python: Use ashd.serve for ashd-wsgi.
[ashd.git] / python / ashd-wsgi
index 6361111..807fe2b 100755 (executable)
@@ -1,32 +1,45 @@
-#!/usr/bin/python3
+#!/usr/bin/python
 
-import sys, os, getopt, threading, time, locale, collections
-import ashd.proto, ashd.util
+import sys, os, getopt, threading, logging, time
+import ashd.proto, ashd.util, ashd.perf, ashd.serve
+try:
+    import pdm.srv
+except:
+    pdm = None
 
 def usage(out):
-    out.write("usage: ashd-wsgi [-hA] [-p MODPATH] [-l REQLIMIT] HANDLER-MODULE [ARGS...]\n")
+    out.write("usage: ashd-wsgi [-hAL] [-m PDM-SPEC] [-p MODPATH] [-l REQLIMIT] HANDLER-MODULE [ARGS...]\n")
 
 reqlimit = 0
 modwsgi_compat = False
-opts, args = getopt.getopt(sys.argv[1:], "+hAp:l:")
+setlog = True
+opts, args = getopt.getopt(sys.argv[1:], "+hAp:l:m:")
 for o, a in opts:
     if o == "-h":
         usage(sys.stdout)
         sys.exit(0)
     elif o == "-p":
         sys.path.insert(0, a)
+    elif o == "-L":
+        setlog = False
     elif o == "-A":
         modwsgi_compat = True
     elif o == "-l":
         reqlimit = int(a)
+    elif o == "-m":
+        if pdm is not None:
+            pdm.srv.listen(a)
 if len(args) < 1:
     usage(sys.stderr)
     sys.exit(1)
+if setlog:
+    logging.basicConfig(format="ashd-wsgi(%(name)s): %(levelname)s: %(message)s")
+log = logging.getLogger("ashd-wsgi")
 
 try:
     handlermod = __import__(args[0], fromlist = ["dummy"])
-except ImportError as exc:
-    sys.stderr.write("ashd-wsgi: handler %s not found: %s\n" % (args[0], exc.args[0]))
+except ImportError, exc:
+    sys.stderr.write("ashd-wsgi: handler %s not found: %s\n" % (args[0], exc.message))
     sys.exit(1)
 if not modwsgi_compat:
     if not hasattr(handlermod, "wmain"):
@@ -39,10 +52,6 @@ else:
         sys.exit(1)
     handler = handlermod.application
 
-class closed(IOError):
-    def __init__(self):
-        super().__init__("The client has closed the connection.")
-
 cwd = os.getcwd()
 def absolutify(path):
     if path[0] != '/':
@@ -50,172 +59,138 @@ def absolutify(path):
     return path
 
 def unquoteurl(url):
-    buf = bytearray()
+    buf = ""
     i = 0
     while i < len(url):
         c = url[i]
         i += 1
-        if c == ord(b'%'):
+        if c == '%':
             if len(url) >= i + 2:
                 c = 0
-                if ord(b'0') <= url[i] <= ord(b'9'):
-                    c |= (url[i] - ord(b'0')) << 4
-                elif ord(b'a') <= url[i] <= ord(b'f'):
-                    c |= (url[i] - ord(b'a') + 10) << 4
-                elif ord(b'A') <= url[i] <= ord(b'F'):
-                    c |= (url[i] - ord(b'A') + 10) << 4
+                if '0' <= url[i] <= '9':
+                    c |= (ord(url[i]) - ord('0')) << 4
+                elif 'a' <= url[i] <= 'f':
+                    c |= (ord(url[i]) - ord('a') + 10) << 4
+                elif 'A' <= url[i] <= 'F':
+                    c |= (ord(url[i]) - ord('A') + 10) << 4
                 else:
                     raise ValueError("Illegal URL escape character")
-                if ord(b'0') <= url[i + 1] <= ord(b'9'):
-                    c |= url[i + 1] - ord('0')
-                elif ord(b'a') <= url[i + 1] <= ord(b'f'):
-                    c |= url[i + 1] - ord(b'a') + 10
-                elif ord(b'A') <= url[i + 1] <= ord(b'F'):
-                    c |= url[i + 1] - ord(b'A') + 10
+                if '0' <= url[i + 1] <= '9':
+                    c |= ord(url[i + 1]) - ord('0')
+                elif 'a' <= url[i + 1] <= 'f':
+                    c |= ord(url[i + 1]) - ord('a') + 10
+                elif 'A' <= url[i + 1] <= 'F':
+                    c |= ord(url[i + 1]) - ord('A') + 10
                 else:
                     raise ValueError("Illegal URL escape character")
-                buf.append(c)
+                buf += chr(c)
                 i += 2
             else:
                 raise ValueError("Incomplete URL escape character")
         else:
-            buf.append(c)
+            buf += c
     return buf
 
-def dowsgi(req):
+def mkenv(req):
     env = {}
     env["wsgi.version"] = 1, 0
     for key, val in req.headers:
-        env["HTTP_" + key.upper().replace(b"-", b"_").decode("latin-1")] = val.decode("latin-1")
+        env["HTTP_" + key.upper().replace("-", "_")] = val
     env["SERVER_SOFTWARE"] = "ashd-wsgi/1"
     env["GATEWAY_INTERFACE"] = "CGI/1.1"
-    env["SERVER_PROTOCOL"] = req.ver.decode("latin-1")
-    env["REQUEST_METHOD"] = req.method.decode("latin-1")
-    try:
-        rawpi = unquoteurl(req.rest)
-    except:
-        rawpi = req.rest
-    try:
-        name, rest, pi = (v.decode("utf-8") for v in (req.url, req.rest, rawpi))
-        env["wsgi.uri_encoding"] = "utf-8"
-    except UnicodeError as exc:
-        name, rest, pi = (v.decode("latin-1") for v in (req.url, req.rest, rawpi))
-        env["wsgi.uri_encoding"] = "latin-1"
-    env["REQUEST_URI"] = name
+    env["SERVER_PROTOCOL"] = req.ver
+    env["REQUEST_METHOD"] = req.method
+    env["REQUEST_URI"] = req.url
+    name = req.url
     p = name.find('?')
     if p >= 0:
         env["QUERY_STRING"] = name[p + 1:]
         name = name[:p]
     else:
         env["QUERY_STRING"] = ""
-    if name[-len(rest):] == rest:
+    if name[-len(req.rest):] == req.rest:
         # This is the same hack used in call*cgi.
-        name = name[:-len(rest)]
-    if name == "/":
+        name = name[:-len(req.rest)]
+    try:
+        pi = unquoteurl(req.rest)
+    except:
+        pi = req.rest
+    if name == '/':
         # This seems to be normal CGI behavior, but see callcgi.c for
         # details.
         pi = "/" + pi
         name = ""
     env["SCRIPT_NAME"] = name
     env["PATH_INFO"] = pi
-    for src, tgt in [("HTTP_HOST", "SERVER_NAME"), ("HTTP_X_ASH_SERVER_PORT", "SERVER_PORT"),
-                     ("HTTP_X_ASH_ADDRESS", "REMOTE_ADDR"), ("HTTP_CONTENT_TYPE", "CONTENT_TYPE"),
-                     ("HTTP_CONTENT_LENGTH", "CONTENT_LENGTH"), ("HTTP_X_ASH_PROTOCOL", "wsgi.url_scheme")]:
-        if src in env: env[tgt] = env[src]
-    if "X-Ash-Protocol" in req and req["X-Ash-Protocol"] == b"https": env["HTTPS"] = "on"
-    if "X-Ash-File" in req: env["SCRIPT_FILENAME"] = absolutify(req["X-Ash-File"].decode(locale.getpreferredencoding()))
+    if "Host" in req: env["SERVER_NAME"] = req["Host"]
+    if "X-Ash-Server-Address" in req: env["SERVER_ADDR"] = req["X-Ash-Server-Address"]
+    if "X-Ash-Server-Port" in req: env["SERVER_PORT"] = req["X-Ash-Server-Port"]
+    if "X-Ash-Protocol" in req and req["X-Ash-Protocol"] == "https": env["HTTPS"] = "on"
+    if "X-Ash-Address" in req: env["REMOTE_ADDR"] = req["X-Ash-Address"]
+    if "X-Ash-Port" in req: env["REMOTE_PORT"] = req["X-Ash-Port"]
+    if "Content-Type" in req:
+        env["CONTENT_TYPE"] = req["Content-Type"]
+        # The CGI specification does not strictly require this, but
+        # many actualy programs and libraries seem to.
+        del env["HTTP_CONTENT_TYPE"]
+    if "Content-Length" in req:
+        env["CONTENT_LENGTH"] = req["Content-Length"]
+        del env["HTTP_CONTENT_LENGTH"]
+    if "X-Ash-File" in req: env["SCRIPT_FILENAME"] = absolutify(req["X-Ash-File"])
+    if "X-Ash-Protocol" in req: env["wsgi.url_scheme"] = req["X-Ash-Protocol"]
     env["wsgi.input"] = req.sk
     env["wsgi.errors"] = sys.stderr
     env["wsgi.multithread"] = True
     env["wsgi.multiprocess"] = False
     env["wsgi.run_once"] = False
+    return env
 
-    resp = []
-    respsent = []
+if reqlimit != 0:
+    guard = ashd.serve.calllimiter(reqlimit).call
+else:
+    guard = lambda fun: fun()
 
-    def recode(thing):
-        if isinstance(thing, collections.ByteString):
-            return thing
-        else:
-            return str(thing).encode("latin-1")
+class reqthread(ashd.serve.wsgithread):
+    def __init__(self, req):
+        super(reqthread, self).__init__()
+        self.req = req.dup()
+    
+    def handlewsgi(self):
+        return handler(self.env, self.startreq)
 
-    def flushreq():
-        if not respsent:
-            if not resp:
-                raise Exception("Trying to write data before starting response.")
-            status, headers = resp
-            respsent[:] = [True]
-            buf = bytearray()
-            buf += b"HTTP/1.1 " + recode(status) + b"\n"
+    def writehead(self, status, headers):
+        try:
+            self.req.sk.write("HTTP/1.1 %s\n" % status)
             for nm, val in headers:
-                buf += recode(nm) + b": " + recode(val) + b"\n"
-            buf += b"\n"
-            try:
-                req.sk.write(buf)
-            except IOError:
-                raise closed()
+                self.req.sk.write("%s: %s\n" % (nm, val))
+            self.req.sk.write("\n")
+        except IOError:
+            raise ashd.serve.closed()
 
-    def write(data):
-        if not data:
-            return
-        flushreq()
+    def writedata(self, data):
         try:
-            req.sk.write(data)
-            req.sk.flush()
+            self.req.sk.write(data)
+            self.req.sk.flush()
         except IOError:
-            raise closed()
+            raise ashd.serve.closed()
 
-    def startreq(status, headers, exc_info = None):
-        if resp:
-            if exc_info:                # Interesting, this...
-                try:
-                    if respsent:
-                        raise exc_info[1]
-                finally:
-                    exc_info = None     # CPython GC bug?
-            else:
-                raise Exception("Can only start responding once.")
-        resp[:] = status, headers
-        return write
-
-    respiter = handler(env, startreq)
-    try:
+    def handle(self):
+        self.env = mkenv(self.req)
+        reqevent = ashd.perf.request(self.env)
+        exc = (None, None, None)
         try:
-            for data in respiter:
-                write(data)
-            if resp:
-                flushreq()
-        except closed:
-            pass
-    finally:
-        if hasattr(respiter, "close"):
-            respiter.close()
-
-flightlock = threading.Condition()
-inflight = 0
+            super(reqthread, self).handle()
+            if self.status:
+                reqevent.response([self.status, self.headers])
+        except:
+            exc = sys.exc_info()
+            raise
+        finally:
+            reqevent.__exit__(*exc)
 
-class reqthread(threading.Thread):
-    def __init__(self, req):
-        super().__init__(name = "Request handler")
-        self.req = req.dup()
-    
     def run(self):
-        global inflight
         try:
-            with flightlock:
-                if reqlimit != 0:
-                    start = time.time()
-                    while inflight >= reqlimit:
-                        flightlock.wait(10)
-                        if time.time() - start > 10:
-                            os.abort()
-                inflight += 1
-            try:
-                dowsgi(self.req)
-            finally:
-                with flightlock:
-                    inflight -= 1
-                    flightlock.notify()
+            guard(super(reqthread, self).run)
         finally:
             self.req.close()