annotate zopeSolr.py @ 4:10733d367831

date ranges links auf fulltext
author dwinter
date Fri, 26 Apr 2013 17:52:24 +0200
parents 156c1db5a701
children f695be8f4f34
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
834706423ac1 initial
dwinter
parents:
diff changeset
1 # -*- coding: utf-8 -*-
834706423ac1 initial
dwinter
parents:
diff changeset
2
834706423ac1 initial
dwinter
parents:
diff changeset
3 #Verbindet Zope mit solr. Vorraussetzung ist das Paket sunburnt, @see http://opensource.timetric.com/sunburnt/
834706423ac1 initial
dwinter
parents:
diff changeset
4
834706423ac1 initial
dwinter
parents:
diff changeset
5
834706423ac1 initial
dwinter
parents:
diff changeset
6 from OFS.SimpleItem import SimpleItem
834706423ac1 initial
dwinter
parents:
diff changeset
7 from Products.PageTemplates.PageTemplateFile import PageTemplateFile
834706423ac1 initial
dwinter
parents:
diff changeset
8 import os.path
834706423ac1 initial
dwinter
parents:
diff changeset
9 import sunburnt
834706423ac1 initial
dwinter
parents:
diff changeset
10 from Globals import package_home
834706423ac1 initial
dwinter
parents:
diff changeset
11 import httplib2
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
12 import urlparse
0
834706423ac1 initial
dwinter
parents:
diff changeset
13 import urllib
834706423ac1 initial
dwinter
parents:
diff changeset
14 import re
834706423ac1 initial
dwinter
parents:
diff changeset
15 import xml.etree.ElementTree as ET
834706423ac1 initial
dwinter
parents:
diff changeset
16 import json
834706423ac1 initial
dwinter
parents:
diff changeset
17
834706423ac1 initial
dwinter
parents:
diff changeset
18 #Worte die nicht in der Termliste angezeigt werden sollen #TODO: make this configurable
834706423ac1 initial
dwinter
parents:
diff changeset
19
834706423ac1 initial
dwinter
parents:
diff changeset
20 STOPLIST={'main_content':['forward','drucken','history','science','part','publications','projects',
834706423ac1 initial
dwinter
parents:
diff changeset
21 'project','new','geschichte','institute','related','boltzmannstraße','14195'],
834706423ac1 initial
dwinter
parents:
diff changeset
22
834706423ac1 initial
dwinter
parents:
diff changeset
23 'title':['bd','10','11','12','18','19','20','abt','ad','di','history','geschichte','science']
834706423ac1 initial
dwinter
parents:
diff changeset
24 }
834706423ac1 initial
dwinter
parents:
diff changeset
25
834706423ac1 initial
dwinter
parents:
diff changeset
26
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
27 PURLSERVER="http://md.mpiwg-berlin.mpg.de/purls/"
10733d367831 date ranges
dwinter
parents: 3
diff changeset
28
10733d367831 date ranges
dwinter
parents: 3
diff changeset
29 #TODO: only needed for getTermsAsJSON, solle irgendwie aus dem sunburnt kommen
10733d367831 date ranges
dwinter
parents: 3
diff changeset
30 SOLRSERVER="/select?q=%s:%s&wt=xml&tv=on&qt=tvrh&fl=%s&tv.tf=true"
0
834706423ac1 initial
dwinter
parents:
diff changeset
31
834706423ac1 initial
dwinter
parents:
diff changeset
32 def zptFile(self, path, orphaned=False):
834706423ac1 initial
dwinter
parents:
diff changeset
33 """returns a page template file from the product"""
834706423ac1 initial
dwinter
parents:
diff changeset
34 if orphaned:
834706423ac1 initial
dwinter
parents:
diff changeset
35 # unusual case
834706423ac1 initial
dwinter
parents:
diff changeset
36 pt=PageTemplateFile(os.path.join(package_home(globals()), path))
834706423ac1 initial
dwinter
parents:
diff changeset
37 else:
834706423ac1 initial
dwinter
parents:
diff changeset
38
834706423ac1 initial
dwinter
parents:
diff changeset
39 pt=PageTemplateFile(os.path.join(package_home(globals()), path)).__of__(self)
834706423ac1 initial
dwinter
parents:
diff changeset
40 return pt
834706423ac1 initial
dwinter
parents:
diff changeset
41
834706423ac1 initial
dwinter
parents:
diff changeset
42
834706423ac1 initial
dwinter
parents:
diff changeset
43
834706423ac1 initial
dwinter
parents:
diff changeset
44 class ZopeSolr(SimpleItem):
834706423ac1 initial
dwinter
parents:
diff changeset
45
834706423ac1 initial
dwinter
parents:
diff changeset
46
834706423ac1 initial
dwinter
parents:
diff changeset
47 meta_type="ZopeSolr"
834706423ac1 initial
dwinter
parents:
diff changeset
48
834706423ac1 initial
dwinter
parents:
diff changeset
49 manage_options= ({'label':'Main Config','action': 'changeMain'},) + SimpleItem.manage_options
834706423ac1 initial
dwinter
parents:
diff changeset
50
834706423ac1 initial
dwinter
parents:
diff changeset
51 def __init__(self,id,title,solrURL):
834706423ac1 initial
dwinter
parents:
diff changeset
52 self.id=id
834706423ac1 initial
dwinter
parents:
diff changeset
53 self.title=title
834706423ac1 initial
dwinter
parents:
diff changeset
54 self.solrURL=solrURL #URL einer solr instance bzw. einer collection, falls nicht die default collection benutzt werden soll
834706423ac1 initial
dwinter
parents:
diff changeset
55
834706423ac1 initial
dwinter
parents:
diff changeset
56
834706423ac1 initial
dwinter
parents:
diff changeset
57 #Verbinde mit der solt Instance
834706423ac1 initial
dwinter
parents:
diff changeset
58
834706423ac1 initial
dwinter
parents:
diff changeset
59 def connect(self):
834706423ac1 initial
dwinter
parents:
diff changeset
60 self._v_solr=sunburnt.SolrInterface(url=self.solrURL)
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
61
10733d367831 date ranges
dwinter
parents: 3
diff changeset
62
10733d367831 date ranges
dwinter
parents: 3
diff changeset
63 #erzeuge den link auf die fulltext display version des bildviewers
10733d367831 date ranges
dwinter
parents: 3
diff changeset
64 def createFullTextLink(self,page,facetSerch,search):
10733d367831 date ranges
dwinter
parents: 3
diff changeset
65
10733d367831 date ranges
dwinter
parents: 3
diff changeset
66 fqs=[]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
67 for key in facetSerch.keys():
10733d367831 date ranges
dwinter
parents: 3
diff changeset
68 values = self.getList(facetSerch.get(key))
10733d367831 date ranges
dwinter
parents: 3
diff changeset
69 for value in values:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
70 if value!="*":
10733d367831 date ranges
dwinter
parents: 3
diff changeset
71 fqs.append("%s=%s"%(key,urllib.quote(value)));
10733d367831 date ranges
dwinter
parents: 3
diff changeset
72
10733d367831 date ranges
dwinter
parents: 3
diff changeset
73 ret="pf=%s"%page
10733d367831 date ranges
dwinter
parents: 3
diff changeset
74 ret+="&query=("
10733d367831 date ranges
dwinter
parents: 3
diff changeset
75
10733d367831 date ranges
dwinter
parents: 3
diff changeset
76
10733d367831 date ranges
dwinter
parents: 3
diff changeset
77
10733d367831 date ranges
dwinter
parents: 3
diff changeset
78 ret+=" OR ".join(fqs)
10733d367831 date ranges
dwinter
parents: 3
diff changeset
79 #if len(fqs)>0 and len(qs)>0 and qs[0]!="":
10733d367831 date ranges
dwinter
parents: 3
diff changeset
80 if len(fqs)>0 and len(search)>0: #TODO das muss noch geandert werden, wenn das der normale suchstring nach feldern müssen alle "AND" in "OR" getauscht werden
10733d367831 date ranges
dwinter
parents: 3
diff changeset
81 #da ja in der volltext-line die Zeile gedunfen werden soll wenn eines der Worte drin ist, vorher wurd eventuell und über die ganee seite gesucht
10733d367831 date ranges
dwinter
parents: 3
diff changeset
82 #die worte müssen aber nicht in einer zeile sein und werden jetzt nicht gefunden
10733d367831 date ranges
dwinter
parents: 3
diff changeset
83 ret+=" OR "
10733d367831 date ranges
dwinter
parents: 3
diff changeset
84
10733d367831 date ranges
dwinter
parents: 3
diff changeset
85
10733d367831 date ranges
dwinter
parents: 3
diff changeset
86
10733d367831 date ranges
dwinter
parents: 3
diff changeset
87 ret+="("+search+")"
10733d367831 date ranges
dwinter
parents: 3
diff changeset
88
10733d367831 date ranges
dwinter
parents: 3
diff changeset
89 ret+=")&viewLayer=search"
10733d367831 date ranges
dwinter
parents: 3
diff changeset
90
10733d367831 date ranges
dwinter
parents: 3
diff changeset
91 return ret
10733d367831 date ranges
dwinter
parents: 3
diff changeset
92
0
834706423ac1 initial
dwinter
parents:
diff changeset
93 #erzeuge einen Link, insbesonder für faceted suche
834706423ac1 initial
dwinter
parents:
diff changeset
94 #@param search: ist suchparameter, wird einfach als search=%s weitergereicht
834706423ac1 initial
dwinter
parents:
diff changeset
95 #@param facetSearch: bekommt einen hash (feldname, liste der suchworte)
834706423ac1 initial
dwinter
parents:
diff changeset
96
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
97 def generateLink(self,search,facetSearch={},ranges={},dateranges={},args={}):
0
834706423ac1 initial
dwinter
parents:
diff changeset
98 ret="?search=%s"%search
834706423ac1 initial
dwinter
parents:
diff changeset
99
834706423ac1 initial
dwinter
parents:
diff changeset
100 for facet in facetSearch.keys():
834706423ac1 initial
dwinter
parents:
diff changeset
101
834706423ac1 initial
dwinter
parents:
diff changeset
102 searchTerms = facetSearch[facet]
834706423ac1 initial
dwinter
parents:
diff changeset
103
834706423ac1 initial
dwinter
parents:
diff changeset
104 if isinstance(searchTerms, basestring):
834706423ac1 initial
dwinter
parents:
diff changeset
105 searchTerms=[searchTerms]
834706423ac1 initial
dwinter
parents:
diff changeset
106
834706423ac1 initial
dwinter
parents:
diff changeset
107 for searchTerm in searchTerms:
834706423ac1 initial
dwinter
parents:
diff changeset
108 try:
834706423ac1 initial
dwinter
parents:
diff changeset
109 ret+="&%s_fc=%s"%(facet,searchTerm.encode('utf-8'))
834706423ac1 initial
dwinter
parents:
diff changeset
110 except:
834706423ac1 initial
dwinter
parents:
diff changeset
111 ret+="&%s_fc=%s"%(facet,searchTerm)
834706423ac1 initial
dwinter
parents:
diff changeset
112
834706423ac1 initial
dwinter
parents:
diff changeset
113 for key,values in ranges.items():
834706423ac1 initial
dwinter
parents:
diff changeset
114
834706423ac1 initial
dwinter
parents:
diff changeset
115 if isinstance(values, basestring):
834706423ac1 initial
dwinter
parents:
diff changeset
116 values=[values]
834706423ac1 initial
dwinter
parents:
diff changeset
117
834706423ac1 initial
dwinter
parents:
diff changeset
118 for value in values:
834706423ac1 initial
dwinter
parents:
diff changeset
119 ret+="&%s_rg=%s"%(key,value)
834706423ac1 initial
dwinter
parents:
diff changeset
120
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
121 for key,values in dateranges.items():
10733d367831 date ranges
dwinter
parents: 3
diff changeset
122
10733d367831 date ranges
dwinter
parents: 3
diff changeset
123 if isinstance(values, basestring):
10733d367831 date ranges
dwinter
parents: 3
diff changeset
124 values=[values]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
125
10733d367831 date ranges
dwinter
parents: 3
diff changeset
126 for value in values:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
127 ret+="&%s_drg=%s"%(key,value)
10733d367831 date ranges
dwinter
parents: 3
diff changeset
128
0
834706423ac1 initial
dwinter
parents:
diff changeset
129
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
130 for key,value in args.items():
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
131
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
132 if isinstance(value, basestring):
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
133 value=[value]
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
134
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
135 for val in value:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
136 ret+="&%s=%s"%(key,val)
0
834706423ac1 initial
dwinter
parents:
diff changeset
137 return ret
834706423ac1 initial
dwinter
parents:
diff changeset
138
834706423ac1 initial
dwinter
parents:
diff changeset
139 #hilfsmethode erzeuget immer eine liste von einträgen
834706423ac1 initial
dwinter
parents:
diff changeset
140 def getList(self,param):
834706423ac1 initial
dwinter
parents:
diff changeset
141 if isinstance(param, basestring):
834706423ac1 initial
dwinter
parents:
diff changeset
142 param=[param]
834706423ac1 initial
dwinter
parents:
diff changeset
143
834706423ac1 initial
dwinter
parents:
diff changeset
144 return param
834706423ac1 initial
dwinter
parents:
diff changeset
145
834706423ac1 initial
dwinter
parents:
diff changeset
146 #erzeugt analog zu den Ranges in Velocity für ein numerisches Feld die Suche in ranges
834706423ac1 initial
dwinter
parents:
diff changeset
147 #@param field:ist der name des Feldes in dem in Ranges gesucht werden soll
834706423ac1 initial
dwinter
parents:
diff changeset
148 #@param begin anfang der ranges
834706423ac1 initial
dwinter
parents:
diff changeset
149 #@param end ende der Raanges
834706423ac1 initial
dwinter
parents:
diff changeset
150 #@param increment größe eines ranges
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
151 #@param date wenn wahr, dann ist field ein Datumsfeld, begin, end,increment sind trotzdem zur Zeit, dan Integer für Jahre !!, d.h. zur Zeit sind nur JAhre als Schritte möglich
0
834706423ac1 initial
dwinter
parents:
diff changeset
152 # für alle andere paramter @see prepareSearch
834706423ac1 initial
dwinter
parents:
diff changeset
153 #@return gibt ein hash: mit range -> anzahl der Treffer in dem Ranage. e.g 1921-1950 -> 21
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
154
10733d367831 date ranges
dwinter
parents: 3
diff changeset
155 def ranges(self,field,begin,end,increment,solrQuery="",facetFields=[],start=0,rows=10,facetSearch=None,sortFacets=True,date=False,storeFields=False):
0
834706423ac1 initial
dwinter
parents:
diff changeset
156
834706423ac1 initial
dwinter
parents:
diff changeset
157 q=self.prepareSearch(solrQuery,facetFields,start,rows,facetSearch,sortFacets)
834706423ac1 initial
dwinter
parents:
diff changeset
158
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
159 res={} #speiceher abzahl
10733d367831 date ranges
dwinter
parents: 3
diff changeset
160
10733d367831 date ranges
dwinter
parents: 3
diff changeset
161 fls={} #speichere felder
10733d367831 date ranges
dwinter
parents: 3
diff changeset
162
10733d367831 date ranges
dwinter
parents: 3
diff changeset
163 #for fl in storeFields: #initialisiere den hash
10733d367831 date ranges
dwinter
parents: 3
diff changeset
164 # fls[fl]={}
10733d367831 date ranges
dwinter
parents: 3
diff changeset
165
0
834706423ac1 initial
dwinter
parents:
diff changeset
166 if not getattr(self,'_v_solr_',None):
834706423ac1 initial
dwinter
parents:
diff changeset
167 self.connect()
834706423ac1 initial
dwinter
parents:
diff changeset
168
834706423ac1 initial
dwinter
parents:
diff changeset
169
834706423ac1 initial
dwinter
parents:
diff changeset
170
834706423ac1 initial
dwinter
parents:
diff changeset
171 for x in range(begin,end,increment):
834706423ac1 initial
dwinter
parents:
diff changeset
172 query={}
834706423ac1 initial
dwinter
parents:
diff changeset
173 #query["%s__gt"%field]=x
834706423ac1 initial
dwinter
parents:
diff changeset
174 #TODO __gt scheint nicht zu funktionieren wird zu gte (???)
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
175
10733d367831 date ranges
dwinter
parents: 3
diff changeset
176 if not date:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
177 query["%s__gte"%field]=int(x)+1
10733d367831 date ranges
dwinter
parents: 3
diff changeset
178
10733d367831 date ranges
dwinter
parents: 3
diff changeset
179 query["%s__lte"%field]=x+increment
10733d367831 date ranges
dwinter
parents: 3
diff changeset
180
10733d367831 date ranges
dwinter
parents: 3
diff changeset
181 key="%s-%s"%(x,x+increment)
0
834706423ac1 initial
dwinter
parents:
diff changeset
182
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
183 else:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
184 year="%s-01-01T00:00:00Z"
10733d367831 date ranges
dwinter
parents: 3
diff changeset
185 query["%s__gte"%field]=year%(int(x))
10733d367831 date ranges
dwinter
parents: 3
diff changeset
186
10733d367831 date ranges
dwinter
parents: 3
diff changeset
187 query["%s__lte"%field]=year%(x+increment)
10733d367831 date ranges
dwinter
parents: 3
diff changeset
188
10733d367831 date ranges
dwinter
parents: 3
diff changeset
189 key="%s@%s"%(year%(int(x)),year%(x+increment))
10733d367831 date ranges
dwinter
parents: 3
diff changeset
190
10733d367831 date ranges
dwinter
parents: 3
diff changeset
191 result = q.query(**query).highlight("text_german").execute()
0
834706423ac1 initial
dwinter
parents:
diff changeset
192
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
193 res[key]=result.result.numFound
10733d367831 date ranges
dwinter
parents: 3
diff changeset
194
10733d367831 date ranges
dwinter
parents: 3
diff changeset
195 if storeFields:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
196 fls[key] = result.highlighting;
10733d367831 date ranges
dwinter
parents: 3
diff changeset
197
0
834706423ac1 initial
dwinter
parents:
diff changeset
198
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
199
10733d367831 date ranges
dwinter
parents: 3
diff changeset
200 if storeFields:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
201 return res,fls
10733d367831 date ranges
dwinter
parents: 3
diff changeset
202
10733d367831 date ranges
dwinter
parents: 3
diff changeset
203 else:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
204 return res
0
834706423ac1 initial
dwinter
parents:
diff changeset
205
834706423ac1 initial
dwinter
parents:
diff changeset
206
834706423ac1 initial
dwinter
parents:
diff changeset
207 #prepareSearch erzeugt die Suchabfrage
834706423ac1 initial
dwinter
parents:
diff changeset
208
834706423ac1 initial
dwinter
parents:
diff changeset
209 #solrQuery sucht im in schema.xml bzw. solrconfig.xml festgelegt generischen Feld, hierbei werden mit blanks getrennte eintrage in "AND" zerlegt.
834706423ac1 initial
dwinter
parents:
diff changeset
210 #TODO: erlaube auch suche nach phrasen mit ""
834706423ac1 initial
dwinter
parents:
diff changeset
211 #facetFields:Liste der Felder, nach denen facitiert werden sollen, wirdn in facet_by in solrburn uebersetzt.
834706423ac1 initial
dwinter
parents:
diff changeset
212 #start: Erste Eintrag für paginierung, anzahl der Treffer
834706423ac1 initial
dwinter
parents:
diff changeset
213 #rows: anzahl der Treffer
834706423ac1 initial
dwinter
parents:
diff changeset
214 #facetSearch: Hash mit Feldnamen: suchwort oder Feldname: liste von Suchworten, hast wird direkt an query von solrburnt weitergegeben,
834706423ac1 initial
dwinter
parents:
diff changeset
215 #sortfacets: if true, dann werden die Ergebnisse der facetierten suche alphabetisch sortiert, ACHTUNG: das ist nicht gleich der Funktion in solr die Liste
834706423ac1 initial
dwinter
parents:
diff changeset
216 #direkt sortiert zurückzubekommen, hier werden die haufigsten Werte genommen (einstellt in solrconfig.xml) und dann nur diese sortiert!
834706423ac1 initial
dwinter
parents:
diff changeset
217 #ausserdem werden beim sortieren, die stopworte gefiltert!
834706423ac1 initial
dwinter
parents:
diff changeset
218 #
834706423ac1 initial
dwinter
parents:
diff changeset
219 #neben den direkten parameter koennen auch parameter fuer die facetierte Suche über FORM im REQUEST übergeben werden, diese Felder müssen dann
834706423ac1 initial
dwinter
parents:
diff changeset
220 #mit "_fc" enden.
834706423ac1 initial
dwinter
parents:
diff changeset
221 # gibt als ergebnis den folgenden Hash, so wie in http://opensource.timetric.com/sunburnt/queryingsolr.html#executing-queries-and-interpreting-the-response,
834706423ac1 initial
dwinter
parents:
diff changeset
222 # http://opensource.timetric.com/sunburnt/queryingsolr.html#highlighting
834706423ac1 initial
dwinter
parents:
diff changeset
223 # und http://opensource.timetric.com/sunburnt/queryingsolr.html#faceting
834706423ac1 initial
dwinter
parents:
diff changeset
224 #dokumentiert. Highlighting selbst wird in sorlconfig.xml konfiguriert.
834706423ac1 initial
dwinter
parents:
diff changeset
225 #return ein Queryobjet, zur eigentlichen Suche muss darauf noch execute ausgeführt werden.
834706423ac1 initial
dwinter
parents:
diff changeset
226
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
227 def prepareSearch(self,solrQuery,facetFields=[],start=0,rows=10,facetSearch=None,sortFacets=True,orSearch=None):
0
834706423ac1 initial
dwinter
parents:
diff changeset
228 "search solr"
834706423ac1 initial
dwinter
parents:
diff changeset
229
834706423ac1 initial
dwinter
parents:
diff changeset
230
834706423ac1 initial
dwinter
parents:
diff changeset
231 ranges={}
834706423ac1 initial
dwinter
parents:
diff changeset
232 ## deal with a form
834706423ac1 initial
dwinter
parents:
diff changeset
233 if self.REQUEST:
834706423ac1 initial
dwinter
parents:
diff changeset
234 constr = self.REQUEST.form
3
156c1db5a701 print statements deleted
dwinter
parents: 2
diff changeset
235
0
834706423ac1 initial
dwinter
parents:
diff changeset
236 for field in constr.keys():
834706423ac1 initial
dwinter
parents:
diff changeset
237
834706423ac1 initial
dwinter
parents:
diff changeset
238 #facetes
834706423ac1 initial
dwinter
parents:
diff changeset
239 if field.endswith("_fc"):
834706423ac1 initial
dwinter
parents:
diff changeset
240 if facetSearch is None:
834706423ac1 initial
dwinter
parents:
diff changeset
241 facetSearch={}
834706423ac1 initial
dwinter
parents:
diff changeset
242
834706423ac1 initial
dwinter
parents:
diff changeset
243 vals = constr[field]
834706423ac1 initial
dwinter
parents:
diff changeset
244 if not isinstance(vals,basestring):
834706423ac1 initial
dwinter
parents:
diff changeset
245
834706423ac1 initial
dwinter
parents:
diff changeset
246 vals=[x.decode('utf-8') for x in vals]
834706423ac1 initial
dwinter
parents:
diff changeset
247 else:
834706423ac1 initial
dwinter
parents:
diff changeset
248 vals=vals.decode('utf-8')
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
249
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
250
0
834706423ac1 initial
dwinter
parents:
diff changeset
251 facetSearch[field.replace("_fc",'')]=vals
834706423ac1 initial
dwinter
parents:
diff changeset
252 #ranges form a-b
834706423ac1 initial
dwinter
parents:
diff changeset
253
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
254
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
255 if field.endswith("_or"):
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
256 if orSearch is None:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
257 orSearch={}
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
258
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
259 vals = constr[field]
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
260 if not isinstance(vals,basestring):
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
261
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
262 vals=[x.decode('utf-8') for x in vals]
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
263 else:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
264 vals=vals.decode('utf-8')
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
265
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
266
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
267 orSearch[field.replace("_or",'')]=vals
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
268 #ranges form a-b
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
269
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
270
0
834706423ac1 initial
dwinter
parents:
diff changeset
271
834706423ac1 initial
dwinter
parents:
diff changeset
272 if field.endswith("_rg"):
834706423ac1 initial
dwinter
parents:
diff changeset
273
834706423ac1 initial
dwinter
parents:
diff changeset
274
834706423ac1 initial
dwinter
parents:
diff changeset
275 splitted = "_".split(field)
834706423ac1 initial
dwinter
parents:
diff changeset
276 if len(splitted)==2:
834706423ac1 initial
dwinter
parents:
diff changeset
277 #ranges[field.replace("_rg","__gt")]=splitted[0]
834706423ac1 initial
dwinter
parents:
diff changeset
278 #TODO __gt scheint nicht zu funktionieren wird zu gte (???)
834706423ac1 initial
dwinter
parents:
diff changeset
279 ranges[field.replace("_rg","__gte")]=int(splitted[0])+1
834706423ac1 initial
dwinter
parents:
diff changeset
280 ranges[field.replace("_rg","__lte")]=splitted[1]
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
281
10733d367831 date ranges
dwinter
parents: 3
diff changeset
282 elif field.endswith("_drg"):
10733d367831 date ranges
dwinter
parents: 3
diff changeset
283 splitted = "@".split(field)
10733d367831 date ranges
dwinter
parents: 3
diff changeset
284 if len(splitted)==2:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
285 #ranges[field.replace("_rg","__gt")]=splitted[0]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
286 #TODO __gt scheint nicht zu funktionieren wird zu gte (???)
10733d367831 date ranges
dwinter
parents: 3
diff changeset
287 ranges[field.replace("_drg","__gte")]=splitted[0]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
288 ranges[field.replace("_drg","__lte")]=splitted[1]
0
834706423ac1 initial
dwinter
parents:
diff changeset
289
834706423ac1 initial
dwinter
parents:
diff changeset
290
834706423ac1 initial
dwinter
parents:
diff changeset
291 #teste verbindung zu solr
834706423ac1 initial
dwinter
parents:
diff changeset
292 if not getattr(self,'_v_solr_',None):
834706423ac1 initial
dwinter
parents:
diff changeset
293 self.connect()
834706423ac1 initial
dwinter
parents:
diff changeset
294
834706423ac1 initial
dwinter
parents:
diff changeset
295
834706423ac1 initial
dwinter
parents:
diff changeset
296
834706423ac1 initial
dwinter
parents:
diff changeset
297 solrQuery = solrQuery.decode('utf-8')
834706423ac1 initial
dwinter
parents:
diff changeset
298
834706423ac1 initial
dwinter
parents:
diff changeset
299 #teile die suche nach " " daraus wird dann eine AND suche
834706423ac1 initial
dwinter
parents:
diff changeset
300 #TODO: sollte flexibler sein. insbesondere phrasen
834706423ac1 initial
dwinter
parents:
diff changeset
301 splitted= solrQuery.split(" ")
834706423ac1 initial
dwinter
parents:
diff changeset
302
834706423ac1 initial
dwinter
parents:
diff changeset
303 res = self._v_solr
834706423ac1 initial
dwinter
parents:
diff changeset
304
834706423ac1 initial
dwinter
parents:
diff changeset
305
834706423ac1 initial
dwinter
parents:
diff changeset
306
834706423ac1 initial
dwinter
parents:
diff changeset
307
834706423ac1 initial
dwinter
parents:
diff changeset
308 res=res.query(splitted)
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
309
0
834706423ac1 initial
dwinter
parents:
diff changeset
310 if len(ranges.keys())>0:
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
311
10733d367831 date ranges
dwinter
parents: 3
diff changeset
312
0
834706423ac1 initial
dwinter
parents:
diff changeset
313 res=res.query(ranges)
834706423ac1 initial
dwinter
parents:
diff changeset
314
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
315
0
834706423ac1 initial
dwinter
parents:
diff changeset
316 #ubergebe alle weiteren feld an die suche.
834706423ac1 initial
dwinter
parents:
diff changeset
317 if facetSearch:
834706423ac1 initial
dwinter
parents:
diff changeset
318 for key,vals in facetSearch.items():
834706423ac1 initial
dwinter
parents:
diff changeset
319
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
320 if key.endswith("_rg"): #range
0
834706423ac1 initial
dwinter
parents:
diff changeset
321
834706423ac1 initial
dwinter
parents:
diff changeset
322 if not isinstance(vals,basestring):
834706423ac1 initial
dwinter
parents:
diff changeset
323
834706423ac1 initial
dwinter
parents:
diff changeset
324 vals=[x.decode('utf-8') for x in vals]
834706423ac1 initial
dwinter
parents:
diff changeset
325 else:
834706423ac1 initial
dwinter
parents:
diff changeset
326 vals=[vals.decode('utf-8')]
834706423ac1 initial
dwinter
parents:
diff changeset
327
834706423ac1 initial
dwinter
parents:
diff changeset
328 for val in vals:
834706423ac1 initial
dwinter
parents:
diff changeset
329
834706423ac1 initial
dwinter
parents:
diff changeset
330 splitted = val.split("-")
3
156c1db5a701 print statements deleted
dwinter
parents: 2
diff changeset
331
0
834706423ac1 initial
dwinter
parents:
diff changeset
332 if len(splitted)==2:
834706423ac1 initial
dwinter
parents:
diff changeset
333 #TODO __gt scheint nicht zu funktionieren wird zu gte (???)
834706423ac1 initial
dwinter
parents:
diff changeset
334 facetSearch[key.replace("_rg","__gte")]=int(splitted[0])+1
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
335 facetSearch[key.replace("_rg","__lte")]=int(splitted[1])
3
156c1db5a701 print statements deleted
dwinter
parents: 2
diff changeset
336
0
834706423ac1 initial
dwinter
parents:
diff changeset
337 del facetSearch[key] # loesche das urspuerngliche feld
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
338
10733d367831 date ranges
dwinter
parents: 3
diff changeset
339 elif key.endswith("_drg"): #daterange
10733d367831 date ranges
dwinter
parents: 3
diff changeset
340
10733d367831 date ranges
dwinter
parents: 3
diff changeset
341 if not isinstance(vals,basestring):
10733d367831 date ranges
dwinter
parents: 3
diff changeset
342
10733d367831 date ranges
dwinter
parents: 3
diff changeset
343 vals=[x.decode('utf-8') for x in vals]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
344 else:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
345 vals=[vals.decode('utf-8')]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
346
10733d367831 date ranges
dwinter
parents: 3
diff changeset
347 for val in vals:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
348
10733d367831 date ranges
dwinter
parents: 3
diff changeset
349 splitted = val.split("@")
10733d367831 date ranges
dwinter
parents: 3
diff changeset
350
10733d367831 date ranges
dwinter
parents: 3
diff changeset
351 if len(splitted)==2:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
352 #TODO __gt scheint nicht zu funktionieren wird zu gte (???)
10733d367831 date ranges
dwinter
parents: 3
diff changeset
353 facetSearch[key.replace("_drg","__gte")]=splitted[0]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
354 facetSearch[key.replace("_drg","__lte")]=splitted[1]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
355
10733d367831 date ranges
dwinter
parents: 3
diff changeset
356 del facetSearch[key] # loesche das urspuerngliche feld
10733d367831 date ranges
dwinter
parents: 3
diff changeset
357
0
834706423ac1 initial
dwinter
parents:
diff changeset
358
834706423ac1 initial
dwinter
parents:
diff changeset
359 else:
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
360 if not isinstance(vals, basestring):
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
361 val = [x for x in vals if x!="*"] #siehe oben
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
362 else:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
363 val = [vals]
0
834706423ac1 initial
dwinter
parents:
diff changeset
364 facetSearch[key]=val
834706423ac1 initial
dwinter
parents:
diff changeset
365
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
366
0
834706423ac1 initial
dwinter
parents:
diff changeset
367 res=res.query(**facetSearch)
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
368
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
369
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
370 #felder mit ODER-Suche
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
371 if orSearch:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
372
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
373 for key,vals in orSearch.items():
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
374 qr = None
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
375 if not "*" in vals: #dann ohne einschränkung (key:*) sucht nur nach eintragen in denen etwas im feld steht, wir wollen aber alle
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
376 if isinstance(vals, basestring):
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
377 vals = [vals]
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
378
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
379
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
380 for val in vals:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
381 if not qr:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
382 qr=self._v_solr.Q(**{key:val})
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
383
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
384 else:
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
385 qr=qr|self._v_solr.Q(**{key:val})
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
386
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
387
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
388 res=res.query(qr)
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
389
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
390
0
834706423ac1 initial
dwinter
parents:
diff changeset
391
834706423ac1 initial
dwinter
parents:
diff changeset
392 #wenn facetField existieren dann rufe facetierung auf
834706423ac1 initial
dwinter
parents:
diff changeset
393 if len(facetFields)>0:
834706423ac1 initial
dwinter
parents:
diff changeset
394 #for facet in facetFields:
834706423ac1 initial
dwinter
parents:
diff changeset
395
834706423ac1 initial
dwinter
parents:
diff changeset
396 # res = res.facet_by(facet)
834706423ac1 initial
dwinter
parents:
diff changeset
397 res = res.facet_by(facetFields)
834706423ac1 initial
dwinter
parents:
diff changeset
398
834706423ac1 initial
dwinter
parents:
diff changeset
399 #res=res.paginate(start=start, rows=rows).highlight("main_content")
834706423ac1 initial
dwinter
parents:
diff changeset
400
834706423ac1 initial
dwinter
parents:
diff changeset
401 res=res.paginate(start=start, rows=rows)
834706423ac1 initial
dwinter
parents:
diff changeset
402
834706423ac1 initial
dwinter
parents:
diff changeset
403
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
404 #only highlighting if not searhc for only "*" - avoid max clause error
2
de6b7ed0c34d highlighting only enabled if not search for "*".
dwinter
parents: 1
diff changeset
405 if solrQuery=="*":
3
156c1db5a701 print statements deleted
dwinter
parents: 2
diff changeset
406
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
407
2
de6b7ed0c34d highlighting only enabled if not search for "*".
dwinter
parents: 1
diff changeset
408 res=res.paginate(start=start, rows=rows)
de6b7ed0c34d highlighting only enabled if not search for "*".
dwinter
parents: 1
diff changeset
409 else:
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
410
2
de6b7ed0c34d highlighting only enabled if not search for "*".
dwinter
parents: 1
diff changeset
411 res=res.paginate(start=start, rows=rows).highlight(usePhraseHighlighter=True)
0
834706423ac1 initial
dwinter
parents:
diff changeset
412
834706423ac1 initial
dwinter
parents:
diff changeset
413 return res
834706423ac1 initial
dwinter
parents:
diff changeset
414
834706423ac1 initial
dwinter
parents:
diff changeset
415
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
416 def replaceParameter(self,paramsNeu,queryString,ignore=[]):
10733d367831 date ranges
dwinter
parents: 3
diff changeset
417 params=urlparse.parse_qs(queryString);
10733d367831 date ranges
dwinter
parents: 3
diff changeset
418
10733d367831 date ranges
dwinter
parents: 3
diff changeset
419 for key in paramsNeu.keys():
10733d367831 date ranges
dwinter
parents: 3
diff changeset
420
10733d367831 date ranges
dwinter
parents: 3
diff changeset
421 params[key]=paramsNeu[key]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
422
10733d367831 date ranges
dwinter
parents: 3
diff changeset
423 for key in ignore:
10733d367831 date ranges
dwinter
parents: 3
diff changeset
424 del params[key]
10733d367831 date ranges
dwinter
parents: 3
diff changeset
425
10733d367831 date ranges
dwinter
parents: 3
diff changeset
426 retArray = []
10733d367831 date ranges
dwinter
parents: 3
diff changeset
427 for x in params.keys():
10733d367831 date ranges
dwinter
parents: 3
diff changeset
428 for y in self.getList(params[x]):
10733d367831 date ranges
dwinter
parents: 3
diff changeset
429 retArray.append("%s=%s"%(x,urllib.quote(y)))
10733d367831 date ranges
dwinter
parents: 3
diff changeset
430
10733d367831 date ranges
dwinter
parents: 3
diff changeset
431 print retArray
10733d367831 date ranges
dwinter
parents: 3
diff changeset
432 return "&".join(retArray);
0
834706423ac1 initial
dwinter
parents:
diff changeset
433 #für die parameter @see prepareSearch
834706423ac1 initial
dwinter
parents:
diff changeset
434 #erzeugt eine Suchabfrage und führt diese aus.
834706423ac1 initial
dwinter
parents:
diff changeset
435 #return {"result":response.result, "hl":response.highlighting,"facetFields":facetedFields}
834706423ac1 initial
dwinter
parents:
diff changeset
436
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
437 def search(self,solrQuery,facetFields=[],start=0,rows=10,facetSearch=None,sortFacets=True,orSearch=None):
0
834706423ac1 initial
dwinter
parents:
diff changeset
438
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
439 res=self.prepareSearch(solrQuery,facetFields,start,rows,facetSearch,sortFacets,orSearch=orSearch)
0
834706423ac1 initial
dwinter
parents:
diff changeset
440 response= res.execute()
834706423ac1 initial
dwinter
parents:
diff changeset
441
834706423ac1 initial
dwinter
parents:
diff changeset
442 #speichere faceted fields
834706423ac1 initial
dwinter
parents:
diff changeset
443 if len(facetFields)>0:
834706423ac1 initial
dwinter
parents:
diff changeset
444 facetedFields=response.facet_counts.facet_fields
834706423ac1 initial
dwinter
parents:
diff changeset
445
834706423ac1 initial
dwinter
parents:
diff changeset
446 if sortFacets:
834706423ac1 initial
dwinter
parents:
diff changeset
447 facetedFields=self.sortFacetedFields(facetedFields)
834706423ac1 initial
dwinter
parents:
diff changeset
448
834706423ac1 initial
dwinter
parents:
diff changeset
449 else:
834706423ac1 initial
dwinter
parents:
diff changeset
450 facetedFields={}
834706423ac1 initial
dwinter
parents:
diff changeset
451
834706423ac1 initial
dwinter
parents:
diff changeset
452
834706423ac1 initial
dwinter
parents:
diff changeset
453
834706423ac1 initial
dwinter
parents:
diff changeset
454
834706423ac1 initial
dwinter
parents:
diff changeset
455 return {"result":response.result, "hl":response.highlighting,"facetFields":facetedFields}
834706423ac1 initial
dwinter
parents:
diff changeset
456
834706423ac1 initial
dwinter
parents:
diff changeset
457
834706423ac1 initial
dwinter
parents:
diff changeset
458
834706423ac1 initial
dwinter
parents:
diff changeset
459 #hilfsmethode zum sortieren über der ranges
834706423ac1 initial
dwinter
parents:
diff changeset
460 def sortRanges(self,ranges):
834706423ac1 initial
dwinter
parents:
diff changeset
461 x=list(ranges)
834706423ac1 initial
dwinter
parents:
diff changeset
462 x.sort()
834706423ac1 initial
dwinter
parents:
diff changeset
463 return x
834706423ac1 initial
dwinter
parents:
diff changeset
464
834706423ac1 initial
dwinter
parents:
diff changeset
465 #sortiert die Werte der FacetedFields
834706423ac1 initial
dwinter
parents:
diff changeset
466 #(facetedFields ist ein hast mit feldname -> liste der (wert für das feld, anzahl der treffer für den wert)
834706423ac1 initial
dwinter
parents:
diff changeset
467 #ausserdem werden die werte gemäß des angegebenen Filter gefiltert.
834706423ac1 initial
dwinter
parents:
diff changeset
468
834706423ac1 initial
dwinter
parents:
diff changeset
469 def sortFacetedFields(self, facetedFields,filter=STOPLIST):
834706423ac1 initial
dwinter
parents:
diff changeset
470 ret={}
834706423ac1 initial
dwinter
parents:
diff changeset
471
834706423ac1 initial
dwinter
parents:
diff changeset
472 def cmpTuple(x,y):
834706423ac1 initial
dwinter
parents:
diff changeset
473 return cmp(x[0],y[0])
834706423ac1 initial
dwinter
parents:
diff changeset
474
834706423ac1 initial
dwinter
parents:
diff changeset
475
834706423ac1 initial
dwinter
parents:
diff changeset
476 for key in facetedFields.keys():
834706423ac1 initial
dwinter
parents:
diff changeset
477 ls = facetedFields[key]
834706423ac1 initial
dwinter
parents:
diff changeset
478 ls.sort(cmpTuple)
834706423ac1 initial
dwinter
parents:
diff changeset
479 ret[key]=ls
834706423ac1 initial
dwinter
parents:
diff changeset
480
834706423ac1 initial
dwinter
parents:
diff changeset
481 if filter.get(key,None):
834706423ac1 initial
dwinter
parents:
diff changeset
482 ls2=[]
834706423ac1 initial
dwinter
parents:
diff changeset
483 for x in ls:
834706423ac1 initial
dwinter
parents:
diff changeset
484 if x[0].encode('utf-8') not in filter[key]:
834706423ac1 initial
dwinter
parents:
diff changeset
485 ls2.append(x)
834706423ac1 initial
dwinter
parents:
diff changeset
486 ret[key]=ls2
834706423ac1 initial
dwinter
parents:
diff changeset
487
834706423ac1 initial
dwinter
parents:
diff changeset
488
834706423ac1 initial
dwinter
parents:
diff changeset
489 ret[key]=[x for x in ret[key] if x[1]!=0]
834706423ac1 initial
dwinter
parents:
diff changeset
490 return ret
834706423ac1 initial
dwinter
parents:
diff changeset
491
834706423ac1 initial
dwinter
parents:
diff changeset
492 def changeMain(self,solrURL=None,title=None,REQUEST=None,RESPONSE=None):
834706423ac1 initial
dwinter
parents:
diff changeset
493 """change main settings"""
834706423ac1 initial
dwinter
parents:
diff changeset
494 if solrURL:
834706423ac1 initial
dwinter
parents:
diff changeset
495 self.solrURL=solrURL
834706423ac1 initial
dwinter
parents:
diff changeset
496 self.title=title
834706423ac1 initial
dwinter
parents:
diff changeset
497 self._v_solr=sunburnt.SolrInterface(url=solrURL)
834706423ac1 initial
dwinter
parents:
diff changeset
498
834706423ac1 initial
dwinter
parents:
diff changeset
499 if RESPONSE is not None:
834706423ac1 initial
dwinter
parents:
diff changeset
500 RESPONSE.redirect('manage_main')
834706423ac1 initial
dwinter
parents:
diff changeset
501
834706423ac1 initial
dwinter
parents:
diff changeset
502
834706423ac1 initial
dwinter
parents:
diff changeset
503 else:
834706423ac1 initial
dwinter
parents:
diff changeset
504 pt=zptFile(self, 'zpt/ChangeZopeSolr.zpt')
834706423ac1 initial
dwinter
parents:
diff changeset
505 return pt()
834706423ac1 initial
dwinter
parents:
diff changeset
506
834706423ac1 initial
dwinter
parents:
diff changeset
507 #sucht die je nach einsteillung in solrconfig.xml Werte mit den häufigsten Treffern oder alphabetisch sortiert zu einem
834706423ac1 initial
dwinter
parents:
diff changeset
508 #bestimmten eintrag aus solr
834706423ac1 initial
dwinter
parents:
diff changeset
509 #gedacht ist die methoden für die Anwendung nach dem Harvesten einer Website, es wird daher davon ausgegangen, dass sie hinter der
834706423ac1 initial
dwinter
parents:
diff changeset
510 #idfield eine url steht. Wir es z.b. bei Nutch passiert. index.htm/index_html als Teil der url wird dabei unterdrückt (analog zu den harbest einstellunge für
834706423ac1 initial
dwinter
parents:
diff changeset
511 # nutch für zope webseiten.
834706423ac1 initial
dwinter
parents:
diff changeset
512 #@param @idfield is hierbei der Name des Feldes, das in solrschema als id defniert wurde
834706423ac1 initial
dwinter
parents:
diff changeset
513 #field der Feldname von dem die Treffer gesurcht wernde sollen
834706423ac1 initial
dwinter
parents:
diff changeset
514 #url die url des textes
834706423ac1 initial
dwinter
parents:
diff changeset
515
834706423ac1 initial
dwinter
parents:
diff changeset
516 def getTermsAsJSON(self,idfield,field,url):
834706423ac1 initial
dwinter
parents:
diff changeset
517 """getTerms"""
834706423ac1 initial
dwinter
parents:
diff changeset
518
834706423ac1 initial
dwinter
parents:
diff changeset
519 ret=[]
834706423ac1 initial
dwinter
parents:
diff changeset
520 h = httplib2.Http()
834706423ac1 initial
dwinter
parents:
diff changeset
521 url = url.replace("/index.html","").replace("/index_html","")
834706423ac1 initial
dwinter
parents:
diff changeset
522 if url[-1]=="/":
834706423ac1 initial
dwinter
parents:
diff changeset
523 url=url[0:-1]
834706423ac1 initial
dwinter
parents:
diff changeset
524
834706423ac1 initial
dwinter
parents:
diff changeset
525 urlq=url.replace(":","\:")
834706423ac1 initial
dwinter
parents:
diff changeset
526 urlq=urlq.replace("/","\/")
834706423ac1 initial
dwinter
parents:
diff changeset
527
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
528 #q ="http://localhost:8983/solr/mpiwgweb/select?q=%s:%s&wt=xml&tv=on&qt=tvrh&fl=%s&tv.tf=true"%(idfield,urlq,field)
10733d367831 date ranges
dwinter
parents: 3
diff changeset
529 q =self.solrURL+SOLRSERVER%(idfield,urlq,field)
0
834706423ac1 initial
dwinter
parents:
diff changeset
530
834706423ac1 initial
dwinter
parents:
diff changeset
531 resp, content = h.request(q)
834706423ac1 initial
dwinter
parents:
diff changeset
532
834706423ac1 initial
dwinter
parents:
diff changeset
533 root = ET.fromstring(content)
834706423ac1 initial
dwinter
parents:
diff changeset
534 #uri = "http://127.0.0.1:18080/www_neu/de/aktuelles/features/feature28"
834706423ac1 initial
dwinter
parents:
diff changeset
535
834706423ac1 initial
dwinter
parents:
diff changeset
536 xpstr = ".//lst[@name='termVectors']/lst[@name='%s']/lst[@name='%s']/lst"%(url,field)
834706423ac1 initial
dwinter
parents:
diff changeset
537
834706423ac1 initial
dwinter
parents:
diff changeset
538
834706423ac1 initial
dwinter
parents:
diff changeset
539 tvs = root.findall(xpstr)
834706423ac1 initial
dwinter
parents:
diff changeset
540
834706423ac1 initial
dwinter
parents:
diff changeset
541 for tv in tvs:
834706423ac1 initial
dwinter
parents:
diff changeset
542 wd = tv.attrib['name']
834706423ac1 initial
dwinter
parents:
diff changeset
543 for f in tv.findall("./int[@name='tf']"):
834706423ac1 initial
dwinter
parents:
diff changeset
544 fre = f.text
834706423ac1 initial
dwinter
parents:
diff changeset
545
834706423ac1 initial
dwinter
parents:
diff changeset
546
834706423ac1 initial
dwinter
parents:
diff changeset
547 if int(fre)>2:
834706423ac1 initial
dwinter
parents:
diff changeset
548
834706423ac1 initial
dwinter
parents:
diff changeset
549 ret.append('{"text":"%s","size":%s}'%(wd,fre))
834706423ac1 initial
dwinter
parents:
diff changeset
550
834706423ac1 initial
dwinter
parents:
diff changeset
551 retStr="["+",".join(ret)+"]"
834706423ac1 initial
dwinter
parents:
diff changeset
552
834706423ac1 initial
dwinter
parents:
diff changeset
553 return retStr
834706423ac1 initial
dwinter
parents:
diff changeset
554
834706423ac1 initial
dwinter
parents:
diff changeset
555 #tauscht im request die in neewparams angegeben parameter aus.
834706423ac1 initial
dwinter
parents:
diff changeset
556 def replaceParam(self, newparams):
834706423ac1 initial
dwinter
parents:
diff changeset
557 x = self.REQUEST.form.copy()
834706423ac1 initial
dwinter
parents:
diff changeset
558
834706423ac1 initial
dwinter
parents:
diff changeset
559 for key,value in newparams.items():
834706423ac1 initial
dwinter
parents:
diff changeset
560 x[key]=value
834706423ac1 initial
dwinter
parents:
diff changeset
561
834706423ac1 initial
dwinter
parents:
diff changeset
562
834706423ac1 initial
dwinter
parents:
diff changeset
563 retls=[]
834706423ac1 initial
dwinter
parents:
diff changeset
564 for k,v in x.items():
834706423ac1 initial
dwinter
parents:
diff changeset
565 if not isinstance(v,basestring):
834706423ac1 initial
dwinter
parents:
diff changeset
566 for y in v:
834706423ac1 initial
dwinter
parents:
diff changeset
567 retls.append((k,y))
834706423ac1 initial
dwinter
parents:
diff changeset
568 else:
834706423ac1 initial
dwinter
parents:
diff changeset
569 retls.append((k,v))
834706423ac1 initial
dwinter
parents:
diff changeset
570
834706423ac1 initial
dwinter
parents:
diff changeset
571 return "?"+"&".join(["%s=%s"%(k,urllib.quote_plus(v,'/')) for (k, v) in retls])
834706423ac1 initial
dwinter
parents:
diff changeset
572
834706423ac1 initial
dwinter
parents:
diff changeset
573
834706423ac1 initial
dwinter
parents:
diff changeset
574
834706423ac1 initial
dwinter
parents:
diff changeset
575
834706423ac1 initial
dwinter
parents:
diff changeset
576 #ruft @set ranges aus, gibt das ergebnis als json zurück
834706423ac1 initial
dwinter
parents:
diff changeset
577 def getRangesAsJSON(self,field,begin,end,increment):
834706423ac1 initial
dwinter
parents:
diff changeset
578 """ getRangesAsJSON"""
834706423ac1 initial
dwinter
parents:
diff changeset
579 res = self.ranges(field, int(begin), int(end), int(increment))
834706423ac1 initial
dwinter
parents:
diff changeset
580 return json.dumps(res)
834706423ac1 initial
dwinter
parents:
diff changeset
581
834706423ac1 initial
dwinter
parents:
diff changeset
582
834706423ac1 initial
dwinter
parents:
diff changeset
583 #return only the values of resultList whicht start with startLetter or if starLetterNonAscii all values which
834706423ac1 initial
dwinter
parents:
diff changeset
584 #start with an non ascii character
834706423ac1 initial
dwinter
parents:
diff changeset
585 def filter (self,resultList,startLetter=None,startLetterNonAscii=0):
1
9c356845613a improvements for the search or search fields added
dwinter
parents: 0
diff changeset
586 ls=[]
0
834706423ac1 initial
dwinter
parents:
diff changeset
587 if startLetter:
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
588 startLetter=startLetter.lower()
0
834706423ac1 initial
dwinter
parents:
diff changeset
589
834706423ac1 initial
dwinter
parents:
diff changeset
590 matchStr = "[\[\]'\"]*"+startLetter
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
591 ls = [x for x in resultList if re.match(matchStr,x[0].lower())]
0
834706423ac1 initial
dwinter
parents:
diff changeset
592
834706423ac1 initial
dwinter
parents:
diff changeset
593 if startLetterNonAscii ==1:
834706423ac1 initial
dwinter
parents:
diff changeset
594
834706423ac1 initial
dwinter
parents:
diff changeset
595 ls = [x for x in resultList if not re.match("[\[\]'\"a-zA-Z].*",x[0])]
834706423ac1 initial
dwinter
parents:
diff changeset
596 return ls
834706423ac1 initial
dwinter
parents:
diff changeset
597
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
598
10733d367831 date ranges
dwinter
parents: 3
diff changeset
599
0
834706423ac1 initial
dwinter
parents:
diff changeset
600 def manage_addZopeSolrForm(self):
834706423ac1 initial
dwinter
parents:
diff changeset
601 """Form for external Links"""
834706423ac1 initial
dwinter
parents:
diff changeset
602 pt=zptFile(self, 'zpt/AddZopeSolr.zpt')
834706423ac1 initial
dwinter
parents:
diff changeset
603 return pt()
834706423ac1 initial
dwinter
parents:
diff changeset
604
834706423ac1 initial
dwinter
parents:
diff changeset
605
834706423ac1 initial
dwinter
parents:
diff changeset
606 def manage_addZopeSolr(self,id,title,solrURL,RESPONSE=None):
834706423ac1 initial
dwinter
parents:
diff changeset
607 """Add an external Link"""
834706423ac1 initial
dwinter
parents:
diff changeset
608
834706423ac1 initial
dwinter
parents:
diff changeset
609 newObj=ZopeSolr(id,title,solrURL)
834706423ac1 initial
dwinter
parents:
diff changeset
610
834706423ac1 initial
dwinter
parents:
diff changeset
611 self._setObject(id,newObj)
834706423ac1 initial
dwinter
parents:
diff changeset
612
834706423ac1 initial
dwinter
parents:
diff changeset
613
834706423ac1 initial
dwinter
parents:
diff changeset
614 if RESPONSE is not None:
834706423ac1 initial
dwinter
parents:
diff changeset
615 RESPONSE.redirect('manage_main')
4
10733d367831 date ranges
dwinter
parents: 3
diff changeset
616