Mercurial > hg > MPIWGWeb

annotate xmlhelper.py @ 0:bca61e893fcc

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first checkin of MPIWGWeb r2 branch from CVS into mercurial
author casties
date Thu, 10 Jan 2013 17:52:13 +0100
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
1
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
2 from sys import argv
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
3
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
4 import string
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
5 import xml.dom.minidom
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
6 #import Ft.Xml.XLink.Processor
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
7 #import Ft.Xml.XLink.XLinkElements
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
8 #
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
9 #from Ft.Xml import XPath
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
10 #from Ft.Xml.XPath import Evaluate
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
11 #from Ft.Xml.XLink import XLINK_NAMESPACE
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
12 #from Ft.Xml.XLink import XLinkElements
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
13
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
14 #from Ft.Xml.Domlette import NonvalidatingReader,InputSource
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
15 #from Ft.Xml import EMPTY_NAMESPACE
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
16
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
17 #from Ft.Lib import Uri
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
18
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
19 from xml.etree import ElementTree
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
20 import logging
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
21
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
22 xml2html={'WEB_normal':('<p>','</p>'),'Normal':('<p>','</p>'),'WEB_picture':('<p class=\"picture\">','</p>'),'WEB_figuretitle':('<p class=\"picturetitle\">','</p>'),'WEB_bibliography':('<p><i>','</i></p>'),'Web_kursiv':('<i>','</i>'),'WEB_kursiv':('<i>','</i>'),'WEB_hyperlink':('',''),'Hyperlink':('','')}
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
23
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
24 def addToDict(dict,name,value):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
25 if name=="":
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
26 return 0
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
27 else:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
28
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
29 if not dict.has_key(name):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
30 dict[name]=[] # als array anlegen
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
31
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
32 dict[name].append(value)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
33 return 1
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
34
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
35 def proj2hash(xmlstring):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
36 """wandelt xml-files fuer die projekte in ein hash"""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
37
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
38 #dom=xml.dom.minidom.parseString(xmlstring)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
39
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
40 tree = ElementTree.fromstring(xmlstring)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
41
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
42
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
43 pars = tree.findall(".//part[0]/par")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
44
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
45 list={}
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
46
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
47 #gettitle
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
48 #part= dom.getElementsByTagName('part')[0]
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
49 #pars=part.getElementsByTagName('par')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
50 #pars=Evaluate('par',dom.getElementsByTagName('part')[0])
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
51 logging.debug(pars)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
52 for par in pars:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
53 logging.debug(par)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
54 className=par.attrib['class']
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
55 #.getAttribute('class')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
56 content=par.text
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
57 addToDict(list,className,content)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
58
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
59 list.update(parseTable(tree.find('.//{http://www.w3.org/HTML/1998/html4}table'))) # Parse the Table
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
60
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
61 #evaluate level 1
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
62 sections = tree.findall(".//part[0]/section")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
63 #sections=Evaluate('section',dom.getElementsByTagName('part')[0])# Parse all Sections
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
64 #sections=Evaluate('section',dom.getElementsByTagName('part')[0])# Parse all Sections
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
65 #print sections,dom.getElementsByTagName('part')[0]
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
66 for section in sections:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
67
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
68 sec=parseSection(section)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
69 if sec[0]=="WEB_project_header": # Sonderfall project
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
70
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
71 addToDict(list,'WEB_project_header',sec[1]) # store title
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
72 addToDict(list,'WEB_project_description',sec[2]) #store description
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
73 else: # no information in heading
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
74
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
75 addToDict(list,sec[0],sec[2])
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
76
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
77 #evaluate higher level sections
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
78 sections = tree.findall(".//part[0]/section/section")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
79 #sections=Evaluate('section/section',dom.getElementsByTagName('part')[0])
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
80
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
81 for section in sections:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
82 logging.debug("sections2:"+repr(section))
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
83 sec=parseSection(section)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
84
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
85 if sec[0]=="WEB_project_header": # Sonderfall project
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
86 addToDict(list,'WEB_project_header',sec[1]) # store title
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
87 addToDict(list,'WEB_project_description',sec[2]) #store description
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
88 else: # no information in heading
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
89 addToDict(list,sec[0],sec[2])
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
90
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
91
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
92 return list
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
93
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
94
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
95 def parseSection(section):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
96 type=""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
97 header=""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
98 #for heading in section.childNodes:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
99
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
100 heading=section.find(".//heading")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
101 # if getattr(heading,'tagName','')=="heading":
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
102
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
103
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
104 type=heading.attrib['class']
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
105 logging.debug("parseSection (class):"+type)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
106 header=heading.text
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
107 logging.debug("parseSection (header):"+header)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
108
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
109 if type=="": # falls heading fehlt, pruefe ob erster par richtig
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
110 par=section.find(".//par")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
111 #par=section.getElementsByTagName('par')[0]
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
112 type=par.attrib['class']
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
113 header=par.text
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
114
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
115 #print section.childNodes
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
116 pars=section.findall(".//par")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
117 #pars=Evaluate('par',section)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
118 content=par2html(pars)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
119
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
120 return (type,header,content)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
121
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
122 def parseTable(table):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
123 fields={}
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
124 rows=table.findall('.//{http://www.w3.org/HTML/1998/html4}tr')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
125 #rows=table.getElementsByTagName('html:tr')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
126 for row in rows:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
127 logging.debug("ROW")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
128 cols=row.findall('.//{http://www.w3.org/HTML/1998/html4}td')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
129 #cols=row.getElementsByTagName('html:td')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
130
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
131 #Name des Datenfeldes einlesen
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
132 try:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
133 field=cols[0].find('.//par').attrib['class']
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
134 #field=cols[0].getElementsByTagName('par')[0].getAttribute('class')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
135 #print "field",field
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
136 except:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
137 logging.debug("error")
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
138 field=""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
139
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
140 #Wandeln der Eintrge in HTML
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
141
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
142 pars=cols[1].findall('.//par')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
143 #pars=cols[1].getElementsByTagName('par')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
144
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
145
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
146 html=par2html(pars,tags=("",";"))
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
147 logging.debug("field:"+field)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
148 logging.debug("html:"+html)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
149 addToDict(fields,field,html)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
150 #print fields
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
151 return fields
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
152
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
153 def par2html(pars,tags=None):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
154 #html=""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
155 logging.debug("part2html:"+repr(pars))
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
156 if pars is None:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
157 return ""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
158 for par in pars:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
159 logging.debug("part2html:"+repr(par))
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
160 if not tags:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
161 try:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
162 tag=xml2html[par.attrib['class']]
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
163 except:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
164 tag=('<p>','</p>')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
165 else:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
166 tag=tags
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
167
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
168 content=par.text
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
169 if content is None:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
170 content=""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
171 logging.debug("part2html:"+content)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
172 #print "CONTETN",content
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
173
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
174 #print par.getAttribute('class'),node
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
175 try:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
176 html=html+tag[0]+content+tag[1]
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
177 except:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
178 html=tag[0]+content+tag[1]
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
179
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
180 try:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
181 return html
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
182 except:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
183 return ""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
184
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
185 def getXlink(nodes):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
186 """searches xlinks and gives them back as html"""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
187 ret=""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
188 for node in nodes:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
189 if node.attributes:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
190 if 'xlink:type' in node.attributes.keys(): #is a xlink?
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
191 ret +=xlink2html(node)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
192 return ret
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
193
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
194 def xlink2html(xlink):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
195 ret=""
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
196 attributes=xlink.attributes
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
197
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
198 if xlink.tagName.lower()=="image":
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
199 ret +="<img src=%s />"%xlink.getAttribute('xlink:href')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
200 elif xlink.tagName.lower()=="link":
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
201 ret +="<a href='%s' >%s</a>"%(xlink.getAttribute('xlink:href'),getText(xlink.childNodes))
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
202
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
203
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
204
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
205
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
206 return ret
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
207
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
208 def getText(nodelist):
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
209
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
210 rc = u''
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
211 for node in nodelist:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
212 if node.nodeType == node.TEXT_NODE:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
213 #print "node",node
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
214 #print "NODE",node.data.encode('utf-8','ignore'),"V"
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
215 #print "HALII"
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
216 try:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
217 try:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
218 #rc += node.data.encode('utf-8','ignore')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
219 rc += node.data
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
220
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
221 except:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
222 #rc= node.data.encode('utf-8','ignore')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
223 rc=node.data
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
224 except:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
225 rc="ERROR"
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
226 #node.data.decode('utf-8','ignore')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
227 print "ERROR"
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
228 node.data.encode('utf-8','ignore')
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
229 #print "RC",rc
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
230 elif node.tagName =="inline":
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
231 rc+=par2html([node])
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
232 elif node.attributes:
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
233
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
234 if 'xlink:type' in node.attributes.keys(): #is a xlink?
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
235 rc +=xlink2html(node)
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
236 #print "RWT",rc
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
237 return rc
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
238
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
239
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
240 #filename=argv[1]
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
241 #fileString=file(filename).read()
bca61e893fcc first checkin of MPIWGWeb r2 branch from CVS into mercurial
casties
parents:
diff changeset
242 #print proj2hash(fileString)