Dome/support.py

   1 import rox
   2
   3 import os, sys
   4 import traceback
   5 from xml.dom import Node
   6 from Ft.Xml.Domlette import implementation
   7 from Ft.Xml import XMLNS_NAMESPACE
   8 from Ft.Xml.Lib.Nss import GetAllNs
   9
  10 from string import find, lower, join
  11 from socket import gethostbyaddr, gethostname
  12
  13 import re
  14 entrefpattern = re.compile('&(\D\S+);')
  15
  16 def node_to_xml(node):
  17         "Takes an XML node and returns an XML documentElement suitable for saving."
  18         root = implementation.createDocument(None, 'root', None)
  19         new = node.cloneNode(1)
  20         new = root.importNode(new, 1)
  21         root.replaceChild(new, root.documentElement)
  22         return root
  23
  24 def node_to_html(node):
  25         "Takes an XML node and returns an HTML documentElement suitable for saving."
  26         root = implementation.createHTMLDocument('HTML document')
  27         def html(doc, node, html):
  28                 new = doc.importNode(node.cloneNode(deep = 0), deep = 0)
  29                 if node.nodeType == Node.ELEMENT_NODE:
  30                         for a in node.attributes:
  31                                 new.setAttribute(a.localName, a.value)
  32                         for k in node.childNodes:
  33                                 new.appendChild(html(doc, k, html))
  34                 return new
  35         new = html(root, node, html)
  36         root.replaceChild(new, root.documentElement)
  37         return root
  38
  39 def send_to_file(data, path):
  40         try:
  41                 file = open(path, 'wb')
  42                 try:
  43                         file.write(data)
  44                 finally:
  45                         file.close()
  46         except:
  47                 rox.report_exception()
  48                 return 0
  49
  50         return 1
  51
  52 def fix_broken_html(data):
  53         """Pre-parse the data before sending to tidy to fix really really broken
  54 stuff (eg, MS Word output). Returns None if data is OK"""
  55         if data.find('<o:p>') == -1:
  56                 return          # Doesn't need fixing?
  57         import re
  58         data = data.replace('<o:p></o:p>', '')
  59         data = re.sub('<!\[[^]]*\]>', '', data)
  60         return data
  61
  62 def to_html_doc(data):
  63         "Run data though tidy and return the resulting XML text"
  64         (r, w) = os.pipe()
  65         child = os.fork()
  66         #data = data.replace('&nbsp;', ' ')
  67         #data = data.replace('&copy;', '(c)')
  68         #data = data.replace('&auml;', '(auml)')
  69         #data = data.replace('&ouml;', '(ouml)')
  70         fixed = fix_broken_html(data)
  71         if child == 0:
  72                 # We are the child
  73                 try:
  74                         os.close(r)
  75                         os.dup2(w, 1)
  76                         os.close(w)
  77                         if fixed:
  78                                 tin = os.popen('tidy --force-output yes -q -utf8 -asxml 2>/dev/null', 'w')
  79                         else:
  80                                 tin = os.popen('tidy --force-output yes -q -asxml 2>/dev/null', 'w')
  81                         tin.write(fixed or data)
  82                         tin.close()
  83                 finally:
  84                         os._exit(0)
  85         os.close(w)
  86
  87         data = os.fdopen(r).read()
  88         os.waitpid(child, 0)
  89
  90         return data
  91
  92 def parse_data(data, path):
  93         """Convert and XML document into a DOM Document."""
  94         from Ft.Xml.InputSource import InputSourceFactory
  95         #from Ft.Xml.cDomlette import nonvalParse
  96         from Ft.Xml.FtMiniDom import nonvalParse
  97         isrc = InputSourceFactory()
  98
  99         try:
 100                 try:
 101                         print "Parsing (with entities)..."
 102                         doc = nonvalParse(isrc.fromString(data, path))
 103                 except:
 104                         print "Parse failed.. retry without entities..."
 105                         data = entrefpattern.sub('&amp;\\1;',data)
 106                         doc = nonvalParse(isrc.fromString(data, path))
 107         except:
 108                 type, val, tb = sys.exc_info()
 109                 traceback.print_exception(type, val, tb)
 110                 print "parsing failed!"
 111                 print "Data was:"
 112                 print data
 113                 #rox.report_exception()
 114                 raise
 115         else:
 116                 print "parse OK...",
 117         return doc
 118