--- documentViewer/MpdlXmlTextServer.py 2011/06/14 09:57:11 1.238
+++ documentViewer/MpdlXmlTextServer.py 2011/08/12 14:41:39 1.238.2.13
@@ -1,15 +1,56 @@
-
from OFS.SimpleItem import SimpleItem
from Products.PageTemplates.PageTemplateFile import PageTemplateFile
+
from Ft.Xml import EMPTY_NAMESPACE, Parse
from Ft.Xml.Domlette import NonvalidatingReader
+import Ft.Xml.Domlette
+import cStringIO
+
+import xml.etree.ElementTree as ET
-import md5
-import sys
+import re
import logging
import urllib
-import documentViewer
-from documentViewer import getTextFromNode, serializeNode
+
+from SrvTxtUtils import getInt, getText, getHttpData
+
+def serialize(node):
+ """returns a string containing an XML snippet of node"""
+ s = ET.tostring(node, 'UTF-8')
+ # snip off XML declaration
+ if s.startswith('')
+ return s[i+3:]
+
+ return s
+
+
+def getTextFromNode(node):
+ """get the cdata content of a node"""
+ if node is None:
+ return ""
+
+ # 4Suite:
+ nodelist=node.childNodes
+ text = ""
+ for n in nodelist:
+ if n.nodeType == node.TEXT_NODE:
+ text = text + n.data
+
+ return text
+
+def serializeNode(node, encoding="utf-8"):
+ """returns a string containing node as XML"""
+ #s = ET.tostring(node)
+
+ # 4Suite:
+ stream = cStringIO.StringIO()
+ Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding)
+ s = stream.getvalue()
+ stream.close()
+
+ return s
+
class MpdlXmlTextServer(SimpleItem):
"""TextServer implementation for MPDL-XML eXist server"""
@@ -21,9 +62,7 @@ class MpdlXmlTextServer(SimpleItem):
manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
- def __init__(self,id,title="",serverUrl="http://mpdl-system.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
- #def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/", serverName=None, timeout=40):
-
+ def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
"""constructor"""
self.id=id
self.title=title
@@ -35,20 +74,22 @@ class MpdlXmlTextServer(SimpleItem):
def getHttpData(self, url, data=None):
"""returns result from url+data HTTP request"""
- return documentViewer.getHttpData(url,data,timeout=self.timeout)
+ return getHttpData(url,data,timeout=self.timeout)
def getServerData(self, method, data=None):
"""returns result from text server for method+data"""
url = self.serverUrl+method
- return documentViewer.getHttpData(url,data,timeout=self.timeout)
+ return getHttpData(url,data,timeout=self.timeout)
+ # WTF: what does this really do? can it be integrated in getPage?
def getSearch(self, pageinfo=None, docinfo=None):
"""get search list"""
+ logging.debug("getSearch()")
docpath = docinfo['textURLPath']
url = docinfo['url']
pagesize = pageinfo['queryPageSize']
pn = pageinfo.get('searchPN',1)
- sn = pageinfo['sn']
+ sn = pageinfo.get('sn',None) #TODO: is this s now?
highlightQuery = pageinfo['highlightQuery']
query =pageinfo['query']
queryType =pageinfo['queryType']
@@ -140,18 +181,16 @@ class MpdlXmlTextServer(SimpleItem):
if not docpath:
return None
- url = docinfo['url']
- selfurl = self.absolute_url()
pn = pageinfo['current']
hrefList=[]
myList= ""
text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
- dom = Parse(text)
- result = dom.xpath("//result/resultPage/place")
+ dom = ET.fromstring(text)
+ result = dom.findall(".//result/resultPage/place")
for l in result:
- hrefNode= l.getAttributeNodeNS(None, u"id")
- href= hrefNode.nodeValue
+ href = l.get("id")
hrefList.append(href)
+ # WTF: what does this do?
myList = ",".join(hrefList)
#logging.debug("getGisPlaces :%s"%(myList))
return myList
@@ -159,271 +198,227 @@ class MpdlXmlTextServer(SimpleItem):
def getAllGisPlaces (self, docinfo=None, pageinfo=None):
"""Show all Gis Places of whole Book """
xpath ='//echo:place'
- docpath =docinfo['textURLPath']
- url = docinfo['url']
- selfurl =self.absolute_url()
- pn =pageinfo['current']
hrefList=[]
myList=""
text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
- dom =Parse(text)
- result = dom.xpath("//result/resultPage/place")
+ dom = ET.fromstring(text)
+ result = dom.findall(".//result/resultPage/place")
for l in result:
- hrefNode = l.getAttributeNodeNS(None, u"id")
- href= hrefNode.nodeValue
+ href = l.get("id")
hrefList.append(href)
+ # WTF: what does this do?
myList = ",".join(hrefList)
#logging.debug("getALLGisPlaces :%s"%(myList))
return myList
+ def processPageInfo(self, dom, docinfo, pageinfo):
+ """processes page info divs from dom and stores in docinfo and pageinfo"""
+ # assume first second level div is pageMeta
+ alldivs = dom.find("div")
+
+ if alldivs is None or alldivs.get('class', '') != 'pageMeta':
+ logging.error("processPageInfo: pageMeta div not found!")
+ return
+
+ for div in alldivs:
+ dc = div.get('class')
+
+ # pageNumberOrig
+ if dc == 'pageNumberOrig':
+ pageinfo['pageNumberOrig'] = div.text
+
+ # pageNumberOrigNorm
+ elif dc == 'pageNumberOrigNorm':
+ pageinfo['pageNumberOrigNorm'] = div.text
+
+ # pageHeaderTitle
+ elif dc == 'pageHeaderTitle':
+ pageinfo['pageHeaderTitle'] = div.text
+
+ # numFigureEntries
+ elif dc == 'countFigureEntries':
+ docinfo['numFigureEntries'] = getInt(div.text)
+
+ # numTocEntries
+ elif dc == 'countTocEntries':
+ # WTF: s1 = int(s)/30+1
+ docinfo['numTocEntries'] = getInt(div.text)
+
+ # numPlaces
+ elif dc == 'countPlaces':
+ docinfo['numPlaces'] = getInt(div.text)
+
+ # numTextPages
+ elif dc == 'countPages':
+ np = getInt(div.text)
+ if np > 0:
+ docinfo['numTextPages'] = np
+ if docinfo.get('numPages', 0) == 0:
+ # seems to be text-only - update page count
+ docinfo['numPages'] = np
+ #pageinfo['end'] = min(pageinfo['end'], np)
+ pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
+ if np % pageinfo['groupsize'] > 0:
+ pageinfo['numgroups'] += 1
+
+ #logging.debug("processPageInfo: pageinfo=%s"%repr(pageinfo))
+ return
+
- def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None):
+ def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None):
"""returns single page from fulltext"""
+ logging.debug("getTextPage mode=%s, pn=%s"%(mode,pn))
+ # check for cached text -- but ideally this shouldn't be called twice
+ if pageinfo.has_key('textPage'):
+ logging.debug("getTextPage: using cached text")
+ return pageinfo['textPage']
+
docpath = docinfo['textURLPath']
- path = docinfo['textURLPath']
- url = docinfo.get('url',None)
- name = docinfo.get('name',None)
- pn =pageinfo['current']
- sn = pageinfo['sn']
- #optionToggle =pageinfo ['optionToggle']
- highlightQuery = pageinfo['highlightQuery']
- #mode = pageinfo ['viewMode']
- tocMode = pageinfo['tocMode']
- characterNormalization=pageinfo['characterNormalization']
- tocPN = pageinfo['tocPN']
- selfurl = self.absolute_url()
- if mode == "text_dict":
+ # just checking
+ if pageinfo['current'] != pn:
+ logging.warning("getTextPage: current!=pn!")
+
+ # stuff for constructing full urls
+ url = docinfo['url']
+ urlmode = docinfo['mode']
+ sn = pageinfo.get('sn', None)
+ highlightQuery = pageinfo.get('highlightQuery', None)
+ tocMode = pageinfo.get('tocMode', None)
+ tocPN = pageinfo.get('tocPN',None)
+ characterNormalization = pageinfo.get('characterNormalization', None)
+
+ selfurl = docinfo['viewerUrl']
+
+ if mode == "dict" or mode == "text_dict":
+ # dict is called textPollux in the backend
textmode = "textPollux"
+ elif not mode:
+ # default is text
+ mode = "text"
+ textmode = "text"
else:
textmode = mode
textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
- if highlightQuery is not None:
+ if highlightQuery:
textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
+ # fetch the page
pagexml = self.getServerData("page-fragment.xql",textParam)
- dom = Parse(pagexml)
- #dom = NonvalidatingReader.parseStream(pagexml)
-
- #original Pages
- pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
-
- """if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
- if len(pagedivs)>0:
- docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
- logging.debug("ORIGINAL PAGE: %s"%(docinfo['pageNumberOrig']))
-
- #original Pages Norm
- pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
- if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
- if len(pagedivs)>0:
- docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
- logging.debug("ORIGINAL PAGE NORM: %s"%(docinfo['pageNumberOrigNorm']))
- """
- #figureEntries
- 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])
- if s=='0':
- try:
- docinfo['countFigureEntries'] = int(s)
- except:
- docinfo['countFigureEntries'] = 0
- else:
- s1 = int(s)/30+1
- try:
- docinfo['countFigureEntries'] = int(s1)
- except:
- docinfo['countFigureEntries'] = 0
-
- #allPlaces
- 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])
- try:
- docinfo['countPlaces'] = int(s)
- except:
- 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:
- s1 = int(s)/30+1
- try:
- docinfo['countTocEntries'] = int(s1)
- except:
- docinfo['countTocEntries'] = 0
-
- #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))
-
- np = docinfo['numPages']
- pageinfo['end'] = min(pageinfo['end'], np)
- pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
- if np % pageinfo['groupsize'] > 0:
- pageinfo['numgroups'] += 1
- except:
- docinfo['numPages'] = 0
-
- else:
- #no full text -- init to 0
- docinfo['pageNumberOrig'] = 0
- docinfo['countFigureEntries'] = 0
- docinfo['countPlaces'] = 0
- docinfo['countTocEntries'] = 0
- docinfo['numPages'] = 0
- docinfo['pageNumberOrigNorm'] = 0
- #return docinfo
+ dom = ET.fromstring(pagexml)
+ # extract additional info
+ self.processPageInfo(dom, docinfo, pageinfo)
+ # page content is in
+ pagediv = None
+ # ElementTree 1.2 in Python 2.6 can't do div[@class='pageContent']
+ # so we look at the second level divs
+ alldivs = dom.findall("div")
+ for div in alldivs:
+ dc = div.get('class')
+ # page content div
+ if dc == 'pageContent':
+ pagediv = div
+ break
# plain text mode
if mode == "text":
- # first div contains text
- pagedivs = dom.xpath("/div")
- if len(pagedivs) > 0:
- pagenode = pagedivs[0]
- links = pagenode.xpath("//a")
+ # get full url assuming documentViewer is parent
+ selfurl = self.getLink()
+ if pagediv is not None:
+ links = pagediv.findall(".//a")
for l in links:
- hrefNode = l.getAttributeNodeNS(None, u"href")
- if hrefNode:
- href= hrefNode.nodeValue
- if href.startswith('#note-'):
- hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn))
- return serializeNode(pagenode)
- if mode == "xml":
- # first div contains text
- pagedivs = dom.xpath("/div")
- if len(pagedivs) > 0:
- pagenode = pagedivs[0]
- return serializeNode(pagenode)
- if mode == "gis":
- # first div contains text
- pagedivs = dom.xpath("/div")
- if len(pagedivs) > 0:
- pagenode = pagedivs[0]
- links =pagenode.xpath("//a")
- for l in links:
- hrefNode =l.getAttributeNodeNS(None, u"href")
- if hrefNode:
- href=hrefNode.nodeValue
- if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
- hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
- l.setAttributeNS(None, 'target', '_blank')
- return serializeNode(pagenode)
-
- if mode == "pureXml":
- # first div contains text
- pagedivs = dom.xpath("/div")
- if len(pagedivs) > 0:
- pagenode = pagedivs[0]
- return serializeNode(pagenode)
+ href = l.get('href')
+ if href and href.startswith('#note-'):
+ href = href.replace('#note-',"%s#note-"%selfurl)
+ l.set('href', href)
+
+ return serialize(pagediv)
+
# text-with-links mode
- if mode == "text_dict":
- # first div contains text
- #mode = pageinfo ['viewMode']
- pagedivs = dom.xpath("/div")
- if len(pagedivs) > 0:
- pagenode = pagedivs[0]
+ elif mode == "dict":
+ if pagediv is not None:
+ viewerurl = docinfo['viewerUrl']
+ selfurl = self.getLink()
# check all a-tags
- links = pagenode.xpath("//a")
-
+ links = pagediv.findall(".//a")
for l in links:
- hrefNode = l.getAttributeNodeNS(None, u"href")
+ href = l.get('href')
- if hrefNode:
+ if href:
# is link with href
- href = hrefNode.nodeValue
if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql'):
- # is pollux link
- selfurl = self.absolute_url()
- # change href
- hrefNode.nodeValue = href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/head_main_voc'%selfurl)
- # add target
- l.setAttributeNS(None, 'target', '_blank')
- #l.setAttributeNS(None, 'onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;")
- #l.setAttributeNS(None, "ondblclick", "popupWin.focus();")
- #window.open("this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=yes, scrollbars=1'"); return false;")
+ # is dictionary link - change href (keeping parameters)
+ l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/template/viewer_wordinfo'%viewerurl))
+ # add target to open new page
+ l.set('target', '_blank')
- if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql'):
+ # TODO: is this needed?
+ if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql'):
selfurl = self.absolute_url()
- hrefNode.nodeValue = href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl)
- l.setAttributeNS(None, 'target', '_blank')
- l.setAttributeNS(None, 'onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;")
- l.setAttributeNS(None, 'ondblclick', 'popupWin.focus();')
+ l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl))
+ l.set('target', '_blank')
+ l.set('onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;")
+ l.set('ondblclick', 'popupWin.focus();')
if href.startswith('#note-'):
- hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn))
+ # note link
+ l.set('href', href.replace('#note-',"%s#note-"%selfurl))
- return serializeNode(pagenode)
+ return serialize(pagediv)
+
+ # xml mode
+ elif mode == "xml":
+ if pagediv is not None:
+ return serialize(pagediv)
+
+ # pureXml mode
+ elif mode == "pureXml":
+ if pagediv is not None:
+ return serialize(pagediv)
+
+ # gis mode
+ elif mode == "gis":
+ name = docinfo['name']
+ if pagediv is not None:
+ # check all a-tags
+ links = pagediv.findall(".//a")
+ for l in links:
+ href = l.get('href')
+ if href:
+ if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
+ l.set('href', href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name))
+ l.set('target', '_blank')
+
+ return serialize(pagediv)
+
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, word=None, language=None):
- """translate into another languages"""
- data = self.getServerData("lt/wordInfo.xql","language="+str(language)+"&word="+urllib.quote(word)+"&output=html")
- #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
+ # TODO: should be getWordInfo
+ def getWordInfo(self, word='', language='', display=''):
+ """show information (like dictionaries) about word"""
+ data = self.getServerData("lt/wordInfo.xql","language=%s&word=%s&display=%s&output=html"%(language,urllib.quote(word),urllib.quote(display)))
return data
+ # WTF: what does this do?
def getLemma(self, lemma=None, language=None):
"""simular words lemma """
data = self.getServerData("lt/lemma.xql","language="+str(language)+"&lemma="+urllib.quote(lemma)+"&output=html")
return data
+ # WTF: what does this do?
def getLemmaQuery(self, query=None, language=None):
"""simular words lemma """
data = self.getServerData("lt/lemma.xql","language="+str(language)+"&query="+urllib.quote(query)+"&output=html")
return data
+ # WTF: what does this do?
def getLex(self, query=None, language=None):
#simular words lemma
data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
return data
-
+
+ # WTF: what does this do?
def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
#number of
docpath = docinfo['textURLPath']
@@ -442,9 +437,11 @@ class MpdlXmlTextServer(SimpleItem):
return tc
def getToc(self, mode="text", docinfo=None):
- """loads table of contents and stores in docinfo"""
+ """loads table of contents and stores XML in docinfo"""
+ logging.debug("getToc mode=%s"%mode)
if mode == "none":
- return docinfo
+ return docinfo
+
if 'tocSize_%s'%mode in docinfo:
# cached toc
return docinfo
@@ -460,44 +457,78 @@ class MpdlXmlTextServer(SimpleItem):
# number of entries in toc
tocSize = 0
tocDiv = None
-
+ # fetch full toc
pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
-
- # post-processing downloaded xml
- pagedom = Parse(pagexml)
- # get number of entries
- numdivs = pagedom.xpath("//div[@class='queryResultHits']")
- if len(numdivs) > 0:
- tocSize = int(getTextFromNode(numdivs[0]))
- docinfo['tocSize_%s'%mode] = tocSize
+ dom = ET.fromstring(pagexml)
+ # page content is in
+ pagediv = None
+ # ElementTree 1.2 in Python 2.6 can't do div[@class='queryResultPage']
+ alldivs = dom.findall("div")
+ for div in alldivs:
+ dc = div.get('class')
+ # page content div
+ if dc == 'queryResultPage':
+ pagediv = div
+
+ elif dc == 'queryResultHits':
+ docinfo['tocSize_%s'%mode] = getInt(div.text)
+
+ if pagediv:
+ # store XML in docinfo
+ docinfo['tocXML_%s'%mode] = ET.tostring(pagediv, 'UTF-8')
+
return docinfo
- def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
+ def getTocPage(self, mode="text", pn=0, pageinfo=None, docinfo=None):
"""returns single page from the table of contents"""
- # TODO: this should use the cached TOC
+ logging.debug("getTocPage mode=%s, pn=%s"%(mode,pn))
if mode == "text":
queryType = "toc"
else:
queryType = mode
- docpath = docinfo['textURLPath']
- path = docinfo['textURLPath']
+
+ # check for cached TOC
+ if not docinfo.has_key('tocXML_%s'%mode):
+ self.getToc(mode=mode, docinfo=docinfo)
+
+ tocxml = docinfo.get('tocXML_%s'%mode, None)
+ if not tocxml:
+ logging.error("getTocPage: unable to find tocXML")
+ return "No ToC"
+
pagesize = pageinfo['tocPageSize']
- pn = pageinfo['tocPN']
- url = docinfo['url']
- selfurl = self.absolute_url()
- viewMode= pageinfo['viewMode']
- characterNormalization = pageinfo ['characterNormalization']
- #optionToggle =pageinfo ['optionToggle']
- tocMode = pageinfo['tocMode']
- tocPN = pageinfo['tocPN']
+ tocPN = pageinfo['tocPN']
+ if not pn:
+ pn = tocPN
+
+ fulltoc = ET.fromstring(tocxml)
- data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm"%(docpath,queryType, pagesize, pn))
- page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
- text = page.replace('mode=image','mode=texttool')
- return text
+ if fulltoc:
+ # paginate
+ start = (pn - 1) * pagesize * 2
+ len = pagesize * 2
+ del fulltoc[:start]
+ del fulltoc[len:]
+ tocdivs = fulltoc
+
+ # check all a-tags
+ links = tocdivs.findall(".//a")
+ for l in links:
+ href = l.get('href')
+ if href:
+ # take pn from href
+ m = re.match(r'page-fragment\.xql.*pn=(\d+)', href)
+ if m is not None:
+ # and create new url (assuming parent is documentViewer)
+ url = self.getLink('pn', m.group(1))
+ l.set('href', url)
+ else:
+ logging.warning("getTocPage: Problem with link=%s"%href)
+
+ return serialize(tocdivs)
+
def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
- #def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None):
"""change settings"""
self.title=title
self.timeout = timeout
@@ -517,4 +548,6 @@ def manage_addMpdlXmlTextServer(self,id,
newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
self.Destination()._setObject(id, newObj)
if RESPONSE is not None:
- RESPONSE.redirect('manage_main')
\ No newline at end of file
+ RESPONSE.redirect('manage_main')
+
+
\ No newline at end of file