version 1.210, 2010/11/09 12:50:06
|
version 1.218, 2010/12/23 16:12:23
|
Line 71 class MpdlXmlTextServer(SimpleItem):
|
Line 71 class MpdlXmlTextServer(SimpleItem):
|
href = hrefNode.nodeValue |
href = hrefNode.nodeValue |
if href.startswith('page-fragment.xql'): |
if href.startswith('page-fragment.xql'): |
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'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle)) |
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) |
return serializeNode(pagenode) |
return serializeNode(pagenode) |
if (queryType=="fulltextMorph"): |
if (queryType=="fulltextMorph"): |
Line 85 class MpdlXmlTextServer(SimpleItem):
|
Line 85 class MpdlXmlTextServer(SimpleItem):
|
href = hrefNode.nodeValue |
href = hrefNode.nodeValue |
if href.startswith('page-fragment.xql'): |
if href.startswith('page-fragment.xql'): |
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'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle)) |
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) |
if href.startswith('../lt/lemma.xql'): |
if href.startswith('../lt/lemma.xql'): |
hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl)) |
hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl)) |
Line 103 class MpdlXmlTextServer(SimpleItem):
|
Line 103 class MpdlXmlTextServer(SimpleItem):
|
hrefNode = l.getAttributeNodeNS(None, u"href") |
hrefNode = l.getAttributeNodeNS(None, u"href") |
if hrefNode: |
if hrefNode: |
href = hrefNode.nodeValue |
href = hrefNode.nodeValue |
hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&optionToggle=%s'%(viewMode,tocMode,tocPN,pn,optionToggle)) |
hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,optionToggle,characterNormalization)) |
if href.startswith('../lt/lex.xql'): |
if href.startswith('../lt/lex.xql'): |
hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl) |
hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl) |
l.setAttributeNS(None, 'target', '_blank') |
l.setAttributeNS(None, 'target', '_blank') |
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""" |
|
logging.debug("getNumTextPages") |
|
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""" |
|
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' |
docpath = docinfo['textURLPath'] |
docpath = docinfo.get('textURLPath',None) |
|
if not docpath: |
|
return None |
|
|
url = docinfo['url'] |
url = docinfo['url'] |
selfurl = self.absolute_url() |
selfurl = self.absolute_url() |
pn = pageinfo['current'] |
pn = pageinfo['current'] |
Line 181 class MpdlXmlTextServer(SimpleItem):
|
Line 190 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['textURLPath'] |
docpath = docinfo.get('textURLPath',None) |
logging.debug ("docinfo['textURLPath']=%s"%(docinfo['textURLPath'])) |
if not docpath: |
#url = docinfo['url'] |
return None |
selfurl = self.absolute_url() |
selfurl = self.absolute_url() |
pn =pageinfo['current'] |
pn =pageinfo['current'] |
|
|
Line 208 class MpdlXmlTextServer(SimpleItem):
|
Line 202 class MpdlXmlTextServer(SimpleItem):
|
text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn)) |
text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn)) |
dom =Parse(text) |
dom =Parse(text) |
pagedivs = dom.xpath("//div[@class='pageNumberOrig']") |
pagedivs = dom.xpath("//div[@class='pageNumberOrig']") |
logging.debug("YYYYYYpagedivs :%s"%(pagedivs)) |
|
if len(pagedivs)>0: |
if len(pagedivs)>0: |
originalPage= getTextFromNode(pagedivs[0]) |
originalPage= getTextFromNode(pagedivs[0]) |
#return docinfo['originalPage'] |
#return docinfo['originalPage'] |
return originalPage |
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): |
|
""" extract diverse info from page-fragment""" |
|
docpath = docinfo['textURLPath'] |
|
if 'allPlaces' in docinfo: |
|
# allredy there |
|
return docinfo |
|
|
|
if (docpath is not None): |
|
text = self.getServerData("page-fragment.xql","document=%s"%(docinfo['textURLPath'])) |
|
dom = Parse(text) |
|
# figureEntries |
|
pagedivs = dom.xpath("//div[@class='countFigureEntries']") |
|
docinfo['figureEntries'] = getTextFromNode(pagedivs[0]) |
|
# tocEntries |
|
pagedivs = dom.xpath("//div[@class='countTocEntries']") |
|
docinfo['tocEntries'] = getTextFromNode(pagedivs[0]) |
|
# allPlaces |
|
pagedivs = dom.xpath("//div[@class='countPlaces']") |
|
docinfo['allPlaces']= getTextFromNode(pagedivs[0]) |
|
|
|
return docinfo |
|
|
|
|
def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="", 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 376 class MpdlXmlTextServer(SimpleItem):
|
Line 396 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/10)+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/10)+1) |
|
return tc |
|
|
|
|
|
def getToc(self, mode="text", docinfo=None): |
def getToc(self, mode="text", docinfo=None): |
Line 466 class MpdlXmlTextServer(SimpleItem):
|
Line 477 class MpdlXmlTextServer(SimpleItem):
|
tocMode = pageinfo['tocMode'] |
tocMode = pageinfo['tocMode'] |
tocPN = pageinfo['tocPN'] |
tocPN = pageinfo['tocPN'] |
|
|
data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s&optionToggle=%s"%(docpath,queryType, pagesize, pn,characterNormalization,optionToggle)) |
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=%s'%(selfurl,url, viewMode, tocMode, tocPN,optionToggle)) |
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) characterNormalization: %s"%(characterNormalization)) |
#logging.debug("documentViewer (characterNormalization) text: %s"%(text)) |
#logging.debug("documentViewer (characterNormalization) text: %s"%(text)) |