--- cdli/cdli_files.py 2007/03/19 18:34:01 1.67
+++ cdli/cdli_files.py 2007/03/22 19:56:22 1.72
@@ -425,11 +425,13 @@ 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)
@@ -453,7 +455,7 @@ class uploadATFThread(Thread):
stObj.returnValue={}
stObj.returnValue['errors']=errors
-
+
stObj.returnValue['newPs']=newPs
stObj.returnValue['tmpdir']=dir
stObj.returnValue['basketLen']=basketLen
@@ -504,12 +506,19 @@ 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="",REQUEST=None,RESPONSE=None):
"""manage baskets, delete or copy"""
if submit=="delete":
self.deleteBaskets(ids)
-
+ elif submit=="join":
+ flag,msg=self.joinBasket("joinedBasket", ids)
+ logging.info("joining %s %s"%(flag,msg))
+
+ elif submit=="subtract":
+ logging.info("BBBb %s %s"%(basket1,basket2))
+ flag,msg=self.subtractBasket("subtractedBasket", basket1,basket2)
+ logging.info("subtract %s %s"%(flag,msg))
if RESPONSE:
RESPONSE.redirect(self.absolute_url())
@@ -632,9 +641,83 @@ 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)
@@ -676,7 +759,7 @@ class CDLIBasketContainer(OrderedFolder)
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:
@@ -778,7 +861,11 @@ 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"""
@@ -842,6 +929,56 @@ 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()))
@@ -881,7 +1018,17 @@ class CDLIBasket(Folder,CatalogAware):
-
+ 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=[]
@@ -968,6 +1115,21 @@ class CDLIBasketVersion(Implicit,Persist
return True
return False
+ 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('View','downloadObjectsAsOneFile')
def downloadObjectsAsOneFile(self,lock=None,procedure=None,REQUEST=None,check="yes",current="no"):
"""download all selected files in one file"""
@@ -1183,7 +1345,8 @@ 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"""
@@ -1420,7 +1583,14 @@ 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"""
@@ -1431,17 +1601,11 @@ 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"""
@@ -1544,7 +1708,7 @@ class CDLIFileFolder(extVersionedFileFol
return []
else:
obj=founds[0].getObject().getLastVersion()
- logging.error(obj.getData())
+
return obj.getData()[0:]
def checkCatalog(self,fn):
@@ -1672,6 +1836,7 @@ class CDLIFileFolder(extVersionedFileFol
#os.write(tf,obj.getLastVersion().data)
if RESPONSE:
RESPONSE.write(obj.getLastVersion().getData()[0:])
+ RESPONSE.write("\n")
self.temp_folder.downloadCounter-=1
self._p_changed=1
transaction.get().commit()
@@ -1756,78 +1921,113 @@ class CDLIRoot(Folder):
meta_type="CDLIRoot"
downloadCounterBaskets=0# counts the current basket downloads if counter > 10 no downloads are possible
- def findWordRegExp(self,searchTerm):
- """find all words in index which match regexp in SearchTerm"""
+ def findWordRegExp(self,indexName,searchTerm):
+ """find all words in index which match regexp in SearchTerm
+ @param indexName: name of the index to be searched in
+ @param searchTerm: word to be searched"""
+
ret=[]
- for x in self.lineIndex.iterkeys():
+ for x in self.lineIndexes[indexName].iterkeys():
if re.match(searchTerm,x):
ret.append(x)
return ret
- def searchRegExpInLineIndexDocs(self,searchTerm):
- """search in inLineIndex with regexp"""
+ def searchRegExpInLineIndexDocs(self,indexName,searchTerm):
+ """search in inLineIndex with regexp
+ @param indexName: name of the index to be searched in
+ @param searchTerm: term to be searched
+ """
if not searchTerm:
return []
ret=[]
- words=self.findWordRegExp(searchTerm) # suche nach allen Treffern
+ words=self.findWordRegExp(indexName,searchTerm) # suche nach allen Treffern
logging.info("wd:%s"%words)
for word in words:
- ret+=self.searchInLineIndexDocs(word)
+ ret+=self.searchInLineIndexDocs(indexName,word)
return unique(ret)
def showInLineIndex(self):
"""get the index for debug purposes"""
print "show"
- for x in self.lineIndex.iterkeys():
- logging.info("word:%s"%repr(x))
- #for y in self.lineIndex[x].iterkeys():
- # print "doc",repr(y),repr(self.lineIndex[x][y])
-
- return self.lineIndex
+ for key in self.lineIndexes.keys():
+ logging.info("index:%s"%key)
+ for x in self.lineIndexes[key].iterkeys():
+ logging.info("word:%s"%repr(x))
+ #for y in self.lineIndex[x].iterkeys():
+ # print "doc",repr(y),repr(self.lineIndex[x][y])
+
+ return self.lineIndexes
- def searchInLineIndexDocs(self,word,uniq=True,regExp=False):
- """search occurences"""
+ def searchInLineIndexDocs(self,indexName,word,uniq=True,regExp=False):
+ """search occurences in an index
+ @param indexName: name of the index to be searched in
+ @param word: word to be searched
+ @param unique: (optional) unify the list of results
+ @param regExp: (optional) use regular expressions
+ """
if regExp:
- return self.searchRegExpInLineIndexDocs(word)
+ return self.searchRegExpInLineIndexDocs(indexName,word)
try:
- lst=list(self.lineIndex.get(word).keys())
+
+ lst=list(self.lineIndexes[indexName].get(word).keys())
except:
+ logging.error("error: searchInLineIndexDocs (%s %s)"%(sys.exc_info()[0:2]))
lst=[]
if uniq:
return unique(lst)
else:
return lst
- def getLinesFromIndex(self,word,doc,regExp=False):
- """get lines"""
+ def getLinesFromIndex(self,indexName,word,doc,regExp=False):
+ """return all lines from a document where word is found
+ @param indexName: Name of the index
+ @param word: word to be searched
+ @param doc: name of the document (usuallay the p-number)
+ @param regExp: (optional) use regExp
+ """
+
if not regExp:
- return self.lineIndex.get(word)[doc]
+ return self.lineIndexes[indexName].get(word)[doc]
else: # wenn regexp, suche welches word
- for w in self.findWordRegExp(word):
- if self.lineIndex.get(w): # ein word in im dex gefunden
+ for w in self.findWordRegExp(indexName,word):
+ if self.lineIndexes[indexName].get(w): # ein word in im dex gefunden
try:
- dc=self.lineIndex.get(word)[doc]
+ dc=self.lineIndex[indexName].get(word)[doc]
return dc # und ein document dann gib es zurueck
except:
pass #andernfalls weiter
- def cleanInLineIndex(self):
- """delete InlineIndex"""
- for x in list(self.lineIndex.keys()):
- del(self.lineIndex[x])
- print [x for x in self.lineIndex.keys()]
+ def cleanInLineIndex(self,indexName):
+ """empty an InlineIndex
+ @param indexName: name of the index
+ """
+ for x in list(self.lineIndexes[indexName].keys()):
+ del(self.lineIndexes[indexName][x])
+ print [x for x in self.lineIndexes[indexName].keys()]
return "ok"
- def storeInLineIndex(self,key,value):
- """store in index"""
-
- if (not hasattr(self,'lineIndex')) or (type(self.lineIndex) is DictType):
- self.lineIndex=OOBTree()
- li=self.lineIndex
+ def storeInLineIndex(self,indexName,key,value):
+ """store in index, key is normally a word or grapheme
+ and value is a tuple (documentname, line) where the word can be found
+ @param indexName: name of the index
+ @param key: key in index
+ @param value: value in index, value is a tuple (document name, line)
+ """
+ logging.error("indexing: %s %s"%(indexName,key))
+ if (not hasattr(self,'lineIndexes')):
+
+ self.lineIndexes={}
+
+ if self.lineIndexes.get(indexName,None) is None:
+ #index exisitiert noch nicht dann anlegen
+
+ self.lineIndexes[indexName]=OOBTree()
+ lis=self.lineIndexes
+ li=lis[indexName]
if li.has_key(key):
@@ -1845,51 +2045,36 @@ class CDLIRoot(Folder):
li[key][value[0]]=[value[1]]
- self.lineIndex=li
+ self.lineIndexes=lis
transaction.get().commit()
def showFile(self,fileId):
- """show a file"""
+ """show a file
+ @param fileId: P-Number of the document to be displayed
+ """
f=self.CDLICatalog({'title':fileId})
if not f:
return ""
return f[0].getObject().getLastVersionFormattedData()
- def showLineFromFile(self,fileId,lineNum,word):
- """get line lineNum fromFileId"""
-
- file=self.showFile(fileId)
- #str="^%s\.[^%s\.]*%s[^\n]*\n"%(lineNum,lineNum,word)
- #str="^%s\..*?%s[^\n]*\n"%(lineNum,word)
-
- #print str
- #m=re.search(str,file,flags=re.M|re.DOTALL)
- #if m:
- # return m.group()
- #else:
- # return ""
- #ret=lineNum+"."
- #splitted=file.split(lineNum+".")
- #if len(splitted)>1:
- #for part in splitted[1:]:
- #if part.find(word)>-1:
- # for x in part.split("\n"):
- #ret+=x
- #if x.find(word)>-1:
- #break
- #break;
- #return ret
- def showWordInFile(self,fileId,word,lineList=None):
+ def showWordInFile(self,fileId,word,lineList=None,regExp=True,indexName=""):
"""get lines with word fromFileId"""
file=self.showFile(fileId)
ret=[]
+ if regExp: # wenn regexp dann generiere alle worte aus der list die der regexp entsprechen
+ wordlist=self.findWordRegExp(indexName,word)
+ else:
+ wordlist=[word]
+
for line in file.split("\n"):
+ found=False
+ for word in wordlist:
if line.find(word)>-1:
if lineList: #liste of moeglichen Zeilennummern
num=line.split(".")[0] #Zeilenummer ist alles vor dem . in der Zeile
@@ -1899,26 +2084,40 @@ class CDLIRoot(Folder):
ret.append(line)
else: # nimm alles ohne line check
ret.append(line)
+
+ break;
return ret
- def tagWordInFile(self,fileId,word,lineList=None):
+ def tagWordInFile(self,fileId,word,lineList=None,regExp=True,indexName=""):
"""get lines with word fromFileId"""
file=self.showFile(fileId)
tagStr="""%s"""
ret=[]
+
+ if regExp: # wenn regexp dann generiere alle worte aus der list die der regexp entsprechen
+ wordlist=self.findWordRegExp(indexName,word)
+ else:
+ wordlist=[word]
+
for line in file.split("\n"):
- if line.find(word)>-1:
+ found=False
+ for word in wordlist:
+ if line.find(word)>-1: #word ist gefunden dann makiere und breche die Schleife ab
if lineList: #liste of moeglichen Zeilennummern
num=line.split(".")[0] #Zeilenummer ist alles vor dem . in der Zeile
if num in lineList:
ret.append(line.replace(word,tagStr%word))
+
else: # nimm alles ohne line check
ret.append(line.replace(word,tagStr%word))
- else:
+ found=True
+ break
+ if not found: #word wurde nicht gefunden keine makierung
ret.append(line)
+
return "
\n".join(ret)
def URLquote(self,str):
@@ -1948,19 +2147,8 @@ class CDLIRoot(Folder):
ret.append((f[0],un))
return ret
-
- def forceDahl(self):
- "break all locks"
- ret=[]
- for f in self.ZopeFind(self,obj_metatypes="CDLI file",search_sub=1):
- if str(f[1].lockedBy)=="dahl":
- un=f[1].forceunlock()
-
- if un and un !="":
- ret.append((f[0],un))
+
- return ret
-
def getChangesByAuthor(self,author,n=100):
"""getChangesByAuthor"""
zcat=self.CDLIObjectsCatalog
@@ -2024,8 +2212,10 @@ class CDLIRoot(Folder):
def uploadATF(self,repeat=None,upload=None,basketId=0,RESPONSE=None):
- """standard ausgabe"""
+ """upload an atf file / basket file"""
#self._v_uploadATF.returnValue=None
+
+ #generate an random id for the upload thread
from random import randint
if (not self.REQUEST.SESSION.get('idTmp',None)):
@@ -2034,8 +2224,10 @@ class CDLIRoot(Folder):
else:
idTmp=self.REQUEST.SESSION.get('idTmp',None)
+
threadName=repeat
if not threadName or threadName=="":
+ #new thread not called from the waiting page
tmpVar=False
thread=uploadATFThread()
@@ -2083,26 +2275,8 @@ class CDLIRoot(Folder):
return pt(txt='/uploadATF',threadName=threadName)
else:
-# tmp={}
-# for key in self._v_uploadATF[threadName].returnValue.keys():
-# t=self._v_uploadATF[threadName].returnValue[key]
-# if type(t) is ListType:
-# tmp[key]=self._v_uploadATF[threadName].returnValue[key][0:]
-# else:
-# tmp[key]=self._v_uploadATF[threadName].returnValue[key]
-# repr(tmp[key]),repr(key)
-#
-# #
- #tmp=self.cdli_main.tmpStore2[threadName]
-
tmp=getattr(self.temp_folder,idTmp).returnValue
-
-
-
-
- #del(self.cdli_main.tmpStore2[threadName])
-
-
+
pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','uploadCheck.zpt')).__of__(self)
return pt(changed=tmp['changed'],lockerrors=tmp['lockerrors'],errors=tmp['errors'],dir=tmp['dir'],newPs=tmp['newPs'],basketLen=tmp['basketLen'],numberOfFiles=tmp['numberOfFiles'],