endCheck.restore(); } const char* nameStart = paren; char ch; bool expectOperator = false; bool isConstructor = false; const char* nameEnd = nullptr; while (nameStart > fChar && ' ' != (ch = *--nameStart)) { if (!isalnum(ch) && '_' != ch) { if (nameEnd) { break; } expectOperator = true; continue; } if (!nameEnd) { nameEnd = nameStart + 1; } } if (!nameEnd) { return this->reportError("unexpected method name char"); } if (' ' == nameStart[0]) { ++nameStart; } if (nameEnd <= nameStart) { return this->reportError("missing method name"); } if (nameStart >= paren) { return this->reportError("missing method name length"); } string name(nameStart, nameEnd - nameStart); bool allLower = true; for (int index = 0; index < (int) (nameEnd - nameStart); ++index) { if (!islower(nameStart[index])) { allLower = false; break; } } if (expectOperator && "operator" != name) { return this->reportError("expected operator"); } const Definition* parent = this->parentSpace(); if (parent && parent->fName.length() > 0) { size_t parentNameIndex = parent->fName.rfind(':'); parentNameIndex = string::npos == parentNameIndex ? 0 : parentNameIndex + 1; string parentName = parent->fName.substr(parentNameIndex); if (parentName == name) { isConstructor = true; } else if ('~' == name[0]) { if (parentName != name.substr(1)) { return this->reportError("expected destructor"); } isConstructor = true; } builder = parent->fName + "::"; } bool addConst = false; if (isConstructor || expectOperator) { paren = this->strnchr(')', end) + 1; TextParserSave saveState(this); this->skipTo(paren); if (this->skipExact(" const")) { addConst = true; } saveState.restore(); } builder.append(nameStart, paren - nameStart); if (addConst) { builder.append(" const"); } if (!expectOperator && allLower) { builder.append("()"); } int parens = 0; while (fChar < end || parens > 0) { if ('(' == this->peek()) { ++parens; } else if (')' == this->peek()) { --parens; } this->next(); } TextParserSave saveState(this); this->skipWhiteSpace(); if (this->startsWith("const")) { this->skipName("const"); } else { saveState.restore(); } // this->next(); if (string::npos != builder.find('\n')) { builder.erase(std::remove(builder.begin(), builder.end(), '\n'), builder.end()); } return uniqueRootName(builder, MarkType::kMethod); } const Definition* BmhParser::parentSpace() const { Definition* parent = nullptr; Definition* test = fParent; while (test) { if (MarkType::kClass == test->fMarkType || MarkType::kEnumClass == test->fMarkType || MarkType::kStruct == test->fMarkType) { parent = test; break; } test = test->fParent; } return parent; } // A full terminal statement is in the form: // \n optional-white-space #MarkType white-space #[# white-space] // \n optional-white-space #MarkType white-space Name white-space #[# white-space] // MarkType must match definition->fMarkType const char* BmhParser::checkForFullTerminal(const char* end, const Definition* definition) const { const char* start = end; while ('\n' != start[0] && start > fStart) { --start; } SkASSERT (start < end); // if end is preceeeded by \n#MarkType ## backup to there TextParser parser(fFileName, start, fChar, fLineCount); parser.skipWhiteSpace(); if (parser.eof() || fMC != parser.next()) { return end; } const char* markName = kMarkProps[(int) definition->fMarkType].fName; if (!parser.skipExact(markName)) { return end; } parser.skipWhiteSpace(); TextParser startName(fFileName, definition->fStart, definition->fContentStart, definition->fLineCount); if ('#' == startName.next()) { startName.skipToSpace(); if (!startName.eof() && startName.skipSpace()) { const char* nameBegin = startName.fChar; startName.skipToWhiteSpace(); string name(nameBegin, (int) (startName.fChar - nameBegin)); if (fMC != parser.peek() && !parser.skipExact(name.c_str())) { return end; } parser.skipSpace(); } } if (parser.eof() || fMC != parser.next()) { return end; } if (!parser.eof() && fMC != parser.next()) { return end; } SkASSERT(parser.eof()); return start; } void BmhParser::parseHashAnchor(Definition* definition) { this->skipToEndBracket(fMC); fMarkup.emplace_front(MarkType::kLink, fChar, fLineCount, definition, fMC); SkAssertResult(fMC == this->next()); this->skipWhiteSpace(); Definition* link = &fMarkup.front(); link->fContentStart = fChar; link->fContentEnd = this->trimmedBracketEnd(fMC); this->skipToEndBracket(fMC); SkAssertResult(fMC == this->next()); SkAssertResult(fMC == this->next()); link->fTerminator = fChar; definition->fContentEnd = link->fContentEnd; definition->fTerminator = fChar; definition->fChildren.emplace_back(link); } void BmhParser::parseHashFormula(Definition* definition) { const char* start = definition->fContentStart; definition->trimEnd(); const char* end = definition->fContentEnd; fMarkup.emplace_front(MarkType::kText, start, fLineCount, definition, fMC); Definition* text = &fMarkup.front(); text->fContentStart = start; text->fContentEnd = end; text->fTerminator = definition->fTerminator; definition->fChildren.emplace_back(text); } void BmhParser::parseHashLine(Definition* definition) { const char* nextLF = this->strnchr('\n', this->fEnd); const char* start = fChar; const char* end = this->trimmedBracketEnd(fMC); this->skipToEndBracket(fMC, nextLF); if (fMC != this->next() || fMC != this->next()) { return this->reportError("expected ## to delineate line"); } fMarkup.emplace_front(MarkType::kText, start, fLineCount, definition, fMC); Definition* text = &fMarkup.front(); if (!islower(start[0]) && (!isdigit(start[0]) || MarkType::kConst != definition->fParent->fMarkType)) { return this->reportError("expect lower case start"); } string contents = string(start, end - start); size_t firstSpace = contents.find(' '); if (string::npos == firstSpace || 0 == firstSpace || 's' != start[firstSpace - 1]) { if (MarkType::kMethod == fParent->fMarkType && "experimental" != contents && "incomplete" != contents) { return this->reportError( "expect phrase in third person present" " tense (1st word should end in 's'"); } } text->fContentStart = start; text->fContentEnd = end; text->fTerminator = fChar; definition->fContentEnd = text->fContentEnd; definition->fTerminator = fChar; definition->fChildren.emplace_back(text); } bool BmhParser::popParentStack(Definition* definition) { if (!fParent) { return this->reportError("missing parent"); } if (definition != fParent) { return this->reportError("definition end is not parent"); } if (!definition->fStart) { return this->reportError("definition missing start"); } if (definition->fContentEnd) { return this->reportError("definition already ended"); } // more to figure out to handle table columns, at minimum const char* end = fChar; if (fMC != end[0]) { while (end > definition->fContentStart && ' ' >= end[-1]) { --end; } SkASSERT(&end[-1] >= definition->fContentStart && fMC == end[-1] && (MarkType::kColumn == definition->fMarkType || (&end[-2] >= definition->fContentStart && fMC == end[-2]))); end -= 2; } end = checkForFullTerminal(end, definition); definition->fContentEnd = end; definition->fTerminator = fChar; fParent = definition->fParent; if (!fParent || (MarkType::kTopic == fParent->fMarkType && !fParent->fParent)) { fRoot = nullptr; } return true; } bool BmhParser::skipNoName() { if ('\n' == this->peek()) { this->next(); return true; } this->skipWhiteSpace(); if (fMC != this->peek()) { return this->reportError("expected end mark 1"); } this->next(); if (fMC != this->peek()) { return this->reportError("expected end mark 2"); } this->next(); return true; } bool BmhParser::skipToDefinitionEnd(MarkType markType) { if (this->eof()) { return this->reportError("missing end"); } const char* start = fLine; int startLineCount = fLineCount; int stack = 1; ptrdiff_t lineLen; bool foundEnd = false; do { lineLen = this->lineLength(); if (fMC != *fChar++) { continue; } if (fMC == *fChar) { continue; } if (' ' == *fChar) { continue; } MarkType nextType = this->getMarkType(MarkLookup::kAllowUnknown); if (markType != nextType) { continue; } bool hasEnd = this->hasEndToken(); if (hasEnd) { if (!--stack) { foundEnd = true; continue; } } else { ++stack; } } while ((void) ++fLineCount, (void) (fLine += lineLen), (void) (fChar = fLine), !this->eof() && !foundEnd); if (foundEnd) { return true; } fLineCount = startLineCount; fLine = start; fChar = start; return this->reportError("unbalanced stack"); } bool BmhParser::skipToString() { this->skipSpace(); if (fMC != this->peek()) { return this->reportError("expected end mark 3"); } this->next(); this->skipSpace(); // body is text from here to double fMC // no single fMC allowed, no linefeed allowed return true; } vector BmhParser::topicName() { vector result; this->skipWhiteSpace(); const char* lineEnd = fLine + this->lineLength(); const char* nameStart = fChar; while (fChar < lineEnd) { char ch = this->next(); SkASSERT(',' != ch); if ('\n' == ch) { break; } if (fMC == ch) { break; } } if (fChar - 1 > nameStart) { string builder(nameStart, fChar - nameStart - 1); trim_start_end(builder); result.push_back(builder); } if (fChar < lineEnd && fMC == this->peek()) { this->next(); } return result; } // typeName parsing rules depend on mark type vector BmhParser::typeName(MarkType markType, bool* checkEnd) { fAnonymous = false; fCloned = false; vector result; string builder; if (fParent) { builder = fParent->fName; } switch (markType) { case MarkType::kDefine: case MarkType::kEnum: // enums may be nameless case MarkType::kConst: case MarkType::kEnumClass: case MarkType::kClass: case MarkType::kStruct: case MarkType::kTemplate: // expect name builder = this->className(markType); break; case MarkType::kExample: // check to see if one already exists -- if so, number this one builder = this->uniqueName(string(), markType); this->skipNoName(); break; case MarkType::kCode: case MarkType::kDescription: case MarkType::kExternal: case MarkType::kFunction: case MarkType::kLegend: case MarkType::kList: case MarkType::kNoExample: this->skipNoName(); break; case MarkType::kFormula: case MarkType::kLine: this->skipToString(); break; case MarkType::kAlias: case MarkType::kAnchor: case MarkType::kBug: // fixme: expect number case MarkType::kDetails: case MarkType::kDuration: case MarkType::kFile: case MarkType::kFilter: case MarkType::kHeight: case MarkType::kIllustration: case MarkType::kImage: case MarkType::kIn: case MarkType::kLiteral: case MarkType::kNoJustify: case MarkType::kOutdent: case MarkType::kPlatform: case MarkType::kPopulate: case MarkType::kReturn: case MarkType::kSeeAlso: case MarkType::kSet: case MarkType::kSubstitute: case MarkType::kToDo: case MarkType::kVolatile: case MarkType::kWidth: *checkEnd = false; // no name, may have text body break; case MarkType::kStdOut: this->skipNoName(); break; // unnamed case MarkType::kMember: builder = this->memberName(); break; case MarkType::kMethod: builder = this->methodName(); break; case MarkType::kTypedef: builder = this->typedefName(); break; case MarkType::kParam: // fixme: expect camelCase for param builder = this->word("", ""); this->skipSpace(); *checkEnd = false; break; case MarkType::kPhraseDef: { const char* nameEnd = this->anyOf("(\n"); builder = string(fChar, nameEnd - fChar); this->skipLower(); if (fChar != nameEnd) { this->reportError("expect lower case only"); break; } this->skipTo(nameEnd); *checkEnd = false; } break; case MarkType::kTable: this->skipNoName(); break; // unnamed case MarkType::kSubtopic: case MarkType::kTopic: // fixme: start with cap, allow space, hyphen, stop on comma // one topic can have multiple type names delineated by comma result = this->topicName(); if (result.size() == 0 && this->hasEndToken()) { break; } return result; default: // fixme: don't allow silent failures SkASSERT(0); } result.push_back(builder); return result; } string BmhParser::typedefName() { if (this->hasEndToken()) { if (!fParent || !fParent->fName.length()) { return this->reportError("missing parent typedef name"); } SkASSERT(fMC == this->peek()); this->next(); SkASSERT(fMC == this->peek()); this->next(); SkASSERT(fMC != this->peek()); return fParent->fName; } string builder; const Definition* parent = this->parentSpace(); if (parent && parent->fName.length() > 0) { builder = parent->fName + "::"; } builder += TextParser::typedefName(); return uniqueRootName(builder, MarkType::kTypedef); } string BmhParser::uniqueName(string base, MarkType markType) { string builder(base); if (!builder.length()) { builder = fParent->fName; } if (!fParent) { return builder; } int number = 2; string numBuilder(builder); do { for (auto& iter : fParent->fChildren) { if (markType == iter->fMarkType) { if (iter->fName == numBuilder) { fCloned = true; numBuilder = builder + '_' + to_string(number); goto tryNext; } } } break; tryNext: ; } while (++number); return numBuilder; } string BmhParser::uniqueRootName(string base, MarkType markType) { auto checkName = [markType](const Definition& def, string numBuilder) -> bool { return markType == def.fMarkType && def.fName == numBuilder; }; string builder(base); if (!builder.length()) { builder = fParent->fName; } int number = 2; string numBuilder(builder); Definition* cloned = nullptr; do { if (fRoot) { for (auto& iter : fRoot->fBranches) { if (checkName(*iter.second, numBuilder)) { cloned = iter.second; goto tryNext; } } for (auto& iter : fRoot->fLeaves) { if (checkName(iter.second, numBuilder)) { cloned = &iter.second; goto tryNext; } } } else if (fParent) { for (auto& iter : fParent->fChildren) { if (checkName(*iter, numBuilder)) { cloned = &*iter; goto tryNext; } } } break; tryNext: ; if ("()" == builder.substr(builder.length() - 2)) { builder = builder.substr(0, builder.length() - 2); } if (MarkType::kMethod == markType) { cloned->fCloned = true; } fCloned = true; numBuilder = builder + '_' + to_string(number); } while (++number); return numBuilder; } void BmhParser::validate() const { for (int index = 0; index <= (int) Last_MarkType; ++index) { SkASSERT(kMarkProps[index].fMarkType == (MarkType) index); } const char* last = ""; for (int index = 0; index <= (int) Last_MarkType; ++index) { const char* next = kMarkProps[index].fName; if (!last[0]) { last = next; continue; } if (!next[0]) { continue; } SkASSERT(strcmp(last, next) < 0); last = next; } } string BmhParser::word(string prefix, string delimiter) { string builder(prefix); this->skipWhiteSpace(); const char* lineEnd = fLine + this->lineLength(); const char* nameStart = fChar; while (fChar < lineEnd) { char ch = this->next(); if (' ' >= ch) { break; } if (',' == ch) { return this->reportError("no comma needed"); break; } if (fMC == ch) { return builder; } if (!isalnum(ch) && '_' != ch && ':' != ch && '-' != ch) { return this->reportError("unexpected char"); } if (':' == ch) { // expect pair, and expect word to start with Sk if (nameStart[0] != 'S' || nameStart[1] != 'k') { return this->reportError("expected Sk"); } if (':' != this->peek()) { return this->reportError("expected ::"); } this->next(); } else if ('-' == ch) { // expect word not to start with Sk or kX where X is A-Z if (nameStart[0] == 'k' && nameStart[1] >= 'A' && nameStart[1] <= 'Z') { return this->reportError("didn't expected kX"); } if (nameStart[0] == 'S' && nameStart[1] == 'k') { return this->reportError("expected Sk"); } } } if (prefix.size()) { builder += delimiter; } builder.append(nameStart, fChar - nameStart - 1); return builder; }