[ashd.git] / python3 / ashd-wsgi3

#!/usr/bin/python3

import sys, os, getopt, threading, logging, time, locale, collections
import ashd.proto, ashd.util, ashd.perf
try:
    import pdm.srv
except:
    pdm = None

def usage(out):
    out.write("usage: ashd-wsgi3 [-hAL] [-m PDM-SPEC] [-p MODPATH] [-l REQLIMIT] HANDLER-MODULE [ARGS...]\n")

reqlimit = 0
modwsgi_compat = False
setlog = True
opts, args = getopt.getopt(sys.argv[1:], "+hALp: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-wsgi3(%(name)s): %(levelname)s: %(message)s")
log = logging.getLogger("ashd-wsgi3")

try:
    handlermod = __import__(args[0], fromlist = ["dummy"])
except ImportError as exc:
    sys.stderr.write("ashd-wsgi3: handler %s not found: %s\n" % (args[0], exc.args[0]))
    sys.exit(1)
if not modwsgi_compat:
    if not hasattr(handlermod, "wmain"):
        sys.stderr.write("ashd-wsgi3: handler %s has no `wmain' function\n" % args[0])
        sys.exit(1)
    handler = handlermod.wmain(*args[1:])
else:
    if not hasattr(handlermod, "application"):
        sys.stderr.write("ashd-wsgi3: handler %s has no `application' object\n" % args[0])
        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] != '/':
        return os.path.join(cwd, path)
    return path

def unquoteurl(url):
    buf = bytearray()
    i = 0
    while i < len(url):
        c = url[i]
        i += 1
        if c == ord(b'%'):
            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
                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
                else:
                    raise ValueError("Illegal URL escape character")
                buf.append(c)
                i += 2
            else:
                raise ValueError("Incomplete URL escape character")
        else:
            buf.append(c)
    return buf

def dowsgi(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["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
    p = name.find('?')
    if p >= 0:
        env["QUERY_STRING"] = name[p + 1:]
        name = name[:p]
    else:
        env["QUERY_STRING"] = ""
    if name[-len(rest):] == rest:
        # This is the same hack used in call*cgi.
        name = name[:-len(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_PROTOCOL", "wsgi.url_scheme"),
                     ("HTTP_X_ASH_SERVER_ADDRESS", "SERVER_ADDR"), ("HTTP_X_ASH_SERVER_PORT", "SERVER_PORT"),
                     ("HTTP_X_ASH_ADDRESS", "REMOTE_ADDR"), ("HTTP_X_ASH_PORT", "REMOTE_PORT"),
                     ("HTTP_CONTENT_TYPE", "CONTENT_TYPE"), ("HTTP_CONTENT_LENGTH", "CONTENT_LENGTH")]:
        if src in env: env[tgt] = env[src]
    for key in ["HTTP_CONTENT_TYPE", "HTTP_CONTENT_LENGTH"]:
        # The CGI specification does not strictly require this, but
        # many actualy programs and libraries seem to.
        if key in env: del env[key]
    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()))
    env["wsgi.input"] = req.sk
    env["wsgi.errors"] = sys.stderr
    env["wsgi.multithread"] = True
    env["wsgi.multiprocess"] = False
    env["wsgi.run_once"] = False

    resp = []
    respsent = []

    def recode(thing):
        if isinstance(thing, collections.ByteString):
            return thing
        else:
            return str(thing).encode("latin-1")

    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"
            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()

    def write(data):
        if not data:
            return
        flushreq()
        try:
            req.sk.write(data)
            req.sk.flush()
        except IOError:
            raise 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

    with ashd.perf.request(env) as reqevent:
        try:
            respiter = handler(env, startreq)
            try:
                for data in respiter:
                    write(data)
                if resp:
                    flushreq()
            finally:
                if hasattr(respiter, "close"):
                    respiter.close()
        except closed:
            pass
        if resp:
            reqevent.response(resp)

flightlock = threading.Condition()
inflight = 0

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()
        except:
            log.error("exception occurred in handler thread", exc_info=True)
        finally:
            self.req.close()
            sys.stderr.flush()
    
def handle(req):
    reqthread(req).start()

ashd.util.serveloop(handle)
Commit	Line	Data
	1	#!/usr/bin/python3
	2
	3	import sys, os, getopt, threading, logging, time, locale, collections
	4	import ashd.proto, ashd.util, ashd.perf
	5	try:
	6	import pdm.srv
	7	except:
	8	pdm = None
	9
	10	def usage(out):
	11	out.write("usage: ashd-wsgi3 [-hAL] [-m PDM-SPEC] [-p MODPATH] [-l REQLIMIT] HANDLER-MODULE [ARGS...]\n")
	12
	13	reqlimit = 0
	14	modwsgi_compat = False
	15	setlog = True
	16	opts, args = getopt.getopt(sys.argv[1:], "+hALp:l:m:")
	17	for o, a in opts:
	18	if o == "-h":
	19	usage(sys.stdout)
	20	sys.exit(0)
	21	elif o == "-p":
	22	sys.path.insert(0, a)
	23	elif o == "-L":
	24	setlog = False
	25	elif o == "-A":
	26	modwsgi_compat = True
	27	elif o == "-l":
	28	reqlimit = int(a)
	29	elif o == "-m":
	30	if pdm is not None:
	31	pdm.srv.listen(a)
	32	if len(args) < 1:
	33	usage(sys.stderr)
	34	sys.exit(1)
	35	if setlog:
	36	logging.basicConfig(format="ashd-wsgi3(%(name)s): %(levelname)s: %(message)s")
	37	log = logging.getLogger("ashd-wsgi3")
	38
	39	try:
	40	handlermod = __import__(args[0], fromlist = ["dummy"])
	41	except ImportError as exc:
	42	sys.stderr.write("ashd-wsgi3: handler %s not found: %s\n" % (args[0], exc.args[0]))
	43	sys.exit(1)
	44	if not modwsgi_compat:
	45	if not hasattr(handlermod, "wmain"):
	46	sys.stderr.write("ashd-wsgi3: handler %s has no `wmain' function\n" % args[0])
	47	sys.exit(1)
	48	handler = handlermod.wmain(*args[1:])
	49	else:
	50	if not hasattr(handlermod, "application"):
	51	sys.stderr.write("ashd-wsgi3: handler %s has no `application' object\n" % args[0])
	52	sys.exit(1)
	53	handler = handlermod.application
	54
	55	class closed(IOError):
	56	def __init__(self):
	57	super().__init__("The client has closed the connection.")
	58
	59	cwd = os.getcwd()
	60	def absolutify(path):
	61	if path[0] != '/':
	62	return os.path.join(cwd, path)
	63	return path
	64
	65	def unquoteurl(url):
	66	buf = bytearray()
	67	i = 0
	68	while i < len(url):
	69	c = url[i]
	70	i += 1
	71	if c == ord(b'%'):
	72	if len(url) >= i + 2:
	73	c = 0
	74	if ord(b'0') <= url[i] <= ord(b'9'):
	75	c \|= (url[i] - ord(b'0')) << 4
	76	elif ord(b'a') <= url[i] <= ord(b'f'):
	77	c \|= (url[i] - ord(b'a') + 10) << 4
	78	elif ord(b'A') <= url[i] <= ord(b'F'):
	79	c \|= (url[i] - ord(b'A') + 10) << 4
	80	else:
	81	raise ValueError("Illegal URL escape character")
	82	if ord(b'0') <= url[i + 1] <= ord(b'9'):
	83	c \|= url[i + 1] - ord('0')
	84	elif ord(b'a') <= url[i + 1] <= ord(b'f'):
	85	c \|= url[i + 1] - ord(b'a') + 10
	86	elif ord(b'A') <= url[i + 1] <= ord(b'F'):
	87	c \|= url[i + 1] - ord(b'A') + 10
	88	else:
	89	raise ValueError("Illegal URL escape character")
	90	buf.append(c)
	91	i += 2
	92	else:
	93	raise ValueError("Incomplete URL escape character")
	94	else:
	95	buf.append(c)
	96	return buf
	97
	98	def dowsgi(req):
	99	env = {}
	100	env["wsgi.version"] = 1, 0
	101	for key, val in req.headers:
	102	env["HTTP_" + key.upper().replace(b"-", b"_").decode("latin-1")] = val.decode("latin-1")
	103	env["SERVER_SOFTWARE"] = "ashd-wsgi/1"
	104	env["GATEWAY_INTERFACE"] = "CGI/1.1"
	105	env["SERVER_PROTOCOL"] = req.ver.decode("latin-1")
	106	env["REQUEST_METHOD"] = req.method.decode("latin-1")
	107	try:
	108	rawpi = unquoteurl(req.rest)
	109	except:
	110	rawpi = req.rest
	111	try:
	112	name, rest, pi = (v.decode("utf-8") for v in (req.url, req.rest, rawpi))
	113	env["wsgi.uri_encoding"] = "utf-8"
	114	except UnicodeError as exc:
	115	name, rest, pi = (v.decode("latin-1") for v in (req.url, req.rest, rawpi))
	116	env["wsgi.uri_encoding"] = "latin-1"
	117	env["REQUEST_URI"] = name
	118	p = name.find('?')
	119	if p >= 0:
	120	env["QUERY_STRING"] = name[p + 1:]
	121	name = name[:p]
	122	else:
	123	env["QUERY_STRING"] = ""
	124	if name[-len(rest):] == rest:
	125	# This is the same hack used in call*cgi.
	126	name = name[:-len(rest)]
	127	if name == "/":
	128	# This seems to be normal CGI behavior, but see callcgi.c for
	129	# details.
	130	pi = "/" + pi
	131	name = ""
	132	env["SCRIPT_NAME"] = name
	133	env["PATH_INFO"] = pi
	134	for src, tgt in [("HTTP_HOST", "SERVER_NAME"), ("HTTP_X_ASH_PROTOCOL", "wsgi.url_scheme"),
	135	("HTTP_X_ASH_SERVER_ADDRESS", "SERVER_ADDR"), ("HTTP_X_ASH_SERVER_PORT", "SERVER_PORT"),
	136	("HTTP_X_ASH_ADDRESS", "REMOTE_ADDR"), ("HTTP_X_ASH_PORT", "REMOTE_PORT"),
	137	("HTTP_CONTENT_TYPE", "CONTENT_TYPE"), ("HTTP_CONTENT_LENGTH", "CONTENT_LENGTH")]:
	138	if src in env: env[tgt] = env[src]
	139	for key in ["HTTP_CONTENT_TYPE", "HTTP_CONTENT_LENGTH"]:
	140	# The CGI specification does not strictly require this, but
	141	# many actualy programs and libraries seem to.
	142	if key in env: del env[key]
	143	if "X-Ash-Protocol" in req and req["X-Ash-Protocol"] == b"https": env["HTTPS"] = "on"
	144	if "X-Ash-File" in req: env["SCRIPT_FILENAME"] = absolutify(req["X-Ash-File"].decode(locale.getpreferredencoding()))
	145	env["wsgi.input"] = req.sk
	146	env["wsgi.errors"] = sys.stderr
	147	env["wsgi.multithread"] = True
	148	env["wsgi.multiprocess"] = False
	149	env["wsgi.run_once"] = False
	150
	151	resp = []
	152	respsent = []
	153
	154	def recode(thing):
	155	if isinstance(thing, collections.ByteString):
	156	return thing
	157	else:
	158	return str(thing).encode("latin-1")
	159
	160	def flushreq():
	161	if not respsent:
	162	if not resp:
	163	raise Exception("Trying to write data before starting response.")
	164	status, headers = resp
	165	respsent[:] = [True]
	166	buf = bytearray()
	167	buf += b"HTTP/1.1 " + recode(status) + b"\n"
	168	for nm, val in headers:
	169	buf += recode(nm) + b": " + recode(val) + b"\n"
	170	buf += b"\n"
	171	try:
	172	req.sk.write(buf)
	173	except IOError:
	174	raise closed()
	175
	176	def write(data):
	177	if not data:
	178	return
	179	flushreq()
	180	try:
	181	req.sk.write(data)
	182	req.sk.flush()
	183	except IOError:
	184	raise closed()
	185
	186	def startreq(status, headers, exc_info = None):
	187	if resp:
	188	if exc_info: # Interesting, this...
	189	try:
	190	if respsent:
	191	raise exc_info[1]
	192	finally:
	193	exc_info = None # CPython GC bug?
	194	else:
	195	raise Exception("Can only start responding once.")
	196	resp[:] = status, headers
	197	return write
	198
	199	with ashd.perf.request(env) as reqevent:
	200	try:
	201	respiter = handler(env, startreq)
	202	try:
	203	for data in respiter:
	204	write(data)
	205	if resp:
	206	flushreq()
	207	finally:
	208	if hasattr(respiter, "close"):
	209	respiter.close()
	210	except closed:
	211	pass
	212	if resp:
	213	reqevent.response(resp)
	214
	215	flightlock = threading.Condition()
	216	inflight = 0
	217
	218	class reqthread(threading.Thread):
	219	def __init__(self, req):
	220	super().__init__(name = "Request handler")
	221	self.req = req.dup()
	222
	223	def run(self):
	224	global inflight
	225	try:
	226	with flightlock:
	227	if reqlimit != 0:
	228	start = time.time()
	229	while inflight >= reqlimit:
	230	flightlock.wait(10)
	231	if time.time() - start > 10:
	232	os.abort()
	233	inflight += 1
	234	try:
	235	dowsgi(self.req)
	236	finally:
	237	with flightlock:
	238	inflight -= 1
	239	flightlock.notify()
	240	except:
	241	log.error("exception occurred in handler thread", exc_info=True)
	242	finally:
	243	self.req.close()
	244	sys.stderr.flush()
	245
	246	def handle(req):
	247	reqthread(req).start()
	248
	249	ashd.util.serveloop(handle)