| 1 | // Copyright (C) 2024 The Qt Company Ltd. |
| 2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only |
| 3 | |
| 4 | #include <qdochtmlparser_p.h> |
| 5 | #include <QtCore/qregularexpression.h> |
| 6 | |
| 7 | QT_BEGIN_NAMESPACE |
| 8 | |
| 9 | using namespace Qt::StringLiterals; |
| 10 | |
| 11 | // An emprical value to avoid too much content |
| 12 | static constexpr qsizetype firstIndexOfParagraphTag = 400; |
| 13 | |
| 14 | // A paragraph can start with <p><i>, or <p><tt> |
| 15 | // We need smallest value to use QString::indexOf |
| 16 | static constexpr auto lengthOfStartParagraphTag = qsizetype(std::char_traits<char>::length(s: "<p>" )); |
| 17 | static constexpr auto lengthOfEndParagraphTag = qsizetype(std::char_traits<char>::length(s: "</p>" )); |
| 18 | |
| 19 | static QString getContentsByMarks(const QString &html, QString startMark, QString endMark) |
| 20 | { |
| 21 | startMark.prepend(s: "$$$"_L1 ); |
| 22 | endMark.prepend(s: "<!-- @@@"_L1 ); |
| 23 | |
| 24 | QString contents; |
| 25 | qsizetype start = html.indexOf(s: startMark); |
| 26 | if (start != -1) { |
| 27 | start = html.indexOf(s: "-->"_L1 , from: start); |
| 28 | if (start != -1) { |
| 29 | qsizetype end = html.indexOf(s: endMark, from: start); |
| 30 | if (end != -1) { |
| 31 | start += qsizetype(std::char_traits<char>::length(s: "-->" )); |
| 32 | contents = html.mid(position: start, n: end - start); |
| 33 | } |
| 34 | } |
| 35 | } |
| 36 | return contents; |
| 37 | } |
| 38 | |
| 39 | |
| 40 | static void stripAllHtml(QString *html) |
| 41 | { |
| 42 | Q_ASSERT(html); |
| 43 | html->remove(re: QRegularExpression("<.*?>"_L1 )); |
| 44 | } |
| 45 | |
| 46 | /*! \internal |
| 47 | \brief Process the string obtained from start mark to end mark. |
| 48 | This is duplicated from QtC's Utils::HtmlExtractor, modified on top of it. |
| 49 | */ |
| 50 | static void processOutput(QString *html) |
| 51 | { |
| 52 | Q_ASSERT(html); |
| 53 | if (html->isEmpty()) |
| 54 | return; |
| 55 | |
| 56 | // Do not write the first paragraph in case it has extra tags below. |
| 57 | // <p><i>This is only used on the Maemo platform.</i></p> |
| 58 | // or: <p><tt>This is used on Windows only.</tt></p> |
| 59 | // or: <p>[Conditional]</p> |
| 60 | const auto skipFirstParagraphIfNeeded = [html](qsizetype &index){ |
| 61 | const bool shouldSkipFirstParagraph = html->indexOf(s: QLatin1String("<p><i>" )) == index || |
| 62 | html->indexOf(s: QLatin1String("<p><tt>" )) == index || |
| 63 | html->indexOf(s: QLatin1String("<p><b>" )) == index || |
| 64 | html->indexOf(s: QLatin1String("<p>[Conditional]</p>" )) == index; |
| 65 | |
| 66 | if (shouldSkipFirstParagraph) |
| 67 | index = html->indexOf(s: QLatin1String("</p>" ), from: index) + lengthOfEndParagraphTag; |
| 68 | }; |
| 69 | |
| 70 | // Try to get the entire first paragraph, but if one is not found or if its opening |
| 71 | // tag is not in the very beginning (using an empirical value as the limit) |
| 72 | // the html is cleared out to avoid too much content. |
| 73 | qsizetype index = html->indexOf(s: QLatin1String("<p>" )); |
| 74 | if (index != -1 && index < firstIndexOfParagraphTag) { |
| 75 | skipFirstParagraphIfNeeded(index); |
| 76 | qsizetype endIndex = html->indexOf(s: QLatin1String("</p>" ), from: index + lengthOfStartParagraphTag); |
| 77 | if (endIndex != -1) { |
| 78 | *html = html->mid(position: index, n: endIndex - index); |
| 79 | } else { |
| 80 | html->clear(); |
| 81 | } |
| 82 | } else { |
| 83 | html->clear(); |
| 84 | } |
| 85 | } |
| 86 | |
| 87 | class : public HtmlExtractor |
| 88 | { |
| 89 | public: |
| 90 | QString extract(const QString &code, const QString &keyword, ExtractionMode mode) override; |
| 91 | }; |
| 92 | |
| 93 | class : public HtmlExtractor |
| 94 | { |
| 95 | public: |
| 96 | QString extract(const QString &code, const QString &keyword, ExtractionMode mode) override; |
| 97 | }; |
| 98 | |
| 99 | class : public HtmlExtractor |
| 100 | { |
| 101 | public: |
| 102 | QString extract(const QString &code, const QString &keyword, ExtractionMode mode) override; |
| 103 | }; |
| 104 | |
| 105 | QString ExtractQmlType::(const QString &code, const QString &element, ExtractionMode mode) |
| 106 | { |
| 107 | QString result; |
| 108 | // Get brief description |
| 109 | if (mode == ExtractionMode::Simplified) { |
| 110 | result = getContentsByMarks(html: code, startMark: element + "-brief"_L1 , endMark: element); |
| 111 | // Remove More... |
| 112 | if (!result.isEmpty()) { |
| 113 | const auto tailToRemove = "More..."_L1 ; |
| 114 | const auto lastIndex = result.lastIndexOf(s: tailToRemove); |
| 115 | if (lastIndex != -1) |
| 116 | result.remove(i: lastIndex, len: tailToRemove.length()); |
| 117 | } |
| 118 | } else { |
| 119 | result = getContentsByMarks(html: code, startMark: element + "-description"_L1 , endMark: element); |
| 120 | // Remove header |
| 121 | if (!result.isEmpty()) { |
| 122 | const auto = "Detailed Description"_L1 ; |
| 123 | const auto firstIndex = result.indexOf(s: headerToRemove); |
| 124 | if (firstIndex != -1) |
| 125 | result.remove(i: firstIndex, len: headerToRemove.length()); |
| 126 | } |
| 127 | } |
| 128 | |
| 129 | stripAllHtml(html: &result); |
| 130 | return result.trimmed(); |
| 131 | } |
| 132 | |
| 133 | QString ExtractQmlProperty::(const QString &code, const QString &keyword, ExtractionMode mode) |
| 134 | { |
| 135 | QString result; |
| 136 | // Qt 5.15 way of finding properties in doc |
| 137 | QString startMark = QString::fromLatin1(ba: "<a name=\"%1-prop\">" ).arg(a: keyword); |
| 138 | qsizetype startIndex = code.indexOf(s: startMark); |
| 139 | if (startIndex == -1) { |
| 140 | // if not found, try Qt6 |
| 141 | startMark = QString::fromLatin1( |
| 142 | ba: "<td class=\"tblQmlPropNode\"><p>\n<span class=\"name\">%1</span>" ) |
| 143 | .arg(a: keyword); |
| 144 | startIndex = code.indexOf(s: startMark); |
| 145 | } |
| 146 | |
| 147 | if (startIndex != -1) { |
| 148 | result = code.mid(position: startIndex + startMark.size()); |
| 149 | startIndex = result.indexOf(s: QLatin1String("<div class=\"qmldoc\"><p>" )); |
| 150 | } else { |
| 151 | result = getContentsByMarks(html: code, startMark: keyword + "-prop"_L1 , endMark: keyword ); |
| 152 | startIndex = result.indexOf(s: QLatin1String("<p>" )); |
| 153 | } |
| 154 | |
| 155 | if (startIndex == -1) |
| 156 | return {}; |
| 157 | result = result.mid(position: startIndex); |
| 158 | if (mode == ExtractionMode::Simplified) |
| 159 | processOutput(html: &result); |
| 160 | stripAllHtml(html: &result); |
| 161 | return result.trimmed(); |
| 162 | } |
| 163 | |
| 164 | QString ExtractQmlMethodOrSignal::(const QString &code, const QString &keyword, ExtractionMode mode) |
| 165 | { |
| 166 | // the case with <!-- $$$childAt[overload1]$$$childAtrealreal --> |
| 167 | QString mark = QString::fromLatin1(ba: "$$$%1[overload1]$$$%1" ).arg(a: keyword); |
| 168 | qsizetype startIndex = code.indexOf(s: mark); |
| 169 | if (startIndex != -1) { |
| 170 | startIndex = code.indexOf(s: "-->"_L1 , from: startIndex + mark.length()); |
| 171 | if (startIndex == -1) |
| 172 | return {}; |
| 173 | } else { |
| 174 | // it could be part of the method list |
| 175 | mark = QString::fromLatin1(ba: "<span class=\"name\">%1</span>" ) |
| 176 | .arg(a: keyword); |
| 177 | startIndex = code.indexOf(s: mark); |
| 178 | if (startIndex != -1) |
| 179 | startIndex += mark.length(); |
| 180 | else |
| 181 | return {}; |
| 182 | } |
| 183 | |
| 184 | startIndex = code.indexOf(s: QLatin1String("<div class=\"qmldoc\"><p>" ), from: startIndex); |
| 185 | if (startIndex == -1) |
| 186 | return {}; |
| 187 | |
| 188 | QString endMark = QString::fromLatin1(ba: "<!-- @@@" ); |
| 189 | qsizetype endIndex = code.indexOf(s: endMark, from: startIndex); |
| 190 | QString contents = code.mid(position: startIndex, n: endIndex); |
| 191 | if (mode == ExtractionMode::Simplified) |
| 192 | processOutput(html: &contents); |
| 193 | stripAllHtml(html: &contents); |
| 194 | return contents.trimmed(); |
| 195 | } |
| 196 | |
| 197 | ExtractDocumentation::(QQmlJS::Dom::DomType domType) |
| 198 | { |
| 199 | using namespace QQmlJS::Dom; |
| 200 | switch (domType) { |
| 201 | case DomType::QmlObject: |
| 202 | m_extractor = std::make_unique<ExtractQmlType>(); |
| 203 | break; |
| 204 | case DomType::Binding: |
| 205 | case DomType::PropertyDefinition: |
| 206 | m_extractor = std::make_unique<ExtractQmlProperty>(); |
| 207 | break; |
| 208 | case DomType::MethodInfo: |
| 209 | m_extractor = std::make_unique<ExtractQmlMethodOrSignal>(); |
| 210 | break; |
| 211 | default: |
| 212 | break; |
| 213 | } |
| 214 | } |
| 215 | |
| 216 | QString ExtractDocumentation::(const QString &code, const QString &keyword, HtmlExtractor::ExtractionMode mode) |
| 217 | { |
| 218 | Q_ASSERT(m_extractor); |
| 219 | return m_extractor->extract(code, keyword, mode); |
| 220 | } |
| 221 | |
| 222 | QT_END_NAMESPACE |
| 223 | |