client: cleanup

client/ResponseHandler.py

@@ -1,6 +1,7 @@
 import logging
 import os
 import re
+from abc import ABC, abstractmethod
 from typing import Dict
 from urllib.parse import urlparse
 
@@ -10,21 +11,7 @@ from client.Retriever import Retriever
 from client.httpclient import HTTPClient, UnsupportedEncoding, FORMAT, InvalidResponse, InvalidStatusLine
 
 
-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
-
-
-class ResponseHandler:
+class ResponseHandler(ABC):
     client: HTTPClient
     headers: Dict[str, str]
     status_code: int
@@ -38,6 +25,7 @@ class ResponseHandler:
         self.retriever = retriever
         pass
 
+    @abstractmethod
     def handle(self):
         pass
 
@@ -133,8 +121,22 @@ class ResponseHandler:
                 logging.error("Invalid content-length value: %r", next_value)
                 raise InvalidResponse()
 
+    @staticmethod
+    def parse_uri(uri: str):
+        parsed = urlparse(uri)
 
-class DownloadHandler(ResponseHandler):
+        # 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
+
+
+class DownloadHandler(ResponseHandler, ABC):
     path: str
 
     def __init__(self, retriever: Retriever, client: HTTPClient, headers: Dict[str, str], url: str, dir=None):
@@ -152,9 +154,6 @@ class DownloadHandler(ResponseHandler):
             return HTMLDownloadHandler(retriever, client, headers, url, dir)
         return RawDownloadHandler(retriever, client, headers, url, dir)
 
-    def handle(self) -> str:
-        pass
-
     def _create_directory(self):
         path = self._get_duplicate_name(os.path.abspath(self.client.host))
         os.mkdir(path)
@@ -248,7 +247,7 @@ class HTMLDownloadHandler(DownloadHandler):
 
     def __download_images(self, tmp_filename, target_filename):
 
-        (host, path) = parse_uri(self.url)
+        (host, path) = ResponseHandler.parse_uri(self.url)
         with open(tmp_filename, "rb") as fp:
             soup = BeautifulSoup(fp, 'html.parser')
 
@@ -286,7 +285,7 @@ class HTMLDownloadHandler(DownloadHandler):
             img_path = parsed.path
         else:
             same_host = False
-            (img_host, img_path) = parse_uri(img_src)
+            (img_host, img_path) = ResponseHandler.parse_uri(img_src)
 
         message = "GET {path} HTTP/1.1\r\n".format(path=img_path)
         message += "Accept: */*\r\nAccept-Encoding: identity\r\n"