Support encrypted log items in the cleaner.

[bluesky.git] / cleaner / cleaner

diff --git a/cleaner/cleaner b/cleaner/cleaner
index 58292e4..95bbb08 100755 (executable)
--- a/cleaner/cleaner
+++ b/cleaner/cleaner
@@ -15,7 +15,8 @@ from boto.s3.key import Key
 # The BlueSky 'struct cloudlog_header' data type.
 HEADER_FORMAT = '<4s48sb16sQIII'
 HEADER_CRYPTBYTES = 48
-HEADER_MAGIC = 'AgI-'
+HEADER_MAGIC1 = 'AgI-'          # Unencrypted data
+HEADER_MAGIC2 = 'AgI='          # Encrypted data
 HEADER_SIZE = struct.calcsize(HEADER_FORMAT)
 
 class ITEM_TYPE:
@@ -42,9 +43,14 @@ class FileBackend:
         return [(f, os.stat(os.path.join(self.path, f)).st_size)
                 for f in files]
 
-    def read(self, filename):
+    def read(self, filename, offset=0, length=None):
         fp = open(os.path.join(self.path, filename), 'rb')
-        return fp.read()
+        if offset > 0:
+            fp.seek(offset)
+        if legnth is None:
+            return fp.read()
+        else:
+            return fp.read(length)
 
     def write(self, filename, data):
         fp = open(os.path.join(self.path, filename), 'wb')
@@ -77,10 +83,15 @@ class S3Backend:
             files.append((k.key, k.size))
         return files
 
-    def read(self, filename):
+    def read(self, filename, offset=0, length=None):
         if filename in self.cache:
             fp = open(os.path.join(self.cachedir, filename), 'rb')
-            return fp.read()
+            if offset > 0:
+                fp.seek(offset)
+            if length is None:
+                return fp.read()
+            else:
+                return fp.read(length)
         else:
             k = Key(self.bucket)
             k.key = self.path + filename
@@ -89,6 +100,10 @@ class S3Backend:
             fp.write(data)
             fp.close()
             self.cache[filename] = True
+            if offset > 0:
+                data = data[offset:]
+            if length is not None:
+                data = data[0:length]
             return data
 
     def write(self, filename, data):
@@ -117,9 +132,10 @@ class LogItem:
 
     def __init__(self):
         self.cryptkeys = '\0' * HEADER_CRYPTBYTES
+        self.encrypted = False
 
     def __str__(self):
-        return "<Item ty=%s location=%s size=%d id=%s...>" % (self.type, self.location, self.size, base64.b16encode(self.id).lower()[0:8])
+        return "<Item%s ty=%s location=%s size=%d id=%s...>" % (self.encrypted and '$' or '', self.type, self.location, self.size, base64.b16encode(self.id).lower()[0:8])
 
     @staticmethod
     def random_id():
@@ -135,8 +151,12 @@ class LogItem:
         link_ids = ''.join(link_ids)
         link_locs = ''.join(link_locs)
 
+        if self.encrypted:
+            magic = HEADER_MAGIC2
+        else:
+            magic = HEADER_MAGIC1
         header = struct.pack(HEADER_FORMAT,
-                             HEADER_MAGIC, self.cryptkeys,
+                             magic, self.cryptkeys,
                              ord(self.type), self.id, self.inum,
                              len(self.data), len(link_ids), len(link_locs))
         return header + self.data + link_ids + link_locs
@@ -218,7 +238,7 @@ def parse_item(data):
     header = struct.unpack_from(HEADER_FORMAT, data, 0)
     size = HEADER_SIZE + sum(header[5:8])
 
-    if header[0] != HEADER_MAGIC:
+    if header[0] not in (HEADER_MAGIC1, HEADER_MAGIC2):
         print "Bad header magic!"
         return
 
@@ -227,6 +247,7 @@ def parse_item(data):
         return
 
     item = LogItem()
+    if header[0] == HEADER_MAGIC2: item.encrypted = True
     item.cryptkeys = header[1]
     item.id = header[3]
     item.inum = header[4]
@@ -256,7 +277,7 @@ def load_item(backend, location):
     The elements of the tuple are (directory, sequence, offset, size)."""
 
     filename = backend.loc_to_name((location[0], location[1]))
-    data = backend.read(filename)[location[2] : location[2] + location[3]]
+    data = backend.read(filename, location[2], location[3])
     item = parse_item(data)
     item.location = location
     return item
@@ -275,7 +296,7 @@ def parse_log(data, location=None):
     while len(data) - offset >= HEADER_SIZE:
         header = struct.unpack_from(HEADER_FORMAT, data, offset)
         size = HEADER_SIZE + sum(header[5:8])
-        if header[0] != HEADER_MAGIC:
+        if header[0] not in (HEADER_MAGIC1, HEADER_MAGIC2):
             print "Bad header magic!"
             break
         if size + offset > len(data):