annotate MpdlXmlTextServer.py @ 445:93b9a5dddb8b Root_elementtree

*** empty log message ***
author abukhman
date Tue, 14 Jun 2011 11:57:11 +0200
parents f160791b0074
children 48f33478acec
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
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
2 from OFS.SimpleItem import SimpleItem
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
3 from Products.PageTemplates.PageTemplateFile import PageTemplateFile
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
4 from Ft.Xml import EMPTY_NAMESPACE, Parse
445
93b9a5dddb8b *** empty log message ***
abukhman
parents: 444
diff changeset
5 from Ft.Xml.Domlette import NonvalidatingReader
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
6
427
70db19d7b3b1 *** empty log message ***
abukhman
parents: 426
diff changeset
7 import md5
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
8 import sys
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
9 import logging
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
10 import urllib
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
11 import documentViewer
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
12 from documentViewer import getTextFromNode, serializeNode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
13
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
14 class MpdlXmlTextServer(SimpleItem):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
15 """TextServer implementation for MPDL-XML eXist server"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
16 meta_type="MPDL-XML TextServer"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
17
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
18 manage_options=(
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
19 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
20 )+SimpleItem.manage_options
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
21
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
22 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
23
443
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
24 def __init__(self,id,title="",serverUrl="http://mpdl-system.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
441
542d0aaf30bb *** empty log message ***
abukhman
parents: 440
diff changeset
25 #def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/", serverName=None, timeout=40):
437
da43c361c0ff *** empty log message ***
abukhman
parents: 436
diff changeset
26
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
27 """constructor"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
28 self.id=id
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
29 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
30 self.timeout = timeout
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
31 if serverName is None:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
32 self.serverUrl = serverUrl
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
33 else:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
34 self.serverUrl = "http://%s/mpdl/interface/"%serverName
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
35
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
36 def getHttpData(self, url, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
37 """returns result from url+data HTTP request"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
38 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
39
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
40 def getServerData(self, method, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
41 """returns result from text server for method+data"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
42 url = self.serverUrl+method
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
43 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
44
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
45 def getSearch(self, pageinfo=None, docinfo=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
46 """get search list"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
47 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
48 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
49 pagesize = pageinfo['queryPageSize']
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
50 pn = pageinfo.get('searchPN',1)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
51 sn = pageinfo['sn']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
52 highlightQuery = pageinfo['highlightQuery']
181
bc44cf97dfdc *** empty log message ***
abukhman
parents: 180
diff changeset
53 query =pageinfo['query']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
54 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
55 viewMode= pageinfo['viewMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
56 tocMode = pageinfo['tocMode']
169
21ac9164eedf characterNormalization
abukhman
parents: 167
diff changeset
57 characterNormalization = pageinfo['characterNormalization']
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
58 #optionToggle = pageinfo['optionToggle']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
59 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
60 selfurl = self.absolute_url()
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
61 data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization, urllib.quote(highlightQuery)))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
62 pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
63 pagedom = Parse(pagexml)
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
64
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
65 """
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
66 pagedivs = pagedom.xpath("//div[@class='queryResultHits']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
67 if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
68 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
69 docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0]))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
70 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
71 s1 = int(s)/10+1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
72 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
73 docinfo['queryResultHits'] = int(s1)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
74 logging.debug("SEARCH ENTRIES: %s"%(s1))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
75 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
76 docinfo['queryResultHits'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
77 """
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
78 if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
79 pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
80 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
81 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
82 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
83 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
84 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
85 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
86 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
87 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
88 selfurl = self.absolute_url()
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
89 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN, characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
90 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
436
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
91 #logging.debug("PUREXML :%s"%(serializeNode(pagenode)))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
92 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
93 if (queryType=="fulltextMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
94 pagedivs = pagedom.xpath("//div[@class='queryResult']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
95 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
96 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
97 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
98 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
99 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
100 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
101 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
102 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
103 selfurl = self.absolute_url()
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
104 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
105 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
106 if href.startswith('../lt/lemma.xql'):
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
107 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_query'%(selfurl))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
108 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
109 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
110 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
111 pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
112 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
113 if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
114 pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
115 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
116 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
117 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
118 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
119 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
120 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
121 href = hrefNode.nodeValue
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
122 hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
123 if href.startswith('../lt/lex.xql'):
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
124 hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_lex'%selfurl)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
125 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
126 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
127 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
128 if href.startswith('../lt/lemma.xql'):
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
129 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%(selfurl))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
130 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
131 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
132 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
133 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
134 return "no text here"
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
135
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
136 def getGisPlaces(self, docinfo=None, pageinfo=None):
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
137 """ Show all Gis Places of whole Page"""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
138 xpath='//place'
409
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
139 docpath = docinfo.get('textURLPath',None)
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
140 if not docpath:
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
141 return None
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
142
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
143 url = docinfo['url']
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
144 selfurl = self.absolute_url()
257
d1eb63cf3410 *** empty log message ***
abukhman
parents: 256
diff changeset
145 pn = pageinfo['current']
291
3b54df7ca88e *** empty log message ***
abukhman
parents: 290
diff changeset
146 hrefList=[]
306
ec293c61f8a9 *** empty log message ***
abukhman
parents: 305
diff changeset
147 myList= ""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
148 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
149 dom = Parse(text)
265
4a0dd5ff6859 *** empty log message ***
abukhman
parents: 264
diff changeset
150 result = dom.xpath("//result/resultPage/place")
236
ccdce5aca47b *** empty log message ***
abukhman
parents: 233
diff changeset
151 for l in result:
250
7775d87016b5 *** empty log message ***
abukhman
parents: 249
diff changeset
152 hrefNode= l.getAttributeNodeNS(None, u"id")
272
7a52ace035e6 *** empty log message ***
abukhman
parents: 271
diff changeset
153 href= hrefNode.nodeValue
292
d97ab809de6f *** empty log message ***
abukhman
parents: 291
diff changeset
154 hrefList.append(href)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
155 myList = ",".join(hrefList)
436
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
156 #logging.debug("getGisPlaces :%s"%(myList))
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
157 return myList
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
158
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
159 def getAllGisPlaces (self, docinfo=None, pageinfo=None):
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
160 """Show all Gis Places of whole Book """
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
161 xpath ='//echo:place'
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
162 docpath =docinfo['textURLPath']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
163 url = docinfo['url']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
164 selfurl =self.absolute_url()
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
165 pn =pageinfo['current']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
166 hrefList=[]
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
167 myList=""
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
168 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
169 dom =Parse(text)
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
170 result = dom.xpath("//result/resultPage/place")
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
171
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
172 for l in result:
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
173 hrefNode = l.getAttributeNodeNS(None, u"id")
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
174 href= hrefNode.nodeValue
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
175 hrefList.append(href)
300
e67f50a09259 *** empty log message ***
abukhman
parents: 299
diff changeset
176 myList = ",".join(hrefList)
436
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
177 #logging.debug("getALLGisPlaces :%s"%(myList))
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
178 return myList
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
179
411
691f0c341b33 *** empty log message ***
abukhman
parents: 409
diff changeset
180
432
8d8f2cbd6213 *** empty log message ***
abukhman
parents: 431
diff changeset
181 def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
182 """returns single page from fulltext"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
183 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
184 path = docinfo['textURLPath']
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
185 url = docinfo.get('url',None)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
186 name = docinfo.get('name',None)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
187 pn =pageinfo['current']
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
188 sn = pageinfo['sn']
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
189 #optionToggle =pageinfo ['optionToggle']
367
f991ddd3a045 *** empty log message ***
abukhman
parents: 364
diff changeset
190 highlightQuery = pageinfo['highlightQuery']
428
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
191 #mode = pageinfo ['viewMode']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
192 tocMode = pageinfo['tocMode']
164
530f6983f72d characterNormalization
abukhman
parents: 162
diff changeset
193 characterNormalization=pageinfo['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
194 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
195 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
196 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
197 textmode = "textPollux"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
198 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
199 textmode = mode
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
200
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
201 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
370
33adbe140534 *** empty log message ***
abukhman
parents: 369
diff changeset
202 if highlightQuery is not None:
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
203 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
204
185
6fb9a06b8bee *** empty log message ***
abukhman
parents: 184
diff changeset
205 pagexml = self.getServerData("page-fragment.xql",textParam)
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
206 dom = Parse(pagexml)
445
93b9a5dddb8b *** empty log message ***
abukhman
parents: 444
diff changeset
207 #dom = NonvalidatingReader.parseStream(pagexml)
443
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
208
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
209 #original Pages
440
3940845b2474 *** empty log message ***
abukhman
parents: 438
diff changeset
210 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
443
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
211
436
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
212 """if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
213 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
214 docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
436
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
215 logging.debug("ORIGINAL PAGE: %s"%(docinfo['pageNumberOrig']))
431
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
216
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
217 #original Pages Norm
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
218 pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
219 if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
220 if len(pagedivs)>0:
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
221 docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
436
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
222 logging.debug("ORIGINAL PAGE NORM: %s"%(docinfo['pageNumberOrigNorm']))
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
223 """
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
224 #figureEntries
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
225 pagedivs = dom.xpath("//div[@class='countFigureEntries']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
226 if pagedivs == dom.xpath("//div[@class='countFigureEntries']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
227 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
228 docinfo['countFigureEntries'] = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
229 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
230 if s=='0':
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
231 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
232 docinfo['countFigureEntries'] = int(s)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
233 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
234 docinfo['countFigureEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
235 else:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
236 s1 = int(s)/30+1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
237 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
238 docinfo['countFigureEntries'] = int(s1)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
239 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
240 docinfo['countFigureEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
241
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
242 #allPlaces
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
243 pagedivs = dom.xpath("//div[@class='countPlaces']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
244 if pagedivs == dom.xpath("//div[@class='countPlaces']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
245 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
246 docinfo['countPlaces']= getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
247 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
248 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
249 docinfo['countPlaces'] = int(s)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
250 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
251 docinfo['countPlaces'] = 0
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
252
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
253 #tocEntries
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
254 pagedivs = dom.xpath("//div[@class='countTocEntries']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
255 if pagedivs == dom.xpath("//div[@class='countTocEntries']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
256 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
257 docinfo['countTocEntries'] = int(getTextFromNode(pagedivs[0]))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
258 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
259 if s=='0':
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
260 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
261 docinfo['countTocEntries'] = int(s)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
262 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
263 docinfo['countTocEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
264 else:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
265 s1 = int(s)/30+1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
266 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
267 docinfo['countTocEntries'] = int(s1)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
268 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
269 docinfo['countTocEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
270
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
271 #numTextPages
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
272 pagedivs = dom.xpath("//div[@class='countPages']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
273 if pagedivs == dom.xpath("//div[@class='countPages']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
274 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
275 docinfo['numPages'] = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
276 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
277
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
278 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
279 docinfo['numPages'] = int(s)
434
44ccb7bd0938 *** empty log message ***
abukhman
parents: 432
diff changeset
280 #logging.debug("PAGE NUMBER: %s"%(s))
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
281
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
282 np = docinfo['numPages']
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
283 pageinfo['end'] = min(pageinfo['end'], np)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
284 pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
285 if np % pageinfo['groupsize'] > 0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
286 pageinfo['numgroups'] += 1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
287 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
288 docinfo['numPages'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
289
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
290 else:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
291 #no full text -- init to 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
292 docinfo['pageNumberOrig'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
293 docinfo['countFigureEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
294 docinfo['countPlaces'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
295 docinfo['countTocEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
296 docinfo['numPages'] = 0
431
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
297 docinfo['pageNumberOrigNorm'] = 0
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
298 #return docinfo
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
299
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
300 # plain text mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
301 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
302 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
303 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
304 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
305 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
306 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
307 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
308 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
309 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
310 href= hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
311 if href.startswith('#note-'):
432
8d8f2cbd6213 *** empty log message ***
abukhman
parents: 431
diff changeset
312 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
313 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
314 if mode == "xml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
315 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
316 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
317 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
318 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
319 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
320 if mode == "gis":
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
321 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
322 pagedivs = dom.xpath("/div")
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
323 if len(pagedivs) > 0:
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
324 pagenode = pagedivs[0]
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
325 links =pagenode.xpath("//a")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
326 for l in links:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
327 hrefNode =l.getAttributeNodeNS(None, u"href")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
328 if hrefNode:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
329 href=hrefNode.nodeValue
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
330 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
239
1f3badb3f1e2 *** empty log message ***
abukhman
parents: 238
diff changeset
331 hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
211
bc6a9fdc1ff8 *** empty log message ***
abukhman
parents: 210
diff changeset
332 l.setAttributeNS(None, 'target', '_blank')
210
50a666145096 *** empty log message ***
abukhman
parents: 209
diff changeset
333 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
334
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
335 if mode == "pureXml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
336 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
337 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
338 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
339 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
340 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
341 # text-with-links mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
342 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
343 # first div contains text
432
8d8f2cbd6213 *** empty log message ***
abukhman
parents: 431
diff changeset
344 #mode = pageinfo ['viewMode']
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
345 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
346 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
347 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
348 # check all a-tags
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
349 links = pagenode.xpath("//a")
443
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
350
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
351 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
352 hrefNode = l.getAttributeNodeNS(None, u"href")
443
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
353
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
354 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
355 # is link with href
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
356 href = hrefNode.nodeValue
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
357 if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql'):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
358 # is pollux link
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
359 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
360 # change href
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
361 hrefNode.nodeValue = href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/head_main_voc'%selfurl)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
362 # add target
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
363 l.setAttributeNS(None, 'target', '_blank')
445
93b9a5dddb8b *** empty log message ***
abukhman
parents: 444
diff changeset
364 #l.setAttributeNS(None, 'onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;")
93b9a5dddb8b *** empty log message ***
abukhman
parents: 444
diff changeset
365 #l.setAttributeNS(None, "ondblclick", "popupWin.focus();")
93b9a5dddb8b *** empty log message ***
abukhman
parents: 444
diff changeset
366 #window.open("this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=yes, scrollbars=1'"); return false;")
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
367
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
368 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
369 selfurl = self.absolute_url()
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
370 hrefNode.nodeValue = href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
371 l.setAttributeNS(None, 'target', '_blank')
443
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
372 l.setAttributeNS(None, 'onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;")
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
373 l.setAttributeNS(None, 'ondblclick', 'popupWin.focus();')
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
374
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
375 if href.startswith('#note-'):
443
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
376 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn))
ad5c3dedfc18 *** empty log message ***
abukhman
parents: 442
diff changeset
377
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
378 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
379 return "no text here"
428
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
380
436
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
381 def getOrigPages(self, docinfo=None, pageinfo=None):
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
382 docpath = docinfo['textURLPath']
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
383 pn =pageinfo['current']
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
384 selfurl = self.absolute_url()
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
385 pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn))
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
386 dom = Parse(pagexml)
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
387 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
388 if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
389 if len(pagedivs)>0:
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
390 docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
391 return docinfo['pageNumberOrig']
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
392
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
393 def getOrigPagesNorm(self, docinfo=None, pageinfo=None):
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
394 docpath = docinfo['textURLPath']
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
395 pn =pageinfo['current']
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
396 selfurl = self.absolute_url()
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
397 pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn))
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
398 dom = Parse(pagexml)
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
399 pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
400 if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
401 if len(pagedivs)>0:
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
402 docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
403 return docinfo['pageNumberOrigNorm']
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
404
0baf4e761be7 *** empty log message ***
abukhman
parents: 435
diff changeset
405
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
406 def getTranslate(self, word=None, language=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
407 """translate into another languages"""
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
408 data = self.getServerData("lt/wordInfo.xql","language="+str(language)+"&word="+urllib.quote(word)+"&output=html")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
409 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
410 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
411
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
412 def getLemma(self, lemma=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
413 """simular words lemma """
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
414 data = self.getServerData("lt/lemma.xql","language="+str(language)+"&lemma="+urllib.quote(lemma)+"&output=html")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
415 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
416
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
417 def getLemmaQuery(self, query=None, language=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
418 """simular words lemma """
442
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
419 data = self.getServerData("lt/lemma.xql","language="+str(language)+"&query="+urllib.quote(query)+"&output=html")
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
420 return data
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
421
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
422 def getLex(self, query=None, language=None):
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
423 #simular words lemma
9148eac0cab3 *** empty log message ***
abukhman
parents: 441
diff changeset
424 data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
425 return data
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
426
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
427 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
428 #number of
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
429 docpath = docinfo['textURLPath']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
430 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
431 pn = pageinfo['searchPN']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
432 query =pageinfo['query']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
433 queryType =pageinfo['queryType']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
434 tocSearch = 0
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
435 tocDiv = None
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
436
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
437 pagexml = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
438 pagedom = Parse(pagexml)
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
439 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
440 tocSearch = int(getTextFromNode(numdivs[0]))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
441 tc=int((tocSearch/10)+1)
402
d3bd3c2674ee *** empty log message ***
abukhman
parents: 401
diff changeset
442 return tc
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
443
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
444 def getToc(self, mode="text", docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
445 """loads table of contents and stores in docinfo"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
446 if mode == "none":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
447 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
448 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
449 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
450 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
451
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
452 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
453 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
454 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
455 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
456 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
457 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
458 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
459 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
460 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
461 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
462 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
463
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
464 pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
465
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
466 # post-processing downloaded xml
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
467 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
468 # get number of entries
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
469 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
470 if len(numdivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
471 tocSize = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
472 docinfo['tocSize_%s'%mode] = tocSize
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
473 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
474
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
475 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
476 """returns single page from the table of contents"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
477 # TODO: this should use the cached TOC
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
478 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
479 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
480 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
481 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
482 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
483 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
484 pagesize = pageinfo['tocPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
485 pn = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
486 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
487 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
488 viewMode= pageinfo['viewMode']
171
77d5ff742bb4 characterNormalization
abukhman
parents: 170
diff changeset
489 characterNormalization = pageinfo ['characterNormalization']
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
490 #optionToggle =pageinfo ['optionToggle']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
491 tocMode = pageinfo['tocMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
492 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
493
444
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
494 data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm"%(docpath,queryType, pagesize, pn))
f160791b0074 *** empty log message ***
abukhman
parents: 443
diff changeset
495 page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
496 text = page.replace('mode=image','mode=texttool')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
497 return text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
498
441
542d0aaf30bb *** empty log message ***
abukhman
parents: 440
diff changeset
499 def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
542d0aaf30bb *** empty log message ***
abukhman
parents: 440
diff changeset
500 #def manage_changeMpdlXmlTextServer(self,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
501 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
502 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
503 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
504 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
505 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
506 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
507
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
508 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
509 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
510 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
511 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
512 return pt()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
513
441
542d0aaf30bb *** empty log message ***
abukhman
parents: 440
diff changeset
514 def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
542d0aaf30bb *** empty log message ***
abukhman
parents: 440
diff changeset
515 #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
516 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
517 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
518 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
519 if RESPONSE is not None:
427
70db19d7b3b1 *** empty log message ***
abukhman
parents: 426
diff changeset
520 RESPONSE.redirect('manage_main')