documentViewer: MpdlXmlTextServer.py annotate

annotate MpdlXmlTextServer.py @ 409:f7d73ea4b38b

quick fix for broken texturlpath

author	casties
date	Thu, 25 Nov 2010 10:03:50 +0100
parents	1cea48640992
children	691f0c341b33

rev	line source
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	1
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	2 from OFS.SimpleItem import SimpleItem
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	3 from Products.PageTemplates.PageTemplateFile import PageTemplateFile
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	4 from Ft.Xml import EMPTY_NAMESPACE, Parse
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	5
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	6 import sys
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	7 import logging
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	8 import urllib
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	9 import documentViewer
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	10 from documentViewer import getTextFromNode, serializeNode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	11
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	12 class MpdlXmlTextServer(SimpleItem):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	13 """TextServer implementation for MPDL-XML eXist server"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	14 meta_type="MPDL-XML TextServer"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	15
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	16 manage_options=(
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	17 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	18 )+SimpleItem.manage_options
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	19
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	20 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	21
132 39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	22 def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	23 """constructor"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	24 self.id=id
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	25 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	26 self.timeout = timeout
132 39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	27 if serverName is None:
39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	28 self.serverUrl = serverUrl
39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	29 else:
39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	30 self.serverUrl = "http://%s/mpdl/interface/"%serverName
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	31
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	32 def getHttpData(self, url, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	33 """returns result from url+data HTTP request"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	34 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	35
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	36 def getServerData(self, method, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	37 """returns result from text server for method+data"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	38 url = self.serverUrl+method
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	39 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	40
398 55f9eeebc338 * empty log message * abukhman parents: 395 diff changeset	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 bc44cf97dfdc * empty log message * abukhman parents: 180 diff changeset	49 query =pageinfo['query']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	50 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	51 viewMode= pageinfo['viewMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	52 tocMode = pageinfo['tocMode']
169 21ac9164eedf characterNormalization abukhman parents: 167 diff changeset	53 characterNormalization = pageinfo['characterNormalization']
399 8abb09665ae5 * empty log message * abukhman parents: 398 diff changeset	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 8abb09665ae5 * empty log message * abukhman parents: 398 diff changeset	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 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	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 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	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 e6beb338f85f onDblclick to onClick abukhman parents: 134 diff changeset	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 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	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 e6beb338f85f onDblclick to onClick abukhman parents: 134 diff changeset	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 e6beb338f85f onDblclick to onClick abukhman parents: 134 diff changeset	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 0f6a377679d5 * empty log message * abukhman parents: 206 diff changeset	120 """def getNumPages(self, docinfo):
0f6a377679d5 * empty log message * abukhman parents: 206 diff changeset	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 c92e6b923913 * empty log message * abukhman parents: 189 diff changeset	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 0f6a377679d5 * empty log message * abukhman parents: 206 diff changeset	129 """
190 c92e6b923913 * empty log message * abukhman parents: 189 diff changeset	130 def getNumTextPages (self, docinfo):
c92e6b923913 * empty log message * abukhman parents: 189 diff changeset	131 """get list of pages from fulltext (texts without images) and put in docinfo"""
205 1fc7d73ce7ce * empty log message * abukhman parents: 204 diff changeset	132 if 'numPages' in docinfo:
190 c92e6b923913 * empty log message * abukhman parents: 189 diff changeset	133 # allredy there
205 1fc7d73ce7ce * empty log message * abukhman parents: 204 diff changeset	134 return docinfo
1fc7d73ce7ce * empty log message * abukhman parents: 204 diff changeset	135 xpath ='/count(//pb)'
1fc7d73ce7ce * empty log message * abukhman parents: 204 diff changeset	136 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
1fc7d73ce7ce * empty log message * abukhman parents: 204 diff changeset	137 dom = Parse(text)
1fc7d73ce7ce * empty log message * abukhman parents: 204 diff changeset	138 result= dom.xpath("//result/resultPage")
388 0265ab93716a * empty log message * abukhman parents: 386 diff changeset	139
205 1fc7d73ce7ce * empty log message * abukhman parents: 204 diff changeset	140 docinfo['numPages']=int(getTextFromNode(result[0]))
190 c92e6b923913 * empty log message * abukhman parents: 189 diff changeset	141 return docinfo
395 e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	142
253 d6ca167ccb1a * empty log message * abukhman parents: 252 diff changeset	143 def getGisPlaces(self, docinfo=None, pageinfo=None):
207 0f6a377679d5 * empty log message * abukhman parents: 206 diff changeset	144 """ Show all Gis Places of whole Page"""
264 ac93e115d902 * empty log message * abukhman parents: 263 diff changeset	145 xpath='//place'
409 f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	146 docpath = docinfo.get('textURLPath',None)
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	147 if not docpath:
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	148 return None
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	149
253 d6ca167ccb1a * empty log message * abukhman parents: 252 diff changeset	150 url = docinfo['url']
d6ca167ccb1a * empty log message * abukhman parents: 252 diff changeset	151 selfurl = self.absolute_url()
257 d1eb63cf3410 * empty log message * abukhman parents: 256 diff changeset	152 pn = pageinfo['current']
291 3b54df7ca88e * empty log message * abukhman parents: 290 diff changeset	153 hrefList=[]
306 ec293c61f8a9 * empty log message * abukhman parents: 305 diff changeset	154 myList= ""
264 ac93e115d902 * empty log message * abukhman parents: 263 diff changeset	155 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
ac93e115d902 * empty log message * abukhman parents: 263 diff changeset	156 dom = Parse(text)
265 4a0dd5ff6859 * empty log message * abukhman parents: 264 diff changeset	157 result = dom.xpath("//result/resultPage/place")
236 ccdce5aca47b * empty log message * abukhman parents: 233 diff changeset	158 for l in result:
250 7775d87016b5 * empty log message * abukhman parents: 249 diff changeset	159 hrefNode= l.getAttributeNodeNS(None, u"id")
272 7a52ace035e6 * empty log message * abukhman parents: 271 diff changeset	160 href= hrefNode.nodeValue
292 d97ab809de6f * empty log message * abukhman parents: 291 diff changeset	161 hrefList.append(href)
309 ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	162 myList = ",".join(hrefList)
ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	163 logging.debug("getGisPlaces :%s"%(myList))
307 ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	164 return myList
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	165
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	166 def getAllGisPlaces (self, docinfo=None, pageinfo=None):
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	167 """Show all Gis Places of whole Book """
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	168 xpath ='//echo:place'
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	169 docpath =docinfo['textURLPath']
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	170 url = docinfo['url']
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	171 selfurl =self.absolute_url()
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	172 pn =pageinfo['current']
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	173 hrefList=[]
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	174 myList=""
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	175 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	176 dom =Parse(text)
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	177 result = dom.xpath("//result/resultPage/place")
395 e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	178
307 ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	179 for l in result:
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	180 hrefNode = l.getAttributeNodeNS(None, u"id")
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	181 href= hrefNode.nodeValue
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	182 hrefList.append(href)
300 e67f50a09259 * empty log message * abukhman parents: 299 diff changeset	183 myList = ",".join(hrefList)
309 ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	184 logging.debug("getALLGisPlaces :%s"%(myList))
ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	185 return myList
ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	186
385 7a6a2e4476eb * empty log message * abukhman parents: 384 diff changeset	187
7a6a2e4476eb * empty log message * abukhman parents: 384 diff changeset	188 def getPDF (self, docinfo=None, pageinfo=None):
7a6a2e4476eb * empty log message * abukhman parents: 384 diff changeset	189 """Show and Save different Pages as PDF in Options"""
7a6a2e4476eb * empty log message * abukhman parents: 384 diff changeset	190 selfurl=self.absolute_url()
7a6a2e4476eb * empty log message * abukhman parents: 384 diff changeset	191 pn=pageinfo['current']
386 48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	192
48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	193 viewMode =pageinfo['viewMode']
385 7a6a2e4476eb * empty log message * abukhman parents: 384 diff changeset	194
386 48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	195 #text = ("page-fragment.xql","document=%s&mode=%s&pn=%s&export=%s"%(docinfo['textURLPath'], 'text', pn,'pdf'))
48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	196 #text = self.getServerData("page-fragment.xql", "document=%s&mode=%s&pn=%s&export=%s"(docinfo['textURLPath'],'text', pn,'pdf'))
48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	197 #logging.debug("text :%s"%(text))
48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	198 #dom =Parse(text)
48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	199 #logging.debug("text :%s"%(text))
388 0265ab93716a * empty log message * abukhman parents: 386 diff changeset	200 #return text
0265ab93716a * empty log message * abukhman parents: 386 diff changeset	201
360 a0fd5e95772c * empty log message * abukhman parents: 359 diff changeset	202 def getOrigPages (self, docinfo=None, pageinfo=None):
310 80ae2324562b * empty log message * abukhman parents: 309 diff changeset	203 """Show original page """
409 f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	204 docpath = docinfo.get('textURLPath',None)
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	205 if not docpath:
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	206 return None
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	207
386 48d27e100d19 * empty log message * abukhman parents: 385 diff changeset	208 logging.debug ("docinfo['textURLPath']=%s"%(docinfo['textURLPath']))
358 0b660cb2889e * empty log message * abukhman parents: 357 diff changeset	209 #url = docinfo['url']
310 80ae2324562b * empty log message * abukhman parents: 309 diff changeset	210 selfurl = self.absolute_url()
349 d86fa52ee13c * empty log message * abukhman parents: 348 diff changeset	211 pn =pageinfo['current']
325 c7e38112c23a * empty log message * abukhman parents: 324 diff changeset	212
349 d86fa52ee13c * empty log message * abukhman parents: 348 diff changeset	213 viewMode= pageinfo['viewMode']
d86fa52ee13c * empty log message * abukhman parents: 348 diff changeset	214 text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn))
310 80ae2324562b * empty log message * abukhman parents: 309 diff changeset	215 dom =Parse(text)
80ae2324562b * empty log message * abukhman parents: 309 diff changeset	216 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
361 c1381556f389 * empty log message * abukhman parents: 360 diff changeset	217 logging.debug("YYYYYYpagedivs :%s"%(pagedivs))
324 9d69b218f74c * empty log message * abukhman parents: 323 diff changeset	218 if len(pagedivs)>0:
360 a0fd5e95772c * empty log message * abukhman parents: 359 diff changeset	219 originalPage= getTextFromNode(pagedivs[0])
358 0b660cb2889e * empty log message * abukhman parents: 357 diff changeset	220 #return docinfo['originalPage']
355 929211c057eb * empty log message * abukhman parents: 354 diff changeset	221 return originalPage
360 a0fd5e95772c * empty log message * abukhman parents: 359 diff changeset	222
361 c1381556f389 * empty log message * abukhman parents: 360 diff changeset	223
406 8511e5bf4b08 * empty log message * abukhman parents: 403 diff changeset	224 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	225 """returns single page from fulltext"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	226 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	227 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	228 url = docinfo['url']
237 d596d6839fb8 * empty log message * abukhman parents: 236 diff changeset	229 name = docinfo['name']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	230 viewMode= pageinfo['viewMode']
376 4f381768909a * empty log message * abukhman parents: 375 diff changeset	231 sn = pageinfo['sn']
367 f991ddd3a045 * empty log message * abukhman parents: 364 diff changeset	232 highlightQuery = pageinfo['highlightQuery']
376 4f381768909a * empty log message * abukhman parents: 375 diff changeset	233
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	234 tocMode = pageinfo['tocMode']
164 530f6983f72d characterNormalization abukhman parents: 162 diff changeset	235 characterNormalization=pageinfo['characterNormalization']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	236 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	237 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	238 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	239 textmode = "textPollux"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	240 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	241 textmode = mode
376 4f381768909a * empty log message * abukhman parents: 375 diff changeset	242 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
373 0aa294b31236 * empty log message * abukhman parents: 372 diff changeset	243 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
370 33adbe140534 * empty log message * abukhman parents: 369 diff changeset	244 if highlightQuery is not None:
376 4f381768909a * empty log message * abukhman parents: 375 diff changeset	245 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
379 4b1b5dd7578f * empty log message * abukhman parents: 378 diff changeset	246 #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
185 6fb9a06b8bee * empty log message * abukhman parents: 184 diff changeset	247 pagexml = self.getServerData("page-fragment.xql",textParam)
379 4b1b5dd7578f * empty log message * abukhman parents: 378 diff changeset	248 logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	249 #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	250
186 3b3476df97d4 * empty log message * abukhman parents: 185 diff changeset	251 pagedom = Parse(pagexml)
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	252 # plain text mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	253 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	254 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	255 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	256 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	257 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	258 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	259 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	260 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	261 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	262 href= hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	263 if href.startswith('#note-'):
172 11cb88e7f523 characterNormalization abukhman parents: 171 diff changeset	264 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	265 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	266 if mode == "xml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	267 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	268 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	269 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	270 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	271 return serializeNode(pagenode)
140 6345583d809a Gis mode abukhman parents: 138 diff changeset	272 if mode == "gis":
6345583d809a Gis mode abukhman parents: 138 diff changeset	273 # first div contains text
6345583d809a Gis mode abukhman parents: 138 diff changeset	274 pagedivs = pagedom.xpath("/div")
6345583d809a Gis mode abukhman parents: 138 diff changeset	275 if len(pagedivs) > 0:
6345583d809a Gis mode abukhman parents: 138 diff changeset	276 pagenode = pagedivs[0]
175 d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	277 links =pagenode.xpath("//a")
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	278 for l in links:
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	279 hrefNode =l.getAttributeNodeNS(None, u"href")
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	280 if hrefNode:
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	281 href=hrefNode.nodeValue
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	282 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
239 1f3badb3f1e2 * empty log message * abukhman parents: 238 diff changeset	283 hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
211 bc6a9fdc1ff8 * empty log message * abukhman parents: 210 diff changeset	284 l.setAttributeNS(None, 'target', '_blank')
210 50a666145096 * empty log message * abukhman parents: 209 diff changeset	285 return serializeNode(pagenode)
140 6345583d809a Gis mode abukhman parents: 138 diff changeset	286
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	287 if mode == "pureXml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	288 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	289 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	290 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	291 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	292 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	293 # text-with-links mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	294 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	295 # first div contains text
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	296 pagedivs = pagedom.xpath("/div")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	297 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	298 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	299 # check all a-tags
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	300 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	301 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	302 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	303 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	304 # is link with href
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	305 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	306 if href.startswith('lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	307 # is pollux link
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	308 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	309 # change href
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	310 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	311 # add target
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	312 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	313 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
138 e6beb338f85f onDblclick to onClick abukhman parents: 134 diff changeset	314 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	315 if href.startswith('lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	316 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	317 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	318 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	319 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
138 e6beb338f85f onDblclick to onClick abukhman parents: 134 diff changeset	320 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	321 if href.startswith('#note-'):
162 e6f0553ad176 characterNormalization abukhman parents: 161 diff changeset	322 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	323 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	324 return "no text here"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	325
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	326 def getTranslate(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	327 """translate into another languages"""
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	328 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	329 #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	330 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	331
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	332 def getLemma(self, lemma=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	333 """simular words lemma """
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	334 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	335 #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	336 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	337
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	338 def getLemmaNew(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	339 """simular words lemma """
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	340 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	341 #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	342 return data
175 d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	343
398 55f9eeebc338 * empty log message * abukhman parents: 395 diff changeset	344 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	345 """number of"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	346 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	347 pagesize = pageinfo['queryPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	348 pn = pageinfo['searchPN']
181 bc44cf97dfdc * empty log message * abukhman parents: 180 diff changeset	349 query =pageinfo['query']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	350 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	351 tocSearch = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	352 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	353
179 ac715583385b urllib.quote abukhman parents: 178 diff changeset	354 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	355 #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	356 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	357 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	358 tocSearch = int(getTextFromNode(numdivs[0]))
393 bb317af1c197 * empty log message * abukhman parents: 392 diff changeset	359 logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	360 tc=int((tocSearch/10)+1)
167 7e2b97941a66 characterNormalization abukhman parents: 166 diff changeset	361 logging.debug("documentViewer (gettoc) tc: %s"%(tc))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	362 return tc
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	363
395 e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	364 def getQueryResultHits(self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionsClose=None):
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	365
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	366 """number of hits in Search mode"""
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	367 docpath = docinfo['textURLPath']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	368 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	369 pn = pageinfo['searchPN']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	370 query =pageinfo['query']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	371 queryType =pageinfo['queryType']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	372 tocSearch = 0
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	373 tocDiv = None
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	374
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	375 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))
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	376 #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)
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	377 pagedom = Parse(pagexml)
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	378 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
402 d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	379 tocSearch = int(getTextFromNode(numdivs[0]))
d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	380 tc=int((tocSearch/10)+1)
d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	381 return tc
395 e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	382
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	383 def getQueryResultHitsText(self, docinfo=None, pageinfo=None):
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	384 """number of hits in Text of Contents mode"""
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	385
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	386 docpath = docinfo['textURLPath']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	387 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	388 pn = pageinfo['searchPN']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	389 query =pageinfo['query']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	390 queryType =pageinfo['queryType']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	391 tocSearch = 0
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	392 tocDiv = None
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	393 tocMode = pageinfo['tocMode']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	394 tocPN = pageinfo['tocPN']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	395 pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'toc'))
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	396 pagedom = Parse(pagexml)
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	397 logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom))
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	398 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
402 d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	399 tocSearch = int(getTextFromNode(numdivs[0]))
403 a4ad4020ef6a * empty log message * abukhman parents: 402 diff changeset	400 tc=int((tocSearch/30)+1)
402 d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	401 return tc
395 e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	402
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	403 def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None):
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	404 """number of hits in Text of Figures mode"""
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	405
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	406 docpath = docinfo['textURLPath']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	407 pagesize = pageinfo['queryPageSize']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	408 pn = pageinfo['searchPN']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	409 query =pageinfo['query']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	410 queryType =pageinfo['queryType']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	411 tocSearch = 0
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	412 tocDiv = None
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	413 tocMode = pageinfo['tocMode']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	414 tocPN = pageinfo['tocPN']
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	415 pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'figures'))
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	416 pagedom = Parse(pagexml)
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	417 logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom))
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	418 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
402 d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	419 tocSearch = int(getTextFromNode(numdivs[0]))
403 a4ad4020ef6a * empty log message * abukhman parents: 402 diff changeset	420 tc=int((tocSearch/30)+1)
402 d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	421 return tc
395 e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	422
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	423
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	424 def getToc(self, mode="text", docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	425 """loads table of contents and stores in docinfo"""
167 7e2b97941a66 characterNormalization abukhman parents: 166 diff changeset	426 logging.debug("documentViewer (gettoc) mode: %s"%(mode))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	427 if mode == "none":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	428 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	429 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	430 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	431 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	432
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	433 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	434 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	435 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	436 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	437 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	438 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	439 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	440 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	441 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	442 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	443 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	444
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	445 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	446 #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	447 # post-processing downloaded xml
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	448 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	449 # get number of entries
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	450 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	451 if len(numdivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	452 tocSize = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	453 docinfo['tocSize_%s'%mode] = tocSize
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	454 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	455
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	456 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	457 """returns single page from the table of contents"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	458 # TODO: this should use the cached TOC
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	459 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	460 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	461 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	462 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	463 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	464 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	465 pagesize = pageinfo['tocPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	466 pn = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	467 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	468 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	469 viewMode= pageinfo['viewMode']
171 77d5ff742bb4 characterNormalization abukhman parents: 170 diff changeset	470 characterNormalization = pageinfo ['characterNormalization']
398 55f9eeebc338 * empty log message * abukhman parents: 395 diff changeset	471 optionToggle =pageinfo ['optionToggle']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	472 tocMode = pageinfo['tocMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	473 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	474
407 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	475 data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn))
1cea48640992 * empty log message * abukhman parents: 406 diff changeset	476 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	477 text = page.replace('mode=image','mode=texttool')
165 820a2a4b23c3 characterNormalization abukhman parents: 164 diff changeset	478 logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
162 e6f0553ad176 characterNormalization abukhman parents: 161 diff changeset	479 #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	480 return text
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	481
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	482 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	483 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	484 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	485 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	486 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	487 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	488 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	489
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	490 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	491 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	492 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	493 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	494 return pt()
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_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	497 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	498 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	499 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	500 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	501 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	502
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	503
133 40730cfbcb8e fixed problem with getnumpages and *args casties* parents: 132 diff changeset	504

Mercurial > hg > documentViewer

annotate MpdlXmlTextServer.py @ 409:f7d73ea4b38b