documentViewer: MpdlXmlTextServer.py annotate

annotate MpdlXmlTextServer.py @ 435:94baa8dbc7e8

*** empty log message ***

author	abukhman
date	Tue, 01 Mar 2011 15:53:15 +0100
parents	44ccb7bd0938
children	0baf4e761be7

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
427 70db19d7b3b1 * empty log message * abukhman parents: 426 diff changeset	6 import md5
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	7 import sys
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	8 import logging
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	9 import urllib
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	10 import documentViewer
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	11 from documentViewer import getTextFromNode, serializeNode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	12
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	13 class MpdlXmlTextServer(SimpleItem):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	14 """TextServer implementation for MPDL-XML eXist server"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	15 meta_type="MPDL-XML TextServer"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	16
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	17 manage_options=(
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	18 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	19 )+SimpleItem.manage_options
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	20
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	21 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	22
132 39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	23 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	24 """constructor"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	25 self.id=id
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	26 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	27 self.timeout = timeout
132 39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	28 if serverName is None:
39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	29 self.serverUrl = serverUrl
39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	30 else:
39167e96e582 fixed creation of text server on __init__ casties parents: 129 diff changeset	31 self.serverUrl = "http://%s/mpdl/interface/"%serverName
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	32
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	33 def getHttpData(self, url, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	34 """returns result from url+data HTTP request"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	35 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	36
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	37 def getServerData(self, method, data=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	38 """returns result from text server for method+data"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	39 url = self.serverUrl+method
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	40 return documentViewer.getHttpData(url,data,timeout=self.timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	41
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	42 def getSearch(self, pageinfo=None, docinfo=None, lemma=None):
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	43 """get search list"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	44 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	45 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	46 pagesize = pageinfo['queryPageSize']
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	47 pn = pageinfo.get('searchPN',1)
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	48 sn = pageinfo['sn']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	49 highlightQuery = pageinfo['highlightQuery']
181 bc44cf97dfdc * empty log message * abukhman parents: 180 diff changeset	50 query =pageinfo['query']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	51 queryType =pageinfo['queryType']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	52 viewMode= pageinfo['viewMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	53 tocMode = pageinfo['tocMode']
169 21ac9164eedf characterNormalization abukhman parents: 167 diff changeset	54 characterNormalization = pageinfo['characterNormalization']
399 8abb09665ae5 * empty log message * abukhman parents: 398 diff changeset	55 optionToggle = pageinfo['optionToggle']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	56 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	57 selfurl = self.absolute_url()
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 pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	60 pagedom = Parse(pagexml)
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	61
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	62 """
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	63 pagedivs = pagedom.xpath("//div[@class='queryResultHits']")
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	64 if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")):
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	65 if len(pagedivs)>0:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	66 docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0]))
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	67 s = getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	68 s1 = int(s)/10+1
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	69 try:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	70 docinfo['queryResultHits'] = int(s1)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	71 logging.debug("SEARCH ENTRIES: %s"%(s1))
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	72 except:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	73 docinfo['queryResultHits'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	74 """
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	75 if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	76 pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	77 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	78 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	79 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	80 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	81 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	82 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	83 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	84 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	85 selfurl = self.absolute_url()
407 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	86 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	87 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
428 672f8857ece8 bibliograpfical info abukhman parents: 427 diff changeset	88 logging.debug("PUREXML :%s"%(serializeNode(pagenode)))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	89 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	90 if (queryType=="fulltextMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	91 pagedivs = pagedom.xpath("//div[@class='queryResult']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	92 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	93 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	94 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	95 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	96 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	97 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	98 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	99 if href.startswith('page-fragment.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	100 selfurl = self.absolute_url()
407 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	101 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	102 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	103 if href.startswith('../lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	104 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	105 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	106 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	107 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	108 pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	109 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	110 if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	111 pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	112 if len(pagedivs)>0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	113 pagenode=pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	114 links=pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	115 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	116 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	117 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	118 href = hrefNode.nodeValue
407 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	119 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	120 if href.startswith('../lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	121 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	122 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	123 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	124 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	125 if href.startswith('../lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	126 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	127 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	128 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	129 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	130 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	131 return "no text here"
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	132
253 d6ca167ccb1a * empty log message * abukhman parents: 252 diff changeset	133 def getGisPlaces(self, docinfo=None, pageinfo=None):
207 0f6a377679d5 * empty log message * abukhman parents: 206 diff changeset	134 """ Show all Gis Places of whole Page"""
264 ac93e115d902 * empty log message * abukhman parents: 263 diff changeset	135 xpath='//place'
409 f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	136 docpath = docinfo.get('textURLPath',None)
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	137 if not docpath:
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	138 return None
f7d73ea4b38b quick fix for broken texturlpath casties parents: 407 diff changeset	139
253 d6ca167ccb1a * empty log message * abukhman parents: 252 diff changeset	140 url = docinfo['url']
d6ca167ccb1a * empty log message * abukhman parents: 252 diff changeset	141 selfurl = self.absolute_url()
257 d1eb63cf3410 * empty log message * abukhman parents: 256 diff changeset	142 pn = pageinfo['current']
291 3b54df7ca88e * empty log message * abukhman parents: 290 diff changeset	143 hrefList=[]
306 ec293c61f8a9 * empty log message * abukhman parents: 305 diff changeset	144 myList= ""
264 ac93e115d902 * empty log message * abukhman parents: 263 diff changeset	145 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
ac93e115d902 * empty log message * abukhman parents: 263 diff changeset	146 dom = Parse(text)
265 4a0dd5ff6859 * empty log message * abukhman parents: 264 diff changeset	147 result = dom.xpath("//result/resultPage/place")
236 ccdce5aca47b * empty log message * abukhman parents: 233 diff changeset	148 for l in result:
250 7775d87016b5 * empty log message * abukhman parents: 249 diff changeset	149 hrefNode= l.getAttributeNodeNS(None, u"id")
272 7a52ace035e6 * empty log message * abukhman parents: 271 diff changeset	150 href= hrefNode.nodeValue
292 d97ab809de6f * empty log message * abukhman parents: 291 diff changeset	151 hrefList.append(href)
309 ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	152 myList = ",".join(hrefList)
ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	153 logging.debug("getGisPlaces :%s"%(myList))
307 ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	154 return myList
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	155
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	156 def getAllGisPlaces (self, docinfo=None, pageinfo=None):
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	157 """Show all Gis Places of whole Book """
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	158 xpath ='//echo:place'
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	159 docpath =docinfo['textURLPath']
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	160 url = docinfo['url']
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	161 selfurl =self.absolute_url()
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	162 pn =pageinfo['current']
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	163 hrefList=[]
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	164 myList=""
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	165 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	166 dom =Parse(text)
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	167 result = dom.xpath("//result/resultPage/place")
395 e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	168
307 ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	169 for l in result:
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	170 hrefNode = l.getAttributeNodeNS(None, u"id")
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	171 href= hrefNode.nodeValue
ec5e920a61e6 * empty log message * abukhman parents: 306 diff changeset	172 hrefList.append(href)
300 e67f50a09259 * empty log message * abukhman parents: 299 diff changeset	173 myList = ",".join(hrefList)
309 ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	174 logging.debug("getALLGisPlaces :%s"%(myList))
ea18683b2bff * empty log message * abukhman parents: 308 diff changeset	175 return myList
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	176
411 691f0c341b33 * empty log message * abukhman parents: 409 diff changeset	177
432 8d8f2cbd6213 * empty log message * abukhman parents: 431 diff changeset	178 def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None):
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	179 """returns single page from fulltext"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	180 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	181 path = docinfo['textURLPath']
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	182 url = docinfo.get('url',None)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	183 name = docinfo.get('name',None)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	184 pn =pageinfo['current']
376 4f381768909a * empty log message * abukhman parents: 375 diff changeset	185 sn = pageinfo['sn']
367 f991ddd3a045 * empty log message * abukhman parents: 364 diff changeset	186 highlightQuery = pageinfo['highlightQuery']
428 672f8857ece8 bibliograpfical info abukhman parents: 427 diff changeset	187 #mode = pageinfo ['viewMode']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	188 tocMode = pageinfo['tocMode']
164 530f6983f72d characterNormalization abukhman parents: 162 diff changeset	189 characterNormalization=pageinfo['characterNormalization']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	190 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	191 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	192 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	193 textmode = "textPollux"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	194 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	195 textmode = mode
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	196
373 0aa294b31236 * empty log message * abukhman parents: 372 diff changeset	197 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
370 33adbe140534 * empty log message * abukhman parents: 369 diff changeset	198 if highlightQuery is not None:
376 4f381768909a * empty log message * abukhman parents: 375 diff changeset	199 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	200
185 6fb9a06b8bee * empty log message * abukhman parents: 184 diff changeset	201 pagexml = self.getServerData("page-fragment.xql",textParam)
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	202 dom = Parse(pagexml)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	203 #original Pages
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	204 pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	205 if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	206 if len(pagedivs)>0:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	207 docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
431 630a4581992a * empty log message * abukhman parents: 428 diff changeset	208
630a4581992a * empty log message * abukhman parents: 428 diff changeset	209 #original Pages Norm
630a4581992a * empty log message * abukhman parents: 428 diff changeset	210 pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
630a4581992a * empty log message * abukhman parents: 428 diff changeset	211 if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
630a4581992a * empty log message * abukhman parents: 428 diff changeset	212 if len(pagedivs)>0:
630a4581992a * empty log message * abukhman parents: 428 diff changeset	213 docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
630a4581992a * empty log message * abukhman parents: 428 diff changeset	214
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	215 #figureEntries
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	216 pagedivs = dom.xpath("//div[@class='countFigureEntries']")
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	217 if pagedivs == dom.xpath("//div[@class='countFigureEntries']"):
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	218 if len(pagedivs)>0:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	219 docinfo['countFigureEntries'] = getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	220 s = getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	221 if s=='0':
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	222 try:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	223 docinfo['countFigureEntries'] = int(s)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	224 except:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	225 docinfo['countFigureEntries'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	226 else:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	227 s1 = int(s)/30+1
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	228 try:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	229 docinfo['countFigureEntries'] = int(s1)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	230 except:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	231 docinfo['countFigureEntries'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	232
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	233 #allPlaces
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	234 pagedivs = dom.xpath("//div[@class='countPlaces']")
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	235 if pagedivs == dom.xpath("//div[@class='countPlaces']"):
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	236 if len(pagedivs)>0:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	237 docinfo['countPlaces']= getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	238 s = getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	239 try:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	240 docinfo['countPlaces'] = int(s)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	241 except:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	242 docinfo['countPlaces'] = 0
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	243
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	244 #tocEntries
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	245 pagedivs = dom.xpath("//div[@class='countTocEntries']")
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	246 if pagedivs == dom.xpath("//div[@class='countTocEntries']"):
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	247 if len(pagedivs)>0:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	248 docinfo['countTocEntries'] = int(getTextFromNode(pagedivs[0]))
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	249 s = getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	250 if s=='0':
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	251 try:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	252 docinfo['countTocEntries'] = int(s)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	253 except:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	254 docinfo['countTocEntries'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	255 else:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	256 s1 = int(s)/30+1
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	257 try:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	258 docinfo['countTocEntries'] = int(s1)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	259 except:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	260 docinfo['countTocEntries'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	261
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	262 #numTextPages
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	263 pagedivs = dom.xpath("//div[@class='countPages']")
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	264 if pagedivs == dom.xpath("//div[@class='countPages']"):
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	265 if len(pagedivs)>0:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	266 docinfo['numPages'] = getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	267 s = getTextFromNode(pagedivs[0])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	268
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	269 try:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	270 docinfo['numPages'] = int(s)
434 44ccb7bd0938 * empty log message * abukhman parents: 432 diff changeset	271 #logging.debug("PAGE NUMBER: %s"%(s))
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	272
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	273 np = docinfo['numPages']
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	274 pageinfo['end'] = min(pageinfo['end'], np)
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	275 pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	276 if np % pageinfo['groupsize'] > 0:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	277 pageinfo['numgroups'] += 1
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	278 except:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	279 docinfo['numPages'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	280
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	281 else:
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	282 #no full text -- init to 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	283 docinfo['pageNumberOrig'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	284 docinfo['countFigureEntries'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	285 docinfo['countPlaces'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	286 docinfo['countTocEntries'] = 0
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	287 docinfo['numPages'] = 0
431 630a4581992a * empty log message * abukhman parents: 428 diff changeset	288 docinfo['pageNumberOrigNorm'] = 0
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	289 #return docinfo
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	290
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	291 # plain text mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	292 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	293 # first div contains text
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	294 pagedivs = dom.xpath("/div")
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	295 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	296 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	297 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	298 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	299 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	300 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	301 href= hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	302 if href.startswith('#note-'):
432 8d8f2cbd6213 * empty log message * abukhman parents: 431 diff changeset	303 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	304 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	305 if mode == "xml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	306 # first div contains text
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	307 pagedivs = dom.xpath("/div")
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	308 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	309 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	310 return serializeNode(pagenode)
140 6345583d809a Gis mode abukhman parents: 138 diff changeset	311 if mode == "gis":
6345583d809a Gis mode abukhman parents: 138 diff changeset	312 # first div contains text
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	313 pagedivs = dom.xpath("/div")
140 6345583d809a Gis mode abukhman parents: 138 diff changeset	314 if len(pagedivs) > 0:
6345583d809a Gis mode abukhman parents: 138 diff changeset	315 pagenode = pagedivs[0]
175 d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	316 links =pagenode.xpath("//a")
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	317 for l in links:
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	318 hrefNode =l.getAttributeNodeNS(None, u"href")
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	319 if hrefNode:
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	320 href=hrefNode.nodeValue
d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	321 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
239 1f3badb3f1e2 * empty log message * abukhman parents: 238 diff changeset	322 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	323 l.setAttributeNS(None, 'target', '_blank')
210 50a666145096 * empty log message * abukhman parents: 209 diff changeset	324 return serializeNode(pagenode)
140 6345583d809a Gis mode abukhman parents: 138 diff changeset	325
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	326 if mode == "pureXml":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	327 # first div contains text
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	328 pagedivs = dom.xpath("/div")
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	329 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	330 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	331 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	332 # text-with-links mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	333 if mode == "text_dict":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	334 # first div contains text
432 8d8f2cbd6213 * empty log message * abukhman parents: 431 diff changeset	335 #mode = pageinfo ['viewMode']
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	336 pagedivs = dom.xpath("/div")
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	337 if len(pagedivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	338 pagenode = pagedivs[0]
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	339 # check all a-tags
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	340 links = pagenode.xpath("//a")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	341 for l in links:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	342 hrefNode = l.getAttributeNodeNS(None, u"href")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	343 if hrefNode:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	344 # is link with href
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	345 href = hrefNode.nodeValue
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	346 if href.startswith('lt/lex.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	347 # is pollux link
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	348 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	349 # change href
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	350 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	351 # add target
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	352 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	353 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	354 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	355 if href.startswith('lt/lemma.xql'):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	356 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	357 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	358 l.setAttributeNS(None, 'target', '_blank')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	359 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	360 l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	361 if href.startswith('#note-'):
432 8d8f2cbd6213 * empty log message * abukhman parents: 431 diff changeset	362 hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn))
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	363 return serializeNode(pagenode)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	364 return "no text here"
428 672f8857ece8 bibliograpfical info abukhman parents: 427 diff changeset	365
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	366 def getTranslate(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	367 """translate into another languages"""
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	368 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	369 #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	370 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	371
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	372 def getLemma(self, lemma=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	373 """simular words lemma """
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	374 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	375 #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	376 return data
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	377
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	378 def getLemmaNew(self, query=None, language=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	379 """simular words lemma """
134 6a33aa624ba4 fixed more oopsies casties parents: 133 diff changeset	380 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	381 #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	382 return data
175 d1c13a6ba20d gis ?nderung abukhman parents: 172 diff changeset	383
398 55f9eeebc338 * empty log message * abukhman parents: 395 diff changeset	384 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None):
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	385 #number of
395 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
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	394 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	395 pagedom = Parse(pagexml)
e6f3fcb2058f * empty log message * abukhman parents: 393 diff changeset	396 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	397 tocSearch = int(getTextFromNode(numdivs[0]))
69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	398 tc=int((tocSearch/10)+1)
402 d3bd3c2674ee * empty log message * abukhman parents: 401 diff changeset	399 return tc
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	400
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	401 def getToc(self, mode="text", docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	402 """loads table of contents and stores in docinfo"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	403 if mode == "none":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	404 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	405 if 'tocSize_%s'%mode in docinfo:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	406 # cached toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	407 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	408
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	409 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	410 # we need to set a result set size
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	411 pagesize = 1000
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	412 pn = 1
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	413 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	414 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	415 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	416 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	417 # number of entries in toc
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	418 tocSize = 0
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	419 tocDiv = None
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	420
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	421 pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
419 69205c9d9404 * empty log message * abukhman parents: 418 diff changeset	422
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	423 # post-processing downloaded xml
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	424 pagedom = Parse(pagexml)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	425 # get number of entries
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	426 numdivs = pagedom.xpath("//div[@class='queryResultHits']")
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	427 if len(numdivs) > 0:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	428 tocSize = int(getTextFromNode(numdivs[0]))
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	429 docinfo['tocSize_%s'%mode] = tocSize
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	430 return docinfo
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	431
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	432 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	433 """returns single page from the table of contents"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	434 # TODO: this should use the cached TOC
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	435 if mode == "text":
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	436 queryType = "toc"
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	437 else:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	438 queryType = mode
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	439 docpath = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	440 path = docinfo['textURLPath']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	441 pagesize = pageinfo['tocPageSize']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	442 pn = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	443 url = docinfo['url']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	444 selfurl = self.absolute_url()
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	445 viewMode= pageinfo['viewMode']
171 77d5ff742bb4 characterNormalization abukhman parents: 170 diff changeset	446 characterNormalization = pageinfo ['characterNormalization']
398 55f9eeebc338 * empty log message * abukhman parents: 395 diff changeset	447 optionToggle =pageinfo ['optionToggle']
129 9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	448 tocMode = pageinfo['tocMode']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	449 tocPN = pageinfo['tocPN']
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	450
407 1cea48640992 * empty log message * abukhman parents: 406 diff changeset	451 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	452 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	453 text = page.replace('mode=image','mode=texttool')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	454 return text
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 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	457 """change settings"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	458 self.title=title
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	459 self.timeout = timeout
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	460 self.serverUrl = serverUrl
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	461 if RESPONSE is not None:
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	462 RESPONSE.redirect('manage_main')
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	463
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	464 # management methods
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	465 def manage_addMpdlXmlTextServerForm(self):
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	466 """Form for adding"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	467 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	468 return pt()
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 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	471 """add zogiimage"""
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	472 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	473 self.Destination()._setObject(id, newObj)
9404b6c37920 more modular version with separate object MpdlXmlTextServer casties parents: diff changeset	474 if RESPONSE is not None:
427 70db19d7b3b1 * empty log message * abukhman parents: 426 diff changeset	475 RESPONSE.redirect('manage_main')

Mercurial > hg > documentViewer

annotate MpdlXmlTextServer.py @ 435:94baa8dbc7e8