annotate MpdlXmlTextServer.py @ 453:beb7ccb92564 elementtree

first version using elementtree instead of 4suite xml
author casties
date Thu, 14 Jul 2011 19:43:56 +0200
parents 1cea48640992
children 0a53fea83df7
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
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
5 from Ft.Xml.Domlette import NonvalidatingReader
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
6 import Ft.Xml.Domlette
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
7 import cStringIO
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
8
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
9 import xml.etree.ElementTree as ET
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
10
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
11 import md5
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
12 import sys
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
13 import logging
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
14 import urllib
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
15 import documentViewer
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
16 #from documentViewer import getTextFromNode, serializeNode
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
17
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
18 def getText(node):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
19 """get the cdata content of a node"""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
20 if node is None:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
21 return ""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
22 # ET:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
23 text = node.text or ""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
24 for e in node:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
25 text += gettext(e)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
26 if e.tail:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
27 text += e.tail
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
28
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
29 return text
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
30
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
31 def serialize(node):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
32 """returns a string containing an XML snippet of node"""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
33 s = ET.tostring(node, 'UTF-8')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
34 # snip off XML declaration
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
35 if s.startswith('<?xml'):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
36 i = s.find('?>')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
37 return s[i+3:]
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
38
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
39 return s
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
40
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
41
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
42 def getTextFromNode(node):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
43 """get the cdata content of a node"""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
44 if node is None:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
45 return ""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
46 # ET:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
47 #text = node.text or ""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
48 #for e in node:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
49 # text += gettext(e)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
50 # if e.tail:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
51 # text += e.tail
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
52
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
53 # 4Suite:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
54 nodelist=node.childNodes
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
55 text = ""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
56 for n in nodelist:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
57 if n.nodeType == node.TEXT_NODE:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
58 text = text + n.data
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
59
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
60 return text
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
61
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
62 def serializeNode(node, encoding="utf-8"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
63 """returns a string containing node as XML"""
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
64 #s = ET.tostring(node)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
65
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
66 # 4Suite:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
67 stream = cStringIO.StringIO()
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
68 Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
69 s = stream.getvalue()
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
70 stream.close()
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
71
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
72 return s
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
73
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
74
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
75 class MpdlXmlTextServer(SimpleItem):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
76 """TextServer implementation for MPDL-XML eXist server"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
77 meta_type="MPDL-XML TextServer"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
78
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
79 manage_options=(
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
80 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
81 )+SimpleItem.manage_options
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
82
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
83 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
84
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
85 def __init__(self,id,title="",serverUrl="http://mpdl-system.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
86 #def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/", serverName=None, timeout=40):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
87
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
88 """constructor"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
89 self.id=id
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
90 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
91 self.timeout = timeout
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
92 if serverName is None:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
93 self.serverUrl = serverUrl
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
94 else:
39167e96e582 fixed creation of text server on __init__
casties
parents: 129
diff changeset
95 self.serverUrl = "http://%s/mpdl/interface/"%serverName
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
96
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
97 def getHttpData(self, url, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
98 """returns result from url+data HTTP request"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
99 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
100
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
101 def getServerData(self, method, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
102 """returns result from text server for method+data"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
103 url = self.serverUrl+method
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
104 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
105
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
106 def getSearch(self, pageinfo=None, docinfo=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
107 """get search list"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
108 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
109 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
110 pagesize = pageinfo['queryPageSize']
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
111 pn = pageinfo.get('searchPN',1)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
112 sn = pageinfo['sn']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
113 highlightQuery = pageinfo['highlightQuery']
181
bc44cf97dfdc *** empty log message ***
abukhman
parents: 180
diff changeset
114 query =pageinfo['query']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
115 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
116 viewMode= pageinfo['viewMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
117 tocMode = pageinfo['tocMode']
169
21ac9164eedf characterNormalization
abukhman
parents: 167
diff changeset
118 characterNormalization = pageinfo['characterNormalization']
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
119 #optionToggle = pageinfo['optionToggle']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
120 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
121 selfurl = self.absolute_url()
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
122 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
123 pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
124 pagedom = Parse(pagexml)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
125
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
126 """
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
127 pagedivs = pagedom.xpath("//div[@class='queryResultHits']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
128 if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
129 if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
130 docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0]))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
131 s = getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
132 s1 = int(s)/10+1
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
133 try:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
134 docinfo['queryResultHits'] = int(s1)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
135 logging.debug("SEARCH ENTRIES: %s"%(s1))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
136 except:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
137 docinfo['queryResultHits'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
138 """
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
139 if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
140 pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
141 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
142 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
143 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
144 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
145 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
146 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
147 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
148 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
149 selfurl = self.absolute_url()
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
150 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN, characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
151 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
152 #logging.debug("PUREXML :%s"%(serializeNode(pagenode)))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
153 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
154 if (queryType=="fulltextMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
155 pagedivs = pagedom.xpath("//div[@class='queryResult']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
156 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
157 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
158 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
159 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
160 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
161 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
162 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
163 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
164 selfurl = self.absolute_url()
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
165 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
166 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
167 if href.startswith('../lt/lemma.xql'):
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
168 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_query'%(selfurl))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
169 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
170 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
171 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
172 pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
173 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
174 if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
175 pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
176 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
177 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
178 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
179 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
180 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
181 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
182 href = hrefNode.nodeValue
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
183 hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,characterNormalization))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
184 if href.startswith('../lt/lex.xql'):
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
185 hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_lex'%selfurl)
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
186 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
187 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
188 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
189 if href.startswith('../lt/lemma.xql'):
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
190 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%(selfurl))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
191 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
192 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
193 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();')
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
194 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
195 return "no text here"
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
196
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
197 def getGisPlaces(self, docinfo=None, pageinfo=None):
207
0f6a377679d5 *** empty log message ***
abukhman
parents: 206
diff changeset
198 """ Show all Gis Places of whole Page"""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
199 xpath='//place'
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
200 docpath = docinfo.get('textURLPath',None)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
201 if not docpath:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
202 return None
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
203
253
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
204 url = docinfo['url']
d6ca167ccb1a *** empty log message ***
abukhman
parents: 252
diff changeset
205 selfurl = self.absolute_url()
257
d1eb63cf3410 *** empty log message ***
abukhman
parents: 256
diff changeset
206 pn = pageinfo['current']
291
3b54df7ca88e *** empty log message ***
abukhman
parents: 290
diff changeset
207 hrefList=[]
306
ec293c61f8a9 *** empty log message ***
abukhman
parents: 305
diff changeset
208 myList= ""
264
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
209 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
ac93e115d902 *** empty log message ***
abukhman
parents: 263
diff changeset
210 dom = Parse(text)
265
4a0dd5ff6859 *** empty log message ***
abukhman
parents: 264
diff changeset
211 result = dom.xpath("//result/resultPage/place")
236
ccdce5aca47b *** empty log message ***
abukhman
parents: 233
diff changeset
212 for l in result:
250
7775d87016b5 *** empty log message ***
abukhman
parents: 249
diff changeset
213 hrefNode= l.getAttributeNodeNS(None, u"id")
272
7a52ace035e6 *** empty log message ***
abukhman
parents: 271
diff changeset
214 href= hrefNode.nodeValue
292
d97ab809de6f *** empty log message ***
abukhman
parents: 291
diff changeset
215 hrefList.append(href)
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
216 myList = ",".join(hrefList)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
217 #logging.debug("getGisPlaces :%s"%(myList))
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
218 return myList
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
219
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
220 def getAllGisPlaces (self, docinfo=None, pageinfo=None):
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
221 """Show all Gis Places of whole Book """
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
222 xpath ='//echo:place'
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
223 docpath =docinfo['textURLPath']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
224 url = docinfo['url']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
225 selfurl =self.absolute_url()
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
226 pn =pageinfo['current']
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
227 hrefList=[]
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
228 myList=""
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
229 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
230 dom =Parse(text)
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
231 result = dom.xpath("//result/resultPage/place")
395
e6f3fcb2058f *** empty log message ***
abukhman
parents: 393
diff changeset
232
307
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
233 for l in result:
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
234 hrefNode = l.getAttributeNodeNS(None, u"id")
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
235 href= hrefNode.nodeValue
ec5e920a61e6 *** empty log message ***
abukhman
parents: 306
diff changeset
236 hrefList.append(href)
300
e67f50a09259 *** empty log message ***
abukhman
parents: 299
diff changeset
237 myList = ",".join(hrefList)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
238 #logging.debug("getALLGisPlaces :%s"%(myList))
309
ea18683b2bff *** empty log message ***
abukhman
parents: 308
diff changeset
239 return myList
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
240
388
0265ab93716a *** empty log message ***
abukhman
parents: 386
diff changeset
241
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
242 def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
243 """returns single page from fulltext"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
244 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
245 path = docinfo['textURLPath']
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
246 url = docinfo.get('url',None)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
247 name = docinfo.get('name',None)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
248 pn =pageinfo['current']
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
249 sn = pageinfo['sn']
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
250 #optionToggle =pageinfo ['optionToggle']
367
f991ddd3a045 *** empty log message ***
abukhman
parents: 364
diff changeset
251 highlightQuery = pageinfo['highlightQuery']
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
252 #mode = pageinfo ['viewMode']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
253 tocMode = pageinfo['tocMode']
164
530f6983f72d characterNormalization
abukhman
parents: 162
diff changeset
254 characterNormalization=pageinfo['characterNormalization']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
255 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
256 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
257 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
258 textmode = "textPollux"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
259 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
260 textmode = mode
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
261
373
0aa294b31236 *** empty log message ***
abukhman
parents: 372
diff changeset
262 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
370
33adbe140534 *** empty log message ***
abukhman
parents: 369
diff changeset
263 if highlightQuery is not None:
376
4f381768909a *** empty log message ***
abukhman
parents: 375
diff changeset
264 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
265
185
6fb9a06b8bee *** empty log message ***
abukhman
parents: 184
diff changeset
266 pagexml = self.getServerData("page-fragment.xql",textParam)
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
267 dom = ET.fromstring(pagexml)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
268 #dom = NonvalidatingReader.parseStream(pagexml)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
269
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
270 #original Pages
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
271 #pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
272
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
273 """if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
274 if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
275 docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
276 logging.debug("ORIGINAL PAGE: %s"%(docinfo['pageNumberOrig']))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
277
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
278 #original Pages Norm
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
279 pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
280 if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
281 if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
282 docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
283 logging.debug("ORIGINAL PAGE NORM: %s"%(docinfo['pageNumberOrigNorm']))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
284 """
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
285 #figureEntries
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
286 # pagedivs = dom.xpath("//div[@class='countFigureEntries']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
287 # if pagedivs == dom.xpath("//div[@class='countFigureEntries']"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
288 # if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
289 # docinfo['countFigureEntries'] = getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
290 # s = getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
291 # if s=='0':
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
292 # try:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
293 # docinfo['countFigureEntries'] = int(s)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
294 # except:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
295 # docinfo['countFigureEntries'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
296 # else:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
297 # s1 = int(s)/30+1
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
298 # try:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
299 # docinfo['countFigureEntries'] = int(s1)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
300 # except:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
301 # docinfo['countFigureEntries'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
302 #
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
303 # #allPlaces
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
304 # pagedivs = dom.xpath("//div[@class='countPlaces']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
305 # if pagedivs == dom.xpath("//div[@class='countPlaces']"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
306 # if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
307 # docinfo['countPlaces']= getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
308 # s = getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
309 # try:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
310 # docinfo['countPlaces'] = int(s)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
311 # except:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
312 # docinfo['countPlaces'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
313 #
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
314 # #tocEntries
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
315 # pagedivs = dom.xpath("//div[@class='countTocEntries']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
316 # if pagedivs == dom.xpath("//div[@class='countTocEntries']"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
317 # if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
318 # docinfo['countTocEntries'] = int(getTextFromNode(pagedivs[0]))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
319 # s = getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
320 # if s=='0':
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
321 # try:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
322 # docinfo['countTocEntries'] = int(s)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
323 # except:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
324 # docinfo['countTocEntries'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
325 # else:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
326 # s1 = int(s)/30+1
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
327 # try:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
328 # docinfo['countTocEntries'] = int(s1)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
329 # except:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
330 # docinfo['countTocEntries'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
331
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
332 #numTextPages
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
333 #pagedivs = dom.xpath("//div[@class='countPages']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
334 alldivs = dom.findall(".//div")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
335 pagediv = None
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
336 for div in alldivs:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
337 dc = div.get('class')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
338 if dc == 'pageContent':
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
339 pagediv = div
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
340
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
341 if dc == 'countPages':
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
342 try:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
343 np = int(div.text)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
344 docinfo['numPages'] = np
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
345 pageinfo['end'] = min(pageinfo['end'], np)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
346 pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
347 if np % pageinfo['groupsize'] > 0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
348 pageinfo['numgroups'] += 1
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
349
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
350 except:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
351 docinfo['numPages'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
352
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
353 break
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
354
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
355 # ROC: why?
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
356 # else:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
357 # #no full text -- init to 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
358 # docinfo['pageNumberOrig'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
359 # docinfo['countFigureEntries'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
360 # docinfo['countPlaces'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
361 # docinfo['countTocEntries'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
362 # docinfo['numPages'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
363 # docinfo['pageNumberOrigNorm'] = 0
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
364 # #return docinfo
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
365
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
366 # plain text mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
367 if mode == "text":
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
368 #pagedivs = dom.xpath("/div")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
369 if pagediv:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
370 links = pagediv.findall(".//a")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
371 for l in links:
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
372 href = l.get('href')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
373 if href and href.startswith('#note-'):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
374 href = href.replace('#note-',"?url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
375 l.set('href', href)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
376 logging.debug("page=%s"%ET.tostring(pagediv, 'UTF-8'))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
377 return serialize(pagediv)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
378
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
379 if mode == "xml":
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
380 if pagediv:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
381 return serialize(pagediv)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
382
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
383 if mode == "pureXml":
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
384 if pagediv:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
385 return serialize(pagediv)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
386
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
387 if mode == "gis":
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
388 if pagediv:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
389 # check all a-tags
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
390 links = pagediv.findall(".//a")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
391 for l in links:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
392 href = l.get('href')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
393 if href:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
394 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
395 l.set('href', href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
396 l.set('target', '_blank')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
397
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
398 return serialize(pagenode)
140
6345583d809a Gis mode
abukhman
parents: 138
diff changeset
399
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
400 # text-with-links mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
401 if mode == "text_dict":
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
402 if pagediv:
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
403 # check all a-tags
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
404 links = pagediv.findall(".//a")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
405 for l in links:
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
406 href = l.get('href')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
407
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
408 if href:
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
409 # is link with href
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
410 if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql'):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
411 # is pollux link
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
412 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
413 # change href
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
414 l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/head_main_voc'%selfurl))
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
415 # add target
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
416 l.set('target', '_blank')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
417
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
418 if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql'):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
419 selfurl = self.absolute_url()
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
420 l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
421 l.set('target', '_blank')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
422 l.set('onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
423 l.set('ondblclick', 'popupWin.focus();')
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
424
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
425 if href.startswith('#note-'):
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
426 l.set('href', href.replace('#note-',"?url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn)))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
427
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
428 return serialize(pagediv)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
429
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
430 return "no text here"
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
431
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
432 def getOrigPages(self, docinfo=None, pageinfo=None):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
433 docpath = docinfo['textURLPath']
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
434 pn =pageinfo['current']
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
435 selfurl = self.absolute_url()
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
436 pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
437 dom = Parse(pagexml)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
438 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
439 if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
440 if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
441 docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
442 return docinfo['pageNumberOrig']
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
443
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
444 def getOrigPagesNorm(self, docinfo=None, pageinfo=None):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
445 docpath = docinfo['textURLPath']
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
446 pn =pageinfo['current']
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
447 selfurl = self.absolute_url()
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
448 pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
449 dom = Parse(pagexml)
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
450 pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
451 if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
452 if len(pagedivs)>0:
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
453 docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
454 return docinfo['pageNumberOrigNorm']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
455
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
456
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
457 def getTranslate(self, word=None, language=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
458 """translate into another languages"""
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
459 data = self.getServerData("lt/wordInfo.xql","language="+str(language)+"&word="+urllib.quote(word)+"&output=html")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
460 #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
461 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
462
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
463 def getLemma(self, lemma=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
464 """simular words lemma """
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
465 data = self.getServerData("lt/lemma.xql","language="+str(language)+"&lemma="+urllib.quote(lemma)+"&output=html")
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
466 return data
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
467
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
468 def getLemmaQuery(self, query=None, language=None):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
469 """simular words lemma """
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
470 data = self.getServerData("lt/lemma.xql","language="+str(language)+"&query="+urllib.quote(query)+"&output=html")
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
471 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
472
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
473 def getLex(self, query=None, language=None):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
474 #simular words lemma
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
475 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
476 return data
175
d1c13a6ba20d gis ?nderung
abukhman
parents: 172
diff changeset
477
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
478 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
479 #number of
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
480 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
481 pagesize = pageinfo['queryPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
482 pn = pageinfo['searchPN']
181
bc44cf97dfdc *** empty log message ***
abukhman
parents: 180
diff changeset
483 query =pageinfo['query']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
484 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
485 tocSearch = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
486 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
487
179
ac715583385b urllib.quote
abukhman
parents: 178
diff changeset
488 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
489 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
490 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
491 tocSearch = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
492 tc=int((tocSearch/10)+1)
402
d3bd3c2674ee *** empty log message ***
abukhman
parents: 401
diff changeset
493 return tc
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
494
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
495 def getToc(self, mode="text", docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
496 """loads table of contents and stores in docinfo"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
497 if mode == "none":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
498 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
499 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
500 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
501 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
502
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
503 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
504 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
505 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
506 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
507 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
508 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
509 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
510 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
511 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
512 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
513 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
514
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
515 pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
516
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
517 # post-processing downloaded xml
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
518 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
519 # get number of entries
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
520 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
521 if len(numdivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
522 tocSize = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
523 docinfo['tocSize_%s'%mode] = tocSize
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
524 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
525
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
526 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
527 """returns single page from the table of contents"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
528 # TODO: this should use the cached TOC
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
529 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
530 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
531 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
532 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
533 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
534 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
535 pagesize = pageinfo['tocPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
536 pn = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
537 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
538 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
539 viewMode= pageinfo['viewMode']
171
77d5ff742bb4 characterNormalization
abukhman
parents: 170
diff changeset
540 characterNormalization = pageinfo ['characterNormalization']
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
541 #optionToggle =pageinfo ['optionToggle']
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
542 tocMode = pageinfo['tocMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
543 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
544
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
545 data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm"%(docpath,queryType, pagesize, pn))
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
546 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
547 text = page.replace('mode=image','mode=texttool')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
548 return text
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
549
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
550 def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
551 #def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
552 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
553 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
554 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
555 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
556 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
557 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
558
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
559 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
560 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
561 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
562 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
563 return pt()
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
564
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
565 def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
566 #def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None):
129
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
567 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
568 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
569 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer
casties
parents:
diff changeset
570 if RESPONSE is not None:
453
beb7ccb92564 first version using elementtree instead of 4suite xml
casties
parents: 407
diff changeset
571 RESPONSE.redirect('manage_main')