Diffstat (limited to 'pwmanager/pwmanager/htmlparse.cpp') (more/less context) (show whitespace changes)
-rw-r--r-- | pwmanager/pwmanager/htmlparse.cpp | 184 |
1 files changed, 184 insertions, 0 deletions
diff --git a/pwmanager/pwmanager/htmlparse.cpp b/pwmanager/pwmanager/htmlparse.cpp new file mode 100644 index 0000000..acde2e3 --- a/dev/null +++ b/pwmanager/pwmanager/htmlparse.cpp | |||
@@ -0,0 +1,184 @@ | |||
1 | /*************************************************************************** | ||
2 | * * | ||
3 | * copyright (C) 2004 by Michael Buesch * | ||
4 | * email: mbuesch@freenet.de * | ||
5 | * * | ||
6 | * This program is free software; you can redistribute it and/or modify * | ||
7 | * it under the terms of the GNU General Public License version 2 * | ||
8 | * as published by the Free Software Foundation. * | ||
9 | * * | ||
10 | ***************************************************************************/ | ||
11 | |||
12 | #include "htmlparse.h" | ||
13 | #include "pwmexception.h" | ||
14 | |||
15 | #include <dom/html_document.h> | ||
16 | #include <dom/html_element.h> | ||
17 | #include <dom/html_misc.h> | ||
18 | |||
19 | |||
20 | /** enable/disable HTML-parser debugging (0/1) */ | ||
21 | #define HTMLPARSE_DEBUG 1 | ||
22 | |||
23 | |||
24 | #if !defined(PWM_DEBUG) || HTMLPARSE_DEBUG == 0 | ||
25 | # undef printDebug | ||
26 | # define printDebug(x)do {} while (0) | ||
27 | #endif | ||
28 | |||
29 | HtmlParse::HtmlParse() | ||
30 | { | ||
31 | } | ||
32 | |||
33 | HtmlParse::~HtmlParse() | ||
34 | { | ||
35 | } | ||
36 | |||
37 | bool HtmlParse::parseHtmlComment(const QString &dta, | ||
38 | HtmlComment *ret) | ||
39 | { | ||
40 | PWM_ASSERT(ret); | ||
41 | ret->clear(); | ||
42 | DOM::HTMLDocument curDoc(static_cast<KHTMLView *>(0)); | ||
43 | curDoc.open(); | ||
44 | curDoc.write(dta); | ||
45 | curDoc.close(); | ||
46 | DOM::HTMLElement body(curDoc.body()); | ||
47 | DOM::HTMLCollection children(body.children()); | ||
48 | |||
49 | unsigned long i, numCh = children.length(); | ||
50 | if (numCh != 1) { | ||
51 | /* we currently support only one global table */ | ||
52 | printDebug("HtmlParse::parseHtmlComment(): global children cnt != 1"); | ||
53 | return false; | ||
54 | } | ||
55 | DOM::DOMString nodeName; | ||
56 | DOM::Node curNode; | ||
57 | for (i = 0; i < numCh; ++i) { | ||
58 | curNode = children.item(i); | ||
59 | nodeName = curNode.nodeName(); | ||
60 | if (nodeName == "table") { | ||
61 | if (!parseCommentGlobTbl(curNode, ret)) | ||
62 | return false; | ||
63 | } else { | ||
64 | // We don't support something else than tables, yet. | ||
65 | printDebug("HtmlParse::parseHtmlComment(): unknown node"); | ||
66 | } | ||
67 | } | ||
68 | return true; | ||
69 | } | ||
70 | |||
71 | bool HtmlParse::parseCommentGlobTbl(const DOM::Node &node, | ||
72 | HtmlComment *ret) | ||
73 | { | ||
74 | PWM_ASSERT(node.nodeName() == "table"); | ||
75 | DOM::Node bodyNode(node.firstChild()); | ||
76 | PWM_ASSERT(bodyNode.nodeName() == "tbody"); | ||
77 | DOM::Node curNode(bodyNode.firstChild()); | ||
78 | DOM::DOMString nodeName; | ||
79 | ret->clear(); | ||
80 | while (!curNode.isNull()) { | ||
81 | nodeName = curNode.nodeName(); | ||
82 | if (nodeName == "tr") { | ||
83 | if (!parseCommentGlobTblRow(curNode, ret)) | ||
84 | return false; | ||
85 | } else { | ||
86 | printDebug("HtmlParse::parseCommentGlobTbl(): node unknown"); | ||
87 | } | ||
88 | curNode = curNode.nextSibling(); | ||
89 | } | ||
90 | return true; | ||
91 | } | ||
92 | |||
93 | bool HtmlParse::parseCommentGlobTblRow(const DOM::Node &node, | ||
94 | HtmlComment *ret) | ||
95 | { | ||
96 | DOM::Node curNode(node.firstChild()), child; | ||
97 | DOM::DOMString nodeName; | ||
98 | HtmlComment::SubTable subTbl; | ||
99 | while (!curNode.isNull()) { | ||
100 | nodeName = curNode.nodeName(); | ||
101 | if (nodeName == "th") { | ||
102 | // global title | ||
103 | ret->setTitle(curNode.firstChild().nodeValue().string()); | ||
104 | } else if (nodeName == "td") { | ||
105 | child = curNode.firstChild(); | ||
106 | if (child.nodeName() == "table" && | ||
107 | child.firstChild().nodeName() == "tbody") { | ||
108 | // we have a sub-table | ||
109 | if (!parseCommentSubTbl(child.firstChild(), &subTbl)) | ||
110 | return false; | ||
111 | ret->addSubtable(subTbl); | ||
112 | } else { | ||
113 | printDebug("HtmlParse::parseCommentGlobTblRow(): subelement unknown"); | ||
114 | } | ||
115 | } else { | ||
116 | printDebug("HtmlParse::parseCommentGlobTblRow(): node unknown"); | ||
117 | } | ||
118 | curNode = curNode.nextSibling(); | ||
119 | } | ||
120 | return true; | ||
121 | } | ||
122 | |||
123 | bool HtmlParse::parseCommentSubTbl(const DOM::Node &node, | ||
124 | HtmlComment::SubTable *ret) | ||
125 | { | ||
126 | PWM_ASSERT(node.nodeName() == "tbody"); | ||
127 | DOM::Node curNode(node.firstChild()); | ||
128 | DOM::DOMString nodeName; | ||
129 | ret->clear(); | ||
130 | while (!curNode.isNull()) { | ||
131 | nodeName = curNode.nodeName(); | ||
132 | if (nodeName == "tr") { | ||
133 | if (!parseCommentSubTblRow(curNode, ret)) | ||
134 | return false; | ||
135 | } else { | ||
136 | printDebug("HtmlParse::parseCommentSubTbl(): node unknown"); | ||
137 | } | ||
138 | curNode = curNode.nextSibling(); | ||
139 | } | ||
140 | return true; | ||
141 | } | ||
142 | |||
143 | bool HtmlParse::parseCommentSubTblRow(const DOM::Node &node, | ||
144 | HtmlComment::SubTable *ret) | ||
145 | { | ||
146 | DOM::Node curNode(node.firstChild()), child; | ||
147 | DOM::DOMString nodeName; | ||
148 | pair<QString, QString> curEntr; | ||
149 | while (!curNode.isNull()) { | ||
150 | nodeName = curNode.nodeName(); | ||
151 | if (nodeName == "th") { | ||
152 | // sub title | ||
153 | ret->setTitle(curNode.firstChild().nodeValue().string()); | ||
154 | } else if (nodeName == "td") { | ||
155 | child = curNode.firstChild(); | ||
156 | if (child.nodeName() == "#text") { | ||
157 | if (!parseCommentSubTblEntry(curNode, &curEntr)) | ||
158 | return false; | ||
159 | ret->addEntry(curEntr); | ||
160 | return true; | ||
161 | } else { | ||
162 | printDebug("HtmlParse::parseCommentSubTblRow(): subelement unknown"); | ||
163 | } | ||
164 | } else { | ||
165 | printDebug("HtmlParse::parseCommentGlobTblRow(): node unknown"); | ||
166 | } | ||
167 | curNode = curNode.nextSibling(); | ||
168 | } | ||
169 | return true; | ||
170 | } | ||
171 | |||
172 | bool HtmlParse::parseCommentSubTblEntry(const DOM::Node &node, | ||
173 | pair<QString, QString> *ret) | ||
174 | { | ||
175 | DOM::Node curChild(node); | ||
176 | if (curChild.isNull()) | ||
177 | return false; | ||
178 | ret->first = curChild.firstChild().nodeValue().string(); | ||
179 | curChild = curChild.nextSibling(); | ||
180 | if (curChild.isNull()) | ||
181 | return false; | ||
182 | ret->second = curChild.firstChild().nodeValue().string(); | ||
183 | return true; | ||
184 | } | ||