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