update client parser

client.py

@@ -3,11 +3,20 @@ import argparse
 import logging
 import sys
 import socket
+import re
 
 FORMAT = 'utf-8'
 BUFSIZE = 4096
 
 
+def receive_bytes_chunk(client: socket.socket):
+    buffering = True
+    buffer = b''
+    while buffering:
+        received = client.recv(BUFSIZE)
+        received_size = len(received)
+        logging.debug("Received size: %s", received_size)
+        logging.debug("Received: %r", received)
 def receive_bytes(client: socket.socket):
     buffering = True
     buffer = b''
@@ -21,23 +30,83 @@ def receive_bytes(client: socket.socket):
             buffering = False
 
         buffer += received
-        buffer_split = buffer.split(b"\r\n\r\n")
+        lf_pos = buffer.find(b"\n\n")
+        crlf_pos = buffer.find(b"\r\n\r\n")
+        if lf_pos != -1 and lf_pos < crlf_pos:
+            buffer_split = buffer.split(b"\n\n")
+        else:
+            buffer_split = buffer.split(b"\r\n\r\n")
         buffer = buffer_split[-1]
 
         for part in buffer_split[:-1]:
             yield part + b"\r\n\r\n"
 
+        if buffer:
+            buffering = True
+
     if buffer:
         yield buffer
 
+
+def parse_header(data: bytes):
+    headers = {}
+
+    # decode bytes, split into lines and filter
+    header_split = list(
+        filter(lambda l: l is not "" and not l[0].isspace(), map(str.strip, data.decode("utf-8").split("\n"))))
+
+    if len(header_split) == 0:
+        raise Exception("No start-line")
+    start_line = header_split.pop(0)
+    logging.debug("start-line: %r", start_line)
+
+    for line in header_split:
+        pos = line.find(":")
+
+        if pos <= 0 or pos >= len(line) - 1:
+            continue
+
+        (header, value) = map(str.strip, line.split(":", 1))
+        headers[header.upper()] = value
+
+    logging.debug("Parsed headers: %r", headers)
+
+    return start_line, headers
+
+
+def validate_status_line(status_line: str):
+    split = list(filter(None, status_line.split(" ")))
+    if len(split) < 3:
+        return False
+
+    # Check HTTP version
+    http_version = split.pop(0)
+    if len(http_version) < 8 or http_version[4] != "/":
+        return False
+    (name, version) = http_version[:4], http_version[5:]
+    if name != "HTTP" or not re.match(r"1\.[0|1]", version):
+        return False
+
+    if not re.match(r"\d{3}", split[0]):
+        return False
+
+    return True
+
+
 def http_parser(client: socket.socket):
     headers = {}
     start_line = ""
 
-    counter = 0
-    for received in receive_bytes(client):
-        if counter == 0:
+    receiver = receive_bytes(client)
 
+    (status_line, headers) = parse_header(next(receiver))
+    if not validate_status_line(status_line):
+        raise Exception("Invalid header")
+
+    logging.debug("valid status-line: %r", status_line)
+
+    for chunk in receiver:
+        logging.debug("chunk: %r", chunk)
 
 
 
@@ -56,42 +125,43 @@ def main():
     client = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
     client.connect((arguments.URI, arguments.port))
 
-    message = "GET /httpgallery/chunked/chunkedimage.aspx HTTP/1.1\r\nHost: www.httpwatch.com:80\r\n\r\n".encode(FORMAT)
+    message = "GET / HTTP/1.1\r\nHost: www.google.com:80\r\n\r\n".encode(FORMAT)
     client.sendall(message)
 
-    tmp = b''
-    keep = False
-    count = 0
-    for line in receive_lines(client):
-
-        if count > 0:
-            tmp += line.rstrip(b"\r\n")
-        if keep:
-            count += 1
-
-        if line == b'\r\n':
-            keep = True
-
-    logging.debug('end of part 1')
-
-    logging.debug("attempt 2")
-    while True:
-        logging.debug("attempt")
-        keep = False
-        for line in receive_lines(client):
-            if line == b"0\r\n":
-                break
-            if keep:
-                tmp += line.rstrip(b"\r\n")
-            keep = True
-
-        if b"0\r\n" == line:
-            break
-    logging.debug("content: %s", tmp)
-    # logging.debug("content: %r", tmp.replace(b"\r\n", b"").decode("utf-8"))
-
-    f = open("test.jpeg", "wb")
-    f.write(tmp)
+    http_parser(client)
+    # tmp = b''
+    # keep = False
+    # count = 0
+    # for line in receive_bytes(client):
+    #
+    #     if count > 0:
+    #         tmp += line.rstrip(b"\r\n")
+    #     if keep:
+    #         count += 1
+    #
+    #     if line == b'\r\n':
+    #         keep = True
+    #
+    # logging.debug('end of part 1')
+    #
+    # logging.debug("attempt 2")
+    # while True:
+    #     logging.debug("attempt")
+    #     keep = False
+    #     for line in receive_bytes(client):
+    #         if line == b"0\r\n":
+    #             break
+    #         if keep:
+    #             tmp += line.rstrip(b"\r\n")
+    #         keep = True
+    #
+    #     if b"0\r\n" == line:
+    #         break
+    # logging.debug("content: %s", tmp)
+    # # logging.debug("content: %r", tmp.replace(b"\r\n", b"").decode("utf-8"))
+    #
+    # f = open("test.jpeg", "wb")
+    # f.write(tmp)
 
 
 try: