File caching

mailcap-client.py

@@ -5,6 +5,7 @@ import os
 import sys
 import mimetypes
 import subprocess
+import hashlib
 
 s = socket.socket(socket.AF_UNIX)
 s.connect(os.path.join(os.environ["HOME"], ".mailcap.sock"))
@@ -19,4 +20,11 @@ data = open(filename, "rb").read()
 def strtr(x, c="B"):
     return struct.pack(c, len(x)) + x
 
-s.sendall(strtr(filename.split(os.pathsep)[-1].encode('utf-8')) + strtr(mime.encode('utf-8')) + strtr(data, "!I"))
+hash = hashlib.sha256()
+hash.update(data)
+
+s.sendall(strtr(filename.split(os.pathsep)[-1].encode('utf-8')) + strtr(mime.encode('utf-8')) + hash.digest())
+
+response = s.recv(1)
+if len(response) and response[0] == 1:
+    s.sendall(strtr(data, "!I"))

mailcap-server.py

@@ -7,6 +7,7 @@ import mailcap
 import sys
 import argparse
 import daemon
+import hashlib
 
 sockpath = "mailcap.sock"
 
@@ -28,6 +29,9 @@ class Server(object):
             data = self._conn.recv(size)
             return data
 
+        def send(self, data):
+            return self._conn.send(data)
+    
         def recv_exact(self, length):
             retval = bytes()
             while len(retval) < length:
@@ -37,11 +41,16 @@ class Server(object):
                 retval += data
             return retval
 
-    def __init__(self, path="mailcap.sock"):
+    def __init__(self, path="mailcap.sock", dir = None):
         self._socket = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
         self._socket.bind(path)
         self._socket.listen(3)
-        self._dir = tempfile.TemporaryDirectory()
+        if dir:
+            if not os.path.exists(dir):
+                os.makedirs(dir)
+            self._dir = dir
+        else:
+            self._dir = tempfile.TemporaryDirectory()
 
     def close(self):
         self._socket.close()
@@ -51,6 +60,7 @@ class Server(object):
         conn = self.Connection(self._socket.accept()[0])
         conn.timeout(initial_timeout)
 
+        # Receive filename length
         data = conn.recv_exact(1)
         if data is None:
             return None
@@ -58,31 +68,64 @@ class Server(object):
 
         conn.timeout(timeout)
         
+        # Receive filename
         data = conn.recv_exact(fn_len)
         if data is None:
             return None
         fn = data.decode('utf-8')
+
+        # Strip the filename of /, do sanity checks
         fn = fn.replace("/", "_")
         if len(fn) == 0 or fn in ['.', '..']:
+            s.close()
             return None
 
+        # Receive the mime type length
         data = conn.recv_exact(1)
         if data is None:
             return None
         mime_len = struct.unpack("B", data)[0]
         
+        # Receive the mime type
         data = conn.recv_exact(mime_len)
         if data is None:
             return None
         mime = data.decode('utf-8')
 
+        # Receive the SHA256 hash
+        data = conn.recv_exact(hashlib.sha256().digest_size)
+        if data is None:
+            return None
+
+        full_name = os.path.join(self._dir.name, fn)
+        append = 0
+        # While the filename already exists, check if we have a hash match
+        while os.path.exists(full_name):
+            hash = hashlib.sha256()
+            hash.update(open(full_name, "rb").read())
+            
+            # Found a matching file. Do not receive the file.
+            if data == hash.digest():
+                conn.send(b'\x00')
+                conn.close()
+                return full_name, mime
+
+            append += 1
+            hash = hashlib.sha256()
+            name_parts = fn.split('.')
+            if len(name_parts) > 1:
+                full_name = os.path.join(self._dir.name, "%s-%d.%s" % ('.'.join(name_parts[:-1]), append, name_parts[-1]))
+            else:
+                full_name = os.path.join(self._dir.name, "%s-%d" % (fn, append, name_parts[-1]))
+        conn.send(b'\x01')
+
+        # Receive the file length
         data = conn.recv_exact(4)
         if data is None:
             return None
         file_len = struct.unpack("!I", data)[0]
 
-        print(fn, mime)
+        # Receive the file data
-        
         fn = os.path.join(self._dir.name, fn)
         with open(fn, "wb+") as f:
             while file_len > 0:
@@ -98,12 +141,26 @@ class Server(object):
             os.remove(fn)
             return None
 
+def run_server(args):
+    sv = Server(path=args.socket, dir=args.dir)
+    caps = mailcap.getcaps()
+    open(args.pid, "w+").write(str(os.getpid()))
+    while True:
+        rv = sv.next_file()
+        if rv:
+            f, mime = rv
+            print(f, mime)
+            match = mailcap.findmatch(caps, mime.split(';')[0], filename=f)
+            if match and match[0]:
+                os.system(match[0])
+
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
     parser.add_argument('--pid', '-p', default=os.path.join(os.environ['HOME'], '.mailcap.pid'))
     parser.add_argument('--action', '-a', default='run', choices=['run', 'kill'])
     parser.add_argument('--no-daemonize', '-n', action='store_true')
     parser.add_argument('--socket', '-s', default=os.path.join(os.environ['HOME'], '.mailcap.sock'))
+    parser.add_argument('--dir', '-d', default=None)
     args = parser.parse_args()
 
     if args.action == 'kill':
@@ -125,17 +182,8 @@ if __name__ == "__main__":
         if os.path.exists(args.socket):
             os.remove(args.socket)
 
-        with daemon.DaemonContext(detach_process=not args.no_daemonize, files_preserve=[1,2]):
+        if args.no_daemonize:
-            sv = Server(args.socket)
+            run_server(args)
-            caps = mailcap.getcaps()
+        else:
-            open(args.pid, "w+").write(str(os.getpid()))
+            with daemon.DaemonContext():
-            while True:
+                run_server()
-                rv = sv.next_file()
-                if rv:
-                    f, mime = rv
-                    print(f, mime)
-                    match = mailcap.findmatch(caps, mime.split(';')[0], filename=f)
-                    if match and match[0]:
-                        os.system(match[0])
-                        if 'nodelete' not in match[1]:
-                            os.remove(f)