Fixed a silly bug which caused a spin loop.

darcs-hash:20090617020833-20ca2-94bcc35d11aa3df6c21892a6b9352f4223fdd916.gz

src/FileFormats/HTML.py

@@ -37,10 +37,7 @@ def decoder(x):
         return x.group(1).decode("hex")
 
     ## we can use this to handle arbitrary levels of quoting
-    while "%" in string:
-        try:
-            string = re.sub("%(..)", decoder, string)
-        except: break
+    string = re.sub("%(..)", decoder, string)
 
     ## references seem to _always_ be encoded using utf8 - even if the
     ## page is encoded using a different charset??? This whole quoting

src/pyflag/Farm.py

@@ -554,7 +554,13 @@ def wake_workers():
 
     Registry.Init()
 
-    #config.parse_options()
-
+    config.parse_options()
+
+    def handler(sig, frame):
+        import pdb
+        pdb.set_trace()
+
+    print "Setting int"
+    signal.signal(signal.SIGINT, handler)
     pyflaglog.log(pyflaglog.INFO,"Starting PyFlag worker")
-    nanny(worker_run)
+    worker_run()

utilities/incremental_load.py

@@ -170,14 +170,19 @@ def load_file(filename, processor,  pcap_dbh):
         except StopIteration:
             break
 
-        pcap_dbh.mass_insert(
-            id = pcap_id,
-            iosource = config.iosource,
-            offset = offset,
-            length = packet.caplen,
-            _ts_sec =  "from_unixtime('%s')" % packet.ts_sec,
-            ts_usec = packet.ts_usec,
-            )
+        try:
+            pcap_dbh.mass_insert(
+                id = pcap_id,
+                iosource = config.iosource,
+                offset = offset,
+                length = packet.caplen,
+                _ts_sec =  "from_unixtime('%s')" % packet.ts_sec,
+                ts_usec = packet.ts_usec,
+                )
+        except KeyError,e:
+            print packet.list()
+            print e
+            continue
 
         input_file.set_id(pcap_id)