version 1.31, 2010/09/23 15:03:51
|
version 1.94, 2010/10/13 15:45:44
|
Line 48 class MpdlXmlTextServer(SimpleItem):
|
Line 48 class MpdlXmlTextServer(SimpleItem):
|
pn = pageinfo['searchPN'] |
pn = pageinfo['searchPN'] |
sn = pageinfo['sn'] |
sn = pageinfo['sn'] |
highlightQuery = pageinfo['highlightQuery'] |
highlightQuery = pageinfo['highlightQuery'] |
#query =pageinfo['query'] |
query =pageinfo['query'] |
queryType =pageinfo['queryType'] |
queryType =pageinfo['queryType'] |
viewMode= pageinfo['viewMode'] |
viewMode= pageinfo['viewMode'] |
tocMode = pageinfo['tocMode'] |
tocMode = pageinfo['tocMode'] |
Line 56 class MpdlXmlTextServer(SimpleItem):
|
Line 56 class MpdlXmlTextServer(SimpleItem):
|
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&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,highlightQuery)) |
data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,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) |
#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) |
Line 118 class MpdlXmlTextServer(SimpleItem):
|
Line 118 class MpdlXmlTextServer(SimpleItem):
|
return serializeNode(pagenode) |
return serializeNode(pagenode) |
return "no text here" |
return "no text here" |
|
|
def getNumPages(self, docinfo): |
"""def getNumPages(self, docinfo): |
"""get list of pages from fulltext and put in docinfo""" |
""get list of pages from fulltext and put in docinfo"" |
if 'numPages' in docinfo: |
if 'numPages' in docinfo: |
# already there |
# already there |
return docinfo |
return docinfo |
|
|
xquery = '//pb' |
xquery = '//pb' |
text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) |
text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) |
#text = self.template.fulltextclient.eval("/mpdl/interface/xquery.xql", "document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) |
|
docinfo['numPages'] = text.count("<pb ") |
docinfo['numPages'] = text.count("<pb ") |
return docinfo |
return docinfo |
|
""" |
|
def getNumTextPages (self, docinfo): |
|
"""get list of pages from fulltext (texts without images) and put in docinfo""" |
|
if 'numPages' in docinfo: |
|
# allredy there |
|
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])) |
|
return docinfo |
|
|
|
def getGisPlaces(self, docinfo=None, pageinfo=None): |
|
""" Show all Gis Places of whole Page""" |
|
#xpath='//place' |
|
docpath = docinfo['textURLPath'] |
|
url = docinfo['url'] |
|
selfurl = self.absolute_url() |
|
pn = pageinfo['current'] |
|
|
|
text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],'//place',pn)) |
|
logging.debug("documentViewer getGisPlaces (text) text:%s"%(text)) |
|
pagedom = Parse(text) |
|
result =pagedom.xpath("//result/resultPage/place") |
|
for l in result: |
|
hrefNode= l.getAttributeNodeNS(None, u"id") |
|
logging.debug("documentViewer getGisPlaces (characterNormalization) l:%s"%(l)) |
|
if hrefNode: |
|
href= hrefNode.nodeValue |
|
if href.startswith('id='): |
|
hrefNode.nodeValue = href.replace('id=',"?") |
|
return serializeNode(pagenode) |
|
|
def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""): |
def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""): |
"""returns single page from fulltext""" |
"""returns single page from fulltext""" |
docpath = docinfo['textURLPath'] |
docpath = docinfo['textURLPath'] |
path = docinfo['textURLPath'] |
path = docinfo['textURLPath'] |
url = docinfo['url'] |
url = docinfo['url'] |
|
name = docinfo['name'] |
viewMode= pageinfo['viewMode'] |
viewMode= pageinfo['viewMode'] |
tocMode = pageinfo['tocMode'] |
tocMode = pageinfo['tocMode'] |
characterNormalization=pageinfo['characterNormalization'] |
characterNormalization=pageinfo['characterNormalization'] |
Line 147 class MpdlXmlTextServer(SimpleItem):
|
Line 179 class MpdlXmlTextServer(SimpleItem):
|
#logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) |
#logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) |
textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
if highlightQuery is not None: |
if highlightQuery is not None: |
textParam +="&highlightQuery=%s&sn=%s"%(highlightQuery,sn) |
textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) |
|
|
pagexml = self.getServerData("page-fragment.xql",textParam) |
pagexml = self.getServerData("page-fragment.xql",textParam) |
#pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False) |
#pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False) |
Line 184 class MpdlXmlTextServer(SimpleItem):
|
Line 216 class MpdlXmlTextServer(SimpleItem):
|
if hrefNode: |
if hrefNode: |
href=hrefNode.nodeValue |
href=hrefNode.nodeValue |
if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): |
if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): |
hrefNode.nodeValue =href.replace('chinagis_REST','chinagis') |
hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name) |
|
l.setAttributeNS(None, 'target', '_blank') |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
|
|
if mode == "pureXml": |
if mode == "pureXml": |
Line 254 class MpdlXmlTextServer(SimpleItem):
|
Line 287 class MpdlXmlTextServer(SimpleItem):
|
tocSearch = 0 |
tocSearch = 0 |
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, 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) |
#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']") |