Mercurial > hg > documentViewer

annotate MpdlXmlTextServer.py @ 510:4fb35343d2e7 elementtree

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
more search. nicer css.
author casties
date Tue, 28 Feb 2012 10:39:21 +0100
parents 9d05befdd462
children 551ca1641a5e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
1 from OFS.SimpleItem import SimpleItem
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
2 from Products.PageTemplates.PageTemplateFile import PageTemplateFile
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
3
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
4 import xml.etree.ElementTree as ET
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
5
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
6 import re
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
7 import logging
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
8 import urllib
506
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
9 import base64
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
10
458
48b135b089c8 more renovation
casties
parents: 456
diff changeset
11 from SrvTxtUtils import getInt, getText, getHttpData
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
12
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
13 def serialize(node):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
14 """returns a string containing an XML snippet of node"""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
15 s = ET.tostring(node, 'UTF-8')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
16 # snip off XML declaration
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
17 if s.startswith('<?xml'):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
18 i = s.find('?>')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
19 return s[i+3:]
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
20
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
21 return s
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
22
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
23
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
24 class MpdlXmlTextServer(SimpleItem):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
25 """TextServer implementation for MPDL-XML eXist server"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
26 meta_type="MPDL-XML TextServer"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
27
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
28 manage_options=(
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
29 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
30 )+SimpleItem.manage_options
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
31
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
32 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
33
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
34 def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
35 """constructor"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
36 self.id=id
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
37 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
38 self.timeout = timeout
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
39 if serverName is None:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
40 self.serverUrl = serverUrl
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
41 else:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
42 self.serverUrl = "http://%s/mpdl/interface/"%serverName
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
43
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
44 def getHttpData(self, url, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
45 """returns result from url+data HTTP request"""
458
48b135b089c8 more renovation
casties
parents: 456
diff changeset
46 return getHttpData(url,data,timeout=self.timeout)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
47
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
48 def getServerData(self, method, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
49 """returns result from text server for method+data"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
50 url = self.serverUrl+method
458
48b135b089c8 more renovation
casties
parents: 456
diff changeset
51 return getHttpData(url,data,timeout=self.timeout)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
52
506
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
53
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
54 def getPlacesOnPage(self, docinfo=None, pn=None):
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
55 """Returns list of GIS places of page pn"""
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
56 docpath = docinfo.get('textURLPath',None)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
57 if not docpath:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
58 return None
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
59
506
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
60 places=[]
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
61 text=self.getServerData("xpath.xql", "document=%s&xpath=//place&pn=%s"%(docpath,pn))
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
62 dom = ET.fromstring(text)
506
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
63 result = dom.findall(".//resultPage/place")
236
ccdce5aca47b *** empty log message ***
abukhman
parents: 233
diff changeset
64 for l in result:
506
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
65 id = l.get("id")
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
66 name = l.text
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
67 place = {'id': id, 'name': name}
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
68 places.append(place)
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
69
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
70 return places
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
71
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
72
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
73 def processPageInfo(self, dom, docinfo, pageinfo):
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
74 """processes page info divs from dom and stores in docinfo and pageinfo"""
469
15394486ab75 working with new templates
casties
parents: 465
diff changeset
75 # assume first second level div is pageMeta
15394486ab75 working with new templates
casties
parents: 465
diff changeset
76 alldivs = dom.find("div")
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
77
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
78 if alldivs is None or alldivs.get('class', '') != 'pageMeta':
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
79 logging.error("processPageInfo: pageMeta div not found!")
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
80 return
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
81
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
82 for div in alldivs:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
83 dc = div.get('class')
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
84
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
85 # pageNumberOrig
469
15394486ab75 working with new templates
casties
parents: 465
diff changeset
86 if dc == 'pageNumberOrig':
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
87 pageinfo['pageNumberOrig'] = div.text
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
88
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
89 # pageNumberOrigNorm
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
90 elif dc == 'pageNumberOrigNorm':
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
91 pageinfo['pageNumberOrigNorm'] = div.text
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
92
474
0bc4a153863a more cleanup (before reversing macro order)
casties
parents: 473
diff changeset
93 # pageHeaderTitle
0bc4a153863a more cleanup (before reversing macro order)
casties
parents: 473
diff changeset
94 elif dc == 'pageHeaderTitle':
0bc4a153863a more cleanup (before reversing macro order)
casties
parents: 473
diff changeset
95 pageinfo['pageHeaderTitle'] = div.text
0bc4a153863a more cleanup (before reversing macro order)
casties
parents: 473
diff changeset
96
0bc4a153863a more cleanup (before reversing macro order)
casties
parents: 473
diff changeset
97 # numFigureEntries
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
98 elif dc == 'countFigureEntries':
469
15394486ab75 working with new templates
casties
parents: 465
diff changeset
99 docinfo['numFigureEntries'] = getInt(div.text)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
100
474
0bc4a153863a more cleanup (before reversing macro order)
casties
parents: 473
diff changeset
101 # numTocEntries
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
102 elif dc == 'countTocEntries':
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
103 # WTF: s1 = int(s)/30+1
469
15394486ab75 working with new templates
casties
parents: 465
diff changeset
104 docinfo['numTocEntries'] = getInt(div.text)
15394486ab75 working with new templates
casties
parents: 465
diff changeset
105
475
5a3ab27385ce more new template stuff
casties
parents: 474
diff changeset
106 # numPlaces
5a3ab27385ce more new template stuff
casties
parents: 474
diff changeset
107 elif dc == 'countPlaces':
5a3ab27385ce more new template stuff
casties
parents: 474
diff changeset
108 docinfo['numPlaces'] = getInt(div.text)
5a3ab27385ce more new template stuff
casties
parents: 474
diff changeset
109
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
110 # numTextPages
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
111 elif dc == 'countPages':
458
48b135b089c8 more renovation
casties
parents: 456
diff changeset
112 np = getInt(div.text)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
113 if np > 0:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
114 docinfo['numTextPages'] = np
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
115 if docinfo.get('numPages', 0) == 0:
465
224aad394350 really works with new getDocinfo
casties
parents: 458
diff changeset
116 # seems to be text-only - update page count
224aad394350 really works with new getDocinfo
casties
parents: 458
diff changeset
117 docinfo['numPages'] = np
477
17f0290b2327 small fixes.
casties
parents: 476
diff changeset
118 #pageinfo['end'] = min(pageinfo['end'], np)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
119 pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
120 if np % pageinfo['groupsize'] > 0:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
121 pageinfo['numgroups'] += 1
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
122
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
123 #logging.debug("processPageInfo: pageinfo=%s"%repr(pageinfo))
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
124 return
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
125
388
0265ab93716a *** empty log message ***
abukhman
parents: 386
diff changeset
126
471
415a7026eeda split viewMode in viewMode and viewType
casties
parents: 469
diff changeset
127 def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
128 """returns single page from fulltext"""
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
129
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
130 logging.debug("getTextPage mode=%s, pn=%s"%(mode,pn))
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
131 # check for cached text -- but ideally this shouldn't be called twice
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
132 if pageinfo.has_key('textPage'):
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
133 logging.debug("getTextPage: using cached text")
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
134 return pageinfo['textPage']
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
135
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
136 docpath = docinfo['textURLPath']
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
137 # just checking
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
138 if pageinfo['current'] != pn:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
139 logging.warning("getTextPage: current!=pn!")
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
140
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
141 # stuff for constructing full urls
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
142 characterNormalization = pageinfo.get('characterNormalization', None)
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
143 moreTextParam = ''
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
144 selfurl = docinfo['viewerUrl']
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
145
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
146 if not mode:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
147 # default is dict
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
148 mode = 'text'
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
149
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
150 modes = mode.split(',')
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
151 # check for multiple layers
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
152 if len(modes) > 1:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
153 logging.debug("getTextPage: more than one mode=%s"%mode)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
154
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
155 # search mode
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
156 if 'search' in modes:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
157 # add highlighting
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
158 highlightQuery = pageinfo.get('highlightQuery', None)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
159 sn = pageinfo.get('sn', None)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
160 if highlightQuery and sn:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
161 moreTextParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
162
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
163 # remove mode
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
164 modes.remove('search')
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
165
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
166 # other modes don't combine
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
167 if 'dict' in modes:
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
168 # dict is called textPollux in the backend
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
169 textmode = 'textPollux'
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
170 elif len(modes) == 0:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
171 # text is default mode
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
172 textmode = 'text'
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
173 else:
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
174 # just take first mode
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
175 textmode = modes[0]
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
176
373
0aa294b31236 *** empty log message ***
abukhman
parents: 372
diff changeset
177 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
178 textParam += moreTextParam
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
179
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
180 # fetch the page
185
6fb9a06b8bee *** empty log message ***
abukhman
parents: 184
diff changeset
181 pagexml = self.getServerData("page-fragment.xql",textParam)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
182 dom = ET.fromstring(pagexml)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
183 # extract additional info
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
184 self.processPageInfo(dom, docinfo, pageinfo)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
185 # page content is in <div class="pageContent">
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
186 pagediv = None
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
187 # ElementTree 1.2 in Python 2.6 can't do div[@class='pageContent']
469
15394486ab75 working with new templates
casties
parents: 465
diff changeset
188 # so we look at the second level divs
15394486ab75 working with new templates
casties
parents: 465
diff changeset
189 alldivs = dom.findall("div")
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
190 for div in alldivs:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
191 dc = div.get('class')
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
192 # page content div
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
193 if dc == 'pageContent':
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
194 pagediv = div
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
195 break
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
196
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
197 # plain text mode
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
198 if textmode == "text":
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
199 # get full url assuming documentViewer is parent
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
200 selfurl = self.getLink()
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
201 if pagediv is not None:
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
202 links = pagediv.findall(".//a")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
203 for l in links:
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
204 href = l.get('href')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
205 if href and href.startswith('#note-'):
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
206 href = href.replace('#note-',"%s#note-"%selfurl)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
207 l.set('href', href)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
208
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
209 return serialize(pagediv)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
210
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
211 # text-with-links mode
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
212 elif textmode == "textPollux":
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
213 if pagediv is not None:
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
214 viewerurl = docinfo['viewerUrl']
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
215 selfurl = self.getLink()
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
216 # check all a-tags
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
217 links = pagediv.findall(".//a")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
218 for l in links:
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
219 href = l.get('href')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
220
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
221 if href:
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
222 # is link with href
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
223 if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql'):
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
224 # is dictionary link - change href (keeping parameters)
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
225 l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/template/viewer_wordinfo'%viewerurl))
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
226 # add target to open new page
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
227 l.set('target', '_blank')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
228
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
229 # TODO: is this needed?
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
230 if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql'):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
231 selfurl = self.absolute_url()
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
232 l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
233 l.set('target', '_blank')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
234 l.set('onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
235 l.set('ondblclick', 'popupWin.focus();')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
236
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
237 if href.startswith('#note-'):
478
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
238 # note link
cd37d6f8d5e8 more cleanup
casties
parents: 477
diff changeset
239 l.set('href', href.replace('#note-',"%s#note-"%selfurl))
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
240
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
241 return serialize(pagediv)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
242
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
243 # xml mode
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
244 elif textmode == "xml":
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
245 if pagediv is not None:
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
246 return serialize(pagediv)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
247
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
248 # pureXml mode
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
249 elif textmode == "pureXml":
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
250 if pagediv is not None:
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
251 return serialize(pagediv)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
252
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
253 # gis mode
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
254 elif textmode == "gis":
473
74e9e74277e9 smaller improvements
casties
parents: 471
diff changeset
255 if pagediv is not None:
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
256 # check all a-tags
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
257 links = pagediv.findall(".//a")
506
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
258 # add our URL as backlink
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
259 selfurl = self.getLink()
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
260 doc = base64.b64encode(selfurl)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
261 for l in links:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
262 href = l.get('href')
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
263 if href:
506
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
264 if href.startswith('http://mappit.mpiwg-berlin.mpg.de'):
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
265 l.set('href', re.sub(r'doc=[\w+/=]+', 'doc=%s'%doc, href))
67014399894d cleaned out all 4suite code and weird methods.
casties
parents: 501
diff changeset
266 l.set('target', '_blank')
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
267
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
268 return serialize(pagediv)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
269
501
29c6d09a506c more cleanup.
casties
parents: 482
diff changeset
270 return None
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
271
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
272
509
9d05befdd462 try to get characterNormalization in search result working.
casties
parents: 508
diff changeset
273 def getSearchResults(self, mode, query=None, pageinfo=None, docinfo=None):
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
274 """loads list of search results and stores XML in docinfo"""
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
275 logging.debug("getSearchResults mode=%s query=%s"%(mode, query))
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
276 if mode == "none":
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
277 return docinfo
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
278
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
279 if 'resultSize_%s_%s'%(mode,query) in docinfo:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
280 # cached result
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
281 return docinfo
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
282
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
283 docpath = docinfo['textURLPath']
509
9d05befdd462 try to get characterNormalization in search result working.
casties
parents: 508
diff changeset
284
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
285 # fetch full results
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
286 params = {'document': docpath,
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
287 'mode': 'text',
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
288 'queryType': mode,
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
289 'query': query,
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
290 'queryResultPageSize': 1000,
509
9d05befdd462 try to get characterNormalization in search result working.
casties
parents: 508
diff changeset
291 'queryResultPN': 1,
9d05befdd462 try to get characterNormalization in search result working.
casties
parents: 508
diff changeset
292 'characterNormalization': pageinfo.get('characterNormalization', 'reg')}
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
293 pagexml = self.getServerData("doc-query.xql",urllib.urlencode(params))
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
294 #pagexml = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&s=%s&viewMode=%s&characterNormalization=%s&highlightElementPos=%s&highlightElement=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, s, viewMode,characterNormalization, highlightElementPos, highlightElement, urllib.quote(highlightQuery)))
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
295 dom = ET.fromstring(pagexml)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
296 # page content is in <div class="queryResultPage">
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
297 pagediv = None
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
298 # ElementTree 1.2 in Python 2.6 can't do div[@class='queryResultPage']
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
299 alldivs = dom.findall("div")
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
300 for div in alldivs:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
301 dc = div.get('class')
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
302 # page content div
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
303 if dc == 'queryResultPage':
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
304 pagediv = div
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
305
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
306 elif dc == 'queryResultHits':
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
307 docinfo['resultSize_%s_%s'%(mode,query)] = getInt(div.text)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
308
510
4fb35343d2e7 more search. nicer css.
casties
parents: 509
diff changeset
309 if pagediv is not None:
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
310 # store XML in docinfo
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
311 docinfo['resultXML_%s_%s'%(mode,query)] = ET.tostring(pagediv, 'UTF-8')
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
312
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
313 return docinfo
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
314
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
315
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
316 def getResultsPage(self, mode="text", query=None, pn=None, start=None, size=None, pageinfo=None, docinfo=None):
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
317 """returns single page from the table of contents"""
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
318 logging.debug("getResultsPage mode=%s, pn=%s"%(mode,pn))
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
319 # check for cached TOC
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
320 #TODO: cache only one search
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
321 if not docinfo.has_key('resultXML_%s_%s'%(mode,query)):
509
9d05befdd462 try to get characterNormalization in search result working.
casties
parents: 508
diff changeset
322 self.getSearchResults(mode=mode, query=query, pageinfo=pageinfo, docinfo=docinfo)
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
323
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
324 resultxml = docinfo.get('resultXML_%s_%s'%(mode,query), None)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
325 if not resultxml:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
326 logging.error("getResultPage: unable to find resultXML")
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
327 return "Error: no result!"
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
328
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
329 if size is None:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
330 size = pageinfo.get('searchResultPageSize', 20)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
331
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
332 if start is None:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
333 start = (pn - 1) * size
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
334
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
335 fullresult = ET.fromstring(resultxml)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
336
510
4fb35343d2e7 more search. nicer css.
casties
parents: 509
diff changeset
337 if fullresult is not None:
508
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
338 # paginate
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
339 first = start
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
340 len = size
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
341 del fullresult[:first]
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
342 del fullresult[len:]
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
343 tocdivs = fullresult
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
344
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
345 # check all a-tags
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
346 links = tocdivs.findall(".//a")
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
347 for l in links:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
348 href = l.get('href')
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
349 if href:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
350 # take pn from href
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
351 m = re.match(r'page-fragment\.xql.*pn=(\d+)', href)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
352 if m is not None:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
353 # and create new url (assuming parent is documentViewer)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
354 #TODO: add highlighting params
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
355 url = self.getLink('pn', m.group(1))
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
356 l.set('href', url)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
357 else:
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
358 logging.warning("getResultPage: Problem with link=%s"%href)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
359
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
360 # fix two-divs-per-row with containing div
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
361 # newtoc = ET.Element('div', {'class':'queryResultPage'})
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
362 # for (d1,d2) in zip(tocdivs[::2],tocdivs[1::2]):
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
363 # e = ET.Element('div',{'class':'tocline'})
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
364 # e.append(d1)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
365 # e.append(d2)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
366 # newtoc.append(e)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
367
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
368 return serialize(tocdivs)
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
369
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
370 return "ERROR: no results!"
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
371
d5a47f82e755 more cleanup.
casties
parents: 506
diff changeset
372
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
373 def getToc(self, mode="text", docinfo=None):
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
374 """loads table of contents and stores XML in docinfo"""
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
375 logging.debug("getToc mode=%s"%mode)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
376 if mode == "none":
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
377 return docinfo
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
378
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
379 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
380 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
381 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
382
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
383 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
384 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
385 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
386 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
387 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
388 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
389 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
390 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
391 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
392 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
393 tocDiv = None
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
394 # fetch full toc
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
395 pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
396 dom = ET.fromstring(pagexml)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
397 # page content is in <div class="queryResultPage">
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
398 pagediv = None
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
399 # ElementTree 1.2 in Python 2.6 can't do div[@class='queryResultPage']
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
400 alldivs = dom.findall("div")
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
401 for div in alldivs:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
402 dc = div.get('class')
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
403 # page content div
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
404 if dc == 'queryResultPage':
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
405 pagediv = div
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
406
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
407 elif dc == 'queryResultHits':
458
48b135b089c8 more renovation
casties
parents: 456
diff changeset
408 docinfo['tocSize_%s'%mode] = getInt(div.text)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
409
510
4fb35343d2e7 more search. nicer css.
casties
parents: 509
diff changeset
410 if pagediv is not None:
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
411 # store XML in docinfo
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
412 docinfo['tocXML_%s'%mode] = ET.tostring(pagediv, 'UTF-8')
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
413
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
414 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
415
482
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
416 def getTocPage(self, mode="text", pn=None, start=None, size=None, pageinfo=None, docinfo=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
417 """returns single page from the table of contents"""
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
418 logging.debug("getTocPage mode=%s, pn=%s"%(mode,pn))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
419 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
420 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
421 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
422 queryType = mode
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
423
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
424 # check for cached TOC
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
425 if not docinfo.has_key('tocXML_%s'%mode):
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
426 self.getToc(mode=mode, docinfo=docinfo)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
427
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
428 tocxml = docinfo.get('tocXML_%s'%mode, None)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
429 if not tocxml:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
430 logging.error("getTocPage: unable to find tocXML")
482
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
431 return "Error: no table of contents!"
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
432
482
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
433 if size is None:
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
434 size = pageinfo.get('tocPageSize', 30)
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
435
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
436 if start is None:
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
437 start = (pn - 1) * size
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
438
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
439 fulltoc = ET.fromstring(tocxml)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
440
510
4fb35343d2e7 more search. nicer css.
casties
parents: 509
diff changeset
441 if fulltoc is not None:
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
442 # paginate
482
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
443 first = (start - 1) * 2
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
444 len = size * 2
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
445 del fulltoc[:first]
456
b27a7d2f06ff even toc pagination works
casties
parents: 455
diff changeset
446 del fulltoc[len:]
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
447 tocdivs = fulltoc
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
448
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
449 # check all a-tags
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
450 links = tocdivs.findall(".//a")
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
451 for l in links:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
452 href = l.get('href')
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
453 if href:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
454 # take pn from href
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
455 m = re.match(r'page-fragment\.xql.*pn=(\d+)', href)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
456 if m is not None:
476
1d93a8cb2d8f more new template stuff
casties
parents: 475
diff changeset
457 # and create new url (assuming parent is documentViewer)
1d93a8cb2d8f more new template stuff
casties
parents: 475
diff changeset
458 url = self.getLink('pn', m.group(1))
1d93a8cb2d8f more new template stuff
casties
parents: 475
diff changeset
459 l.set('href', url)
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
460 else:
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
461 logging.warning("getTocPage: Problem with link=%s"%href)
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
462
482
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
463 # fix two-divs-per-row with containing div
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
464 newtoc = ET.Element('div', {'class':'queryResultPage'})
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
465 for (d1,d2) in zip(tocdivs[::2],tocdivs[1::2]):
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
466 e = ET.Element('div',{'class':'tocline'})
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
467 e.append(d1)
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
468 e.append(d2)
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
469 newtoc.append(e)
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
470
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
471 return serialize(newtoc)
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
472
7ca8ac7db06e more new template stuff. more batching methods in documentViewer.
casties
parents: 478
diff changeset
473 return "ERROR: no table of contents!"
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
474
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
475
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
476 def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
477 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
478 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
479 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
480 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
481 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
482 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
483
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
484 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
485 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
486 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
487 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
488 return pt()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
489
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
490 def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
491 #def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
492 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
493 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
494 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
495 if RESPONSE is not None:
455
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
496 RESPONSE.redirect('manage_main')
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
497
0a53fea83df7 more work renovating
casties
parents: 453
diff changeset
498