Mercurial > hg > documentViewer
annotate MpdlXmlTextServer.py @ 418:bed1ff5eaec1
fix getTocPages et al
author | casties |
---|---|
date | Thu, 23 Dec 2010 18:13:01 +0100 |
parents | a7dc563103c7 |
children | 69205c9d9404 |
rev | line source |
---|---|
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
1 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
2 from OFS.SimpleItem import SimpleItem |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
3 from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
4 from Ft.Xml import EMPTY_NAMESPACE, Parse |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
5 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
6 import sys |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
7 import logging |
134 | 8 import urllib |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
9 import documentViewer |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
10 from documentViewer import getTextFromNode, serializeNode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
11 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
12 class MpdlXmlTextServer(SimpleItem): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
13 """TextServer implementation for MPDL-XML eXist server""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
14 meta_type="MPDL-XML TextServer" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
15 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
16 manage_options=( |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
17 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'}, |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
18 )+SimpleItem.manage_options |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
19 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
20 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals()) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
21 |
132 | 22 def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
23 """constructor""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
24 self.id=id |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
25 self.title=title |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
26 self.timeout = timeout |
132 | 27 if serverName is None: |
28 self.serverUrl = serverUrl | |
29 else: | |
30 self.serverUrl = "http://%s/mpdl/interface/"%serverName | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
31 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
32 def getHttpData(self, url, data=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
33 """returns result from url+data HTTP request""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
34 return documentViewer.getHttpData(url,data,timeout=self.timeout) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
35 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
36 def getServerData(self, method, data=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
37 """returns result from text server for method+data""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
38 url = self.serverUrl+method |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
39 return documentViewer.getHttpData(url,data,timeout=self.timeout) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
40 |
398 | 41 def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None, characterNormalization=None, optionToggle=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
42 """get search list""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
43 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
44 url = docinfo['url'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
45 pagesize = pageinfo['queryPageSize'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
46 pn = pageinfo['searchPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
47 sn = pageinfo['sn'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
48 highlightQuery = pageinfo['highlightQuery'] |
181 | 49 query =pageinfo['query'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
50 queryType =pageinfo['queryType'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
51 viewMode= pageinfo['viewMode'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
52 tocMode = pageinfo['tocMode'] |
169 | 53 characterNormalization = pageinfo['characterNormalization'] |
399 | 54 optionToggle = pageinfo['optionToggle'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
55 tocPN = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
56 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
57 |
399 | 58 data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&optionToggle=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization,optionToggle ,urllib.quote(highlightQuery))) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
59 #page=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, query, pagesize, pn, sn, viewMode,highlightQuery) ,outputUnicode=False) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
60 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
61 pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
62 pagedom = Parse(pagexml) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
63 if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
64 pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
65 if len(pagedivs)>0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
66 pagenode=pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
67 links=pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
68 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
69 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
70 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
71 href = hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
72 if href.startswith('page-fragment.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
73 selfurl = self.absolute_url() |
407 | 74 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
75 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
76 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
77 if (queryType=="fulltextMorph"): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
78 pagedivs = pagedom.xpath("//div[@class='queryResult']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
79 if len(pagedivs)>0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
80 pagenode=pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
81 links=pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
82 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
83 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
84 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
85 href = hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
86 if href.startswith('page-fragment.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
87 selfurl = self.absolute_url() |
407 | 88 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
89 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
90 if href.startswith('../lt/lemma.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
91 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl)) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
92 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
93 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
138 | 94 l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
95 pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
96 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
97 if (queryType=="ftIndex")or(queryType=="ftIndexMorph"): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
98 pagedivs= pagedom.xpath("//div[@class='queryResultPage']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
99 if len(pagedivs)>0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
100 pagenode=pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
101 links=pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
102 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
103 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
104 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
105 href = hrefNode.nodeValue |
407 | 106 hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,optionToggle,characterNormalization)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
107 if href.startswith('../lt/lex.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
108 hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
109 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
110 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
138 | 111 l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
112 if href.startswith('../lt/lemma.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
113 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
114 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
115 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
138 | 116 l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
117 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
118 return "no text here" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
119 |
412 | 120 |
121 def getNumTextPages (self, docinfo=None): | |
190 | 122 """get list of pages from fulltext (texts without images) and put in docinfo""" |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
123 logging.debug("getNumTextPages") |
412 | 124 docpath = docinfo['textURLPath'] |
125 selfurl = self.absolute_url() | |
126 #viewMode= pageinfo['viewMode'] | |
205 | 127 if 'numPages' in docinfo: |
190 | 128 # allredy there |
205 | 129 return docinfo |
412 | 130 |
131 text = self.getServerData("page-fragment.xql","document=%s"%(docinfo['textURLPath'])) | |
132 dom =Parse(text) | |
133 pagedivs = dom.xpath("//div[@class='countPages']") | |
134 logging.debug ("pagedivs=%s"%(pagedivs)) | |
135 if len(pagedivs)>0: | |
136 docinfo['numPages']= int(getTextFromNode(pagedivs[0])) | |
137 return docinfo | |
138 | |
139 def getTocEntries (self, docinfo=None): | |
140 """ number of text entries""" | |
418 | 141 docinfo = self.getInfoFromPage(docinfo) |
142 logging.debug("gettocentries: docinfo=%s"%docinfo) | |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
143 return docinfo['tocEntries'] |
395 | 144 |
412 | 145 def getFigureEntries (self, docinfo=None): |
146 """ number of figure entries""" | |
418 | 147 docinfo = self.getInfoFromPage(docinfo) |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
148 return docinfo['figureEntries'] |
412 | 149 |
253 | 150 def getGisPlaces(self, docinfo=None, pageinfo=None): |
207 | 151 """ Show all Gis Places of whole Page""" |
264 | 152 xpath='//place' |
409 | 153 docpath = docinfo.get('textURLPath',None) |
154 if not docpath: | |
155 return None | |
156 | |
253 | 157 url = docinfo['url'] |
158 selfurl = self.absolute_url() | |
257 | 159 pn = pageinfo['current'] |
291 | 160 hrefList=[] |
306 | 161 myList= "" |
264 | 162 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn)) |
163 dom = Parse(text) | |
265 | 164 result = dom.xpath("//result/resultPage/place") |
236 | 165 for l in result: |
250 | 166 hrefNode= l.getAttributeNodeNS(None, u"id") |
272 | 167 href= hrefNode.nodeValue |
292 | 168 hrefList.append(href) |
309 | 169 myList = ",".join(hrefList) |
170 logging.debug("getGisPlaces :%s"%(myList)) | |
307 | 171 return myList |
172 | |
173 def getAllGisPlaces (self, docinfo=None, pageinfo=None): | |
174 """Show all Gis Places of whole Book """ | |
175 xpath ='//echo:place' | |
176 docpath =docinfo['textURLPath'] | |
177 url = docinfo['url'] | |
178 selfurl =self.absolute_url() | |
179 pn =pageinfo['current'] | |
180 hrefList=[] | |
181 myList="" | |
182 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath)) | |
183 dom =Parse(text) | |
184 result = dom.xpath("//result/resultPage/place") | |
395 | 185 |
307 | 186 for l in result: |
187 hrefNode = l.getAttributeNodeNS(None, u"id") | |
188 href= hrefNode.nodeValue | |
189 hrefList.append(href) | |
300 | 190 myList = ",".join(hrefList) |
309 | 191 logging.debug("getALLGisPlaces :%s"%(myList)) |
192 return myList | |
412 | 193 |
360 | 194 def getOrigPages (self, docinfo=None, pageinfo=None): |
310 | 195 """Show original page """ |
409 | 196 docpath = docinfo.get('textURLPath',None) |
197 if not docpath: | |
198 return None | |
310 | 199 selfurl = self.absolute_url() |
349 | 200 pn =pageinfo['current'] |
325 | 201 |
349 | 202 viewMode= pageinfo['viewMode'] |
203 text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn)) | |
310 | 204 dom =Parse(text) |
205 pagedivs = dom.xpath("//div[@class='pageNumberOrig']") | |
324 | 206 if len(pagedivs)>0: |
360 | 207 originalPage= getTextFromNode(pagedivs[0]) |
358 | 208 #return docinfo['originalPage'] |
355 | 209 return originalPage |
360 | 210 |
414 | 211 def getAllPlaces (self, docinfo=None): |
411 | 212 """Show all Places if no places than 0""" |
418 | 213 docinfo = self.getInfoFromPage(docinfo) |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
214 return docinfo['allPlaces'] |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
215 |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
216 def getInfoFromPage(self, docinfo=None): |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
217 """ extract diverse info from page-fragment""" |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
218 docpath = docinfo['textURLPath'] |
414 | 219 if 'allPlaces' in docinfo: |
220 # allredy there | |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
221 return docinfo |
414 | 222 |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
223 if (docpath is not None): |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
224 text = self.getServerData("page-fragment.xql","document=%s"%(docinfo['textURLPath'])) |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
225 dom = Parse(text) |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
226 # figureEntries |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
227 pagedivs = dom.xpath("//div[@class='countFigureEntries']") |
416
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
228 s = getTextFromNode(pagedivs[0]) |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
229 try: |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
230 docinfo['figureEntries'] = int(s) |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
231 except: |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
232 docinfo['figureEntries'] = 0 |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
233 # tocEntries |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
234 pagedivs = dom.xpath("//div[@class='countTocEntries']") |
416
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
235 s = getTextFromNode(pagedivs[0]) |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
236 try: |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
237 docinfo['tocEntries'] = int(s) |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
238 except: |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
239 docinfo['tocEntries'] = 0 |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
240 # allPlaces |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
241 pagedivs = dom.xpath("//div[@class='countPlaces']") |
416
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
242 s = getTextFromNode(pagedivs[0]) |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
243 try: |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
244 docinfo['allPlaces'] = int(s) |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
245 except: |
1c821283556a
new getInfoFromPage to read multiple parameters at once
casties
parents:
415
diff
changeset
|
246 docinfo['allPlaces'] = 0 |
417
a7dc563103c7
new getInfoFromPage to read multiple parameters at once
casties
parents:
416
diff
changeset
|
247 |
a7dc563103c7
new getInfoFromPage to read multiple parameters at once
casties
parents:
416
diff
changeset
|
248 else: |
a7dc563103c7
new getInfoFromPage to read multiple parameters at once
casties
parents:
416
diff
changeset
|
249 # no full text -- init to 0 |
a7dc563103c7
new getInfoFromPage to read multiple parameters at once
casties
parents:
416
diff
changeset
|
250 docinfo['figureEntries'] = 0 |
a7dc563103c7
new getInfoFromPage to read multiple parameters at once
casties
parents:
416
diff
changeset
|
251 docinfo['tocEntries'] = 0 |
a7dc563103c7
new getInfoFromPage to read multiple parameters at once
casties
parents:
416
diff
changeset
|
252 docinfo['allPlaces'] = 0 |
415
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
253 |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
254 return docinfo |
b93a2e5d5506
new getInfoFromPage to read multiple parameters at once
casties
parents:
414
diff
changeset
|
255 |
411 | 256 |
412 | 257 def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="reg", highlightQuery=None, sn=None, optionToggle=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
258 """returns single page from fulltext""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
259 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
260 path = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
261 url = docinfo['url'] |
237 | 262 name = docinfo['name'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
263 viewMode= pageinfo['viewMode'] |
376 | 264 sn = pageinfo['sn'] |
367 | 265 highlightQuery = pageinfo['highlightQuery'] |
376 | 266 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
267 tocMode = pageinfo['tocMode'] |
164 | 268 characterNormalization=pageinfo['characterNormalization'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
269 tocPN = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
270 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
271 if mode == "text_dict": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
272 textmode = "textPollux" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
273 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
274 textmode = mode |
376 | 275 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
373 | 276 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
370 | 277 if highlightQuery is not None: |
376 | 278 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) |
379 | 279 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
185 | 280 pagexml = self.getServerData("page-fragment.xql",textParam) |
379 | 281 logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
282 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
283 |
186 | 284 pagedom = Parse(pagexml) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
285 # plain text mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
286 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
287 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
288 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
289 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
290 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
291 links = pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
292 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
293 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
294 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
295 href= hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
296 if href.startswith('#note-'): |
172 | 297 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
298 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
299 if mode == "xml": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
300 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
301 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
302 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
303 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
304 return serializeNode(pagenode) |
140 | 305 if mode == "gis": |
306 # first div contains text | |
307 pagedivs = pagedom.xpath("/div") | |
308 if len(pagedivs) > 0: | |
309 pagenode = pagedivs[0] | |
175 | 310 links =pagenode.xpath("//a") |
311 for l in links: | |
312 hrefNode =l.getAttributeNodeNS(None, u"href") | |
313 if hrefNode: | |
314 href=hrefNode.nodeValue | |
315 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): | |
239 | 316 hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name) |
211 | 317 l.setAttributeNS(None, 'target', '_blank') |
210 | 318 return serializeNode(pagenode) |
140 | 319 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
320 if mode == "pureXml": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
321 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
322 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
323 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
324 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
325 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
326 # text-with-links mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
327 if mode == "text_dict": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
328 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
329 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
330 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
331 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
332 # check all a-tags |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
333 links = pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
334 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
335 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
336 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
337 # is link with href |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
338 href = hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
339 if href.startswith('lt/lex.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
340 # is pollux link |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
341 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
342 # change href |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
343 hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
344 # add target |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
345 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
346 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
138 | 347 l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
348 if href.startswith('lt/lemma.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
349 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
350 hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
351 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
352 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
138 | 353 l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
354 if href.startswith('#note-'): |
162 | 355 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
356 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
357 return "no text here" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
358 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
359 def getTranslate(self, query=None, language=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
360 """translate into another languages""" |
134 | 361 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
|
362 #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
|
363 return data |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
364 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
365 def getLemma(self, lemma=None, language=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
366 """simular words lemma """ |
134 | 367 data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
368 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma))) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
369 return data |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
370 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
371 def getLemmaNew(self, query=None, language=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
372 """simular words lemma """ |
134 | 373 data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
374 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query))) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
375 return data |
175 | 376 |
398 | 377 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
378 """number of""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
379 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
380 pagesize = pageinfo['queryPageSize'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
381 pn = pageinfo['searchPN'] |
181 | 382 query =pageinfo['query'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
383 queryType =pageinfo['queryType'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
384 tocSearch = 0 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
385 tocDiv = None |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
386 |
179 | 387 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
|
388 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
389 pagedom = Parse(pagexml) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
390 numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
391 tocSearch = int(getTextFromNode(numdivs[0])) |
393 | 392 logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
393 tc=int((tocSearch/10)+1) |
167 | 394 logging.debug("documentViewer (gettoc) tc: %s"%(tc)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
395 return tc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
396 |
395 | 397 def getQueryResultHits(self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionsClose=None): |
398 | |
399 """number of hits in Search mode""" | |
400 docpath = docinfo['textURLPath'] | |
401 pagesize = pageinfo['queryPageSize'] | |
402 pn = pageinfo['searchPN'] | |
403 query =pageinfo['query'] | |
404 queryType =pageinfo['queryType'] | |
405 tocSearch = 0 | |
406 tocDiv = None | |
407 | |
408 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)) | |
409 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False) | |
410 pagedom = Parse(pagexml) | |
411 numdivs = pagedom.xpath("//div[@class='queryResultHits']") | |
402 | 412 tocSearch = int(getTextFromNode(numdivs[0])) |
413 tc=int((tocSearch/10)+1) | |
414 return tc | |
395 | 415 |
416 def getQueryResultHitsText(self, docinfo=None, pageinfo=None): | |
417 """number of hits in Text of Contents mode""" | |
412 | 418 selfurl = self.absolute_url() |
419 docpath = docinfo['textURLPath'] | |
420 viewMode= pageinfo['viewMode'] | |
421 text = self.getServerData("page-fragment.xql","document=%s&mode=%s"%(docinfo['textURLPath'], 'text')) | |
422 dom =Parse(text) | |
423 pagedivs = dom.xpath("//div[@class='countTocEntries']") | |
424 logging.debug ("pagedivs=%s"%(pagedivs)) | |
425 if len(pagedivs)>0: | |
426 originalPage= (getTextFromNode(pagedivs[0])) | |
427 tc = int (originalPage) | |
428 tc1 =tc/30+1 | |
429 return tc1 | |
395 | 430 |
431 def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None): | |
432 """number of hits in Text of Figures mode""" | |
433 | |
412 | 434 selfurl = self.absolute_url() |
435 docpath = docinfo['textURLPath'] | |
436 viewMode= pageinfo['viewMode'] | |
437 text = self.getServerData("page-fragment.xql","document=%s&mode=%s"%(docinfo['textURLPath'], 'text')) | |
438 dom =Parse(text) | |
439 pagedivs = dom.xpath("//div[@class='countFigureEntries']") | |
440 logging.debug ("pagedivs=%s"%(pagedivs)) | |
441 if len(pagedivs)>0: | |
442 originalPage= (getTextFromNode(pagedivs[0])) | |
443 tc = int (originalPage) | |
444 tc1 =tc/30+1 | |
445 return tc1 | |
395 | 446 |
447 | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
448 def getToc(self, mode="text", docinfo=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
449 """loads table of contents and stores in docinfo""" |
167 | 450 logging.debug("documentViewer (gettoc) mode: %s"%(mode)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
451 if mode == "none": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
452 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
453 if 'tocSize_%s'%mode in docinfo: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
454 # cached toc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
455 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
456 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
457 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
458 # we need to set a result set size |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
459 pagesize = 1000 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
460 pn = 1 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
461 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
462 queryType = "toc" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
463 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
464 queryType = mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
465 # number of entries in toc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
466 tocSize = 0 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
467 tocDiv = None |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
468 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
469 pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn)) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
470 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
471 # post-processing downloaded xml |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
472 pagedom = Parse(pagexml) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
473 # get number of entries |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
474 numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
475 if len(numdivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
476 tocSize = int(getTextFromNode(numdivs[0])) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
477 docinfo['tocSize_%s'%mode] = tocSize |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
478 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
479 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
480 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
481 """returns single page from the table of contents""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
482 # TODO: this should use the cached TOC |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
483 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
484 queryType = "toc" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
485 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
486 queryType = mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
487 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
488 path = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
489 pagesize = pageinfo['tocPageSize'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
490 pn = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
491 url = docinfo['url'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
492 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
493 viewMode= pageinfo['viewMode'] |
171 | 494 characterNormalization = pageinfo ['characterNormalization'] |
398 | 495 optionToggle =pageinfo ['optionToggle'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
496 tocMode = pageinfo['tocMode'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
497 tocPN = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
498 |
407 | 499 data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn)) |
500 page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&optionToggle=1'%(selfurl,url, viewMode, tocMode, tocPN)) | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
501 text = page.replace('mode=image','mode=texttool') |
165 | 502 logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) |
162 | 503 #logging.debug("documentViewer (characterNormalization) text: %s"%(text)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
504 return text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
505 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
506 def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
507 """change settings""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
508 self.title=title |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
509 self.timeout = timeout |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
510 self.serverUrl = serverUrl |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
511 if RESPONSE is not None: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
512 RESPONSE.redirect('manage_main') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
513 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
514 # management methods |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
515 def manage_addMpdlXmlTextServerForm(self): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
516 """Form for adding""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
517 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
518 return pt() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
519 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
520 def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
521 """add zogiimage""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
522 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
523 self.Destination()._setObject(id, newObj) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
524 if RESPONSE is not None: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
525 RESPONSE.redirect('manage_main') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
526 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
527 |
133 | 528 |