examples/network: Split recv- and read-based HTTP servers.

pfalcon · pfalcon · commit 3d19adf9b323 · 2016-08-06T01:59:51.000+03:00
Name recv() based a "simplistic", as it can't work robustly in every
environment. All this is to let people concentreate on proper, read()-
based one (and to turn recv() based into a "negative showcase",
explaining what are the pitfalls of such approach).
diff --git a/examples/network/http_server.py b/examples/network/http_server.py
@@ -10,7 +10,7 @@
 Hello #%d from MicroPython!
 """
 
-def main(use_stream=False):
+def main(micropython_optimize=False):
     s = socket.socket()
 
     # Binding to all interfaces - server will be accessible to other hosts!
@@ -26,26 +26,37 @@ def main(use_stream=False):
     counter = 0
     while True:
         res = s.accept()
-        client_s = res[0]
+        client_sock = res[0]
         client_addr = res[1]
         print("Client address:", client_addr)
-        print("Client socket:", client_s)
-        print("Request:")
-        if use_stream:
-            # MicroPython socket objects support stream (aka file) interface
-            # directly.
-            req = client_s.readline()
-            print(req)
-            while True:
-                h = client_s.readline()
-                if h == b"" or h == b"\r\n":
-                    break
-                print(h)
-            client_s.write(CONTENT % counter)
+        print("Client socket:", client_sock)
+
+        if not micropython_optimize:
+            # To read line-oriented protocol (like HTTP) from a socket (and
+            # avoid short read problem), it must be wrapped in a stream (aka
+            # file-like) object. That's how you do it in CPython:
+            client_stream = client_sock.makefile("rwb")
         else:
-            print(client_s.recv(4096))
-            client_s.send(CONTENT % counter)
-        client_s.close()
+            # .. but MicroPython socket objects support stream interface
+            # directly, so calling .makefile() method is not required. If
+            # you develop application which will run only on MicroPython,
+            # especially on a resource-constrained embedded device, you
+            # may take this shortcut to save resources.
+            client_stream = client_sock
+
+        print("Request:")
+        req = client_stream.readline()
+        print(req)
+        while True:
+            h = client_stream.readline()
+            if h == b"" or h == b"\r\n":
+                break
+            print(h)
+        client_stream.write(CONTENT % counter)
+
+        client_stream.close()
+        if not micropython_optimize:
+            client_sock.close()
         counter += 1
         print()
 
diff --git a/examples/network/http_server_simplistic.py b/examples/network/http_server_simplistic.py
@@ -0,0 +1,38 @@
+try:
+    import usocket as socket
+except:
+    import socket
+
+
+CONTENT = b"""\
+HTTP/1.0 200 OK
+
+Hello #%d from MicroPython!
+"""
+
+def main():
+    s = socket.socket()
+    ai = socket.getaddrinfo("0.0.0.0", 8080)
+    addr = ai[0][-1]
+
+    s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+
+    s.bind(addr)
+    s.listen(5)
+    print("Listening, connect your browser to http://<this_host>:8080/")
+
+    counter = 0
+    while True:
+        res = s.accept()
+        client_s = res[0]
+        client_addr = res[1]
+        req = client_s.recv(4096)
+        print("Request:")
+        print(req)
+        client_s.send(CONTENT % counter)
+        client_s.close()
+        counter += 1
+        print()
+
+
+main()
diff --git a/examples/network/http_server_simplistic_commented.py b/examples/network/http_server_simplistic_commented.py
@@ -0,0 +1,76 @@
+#
+# MicroPython http_server_simplistic.py example
+#
+# This example shows how to write the smallest possible HTTP
+# server in MicroPython. With comments and convenience code
+# removed, this example can be compressed literally to ten
+# lines. There's a catch though - read comments below for
+# details, and use this code only for quick hacks, preferring
+# http_server.py for "real thing".
+#
+try:
+    import usocket as socket
+except:
+    import socket
+
+
+CONTENT = b"""\
+HTTP/1.0 200 OK
+
+Hello #%d from MicroPython!
+"""
+
+def main():
+    s = socket.socket()
+
+    # Bind to (allow to be connected on ) all interfaces. This means
+    # this server will be accessible to other hosts on your local
+    # network, and if your server has direct (non-firewalled) connection
+    # to the Internet, then to anyone on the Internet. We bind to all
+    # interfaces to let this example work easily on embedded MicroPython
+    # targets, which you will likely access from another machine on your
+    # local network. Take care when running this on an Internet-connected
+    # machine though! Replace "0.0.0.0" with "127.0.0.1" if in doubt, to
+    # make the server accessible only on the machine it runs on.
+    ai = socket.getaddrinfo("0.0.0.0", 8080)
+    print("Bind address info:", ai)
+    addr = ai[0][-1]
+
+    # A port on which a socket listened remains inactive during some time.
+    # This means that if you run this sample, terminate it, and run again
+    # you will likely get an error. To avoid this timeout, set SO_REUSEADDR
+    # socket option.
+    s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+
+    s.bind(addr)
+    s.listen(5)
+    print("Listening, connect your browser to http://<this_host>:8080/")
+
+    counter = 0
+    while True:
+        res = s.accept()
+        client_s = res[0]
+        client_addr = res[1]
+        print("Client address:", client_addr)
+        print("Client socket:", client_s)
+        # We assume here that .recv() call will read entire HTTP request
+        # from client. This is usually true, at least on "big OS" systems
+        # like Linux/MacOS/Windows. But that doesn't have to be true in
+        # all cases, in particular on embedded systems, when there can
+        # easily be "short recv", where it returns much less than requested
+        # data size. That's why this example is called "simplistic" - it
+        # shows that writing a web server in Python that *usually works* is
+        # ten lines of code, and you can use this technique for quick hacks
+        # and experimentation. But don't do it like that in production
+        # applications - instead, parse HTTP request properly, as shown
+        # by http_server.py example.
+        req = client_s.recv(4096)
+        print("Request:")
+        print(req)
+        client_s.send(CONTENT % counter)
+        client_s.close()
+        counter += 1
+        print()
+
+
+main()
