1 | import sys
|
---|
2 |
|
---|
3 | from xml.sax import make_parser, handler
|
---|
4 |
|
---|
5 | # --- Templates
|
---|
6 |
|
---|
7 | top = \
|
---|
8 | """
|
---|
9 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
|
---|
10 | <HTML>
|
---|
11 | <HEAD>
|
---|
12 | <TITLE>%s</TITLE>
|
---|
13 | </HEAD>
|
---|
14 |
|
---|
15 | <BODY>
|
---|
16 | <H1>%s</H1>
|
---|
17 | """
|
---|
18 |
|
---|
19 | bottom = \
|
---|
20 | """
|
---|
21 | </ul>
|
---|
22 |
|
---|
23 | <HR>
|
---|
24 | <ADDRESS>
|
---|
25 | Converted to HTML by sax_rss2html.py.
|
---|
26 | </ADDRESS>
|
---|
27 |
|
---|
28 | </BODY>
|
---|
29 | </HTML>
|
---|
30 | """
|
---|
31 |
|
---|
32 | # --- The ContentHandler
|
---|
33 |
|
---|
34 | class RSSHandler(handler.ContentHandler):
|
---|
35 |
|
---|
36 | def __init__(self, out = sys.stdout):
|
---|
37 | handler.ContentHandler.__init__(self)
|
---|
38 | self._out = out
|
---|
39 |
|
---|
40 | self._text = ""
|
---|
41 | self._parent = None
|
---|
42 | self._list_started = 0
|
---|
43 | self._title = None
|
---|
44 | self._link = None
|
---|
45 | self._descr = ""
|
---|
46 |
|
---|
47 | # ContentHandler methods
|
---|
48 |
|
---|
49 | def startElement(self, name, attrs):
|
---|
50 | if name == "channel" or name == "image" or name == "item":
|
---|
51 | self._parent = name
|
---|
52 |
|
---|
53 | self._text = ""
|
---|
54 |
|
---|
55 | def endElement(self, name):
|
---|
56 | if self._parent == "channel":
|
---|
57 | if name == "title":
|
---|
58 | self._out.write(top % (self._text, self._text))
|
---|
59 | elif name == "description":
|
---|
60 | self._out.write("<p>%s</p>\n" % self._text)
|
---|
61 |
|
---|
62 | elif self._parent == "item":
|
---|
63 | if name == "title":
|
---|
64 | self._title = self._text
|
---|
65 | elif name == "link":
|
---|
66 | self._link = self._text
|
---|
67 | elif name == "description":
|
---|
68 | self._descr = self._text
|
---|
69 | elif name == "item":
|
---|
70 | if not self._list_started:
|
---|
71 | self._out.write("<ul>\n")
|
---|
72 | self._list_started = 1
|
---|
73 |
|
---|
74 | self._out.write(' <li><a href="%s">%s</a> %s\n' %
|
---|
75 | (self._link, self._title, self._descr))
|
---|
76 |
|
---|
77 | self._title = None
|
---|
78 | self._link = None
|
---|
79 | self._descr = ""
|
---|
80 |
|
---|
81 | if name == "rss":
|
---|
82 | self._out.write(bottom)
|
---|
83 |
|
---|
84 | def characters(self, content):
|
---|
85 | self._text = self._text + content
|
---|
86 |
|
---|
87 | # --- Main program
|
---|
88 |
|
---|
89 | parser = make_parser()
|
---|
90 | parser.setContentHandler(RSSHandler())
|
---|
91 | parser.parse(sys.argv[1])
|
---|