MoinMoin: attachment:HTML2MoinMoin.py of ActionMarket

   1 #!/usr/bin/python2
   2 
   3 """
   4 Usage:
   5   moinconvert URL
   6 
   7 Retrives the given URL and convert it to MoinMoin markup. The result is
   8 written to stdout.
   9 """
  10 
  11 import htmlentitydefs, sys
  12 
  13 from HTMLParser import HTMLParser
  14 
  15 class HTML2MoinMoin(HTMLParser):
  16 
  17     start_tags = {
  18         "a"     : " [%(0)s ",
  19         "b"     : "'''",
  20         "em"    : "''",
  21         "tt"    : "{{{",
  22         "pre"   : "\n{{{",
  23         "p"     : "\n\n",
  24         "br"    : "\n\n",
  25         "h1"    : "\n\n= ",
  26         "h2"    : "\n\n== ",
  27         "h3"    : "\n\n=== ",
  28         "h4"    : "\n\n==== ",
  29         "h5"    : "\n\n===== ",
  30         "title" : "TITLE: ",
  31         "table" : "\n",
  32         "tr"    : "",
  33         "td"    : "||"
  34         }
  35 
  36     end_tags = {
  37         "a"     : ']',
  38         "b"     : "'''",
  39         "em"    : "''",
  40         "tt"    : "}}}",
  41         "pre"   : "}}}\n",
  42         "p"     : "",
  43         "h1"    : " =\n\n",
  44         "h2"    : " ==\n\n",
  45         "h3"    : " ===\n\n",
  46         "h4"    : " ====\n\n",
  47         "h5"    : " =====\n\n",
  48         "table" : "\n", 
  49         "tr"    : "||\n",
  50         "dt"    : ":: "
  51         }
  52 
  53     def __init__(self):
  54         HTMLParser.__init__(self)
  55         self.output = sys.stdout
  56         self.list_mode = []
  57         self.preformatted = False
  58         self.verbose = 0
  59 
  60     def write(self, text):
  61         self.output.write(text)
  62 
  63     def do_ul_start(self, attrs, tag):
  64         self.list_mode.append("*")
  65 
  66     def do_ol_start(self, attrs, tag):
  67         self.list_mode.append("1.")
  68 
  69     def do_dl_start(self, attrs, tag):
  70         self.list_mode.append("")
  71 
  72     def do_ul_end(self, tag):
  73         self.list_mode = self.list_mode[:-1]
  74 
  75     do_ol_end = do_ul_end
  76     do_dl_end = do_ul_end
  77 
  78     def do_li_start(self, args, tag):
  79         self.write("\n" + " " * len(self.list_mode) + self.list_mode[-1])
  80 
  81     def do_dt_start(self, args, tag):
  82         self.write("\n" + " " * len(self.list_mode) + self.list_mode[-1])
  83 
  84     def do_pre_start(self, args, tag):
  85         self.preformatted = True
  86         self.write(self.start_tags["pre"])
  87 
  88     def do_pre_end(self, tag):
  89         self.preformatted = False
  90         self.write(self.end_tags["pre"])
  91 
  92     def handle_starttag(self, tag, attrs):
  93         func = HTML2MoinMoin.__dict__.get("do_%s_start" % tag,
  94                                          HTML2MoinMoin.do_default_start)
  95         if ((func == HTML2MoinMoin.do_default_start) and
  96             self.start_tags.has_key(tag)):
  97             attr_dict = {}
  98             i = 0
  99             for a in attrs:
 100                 attr_dict[a[0]] = a[1]
 101                 attr_dict[str(i)] = a[1]
 102                 i += 1
 103             self.write(self.start_tags[tag] % attr_dict)
 104         else:
 105             func(self, attrs, tag)
 106 
 107     def handle_endtag(self, tag):
 108         func = HTML2MoinMoin.__dict__.get("do_%s_end" % tag,
 109                                          HTML2MoinMoin.do_default_end)
 110         if ((func == HTML2MoinMoin.do_default_end) and
 111             self.end_tags.has_key(tag)):
 112             self.write(self.end_tags[tag])
 113         else:
 114             func(self, tag)
 115 
 116     def handle_data(self, data):
 117         if self.preformatted:
 118             self.write(data)
 119         else:
 120             self.write(data.replace("\n", " "))
 121 
 122     def handle_charref(self, name):
 123         self.write(name)
 124 
 125     def handle_entityref(self, name):
 126         if htmlentitydefs.entitydefs.has_key(name):
 127             self.write(htmlentitydefs.entitydefs[name])
 128         else:
 129             self.write("&" + name)
 130 
 131     def do_default_start(self, attrs, tag):
 132         if self.verbose:
 133             print "Encountered the beginning of a %s tag" % tag
 134             print "Attribs: %s" % attrs
 135             
 136     def do_default_end(self, tag):
 137         if self.verbose:
 138             print "Encountered the end of a %s tag" % tag
 139 
 140 
 141 def main():
 142     import urllib
 143     htmldata = urllib.urlopen(sys.argv[1]).read()
 144 
 145     p = HTML2MoinMoin()
 146     p.feed(htmldata)
 147     p.close()
 148 
 149 
 150 if __name__ == "__main__":
 151     main()