version 1.47, 2010/04/27 12:58:31
|
version 1.69.2.1, 2010/06/14 10:49:48
|
Line 2
|
Line 2
|
from OFS.Folder import Folder |
from OFS.Folder import Folder |
from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate |
from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate |
from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
|
from Products.PythonScripts.standard import url_quote |
from AccessControl import ClassSecurityInfo |
from AccessControl import ClassSecurityInfo |
from AccessControl import getSecurityManager |
from AccessControl import getSecurityManager |
from Globals import package_home |
from Globals import package_home |
Line 12 from Ft.Xml import EMPTY_NAMESPACE, Pars
|
Line 13 from Ft.Xml import EMPTY_NAMESPACE, Pars
|
|
|
from xml.dom.minidom import parse, parseString |
from xml.dom.minidom import parse, parseString |
|
|
|
from extraFunction import * |
|
|
|
|
import Ft.Xml.XPath |
import Ft.Xml.XPath |
Line 77 def urlopen(url,timeout=2):
|
Line 79 def urlopen(url,timeout=2):
|
## |
## |
## documentViewer class |
## documentViewer class |
## |
## |
class documentViewer(Folder): |
class documentViewer(Folder, extraFunction): |
"""document viewer""" |
"""document viewer""" |
#textViewerUrl="http://127.0.0.1:8080/HFQP/testXSLT/getPage?" |
#textViewerUrl="http://127.0.0.1:8080/HFQP/testXSLT/getPage?" |
|
|
Line 96 class documentViewer(Folder):
|
Line 98 class documentViewer(Folder):
|
page_main_images = PageTemplateFile('zpt/page_main_images', globals()) |
page_main_images = PageTemplateFile('zpt/page_main_images', globals()) |
page_main_text = PageTemplateFile('zpt/page_main_text', globals()) |
page_main_text = PageTemplateFile('zpt/page_main_text', globals()) |
page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals()) |
page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals()) |
|
page_main_xml = PageTemplateFile('zpt/page_main_xml', globals()) |
head_main = PageTemplateFile('zpt/head_main', globals()) |
head_main = PageTemplateFile('zpt/head_main', globals()) |
docuviewer_css = PageTemplateFile('css/docuviewer.css', globals()) |
docuviewer_css = PageTemplateFile('css/docuviewer.css', globals()) |
info_xml = PageTemplateFile('zpt/info_xml', globals()) |
info_xml = PageTemplateFile('zpt/info_xml', globals()) |
|
|
|
|
thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals()) |
thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals()) |
security.declareProtected('View management screens','changeDocumentViewerForm') |
security.declareProtected('View management screens','changeDocumentViewerForm') |
changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals()) |
changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals()) |
Line 172 class documentViewer(Folder):
|
Line 176 class documentViewer(Folder):
|
@param mode: defines how to access the document behind url |
@param mode: defines how to access the document behind url |
@param url: url which contains display information |
@param url: url which contains display information |
@param viewMode: if images display images, if text display text, default is auto (text,images or auto) |
@param viewMode: if images display images, if text display text, default is auto (text,images or auto) |
@param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, search) |
@param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, none) |
@param querySearch: type of different search modes (fulltext, fulltextMorph, xpath, xquery, ftIndex, ftIndexMorph) |
@param querySearch: type of different search modes (fulltext, fulltextMorph, xpath, xquery, ftIndex, ftIndexMorph, fulltextMorphLemma) |
''' |
''' |
|
|
logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
Line 193 class documentViewer(Folder):
|
Line 197 class documentViewer(Folder):
|
# 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_dict" |
else: |
else: |
viewMode="images" |
viewMode="images" |
|
|
|
pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) |
|
|
pt = getattr(self.template, 'viewer_main') |
pt = getattr(self.template, 'viewer_main') |
return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk)) |
return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk)) |
|
|
Line 220 class documentViewer(Folder):
|
Line 224 class documentViewer(Folder):
|
url = self.template.zogilib.getDLBaseUrl() |
url = self.template.zogilib.getDLBaseUrl() |
return url |
return url |
|
|
|
def getDocumentViewerURL(self): |
|
"""returns the URL of this instance""" |
|
return self.absolute_url() |
|
|
def getStyle(self, idx, selected, style=""): |
def getStyle(self, idx, selected, style=""): |
"""returns a string with the given style and append 'sel' if path == selected.""" |
"""returns a string with the given style and append 'sel' if path == selected.""" |
#logger("documentViewer (getstyle)", logging.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style)) |
#logger("documentViewer (getstyle)", logging.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style)) |
Line 545 class documentViewer(Folder):
|
Line 553 class documentViewer(Folder):
|
if textUrls and (len(textUrls) > 0): |
if textUrls and (len(textUrls) > 0): |
textUrl = getTextFromNode(textUrls[0]) |
textUrl = getTextFromNode(textUrls[0]) |
docinfo['textURLPath'] = textUrl |
docinfo['textURLPath'] = textUrl |
|
if not docinfo['imagePath']: |
|
# text-only, no page images |
|
docinfo = self.getNumPages(docinfo) #im moment einfach auf eins setzen, navigation ueber die thumbs geht natuerlich nicht |
|
|
presentationUrls = dom.xpath("//texttool/presentation") |
presentationUrls = dom.xpath("//texttool/presentation") |
docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag |
docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag |
Line 557 class documentViewer(Folder):
|
Line 568 class documentViewer(Folder):
|
presentationUrl = url.replace('index.meta', presentationPath) |
presentationUrl = url.replace('index.meta', presentationPath) |
else: |
else: |
presentationUrl = url + "/" + presentationPath |
presentationUrl = url + "/" + presentationPath |
docinfo = self.getNumPages(docinfo) #im moment einfach auf eins setzen, navigation ueber die thumbs geht natuerlich nicht |
|
docinfo = self.getBibinfoFromTextToolPresentation(presentationUrl, docinfo=docinfo, dom=dom) |
docinfo = self.getBibinfoFromTextToolPresentation(presentationUrl, docinfo=docinfo, dom=dom) |
|
|
docinfo = self.getAuthinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get access info |
docinfo = self.getAuthinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get access info |
Line 631 class documentViewer(Folder):
|
Line 642 class documentViewer(Folder):
|
self.REQUEST.SESSION['docinfo'] = docinfo |
self.REQUEST.SESSION['docinfo'] = docinfo |
return docinfo |
return docinfo |
|
|
|
|
def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None): |
def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None): |
"""returns pageinfo with the given parameters""" |
"""returns pageinfo with the given parameters""" |
pageinfo = {} |
pageinfo = {} |
Line 653 class documentViewer(Folder):
|
Line 663 class documentViewer(Folder):
|
pageinfo['numgroups'] = int(np / grpsize) |
pageinfo['numgroups'] = int(np / grpsize) |
if np % grpsize > 0: |
if np % grpsize > 0: |
pageinfo['numgroups'] += 1 |
pageinfo['numgroups'] += 1 |
|
|
pageinfo['viewMode'] = viewMode |
pageinfo['viewMode'] = viewMode |
pageinfo['tocMode'] = tocMode |
pageinfo['tocMode'] = tocMode |
pageinfo['query'] = self.REQUEST.get('query',' ') |
pageinfo['query'] = self.REQUEST.get('query',' ') |
pageinfo['queryType'] = self.REQUEST.get('queryType',' ') |
pageinfo['queryType'] = self.REQUEST.get('queryType',' ') |
pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext') |
pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext') |
|
pageinfo['textPN'] = self.REQUEST.get('textPN','1') |
|
pageinfo['highlightQuery'] = self.REQUEST.get('highlightQuery','') |
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', '10') |
pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') |
pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') |
#if 'tocSize_%s'%tocMode in docinfo: |
toc = int (pageinfo['tocPN']) |
# cached toc |
pageinfo['textPages'] =int (toc) |
# pageinfo['tocPN'] = min (int (docinfo['tocSize_%s'%tocMode])/int(pageinfo['tocPageSize']),int(pageinfo['tocPN'])) |
|
|
|
|
if 'tocSize_%s'%tocMode in docinfo: |
|
tocSize = int(docinfo['tocSize_%s'%tocMode]) |
|
tocPageSize = int(pageinfo['tocPageSize']) |
|
# cached toc |
|
if tocSize%tocPageSize>0: |
|
tocPages=tocSize/tocPageSize+1 |
|
else: |
|
tocPages=tocSize/tocPageSize |
|
pageinfo['tocPN'] = min (tocPages,toc) |
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','') |
|
|
return pageinfo |
return pageinfo |
|
|
def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None): |
|
"""get search list""" |
|
docpath = docinfo['textURLPath'] |
|
pagesize = pageinfo['queryPageSize'] |
|
pn = pageinfo['searchPN'] |
|
sn = pageinfo['sn'] |
|
query =pageinfo['query'] |
|
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) |
|
pagedom = Parse(pagexml) |
|
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 |
|
|
|
|
|
#if len(pagedivs) > 0: |
|
# pagenode = pagedom[0] |
|
# return serializeNode(pagenode) |
|
#else: |
|
# return "xaxa" |
|
|
|
def getNumPages(self,docinfo=None): |
|
"""get list of pages from fulltext and put in docinfo""" |
|
xquery = '//pb' |
|
text = self.template.fulltextclient.eval("/mpdl/interface/xquery.xql", "document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) |
|
# TODO: better processing of the page list. do we need the info somewhere else also? |
|
docinfo['numPages'] = text.count("<pb ") |
|
return docinfo |
|
|
|
def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None,): |
|
"""returns single page from fulltext""" |
|
docpath = docinfo['textURLPath'] |
|
if mode == "text_dict": |
|
textmode = "textPollux" |
|
else: |
|
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) |
|
####### |
|
#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=text','mode=texttool') |
|
####### |
|
# post-processing downloaded xml |
|
pagedom = Parse(pagexml) |
|
# plain text mode |
|
if mode == "text": |
|
# first div contains text |
|
pagedivs = pagedom.xpath("/div") |
|
#queryResultPage |
|
if len(pagedivs) > 0: |
|
pagenode = pagedivs[0] |
|
return serializeNode(pagenode) |
|
|
|
# text-with-links mode |
|
if mode == "text_dict": |
|
# first div contains text |
|
pagedivs = pagedom.xpath("/div") |
|
if len(pagedivs) > 0: |
|
pagenode = pagedivs[0] |
|
# check all a-tags |
|
links = pagenode.xpath("//a") |
|
for l in links: |
|
hrefNode = l.getAttributeNodeNS(None, u"href") |
|
if hrefNode: |
|
# is link with href |
|
href = hrefNode.nodeValue |
|
if href.startswith('lt/lex.xql'): |
|
# is pollux link |
|
selfurl = self.absolute_url() |
|
# change href |
|
hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl) |
|
# add target |
|
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 "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): |
|
"""loads table of contents and stores in docinfo""" |
|
logging.debug("documentViewer (gettoc) mode: %s"%(mode)) |
|
if 'tocSize_%s'%mode in docinfo: |
|
# cached toc |
|
return docinfo |
|
docpath = docinfo['textURLPath'] |
|
# we need to set a result set size |
|
pagesize = 1000 |
|
pn = 1 |
|
if mode == "text": |
|
queryType = "toc" |
|
else: |
|
queryType = mode |
|
# number of entries in toc |
|
tocSize = 0 |
|
tocDiv = None |
|
pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False) |
|
# 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])) |
|
# div contains text |
|
#pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
|
#if len(pagedivs) > 0: |
|
# tocDiv = pagedivs[0] |
|
|
|
docinfo['tocSize_%s'%mode] = tocSize |
|
#docinfo['tocDiv_%s'%mode] = tocDiv |
|
return docinfo |
|
|
|
def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None): |
|
"""returns single page from the table of contents""" |
|
# TODO: this should use the cached TOC |
|
if mode == "text": |
|
queryType = "toc" |
|
else: |
|
queryType = mode |
|
docpath = docinfo['textURLPath'] |
|
pagesize = pageinfo['tocPageSize'] |
|
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) |
|
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 |
|
#pagedom = Parse(text) |
|
# div contains text |
|
#pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
|
#if len(pagedivs) > 0: |
|
# pagenode = pagedivs[0] |
|
# return serializeNode(pagenode) |
|
#else: |
|
# 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): |
"""init document viewer""" |
"""init document viewer""" |
self.title=title |
self.title=title |
Line 872 class documentViewer(Folder):
|
Line 699 class documentViewer(Folder):
|
if RESPONSE is not None: |
if RESPONSE is not None: |
RESPONSE.redirect('manage_main') |
RESPONSE.redirect('manage_main') |
|
|
|
|
|
|
def manage_AddDocumentViewerForm(self): |
def manage_AddDocumentViewerForm(self): |
"""add the viewer form""" |
"""add the viewer form""" |
pt=PageTemplateFile('zpt/addDocumentViewer', globals()).__of__(self) |
pt=PageTemplateFile('zpt/addDocumentViewer', globals()).__of__(self) |
Line 887 def manage_AddDocumentViewer(self,id,ima
|
Line 712 def manage_AddDocumentViewer(self,id,ima
|
if RESPONSE is not None: |
if RESPONSE is not None: |
RESPONSE.redirect('manage_main') |
RESPONSE.redirect('manage_main') |
|
|
|
|
## |
|
## DocumentViewerTemplate class |
## DocumentViewerTemplate class |
## |
|
class DocumentViewerTemplate(ZopePageTemplate): |
class DocumentViewerTemplate(ZopePageTemplate): |
"""Template for document viewer""" |
"""Template for document viewer""" |
meta_type="DocumentViewer Template" |
meta_type="DocumentViewer Template" |