Client update

client.py

@@ -1,13 +1,14 @@
 #!/usr/bin/env python3
 import argparse
 import logging
-import sys
-import socket
 import re
+import socket
+import sys
 import time
-import os
+from urllib.parse import urlparse
 
-from client.ResponseHandler import ResponseHandler
+from client import ResponseHandler
+from client.httpclient import HTTPClient
 
 FORMAT = 'utf-8'
 BUFSIZE = 4096
@@ -125,22 +126,7 @@ def get_chunk(buffer: bytes):
     return buffer[:split_start], buffer[split_end:]
 
 
-def get_html_filename(headers):
-    if "CONTENT-LOCATION" not in headers:
-        return "index.html"
-
-    filename = headers["CONTENT-LOCATION"]
-    result = os.path.basename(filename).strip()
-
-    if len(result.strip()) == 0:
-        return 'index.html'
-
-    return result
-
-
 def response_parser(client: socket.socket):
-    client.settimeout(3.0)
-
     try:
         buffer = client.recv(BUFSIZE)
     except TimeoutError as err:
@@ -165,7 +151,7 @@ def response_parser(client: socket.socket):
         if payload_size == 0:
             return
 
-        filename = get_html_filename(headers)
+        filename = util.get_html_filename(headers)
 
         f = open(filename, "wb")
         f.write(buffer)
@@ -199,6 +185,20 @@ def http_parser(client: socket.socket):
         logging.debug("chunk: %r", chunk)
 
 
+def parse_uri(uri: str):
+    parsed = urlparse(uri)
+
+    # If there is no netloc, the url is invalid, so prepend `//` and try again
+    if parsed.netloc == "":
+        parsed = urlparse("//" + uri)
+
+    host = parsed.netloc
+    path = parsed.path
+    if len(path) == 0 or path[0] != '/':
+        path = "/" + path
+    return host, path
+
+
 def main():
     parser = argparse.ArgumentParser(description='HTTP Client')
     parser.add_argument("--verbose", "-v", action='count', default=0, help="Increase verbosity level of logging")
@@ -211,13 +211,19 @@ def main():
     logging.basicConfig(level=logging.ERROR - (10 * arguments.verbose))
     logging.debug("Arguments: %s", arguments)
 
-    client = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-    client.connect((arguments.URI, arguments.port))
+    (host, path) = parse_uri(arguments.URI)
+    client = HTTPClient(host)
+    client.connect((host, arguments.port))
 
-    message = "GET /Protocols/HTTP/Performance/microscape/ HTTP/1.1\r\nHost: www.w3.org:80\r\n\r\n".encode(FORMAT)
+    message = "GET {path} HTTP/1.1\r\n".format(path=path)
+    message += "Accept: */*\r\nAccept-Encoding: identity\r\n"
+    message += "Host: {host}\r\n\r\n".format(host=host)
+
+    message = message.encode(FORMAT)
+    logging.debug("Sending HTTP message: %r", message)
     client.sendall(message)
-
-    response_parser(client)
+    ResponseHandler.handle(client, arguments.URI)
+    # response_parser(client)
     # http_parser(client)
     # tmp = b''
     # keep = False