Mercurial > hg > purlService

changeset 39:a33fa2377075

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
outfiles are now set as parameter
author dwinter
date Wed, 05 Mar 2014 10:11:50 +0100
parents 6d0d7f1c11f2
children 671dd1e4bd09
files addDriToIndexMeta.py harvestToPurl.py
diffstat 2 files changed, 23 insertions(+), 15 deletions(-) [+]
line wrap: on
line diff
--- a/addDriToIndexMeta.py	Tue Mar 04 14:16:55 2014 +0100
+++ b/addDriToIndexMeta.py	Wed Mar 05 10:11:50 2014 +0100
@@ -11,9 +11,6 @@
 
 from os.path import join, getsize
 
-errorFile = file("/tmp/addDRIErrors.txt","w")
-parseErrorFile = file("/tmp/addDRIParseErrors.txt","w")
-alreadyExistsFile = file("/tmp/addDRIalreadyExists.txt","w")
 
 
 def correctAuthor(tree):
@@ -36,7 +33,11 @@
     return changed
 
 
-def getDRIfromIndexMeta(fl):
+def getDRIfromIndexMeta(fl,parseErrorFile=None):
+    
+    
+    if parseErrorFile is None:
+          parseErrorFile = file("/tmp/addDRIParseErrors.txt","w")
     try:
         tree = etree.parse(fl)
     except:
@@ -51,11 +52,11 @@
         return dris[0].text
 
 
-def addPURL(fl,purl,test=False):
+def addPURL(fl,purl,efiles,test=False):
     try:
         tree = etree.parse(fl)
     except:
-        parseErrorFile.write("PARSE ERROR:"+fl+"\n")
+        efiles.parseErrorFile.write("PARSE ERROR:"+fl+"\n")
         return False
     
     dris = tree.xpath("/resource/meta/dri[@type='mpiwg']")
@@ -70,13 +71,13 @@
         newDri.text=purl
         metas=tree.xpath("/resource/meta")
         if len(metas)==0:
-            parseErrorFile.write("no resource/meta: %s \n"%fl)
+            efiles.parseErrorFile.write("no resource/meta: %s \n"%fl)
             return False
         else:
             metas[0].append(newDri)
     else:
         dris[0].text=purl
-        alreadyExistsFile.write("%s \n"%fl)
+        efiles.alreadyExistsFile.write("%s \n"%fl)
         if not changed: #nothing has to be done
             return True
 
@@ -95,10 +96,10 @@
             
             print sys.exc_info()[0]
             print sys.exc_info()[1]
-            errorFile.write(fl+"\n")
+            efiles.errorFile.write(fl+"\n")
     return True
     
-def addDriToIndexMeta(path,delpath="",replacepath="",test=False):
+def addDriToIndexMeta(path,efiles,delpath="",replacepath="",test=False):
     
     md=manageIndexMetaPURLs.IndexMetaPURLManager()
     
@@ -111,7 +112,7 @@
                 shortPath=re.sub("^"+delpath,replacepath,fl)
                 purl=md.getPurl(shortPath)
               
-                addPURL(fl,purl,test)
+                addPURL(fl,purl,test,efiles)
                     
         if 'pageimg' in dirs:
             dirs.remove('pageimg')  # don't visit pageimf
@@ -122,4 +123,11 @@
                 dirs.remove(dir)
 
 if __name__ == '__main__':
-     addDriToIndexMeta("/mpiwg/online/permanent/vlp",delpath="/mpiwg/online",test=False)
+    class ef:
+        errorFile = file("/tmp/addDRIErrors.txt","w")
+        parseErrorFile = file("/tmp/addDRIParseErrors.txt","w")
+        alreadyExistsFile = file("/tmp/addDRIalreadyExists.txt","w")
+        
+    efiles = ef()
+
+    addDriToIndexMeta("/mpiwg/online/permanent/vlp",efiles,delpath="/mpiwg/online",test=False)
--- a/harvestToPurl.py	Tue Mar 04 14:16:55 2014 +0100
+++ b/harvestToPurl.py	Wed Mar 05 10:11:50 2014 +0100
@@ -11,7 +11,7 @@
 
 
 import managePurls.manageIndexMetaPURLs as manageIndexMetaPURLs
-import addDriToIndexMeta
+from addDriToIndexMeta import getDRIfromIndexMeta
 
 
 import os
@@ -45,8 +45,8 @@
             if name.endswith(".meta"):
                 fl=join(root, name)
                 
-                
-                driIndexMeta=addDriToIndexMeta.getDRIfromIndexMeta(fl)
+                parseErrorFile = file("/tmp/HarvestToPurlParseErrors.txt","w")
+                driIndexMeta=getDRIfromIndexMeta(fl,parseErrorFile=parseErrorFile)
                 
                 imagePath=createImagePath(fl,root)