version 1.45, 2010/04/19 13:13:13
|
version 1.47, 2010/04/27 12:58:31
|
Line 10 from Ft.Xml.Domlette import Nonvalidatin
|
Line 10 from Ft.Xml.Domlette import Nonvalidatin
|
from Ft.Xml.Domlette import PrettyPrint, Print |
from Ft.Xml.Domlette import PrettyPrint, Print |
from Ft.Xml import EMPTY_NAMESPACE, Parse |
from Ft.Xml import EMPTY_NAMESPACE, Parse |
|
|
|
from xml.dom.minidom import parse, parseString |
|
|
|
|
|
|
import Ft.Xml.XPath |
import Ft.Xml.XPath |
import cStringIO |
import cStringIO |
Line 184 class documentViewer(Folder):
|
Line 187 class documentViewer(Folder):
|
self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" |
self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" |
|
|
docinfo = self.getDocinfo(mode=mode,url=url) |
docinfo = self.getDocinfo(mode=mode,url=url) |
pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) |
|
|
|
if tocMode != "thumbs": |
if tocMode != "thumbs": |
# get table of contents |
# get table of contents |
docinfo = self.getToc(mode=tocMode, docinfo=docinfo) |
docinfo = self.getToc(mode=tocMode, docinfo=docinfo) |
|
|
|
pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) |
|
|
if viewMode=="auto": # automodus gewaehlt |
if viewMode=="auto": # automodus gewaehlt |
if docinfo.get("textURL",''): #texturl gesetzt und textViewer konfiguriert |
if docinfo.get("textURL",''): #texturl gesetzt und textViewer konfiguriert |
viewMode="text" |
viewMode="text" |
Line 656 class documentViewer(Folder):
|
Line 662 class documentViewer(Folder):
|
pageinfo['tocPageSize'] = self.REQUEST.get('tocPageSize', '30') |
pageinfo['tocPageSize'] = self.REQUEST.get('tocPageSize', '30') |
pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '20') |
pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '20') |
pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') |
pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') |
|
#if 'tocSize_%s'%tocMode in docinfo: |
|
# cached toc |
|
# pageinfo['tocPN'] = min (int (docinfo['tocSize_%s'%tocMode])/int(pageinfo['tocPageSize']),int(pageinfo['tocPN'])) |
|
|
pageinfo['searchPN'] =self.REQUEST.get('searchPN','1') |
pageinfo['searchPN'] =self.REQUEST.get('searchPN','1') |
pageinfo['sn'] =self.REQUEST.get('sn','1') |
pageinfo['sn'] =self.REQUEST.get('sn','1') |
|
|
Line 669 class documentViewer(Folder):
|
Line 679 class documentViewer(Folder):
|
sn = pageinfo['sn'] |
sn = pageinfo['sn'] |
query =pageinfo['query'] |
query =pageinfo['query'] |
queryType =pageinfo['queryType'] |
queryType =pageinfo['queryType'] |
|
viewMode= pageinfo['viewMode'] |
|
tocMode = pageinfo['tocMode'] |
|
tocPN = pageinfo['tocPN'] |
pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s"%(docpath, 'text', queryType, query, pagesize, pn, sn) ,outputUnicode=False) |
pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s"%(docpath, 'text', queryType, query, pagesize, pn, sn) ,outputUnicode=False) |
pagedom = Parse(pagexml) |
pagedom = Parse(pagexml) |
#pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
|
|
|
selfurl = self.absolute_url() |
|
|
|
page = pagexml.replace('page-fragment.xql?document=/echo/la/Benedetti_1585.xml','%s?url=/mpiwg/online/permanent/library/163127KK&viewMode=%s&tocMode=%s&tocPN=%s&query=%s&queryType=%s'%(selfurl, viewMode, tocMode, tocPN, query, queryType)) |
|
text =page.replace('mode=text','mode=texttool') |
|
href = text.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl) |
|
lemma= href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl) |
|
#logging.debug("documentViewer (gettoc) lemma: %s"%(lemma)) |
|
|
|
return lemma |
|
|
|
|
return pagexml |
|
#if len(pagedivs) > 0: |
#if len(pagedivs) > 0: |
# pagenode = pagedom[0] |
# pagenode = pagedom[0] |
# return serializeNode(pagenode) |
# return serializeNode(pagenode) |
Line 689 class documentViewer(Folder):
|
Line 711 class documentViewer(Folder):
|
docinfo['numPages'] = text.count("<pb ") |
docinfo['numPages'] = text.count("<pb ") |
return docinfo |
return docinfo |
|
|
def getTextPage(self, mode="text", pn=1, docinfo=None): |
def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None,): |
"""returns single page from fulltext""" |
"""returns single page from fulltext""" |
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
if mode == "text_dict": |
if mode == "text_dict": |
Line 697 class documentViewer(Folder):
|
Line 719 class documentViewer(Folder):
|
else: |
else: |
textmode = mode |
textmode = mode |
|
|
|
#selfurl = self.absolute_url() |
|
#viewMode= pageinfo['viewMode'] |
|
#tocMode = pageinfo['tocMode'] |
|
#tocPN = pageinfo['tocPN'] |
|
|
pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", "document=%s&mode=%s&pn=%s"%(docpath,textmode,pn), outputUnicode=False) |
pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", "document=%s&mode=%s&pn=%s"%(docpath,textmode,pn), outputUnicode=False) |
####### |
####### |
#textpython = pagexml.replace('page-fragment.xql?document=/echo/la/Benedetti_1585.xml','?url=/mpiwg/online/permanent/library/163127KK&tocMode='+str(tocMode)+'&queryResultPN='+str(queryResultPN)+'&viewMode='+str(viewMode)) |
#page = pagexml.replace('page-fragment.xql?document=/echo/la/Benedetti_1585.xml','%s?url=/mpiwg/online/permanent/library/163127KK&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl, viewMode, tocMode, tocPN)) |
#textnew =textpython.replace('mode=text','mode=texttool') |
#text =page.replace('mode=text','mode=texttool') |
####### |
####### |
# post-processing downloaded xml |
# post-processing downloaded xml |
pagedom = Parse(pagexml) |
pagedom = Parse(pagexml) |
Line 730 class documentViewer(Folder):
|
Line 757 class documentViewer(Folder):
|
# is pollux link |
# is pollux link |
selfurl = self.absolute_url() |
selfurl = self.absolute_url() |
# change href |
# change href |
hrefNode.nodeValue = href.replace('lt/lex.xql','%s/head_main_voc'%selfurl) |
hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl) |
# add target |
# add target |
l.setAttributeNS(None, 'target', '_blank') |
l.setAttributeNS(None, 'target', '_blank') |
|
|
|
if href.startswith('lt/lemma.xql'): |
|
selfurl = self.absolute_url() |
|
hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl) |
|
l.setAttributeNS(None, 'target', '_blank') |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
|
|
return "no text here" |
return "no text here" |
|
|
|
def getTranslate(self, query=None, language=None): |
|
"""translate into another languages""" |
|
pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","query=%s&language=%s"%(query,language),outputUnicode=False) |
|
return pagexml |
|
|
|
def getLemma(self, lemma=None, language=None): |
|
"""lemma""" |
|
pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","lemma=%s&language=%s"%(lemma,language),outputUnicode=False) |
|
return pagexml |
|
|
|
def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1): |
|
"""number of""" |
|
docpath = docinfo['textURLPath'] |
|
pagesize = pageinfo['queryPageSize'] |
|
pn = pageinfo['searchPN'] |
|
query =pageinfo['query'] |
|
queryType =pageinfo['queryType'] |
|
|
|
tocSearch = 0 |
|
tocDiv = None |
|
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])) |
|
tc=int((tocSearch/20)+1) |
|
logging.debug("documentViewer (gettoc) tc: %s"%(tc)) |
|
return tc |
|
|
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)) |
logging.debug("documentViewer (gettoc) mode: %s"%(mode)) |
if 'tocSize_%s'%mode in docinfo: |
if 'tocSize_%s'%mode in docinfo: |
# cached toc |
# cached toc |
return docinfo |
return docinfo |
|
|
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
# we need to set a result set size |
# we need to set a result set size |
pagesize = 1000 |
pagesize = 1000 |
Line 781 class documentViewer(Folder):
|
Line 841 class documentViewer(Folder):
|
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
pagesize = pageinfo['tocPageSize'] |
pagesize = pageinfo['tocPageSize'] |
pn = pageinfo['tocPN'] |
pn = pageinfo['tocPN'] |
|
|
|
selfurl = self.absolute_url() |
|
viewMode= pageinfo['viewMode'] |
|
tocMode = pageinfo['tocMode'] |
|
tocPN = pageinfo['tocPN'] |
|
|
pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn), outputUnicode=False) |
pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn), outputUnicode=False) |
|
page = pagexml.replace('page-fragment.xql?document=/echo/la/Benedetti_1585.xml','%s?url=/mpiwg/online/permanent/library/163127KK&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl, viewMode, tocMode, tocPN)) |
|
text = page.replace('mode=image','mode=texttool') |
|
return text |
# post-processing downloaded xml |
# post-processing downloaded xml |
pagedom = Parse(pagexml) |
#pagedom = Parse(text) |
# div contains text |
# div contains text |
pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
#pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
if len(pagedivs) > 0: |
#if len(pagedivs) > 0: |
pagenode = pagedivs[0] |
# pagenode = pagedivs[0] |
return serializeNode(pagenode) |
# return serializeNode(pagenode) |
else: |
#else: |
return "No TOC!" |
# return "No TOC!" |
|
|
|
|
def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None): |
def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None): |