version 1.189, 2004/11/24 15:17:04
|
version 1.192, 2004/11/29 20:39:08
|
Line 9 class ECHO_externalLink contains informa
|
Line 9 class ECHO_externalLink contains informa
|
|
|
|
|
""" |
""" |
|
from reportlab.pdfgen import canvas |
|
from reportlab.lib.pagesizes import A4 |
|
|
|
|
|
|
import string |
import string |
|
import tempfile |
import re |
import re |
import os |
import os |
import OFS.Image |
import OFS.Image |
Line 222 def manage_addECHO_layoutTemplate(self,
|
Line 228 def manage_addECHO_layoutTemplate(self,
|
REQUEST.RESPONSE.redirect(u+'/manage_main') |
REQUEST.RESPONSE.redirect(u+'/manage_main') |
return '' |
return '' |
|
|
class ECHO_fullText(ZopePageTemplate,ECHO_basis): |
class ECHO_fullText(ZopePageTemplate,ECHO_basis,ECHO_language): |
"""echo fulltext in xml""" |
"""echo fulltext in xml""" |
|
|
meta_type="ECHO_fullText" |
meta_type="ECHO_fullText" |
|
|
def donatusVariant2Lemma(self,nr='1'): |
|
"""analyze by donatus""" |
|
return donatusVariant2Lemma(donatus(self.lemmatize(nr))) |
|
|
|
def tagLex(self,nr="1"): |
|
"""generate Links""" |
|
global retLex |
|
global toggle |
|
|
|
toggle=0 |
|
retLex="" |
|
|
|
lemmatized=self.lemmatize(nr)[0:] |
|
#print "ho",repr(lemmatized) |
|
variants=donatusVariant2Lemma(donatus(lemmatized)) |
|
|
|
def createTag(name,attrs): |
|
global toggle |
|
|
|
if name=="w": |
|
toggle=1 |
|
return "" |
|
else: |
|
tag="<" |
|
tag+=name |
|
for attr in attrs.keys(): |
|
tag+=""" %s="%s" """%(attr,attrs[attr]) |
|
tag+=">" |
|
return tag |
|
|
|
def createData(data): |
|
global toggle |
|
astring="""<a href="http://141.14.236.86/cgi-bin/toc/dict?step=remotetable;word=%s;lang=de" target="_blank">%s</a> """ |
|
if toggle: # tag war ein w |
|
toggle=0 |
|
if variants.has_key(data): |
|
return astring%(variants[data][0],data) |
|
else: |
|
return astring%(data,data) |
|
|
|
|
|
|
|
# 3 handler functions |
|
def start_element(name, attrs): |
|
global retLex |
|
|
|
retLex+=createTag(name,attrs) |
|
def end_element(name): |
|
global retLex |
|
if not name=="w": |
|
retLex+="</%s>"%(name.encode('utf-8')) |
|
|
|
|
|
def char_data(data): |
|
global retLex |
|
if data: |
|
try: |
|
retLex+=createData(data) |
|
except: |
|
"""no""" |
|
|
|
p = xml.parsers.expat.ParserCreate() |
|
|
|
p.StartElementHandler = start_element |
|
p.EndElementHandler = end_element |
|
p.CharacterDataHandler = char_data |
|
|
|
p.Parse(lemmatized.encode('utf-8'),1) |
|
#print repr(lemmatized.encode('utf-8')) |
|
|
|
return retLex |
|
|
|
|
|
def lemmatize(self,nr='1',lang="de"): |
|
"""lemmatize""" |
|
global ret |
|
ret="" |
|
|
|
def createTag(name,attrs): |
|
tag="<" |
|
tag+=name |
|
for attr in attrs.keys(): |
|
tag+=""" %s="%s" """%(attr,attrs[attr]) |
|
tag+=">" |
|
return tag |
|
|
|
def insertW(str): |
|
splitted=str.split() |
|
wordlist=["<w>%s</w>"%split for split in splitted] |
|
return string.join(wordlist,'\n') |
|
|
|
# 3 handler functions |
|
def start_element(name, attrs): |
|
global ret |
|
ret+=createTag(name,attrs) |
|
def end_element(name): |
|
global ret |
|
ret+="</%s>"%(name.encode('utf-8')) |
|
|
|
def char_data(data): |
|
global ret |
|
ret+=insertW(data) |
|
|
|
p = xml.parsers.expat.ParserCreate() |
|
|
|
p.StartElementHandler = start_element |
|
p.EndElementHandler = end_element |
|
p.CharacterDataHandler = char_data |
|
|
|
p.Parse(self.getPage(nr), 1) |
|
txt="""<wtag locator="xxx"> |
|
<section lang="%s"><s>%s</s></section> |
|
</wtag>""" |
|
ret=txt%(lang,ret) |
|
|
|
return ret |
|
|
|
def getPage(self,nr='1'): |
def getPage(self,nr='1'): |
"""get page n""" |
"""get page n""" |
Line 388 def manage_addECHO_fullText(self, id, ti
|
Line 278 def manage_addECHO_fullText(self, id, ti
|
REQUEST.RESPONSE.redirect(u+'/manage_main') |
REQUEST.RESPONSE.redirect(u+'/manage_main') |
return '' |
return '' |
|
|
|
|
class ECHO_resource(Folder,Persistent,ECHO_basis): |
class ECHO_resource(Folder,Persistent,ECHO_basis): |
"""ECHO Ressource""" |
"""ECHO Ressource""" |
security=ClassSecurityInfo() |
security=ClassSecurityInfo() |
Line 399 class ECHO_resource(Folder,Persistent,EC
|
Line 290 class ECHO_resource(Folder,Persistent,EC
|
|
|
security.declareProtected('View','index_html') |
security.declareProtected('View','index_html') |
|
|
def showRDF(self): |
|
"""showrdf""" |
|
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
|
ret="""<?xml version="1.0" encoding="utf-8"?>\n<RDF:RDF xmlns:RDF="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:ECHONAVIGATION="http://www.echo.eu/rdf#">\n""" |
|
ret+=self.getRDF(urn="echo:colllectionroot")+"\n" |
|
|
|
ret+="""</RDF:RDF>""" |
|
return ret |
|
|
|
def getRDF(self,urn=None): |
security.declareProtected('View','createPDF') |
"""rdf""" |
def createPDF(self,RESPONSE=None,local=None,dpi=150): |
ret=getRDFDescription(self,self.link,urn=urn) |
"""erzeuge pdf file""" |
return ret+self.createSubElementRDF(urn=urn) |
pages=1 |
|
|
|
imagePath=self.getImagePath().replace("/mpiwg/online","") |
|
|
|
|
def createSubElementRDF(self,urn=None): |
image="http://nausikaa2.mpiwg-berlin.mpg.de/digitallibrary/servlet/Scaler?fn="+imagePath+"&dw=%i&dh=%i&pn=%i" |
"""rdf list""" |
xmlInfo="http://nausikaa2.mpiwg-berlin.mpg.de/digitallibrary/dlInfo-xml.jsp?fn="+imagePath |
if not urn: |
dom=xml.dom.minidom.parse(urllib.urlopen(xmlInfo)) |
urn=self.absolute_url() |
for parameter in dom.getElementsByTagName('parameter'): |
ret="" |
if parameter.getAttribute('name')=="pt": |
|
pages=int(parameter.getAttribute('value')) |
|
break |
|
|
rettemp="""<RDF:Seq RDF:about="%s">\n"""%urn |
tempfile.tempdir="/tmp/archivesImageServer" |
flag=0 |
|
|
|
li="""<RDF:li RDF:resource="%s" />\n""" |
tmpPath=tempfile.mktemp() |
if not ('<error>' in self.getFullTextXML(noredirect='Yes')): |
|
nurn=self.absolute_url()+'/getFullTextXML' |
|
rettemp+=li%nurn |
tmpZip=tempfile.mktemp()+".pdf" |
flag=1 |
tmpFn=os.path.split(tmpZip)[1] |
if not ('<error>' in self.getImageView(noredirect='Yes')): |
|
nurn=self.absolute_url()+'/getImageView' |
|
rettemp+=li%nurn |
if not os.path.exists(tempfile.tempdir): |
flag=1 |
os.mkdir(tempfile.tempdir) |
|
|
|
if not os.path.exists(tmpPath): |
if not ('<error>' in self.showMetaDataXML()): |
os.mkdir(tmpPath) |
nurn=self.absolute_url()+'/showMetaDataXML' |
|
rettemp+=li%nurn |
if RESPONSE: |
flag=1 |
RESPONSE.setHeader("Content-Type","text/html") |
|
RESPONSE.write("<h1>I am creating the pdf</h1>") |
rettemp+="</RDF:Seq>" |
txt="<h3>1. step: getting the images( %i pages)</h3>"%pages |
|
RESPONSE.write(txt) |
if flag==1: |
|
ret+=rettemp |
c=canvas.Canvas(tmpZip) |
|
for i in range(1,pages): |
if not ('<error>' in self.getFullTextXML(noredirect='Yes')): |
if RESPONSE: |
nurn=self.absolute_url()+'/getFullTextXML' |
RESPONSE.write(str("<p>Get Page: %i<br>\n"%i)) |
ret+=getRDFDescription(self,self.absolute_url()+'/getFullTextXML',urn=nurn,nameDef="Fulltext",typeName="ECHO_fulltext") |
faktor=dpi/72.0 |
|
|
if not ('<error>' in self.getImageView(noredirect='Yes')): |
fn=tmpPath+"/%i.png"%i |
nurn=self.absolute_url()+'/getImageView' |
|
ret+=getRDFDescription(self,self.absolute_url()+'/getImageView',urn=nurn,nameDef="Image View",typeName="ECHO_imageview") |
width,height=A4 |
|
#print image%(width*faktor,height*faktor,i) |
if not ('<error>' in self.showMetaDataXML()): |
url=urllib.urlopen(image%(width*faktor,height*faktor,i)).read() |
nurn=self.absolute_url()+'/showMetaDataXML' |
fh=file(fn,"w") |
ret+=getRDFDescription(self,self.absolute_url()+'/showMetaDataXML',urn=nurn,nameDef="Metadata",typeName="ECHO_metaData") |
fh.write(url) |
|
fh.close() |
|
|
|
|
|
|
|
c.drawImage(fn,0,0,width=width,height=height) |
|
c.showPage() |
|
c.save() |
|
if RESPONSE: |
|
RESPONSE.write("<p>finished<br>\n") |
|
|
|
if RESPONSE: |
|
len=os.stat(tmpZip)[6] |
|
downloadUrl=self.absolute_url()+"/downloadPDF" |
|
RESPONSE.write("""<h1><a href="downloadPDF?fn=%s">Click here for download ( %i Byte)</a></h1>\n"""%(tmpFn,len)) |
|
RESPONSE.write("""<p>The file will be stored for a while, you can download it later, the URL is:</p> |
|
<p><a href="downloadPDF?fn=%s">%s?fn=%s</a></h1>\n"""%(tmpFn,downloadUrl,tmpFn)) |
|
RESPONSE.close() |
|
|
|
|
|
def downloadPDF(self,fn): |
|
"""download prepared set""" |
|
filename="/tmp/archivesImageServer/"+fn |
|
namePDF=self.getId()+".pdf" |
|
self.REQUEST.RESPONSE.setHeader("Content-Disposition","""attachement; filename="%s" """%namePDF) |
|
self.REQUEST.RESPONSE.setHeader("Content-Type","application/octet-stream") |
|
len=os.stat(filename)[6] |
|
self.REQUEST.RESPONSE.setHeader("Content-Length",len) |
|
images=file(filename).read() |
|
self.REQUEST.RESPONSE.write(images) |
|
self.REQUEST.RESPONSE.close() |
|
|
|
|
|
def getRDF(self,urn=None): |
|
"""rdf""" |
|
ret=getRDFDescription(self,self.link,urn=urn) |
|
return ret+self.createSubElementRDF(urn=urn) |
|
|
return ret |
|
|
|
def getAccessRightSelectorHTML(self,outlook="select"): |
def getAccessRightSelectorHTML(self,outlook="select"): |
"""htmlselector""" |
"""htmlselector""" |
Line 805 class ECHO_resource(Folder,Persistent,EC
|
Line 727 class ECHO_resource(Folder,Persistent,EC
|
else: |
else: |
return "<error>no fulltext available</error>" |
return "<error>no fulltext available</error>" |
|
|
|
def getImagePath(self): |
|
"""gibt pfad zum image aus""" |
|
return self.getImageView(noredirect="yes",onlyPath="yes") |
|
|
def getImageView(self,noredirect=None): |
def getImageView(self,noredirect=None,onlyPath=None): |
"""getImages; give Imageviewr and <error></error> if somthing goes wrong.""" |
"""getImages; give Imageviewr and <error></error> if somthing goes wrong.""" |
try: |
try: |
fh=ECHO_helpers.urlopen(self.metalink) |
fh=ECHO_helpers.urlopen(self.metalink) |
Line 825 class ECHO_resource(Folder,Persistent,EC
|
Line 750 class ECHO_resource(Folder,Persistent,EC
|
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
self.REQUEST.RESPONSE.redirect(imageurl) |
self.REQUEST.RESPONSE.redirect(imageurl) |
else: |
else: |
|
if not onlyPath: |
return imageurl |
return imageurl |
|
else: |
|
return archivepath+"/"+imagetemp |
except: |
except: |
|
|
if not noredirect: |
if not noredirect: |
Line 1067 class ECHO_resource(Folder,Persistent,EC
|
Line 995 class ECHO_resource(Folder,Persistent,EC
|
RESPONSE.redirect('manage_main') |
RESPONSE.redirect('manage_main') |
|
|
|
|
def getMDValue(self,fieldName): |
def getMDValue(self,fieldName,empty=None): |
|
if empty: |
return self.metaDataHash.get(fieldName,'!!NOT USED HERE in Type: %s'%self.contentType) |
return self.metaDataHash.get(fieldName,'!!NOT USED HERE in Type: %s'%self.contentType) |
|
|
|
else: |
|
return self.metaDataHash.get(fieldName,empty) |
|
|
def newMetaXML(self,project=None,startpage=None,xslt=None,thumbtemplate=None,topbar=None,digiLibTemplate=None,digiliburlprefix=None): |
def newMetaXML(self,project=None,startpage=None,xslt=None,thumbtemplate=None,topbar=None,digiLibTemplate=None,digiliburlprefix=None): |
"""new index.meta""" |
"""new index.meta""" |
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
Line 1354 class ECHO_resource(Folder,Persistent,EC
|
Line 1286 class ECHO_resource(Folder,Persistent,EC
|
"""Ausgabe der MD""" |
"""Ausgabe der MD""" |
return getattr(self,item) |
return getattr(self,item) |
|
|
|
def checkRDF(self,path): |
|
"""check if pdf in the path""" |
|
try: |
|
for fileName in os.listdir(path): |
|
if os.path.splitext(fileName)[1]==".pdf": |
|
return os.path.join(path,fileName) |
|
return None |
|
except: |
|
return None |
|
|
|
|
def index_html(self): |
def index_html(self): |
"""standard page""" |
"""standard page""" |
|
pdf=self.checkRDF(self.link) |
|
if pdf: |
|
fh=file(pdf,'r').read() |
|
self.REQUEST.RESPONSE.setHeader('Content-Type','x-Application/pdf') |
|
self.REQUEST.RESPONSE.write(fh) |
|
self.REQUEST.RESPONSE.close() |
|
return |
return self.REQUEST.RESPONSE.redirect(self.link) |
return self.REQUEST.RESPONSE.redirect(self.link) |
|
|
def startpage_html(self): |
def startpage_html(self): |
Line 1572 class ECHO_collection(Folder, Persistent
|
Line 1521 class ECHO_collection(Folder, Persistent
|
return ret+"</RDF:Seq>" |
return ret+"</RDF:Seq>" |
|
|
|
|
def showRDF(self): |
|
"""showrdf""" |
|
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
|
ret="""<?xml version="1.0" encoding="utf-8"?>\n<RDF:RDF xmlns:RDF="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:ECHONAVIGATION="http://www.echo.eu/rdf#">\n""" |
|
ret+=self.getRDF(urn="echo:collectionroot")+"\n" |
|
|
|
ret+="""</RDF:RDF>""" |
|
|
|
return ret |
|
|
|
|
|
|
|
def changeLabels(self): |
def changeLabels(self): |
"""change form""" |
"""change form""" |
Line 2365 class ECHO_group(ECHO_collection):
|
Line 2303 class ECHO_group(ECHO_collection):
|
|
|
security.declareProtected('View','index_html') |
security.declareProtected('View','index_html') |
|
|
def showRDF(self): |
|
"""showrdf""" |
|
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
|
ret="""<?xml version="1.0" encoding="utf-8"?>\n<RDF:RDF xmlns:RDF="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:ECHONAVIGATION="http://www.echo.eu/rdf#">\n""" |
|
ret+=self.getRDF(urn="echo:collectionroot")+"\n" |
|
|
|
ret+="""</RDF:RDF>""" |
|
|
|
return ret |
|
|
|
def getRDF(self,urn=None): |
def getRDF(self,urn=None): |
"""rdf of the collection""" |
"""rdf of the collection""" |
Line 2820 class ECHO_root(Folder,Persistent,Implic
|
Line 2749 class ECHO_root(Folder,Persistent,Implic
|
) |
) |
|
|
|
|
def showRDF(self): |
|
"""showrdf""" |
|
self.REQUEST.RESPONSE.setHeader('Content-Type','text/xml') |
|
ret="""<?xml version="1.0" encoding="utf-8"?>\n<RDF:RDF xmlns:RDF="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:ECHONAVIGATION="http://www.echo.eu/rdf#">\n""" |
|
ret+=self.getRDF(urn="echo:collectionroot")+"\n" |
|
|
|
ret+="""</RDF:RDF>""" |
|
return ret |
|
|
|
|
|
def reloadMetaDataFromStorageWarning(self,RESPONSE=None): |
def reloadMetaDataFromStorageWarning(self,RESPONSE=None): |
"""warning""" |
"""warning""" |