--- cdli/cdli_files.py 2007/02/05 16:20:38 1.64
+++ cdli/cdli_files.py 2007/09/03 11:10:04 1.80
@@ -25,7 +25,65 @@ from BTrees.OOBTree import OOBTree
import logging
import transaction
import copy
+import codecs
+import sys
+def unicodify(s):
+ """decode str (utf-8 or latin-1 representation) into unicode object"""
+ if not s:
+ return u""
+ if isinstance(s, str):
+ try:
+ return s.decode('utf-8')
+ except:
+ return s.decode('latin-1')
+ else:
+ return s
+
+def utf8ify(s):
+ """encode unicode object or string into byte string in utf-8 representation.
+ assumes string objects to be utf-8"""
+ if not s:
+ return ""
+ if isinstance(s, str):
+ return s
+ else:
+ return s.encode('utf-8')
+
+
+def generateXMLReturn(hash):
+ """erzeugt das xml file als returnwert fuer uploadATFRPC"""
+
+ ret="
Changing : %s"%fn+self.result founds[0].getObject().manage_addCDLIFileObject('',comment,SESSION['author'],file=os.path.join(SESSION['tmpdir'],fn),from_tmp=True) if i==200: - i=0 - transaction.get().commit() - logging.info("changing: do commit") - transaction.get().commit() - logging.info("changing: last commit") + i=0 + transaction.get().commit() + logging.info("changing: do commit") + + transaction.get().commit() + logging.info("changing: last commit") #now add the new files newPs=SESSION['newPs'] if len(newPs)>0: tmpDir=SESSION['tmpdir'] + logging.info("adding start") self.result="
Adding files
"+self.result #TODO: make this configurable, at the moment base folder for the files has to be cdli_main ctx2.importFiles(comment=comment,author=str(username) ,folderName=tmpDir, files=newPs,ext=self) - + logging.info("adding finished") #unlock locked files? if unlock: + logging.info("unlocking start") self.result="Unlock files
"+self.result unlockFns=[] for x in os.listdir(SESSION['tmpdir']): if not x in SESSION['errors']: unlockFns.append(x) - + logging.info("unlocking have now what to unlock") + for fn in unlockFns: + #logging.info("will unlock: %s"%fn) founds=ctx2.CDLICatalog.search({'title':fn}) + #logging.info("found it: %s"%repr(founds)) if len(founds)>0: + #logging.info("unlock: %s"%founds[0].getObject().getId()) SESSION['author']=str(username) founds[0].getObject().lockedBy="" + logging.info("unlocking done") - #if a basketname is give, add files to the basket + #if a basketname is given, add files to the basket if not (basketname ==''): - self.result="Add basket
"+self.result + logging.info("add to basket %s"%basketname) + self.result="Add to basket
"+self.result basketId=ctx2.basketContainer.getBasketIdfromName(basketname) if not basketId: # create new basket + logging.info("create basket %s"%basketname) + self.result="Create a new basket
"+self.result ob=ctx2.basketContainer.addBasket(basketname) basketId=ob.getId() basket=getattr(ctx2.basketContainer,str(basketId)) @@ -270,7 +340,7 @@ class uploadATFfinallyThread(Thread): RESPONSE.redirect(self.aq_parent.absolute_url()) - + logging.info("uploadfinally done") return True class tmpStore(SimpleItem): @@ -329,6 +399,7 @@ class uploadATFThread(Thread): ctx = self.getContext(app,serverport=self.serverport) logging.info("run intern") try: + logging.info("created: %s"%idTmp) ctx.temp_folder._setObject(idTmp,tmpStore(idTmp)) except: logging.error("thread upload: %s %s"%sys.exc_info()[0:2]) @@ -338,11 +409,12 @@ class uploadATFThread(Thread): #ctx.cdliRoot.cdli_main.tmpStore2[self.getName()[0:]]=self.returnValue - + transaction.get().commit() conn.close() + return getattr(ctx.temp_folder,idTmp) def getResult(self): """method for accessing result""" @@ -413,11 +485,13 @@ class uploadATFThread(Thread): #if not than add filename to the list of newfiles data=file(os.path.join(dir,fn)).read() - #status,msg=checkFile(fn,data,dir) - status=True - msg="" + status,msg=checkFile(fn,data,dir) + #status=True + + if not status: # error errors.append((fn,msg)) + else: if len(founds)==0: newPs.append(fn) @@ -441,7 +515,7 @@ class uploadATFThread(Thread): stObj.returnValue={} stObj.returnValue['errors']=errors - + stObj.returnValue['newPs']=newPs stObj.returnValue['tmpdir']=dir stObj.returnValue['basketLen']=basketLen @@ -465,6 +539,46 @@ class CDLIBasketContainer(OrderedFolder) security=ClassSecurityInfo() meta_type="CDLIBasketContainer" + def getPNumbersOfBasket(self,basketName): + """get all pnumbers of a basket as a list, returns an empty list if basket not found + @param basketName: name of the basket + """ + ret=[] + basketId=self.getBasketIdfromName(basketName) + if not basketId: + return [] + + ob=getattr(self,basketId).getContent() + + ret=[x[0].split(".")[0] for x in ob] + + return ret + + security.declareProtected('manage','getBasketAsOneFile') + def getBasketAsOneFile(self,basketName,current="no"): + """returns all files of the basket combined in one file + @param basketName: Name of the basket + @param current: (optional) if current is set to "yes" then the most current version of + all files are downloaded and not the versions of the files as stored in the basket + """ + ret="" + basketId=self.getBasketIdfromName(basketName) + if not basketId: + return "" + + ob=getattr(self,basketId).getLastVersion() + for object in ob.getContent(): + if current=="no": #version as they are in the basket + ret+=str(object[0].getData())+"\n" + elif current=="yes": + #search current object + logging.info("crrent: %s"%object[1].getId().split(".")[0]) + founds=self.CDLICatalog.search({'title':object[1].getId().split(".")[0]}) + if len(founds)>0: + ret+=str(founds[0].getObject().getLastVersion().getData())+"\n" + return ret + + security.declareProtected('manage','upDateBaskets') def upDateBaskets(self): """update content in to objects""" @@ -492,12 +606,19 @@ class CDLIBasketContainer(OrderedFolder) trash.manage_pasteObjects(cut) security.declareProtected('manage','manageBaskets') - def manageBaskets(self,ids,submit,REQUEST=None,RESPONSE=None): + def manageBaskets(self,submit,ids=None,basket1="",basket2="",joinBasket="",subtractBasket="",REQUEST=None,RESPONSE=None): """manage baskets, delete or copy""" if submit=="delete": self.deleteBaskets(ids) - + elif submit=="join": + flag,msg=self.joinBasket(joinBasket, ids) + logging.info("joining %s %s"%(flag,msg)) + + elif submit=="subtract": + logging.info("BBBb %s %s"%(basket1,basket2)) + flag,msg=self.subtractBasket(subtractBasket, basket1,basket2) + logging.info("subtract %s %s"%(flag,msg)) if RESPONSE: RESPONSE.redirect(self.absolute_url()) @@ -528,7 +649,7 @@ class CDLIBasketContainer(OrderedFolder) return pt(basketId=basketId,basketName=basketName) - security.declareProtected('View','index_html') + security.declareProtected('manage','index_html') def index_html(self): """stanadard ansicht""" @@ -561,7 +682,7 @@ class CDLIBasketContainer(OrderedFolder) ret+= "BASKET:"+com+"\t"+user+"\t"+time+"\n" for x in values: ret+= x[0]+"\t"+x[1]+"\n" - return ret + return ret def getBaskets(self,sortField='title'): """get all baskets files""" @@ -620,9 +741,83 @@ class CDLIBasketContainer(OrderedFolder) baskets.sort(sortComment) return baskets + + + def subtractBasket(self,newBasket,basket1,basket2): + """subtract basket2 from basket1 + (i.e. newbasket will contain alle elements of basket1 which are not in basket2), + if basket2 contains files which are not in basket1, then theses files fill be ignored + + @param newbasket: name of the new basket + @param basket1: basket where basket2 will be subtracted from + @param basket2: see above + + """ + logging.info("CCCCC %s %s"%(basket1,basket2)) + + try: + newB=self.addBasket(newBasket) + except: + return False, "cannot create the new basket" + + - - + + + bas2= getattr(self,basket2) + bas2content=bas2.getContent() + bas2ids=[x[0] for x in bas2content] + + + + bas1= getattr(self,basket1) + bas1content=bas1.getContent() + + + newBasketContent={} + + for id,version in bas1content: + if not (id in bas2ids): + newBasketContent[id]=version + + username=self.getActualUserName() + + logging.info("sbc %s"%newBasketContent) + newB.addObjectsWithVersion(newBasketContent,username=username,catalog=self.CDLICatalog) + + return True, "" + + + def joinBasket(self,newBasket,oldBaskets): + """join two baskets + @param newbasket: name of the new basket + @param oldbaskets: list of baskets to be joined + """ + try: + newB=self.addBasket(newBasket) + except: + return False, "cannot create the new basket" + + newBasketContent={} + for ob in oldBaskets: + x= getattr(self,ob,None) + if x is None: + return False, "cannot find basket: %s"%ob + + ids=x.getContent() # hole den Inhalt + + for id,version in ids: + if newBasketContent.has_key(id): # p number gibt's schon + newBasketContent[id]=max(newBasketContent[id],version) # speichere die groessere Versionsnumber + else: + newBasketContent[id]=version + username=self.getActualUserName() + + logging.info("nbc %s"%newBasketContent) + newB.addObjectsWithVersion(newBasketContent,username=username,catalog=self.CDLICatalog) + + return True, "" + def getNewId(self): """createIds""" last=getattr(self,'last',0) @@ -636,11 +831,11 @@ class CDLIBasketContainer(OrderedFolder) def setActiveBasket(self,basketId,REQUEST=None): """store active basketId in a cookie""" self.REQUEST.RESPONSE.setCookie("CDLIActiveBasket",basketId,path="/") - try: - qs=cgi.parse_qs(REQUEST['QUERY_STRING']) - del(qs['basketId']) - except: - qs={} + try: + qs=cgi.parse_qs(REQUEST['QUERY_STRING']) + del(qs['basketId']) + except: + qs={} if REQUEST: REQUEST.RESPONSE.redirect(REQUEST['URL1']+'?'+urllib.urlencode(qs)) @@ -658,13 +853,13 @@ class CDLIBasketContainer(OrderedFolder) """get name of the actualuser""" return str(self.REQUEST['AUTHENTICATED_USER']) - + security.declareProtected('manage','addBasket') def addBasket(self,newBasketName): """add a new basket""" ob=manage_addCDLIBasket(self,newBasketName) return ob - + def storeInBasket(self,submit,ids=None,newBasketName=None,fromFileList=None,RESPONSE=None,REQUEST=None): """store it""" if not ids: @@ -714,7 +909,43 @@ class CDLIBasket(Folder,CatalogAware): meta_type="CDLIBasket" default_catalog="CDLIBasketCatalog" + def searchInBasket(self,indexName,searchStr,regExp=False): + """searchInBasket""" + + lst=self.searchInLineIndexDocs(indexName,searchStr,uniq=True,regExp=regExp) + ret={} + + lv=self.getLastVersion() + + for obj in lv.content.getContent(): + id=obj[1].getId().split(".")[0] + if id in lst: + + ret[id]=self.showWordInFile(id,searchStr,lineList=self.getLinesFromIndex(indexName,searchStr,id,regExp=regExp),regExp=regExp,indexName=indexName) + + + pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','searchResultsInBasket')).__of__(self) + return pt(result=ret,indexName=indexName,regExp=regExp,word=searchStr) + + + + + def searchInBasket_v1(self,searchStr): + """search occurences of searchStr in files im basket""" + ret=[] + lv=self.getLastVersion() + logging.info("searching") + for obj in lv.content.getContent(): + txt=obj[0].getData() + for x in txt.split("\n"): + logging.info("search %s"%x) + if re.match(searchStr,x): + ret.append(x) + + return "\n".join(ret) + + def getFile(self,obj): return obj[1] @@ -766,7 +997,11 @@ class CDLIBasket(Folder,CatalogAware): self.shortDescription=shortDescription self.comment=comment - + def getActualUserName(self): + """get name of the actualuser""" + + return str(self.REQUEST['AUTHENTICATED_USER']) + def getLastVersion(self): """hole letzte version""" @@ -830,9 +1065,59 @@ class CDLIBasket(Folder,CatalogAware): return obj + def addObjectsWithVersion(self,ids,deleteOld=None,username=None,catalog=None): + """generate a new version of the basket with objects added, + hier wird jedoch nicht die letzte Version jedes Files hinzugefuegt, s + ondern ids is ein Tupel mit der Id (d.h. der p-number) und der Versionsnummer. + """ + logging.info("add to basket (%s)"%(self.getId())) + lastVersion=self.getLastVersion() + + if not catalog: + catalog=self.CDLICatalog + + if lastVersion is None: + oldContent=[] + else: + oldContent=lastVersion.content.getContent() + + if deleteOld: + oldContent=[] + + newContent=[] + added=0 + + for id,version in ids.iteritems(): + logging.info("adding %s %s"%(id,version)) + id=id.split(".")[0] # title nur die pnumber ohne atf + + try: + founds=catalog.search({'title':id}) + except: + founds=[] + logging.info(" found %s "%(founds)) + for found in founds: + if found.getObject() not in oldContent: + + #TODO: was passiert wenn, man eine Object dazufgt, das schon da ist aber eine neuere version + newContent.append((found.getObject().getVersions()[version-1][1],found.getObject())) + added+=1 + + content=oldContent+newContent + if not username: + logging.error("XXXXXXXXXXX %s"%repr(self)) + user=self.getActualUserName() + else: + user = username + + ob=manage_addCDLIBasketVersion(self,user,comment="",basketContent=content) + logging.info("add to basket (%s) done"%(self.getId())) + return added + + def addObjects(self,ids,deleteOld=None,username=None): """generate a new version of the basket with objects added""" - + logging.info("add to basket (%s)"%(self.getId())) lastVersion=self.getLastVersion() if lastVersion is None: @@ -846,10 +1131,11 @@ class CDLIBasket(Folder,CatalogAware): newContent=[] added=0 for id in ids: - try: - founds=self.CDLICatalog.search({'title':id}) - except: - founds=[] + try: + founds=self.CDLICatalog.search({'title':id}) + except: + founds=[] + for found in founds: if found.getObject() not in oldContent: #TODO: was passiert wenn, man eine Object dazufgt, das schon da ist aber eine neuere version @@ -863,12 +1149,22 @@ class CDLIBasket(Folder,CatalogAware): user = username ob=manage_addCDLIBasketVersion(self,user,comment="",basketContent=content) - + logging.info("add to basket (%s) done"%(self.getId())) return added - + def getContent(self): + """print content""" + ret=[] + + lv=self.getLastVersion() + for obj in lv.content.getContent(): + logging.info("XXXXXXXXXX %s"%repr(obj)) + ret.append((obj[1].getId(),obj[0].versionNumber)) + + return ret + def getContentIds(self): """print basket content""" ret=[] @@ -955,7 +1251,22 @@ class CDLIBasketVersion(Implicit,Persist return True return False - security.declareProtected('View','downloadObjectsAsOneFile') + def downloadListOfPnumbers(self): + """download pnumbers of the basket as list""" + + basket_name=self.aq_parent.title + + ids=self.getContent() # get the list of objects + logging.error(ids) + ret="\n".join([x[1].getId().split(".")[0] for x in ids]) + + self.REQUEST.RESPONSE.setHeader("Content-Disposition","""attachement; filename="%s.txt" """%basket_name) + self.REQUEST.RESPONSE.setHeader("Content-Type","application/octet-stream") + length=len(ret) + self.REQUEST.RESPONSE.setHeader("Content-Length",length) + self.REQUEST.RESPONSE.write(ret) + + security.declareProtected('manage','downloadObjectsAsOneFile') def downloadObjectsAsOneFile(self,lock=None,procedure=None,REQUEST=None,check="yes",current="no"): """download all selected files in one file""" @@ -1027,7 +1338,7 @@ class CDLIBasketVersion(Implicit,Persist ret+=str(object[0].getData())+"\n" elif current=="yes": #search current object - founds=self.CDLICatalog.search({'title':object[0].getId()}) + founds=self.CDLICatalog.search({'title':object[1].getId().split(".")[0]}) if len(founds)>0: ret+=str(founds[0].getObject().getLastVersion().getData())+"\n" @@ -1073,7 +1384,7 @@ class CDLIBasketVersion(Implicit,Persist def __init__(self,id,user,comment="",basketContent=[]): """ init a basket version""" self.id=id - self.coment=comment + self.comment=comment self._setObject('content',BasketContent(basketContent)) #self.basketContent=basketContent[0:]a self.user=user @@ -1087,7 +1398,7 @@ class CDLIBasketVersion(Implicit,Persist """get Comment""" return self.comment - security.declareProtected('View','index_html') + security.declareProtected('manage','index_html') def index_html(self): """view the basket""" @@ -1142,6 +1453,7 @@ class CDLIFileObject(CatalogAware,extVer security.declarePublic('makeThisVersionCurrent') + security.declareProtected('manage','index_html') def PrincipiaSearchSource(self): """Return cataloguable key for ourselves.""" return str(self) @@ -1170,7 +1482,8 @@ class CDLIFileObject(CatalogAware,extVer def getFormattedData(self): """fromat text""" data=self.getData() - return re.sub("\s\#lem"," #lem",data) #remove return vor #lem +# return re.sub("\s\#lem"," #lem",data) #remove return vor #lem + return re.sub("#lem"," #lem",data) #remove return vor #lem def view(self): """view file""" @@ -1224,34 +1537,37 @@ def manage_addCDLIFileObject(self,id,vC= # First, we create the file without data: self._setObject(id, CDLIFileObject(id,title,'',content_type, precondition)) - self._getOb(id).versionComment=str(vC) - self._getOb(id).time=time.localtime() - - setattr(self._getOb(id),'author',author) - + fob = self._getOb(id) + fob.versionComment=str(vC) + fob.time=time.localtime() + setattr(fob,'author',author) # Now we "upload" the data. By doing this in two steps, we # can use a database trick to make the upload more efficient. if file and not from_tmp: - self._getOb(id).manage_upload(file) + fob.manage_upload(file) elif file and from_tmp: - self._getOb(id).manage_upload_from_tmp(file) + fob.manage_upload_from_tmp(file) if content_type: - self._getOb(id).content_type=content_type + fob.content_type=content_type + logging.debug("reindex1: %s"%repr(self)) self.reindex_object() - self._getOb(id).reindex_object() + logging.debug("reindex2: %s in %s"%(repr(fob), repr(fob.default_catalog))) + fob.reindex_object() if REQUEST is not None: REQUEST['RESPONSE'].redirect(self.absolute_url()+'/manage_main') + class CDLIFile(extVersionedFile,CatalogAware): """CDLI file""" + security=ClassSecurityInfo() meta_type="CDLI file" default_catalog='CDLICatalog' - + security.declareProtected('manage','index_html') #security.declarePublic('history') def getLastVersionData(self): """get last version data""" @@ -1407,7 +1723,14 @@ def manage_addCDLIFile(self,id,title,loc if RESPONSE is not None: RESPONSE.redirect('manage_main') - +def checkUTF8(data): + """check utf 8""" + try: + data.encode('utf-8') + return True + except: + return False + def checkFile(filename,data,folder): """check the files""" @@ -1418,17 +1741,11 @@ def checkFile(filename,data,folder): return False,"P missing in the filename" elif len(fn[0])!=7: return False,"P number has not the right length 6" + elif not checkUTF8(data): + return False,"not utf-8" else: - fn=os.path.join(folder,filename) - stin,out=os.popen4("/usr/bin/atfcheck.plx %s"%fn) - value=out.read() - ret= out.close() - - if value: - - return False,"atf checker error: %s"%value - else: - return True,"" + return True,"" + def splitatf(fh,dir=None,ext=None): """split it""" @@ -1436,42 +1753,51 @@ def splitatf(fh,dir=None,ext=None): nf=None i=0 - for lineTmp in fh.readlines(): - for line in lineTmp.split("\r"): - if ext: - i+=1 - if (i%100)==0: - ext.result+="." - if i==10000: - i=0 - ext.result+="adding: %s
"%f + ext.result="adding: %s
"%f+ext.result if not obj: manage_addCDLIFileFolder(root,folder,folder) fobj=getattr(root,folder) @@ -2180,13 +2596,13 @@ class CDLIRoot(Folder): self.CDLICatalog.catalog_object(ob) #self.CDLICatalog.manage_catalogFoundItems(obj_ids=[id],search_sub=1) #self.CDLICatalog.manage_catalogObject(self.REQUEST, self.REQUEST.RESPONSE, 'CDLICatalog', urlparse.urlparse(ob.absolute_url())[1]) - count+=1 + count+=1 - if count > 1000: - print "committing" - transaction.get().commit() - count=0 - transaction.get().commit() + if count > 1000: + print "committing" + transaction.get().commit() + count=0 + transaction.get().commit() return "ok" @@ -2207,9 +2623,9 @@ def manage_addCDLIRoot(self, id, title=' ob.id=str(id) ob.title=title try: - self._setObject(id, ob) + self._setObject(id, ob) except: - pass + pass ob=self._getOb(id) checkPermission=getSecurityManager().checkPermission