add support for file scheme

split url parsing into separate function

browser.py

@@ -1,31 +1,61 @@
+import os.path
 import socket
 import ssl
+import sys
+
+
+SUPPORTED_SCHEMES = [
+    "http",
+    "https",
+    "file",
+]
+
+DEFAULT_FILE = "default.html"
 
 
 class URL:
-    def __init__(self, url: str):
+    def __init__(self, url_string: str | None = None):
-        self.scheme, url = url.split("://", 1)
+        self.scheme: str = ""
-        assert self.scheme in ["http", "https"]
+        self.host: str = ""
-        if "/" not in url:
+        self.port: int = -1
-            url = url + "/"
+        self.path: str = ""
-        self.host, url = url.split("/", 1)
+        self.query: str = ""
-        self.path = "/" + url
+        self.fragment: str = ""
+        self.default_port = False
+        if url_string is not None:
+            parse_url(url_string, self)
 
-        if self.scheme == "http":
+    def to_string(self):
-            self.port = 80
+        url_string = self.scheme + ":"
-        elif self.scheme == "https":
+        if self.host != "":
-            self.port = 443
+            url_string += "//" + self.host
+            if self.port != -1 and not self.default_port:
+                url_string += f":{self.port}"
+        url_string += self.path
+        if self.query != "":
+            url_string += "?" + self.query
+        if self.fragment != "":
+            url_string += "#" + self.fragment
+        return url_string
 
-        if ":" in self.host:
+    def __str__(self):
-            self.host, port = self.host.split(":", 1)
+        return self.to_string()
-            self.port = int(port)
+
+    def __repr__(self):
+        return f"<URL {self.to_string()}>"
+
+
+class Request:
+    def __init__(self, url: URL, method: str = "GET"):
+        self.url = url
+        self.method = method
 
         self.request_string = ""
         self.http_version = "HTTP/1.1"
-        self.headers = {"Host": self.host}
+        self.headers = {"Host": self.url.host}
 
     def add_request_line(self, method):
-        self.request_string += f"{method} {self.path} {self.http_version}\r\n"
+        self.request_string += f"{method} {self.url.path} {self.http_version}\r\n"
 
     def add_header(self, key, value):
         self.headers[key] = value
@@ -44,7 +74,13 @@ class URL:
             self.request_string += f"{key}: {value}\r\n"
         self.request_string += "\r\n"
 
-    def request(self, method: str = "GET", headers: dict = None) -> str:
+    def send_request(self, *args, **kwargs):
+        if self.url.scheme in ["http", "https"]:
+            return self.http_request(*args, **kwargs)
+        elif self.url.scheme == "file":
+            return self.file_request()
+
+    def http_request(self, method: str = "GET", headers: dict = None) -> str:
         if headers is not None:
             self.add_headers(headers)
         s = socket.socket(
@@ -52,10 +88,10 @@ class URL:
             type=socket.SOCK_STREAM,
             proto=socket.IPPROTO_TCP,
         )
-        s.connect((self.host, self.port))
+        s.connect((self.url.host, self.url.port))
-        if self.scheme == "https":
+        if self.url.scheme == "https":
             context = ssl.SSLContext(ssl.PROTOCOL_TLS)
-            s = context.wrap_socket(s, server_hostname=self.host)
+            s = context.wrap_socket(s, server_hostname=self.url.host)
 
         self.add_request_line(method)
         self.add_default_headers()
@@ -81,6 +117,82 @@ class URL:
         s.close()
         return content
 
+    def file_request(self):
+        with open(self.url.path) as f:
+            return f.read()
+
+
+def parse_url(url_string: str, url: URL | None = None) -> tuple[URL, bool]:
+    has_authority = False
+    if url is None:
+        url = URL()
+    try:
+        url.scheme, url_string = url_string.split(":", 1)
+        assert url.scheme in SUPPORTED_SCHEMES
+
+        if url_string.startswith("//"):
+            has_authority = True
+            url_string = url_string[2:]
+            i = 0
+            for char in url_string:
+                if char in ["/", "?", "#"]:
+                    break
+                i += 1
+                url.host += char
+            url_string = url_string[i:]
+        i = 0
+        for char in url_string:
+            if char in ["?", "#"]:
+                break
+            url.path += char
+        url_string = url_string[i:]
+
+        if has_authority and url.host == "" and url.path == "":
+            return url, False
+        elif not has_authority and url.path == "":
+            return url, False
+
+        if url_string.startswith("?"):
+            url_string = url_string[1:]
+            i = 0
+            for char in url_string:
+                if char == "#":
+                    break
+                i += 1
+                url.query += char
+        url_string = url_string[i:]
+        if url_string.startswith("#"):
+            url.fragment = url_string[1:]
+
+        if url.scheme == "http":
+            url.port = 80
+            url.default_port = True
+        elif url.scheme == "https":
+            url.port = 443
+            url.default_port = False
+        if url.scheme in ["http", "https"]:
+            if url.path == "" or url.path is None:
+                url.path = "/"
+        if url.scheme == "file":
+            print(f"{url.scheme=}")
+            print(f"{url.host=}")
+            print(f"{url.port=}")
+            print(f"{url.path=}")
+            print(f"{url.query=}")
+            print(f"{url.fragment=}")
+            if sys.platform == "win32" and url.path.startswith("/") and ":" in url.path:
+                url.path = url.path[1:]
+            if url.path == "" or url.path == "/":
+                return url, False
+
+        if ":" in url.host:
+            url.host, port = url.host.split(":", 1)
+            url.port = int(port)
+        return url, True
+    except Exception as e:
+        print(e)
+        return url, False
+
 
 def show(body: str) -> None:
     in_tag = False
@@ -93,11 +205,17 @@ def show(body: str) -> None:
             print(char, end="")
 
 
-def load(url: URL):
+def load(url_string: str):
-    body = url.request()
+    url, success = parse_url(url_string)
+    if not success:
+        default = os.path.abspath(DEFAULT_FILE)
+        if sys.platform == "win32":
+            default = "/" + default
+        url, _ = parse_url(f"file://{default}")
+
+    body = Request(url).send_request()
     show(body)
 
 
 if __name__ == '__main__':
-    import sys
+    load(sys.argv[1])
-    load(URL(sys.argv[1]))

default.html

@@ -0,0 +1,10 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8">
+  <title>Title</title>
+</head>
+<body>
+<p>Test</p>
+</body>
+</html>

tests/url_test.py

@@ -1,9 +1,11 @@
 import pytest
 
 from browser import URL
+from browser import parse_url
+from browser import Request
 
 @pytest.mark.parametrize(
-    "url_string,scheme,host,port,path",
+    "url_string,scheme,host,port,path,query,fragment,parse_success",
     [
         (
             "http://example.com",
@@ -11,6 +13,9 @@ from browser import URL
             "example.com",
             80,
             "/",
+            "",
+            "",
+            True,
         ),
         (
             "http://example.com/",
@@ -18,6 +23,9 @@ from browser import URL
             "example.com",
             80,
             "/",
+            "",
+            "",
+            True,
         ),
         (
             "https://example.com/",
@@ -25,6 +33,9 @@ from browser import URL
             "example.com",
             443,
             "/",
+            "",
+            "",
+            True,
         ),
         (
             "http://example.com:5000/",
@@ -32,6 +43,9 @@ from browser import URL
             "example.com",
             5000,
             "/",
+            "",
+            "",
+            True,
         ),
         (
             "http://example.com:5000/test/example",
@@ -39,6 +53,9 @@ from browser import URL
             "example.com",
             5000,
             "/test/example",
+            "",
+            "",
+            True,
         ),
         (
             "https://example.com:5000/test/example",
@@ -46,15 +63,101 @@ from browser import URL
             "example.com",
             5000,
             "/test/example",
+            "",
+            "",
+            True,
+        ),
+        (
+            "file:///test.html",
+            "file",
+            "",
+            -1,
+            "/test.html",
+            "",
+            "",
+            True,
+        ),
+        (
+            "file://file_host/test.html",
+            "file",
+            "file_host",
+            -1,
+            "/test.html",
+            "",
+            "",
+            True,
+        ),
+        (
+            "file:///c:/test.txt",
+            "file",
+            "",
+            -1,
+            "c:/test.txt",
+            "",
+            "",
+            True,
+        ),
+        (
+            r"file:///c:\test.txt",
+            "file",
+            "",
+            -1,
+            r"c:\test.txt",
+            "",
+            "",
+            True,
+        ),
+        (
+            "file://file_host/",
+            "file",
+            "file_host",
+            -1,
+            "/",
+            "",
+            "",
+            False,
+        ),
+        (
+            "htp://example.com/",
+            "htp",
+            "",
+            -1,
+            "",
+            "",
+            "",
+            False,
+        ),
+        (
+            "file:test.txt",
+            "file",
+            "",
+            -1,
+            "test.txt",
+            "",
+            "",
+            True,
+        ),
+        (
+            "file:/test.txt",
+            "file",
+            "",
+            -1,
+            "/test.txt",
+            "",
+            "",
+            True,
         ),
     ],
 )
-def test_url_parsing(url_string, scheme, host, port, path):
+def test_url_parsing(url_string, scheme, host, port, path, query, fragment, parse_success):
-    url = URL(url_string)
+    url, success = parse_url(url_string)
     assert url.scheme == scheme
     assert url.host == host
     assert url.port == port
     assert url.path == path
+    assert url.query == query
+    assert url.fragment == fragment
+    assert success == parse_success
 
 
 @pytest.mark.parametrize(
@@ -67,7 +170,7 @@ def test_url_parsing(url_string, scheme, host, port, path):
 )
 def test_http_request(http_server, url_string):
     url = URL(url_string)
-    assert url.request() == "test"
+    assert Request(url).send_request() == "test"
 
 
 @pytest.mark.parametrize(
@@ -80,4 +183,4 @@ def test_http_request(http_server, url_string):
 )
 def test_https_request(https_server, url_string):
     url = URL(url_string)
-    assert url.request() == "test"
+    assert Request(url).send_request() == "test"