Mercurial > hg > documentViewer
annotate MpdlXmlTextServer.py @ 411:691f0c341b33
*** empty log message ***
author | abukhman |
---|---|
date | Mon, 06 Dec 2010 11:41:55 +0100 |
parents | f7d73ea4b38b |
children | 759ed7577eef |
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 |
207 | 120 """def getNumPages(self, docinfo): |
121 ""get list of pages from fulltext and put in docinfo"" | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
122 if 'numPages' in docinfo: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
123 # already there |
190 | 124 return docinfo |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
125 xquery = '//pb' |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
126 text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
127 docinfo['numPages'] = text.count("<pb ") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
128 return docinfo |
207 | 129 """ |
190 | 130 def getNumTextPages (self, docinfo): |
131 """get list of pages from fulltext (texts without images) and put in docinfo""" | |
205 | 132 if 'numPages' in docinfo: |
190 | 133 # allredy there |
205 | 134 return docinfo |
135 xpath ='/count(//pb)' | |
136 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath)) | |
137 dom = Parse(text) | |
138 result= dom.xpath("//result/resultPage") | |
388 | 139 |
205 | 140 docinfo['numPages']=int(getTextFromNode(result[0])) |
190 | 141 return docinfo |
395 | 142 |
253 | 143 def getGisPlaces(self, docinfo=None, pageinfo=None): |
207 | 144 """ Show all Gis Places of whole Page""" |
264 | 145 xpath='//place' |
409 | 146 docpath = docinfo.get('textURLPath',None) |
147 if not docpath: | |
148 return None | |
149 | |
253 | 150 url = docinfo['url'] |
151 selfurl = self.absolute_url() | |
257 | 152 pn = pageinfo['current'] |
291 | 153 hrefList=[] |
306 | 154 myList= "" |
264 | 155 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn)) |
156 dom = Parse(text) | |
265 | 157 result = dom.xpath("//result/resultPage/place") |
236 | 158 for l in result: |
250 | 159 hrefNode= l.getAttributeNodeNS(None, u"id") |
272 | 160 href= hrefNode.nodeValue |
292 | 161 hrefList.append(href) |
309 | 162 myList = ",".join(hrefList) |
163 logging.debug("getGisPlaces :%s"%(myList)) | |
307 | 164 return myList |
165 | |
166 def getAllGisPlaces (self, docinfo=None, pageinfo=None): | |
167 """Show all Gis Places of whole Book """ | |
168 xpath ='//echo:place' | |
169 docpath =docinfo['textURLPath'] | |
170 url = docinfo['url'] | |
171 selfurl =self.absolute_url() | |
172 pn =pageinfo['current'] | |
173 hrefList=[] | |
174 myList="" | |
175 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath)) | |
176 dom =Parse(text) | |
177 result = dom.xpath("//result/resultPage/place") | |
395 | 178 |
307 | 179 for l in result: |
180 hrefNode = l.getAttributeNodeNS(None, u"id") | |
181 href= hrefNode.nodeValue | |
182 hrefList.append(href) | |
300 | 183 myList = ",".join(hrefList) |
309 | 184 logging.debug("getALLGisPlaces :%s"%(myList)) |
185 return myList | |
186 | |
385 | 187 |
188 def getPDF (self, docinfo=None, pageinfo=None): | |
189 """Show and Save different Pages as PDF in Options""" | |
190 selfurl=self.absolute_url() | |
191 pn=pageinfo['current'] | |
386 | 192 |
193 viewMode =pageinfo['viewMode'] | |
385 | 194 |
386 | 195 #text = ("page-fragment.xql","document=%s&mode=%s&pn=%s&export=%s"%(docinfo['textURLPath'], 'text', pn,'pdf')) |
196 #text = self.getServerData("page-fragment.xql", "document=%s&mode=%s&pn=%s&export=%s"(docinfo['textURLPath'],'text', pn,'pdf')) | |
197 #logging.debug("text :%s"%(text)) | |
198 #dom =Parse(text) | |
199 #logging.debug("text :%s"%(text)) | |
388 | 200 #return text |
201 | |
360 | 202 def getOrigPages (self, docinfo=None, pageinfo=None): |
310 | 203 """Show original page """ |
409 | 204 docpath = docinfo.get('textURLPath',None) |
205 if not docpath: | |
206 return None | |
207 | |
386 | 208 logging.debug ("docinfo['textURLPath']=%s"%(docinfo['textURLPath'])) |
358 | 209 #url = docinfo['url'] |
310 | 210 selfurl = self.absolute_url() |
349 | 211 pn =pageinfo['current'] |
325 | 212 |
349 | 213 viewMode= pageinfo['viewMode'] |
214 text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn)) | |
310 | 215 dom =Parse(text) |
216 pagedivs = dom.xpath("//div[@class='pageNumberOrig']") | |
324 | 217 if len(pagedivs)>0: |
360 | 218 originalPage= getTextFromNode(pagedivs[0]) |
358 | 219 #return docinfo['originalPage'] |
355 | 220 return originalPage |
360 | 221 |
411 | 222 def getAllPlaces (self, docinfo=None, pageinfo=None): |
223 """Show all Places if no places than 0""" | |
224 docpath = docinfo['textURLPath'] | |
225 selfurl = self.absolute_url() | |
226 viewMode= pageinfo['viewMode'] | |
227 text = self.getServerData("page-fragment.xql","document=%s&mode=%s"%(docinfo['textURLPath'], 'text')) | |
228 dom =Parse(text) | |
229 pagedivs = dom.xpath("//div[@class='countPlaces']") | |
230 logging.debug ("pagedivs=%s"%(pagedivs)) | |
231 #originalPage= getTextFromNode(pagedivs[0]) | |
232 #return allPlaces | |
233 if len(pagedivs)>0: | |
234 originalPage= getTextFromNode(pagedivs[0]) | |
235 #return docinfo['originalPage'] | |
236 return originalPage | |
237 | |
406 | 238 def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="regPlusNorm", highlightQuery=None, sn=None, optionToggle=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
239 """returns single page from fulltext""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
240 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
241 path = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
242 url = docinfo['url'] |
237 | 243 name = docinfo['name'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
244 viewMode= pageinfo['viewMode'] |
376 | 245 sn = pageinfo['sn'] |
367 | 246 highlightQuery = pageinfo['highlightQuery'] |
376 | 247 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
248 tocMode = pageinfo['tocMode'] |
164 | 249 characterNormalization=pageinfo['characterNormalization'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
250 tocPN = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
251 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
252 if mode == "text_dict": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
253 textmode = "textPollux" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
254 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
255 textmode = mode |
376 | 256 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
373 | 257 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
370 | 258 if highlightQuery is not None: |
376 | 259 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) |
379 | 260 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
185 | 261 pagexml = self.getServerData("page-fragment.xql",textParam) |
379 | 262 logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
263 #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
|
264 |
186 | 265 pagedom = Parse(pagexml) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
266 # plain text mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
267 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
268 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
269 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
270 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
271 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
272 links = pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
273 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
274 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
275 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
276 href= hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
277 if href.startswith('#note-'): |
172 | 278 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
|
279 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
280 if mode == "xml": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
281 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
282 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
283 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
284 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
285 return serializeNode(pagenode) |
140 | 286 if mode == "gis": |
287 # first div contains text | |
288 pagedivs = pagedom.xpath("/div") | |
289 if len(pagedivs) > 0: | |
290 pagenode = pagedivs[0] | |
175 | 291 links =pagenode.xpath("//a") |
292 for l in links: | |
293 hrefNode =l.getAttributeNodeNS(None, u"href") | |
294 if hrefNode: | |
295 href=hrefNode.nodeValue | |
296 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): | |
239 | 297 hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name) |
211 | 298 l.setAttributeNS(None, 'target', '_blank') |
210 | 299 return serializeNode(pagenode) |
140 | 300 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
301 if mode == "pureXml": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
302 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
303 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
304 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
305 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
306 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
307 # text-with-links mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
308 if mode == "text_dict": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
309 # first div contains text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
310 pagedivs = pagedom.xpath("/div") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
311 if len(pagedivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
312 pagenode = pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
313 # check all a-tags |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
314 links = pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
315 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
316 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
317 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
318 # is link with href |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
319 href = hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
320 if href.startswith('lt/lex.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
321 # is pollux link |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
322 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
323 # change href |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
324 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
|
325 # add target |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
326 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
327 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
138 | 328 l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
329 if href.startswith('lt/lemma.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
330 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
331 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
|
332 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
333 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
138 | 334 l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
335 if href.startswith('#note-'): |
162 | 336 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
|
337 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
338 return "no text here" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
339 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
340 def getTranslate(self, query=None, language=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
341 """translate into another languages""" |
134 | 342 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
|
343 #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
|
344 return data |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
345 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
346 def getLemma(self, lemma=None, language=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
347 """simular words lemma """ |
134 | 348 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
|
349 #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
|
350 return data |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
351 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
352 def getLemmaNew(self, query=None, language=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
353 """simular words lemma """ |
134 | 354 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
|
355 #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
|
356 return data |
175 | 357 |
398 | 358 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
|
359 """number of""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
360 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
361 pagesize = pageinfo['queryPageSize'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
362 pn = pageinfo['searchPN'] |
181 | 363 query =pageinfo['query'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
364 queryType =pageinfo['queryType'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
365 tocSearch = 0 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
366 tocDiv = None |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
367 |
179 | 368 pagexml = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
369 #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
370 pagedom = Parse(pagexml) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
371 numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
372 tocSearch = int(getTextFromNode(numdivs[0])) |
393 | 373 logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
374 tc=int((tocSearch/10)+1) |
167 | 375 logging.debug("documentViewer (gettoc) tc: %s"%(tc)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
376 return tc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
377 |
395 | 378 def getQueryResultHits(self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionsClose=None): |
379 | |
380 """number of hits in Search mode""" | |
381 docpath = docinfo['textURLPath'] | |
382 pagesize = pageinfo['queryPageSize'] | |
383 pn = pageinfo['searchPN'] | |
384 query =pageinfo['query'] | |
385 queryType =pageinfo['queryType'] | |
386 tocSearch = 0 | |
387 tocDiv = None | |
388 | |
389 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)) | |
390 #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) | |
391 pagedom = Parse(pagexml) | |
392 numdivs = pagedom.xpath("//div[@class='queryResultHits']") | |
402 | 393 tocSearch = int(getTextFromNode(numdivs[0])) |
394 tc=int((tocSearch/10)+1) | |
395 return tc | |
395 | 396 |
397 def getQueryResultHitsText(self, docinfo=None, pageinfo=None): | |
398 """number of hits in Text of Contents mode""" | |
399 | |
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 tocMode = pageinfo['tocMode'] | |
408 tocPN = pageinfo['tocPN'] | |
409 pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'toc')) | |
410 pagedom = Parse(pagexml) | |
411 logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom)) | |
412 numdivs = pagedom.xpath("//div[@class='queryResultHits']") | |
402 | 413 tocSearch = int(getTextFromNode(numdivs[0])) |
403 | 414 tc=int((tocSearch/30)+1) |
402 | 415 return tc |
395 | 416 |
417 def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None): | |
418 """number of hits in Text of Figures mode""" | |
419 | |
420 docpath = docinfo['textURLPath'] | |
421 pagesize = pageinfo['queryPageSize'] | |
422 pn = pageinfo['searchPN'] | |
423 query =pageinfo['query'] | |
424 queryType =pageinfo['queryType'] | |
425 tocSearch = 0 | |
426 tocDiv = None | |
427 tocMode = pageinfo['tocMode'] | |
428 tocPN = pageinfo['tocPN'] | |
429 pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'figures')) | |
430 pagedom = Parse(pagexml) | |
431 logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom)) | |
432 numdivs = pagedom.xpath("//div[@class='queryResultHits']") | |
402 | 433 tocSearch = int(getTextFromNode(numdivs[0])) |
403 | 434 tc=int((tocSearch/30)+1) |
402 | 435 return tc |
395 | 436 |
437 | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
438 def getToc(self, mode="text", docinfo=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
439 """loads table of contents and stores in docinfo""" |
167 | 440 logging.debug("documentViewer (gettoc) mode: %s"%(mode)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
441 if mode == "none": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
442 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
443 if 'tocSize_%s'%mode in docinfo: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
444 # cached toc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
445 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
446 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
447 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
448 # we need to set a result set size |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
449 pagesize = 1000 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
450 pn = 1 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
451 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
452 queryType = "toc" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
453 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
454 queryType = mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
455 # number of entries in toc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
456 tocSize = 0 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
457 tocDiv = None |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
458 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
459 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
|
460 #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
|
461 # post-processing downloaded xml |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
462 pagedom = Parse(pagexml) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
463 # get number of entries |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
464 numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
465 if len(numdivs) > 0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
466 tocSize = int(getTextFromNode(numdivs[0])) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
467 docinfo['tocSize_%s'%mode] = tocSize |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
468 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
469 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
470 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
471 """returns single page from the table of contents""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
472 # TODO: this should use the cached TOC |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
473 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
474 queryType = "toc" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
475 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
476 queryType = mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
477 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
478 path = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
479 pagesize = pageinfo['tocPageSize'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
480 pn = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
481 url = docinfo['url'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
482 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
483 viewMode= pageinfo['viewMode'] |
171 | 484 characterNormalization = pageinfo ['characterNormalization'] |
398 | 485 optionToggle =pageinfo ['optionToggle'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
486 tocMode = pageinfo['tocMode'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
487 tocPN = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
488 |
407 | 489 data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn)) |
490 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
|
491 text = page.replace('mode=image','mode=texttool') |
165 | 492 logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) |
162 | 493 #logging.debug("documentViewer (characterNormalization) text: %s"%(text)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
494 return text |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
495 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
496 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
|
497 """change settings""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
498 self.title=title |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
499 self.timeout = timeout |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
500 self.serverUrl = serverUrl |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
501 if RESPONSE is not None: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
502 RESPONSE.redirect('manage_main') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
503 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
504 # management methods |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
505 def manage_addMpdlXmlTextServerForm(self): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
506 """Form for adding""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
507 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
508 return pt() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
509 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
510 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
|
511 """add zogiimage""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
512 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
513 self.Destination()._setObject(id, newObj) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
514 if RESPONSE is not None: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
515 RESPONSE.redirect('manage_main') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
516 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
517 |
133 | 518 |