annotate MpdlXmlTextServer.py @ 395:e6f3fcb2058f

*** empty log message ***
author abukhman
date Mon, 08 Nov 2010 16:04:06 +0100
parents bb317af1c197
children 55f9eeebc338
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
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
6 import sys
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
7 import logging
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
8 import urllib
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
9 import documentViewer
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
10 from documentViewer import getTextFromNode, serializeNode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
11
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
12 class MpdlXmlTextServer(SimpleItem):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
13 """TextServer implementation for MPDL-XML eXist server"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
14 meta_type="MPDL-XML TextServer"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
15
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
16 manage_options=(
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
17 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
18 )+SimpleItem.manage_options
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
19
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
20 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
21
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
22 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
23 """constructor"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
24 self.id=id
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
25 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
26 self.timeout = timeout
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
27 if serverName is None:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
28 self.serverUrl = serverUrl
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
29 else:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
30 self.serverUrl = "http://%s/mpdl/interface/"%serverName
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
31
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
32 def getHttpData(self, url, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
33 """returns result from url+data HTTP request"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
34 return documentViewer.getHttpData(url,data,timeout=self.timeout)
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 getServerData(self, method, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
37 """returns result from text server for method+data"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
38 url = self.serverUrl+method
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
39 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
40
392
9b2adae6c6e2 *** empty log message ***
abukhman
parents: 388
diff changeset
41 def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None, characterNormalization=None, optionsClose=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
42 """get search list"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
43 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
44 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
45 pagesize = pageinfo['queryPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
46 pn = pageinfo['searchPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
47 sn = pageinfo['sn']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
48 highlightQuery = pageinfo['highlightQuery']
181
bc44cf97dfdc *** empty log message ***
abukhman
parents: 180
diff changeset
49 query =pageinfo['query']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
50 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
51 viewMode= pageinfo['viewMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
52 tocMode = pageinfo['tocMode']
169
21ac9164eedf characterNormalization
abukhman
parents: 167
diff changeset
53 characterNormalization = pageinfo['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
54 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
55 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
56
363
c89ae357ae48 *** empty log message ***
abukhman
parents: 361
diff changeset
57 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
58 #page=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, query, pagesize, pn, sn, viewMode,highlightQuery) ,outputUnicode=False)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
59
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
60 pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
61 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
62 if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
63 pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
64 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
65 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
66 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
67 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
68 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
69 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
70 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
71 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
72 selfurl = self.absolute_url()
177
393234132abd urllib.quote
abukhman
parents: 176
diff changeset
73 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
74 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
75 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
76 if (queryType=="fulltextMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
77 pagedivs = pagedom.xpath("//div[@class='queryResult']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
78 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
79 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
80 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
81 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
82 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
83 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
84 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
85 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
86 selfurl = self.absolute_url()
177
393234132abd urllib.quote
abukhman
parents: 176
diff changeset
87 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
88 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
89 if href.startswith('../lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
90 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
91 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
92 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
93 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
94 pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
95 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
96 if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
97 pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
98 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
99 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
100 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
101 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
102 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
103 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
104 href = hrefNode.nodeValue
161
1e1733fbe286 characterNormalization
abukhman
parents: 158
diff changeset
105 hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
106 if href.startswith('../lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
107 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
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;")
138
e6beb338f85f onDblclick to onClick
abukhman
parents: 134
diff changeset
110 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
111 if href.startswith('../lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
112 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
113 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
114 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
115 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
116 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
117 return "no text here"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
118
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
119 """def getNumPages(self, docinfo):
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
120 ""get list of pages from fulltext and put in docinfo""
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
121 if 'numPages' in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
122 # already there
190
c92e6b923913 *** empty log message ***
abukhman
parents: 189
diff changeset
123 return docinfo
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
124 xquery = '//pb'
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
125 text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
126 docinfo['numPages'] = text.count("<pb ")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
127 return docinfo
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
128 """
190
c92e6b923913 *** empty log message ***
abukhman
parents: 189
diff changeset
129 def getNumTextPages (self, docinfo):
c92e6b923913 *** empty log message ***
abukhman
parents: 189
diff changeset
130 """get list of pages from fulltext (texts without images) and put in docinfo"""
205
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
131 if 'numPages' in docinfo:
190
c92e6b923913 *** empty log message ***
abukhman
parents: 189
diff changeset
132 # allredy there
205
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
133 return docinfo
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
134 xpath ='/count(//pb)'
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
135 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
136 dom = Parse(text)
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
137 result= dom.xpath("//result/resultPage")
388
0265ab93716a *** empty log message ***
abukhman
parents: 386
diff changeset
138
205
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
139 docinfo['numPages']=int(getTextFromNode(result[0]))
190
c92e6b923913 *** empty log message ***
abukhman
parents: 189
diff changeset
140 return docinfo
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
141
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
142 def getGisPlaces(self, docinfo=None, pageinfo=None):
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
143 """ Show all Gis Places of whole Page"""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
144 xpath='//place'
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
145 docpath = docinfo['textURLPath']
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
146 url = docinfo['url']
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
147 selfurl = self.absolute_url()
257
d1eb63cf3410 *** empty log message ***
abukhman
parents: 256
diff changeset
148 pn = pageinfo['current']
291
3b54df7ca88e *** empty log message ***
abukhman
parents: 290
diff changeset
149 hrefList=[]
306
ec293c61f8a9 *** empty log message ***
abukhman
parents: 305
diff changeset
150 myList= ""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
151 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
152 dom = Parse(text)
265
4a0dd5ff6859 *** empty log message ***
abukhman
parents: 264
diff changeset
153 result = dom.xpath("//result/resultPage/place")
236
ccdce5aca47b *** empty log message ***
abukhman
parents: 233
diff changeset
154 for l in result:
250
7775d87016b5 *** empty log message ***
abukhman
parents: 249
diff changeset
155 hrefNode= l.getAttributeNodeNS(None, u"id")
272
7a52ace035e6 *** empty log message ***
abukhman
parents: 271
diff changeset
156 href= hrefNode.nodeValue
292
d97ab809de6f *** empty log message ***
abukhman
parents: 291
diff changeset
157 hrefList.append(href)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
158 myList = ",".join(hrefList)
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
159 logging.debug("getGisPlaces :%s"%(myList))
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
160 return myList
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
161
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
162 def getAllGisPlaces (self, docinfo=None, pageinfo=None):
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
163 """Show all Gis Places of whole Book """
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
164 xpath ='//echo:place'
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
165 docpath =docinfo['textURLPath']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
166 url = docinfo['url']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
167 selfurl =self.absolute_url()
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
168 pn =pageinfo['current']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
169 hrefList=[]
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
170 myList=""
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
171 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
172 dom =Parse(text)
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
173 result = dom.xpath("//result/resultPage/place")
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
174
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
175 for l in result:
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
176 hrefNode = l.getAttributeNodeNS(None, u"id")
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
177 href= hrefNode.nodeValue
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
178 hrefList.append(href)
300
e67f50a09259 *** empty log message ***
abukhman
parents: 299
diff changeset
179 myList = ",".join(hrefList)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
180 logging.debug("getALLGisPlaces :%s"%(myList))
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
181 return myList
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
182
385
7a6a2e4476eb *** empty log message ***
abukhman
parents: 384
diff changeset
183
7a6a2e4476eb *** empty log message ***
abukhman
parents: 384
diff changeset
184 def getPDF (self, docinfo=None, pageinfo=None):
7a6a2e4476eb *** empty log message ***
abukhman
parents: 384
diff changeset
185 """Show and Save different Pages as PDF in Options"""
7a6a2e4476eb *** empty log message ***
abukhman
parents: 384
diff changeset
186 selfurl=self.absolute_url()
7a6a2e4476eb *** empty log message ***
abukhman
parents: 384
diff changeset
187 pn=pageinfo['current']
386
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
188
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
189 viewMode =pageinfo['viewMode']
385
7a6a2e4476eb *** empty log message ***
abukhman
parents: 384
diff changeset
190
386
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
191 #text = ("page-fragment.xql","document=%s&mode=%s&pn=%s&export=%s"%(docinfo['textURLPath'], 'text', pn,'pdf'))
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
192 #text = self.getServerData("page-fragment.xql", "document=%s&mode=%s&pn=%s&export=%s"(docinfo['textURLPath'],'text', pn,'pdf'))
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
193 #logging.debug("text :%s"%(text))
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
194 #dom =Parse(text)
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
195 #logging.debug("text :%s"%(text))
388
0265ab93716a *** empty log message ***
abukhman
parents: 386
diff changeset
196 #return text
0265ab93716a *** empty log message ***
abukhman
parents: 386
diff changeset
197
360
a0fd5e95772c *** empty log message ***
abukhman
parents: 359
diff changeset
198 def getOrigPages (self, docinfo=None, pageinfo=None):
310
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
199 """Show original page """
360
a0fd5e95772c *** empty log message ***
abukhman
parents: 359
diff changeset
200 docpath = docinfo['textURLPath']
386
48d27e100d19 *** empty log message ***
abukhman
parents: 385
diff changeset
201 logging.debug ("docinfo['textURLPath']=%s"%(docinfo['textURLPath']))
358
0b660cb2889e *** empty log message ***
abukhman
parents: 357
diff changeset
202 #url = docinfo['url']
310
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
203 selfurl = self.absolute_url()
349
d86fa52ee13c *** empty log message ***
abukhman
parents: 348
diff changeset
204 pn =pageinfo['current']
325
c7e38112c23a *** empty log message ***
abukhman
parents: 324
diff changeset
205
349
d86fa52ee13c *** empty log message ***
abukhman
parents: 348
diff changeset
206 viewMode= pageinfo['viewMode']
d86fa52ee13c *** empty log message ***
abukhman
parents: 348
diff changeset
207 text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn))
310
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
208 dom =Parse(text)
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
209 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
361
c1381556f389 *** empty log message ***
abukhman
parents: 360
diff changeset
210 logging.debug("YYYYYYpagedivs :%s"%(pagedivs))
324
9d69b218f74c *** empty log message ***
abukhman
parents: 323
diff changeset
211 if len(pagedivs)>0:
360
a0fd5e95772c *** empty log message ***
abukhman
parents: 359
diff changeset
212 originalPage= getTextFromNode(pagedivs[0])
358
0b660cb2889e *** empty log message ***
abukhman
parents: 357
diff changeset
213 #return docinfo['originalPage']
355
929211c057eb *** empty log message ***
abukhman
parents: 354
diff changeset
214 return originalPage
360
a0fd5e95772c *** empty log message ***
abukhman
parents: 359
diff changeset
215
361
c1381556f389 *** empty log message ***
abukhman
parents: 360
diff changeset
216
392
9b2adae6c6e2 *** empty log message ***
abukhman
parents: 388
diff changeset
217 def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="", highlightQuery=None, sn=None, optionsClose=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
218 """returns single page from fulltext"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
219 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
220 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
221 url = docinfo['url']
237
d596d6839fb8 *** empty log message ***
abukhman
parents: 236
diff changeset
222 name = docinfo['name']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
223 viewMode= pageinfo['viewMode']
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
224 sn = pageinfo['sn']
367
f991ddd3a045 *** empty log message ***
abukhman
parents: 364
diff changeset
225 highlightQuery = pageinfo['highlightQuery']
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
226
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
227 tocMode = pageinfo['tocMode']
164
530f6983f72d characterNormalization
abukhman
parents: 162
diff changeset
228 characterNormalization=pageinfo['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
229 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
230 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
231 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
232 textmode = "textPollux"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
233 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
234 textmode = mode
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
235 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
373
0aa294b31236 *** empty log message ***
abukhman
parents: 372
diff changeset
236 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
370
33adbe140534 *** empty log message ***
abukhman
parents: 369
diff changeset
237 if highlightQuery is not None:
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
238 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
379
4b1b5dd7578f *** empty log message ***
abukhman
parents: 378
diff changeset
239 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
185
6fb9a06b8bee *** empty log message ***
abukhman
parents: 184
diff changeset
240 pagexml = self.getServerData("page-fragment.xql",textParam)
379
4b1b5dd7578f *** empty log message ***
abukhman
parents: 378
diff changeset
241 logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
242 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
243
186
3b3476df97d4 *** empty log message ***
abukhman
parents: 185
diff changeset
244 pagedom = Parse(pagexml)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
245 # plain text mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
246 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
247 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
248 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
249 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
250 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
251 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
252 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
253 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
254 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
255 href= hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
256 if href.startswith('#note-'):
172
11cb88e7f523 characterNormalization
abukhman
parents: 171
diff changeset
257 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
258 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
259 if mode == "xml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
260 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
261 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
262 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
263 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
264 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
265 if mode == "gis":
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
266 # first div contains text
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
267 pagedivs = pagedom.xpath("/div")
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
268 if len(pagedivs) > 0:
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
269 pagenode = pagedivs[0]
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
270 links =pagenode.xpath("//a")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
271 for l in links:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
272 hrefNode =l.getAttributeNodeNS(None, u"href")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
273 if hrefNode:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
274 href=hrefNode.nodeValue
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
275 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
239
1f3badb3f1e2 *** empty log message ***
abukhman
parents: 238
diff changeset
276 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
277 l.setAttributeNS(None, 'target', '_blank')
210
50a666145096 *** empty log message ***
abukhman
parents: 209
diff changeset
278 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
279
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
280 if mode == "pureXml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
281 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
282 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
283 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
284 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
285 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
286 # text-with-links mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
287 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
288 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
289 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
290 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
291 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
292 # check all a-tags
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
293 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
294 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
295 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
296 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
297 # is link with href
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
298 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
299 if href.startswith('lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
300 # is pollux link
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
301 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
302 # change href
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
303 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
304 # add target
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
305 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
306 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
307 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
308 if href.startswith('lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
309 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
310 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
311 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
312 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
313 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
314 if href.startswith('#note-'):
162
e6f0553ad176 characterNormalization
abukhman
parents: 161
diff changeset
315 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
316 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
317 return "no text here"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
318
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
319 def getTranslate(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
320 """translate into another languages"""
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
321 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
322 #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
323 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
324
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
325 def getLemma(self, lemma=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
326 """simular words lemma """
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
327 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
328 #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
329 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
330
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
331 def getLemmaNew(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
332 """simular words lemma """
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
333 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
334 #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
335 return data
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
336
392
9b2adae6c6e2 *** empty log message ***
abukhman
parents: 388
diff changeset
337 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionsClose=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
338 """number of"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
339 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
340 pagesize = pageinfo['queryPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
341 pn = pageinfo['searchPN']
181
bc44cf97dfdc *** empty log message ***
abukhman
parents: 180
diff changeset
342 query =pageinfo['query']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
343 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
344 tocSearch = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
345 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
346
179
ac715583385b urllib.quote
abukhman
parents: 178
diff changeset
347 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))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
348 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
349 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
350 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
351 tocSearch = int(getTextFromNode(numdivs[0]))
393
bb317af1c197 *** empty log message ***
abukhman
parents: 392
diff changeset
352 logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
353 tc=int((tocSearch/10)+1)
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
354 logging.debug("documentViewer (gettoc) tc: %s"%(tc))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
355 return tc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
356
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
357 def getQueryResultHits(self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionsClose=None):
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
358
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
359 """number of hits in Search mode"""
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
360 docpath = docinfo['textURLPath']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
361 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
362 pn = pageinfo['searchPN']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
363 query =pageinfo['query']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
364 queryType =pageinfo['queryType']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
365 tocSearch = 0
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
366 tocDiv = None
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
367
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
368 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
369 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False)
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
370 pagedom = Parse(pagexml)
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
371 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
372 tocSearch = int(getTextFromNode(numdivs[0]))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
373 return tocSearch
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
374
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
375 def getQueryResultHitsText(self, docinfo=None, pageinfo=None):
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
376 """number of hits in Text of Contents mode"""
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
377
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
378 docpath = docinfo['textURLPath']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
379 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
380 pn = pageinfo['searchPN']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
381 query =pageinfo['query']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
382 queryType =pageinfo['queryType']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
383 tocSearch = 0
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
384 tocDiv = None
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
385 tocMode = pageinfo['tocMode']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
386 tocPN = pageinfo['tocPN']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
387 pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'toc'))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
388 pagedom = Parse(pagexml)
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
389 logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
390 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
391 tocSearch = int(getTextFromNode(numdivs[0]))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
392 return tocSearch
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
393
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
394 def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None):
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
395 """number of hits in Text of Figures mode"""
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
396
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
397 docpath = docinfo['textURLPath']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
398 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
399 pn = pageinfo['searchPN']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
400 query =pageinfo['query']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
401 queryType =pageinfo['queryType']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
402 tocSearch = 0
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
403 tocDiv = None
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
404 tocMode = pageinfo['tocMode']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
405 tocPN = pageinfo['tocPN']
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
406 pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'figures'))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
407 pagedom = Parse(pagexml)
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
408 logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
409 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
410 tocSearch = int(getTextFromNode(numdivs[0]))
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
411 return tocSearch
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
412
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
413
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
414 def getToc(self, mode="text", docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
415 """loads table of contents and stores in docinfo"""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
416 logging.debug("documentViewer (gettoc) mode: %s"%(mode))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
417 if mode == "none":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
418 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
419 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
420 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
421 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
422
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
423 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
424 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
425 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
426 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
427 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
428 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
429 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
430 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
431 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
432 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
433 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
434
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
435 pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
436 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
437 # post-processing downloaded xml
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
438 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
439 # get number of entries
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
440 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
441 if len(numdivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
442 tocSize = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
443 docinfo['tocSize_%s'%mode] = tocSize
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
444 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
445
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
446 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
447 """returns single page from the table of contents"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
448 # TODO: this should use the cached TOC
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
449 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
450 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
451 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
452 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
453 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
454 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
455 pagesize = pageinfo['tocPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
456 pn = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
457 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
458 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
459 viewMode= pageinfo['viewMode']
171
77d5ff742bb4 characterNormalization
abukhman
parents: 170
diff changeset
460 characterNormalization = pageinfo ['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
461 tocMode = pageinfo['tocMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
462 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
463
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
464 data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
465 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
466 text = page.replace('mode=image','mode=texttool')
165
820a2a4b23c3 characterNormalization
abukhman
parents: 164
diff changeset
467 logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
162
e6f0553ad176 characterNormalization
abukhman
parents: 161
diff changeset
468 #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
469 return text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
470
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
471 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
472 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
473 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
474 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
475 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
476 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
477 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
478
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
479 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
480 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
481 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
482 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
483 return pt()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
484
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
485 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
486 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
487 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
488 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
489 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
490 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
491
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
492
133
40730cfbcb8e fixed problem with getnumpages and **args
casties
parents: 132
diff changeset
493