implemented maxmemberkb limit for multidoc (e.g. archive) members

src/filters/rclexecm.py

@@ -17,6 +17,11 @@ class RclExecM:
     def __init__(self):
         self.myname = os.path.basename(sys.argv[0])
         self.mimetype = ""
+        self.maxmembersize = int(os.environ.get("RECOLL_FILTER_MAXMEMBERKB"))
+        if self.maxmembersize:
+            self.maxmembersize = self.maxmembersize * 1024
+        else:
+            self.maxmembersize = 50 * 1024 * 1024
 
     def rclog(self, s, doexit = 0, exitvalue = 1):
         print >> sys.stderr, "RCLMFILT:", self.myname, ":", s

src/filters/rclrar

@@ -49,7 +49,7 @@ class RarExtractor:
 
         if not isdir:
             try:
-                if rarinfo.file_size > 50 * 1024 * 1024:
+                if rarinfo.file_size > self.em.maxmembersize:
                     self.em.rclog("extractone: entry %s size %d too big" %
                               (ipath, rarinfo.file_size))
                     docdata = ""

src/filters/rcltar

@@ -24,7 +24,7 @@ class TarExtractor:
         docdata = ""
         try:
             info = self.tar.getmember(ipath)
-            if info.size > 50 * 1024 * 1024:
+            if info.size > self.em.maxmembersize:
                 # skip
                 docdata = ""
                 self.em.rclog("extractone: entry %s size %d too big" %

src/filters/rclzip

@@ -40,14 +40,14 @@ class ZipExtractor:
     def __init__(self, em):
         self.currentindex = 0
         self.em = em
-
+            
     def extractone(self, ipath):
         #self.em.rclog("extractone: [%s]" % ipath)
         docdata = ""
         try:
             info = self.zip.getinfo(ipath)
             # There could be a 4GB Iso in the zip. We have to set a limit
-            if info.file_size > 50 * 1024*1024:
+            if info.file_size > self.em.maxmembersize:
                 self.em.rclog("extractone: entry %s size %d too big" %
                               (ipath, info.file_size))
                 docdata = ""