annotate MpdlXmlTextServer.py @ 321:480097b8e685

*** empty log message ***
author abukhman
date Fri, 15 Oct 2010 15:42:09 +0200
parents 719e6e985677
children 9adbedb6f85b
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
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
41 def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None):
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
188
a7610241ae57 *** empty log message ***
abukhman
parents: 187
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&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,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")
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
138 docinfo['numPages']=int(getTextFromNode(result[0]))
190
c92e6b923913 *** empty log message ***
abukhman
parents: 189
diff changeset
139 return docinfo
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
140
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
141 def getGisPlaces(self, docinfo=None, pageinfo=None):
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
142 """ Show all Gis Places of whole Page"""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
143 xpath='//place'
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
144 docpath = docinfo['textURLPath']
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
145 url = docinfo['url']
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
146 selfurl = self.absolute_url()
257
d1eb63cf3410 *** empty log message ***
abukhman
parents: 256
diff changeset
147 pn = pageinfo['current']
291
3b54df7ca88e *** empty log message ***
abukhman
parents: 290
diff changeset
148 hrefList=[]
306
ec293c61f8a9 *** empty log message ***
abukhman
parents: 305
diff changeset
149 myList= ""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
150 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
151 dom = Parse(text)
265
4a0dd5ff6859 *** empty log message ***
abukhman
parents: 264
diff changeset
152 result = dom.xpath("//result/resultPage/place")
236
ccdce5aca47b *** empty log message ***
abukhman
parents: 233
diff changeset
153 for l in result:
250
7775d87016b5 *** empty log message ***
abukhman
parents: 249
diff changeset
154 hrefNode= l.getAttributeNodeNS(None, u"id")
272
7a52ace035e6 *** empty log message ***
abukhman
parents: 271
diff changeset
155 href= hrefNode.nodeValue
292
d97ab809de6f *** empty log message ***
abukhman
parents: 291
diff changeset
156 hrefList.append(href)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
157 myList = ",".join(hrefList)
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
158 logging.debug("getGisPlaces :%s"%(myList))
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
159 return myList
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
160
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
161 def getAllGisPlaces (self, docinfo=None, pageinfo=None):
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
162 """Show all Gis Places of whole Book """
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
163 xpath ='//echo:place'
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
164 docpath =docinfo['textURLPath']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
165 url = docinfo['url']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
166 selfurl =self.absolute_url()
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
167 pn =pageinfo['current']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
168 hrefList=[]
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
169 myList=""
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
170 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
171 dom =Parse(text)
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
172 result = dom.xpath("//result/resultPage/place")
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
173 for l in result:
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
174 hrefNode = l.getAttributeNodeNS(None, u"id")
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
175 href= hrefNode.nodeValue
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
176 hrefList.append(href)
300
e67f50a09259 *** empty log message ***
abukhman
parents: 299
diff changeset
177 myList = ",".join(hrefList)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
178 logging.debug("getALLGisPlaces :%s"%(myList))
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
179 return myList
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
180
314
6e08398ba38d *** empty log message ***
abukhman
parents: 313
diff changeset
181 def getOrigPages (self, docinfo=None, pageinfo=None):
310
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
182 """Show original page """
312
c774a0c58bc3 *** empty log message ***
abukhman
parents: 311
diff changeset
183 docpath = docinfo['textURLPath']
311
84b69d261131 *** empty log message ***
abukhman
parents: 310
diff changeset
184 #url = docinfo['url']
310
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
185 selfurl = self.absolute_url()
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
186 pn =pageinfo['current']
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
187 viewMode= pageinfo['viewMode']
317
d2ee9f3ab08d *** empty log message ***
abukhman
parents: 316
diff changeset
188 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
189 dom =Parse(text)
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
190 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
80ae2324562b *** empty log message ***
abukhman
parents: 309
diff changeset
191 logging.debug("getOrigPages :%s"%(pagedivs))
321
480097b8e685 *** empty log message ***
abukhman
parents: 320
diff changeset
192 original =getTextFromNode(pagedivs[0])
480097b8e685 *** empty log message ***
abukhman
parents: 320
diff changeset
193 logging.debug("XXXXXXXXXXXgetOrigPagesXXXXXXXX :%s"%(original))
480097b8e685 *** empty log message ***
abukhman
parents: 320
diff changeset
194 return original
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
195
170
caf66e1fccd3 characterNormalization
abukhman
parents: 169
diff changeset
196 def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
197 """returns single page from fulltext"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
198 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
199 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
200 url = docinfo['url']
237
d596d6839fb8 *** empty log message ***
abukhman
parents: 236
diff changeset
201 name = docinfo['name']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
202 viewMode= pageinfo['viewMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
203 tocMode = pageinfo['tocMode']
164
530f6983f72d characterNormalization
abukhman
parents: 162
diff changeset
204 characterNormalization=pageinfo['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
205 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
206 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
207 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
208 textmode = "textPollux"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
209 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
210 textmode = mode
162
e6f0553ad176 characterNormalization
abukhman
parents: 161
diff changeset
211 #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
171
77d5ff742bb4 characterNormalization
abukhman
parents: 170
diff changeset
212 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
213 if highlightQuery is not None:
187
0b27f1c1e34b *** empty log message ***
abukhman
parents: 186
diff changeset
214 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
215
185
6fb9a06b8bee *** empty log message ***
abukhman
parents: 184
diff changeset
216 pagexml = self.getServerData("page-fragment.xql",textParam)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
217 #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
218
186
3b3476df97d4 *** empty log message ***
abukhman
parents: 185
diff changeset
219 pagedom = Parse(pagexml)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
220 # plain text mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
221 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
222 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
223 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
224 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
225 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
226 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
227 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
228 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
229 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
230 href= hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
231 if href.startswith('#note-'):
172
11cb88e7f523 characterNormalization
abukhman
parents: 171
diff changeset
232 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
233 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
234 if mode == "xml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
235 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
236 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
237 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
238 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
239 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
240 if mode == "gis":
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
241 # first div contains text
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
242 pagedivs = pagedom.xpath("/div")
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
243 if len(pagedivs) > 0:
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
244 pagenode = pagedivs[0]
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
245 links =pagenode.xpath("//a")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
246 for l in links:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
247 hrefNode =l.getAttributeNodeNS(None, u"href")
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
248 if hrefNode:
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
249 href=hrefNode.nodeValue
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
250 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
239
1f3badb3f1e2 *** empty log message ***
abukhman
parents: 238
diff changeset
251 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
252 l.setAttributeNS(None, 'target', '_blank')
210
50a666145096 *** empty log message ***
abukhman
parents: 209
diff changeset
253 return serializeNode(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
254
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
255 if mode == "pureXml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
256 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
257 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
258 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
259 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
260 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
261 # text-with-links mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
262 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
263 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
264 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
265 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
266 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
267 # check all a-tags
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
268 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
269 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
270 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
271 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
272 # is link with href
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
273 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
274 if href.startswith('lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
275 # is pollux link
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
276 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
277 # change href
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
278 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
279 # add target
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
280 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
281 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
282 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
283 if href.startswith('lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
284 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
285 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
286 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
287 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
288 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
289 if href.startswith('#note-'):
162
e6f0553ad176 characterNormalization
abukhman
parents: 161
diff changeset
290 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
291 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
292 return "no text here"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
293
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
294 def getTranslate(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
295 """translate into another languages"""
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
296 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
297 #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
298 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
299
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
300 def getLemma(self, lemma=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
301 """simular words lemma """
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
302 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
303 #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
304 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
305
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
306 def getLemmaNew(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
307 """simular words lemma """
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
308 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
309 #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
310 return data
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
311
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
312 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
313 """number of"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
314 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
315 pagesize = pageinfo['queryPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
316 pn = pageinfo['searchPN']
181
bc44cf97dfdc *** empty log message ***
abukhman
parents: 180
diff changeset
317 query =pageinfo['query']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
318 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
319 tocSearch = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
320 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
321
179
ac715583385b urllib.quote
abukhman
parents: 178
diff changeset
322 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
323 #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
324 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
325 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
326 tocSearch = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
327 tc=int((tocSearch/10)+1)
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
328 logging.debug("documentViewer (gettoc) tc: %s"%(tc))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
329 return tc
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 getToc(self, mode="text", docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
332 """loads table of contents and stores in docinfo"""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
333 logging.debug("documentViewer (gettoc) mode: %s"%(mode))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
334 if mode == "none":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
335 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
336 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
337 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
338 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
339
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
340 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
341 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
342 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
343 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
344 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
345 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
346 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
347 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
348 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
349 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
350 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
351
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
352 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
353 #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
354 # post-processing downloaded xml
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
355 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
356 # get number of entries
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
357 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
358 if len(numdivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
359 tocSize = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
360 docinfo['tocSize_%s'%mode] = tocSize
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
361 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
362
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
363 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
364 """returns single page from the table of contents"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
365 # TODO: this should use the cached TOC
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
366 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
367 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
368 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
369 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
370 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
371 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
372 pagesize = pageinfo['tocPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
373 pn = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
374 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
375 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
376 viewMode= pageinfo['viewMode']
171
77d5ff742bb4 characterNormalization
abukhman
parents: 170
diff changeset
377 characterNormalization = pageinfo ['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
378 tocMode = pageinfo['tocMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
379 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
380
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
381 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
382 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
383 text = page.replace('mode=image','mode=texttool')
165
820a2a4b23c3 characterNormalization
abukhman
parents: 164
diff changeset
384 logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
162
e6f0553ad176 characterNormalization
abukhman
parents: 161
diff changeset
385 #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
386 return text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
387
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
388 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
389 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
390 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
391 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
392 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
393 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
394 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
395
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
396 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
397 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
398 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
399 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
400 return pt()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
401
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
402 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
403 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
404 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
405 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
406 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
407 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
408
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
409
133
40730cfbcb8e fixed problem with getnumpages and **args
casties
parents: 132
diff changeset
410