version 1.21, 2007/01/11 20:27:17
|
version 1.49, 2010/05/05 15:42:56
|
Line 1
|
Line 1
|
|
|
|
|
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 |
Line 11 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 |
|
|
import Ft.Xml.XPath |
from xml.dom.minidom import parse, parseString |
|
|
|
|
|
|
|
import Ft.Xml.XPath |
|
import cStringIO |
|
import xmlrpclib |
import os.path |
import os.path |
import sys |
import sys |
import cgi |
import cgi |
import urllib |
import urllib |
import logging |
import logging |
import zLOG |
import math |
|
|
import urlparse |
import urlparse |
|
from types import * |
|
|
|
def logger(txt,method,txt2): |
|
"""logging""" |
|
logging.info(txt+ txt2) |
|
|
|
|
def getInt(number, default=0): |
def getInt(number, default=0): |
"""returns always an int (0 in case of problems)""" |
"""returns always an int (0 in case of problems)""" |
try: |
try: |
return int(number) |
return int(number) |
except: |
except: |
return default |
return int(default) |
|
|
def getTextFromNode(nodename): |
def getTextFromNode(nodename): |
"""get the cdata content of a node""" |
"""get the cdata content of a node""" |
Line 39 def getTextFromNode(nodename):
|
Line 50 def getTextFromNode(nodename):
|
rc = rc + node.data |
rc = rc + node.data |
return rc |
return rc |
|
|
|
def serializeNode(node, encoding='utf-8'): |
|
"""returns a string containing node as XML""" |
|
buf = cStringIO.StringIO() |
|
Print(node, stream=buf, encoding=encoding) |
|
s = buf.getvalue() |
|
buf.close() |
|
return s |
|
|
|
|
def getParentDir(path): |
def getParentDir(path): |
"""returns pathname shortened by one""" |
"""returns pathname shortened by one""" |
Line 71 class documentViewer(Folder):
|
Line 90 class documentViewer(Folder):
|
|
|
# templates and forms |
# templates and forms |
viewer_main = PageTemplateFile('zpt/viewer_main', globals()) |
viewer_main = PageTemplateFile('zpt/viewer_main', globals()) |
thumbs_main = PageTemplateFile('zpt/thumbs_main', globals()) |
toc_thumbs = PageTemplateFile('zpt/toc_thumbs', globals()) |
image_main = PageTemplateFile('zpt/image_main', globals()) |
toc_text = PageTemplateFile('zpt/toc_text', globals()) |
|
toc_figures = PageTemplateFile('zpt/toc_figures', globals()) |
|
page_main_images = PageTemplateFile('zpt/page_main_images', globals()) |
|
page_main_text = PageTemplateFile('zpt/page_main_text', 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()) |
|
|
|
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()) |
|
|
|
|
def __init__(self,id,imageViewerUrl,textViewerUrl=None,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=10,authgroups="mpiwg"): |
def __init__(self,id,imageScalerUrl=None,textServerName=None,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=5,authgroups="mpiwg"): |
"""init document viewer""" |
"""init document viewer""" |
self.id=id |
self.id=id |
self.title=title |
self.title=title |
self.imageViewerUrl=imageViewerUrl |
|
self.textViewerUrl=textViewerUrl |
|
|
|
if not digilibBaseUrl: |
|
self.digilibBaseUrl = self.findDigilibUrl() |
|
else: |
|
self.digilibBaseUrl = digilibBaseUrl |
|
self.thumbcols = thumbcols |
self.thumbcols = thumbcols |
self.thumbrows = thumbrows |
self.thumbrows = thumbrows |
# authgroups is list of authorized groups (delimited by ,) |
# authgroups is list of authorized groups (delimited by ,) |
self.authgroups = [s.strip().lower() for s in authgroups.split(',')] |
self.authgroups = [s.strip().lower() for s in authgroups.split(',')] |
# add template folder so we can always use template.something |
# create template folder so we can always use template.something |
self.manage_addFolder('template') |
|
|
|
|
templateFolder = Folder('template') |
|
#self['template'] = templateFolder # Zope-2.12 style |
|
self._setObject('template',templateFolder) # old style |
|
try: |
|
from Products.XMLRpcTools.XMLRpcTools import XMLRpcServerProxy |
|
xmlRpcClient = XMLRpcServerProxy(id='fulltextclient', serverUrl=textServerName, use_xmlrpc=False) |
|
#templateFolder['fulltextclient'] = xmlRpcClient |
|
templateFolder._setObject('fulltextclient',xmlRpcClient) |
|
except Exception, e: |
|
logging.error("Unable to create XMLRpcTools for fulltextclient: "+str(e)) |
|
try: |
|
from Products.zogiLib.zogiLib import zogiLib |
|
zogilib = zogiLib(id="zogilib", title="zogilib for docuviewer", dlServerURL=imageScalerUrl, layout="book") |
|
#templateFolder['zogilib'] = zogilib |
|
templateFolder._setObject('zogilib',zogilib) |
|
except Exception, e: |
|
logging.error("Unable to create zogiLib for zogilib: "+str(e)) |
|
|
security.declareProtected('View','index_html') |
|
def index_html(self,mode,url,viewMode="auto",start=None,pn=1): |
security.declareProtected('View','thumbs_rss') |
|
def thumbs_rss(self,mode,url,viewMode="auto",start=None,pn=1): |
''' |
''' |
view it |
view it |
@param mode: defines which type of document is behind url (text,images or auto) |
@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 images |
@param viewMode: if images display images, if text display text, default is images (text,images or auto) |
|
|
''' |
''' |
|
logging.debug("HHHHHHHHHHHHHH:load the rss") |
zLOG.LOG("documentViewer (index)", zLOG.INFO, "mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
logger("documentViewer (index)", logging.INFO, "mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
|
|
if not hasattr(self, 'template'): |
if not hasattr(self, 'template'): |
# create template folder if it doesn't exist |
# create template folder if it doesn't exist |
Line 120 class documentViewer(Folder):
|
Line 156 class documentViewer(Folder):
|
|
|
docinfo = self.getDocinfo(mode=mode,url=url) |
docinfo = self.getDocinfo(mode=mode,url=url) |
pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) |
pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) |
pt = getattr(self.template, 'viewer_main') |
pt = getattr(self.template, 'thumbs_main_rss') |
|
|
if viewMode=="auto": # automodus gewaehlt |
if viewMode=="auto": # automodus gewaehlt |
if docinfo.get("textURL",'') and self.textViewerUrl: #texturl gesetzt und textViewer konfiguriert |
if docinfo.get("textURL",'') and self.textViewerUrl: #texturl gesetzt und textViewer konfiguriert |
Line 130 class documentViewer(Folder):
|
Line 166 class documentViewer(Folder):
|
|
|
return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode) |
return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode) |
|
|
|
security.declareProtected('View','index_html') |
|
def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None, query=None, querySearch=None): |
|
''' |
|
view it |
|
@param mode: defines how to access the document behind url |
|
@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 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, fulltextMorphLemma) |
|
''' |
|
|
|
logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
|
|
|
if not hasattr(self, 'template'): |
|
# this won't work |
|
logging.error("template folder missing!") |
|
return "ERROR: template folder missing!" |
|
|
|
if not getattr(self, 'digilibBaseUrl', None): |
|
self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" |
|
|
|
docinfo = self.getDocinfo(mode=mode,url=url) |
|
|
|
|
|
if tocMode != "thumbs": |
|
# get table of contents |
|
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 docinfo.get("textURL",''): #texturl gesetzt und textViewer konfiguriert |
|
viewMode="text" |
|
else: |
|
viewMode="images" |
|
|
|
pt = getattr(self.template, 'viewer_main') |
|
return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk)) |
|
|
|
def generateMarks(self,mk): |
|
ret="" |
|
if mk is None: |
|
return "" |
|
if type(mk) is not ListType: |
|
mk=[mk] |
|
for m in mk: |
|
ret+="mk=%s"%m |
|
return ret |
|
|
|
|
|
def findDigilibUrl(self): |
|
"""try to get the digilib URL from zogilib""" |
|
url = self.template.zogilib.getDLBaseUrl() |
|
return url |
|
|
|
def getStyle(self, idx, selected, style=""): |
|
"""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)) |
|
if idx == selected: |
|
return style + 'sel' |
|
else: |
|
return style |
|
|
def getLink(self,param=None,val=None): |
def getLink(self,param=None,val=None): |
"""link to documentviewer with parameter param set to val""" |
"""link to documentviewer with parameter param set to val""" |
Line 141 class documentViewer(Folder):
|
Line 239 class documentViewer(Folder):
|
else: |
else: |
params[param] = str(val) |
params[param] = str(val) |
|
|
|
if params.get("mode", None) == "filepath": #wenn beim erst Aufruf filepath gesetzt wurde aendere das nun zu imagepath |
|
params["mode"] = "imagepath" |
|
params["url"] = getParentDir(params["url"]) |
|
|
# quote values and assemble into query string |
# quote values and assemble into query string |
ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) |
ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) |
url=self.REQUEST['URL1']+"?"+ps |
url=self.REQUEST['URL1']+"?"+ps |
return url |
return url |
|
|
|
def getLinkAmp(self,param=None,val=None): |
def getStyle(self, idx, selected, style=""): |
"""link to documentviewer with parameter param set to val""" |
"""returns a string with the given style and append 'sel' if path == selected.""" |
params=self.REQUEST.form.copy() |
#zLOG.LOG("documentViewer (getstyle)", zLOG.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style)) |
if param is not None: |
if idx == selected: |
if val is None: |
return style + 'sel' |
if params.has_key(param): |
|
del params[param] |
else: |
else: |
return style |
params[param] = str(val) |
|
|
|
# quote values and assemble into query string |
|
logging.info("XYXXXXX: %s"%repr(params.items())) |
|
ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) |
|
url=self.REQUEST['URL1']+"?"+ps |
|
return url |
|
|
|
def getInfo_xml(self,url,mode): |
|
"""returns info about the document as XML""" |
|
|
|
if not self.digilibBaseUrl: |
|
self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" |
|
|
|
docinfo = self.getDocinfo(mode=mode,url=url) |
|
pt = getattr(self.template, 'info_xml') |
|
return pt(docinfo=docinfo) |
|
|
|
|
def isAccessible(self, docinfo): |
def isAccessible(self, docinfo): |
"""returns if access to the resource is granted""" |
"""returns if access to the resource is granted""" |
access = docinfo.get('accessType', None) |
access = docinfo.get('accessType', None) |
zLOG.LOG("documentViewer (accessOK)", zLOG.INFO, "access type %s"%access) |
logger("documentViewer (accessOK)", logging.INFO, "access type %s"%access) |
if access is not None and access == 'free': |
if access is not None and access == 'free': |
zLOG.LOG("documentViewer (accessOK)", zLOG.INFO, "access is free") |
logger("documentViewer (accessOK)", logging.INFO, "access is free") |
return True |
return True |
elif access is None or access in self.authgroups: |
elif access is None or access in self.authgroups: |
# only local access -- only logged in users |
# only local access -- only logged in users |
Line 172 class documentViewer(Folder):
|
Line 291 class documentViewer(Folder):
|
else: |
else: |
return False |
return False |
|
|
zLOG.LOG("documentViewer (accessOK)", zLOG.INFO, "unknown access type %s"%access) |
logger("documentViewer (accessOK)", logging.INFO, "unknown access type %s"%access) |
return False |
return False |
|
|
|
|
def getDirinfoFromDigilib(self,path,docinfo=None): |
def getDirinfoFromDigilib(self,path,docinfo=None,cut=0): |
"""gibt param von dlInfo aus""" |
"""gibt param von dlInfo aus""" |
num_retries = 3 |
num_retries = 3 |
if docinfo is None: |
if docinfo is None: |
docinfo = {} |
docinfo = {} |
|
|
|
for x in range(cut): |
|
|
|
path=getParentDir(path) |
|
|
infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path |
infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path |
|
|
zLOG.LOG("documentViewer (getparamfromdigilib)", zLOG.INFO, "dirInfo from %s"%(infoUrl)) |
logger("documentViewer (getparamfromdigilib)", logging.INFO, "dirInfo from %s"%(infoUrl)) |
|
|
for cnt in range(num_retries): |
for cnt in range(num_retries): |
try: |
try: |
Line 193 class documentViewer(Folder):
|
Line 316 class documentViewer(Folder):
|
dom = Parse(txt) |
dom = Parse(txt) |
break |
break |
except: |
except: |
zLOG.LOG("documentViewer (getdirinfofromdigilib)", zLOG.ERROR, "error reading %s (try %d)"%(infoUrl,cnt)) |
logger("documentViewer (getdirinfofromdigilib)", logging.ERROR, "error reading %s (try %d)"%(infoUrl,cnt)) |
else: |
else: |
raise IOError("Unable to get dir-info from %s"%(infoUrl)) |
raise IOError("Unable to get dir-info from %s"%(infoUrl)) |
|
|
sizes=dom.xpath("//dir/size") |
sizes=dom.xpath("//dir/size") |
zLOG.LOG("documentViewer (getparamfromdigilib)", zLOG.INFO, "dirInfo:size"%sizes) |
logger("documentViewer (getparamfromdigilib)", logging.INFO, "dirInfo:size"%sizes) |
|
|
if sizes: |
if sizes: |
docinfo['numPages'] = int(getTextFromNode(sizes[0])) |
docinfo['numPages'] = int(getTextFromNode(sizes[0])) |
else: |
else: |
docinfo['numPages'] = 0 |
docinfo['numPages'] = 0 |
|
|
|
# TODO: produce and keep list of image names and numbers |
|
|
return docinfo |
return docinfo |
|
|
|
|
Line 222 class documentViewer(Folder):
|
Line 347 class documentViewer(Folder):
|
metaUrl=server+url.replace("/mpiwg/online","") |
metaUrl=server+url.replace("/mpiwg/online","") |
if not metaUrl.endswith("index.meta"): |
if not metaUrl.endswith("index.meta"): |
metaUrl += "/index.meta" |
metaUrl += "/index.meta" |
print metaUrl |
logging.debug("METAURL: %s"%metaUrl) |
for cnt in range(num_retries): |
for cnt in range(num_retries): |
try: |
try: |
# patch dirk encoding fehler treten dann nicht mehr auf |
# patch dirk encoding fehler treten dann nicht mehr auf |
Line 231 class documentViewer(Folder):
|
Line 356 class documentViewer(Folder):
|
dom = Parse(txt) |
dom = Parse(txt) |
break |
break |
except: |
except: |
zLOG.LOG("ERROR documentViewer (getIndexMata)", zLOG.INFO,"%s (%s)"%sys.exc_info()[0:2]) |
logger("ERROR documentViewer (getIndexMeta)", logging.INFO,"%s (%s)"%sys.exc_info()[0:2]) |
|
|
if dom is None: |
if dom is None: |
raise IOError("Unable to read index meta from %s"%(url)) |
raise IOError("Unable to read index meta from %s"%(url)) |
Line 251 class documentViewer(Folder):
|
Line 376 class documentViewer(Folder):
|
server=self.digilibBaseUrl+"/servlet/Texter?fn=" |
server=self.digilibBaseUrl+"/servlet/Texter?fn=" |
metaUrl=server+url.replace("/mpiwg/online","") |
metaUrl=server+url.replace("/mpiwg/online","") |
|
|
|
|
for cnt in range(num_retries): |
for cnt in range(num_retries): |
try: |
try: |
# patch dirk encoding fehler treten dann nicht mehr auf |
# patch dirk encoding fehler treten dann nicht mehr auf |
Line 260 class documentViewer(Folder):
|
Line 384 class documentViewer(Folder):
|
dom = Parse(txt) |
dom = Parse(txt) |
break |
break |
except: |
except: |
zLOG.LOG("ERROR documentViewer (getPresentationInfoXML)", zLOG.INFO,"%s (%s)"%sys.exc_info()[0:2]) |
logger("ERROR documentViewer (getPresentationInfoXML)", logging.INFO,"%s (%s)"%sys.exc_info()[0:2]) |
|
|
if dom is None: |
if dom is None: |
raise IOError("Unable to read infoXMLfrom %s"%(url)) |
raise IOError("Unable to read infoXMLfrom %s"%(url)) |
Line 268 class documentViewer(Folder):
|
Line 392 class documentViewer(Folder):
|
return dom |
return dom |
|
|
|
|
def getAuthinfoFromIndexMeta(self,path,docinfo=None,dom=None): |
def getAuthinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): |
"""gets authorization info from the index.meta file at path or given by dom""" |
"""gets authorization info from the index.meta file at path or given by dom""" |
zLOG.LOG("documentViewer (getauthinfofromindexmeta)", zLOG.INFO,"path: %s"%(path)) |
logger("documentViewer (getauthinfofromindexmeta)", logging.INFO,"path: %s"%(path)) |
|
|
access = None |
access = None |
|
|
Line 278 class documentViewer(Folder):
|
Line 402 class documentViewer(Folder):
|
docinfo = {} |
docinfo = {} |
|
|
if dom is None: |
if dom is None: |
dom = self.getIndexMeta(getParentDir(path)) |
for x in range(cut): |
|
path=getParentDir(path) |
|
dom = self.getIndexMeta(path) |
|
|
acctype = dom.xpath("//access-conditions/access/@type") |
acctype = dom.xpath("//access-conditions/access/@type") |
if acctype and (len(acctype)>0): |
if acctype and (len(acctype)>0): |
Line 290 class documentViewer(Folder):
|
Line 416 class documentViewer(Folder):
|
return docinfo |
return docinfo |
|
|
|
|
def getBibinfoFromIndexMeta(self,path,docinfo=None,dom=None): |
def getBibinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): |
"""gets bibliographical info from the index.meta file at path or given by dom""" |
"""gets bibliographical info from the index.meta file at path or given by dom""" |
zLOG.LOG("documentViewer (getbibinfofromindexmeta)", zLOG.INFO,"path: %s"%(path)) |
logging.debug("documentViewer (getbibinfofromindexmeta) path: %s"%(path)) |
|
|
if docinfo is None: |
if docinfo is None: |
docinfo = {} |
docinfo = {} |
|
|
if dom is None: |
if dom is None: |
dom = self.getIndexMeta(getParentDir(path)) |
for x in range(cut): |
|
path=getParentDir(path) |
|
dom = self.getIndexMeta(path) |
|
|
|
logging.debug("documentViewer (getbibinfofromindexmeta cutted) path: %s"%(path)) |
|
# put in all raw bib fields as dict "bib" |
|
bib = dom.xpath("//bib/*") |
|
if bib and len(bib)>0: |
|
bibinfo = {} |
|
for e in bib: |
|
bibinfo[e.localName] = getTextFromNode(e) |
|
docinfo['bib'] = bibinfo |
|
|
|
# extract some fields (author, title, year) according to their mapping |
metaData=self.metadata.main.meta.bib |
metaData=self.metadata.main.meta.bib |
bibtype=dom.xpath("//bib/@type") |
bibtype=dom.xpath("//bib/@type") |
if bibtype and (len(bibtype)>0): |
if bibtype and (len(bibtype)>0): |
bibtype=bibtype[0].value |
bibtype=bibtype[0].value |
else: |
else: |
bibtype="generic" |
bibtype="generic" |
|
|
bibtype=bibtype.replace("-"," ") # wrong typesiin index meta "-" instead of " " (not wrong! ROC) |
bibtype=bibtype.replace("-"," ") # wrong typesiin index meta "-" instead of " " (not wrong! ROC) |
|
docinfo['bib_type'] = bibtype |
bibmap=metaData.generateMappingForType(bibtype) |
bibmap=metaData.generateMappingForType(bibtype) |
#print "bibmap: ", bibmap, " for: ", bibtype |
|
# if there is no mapping bibmap is empty (mapping sometimes has empty fields) |
# if there is no mapping bibmap is empty (mapping sometimes has empty fields) |
if len(bibmap) > 0 and len(bibmap['author'][0]) > 0: |
if len(bibmap) > 0 and len(bibmap['author'][0]) > 0: |
|
try: |
docinfo['author']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['author'][0])[0]) |
docinfo['author']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['author'][0])[0]) |
|
except: pass |
|
try: |
docinfo['title']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['title'][0])[0]) |
docinfo['title']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['title'][0])[0]) |
|
except: pass |
|
try: |
docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['year'][0])[0]) |
docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['year'][0])[0]) |
|
except: pass |
|
logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype) |
|
try: |
|
docinfo['lang']=getTextFromNode(dom.xpath("//bib/lang")[0]) |
|
except: |
|
docinfo['lang']='' |
|
|
return docinfo |
return docinfo |
|
|
|
|
def getDocinfoFromTextTool(self,url,dom=None,docinfo=None): |
def getDocinfoFromTextTool(self,url,dom=None,docinfo=None): |
"""parse texttool tag in index meta""" |
"""parse texttool tag in index meta""" |
zLOG.LOG("documentViewer (getdocinfofromtexttool)", zLOG.INFO,"url: %s"%(url)) |
logger("documentViewer (getdocinfofromtexttool)", logging.INFO, "url: %s" % (url)) |
if docinfo is None: |
if docinfo is None: |
docinfo = {} |
docinfo = {} |
|
if docinfo.get('lang', None) is None: |
|
docinfo['lang'] = '' # default keine Sprache gesetzt |
if dom is None: |
if dom is None: |
dom = self.getIndexMeta(url) |
dom = self.getIndexMeta(url) |
|
|
Line 334 class documentViewer(Folder):
|
Line 485 class documentViewer(Folder):
|
if archiveNames and (len(archiveNames)>0): |
if archiveNames and (len(archiveNames)>0): |
archiveName=getTextFromNode(archiveNames[0]) |
archiveName=getTextFromNode(archiveNames[0]) |
else: |
else: |
zLOG.LOG("documentViewer (getdocinfofromtexttool)", zLOG.WARNING,"resource/name missing in: %s"%(url)) |
logger("documentViewer (getdocinfofromtexttool)", logging.WARNING, "resource/name missing in: %s" % (url)) |
|
|
archivePaths=dom.xpath("//resource/archive-path") |
archivePaths=dom.xpath("//resource/archive-path") |
if archivePaths and (len(archivePaths)>0): |
if archivePaths and (len(archivePaths)>0): |
Line 346 class documentViewer(Folder):
|
Line 497 class documentViewer(Folder):
|
archivePath += "/" + archiveName |
archivePath += "/" + archiveName |
else: |
else: |
# try to get archive-path from url |
# try to get archive-path from url |
zLOG.LOG("documentViewer (getdocinfofromtexttool)", zLOG.WARNING,"resource/archive-path missing in: %s"%(url)) |
logger("documentViewer (getdocinfofromtexttool)", logging.WARNING, "resource/archive-path missing in: %s" % (url)) |
if (not url.startswith('http')): |
if (not url.startswith('http')): |
archivePath = url.replace('index.meta', '') |
archivePath = url.replace('index.meta', '') |
|
|
Line 357 class documentViewer(Folder):
|
Line 508 class documentViewer(Folder):
|
imageDirs=dom.xpath("//texttool/image") |
imageDirs=dom.xpath("//texttool/image") |
if imageDirs and (len(imageDirs)>0): |
if imageDirs and (len(imageDirs)>0): |
imageDir=getTextFromNode(imageDirs[0]) |
imageDir=getTextFromNode(imageDirs[0]) |
|
|
else: |
else: |
# we balk with no image tag |
# we balk with no image tag / not necessary anymore because textmode is now standard |
raise IOError("No text-tool info in %s"%(url)) |
#raise IOError("No text-tool info in %s"%(url)) |
|
imageDir = "" |
|
#xquery="//pb" |
|
docinfo['imagePath'] = "" # keine Bilder |
|
docinfo['imageURL'] = "" |
|
|
if imageDir and archivePath: |
if imageDir and archivePath: |
#print "image: ", imageDir, " archivepath: ", archivePath |
#print "image: ", imageDir, " archivepath: ", archivePath |
Line 367 class documentViewer(Folder):
|
Line 523 class documentViewer(Folder):
|
imageDir=imageDir.replace("/mpiwg/online",'') |
imageDir=imageDir.replace("/mpiwg/online",'') |
docinfo=self.getDirinfoFromDigilib(imageDir,docinfo=docinfo) |
docinfo=self.getDirinfoFromDigilib(imageDir,docinfo=docinfo) |
docinfo['imagePath'] = imageDir |
docinfo['imagePath'] = imageDir |
|
|
docinfo['imageURL'] = self.digilibBaseUrl+"/servlet/Scaler?fn="+imageDir |
docinfo['imageURL'] = self.digilibBaseUrl+"/servlet/Scaler?fn="+imageDir |
|
|
viewerUrls=dom.xpath("//texttool/digiliburlprefix") |
viewerUrls=dom.xpath("//texttool/digiliburlprefix") |
Line 379 class documentViewer(Folder):
|
Line 536 class documentViewer(Folder):
|
textUrl=getTextFromNode(textUrls[0]) |
textUrl=getTextFromNode(textUrls[0]) |
if urlparse.urlparse(textUrl)[0]=="": #keine url |
if urlparse.urlparse(textUrl)[0]=="": #keine url |
textUrl=os.path.join(archivePath,textUrl) |
textUrl=os.path.join(archivePath,textUrl) |
|
# fix URLs starting with /mpiwg/online |
|
if textUrl.startswith("/mpiwg/online"): |
|
textUrl = textUrl.replace("/mpiwg/online", '', 1) |
|
|
docinfo['textURL'] = textUrl |
docinfo['textURL'] = textUrl |
|
|
|
textUrls = dom.xpath("//texttool/text-url-path") |
|
if textUrls and (len(textUrls) > 0): |
|
textUrl = getTextFromNode(textUrls[0]) |
|
docinfo['textURLPath'] = textUrl |
|
|
presentationUrls=dom.xpath("//texttool/presentation") |
presentationUrls=dom.xpath("//texttool/presentation") |
if presentationUrls and (len(presentationUrls)>0): |
docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag |
# presentation url ergiebt sich ersetzen von index.meta in der url der fŸr die Metadaten |
|
# durch den relativen Pfad auf die presentation infos |
|
presentationUrl=url.replace('index.meta',getTextFromNode(presentationUrls[0])) |
|
|
|
docinfo = self.getBibinfoFromTextToolPresentation(presentationUrl,docinfo=docinfo,dom=dom) |
if presentationUrls and (len(presentationUrls) > 0): # ueberschreibe diese durch presentation informationen |
|
# presentation url ergiebt sich ersetzen von index.meta in der url der fuer die Metadaten |
|
# durch den relativen Pfad auf die presentation infos |
|
presentationPath = getTextFromNode(presentationUrls[0]) |
|
if url.endswith("index.meta"): |
|
presentationUrl = url.replace('index.meta', presentationPath) |
else: |
else: |
docinfo = self.getBibinfoFromIndexMeta(url,docinfo=docinfo,dom=dom) |
presentationUrl = url + "/" + presentationPath |
docinfo = self.getAuthinfoFromIndexMeta(url,docinfo=docinfo,dom=dom) |
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.getAuthinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get access info |
|
|
return docinfo |
return docinfo |
|
|
|
|
Line 400 class documentViewer(Folder):
|
Line 570 class documentViewer(Folder):
|
"""gets the bibliographical information from the preseantion entry in texttools |
"""gets the bibliographical information from the preseantion entry in texttools |
""" |
""" |
dom=self.getPresentationInfoXML(url) |
dom=self.getPresentationInfoXML(url) |
|
try: |
docinfo['author']=getTextFromNode(dom.xpath("//author")[0]) |
docinfo['author']=getTextFromNode(dom.xpath("//author")[0]) |
|
except: |
|
pass |
|
try: |
docinfo['title']=getTextFromNode(dom.xpath("//title")[0]) |
docinfo['title']=getTextFromNode(dom.xpath("//title")[0]) |
|
except: |
|
pass |
|
try: |
docinfo['year']=getTextFromNode(dom.xpath("//date")[0]) |
docinfo['year']=getTextFromNode(dom.xpath("//date")[0]) |
|
except: |
|
pass |
return docinfo |
return docinfo |
|
|
def getDocinfoFromImagePath(self,path,docinfo=None): |
def getDocinfoFromImagePath(self,path,docinfo=None,cut=0): |
"""path ist the path to the images it assumes that the index.meta file is one level higher.""" |
"""path ist the path to the images it assumes that the index.meta file is one level higher.""" |
zLOG.LOG("documentViewer (getdocinfofromimagepath)", zLOG.INFO,"path: %s"%(path)) |
logger("documentViewer (getdocinfofromimagepath)", logging.INFO,"path: %s"%(path)) |
if docinfo is None: |
if docinfo is None: |
docinfo = {} |
docinfo = {} |
path=path.replace("/mpiwg/online","") |
path=path.replace("/mpiwg/online","") |
docinfo['imagePath'] = path |
docinfo['imagePath'] = path |
docinfo=self.getDirinfoFromDigilib(path,docinfo=docinfo) |
docinfo=self.getDirinfoFromDigilib(path,docinfo=docinfo,cut=cut) |
|
|
|
pathorig=path |
|
for x in range(cut): |
|
path=getParentDir(path) |
|
logging.error("PATH:"+path) |
imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path |
imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path |
docinfo['imageURL'] = imageUrl |
docinfo['imageURL'] = imageUrl |
|
|
docinfo = self.getBibinfoFromIndexMeta(path,docinfo=docinfo) |
#path ist the path to the images it assumes that the index.meta file is one level higher. |
docinfo = self.getAuthinfoFromIndexMeta(path,docinfo=docinfo) |
docinfo = self.getBibinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1) |
|
docinfo = self.getAuthinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1) |
return docinfo |
return docinfo |
|
|
|
|
def getDocinfo(self, mode, url): |
def getDocinfo(self, mode, url): |
"""returns docinfo depending on mode""" |
"""returns docinfo depending on mode""" |
zLOG.LOG("documentViewer (getdocinfo)", zLOG.INFO,"mode: %s, url: %s"%(mode,url)) |
logger("documentViewer (getdocinfo)", logging.INFO,"mode: %s, url: %s"%(mode,url)) |
# look for cached docinfo in session |
# look for cached docinfo in session |
if self.REQUEST.SESSION.has_key('docinfo'): |
if self.REQUEST.SESSION.has_key('docinfo'): |
docinfo = self.REQUEST.SESSION['docinfo'] |
docinfo = self.REQUEST.SESSION['docinfo'] |
# check if its still current |
# check if its still current |
if docinfo is not None and docinfo.get('mode') == mode and docinfo.get('url') == url: |
if docinfo is not None and docinfo.get('mode') == mode and docinfo.get('url') == url: |
zLOG.LOG("documentViewer (getdocinfo)", zLOG.INFO,"docinfo in session: %s"%docinfo) |
logger("documentViewer (getdocinfo)", logging.INFO,"docinfo in session: %s"%docinfo) |
return docinfo |
return docinfo |
# new docinfo |
# new docinfo |
docinfo = {'mode': mode, 'url': url} |
docinfo = {'mode': mode, 'url': url} |
Line 437 class documentViewer(Folder):
|
Line 622 class documentViewer(Folder):
|
docinfo = self.getDocinfoFromTextTool(url, docinfo=docinfo) |
docinfo = self.getDocinfoFromTextTool(url, docinfo=docinfo) |
elif mode=="imagepath": |
elif mode=="imagepath": |
docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo) |
docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo) |
|
elif mode=="filepath": |
|
docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo,cut=1) |
else: |
else: |
zLOG.LOG("documentViewer (getdocinfo)", zLOG.ERROR,"unknown mode!") |
logger("documentViewer (getdocinfo)", logging.ERROR,"unknown mode!") |
raise ValueError("Unknown mode %s"%(mode)) |
raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode)) |
|
|
zLOG.LOG("documentViewer (getdocinfo)", zLOG.INFO,"docinfo: %s"%docinfo) |
logger("documentViewer (getdocinfo)", logging.INFO,"docinfo: %s"%docinfo) |
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): |
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 = {} |
current = getInt(current) |
current = getInt(current) |
Line 457 class documentViewer(Folder):
|
Line 644 class documentViewer(Folder):
|
pageinfo['cols'] = cols |
pageinfo['cols'] = cols |
grpsize = cols * rows |
grpsize = cols * rows |
pageinfo['groupsize'] = grpsize |
pageinfo['groupsize'] = grpsize |
start = getInt(start, default=(int(current / grpsize) * grpsize +1)) |
start = getInt(start, default=(math.ceil(float(current)/float(grpsize))*grpsize-(grpsize-1))) |
|
# int(current / grpsize) * grpsize +1)) |
pageinfo['start'] = start |
pageinfo['start'] = start |
pageinfo['end'] = start + grpsize |
pageinfo['end'] = start + grpsize |
if docinfo is not None: |
if (docinfo is not None) and ('numPages' in docinfo): |
np = int(docinfo['numPages']) |
np = int(docinfo['numPages']) |
pageinfo['end'] = min(pageinfo['end'], np) |
pageinfo['end'] = min(pageinfo['end'], np) |
pageinfo['numgroups'] = int(np / grpsize) |
pageinfo['numgroups'] = int(np / grpsize) |
if np % grpsize > 0: |
if np % grpsize > 0: |
pageinfo['numgroups'] += 1 |
pageinfo['numgroups'] += 1 |
|
|
return pageinfo |
|
|
|
def text(self,mode,url,pn): |
|
"""give text""" |
|
if mode=="texttool": #index.meta with texttool information |
|
(viewerUrl,imagepath,textpath)=parseUrlTextTool(url) |
|
|
|
#print textpath |
|
try: |
|
dom = NonvalidatingReader.parseUri(textpath) |
|
except: |
|
return None |
|
|
|
list=[] |
|
nodes=dom.xpath("//pb") |
|
|
|
node=nodes[int(pn)-1] |
|
|
|
p=node |
pageinfo['viewMode'] = viewMode |
|
pageinfo['tocMode'] = tocMode |
while p.tagName!="p": |
pageinfo['query'] = self.REQUEST.get('query',' ') |
p=p.parentNode |
pageinfo['queryType'] = self.REQUEST.get('queryType',' ') |
|
pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext') |
|
|
endNode=nodes[int(pn)] |
pageinfo['textPN'] = self.REQUEST.get('textPN','1') |
|
|
|
pageinfo['tocPageSize'] = self.REQUEST.get('tocPageSize', '30') |
|
pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '20') |
|
pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') |
|
toc = int (pageinfo['tocPN']) |
|
pageinfo['textPages'] =int (toc) |
|
|
|
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) |
|
|
e=endNode |
pageinfo['searchPN'] =self.REQUEST.get('searchPN','1') |
|
pageinfo['sn'] =self.REQUEST.get('sn','1') |
|
|
while e.tagName!="p": |
return pageinfo |
e=e.parentNode |
|
|
|
|
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'] |
|
selfurl = self.absolute_url() |
|
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"%(docpath, 'text', queryType, query, pagesize, pn, sn, viewMode) ,outputUnicode=False) |
|
pagexml = page.replace('?document=/echo/la/Benedetti_1585.xml','?url=/mpiwg/online/permanent/library/163127KK') |
|
#hrefNode.nodeValue =pagexml.replace('mode=text','%s&mode=texttool'%selfurl) |
|
pagedom = Parse(pagexml) |
|
#logging.debug("documentViewer (gettoc) pagedom: %s"%(pagedom)) |
|
if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): |
|
pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
|
if len(pagedivs)>0: |
|
pagenode=pagedivs[0] |
|
links=pagenode.xpath("//a") |
|
for l in links: |
|
hrefNode = l.getAttributeNodeNS(None, u"href") |
|
if hrefNode: |
|
href = hrefNode.nodeValue |
|
if href.startswith('page-fragment.xql'): |
|
selfurl = self.absolute_url() |
|
#l.setAttributeNS(None, "span class = 'hit highlight'", "background-color: #77DD77;") |
|
pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s'%(viewMode,queryType,query,pagesize,pn,tocMode,pn,tocPN)) |
|
hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
|
l.setAttributeNS(None, "span class = 'hit'", "background-color: #77DD77;") |
|
return serializeNode(pagenode) |
|
|
|
if (queryType=="fulltextMorph"): |
|
pagedivs = pagedom.xpath("//div[@class='queryResult']") |
|
|
|
if len(pagedivs)>0: |
|
pagenode=pagedivs[0] |
|
links=pagenode.xpath("//a") |
|
for l in links: |
|
hrefNode = l.getAttributeNodeNS(None, u"href") |
|
if hrefNode: |
|
href = hrefNode.nodeValue |
|
if href.startswith('page-fragment.xql'): |
|
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'%(viewMode,queryType,query,pagesize,pn,tocMode,pn,tocPN)) |
|
hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
|
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') |
|
l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
|
l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
|
return serializeNode(pagenode) |
|
|
|
if (queryType=="ftIndex")or(queryType=="ftIndexMorph"): |
|
pagedivs= pagedom.xpath("//div[@class='queryResultPage']") |
|
if len(pagedivs)>0: |
|
pagenode=pagedivs[0] |
|
links=pagenode.xpath("//a") |
|
for l in links: |
|
hrefNode = l.getAttributeNodeNS(None, u"href") |
|
if hrefNode: |
|
href = hrefNode.nodeValue |
|
hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn)) |
|
|
|
if href.startswith('../lt/lex.xql'): |
|
selfurl = self.absolute_url() |
|
hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl) |
|
l.setAttributeNS(None, 'target', '_blank') |
|
l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
|
l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
|
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') |
|
l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
|
l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
|
return serializeNode(pagenode) |
|
return "xexe" |
|
|
|
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 |
|
|
next=node.parentNode |
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) |
|
# 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) |
|
if mode == "xml": |
|
# first div contains text |
|
pagedivs = pagedom.xpath("/div") |
|
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') |
|
l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
|
l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
|
|
|
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') |
|
l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") |
|
l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
|
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): |
|
"""simular words 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] |
|
|
#sammle s |
docinfo['tocSize_%s'%mode] = tocSize |
while next and (next!=endNode.parentNode): |
#docinfo['tocDiv_%s'%mode] = tocDiv |
list.append(next) |
return docinfo |
next=next.nextSibling |
|
list.append(endNode.parentNode) |
|
|
|
if p==e:# beide im selben paragraphen |
def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None): |
pass |
"""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: |
# else: |
# next=p |
# return "No TOC!" |
# while next!=e: |
|
# print next,e |
|
# list.append(next) |
|
# next=next.nextSibling |
|
# |
|
# for x in list: |
|
# PrettyPrint(x) |
|
# |
|
# return list |
|
# |
|
|
|
def findDigilibUrl(self): |
|
"""try to get the digilib URL from zogilib""" |
|
url = self.imageViewerUrl[:-1] + "/getScalerUrl" |
|
#print urlparse.urlparse(url)[0] |
|
#print urlparse.urljoin(self.absolute_url(),url) |
|
logging.info("finddigiliburl: %s"%urlparse.urlparse(url)[0]) |
|
logging.info("finddigiliburl: %s"%urlparse.urljoin(self.absolute_url(),url)) |
|
|
|
try: |
|
if urlparse.urlparse(url)[0]=='': #relative path |
|
url=urlparse.urljoin(self.absolute_url()+"/",url) |
|
|
|
scaler = urlopen(url).read() |
def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None): |
return scaler.replace("/servlet/Scaler?", "") |
|
except: |
|
return None |
|
|
|
def changeDocumentViewer(self,imageViewerUrl,textViewerUrl,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=10,authgroups='mpiwg',RESPONSE=None): |
|
"""init document viewer""" |
"""init document viewer""" |
self.title=title |
self.title=title |
self.imageViewerUrl=imageViewerUrl |
|
self.textViewerUrl=textViewerUrl |
|
self.digilibBaseUrl = digilibBaseUrl |
self.digilibBaseUrl = digilibBaseUrl |
self.thumbrows = thumbrows |
self.thumbrows = thumbrows |
self.thumbcols = thumbcols |
self.thumbcols = thumbcols |
Line 554 class documentViewer(Folder):
|
Line 944 class documentViewer(Folder):
|
|
|
|
|
|
|
|
|
# security.declareProtected('View management screens','renameImageForm') |
|
|
|
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) |
return pt() |
return pt() |
|
|
def manage_AddDocumentViewer(self,id,imageViewerUrl="",textViewerUrl="",title="",RESPONSE=None): |
def manage_AddDocumentViewer(self,id,imageScalerUrl="",textServerName="",title="",RESPONSE=None): |
"""add the viewer""" |
"""add the viewer""" |
newObj=documentViewer(id,imageViewerUrl,title=title,textViewerUrl=textViewerUrl) |
newObj=documentViewer(id,imageScalerUrl=imageScalerUrl,title=title,textServerName=textServerName) |
self._setObject(id,newObj) |
self._setObject(id,newObj) |
|
|
if RESPONSE is not None: |
if RESPONSE is not None: |
Line 590 def manage_addDocumentViewerTemplate(sel
|
Line 977 def manage_addDocumentViewerTemplate(sel
|
|
|
self._setObject(id, DocumentViewerTemplate(id)) |
self._setObject(id, DocumentViewerTemplate(id)) |
ob = getattr(self, id) |
ob = getattr(self, id) |
ob.pt_edit(open(os.path.join(package_home(globals()),'zpt/viewer_main.zpt')).read(),None) |
txt=file(os.path.join(package_home(globals()),'zpt/viewer_main.zpt'),'r').read() |
|
logging.info("txt %s:"%txt) |
|
ob.pt_edit(txt,"text/html") |
if title: |
if title: |
ob.pt_setTitle(title) |
ob.pt_setTitle(title) |
try: |
try: |