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 def node_to_xml(node):
  13         "Takes an XML node and returns an XML documentElement suitable for saving."
  14         root = implementation.createDocument(None, 'root', None)
  15         new = node.cloneNode(deep = 1)
  16         new = root.importNode(new, deep = 1)
  17         root.replaceChild(new, root.documentElement)
  18         return root
  19
  20 def node_to_html(node):
  21         "Takes an XML node and returns an HTML documentElement suitable for saving."
  22         root = implementation.createHTMLDocument('HTML document')
  23         def html(doc, node, html):
  24                 new = doc.importNode(node.cloneNode(deep = 0), deep = 0)
  25                 if node.nodeType == Node.ELEMENT_NODE:
  26                         for a in node.attributes:
  27                                 new.setAttribute(a.localName, a.value)
  28                         for k in node.childNodes:
  29                                 new.appendChild(html(doc, k, html))
  30                 return new
  31         new = html(root, node, html)
  32         root.replaceChild(new, root.documentElement)
  33         return root
  34
  35 def send_to_file(data, path):
  36         try:
  37                 file = open(path, 'wb')
  38                 try:
  39                         file.write(data)
  40                 finally:
  41                         file.close()
  42         except:
  43                 rox.report_exception()
  44                 return 0
  45
  46         return 1
  47
  48 def import_with_ns(doc, node):
  49         nss = GetAllNs(node)
  50
  51         node = doc.importNode(node, 1)
  52         for ns in nss.keys():
  53                 if ns == 'xml':
  54                         continue
  55                 uri = nss[ns]
  56                 if ns or uri:
  57                         if ns is None:
  58                                 ns = 'xmlns'
  59                         else:
  60                                 ns = 'xmlns:' + ns
  61                         node.setAttributeNS(XMLNS_NAMESPACE, ns, uri)
  62         return node
  63
  64 def fix_broken_html(data):
  65         """Pre-parse the data before sending to tidy to fix really really broken
  66 stuff (eg, MS Word output). Returns None if data is OK"""
  67         if data.find('<o:p>') == -1:
  68                 return          # Doesn't need fixing?
  69         import re
  70         data = data.replace('<o:p></o:p>', '')
  71         data = re.sub('<!\[[^]]*\]>', '', data)
  72         return data
  73
  74 def to_html_doc(data):
  75         (r, w) = os.pipe()
  76         child = os.fork()
  77         #data = data.replace('&nbsp;', ' ')
  78         #data = data.replace('&copy;', '(c)')
  79         #data = data.replace('&auml;', '(auml)')
  80         #data = data.replace('&ouml;', '(ouml)')
  81         fixed = fix_broken_html(data)
  82         if child == 0:
  83                 # We are the child
  84                 try:
  85                         os.close(r)
  86                         os.dup2(w, 1)
  87                         os.close(w)
  88                         if fixed:
  89                                 tin = os.popen('tidy --force-output yes -q -utf8 -asxml 2>/dev/null', 'w')
  90                         else:
  91                                 tin = os.popen('tidy --force-output yes -q -asxml 2>/dev/null', 'w')
  92                         tin.write(fixed or data)
  93                         tin.close()
  94                 finally:
  95                         os._exit(0)
  96         os.close(w)
  97
  98         data = os.fdopen(r).read()
  99         os.waitpid(child, 0)
 100
 101         return data
 102
 103 def parse_data(data, path):
 104         """Convert and XML document into a DOM Document."""
 105         from Ft.Xml.InputSource import InputSourceFactory
 106         #from Ft.Xml.cDomlette import nonvalParse
 107         from Ft.Xml.FtMiniDom import nonvalParse
 108         isrc = InputSourceFactory()
 109
 110         try:
 111                 try:
 112                         print "Parsing (with entities)..."
 113                         doc = nonvalParse(isrc.fromString(data, path))
 114                 except:
 115                         print "Parse failed.. retry without entities..."
 116                         data = entrefpattern.sub('&amp;\\1;',data)
 117                         doc = nonvalParse(isrc.fromString(data, path))
 118         except:
 119                 type, val, tb = sys.exc_info()
 120                 traceback.print_exception(type, val, tb)
 121                 print "parsing failed!"
 122                 print "Data was:"
 123                 print data
 124                 #rox.report_exception()
 125                 raise Beep
 126         else:
 127                 print "parse OK...",
 128         return doc
 129