version 1.215, 2010/12/06 10:41:55
|
version 1.216, 2010/12/20 13:52:07
|
Line 117 class MpdlXmlTextServer(SimpleItem):
|
Line 117 class MpdlXmlTextServer(SimpleItem):
|
return serializeNode(pagenode) |
return serializeNode(pagenode) |
return "no text here" |
return "no text here" |
|
|
"""def getNumPages(self, docinfo): |
|
""get list of pages from fulltext and put in docinfo"" |
def getNumTextPages (self, docinfo=None): |
if 'numPages' in docinfo: |
|
# already there |
|
return docinfo |
|
xquery = '//pb' |
|
text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) |
|
docinfo['numPages'] = text.count("<pb ") |
|
return docinfo |
|
""" |
|
def getNumTextPages (self, docinfo): |
|
"""get list of pages from fulltext (texts without images) and put in docinfo""" |
"""get list of pages from fulltext (texts without images) and put in docinfo""" |
|
docpath = docinfo['textURLPath'] |
|
selfurl = self.absolute_url() |
|
#viewMode= pageinfo['viewMode'] |
if 'numPages' in docinfo: |
if 'numPages' in docinfo: |
# allredy there |
# allredy there |
return docinfo |
return docinfo |
xpath ='/count(//pb)' |
|
text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath)) |
|
dom = Parse(text) |
|
result= dom.xpath("//result/resultPage") |
|
|
|
docinfo['numPages']=int(getTextFromNode(result[0])) |
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 |
return docinfo |
|
|
|
def getTocEntries (self, docinfo=None): |
|
""" number of text entries""" |
|
docpath = docinfo['textURLPath'] |
|
selfurl = self.absolute_url() |
|
if (docpath!=None): |
|
text = self.getServerData("page-fragment.xql","document=%s"%(docinfo['textURLPath'])) |
|
dom =Parse(text) |
|
pagedivs = dom.xpath("//div[@class='countTocEntries']") |
|
logging.debug ("pagedivs=%s"%(pagedivs)) |
|
originalPage = getTextFromNode(pagedivs[0]) |
|
tc = int (originalPage) |
|
return tc |
|
|
|
def getFigureEntries (self, docinfo=None): |
|
""" number of figure entries""" |
|
docpath = docinfo['textURLPath'] |
|
selfurl = self.absolute_url() |
|
if (docpath!=None): |
|
text = self.getServerData("page-fragment.xql","document=%s"%(docinfo['textURLPath'])) |
|
dom = Parse(text) |
|
pagedivs = dom.xpath("//div[@class='countFigureEntries']") |
|
logging.debug ("pagedivs=%s"%(pagedivs)) |
|
originalPage = getTextFromNode(pagedivs[0]) |
|
tc = int (originalPage) |
|
return tc |
|
|
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 184 class MpdlXmlTextServer(SimpleItem):
|
Line 205 class MpdlXmlTextServer(SimpleItem):
|
logging.debug("getALLGisPlaces :%s"%(myList)) |
logging.debug("getALLGisPlaces :%s"%(myList)) |
return myList |
return myList |
|
|
|
|
def getPDF (self, docinfo=None, pageinfo=None): |
|
"""Show and Save different Pages as PDF in Options""" |
|
selfurl=self.absolute_url() |
|
pn=pageinfo['current'] |
|
|
|
viewMode =pageinfo['viewMode'] |
|
|
|
#text = ("page-fragment.xql","document=%s&mode=%s&pn=%s&export=%s"%(docinfo['textURLPath'], 'text', pn,'pdf')) |
|
#text = self.getServerData("page-fragment.xql", "document=%s&mode=%s&pn=%s&export=%s"(docinfo['textURLPath'],'text', pn,'pdf')) |
|
#logging.debug("text :%s"%(text)) |
|
#dom =Parse(text) |
|
#logging.debug("text :%s"%(text)) |
|
#return text |
|
|
|
def getOrigPages (self, docinfo=None, pageinfo=None): |
def getOrigPages (self, docinfo=None, pageinfo=None): |
"""Show original page """ |
"""Show original page """ |
docpath = docinfo.get('textURLPath',None) |
docpath = docinfo.get('textURLPath',None) |
Line 228 class MpdlXmlTextServer(SimpleItem):
|
Line 234 class MpdlXmlTextServer(SimpleItem):
|
dom =Parse(text) |
dom =Parse(text) |
pagedivs = dom.xpath("//div[@class='countPlaces']") |
pagedivs = dom.xpath("//div[@class='countPlaces']") |
logging.debug ("pagedivs=%s"%(pagedivs)) |
logging.debug ("pagedivs=%s"%(pagedivs)) |
#originalPage= getTextFromNode(pagedivs[0]) |
|
#return allPlaces |
|
if len(pagedivs)>0: |
if len(pagedivs)>0: |
originalPage= getTextFromNode(pagedivs[0]) |
originalPage= getTextFromNode(pagedivs[0]) |
#return docinfo['originalPage'] |
|
return originalPage |
return originalPage |
|
|
def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="regPlusNorm", highlightQuery=None, sn=None, optionToggle=None): |
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): |
"""returns single page from fulltext""" |
"""returns single page from fulltext""" |
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
path = docinfo['textURLPath'] |
path = docinfo['textURLPath'] |
Line 396 class MpdlXmlTextServer(SimpleItem):
|
Line 399 class MpdlXmlTextServer(SimpleItem):
|
|
|
def getQueryResultHitsText(self, docinfo=None, pageinfo=None): |
def getQueryResultHitsText(self, docinfo=None, pageinfo=None): |
"""number of hits in Text of Contents mode""" |
"""number of hits in Text of Contents mode""" |
|
selfurl = self.absolute_url() |
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
pagesize = pageinfo['queryPageSize'] |
viewMode= pageinfo['viewMode'] |
pn = pageinfo['searchPN'] |
text = self.getServerData("page-fragment.xql","document=%s&mode=%s"%(docinfo['textURLPath'], 'text')) |
query =pageinfo['query'] |
dom =Parse(text) |
queryType =pageinfo['queryType'] |
pagedivs = dom.xpath("//div[@class='countTocEntries']") |
tocSearch = 0 |
logging.debug ("pagedivs=%s"%(pagedivs)) |
tocDiv = None |
if len(pagedivs)>0: |
tocMode = pageinfo['tocMode'] |
originalPage= (getTextFromNode(pagedivs[0])) |
tocPN = pageinfo['tocPN'] |
tc = int (originalPage) |
pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'toc')) |
tc1 =tc/30+1 |
pagedom = Parse(pagexml) |
return tc1 |
logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom)) |
|
numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
|
tocSearch = int(getTextFromNode(numdivs[0])) |
|
tc=int((tocSearch/30)+1) |
|
return tc |
|
|
|
def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None): |
def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None): |
"""number of hits in Text of Figures mode""" |
"""number of hits in Text of Figures mode""" |
|
|
|
selfurl = self.absolute_url() |
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
pagesize = pageinfo['queryPageSize'] |
viewMode= pageinfo['viewMode'] |
pn = pageinfo['searchPN'] |
text = self.getServerData("page-fragment.xql","document=%s&mode=%s"%(docinfo['textURLPath'], 'text')) |
query =pageinfo['query'] |
dom =Parse(text) |
queryType =pageinfo['queryType'] |
pagedivs = dom.xpath("//div[@class='countFigureEntries']") |
tocSearch = 0 |
logging.debug ("pagedivs=%s"%(pagedivs)) |
tocDiv = None |
if len(pagedivs)>0: |
tocMode = pageinfo['tocMode'] |
originalPage= (getTextFromNode(pagedivs[0])) |
tocPN = pageinfo['tocPN'] |
tc = int (originalPage) |
pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'figures')) |
tc1 =tc/30+1 |
pagedom = Parse(pagexml) |
return tc1 |
logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom)) |
|
numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
|
tocSearch = int(getTextFromNode(numdivs[0])) |
|
tc=int((tocSearch/30)+1) |
|
return tc |
|
|
|
|
|
def getToc(self, mode="text", docinfo=None): |
def getToc(self, mode="text", docinfo=None): |