annotate MpdlXmlTextServer.py @ 431:630a4581992a

*** empty log message ***
author abukhman
date Tue, 15 Feb 2011 12:12:39 +0100
parents 672f8857ece8
children 8d8f2cbd6213
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
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
5
427
70db19d7b3b1 *** empty log message ***
abukhman
parents: 426
diff changeset
6 import md5
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
7 import sys
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
8 import logging
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
9 import urllib
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
10 import documentViewer
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
11 from documentViewer import getTextFromNode, serializeNode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
12
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
13 class MpdlXmlTextServer(SimpleItem):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
14 """TextServer implementation for MPDL-XML eXist server"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
15 meta_type="MPDL-XML TextServer"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
16
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
17 manage_options=(
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
18 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
19 )+SimpleItem.manage_options
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
20
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
21 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
22
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
23 def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
24 """constructor"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
25 self.id=id
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
26 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
27 self.timeout = timeout
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
28 if serverName is None:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
29 self.serverUrl = serverUrl
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
30 else:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
31 self.serverUrl = "http://%s/mpdl/interface/"%serverName
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
32
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
33 def getHttpData(self, url, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
34 """returns result from url+data HTTP request"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
35 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
36
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
37 def getServerData(self, method, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
38 """returns result from text server for method+data"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
39 url = self.serverUrl+method
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
40 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
41
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
42 def getSearch(self, pageinfo=None, docinfo=None, lemma=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
43 """get search list"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
44 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
45 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
46 pagesize = pageinfo['queryPageSize']
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
47 pn = pageinfo.get('searchPN',1)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
48 sn = pageinfo['sn']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
49 highlightQuery = pageinfo['highlightQuery']
181
bc44cf97dfdc *** empty log message ***
abukhman
parents: 180
diff changeset
50 query =pageinfo['query']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
51 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
52 viewMode= pageinfo['viewMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
53 tocMode = pageinfo['tocMode']
169
21ac9164eedf characterNormalization
abukhman
parents: 167
diff changeset
54 characterNormalization = pageinfo['characterNormalization']
399
8abb09665ae5 *** empty log message ***
abukhman
parents: 398
diff changeset
55 optionToggle = pageinfo['optionToggle']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
56 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
57 selfurl = self.absolute_url()
399
8abb09665ae5 *** empty log message ***
abukhman
parents: 398
diff changeset
58 data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&optionToggle=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization,optionToggle ,urllib.quote(highlightQuery)))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
59 pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
60 pagedom = Parse(pagexml)
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
61
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
62 """
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
63 pagedivs = pagedom.xpath("//div[@class='queryResultHits']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
64 if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
65 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
66 docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0]))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
67 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
68 s1 = int(s)/10+1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
69 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
70 docinfo['queryResultHits'] = int(s1)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
71 logging.debug("SEARCH ENTRIES: %s"%(s1))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
72 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
73 docinfo['queryResultHits'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
74 """
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
75 if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
76 pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
77 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
78 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
79 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
80 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
81 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
82 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
83 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
84 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
85 selfurl = self.absolute_url()
407
1cea48640992 *** empty log message ***
abukhman
parents: 406
diff changeset
86 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
87 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
428
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
88 logging.debug("PUREXML :%s"%(serializeNode(pagenode)))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
89 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
90 if (queryType=="fulltextMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
91 pagedivs = pagedom.xpath("//div[@class='queryResult']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
92 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
93 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
94 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
95 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
96 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
97 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
98 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
99 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
100 selfurl = self.absolute_url()
407
1cea48640992 *** empty log message ***
abukhman
parents: 406
diff changeset
101 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
102 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
103 if href.startswith('../lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
104 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
105 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
106 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
138
e6beb338f85f onDblclick to onClick
abukhman
parents: 134
diff changeset
107 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
108 pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
109 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
110 if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
111 pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
112 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
113 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
114 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
115 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
116 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
117 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
118 href = hrefNode.nodeValue
407
1cea48640992 *** empty log message ***
abukhman
parents: 406
diff changeset
119 hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,optionToggle,characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
120 if href.startswith('../lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
121 hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
122 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
123 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
138
e6beb338f85f onDblclick to onClick
abukhman
parents: 134
diff changeset
124 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
125 if href.startswith('../lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
126 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
127 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
128 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
138
e6beb338f85f onDblclick to onClick
abukhman
parents: 134
diff changeset
129 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
130 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
131 return "no text here"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
132
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
133
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
134 def getGisPlaces(self, docinfo=None, pageinfo=None):
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
135 """ Show all Gis Places of whole Page"""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
136 xpath='//place'
409
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
137 docpath = docinfo.get('textURLPath',None)
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
138 if not docpath:
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
139 return None
f7d73ea4b38b quick fix for broken texturlpath
casties
parents: 407
diff changeset
140
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
141 url = docinfo['url']
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
142 selfurl = self.absolute_url()
257
d1eb63cf3410 *** empty log message ***
abukhman
parents: 256
diff changeset
143 pn = pageinfo['current']
291
3b54df7ca88e *** empty log message ***
abukhman
parents: 290
diff changeset
144 hrefList=[]
306
ec293c61f8a9 *** empty log message ***
abukhman
parents: 305
diff changeset
145 myList= ""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
146 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
147 dom = Parse(text)
265
4a0dd5ff6859 *** empty log message ***
abukhman
parents: 264
diff changeset
148 result = dom.xpath("//result/resultPage/place")
236
ccdce5aca47b *** empty log message ***
abukhman
parents: 233
diff changeset
149 for l in result:
250
7775d87016b5 *** empty log message ***
abukhman
parents: 249
diff changeset
150 hrefNode= l.getAttributeNodeNS(None, u"id")
272
7a52ace035e6 *** empty log message ***
abukhman
parents: 271
diff changeset
151 href= hrefNode.nodeValue
292
d97ab809de6f *** empty log message ***
abukhman
parents: 291
diff changeset
152 hrefList.append(href)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
153 myList = ",".join(hrefList)
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
154 logging.debug("getGisPlaces :%s"%(myList))
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
155 return myList
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
156
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
157 def getAllGisPlaces (self, docinfo=None, pageinfo=None):
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
158 """Show all Gis Places of whole Book """
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
159 xpath ='//echo:place'
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
160 docpath =docinfo['textURLPath']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
161 url = docinfo['url']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
162 selfurl =self.absolute_url()
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
163 pn =pageinfo['current']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
164 hrefList=[]
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
165 myList=""
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
166 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
167 dom =Parse(text)
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
168 result = dom.xpath("//result/resultPage/place")
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
169
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
170 for l in result:
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
171 hrefNode = l.getAttributeNodeNS(None, u"id")
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
172 href= hrefNode.nodeValue
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
173 hrefList.append(href)
300
e67f50a09259 *** empty log message ***
abukhman
parents: 299
diff changeset
174 myList = ",".join(hrefList)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
175 logging.debug("getALLGisPlaces :%s"%(myList))
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
176 return myList
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
177
411
691f0c341b33 *** empty log message ***
abukhman
parents: 409
diff changeset
178
428
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
179 def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
180 """returns single page from fulltext"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
181 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
182 path = docinfo['textURLPath']
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
183 url = docinfo.get('url',None)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
184 name = docinfo.get('name',None)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
185 pn =pageinfo['current']
426
a935129e169f *** empty log message ***
abukhman
parents: 419
diff changeset
186 viewMode= pageinfo['viewMode']
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
187 sn = pageinfo['sn']
367
f991ddd3a045 *** empty log message ***
abukhman
parents: 364
diff changeset
188 highlightQuery = pageinfo['highlightQuery']
428
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
189 #mode = pageinfo ['viewMode']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
190 tocMode = pageinfo['tocMode']
164
530f6983f72d characterNormalization
abukhman
parents: 162
diff changeset
191 characterNormalization=pageinfo['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
192 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
193 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
194 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
195 textmode = "textPollux"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
196 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
197 textmode = mode
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
198
373
0aa294b31236 *** empty log message ***
abukhman
parents: 372
diff changeset
199 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
370
33adbe140534 *** empty log message ***
abukhman
parents: 369
diff changeset
200 if highlightQuery is not None:
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
201 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
202
185
6fb9a06b8bee *** empty log message ***
abukhman
parents: 184
diff changeset
203 pagexml = self.getServerData("page-fragment.xql",textParam)
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
204 dom = Parse(pagexml)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
205
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
206 #original Pages
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
207 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
208 if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
209 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
210 docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
211 logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrig']))
431
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
212
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
213 #original Pages Norm
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
214 pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
215 if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
216 if len(pagedivs)>0:
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
217 docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
218 logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrigNorm']))
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
219
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
220
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
221 #figureEntries
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
222 pagedivs = dom.xpath("//div[@class='countFigureEntries']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
223 if pagedivs == dom.xpath("//div[@class='countFigureEntries']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
224 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
225 docinfo['countFigureEntries'] = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
226 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
227 if s=='0':
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
228 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
229 docinfo['countFigureEntries'] = int(s)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
230 logging.debug("FIGURE ENTRIES: %s"%(s))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
231 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
232 docinfo['countFigureEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
233 else:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
234 s1 = int(s)/30+1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
235 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
236 docinfo['countFigureEntries'] = int(s1)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
237 logging.debug("FIGURE ENTRIES: %s"%(s1))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
238 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
239 docinfo['countFigureEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
240
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
241 #allPlaces
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
242 pagedivs = dom.xpath("//div[@class='countPlaces']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
243 if pagedivs == dom.xpath("//div[@class='countPlaces']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
244 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
245 docinfo['countPlaces']= getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
246 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
247 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
248 docinfo['countPlaces'] = int(s)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
249 logging.debug("PLACES HERE: %s"%(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 logging.debug("TEXT ENTRIES: %s"%(s))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
263 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
264 docinfo['countTocEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
265 else:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
266 s1 = int(s)/30+1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
267 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
268 docinfo['countTocEntries'] = int(s1)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
269 logging.debug("TEXT ENTRIES: %s"%(s1))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
270 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
271 docinfo['countTocEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
272
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
273 #numTextPages
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
274 pagedivs = dom.xpath("//div[@class='countPages']")
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
275 if pagedivs == dom.xpath("//div[@class='countPages']"):
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
276 if len(pagedivs)>0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
277 docinfo['numPages'] = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
278 s = getTextFromNode(pagedivs[0])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
279
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
280 try:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
281 docinfo['numPages'] = int(s)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
282 logging.debug("PAGE NUMBER: %s"%(s))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
283
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
284 np = docinfo['numPages']
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
285 pageinfo['end'] = min(pageinfo['end'], np)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
286 pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
287 if np % pageinfo['groupsize'] > 0:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
288 pageinfo['numgroups'] += 1
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
289 except:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
290 docinfo['numPages'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
291
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
292 else:
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
293 #no full text -- init to 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
294 docinfo['pageNumberOrig'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
295 docinfo['countFigureEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
296 docinfo['countPlaces'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
297 docinfo['countTocEntries'] = 0
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
298 docinfo['numPages'] = 0
431
630a4581992a *** empty log message ***
abukhman
parents: 428
diff changeset
299 docinfo['pageNumberOrigNorm'] = 0
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
300 #return docinfo
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
301
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
302 # plain text mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
303 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
304 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
305 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
306 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
307 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
308 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
309 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
310 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
311 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
312 href= hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
313 if href.startswith('#note-'):
172
11cb88e7f523 characterNormalization
abukhman
parents: 171
diff changeset
314 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
315 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
316 if mode == "xml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
317 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
318 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
319 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
320 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
321 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
322 if mode == "gis":
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
323 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
324 pagedivs = dom.xpath("/div")
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
325 if len(pagedivs) > 0:
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
326 pagenode = pagedivs[0]
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
327 links =pagenode.xpath("//a")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
328 for l in links:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
329 hrefNode =l.getAttributeNodeNS(None, u"href")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
330 if hrefNode:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
331 href=hrefNode.nodeValue
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
332 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
239
1f3badb3f1e2 *** empty log message ***
abukhman
parents: 238
diff changeset
333 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
334 l.setAttributeNS(None, 'target', '_blank')
210
50a666145096 *** empty log message ***
abukhman
parents: 209
diff changeset
335 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
336
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
337 if mode == "pureXml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
338 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
339 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
340 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
341 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
342 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
343 # text-with-links mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
344 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
345 # first div contains text
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
346 pagedivs = dom.xpath("/div")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
347 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
348 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
349 # check all a-tags
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
350 links = pagenode.xpath("//a")
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")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
353 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
354 # is link with href
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
355 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
356 if href.startswith('lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
357 # is pollux link
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
358 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
359 # change href
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
360 hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
361 # add target
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
362 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
363 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
138
e6beb338f85f onDblclick to onClick
abukhman
parents: 134
diff changeset
364 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
365 if href.startswith('lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
366 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
367 hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
368 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
369 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
138
e6beb338f85f onDblclick to onClick
abukhman
parents: 134
diff changeset
370 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
371 if href.startswith('#note-'):
162
e6f0553ad176 characterNormalization
abukhman
parents: 161
diff changeset
372 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
373 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
374 return "no text here"
428
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
375
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
376 def getText(self, mode="text", pn=1, docinfo=None, pageinfo=None):
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
377 data = self.getTextPage(mode="text", pn=1 ,docinfo=docinfo, pageinfo=pageinfo)
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
378 logging.debug("TEXT ALL data: %s"%(data))
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
379 return data
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
380
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
381 def getXML(self, mode="xml", pn=1, docinfo=None, pageinfo=None):
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
382 data = self.getTextPage(mode="xml", pn=1 ,docinfo=docinfo, pageinfo=pageinfo)
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
383 logging.debug("TEXT ALL data: %s"%(data))
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
384 return data
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
385
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
386 def getDictionary(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None):
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
387 data = self.getTextPage(mode="text_dict", pn=1 ,docinfo=docinfo, pageinfo=pageinfo)
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
388 logging.debug("TEXT ALL data: %s"%(data))
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
389 return data
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
390
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
391 def getTranslate(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
392 """translate into another languages"""
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
393 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
394 #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
395 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
396
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
397 def getLemma(self, lemma=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
398 """simular words lemma """
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
399 data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
400 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
401 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
402
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
403 def getLemmaNew(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
404 """simular words lemma """
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
405 data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
406 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
407 return data
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
408
398
55f9eeebc338 *** empty log message ***
abukhman
parents: 395
diff changeset
409 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None):
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
410 #number of
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
411 docpath = docinfo['textURLPath']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
412 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
413 pn = pageinfo['searchPN']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
414 query =pageinfo['query']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
415 queryType =pageinfo['queryType']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
416 tocSearch = 0
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
417 tocDiv = None
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
418
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
419 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
420 pagedom = Parse(pagexml)
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
421 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
422 tocSearch = int(getTextFromNode(numdivs[0]))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
423 logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch))
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
424 tc=int((tocSearch/10)+1)
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
425 logging.debug("documentViewer (gettoc) tc: %s"%(tc))
402
d3bd3c2674ee *** empty log message ***
abukhman
parents: 401
diff changeset
426 return tc
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
427
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
428 def getToc(self, mode="text", docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
429 """loads table of contents and stores in docinfo"""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
430 logging.debug("documentViewer (gettoc) mode: %s"%(mode))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
431 if mode == "none":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
432 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
433 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
434 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
435 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
436
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
437 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
438 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
439 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
440 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
441 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
442 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
443 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
444 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
445 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
446 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
447 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
448
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
449 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
450
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
451 # post-processing downloaded xml
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
452 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
453 # get number of entries
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
454 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
455 if len(numdivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
456 tocSize = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
457 docinfo['tocSize_%s'%mode] = tocSize
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
458 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
459
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
460 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
461 """returns single page from the table of contents"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
462 # TODO: this should use the cached TOC
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
463 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
464 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
465 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
466 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
467 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
468 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
469 pagesize = pageinfo['tocPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
470 pn = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
471 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
472 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
473 viewMode= pageinfo['viewMode']
171
77d5ff742bb4 characterNormalization
abukhman
parents: 170
diff changeset
474 characterNormalization = pageinfo ['characterNormalization']
398
55f9eeebc338 *** empty log message ***
abukhman
parents: 395
diff changeset
475 optionToggle =pageinfo ['optionToggle']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
476 tocMode = pageinfo['tocMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
477 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
478
407
1cea48640992 *** empty log message ***
abukhman
parents: 406
diff changeset
479 data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn))
1cea48640992 *** empty log message ***
abukhman
parents: 406
diff changeset
480 page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&optionToggle=1'%(selfurl,url, viewMode, tocMode, tocPN))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
481 text = page.replace('mode=image','mode=texttool')
419
69205c9d9404 *** empty log message ***
abukhman
parents: 418
diff changeset
482 #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
428
672f8857ece8 bibliograpfical info
abukhman
parents: 427
diff changeset
483 logging.debug("TEXT %s"%(text))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
484 return text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
485
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
486 def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
487 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
488 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
489 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
490 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
491 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
492 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
493
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
494 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
495 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
496 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
497 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
498 return pt()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
499
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
500 def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
501 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
502 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
503 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
504 if RESPONSE is not None:
427
70db19d7b3b1 *** empty log message ***
abukhman
parents: 426
diff changeset
505 RESPONSE.redirect('manage_main')