1 | /* |
2 | This file is part of the syndication library |
3 | SPDX-FileCopyrightText: 2006 Frank Osterfeld <osterfeld@kde.org> |
4 | |
5 | SPDX-License-Identifier: LGPL-2.0-or-later |
6 | */ |
7 | |
8 | #include "parser.h" |
9 | #include "constants.h" |
10 | #include "content.h" |
11 | #include "document.h" |
12 | |
13 | #include <documentsource.h> |
14 | |
15 | #include <QDomAttr> |
16 | #include <QDomDocument> |
17 | #include <QDomElement> |
18 | #include <QDomNamedNodeMap> |
19 | #include <QDomNode> |
20 | #include <QDomNodeList> |
21 | |
22 | #include <QHash> |
23 | #include <QString> |
24 | |
25 | namespace Syndication |
26 | { |
27 | namespace Atom |
28 | { |
29 | class SYNDICATION_NO_EXPORT Parser::ParserPrivate |
30 | { |
31 | public: |
32 | static QDomDocument convertAtom0_3(const QDomDocument &document); |
33 | static QDomNode convertNode(QDomDocument &doc, const QDomNode &node, const QHash<QString, QString> &nameMapper); |
34 | }; |
35 | |
36 | bool Parser::accept(const Syndication::DocumentSource &source) const |
37 | { |
38 | QDomElement root = source.asDomDocument().documentElement(); |
39 | return !root.isNull() && (root.namespaceURI() == atom1Namespace() || root.namespaceURI() == atom0_3Namespace()); |
40 | } |
41 | |
42 | Syndication::SpecificDocumentPtr Parser::parse(const Syndication::DocumentSource &source) const |
43 | { |
44 | QDomDocument doc = source.asDomDocument(); |
45 | |
46 | if (doc.isNull()) { |
47 | // if this is not atom, return an invalid feed document |
48 | return FeedDocumentPtr(new FeedDocument()); |
49 | } |
50 | |
51 | QDomElement feed = doc.namedItem(QStringLiteral("feed" )).toElement(); |
52 | |
53 | bool feedValid = !feed.isNull(); |
54 | |
55 | if (feedValid // |
56 | && feed.attribute(QStringLiteral("version" )) == QLatin1String("0.3" )) { |
57 | doc = ParserPrivate::convertAtom0_3(document: doc); |
58 | feed = doc.namedItem(QStringLiteral("feed" )).toElement(); |
59 | } |
60 | |
61 | feedValid = !feed.isNull() && feed.namespaceURI() == atom1Namespace(); |
62 | |
63 | if (feedValid) { |
64 | return FeedDocumentPtr(new FeedDocument(feed)); |
65 | } |
66 | |
67 | QDomElement entry = doc.namedItem(QStringLiteral("entry" )).toElement(); |
68 | bool entryValid = !entry.isNull() && entry.namespaceURI() == atom1Namespace(); |
69 | |
70 | if (entryValid) { |
71 | return EntryDocumentPtr(new EntryDocument(feed)); |
72 | } |
73 | |
74 | // if this is not atom, return an invalid feed document |
75 | return FeedDocumentPtr(new FeedDocument()); |
76 | } |
77 | |
78 | QString Parser::format() const |
79 | { |
80 | return QStringLiteral("atom" ); |
81 | } |
82 | |
83 | QDomNode Parser::ParserPrivate::convertNode(QDomDocument &doc, const QDomNode &node, const QHash<QString, QString> &nameMapper) |
84 | { |
85 | if (!node.isElement()) { |
86 | return node.cloneNode(deep: true); |
87 | } |
88 | |
89 | bool isAtom03Element = node.namespaceURI() == atom0_3Namespace(); |
90 | QDomElement oldEl = node.toElement(); |
91 | |
92 | // use new namespace |
93 | QString newNS = isAtom03Element ? atom1Namespace() : node.namespaceURI(); |
94 | |
95 | QString newName = node.localName(); |
96 | |
97 | // rename tags that are listed in the nameMapper |
98 | if (isAtom03Element && nameMapper.contains(key: node.localName())) { |
99 | newName = nameMapper[node.localName()]; |
100 | } |
101 | |
102 | QDomElement newEl = doc.createElementNS(nsURI: newNS, qName: newName); |
103 | |
104 | QDomNamedNodeMap attributes = oldEl.attributes(); |
105 | |
106 | // copy over attributes |
107 | const int numberOfAttributes(attributes.count()); |
108 | for (int i = 0; i < numberOfAttributes; ++i) { |
109 | const QDomAttr attr = attributes.item(index: i).toAttr(); |
110 | if (attr.namespaceURI().isEmpty()) { |
111 | newEl.setAttribute(name: attr.name(), value: attr.value()); |
112 | } else { |
113 | newEl.setAttributeNS(nsURI: attr.namespaceURI(), qName: attr.name(), value: attr.value()); |
114 | } |
115 | } |
116 | |
117 | /* clang-format off */ |
118 | bool isTextConstruct = newNS == atom1Namespace() |
119 | && (newName == QLatin1String("title" ) |
120 | || newName == QLatin1String("rights" ) |
121 | || newName == QLatin1String("subtitle" ) |
122 | || newName == QLatin1String("summary" )); |
123 | /* clang-format on */ |
124 | |
125 | // for atom text constructs, map to new type schema (which only allows text, type, xhtml) |
126 | |
127 | if (isTextConstruct) { |
128 | QString oldType = newEl.attribute(QStringLiteral("type" ), QStringLiteral("text/plain" )); |
129 | QString newType; |
130 | |
131 | Content::Format format = Content::mapTypeToFormat(type: oldType); |
132 | switch (format) { |
133 | case Content::XML: |
134 | newType = QStringLiteral("xhtml" ); |
135 | break; |
136 | case Content::EscapedHTML: |
137 | newType = QStringLiteral("html" ); |
138 | break; |
139 | case Content::PlainText: |
140 | case Content::Binary: |
141 | default: |
142 | newType = QStringLiteral("text" ); |
143 | } |
144 | |
145 | newEl.setAttribute(QStringLiteral("type" ), value: newType); |
146 | } else { |
147 | // for generator, rename the "url" attribute to "uri" |
148 | |
149 | bool isGenerator = newNS == atom1Namespace() && newName == QLatin1String("generator" ); |
150 | if (isGenerator && newEl.hasAttribute(QStringLiteral("url" ))) { |
151 | newEl.setAttribute(QStringLiteral("uri" ), value: newEl.attribute(QStringLiteral("url" ))); |
152 | } |
153 | } |
154 | |
155 | // process child nodes recursively and append them |
156 | QDomNodeList children = node.childNodes(); |
157 | for (int i = 0; i < children.count(); ++i) { |
158 | newEl.appendChild(newChild: convertNode(doc, node: children.item(index: i), nameMapper)); |
159 | } |
160 | |
161 | return newEl; |
162 | } |
163 | |
164 | QDomDocument Parser::ParserPrivate::convertAtom0_3(const QDomDocument &doc03) |
165 | { |
166 | QDomDocument doc = doc03.cloneNode(deep: false).toDocument(); |
167 | |
168 | // these are the tags that were renamed in 1.0 |
169 | QHash<QString, QString> nameMapper; |
170 | nameMapper.insert(QStringLiteral("issued" ), QStringLiteral("published" )); |
171 | nameMapper.insert(QStringLiteral("modified" ), QStringLiteral("updated" )); |
172 | nameMapper.insert(QStringLiteral("url" ), QStringLiteral("uri" )); |
173 | nameMapper.insert(QStringLiteral("copyright" ), QStringLiteral("rights" )); |
174 | nameMapper.insert(QStringLiteral("tagline" ), QStringLiteral("subtitle" )); |
175 | |
176 | const QDomNodeList children = doc03.childNodes(); |
177 | |
178 | for (int i = 0; i < children.count(); ++i) { |
179 | doc.appendChild(newChild: convertNode(doc, node: children.item(index: i), nameMapper)); |
180 | } |
181 | |
182 | return doc; |
183 | } |
184 | |
185 | Parser::Parser() |
186 | : d(nullptr) |
187 | { |
188 | Q_UNUSED(d) // silence -Wunused-private-field |
189 | } |
190 | |
191 | Parser::~Parser() = default; |
192 | |
193 | Parser::Parser(const Parser &other) |
194 | : AbstractParser(other) |
195 | , d(nullptr) |
196 | { |
197 | } |
198 | Parser &Parser::operator=(const Parser & /*other*/) |
199 | { |
200 | return *this; |
201 | } |
202 | |
203 | } // namespace Atom |
204 | } // namespace Syndication |
205 | |