version 1.220, 2010/12/23 17:05:01
|
version 1.230, 2011/03/07 14:08:07
|
Line 3 from OFS.SimpleItem import SimpleItem
|
Line 3 from OFS.SimpleItem import SimpleItem
|
from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
from Ft.Xml import EMPTY_NAMESPACE, Parse |
from Ft.Xml import EMPTY_NAMESPACE, Parse |
|
|
|
import md5 |
import sys |
import sys |
import logging |
import logging |
import urllib |
import urllib |
Line 38 class MpdlXmlTextServer(SimpleItem):
|
Line 39 class MpdlXmlTextServer(SimpleItem):
|
url = self.serverUrl+method |
url = self.serverUrl+method |
return documentViewer.getHttpData(url,data,timeout=self.timeout) |
return documentViewer.getHttpData(url,data,timeout=self.timeout) |
|
|
def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None, characterNormalization=None, optionToggle=None): |
def getSearch(self, pageinfo=None, docinfo=None, lemma=None): |
"""get search list""" |
"""get search list""" |
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
url = docinfo['url'] |
url = docinfo['url'] |
pagesize = pageinfo['queryPageSize'] |
pagesize = pageinfo['queryPageSize'] |
pn = pageinfo['searchPN'] |
pn = pageinfo.get('searchPN',1) |
sn = pageinfo['sn'] |
sn = pageinfo['sn'] |
highlightQuery = pageinfo['highlightQuery'] |
highlightQuery = pageinfo['highlightQuery'] |
query =pageinfo['query'] |
query =pageinfo['query'] |
Line 54 class MpdlXmlTextServer(SimpleItem):
|
Line 55 class MpdlXmlTextServer(SimpleItem):
|
optionToggle = pageinfo['optionToggle'] |
optionToggle = pageinfo['optionToggle'] |
tocPN = pageinfo['tocPN'] |
tocPN = pageinfo['tocPN'] |
selfurl = self.absolute_url() |
selfurl = self.absolute_url() |
|
|
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))) |
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))) |
#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) |
|
|
|
pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url) |
pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url) |
pagedom = Parse(pagexml) |
pagedom = Parse(pagexml) |
|
|
|
""" |
|
pagedivs = pagedom.xpath("//div[@class='queryResultHits']") |
|
if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")): |
|
if len(pagedivs)>0: |
|
docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0])) |
|
s = getTextFromNode(pagedivs[0]) |
|
s1 = int(s)/10+1 |
|
try: |
|
docinfo['queryResultHits'] = int(s1) |
|
logging.debug("SEARCH ENTRIES: %s"%(s1)) |
|
except: |
|
docinfo['queryResultHits'] = 0 |
|
""" |
if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): |
if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): |
pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
if len(pagedivs)>0: |
if len(pagedivs)>0: |
Line 73 class MpdlXmlTextServer(SimpleItem):
|
Line 85 class MpdlXmlTextServer(SimpleItem):
|
selfurl = self.absolute_url() |
selfurl = self.absolute_url() |
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)) |
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)) |
hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
|
#logging.debug("PUREXML :%s"%(serializeNode(pagenode))) |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
if (queryType=="fulltextMorph"): |
if (queryType=="fulltextMorph"): |
pagedivs = pagedom.xpath("//div[@class='queryResult']") |
pagedivs = pagedom.xpath("//div[@class='queryResult']") |
Line 117 class MpdlXmlTextServer(SimpleItem):
|
Line 130 class MpdlXmlTextServer(SimpleItem):
|
return serializeNode(pagenode) |
return serializeNode(pagenode) |
return "no text here" |
return "no text here" |
|
|
|
|
def getNumTextPages (self, docinfo=None): |
|
"""get list of pages from fulltext (texts without images) and put in docinfo""" |
|
logging.debug("getNumTextPages") |
|
docpath = docinfo['textURLPath'] |
|
selfurl = self.absolute_url() |
|
#viewMode= pageinfo['viewMode'] |
|
if 'numPages' in docinfo: |
|
# allredy there |
|
return docinfo |
|
|
|
text = self.getServerData("page-fragment.xql","document=%s"%(docinfo['textURLPath'])) |
|
dom =Parse(text) |
|
pagedivs = dom.xpath("//div[@class='countPages']") |
|
logging.debug ("pagedivs=%s"%(pagedivs)) |
|
if len(pagedivs)>0: |
|
docinfo['numPages']= int(getTextFromNode(pagedivs[0])) |
|
return docinfo |
|
|
|
def getTocEntries (self, docinfo=None): |
|
""" number of text entries""" |
|
self.getInfoFromPage(docinfo) |
|
return docinfo['tocEntries'] |
|
|
|
def getFigureEntries (self, docinfo=None): |
|
""" number of figure entries""" |
|
self.getInfoFromPage(docinfo) |
|
return docinfo['figureEntries'] |
|
|
|
def getGisPlaces(self, docinfo=None, pageinfo=None): |
def getGisPlaces(self, docinfo=None, pageinfo=None): |
""" Show all Gis Places of whole Page""" |
""" Show all Gis Places of whole Page""" |
xpath='//place' |
xpath='//place' |
Line 166 class MpdlXmlTextServer(SimpleItem):
|
Line 150 class MpdlXmlTextServer(SimpleItem):
|
href= hrefNode.nodeValue |
href= hrefNode.nodeValue |
hrefList.append(href) |
hrefList.append(href) |
myList = ",".join(hrefList) |
myList = ",".join(hrefList) |
logging.debug("getGisPlaces :%s"%(myList)) |
#logging.debug("getGisPlaces :%s"%(myList)) |
return myList |
return myList |
|
|
def getAllGisPlaces (self, docinfo=None, pageinfo=None): |
def getAllGisPlaces (self, docinfo=None, pageinfo=None): |
Line 187 class MpdlXmlTextServer(SimpleItem):
|
Line 171 class MpdlXmlTextServer(SimpleItem):
|
href= hrefNode.nodeValue |
href= hrefNode.nodeValue |
hrefList.append(href) |
hrefList.append(href) |
myList = ",".join(hrefList) |
myList = ",".join(hrefList) |
logging.debug("getALLGisPlaces :%s"%(myList)) |
#logging.debug("getALLGisPlaces :%s"%(myList)) |
return myList |
return myList |
|
|
def getOrigPages (self, docinfo=None, pageinfo=None): |
|
"""Show original page """ |
def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None): |
docpath = docinfo.get('textURLPath',None) |
"""returns single page from fulltext""" |
if not docpath: |
docpath = docinfo['textURLPath'] |
return None |
path = docinfo['textURLPath'] |
selfurl = self.absolute_url() |
url = docinfo.get('url',None) |
|
name = docinfo.get('name',None) |
pn =pageinfo['current'] |
pn =pageinfo['current'] |
|
sn = pageinfo['sn'] |
|
highlightQuery = pageinfo['highlightQuery'] |
|
#mode = pageinfo ['viewMode'] |
|
tocMode = pageinfo['tocMode'] |
|
characterNormalization=pageinfo['characterNormalization'] |
|
tocPN = pageinfo['tocPN'] |
|
selfurl = self.absolute_url() |
|
if mode == "text_dict": |
|
textmode = "textPollux" |
|
else: |
|
textmode = mode |
|
|
viewMode= pageinfo['viewMode'] |
textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn)) |
if highlightQuery is not None: |
dom =Parse(text) |
textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) |
|
|
|
pagexml = self.getServerData("page-fragment.xql",textParam) |
|
dom = Parse(pagexml) |
|
#original Pages |
pagedivs = dom.xpath("//div[@class='pageNumberOrig']") |
pagedivs = dom.xpath("//div[@class='pageNumberOrig']") |
|
"""if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"): |
if len(pagedivs)>0: |
if len(pagedivs)>0: |
originalPage= getTextFromNode(pagedivs[0]) |
docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0]) |
#return docinfo['originalPage'] |
logging.debug("ORIGINAL PAGE: %s"%(docinfo['pageNumberOrig'])) |
return originalPage |
|
|
|
def getAllPlaces (self, docinfo=None): |
|
"""Show all Places if no places than 0""" |
|
self.getInfoFromPage(docinfo) |
|
return docinfo['allPlaces'] |
|
|
|
def getInfoFromPage(self, docinfo=None): |
#original Pages Norm |
""" extract diverse info from page-fragment""" |
pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']") |
docpath = docinfo['textURLPath'] |
if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"): |
if 'allPlaces' in docinfo: |
if len(pagedivs)>0: |
# allredy there |
docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0]) |
return docinfo |
logging.debug("ORIGINAL PAGE NORM: %s"%(docinfo['pageNumberOrigNorm'])) |
|
""" |
if (docpath is not None): |
|
text = self.getServerData("page-fragment.xql","document=%s"%(docinfo['textURLPath'])) |
|
dom = Parse(text) |
|
# figureEntries |
# figureEntries |
pagedivs = dom.xpath("//div[@class='countFigureEntries']") |
pagedivs = dom.xpath("//div[@class='countFigureEntries']") |
|
if pagedivs == dom.xpath("//div[@class='countFigureEntries']"): |
|
if len(pagedivs)>0: |
|
docinfo['countFigureEntries'] = getTextFromNode(pagedivs[0]) |
s = getTextFromNode(pagedivs[0]) |
s = getTextFromNode(pagedivs[0]) |
|
if s=='0': |
try: |
try: |
docinfo['figureEntries'] = int(s) |
docinfo['countFigureEntries'] = int(s) |
except: |
except: |
docinfo['figureEntries'] = 0 |
docinfo['countFigureEntries'] = 0 |
# tocEntries |
else: |
pagedivs = dom.xpath("//div[@class='countTocEntries']") |
s1 = int(s)/30+1 |
s = getTextFromNode(pagedivs[0]) |
|
try: |
try: |
docinfo['tocEntries'] = int(s) |
docinfo['countFigureEntries'] = int(s1) |
except: |
except: |
docinfo['tocEntries'] = 0 |
docinfo['countFigureEntries'] = 0 |
|
|
# allPlaces |
# allPlaces |
pagedivs = dom.xpath("//div[@class='countPlaces']") |
pagedivs = dom.xpath("//div[@class='countPlaces']") |
|
if pagedivs == dom.xpath("//div[@class='countPlaces']"): |
|
if len(pagedivs)>0: |
|
docinfo['countPlaces']= getTextFromNode(pagedivs[0]) |
s = getTextFromNode(pagedivs[0]) |
s = getTextFromNode(pagedivs[0]) |
try: |
try: |
docinfo['allPlaces'] = int(s) |
docinfo['countPlaces'] = int(s) |
except: |
except: |
docinfo['allPlaces'] = 0 |
docinfo['countPlaces'] = 0 |
|
|
|
#tocEntries |
|
pagedivs = dom.xpath("//div[@class='countTocEntries']") |
|
if pagedivs == dom.xpath("//div[@class='countTocEntries']"): |
|
if len(pagedivs)>0: |
|
docinfo['countTocEntries'] = int(getTextFromNode(pagedivs[0])) |
|
s = getTextFromNode(pagedivs[0]) |
|
if s=='0': |
|
try: |
|
docinfo['countTocEntries'] = int(s) |
|
except: |
|
docinfo['countTocEntries'] = 0 |
else: |
else: |
# no full text -- init to 0 |
s1 = int(s)/30+1 |
docinfo['figureEntries'] = 0 |
try: |
docinfo['tocEntries'] = 0 |
docinfo['countTocEntries'] = int(s1) |
docinfo['allPlaces'] = 0 |
except: |
|
docinfo['countTocEntries'] = 0 |
|
|
return docinfo |
#numTextPages |
|
pagedivs = dom.xpath("//div[@class='countPages']") |
|
if pagedivs == dom.xpath("//div[@class='countPages']"): |
|
if len(pagedivs)>0: |
|
docinfo['numPages'] = getTextFromNode(pagedivs[0]) |
|
s = getTextFromNode(pagedivs[0]) |
|
|
|
try: |
|
docinfo['numPages'] = int(s) |
|
#logging.debug("PAGE NUMBER: %s"%(s)) |
|
|
def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="reg", highlightQuery=None, sn=None, optionToggle=None): |
np = docinfo['numPages'] |
"""returns single page from fulltext""" |
pageinfo['end'] = min(pageinfo['end'], np) |
docpath = docinfo['textURLPath'] |
pageinfo['numgroups'] = int(np / pageinfo['groupsize']) |
path = docinfo['textURLPath'] |
if np % pageinfo['groupsize'] > 0: |
url = docinfo['url'] |
pageinfo['numgroups'] += 1 |
name = docinfo['name'] |
except: |
viewMode= pageinfo['viewMode'] |
docinfo['numPages'] = 0 |
sn = pageinfo['sn'] |
|
highlightQuery = pageinfo['highlightQuery'] |
|
|
|
tocMode = pageinfo['tocMode'] |
|
characterNormalization=pageinfo['characterNormalization'] |
|
tocPN = pageinfo['tocPN'] |
|
selfurl = self.absolute_url() |
|
if mode == "text_dict": |
|
textmode = "textPollux" |
|
else: |
else: |
textmode = mode |
#no full text -- init to 0 |
#logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
docinfo['pageNumberOrig'] = 0 |
textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
docinfo['countFigureEntries'] = 0 |
if highlightQuery is not None: |
docinfo['countPlaces'] = 0 |
textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) |
docinfo['countTocEntries'] = 0 |
#logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
docinfo['numPages'] = 0 |
pagexml = self.getServerData("page-fragment.xql",textParam) |
docinfo['pageNumberOrigNorm'] = 0 |
logging.debug("documentViewer highlightQuery: %s"%(highlightQuery)) |
#return docinfo |
#pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False) |
|
|
|
pagedom = Parse(pagexml) |
|
# plain text mode |
# plain text mode |
if mode == "text": |
if mode == "text": |
# first div contains text |
# first div contains text |
pagedivs = pagedom.xpath("/div") |
pagedivs = dom.xpath("/div") |
if len(pagedivs) > 0: |
if len(pagedivs) > 0: |
pagenode = pagedivs[0] |
pagenode = pagedivs[0] |
links = pagenode.xpath("//a") |
links = pagenode.xpath("//a") |
Line 293 class MpdlXmlTextServer(SimpleItem):
|
Line 302 class MpdlXmlTextServer(SimpleItem):
|
if hrefNode: |
if hrefNode: |
href= hrefNode.nodeValue |
href= hrefNode.nodeValue |
if href.startswith('#note-'): |
if href.startswith('#note-'): |
hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn)) |
hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn)) |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
if mode == "xml": |
if mode == "xml": |
# first div contains text |
# first div contains text |
pagedivs = pagedom.xpath("/div") |
pagedivs = dom.xpath("/div") |
if len(pagedivs) > 0: |
if len(pagedivs) > 0: |
pagenode = pagedivs[0] |
pagenode = pagedivs[0] |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
if mode == "gis": |
if mode == "gis": |
# first div contains text |
# first div contains text |
pagedivs = pagedom.xpath("/div") |
pagedivs = dom.xpath("/div") |
if len(pagedivs) > 0: |
if len(pagedivs) > 0: |
pagenode = pagedivs[0] |
pagenode = pagedivs[0] |
links =pagenode.xpath("//a") |
links =pagenode.xpath("//a") |
Line 318 class MpdlXmlTextServer(SimpleItem):
|
Line 327 class MpdlXmlTextServer(SimpleItem):
|
|
|
if mode == "pureXml": |
if mode == "pureXml": |
# first div contains text |
# first div contains text |
pagedivs = pagedom.xpath("/div") |
pagedivs = dom.xpath("/div") |
if len(pagedivs) > 0: |
if len(pagedivs) > 0: |
pagenode = pagedivs[0] |
pagenode = pagedivs[0] |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
# text-with-links mode |
# text-with-links mode |
if mode == "text_dict": |
if mode == "text_dict": |
# first div contains text |
# first div contains text |
pagedivs = pagedom.xpath("/div") |
#mode = pageinfo ['viewMode'] |
|
pagedivs = dom.xpath("/div") |
if len(pagedivs) > 0: |
if len(pagedivs) > 0: |
pagenode = pagedivs[0] |
pagenode = pagedivs[0] |
# check all a-tags |
# check all a-tags |
Line 351 class MpdlXmlTextServer(SimpleItem):
|
Line 361 class MpdlXmlTextServer(SimpleItem):
|
l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
l.setAttributeNS(None, 'onClick', 'popupWin.focus();') |
if href.startswith('#note-'): |
if href.startswith('#note-'): |
hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn)) |
hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn)) |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
return "no text here" |
return "no text here" |
|
|
|
def getOrigPages(self, docinfo=None, pageinfo=None): |
|
docpath = docinfo['textURLPath'] |
|
pn =pageinfo['current'] |
|
selfurl = self.absolute_url() |
|
pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn)) |
|
dom = Parse(pagexml) |
|
pagedivs = dom.xpath("//div[@class='pageNumberOrig']") |
|
if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"): |
|
if len(pagedivs)>0: |
|
docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0]) |
|
return docinfo['pageNumberOrig'] |
|
|
|
def getOrigPagesNorm(self, docinfo=None, pageinfo=None): |
|
docpath = docinfo['textURLPath'] |
|
pn =pageinfo['current'] |
|
selfurl = self.absolute_url() |
|
pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn)) |
|
dom = Parse(pagexml) |
|
pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']") |
|
if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"): |
|
if len(pagedivs)>0: |
|
docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0]) |
|
return docinfo['pageNumberOrigNorm'] |
|
|
|
|
def getTranslate(self, query=None, language=None): |
def getTranslate(self, query=None, language=None): |
"""translate into another languages""" |
"""translate into another languages""" |
data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query)) |
data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query)) |
Line 374 class MpdlXmlTextServer(SimpleItem):
|
Line 409 class MpdlXmlTextServer(SimpleItem):
|
return data |
return data |
|
|
def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None): |
def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None): |
"""number of""" |
#number of |
docpath = docinfo['textURLPath'] |
|
pagesize = pageinfo['queryPageSize'] |
|
pn = pageinfo['searchPN'] |
|
query =pageinfo['query'] |
|
queryType =pageinfo['queryType'] |
|
tocSearch = 0 |
|
tocDiv = None |
|
|
|
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)) |
|
#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) |
|
pagedom = Parse(pagexml) |
|
numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
|
tocSearch = int(getTextFromNode(numdivs[0])) |
|
logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch)) |
|
tc=int((tocSearch/10)+1) |
|
logging.debug("documentViewer (gettoc) tc: %s"%(tc)) |
|
return tc |
|
|
|
def getQueryResultHits(self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionsClose=None): |
|
|
|
"""number of hits in Search mode""" |
|
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
pagesize = pageinfo['queryPageSize'] |
pagesize = pageinfo['queryPageSize'] |
pn = pageinfo['searchPN'] |
pn = pageinfo['searchPN'] |
Line 405 class MpdlXmlTextServer(SimpleItem):
|
Line 419 class MpdlXmlTextServer(SimpleItem):
|
tocDiv = None |
tocDiv = None |
|
|
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)) |
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)) |
#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) |
|
pagedom = Parse(pagexml) |
pagedom = Parse(pagexml) |
numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
tocSearch = int(getTextFromNode(numdivs[0])) |
tocSearch = int(getTextFromNode(numdivs[0])) |
tc=int((tocSearch/10)+1) |
tc=int((tocSearch/10)+1) |
return tc |
return tc |
|
|
def getQueryResultHitsText(self, docinfo=None, pageinfo=None): |
|
"""number of hits in Text of Contents mode""" |
|
selfurl = self.absolute_url() |
|
docpath = docinfo['textURLPath'] |
|
viewMode= pageinfo['viewMode'] |
|
text = self.getServerData("page-fragment.xql","document=%s&mode=%s"%(docinfo['textURLPath'], 'text')) |
|
dom =Parse(text) |
|
pagedivs = dom.xpath("//div[@class='countTocEntries']") |
|
logging.debug ("pagedivs=%s"%(pagedivs)) |
|
if len(pagedivs)>0: |
|
originalPage= (getTextFromNode(pagedivs[0])) |
|
tc = int (originalPage) |
|
tc1 =tc/30+1 |
|
return tc1 |
|
|
|
def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None): |
|
"""number of hits in Text of Figures mode""" |
|
|
|
selfurl = self.absolute_url() |
|
docpath = docinfo['textURLPath'] |
|
viewMode= pageinfo['viewMode'] |
|
text = self.getServerData("page-fragment.xql","document=%s&mode=%s"%(docinfo['textURLPath'], 'text')) |
|
dom =Parse(text) |
|
pagedivs = dom.xpath("//div[@class='countFigureEntries']") |
|
logging.debug ("pagedivs=%s"%(pagedivs)) |
|
if len(pagedivs)>0: |
|
originalPage= (getTextFromNode(pagedivs[0])) |
|
tc = int (originalPage) |
|
tc1 =tc/30+1 |
|
return tc1 |
|
|
|
|
|
def getToc(self, mode="text", docinfo=None): |
def getToc(self, mode="text", docinfo=None): |
"""loads table of contents and stores in docinfo""" |
"""loads table of contents and stores in docinfo""" |
logging.debug("documentViewer (gettoc) mode: %s"%(mode)) |
|
if mode == "none": |
if mode == "none": |
return docinfo |
return docinfo |
if 'tocSize_%s'%mode in docinfo: |
if 'tocSize_%s'%mode in docinfo: |
Line 466 class MpdlXmlTextServer(SimpleItem):
|
Line 446 class MpdlXmlTextServer(SimpleItem):
|
tocDiv = None |
tocDiv = None |
|
|
pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn)) |
pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn)) |
#pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False) |
|
# post-processing downloaded xml |
# post-processing downloaded xml |
pagedom = Parse(pagexml) |
pagedom = Parse(pagexml) |
# get number of entries |
# get number of entries |
Line 498 class MpdlXmlTextServer(SimpleItem):
|
Line 478 class MpdlXmlTextServer(SimpleItem):
|
data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn)) |
data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn)) |
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)) |
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)) |
text = page.replace('mode=image','mode=texttool') |
text = page.replace('mode=image','mode=texttool') |
logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) |
|
#logging.debug("documentViewer (characterNormalization) text: %s"%(text)) |
|
return text |
return text |
|
|
def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
Line 522 def manage_addMpdlXmlTextServer(self,id,
|
Line 500 def manage_addMpdlXmlTextServer(self,id,
|
self.Destination()._setObject(id, newObj) |
self.Destination()._setObject(id, newObj) |
if RESPONSE is not None: |
if RESPONSE is not None: |
RESPONSE.redirect('manage_main') |
RESPONSE.redirect('manage_main') |
|
|
|
|
|
|