Adding the fetching from URI... still not fully defined...

crypt4gh/cli.py

@@ -27,7 +27,7 @@
 Utility for the cryptographic GA4GH standard, reading from stdin and outputting to stdout.
 
 Usage:
-   {PROG} [-hv] [--log <file>] encrypt [--sk <path>] --recipient_pk <path> [--recipient_pk <path>]... [--range <start-end>] [--header <path>] [--expiration <date>]
+   {PROG} [-hv] [--log <file>] encrypt [--sk <path>] --recipient_pk <path> [--recipient_pk <path>]... [--range <start-end>] [--header <path>] [--expiration <date>] [--uri <path>]
    {PROG} [-hv] [--log <file>] decrypt [--sk <path>] [--sender_pk <path>] [--range <start-end>]
    {PROG} [-hv] [--log <file>] rearrange [--sk <path>] --range <start-end>
    {PROG} [-hv] [--log <file>] reencrypt [--sk <path>] --recipient_pk <path> [--recipient_pk <path>]... [--trim] [--header-only]
@@ -45,6 +45,7 @@
    --header <path>        Where to write the header (default: stdout)
    --header-only          Whether the input data consists only of a header (default: false)
    --expiration <date>    Expiration date (in ISO format)
+   --uri <path>           Fully Qualified URI Path to the payload location
 
 Environment variables:
    C4GH_LOG         If defined, it will be used as the default logger
@@ -80,6 +81,9 @@ def parse_args(argv=sys.argv[1:]):
     for s in ['--log', '--help', '--version']:#, 'help', 'version']:
         del args[s]
 
+    if args['--range'] is not None and args['--uri'] is not None:
+        raise ValueError('Can not mix --range and --uri')
+
     # print(args)
     return args
 
@@ -157,6 +161,8 @@ def build_recipients():
 
     header = args["--header"]
     timestamp = args["--expiration"]
+    uri = args["--uri"]
+
     if timestamp:
         timestamp = make_timestamp(timestamp)
         LOG.debug("timestamp: %s", timestamp)
@@ -170,7 +176,8 @@ def build_recipients():
                     headerfile = header,
                     offset = range_start,
                     span = range_span,
-                    timestamp = timestamp)
+                    timestamp = timestamp,
+                    uri=uri)
     finally:
         if header:
             header.close()

crypt4gh/fetcher.py

@@ -0,0 +1,33 @@
+
+import logging
+from urllib.request import urlopen, Request
+from urllib.error import URLError, HTTPError
+
+LOG = logging.getLogger(__name__)
+
+class Fetcher():
+
+    __slots__ = ('response')
+
+    def __init__(self, link):
+        try:
+            LOG.info('Fetching URI: %s', link)
+            self.response = urlopen(link)
+        except HTTPError as e:
+            LOG.error('%r', e)
+            raise ValueError(f'HTTP Error {e.code}')
+        except URLError as e:
+            LOG.error('%r', e)
+            raise ValueError(e.reason)
+
+    def read(self, size=-1):
+        if size < 0:
+            return self.response.read()
+        return self.response.read(size)
+
+    def close(self):
+        return self.response.close()
+
+    def seek(self, offset, whence):
+        return self.response.seek(offset, whence=whence)
+

crypt4gh/header.py

@@ -192,12 +192,12 @@ def parse_timestamp_packet(packet):
 def make_packet_link(link):
     return (PACKET_TYPE_LINK
             # + len(link).to_bytes(4,'little') 
-            + link)
+            + link.encode())
 
 def parse_link_packet(packet):
     # size = int.from_bytes(packet[:4], byteorder='little')
     # return packet[:size]
-    return packet
+    return packet.decode()
 
 # -------------------------------------
 # Header Encryption Methods Conventions

crypt4gh/lib.py

@@ -16,7 +16,7 @@
 from . import SEGMENT_SIZE
 from .exceptions import close_on_broken_pipe
 from . import header
-from .fetcher import Fetcher, URLFetcher
+from .fetcher import Fetcher
 
 LOG = logging.getLogger(__name__)
 
@@ -49,7 +49,7 @@ def _encrypt_segment(data, process, key):
 
 
 @close_on_broken_pipe
-def encrypt(keys, infile, outfile, headerfile=None, offset=0, span=None, timestamp=None):
+def encrypt(keys, infile, outfile, headerfile=None, offset=None, span=None, timestamp=None, uri=None):
     '''Encrypt infile into outfile, using the list of keys.
 
 
@@ -60,7 +60,7 @@ def encrypt(keys, infile, outfile, headerfile=None, offset=0, span=None, timesta
     '''
 
     LOG.info('Encrypting the file')
-    
+
     headerfile = headerfile or outfile
 
     # Forward to start position
@@ -97,11 +97,20 @@ def encrypt(keys, infile, outfile, headerfile=None, offset=0, span=None, timesta
         header_packets = chain(header_packets,
                                header.encrypt(header.make_packet_timestamp(timestamp),
                                               keys))
+    if uri:
+        header_packets = chain(header_packets,
+                               header.encrypt(header.make_packet_link(uri),
+                                              keys))
+
     header_bytes = header.serialize(header_packets)
 
     LOG.debug('header length: %d', len(header_bytes))
     headerfile.write(header_bytes)
 
+    if uri:
+        LOG.info('Encryption Successful via the URI: %s', uri)
+        return
+
     # ...and cue music
     LOG.debug("Streaming content")
     # oh boy... I buffer a whole segment!
@@ -395,8 +404,8 @@ def decrypt(keys, infile, outfile, sender_pubkey=None, offset=0, span=None):
     # Infile in now positioned at the beginning of the data portion
     # or we fetch the data portion from the link.
     if link:
-        # replacing the infile with a fetcher
-        outfile = URLFetcher(link)
+        # replacing the infile with a fetcher, might raise exception
+        infile = Fetcher(link)
         # Note: the remainder of the infile might not be empty
 
     # Generator to slice the output