Dome/support.py

   1 import rox
   2
   3 import os, sys
   4 import traceback
   5 from Ft.Xml.Domlette import Node, implementation
   6 from Ft.Xml import XMLNS_NAMESPACE
   7 from Ft.Xml.Lib.Nss import GetAllNs
   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(deep = 1)
  19         new = root.importNode(new, deep = 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 import_with_ns(doc, node):
  52         nss = GetAllNs(node)
  53
  54         node = doc.importNode(node, 1)
  55         for ns in nss.keys():
  56                 if ns == 'xml':
  57                         continue
  58                 uri = nss[ns]
  59                 if ns or uri:
  60                         if ns is None:
  61                                 ns = 'xmlns'
  62                         else:
  63                                 ns = 'xmlns:' + ns
  64                         node.setAttributeNS(XMLNS_NAMESPACE, ns, uri)
  65         return node
  66
  67 def fix_broken_html(data):
  68         """Pre-parse the data before sending to tidy to fix really really broken
  69 stuff (eg, MS Word output). Returns None if data is OK"""
  70         if data.find('<o:p>') == -1:
  71                 return          # Doesn't need fixing?
  72         import re
  73         data = data.replace('<o:p></o:p>', '')
  74         data = re.sub('<!\[[^]]*\]>', '', data)
  75         return data
  76
  77 def to_html_doc(data):
  78         (r, w) = os.pipe()
  79         child = os.fork()
  80         #data = data.replace('&nbsp;', ' ')
  81         #data = data.replace('&copy;', '(c)')
  82         #data = data.replace('&auml;', '(auml)')
  83         #data = data.replace('&ouml;', '(ouml)')
  84         fixed = fix_broken_html(data)
  85         if child == 0:
  86                 # We are the child
  87                 try:
  88                         os.close(r)
  89                         os.dup2(w, 1)
  90                         os.close(w)
  91                         if fixed:
  92                                 tin = os.popen('tidy --force-output yes -q -utf8 -asxml 2>/dev/null', 'w')
  93                         else:
  94                                 tin = os.popen('tidy --force-output yes -q -asxml 2>/dev/null', 'w')
  95                         tin.write(fixed or data)
  96                         tin.close()
  97                 finally:
  98                         os._exit(0)
  99         os.close(w)
 100
 101         data = os.fdopen(r).read()
 102         os.waitpid(child, 0)
 103
 104         return data
 105
 106 def parse_data(data, path):
 107         """Convert and XML document into a DOM Document."""
 108         from Ft.Xml.InputSource import InputSourceFactory
 109         #from Ft.Xml.cDomlette import nonvalParse
 110         from Ft.Xml.FtMiniDom import nonvalParse
 111         isrc = InputSourceFactory()
 112
 113         try:
 114                 try:
 115                         print "Parsing (with entities)..."
 116                         doc = nonvalParse(isrc.fromString(data, path))
 117                 except:
 118                         print "Parse failed.. retry without entities..."
 119                         data = entrefpattern.sub('&amp;\\1;',data)
 120                         doc = nonvalParse(isrc.fromString(data, path))
 121         except:
 122                 type, val, tb = sys.exc_info()
 123                 traceback.print_exception(type, val, tb)
 124                 print "parsing failed!"
 125                 print "Data was:"
 126                 print data
 127                 #rox.report_exception()
 128                 raise
 129         else:
 130                 print "parse OK...",
 131         return doc
 132