Show More
@@ -1,668 +1,669 | |||||
1 | # hgweb/request.py - An http request from either CGI or the standalone server. |
|
1 | # hgweb/request.py - An http request from either CGI or the standalone server. | |
2 | # |
|
2 | # | |
3 | # Copyright 21 May 2005 - (c) 2005 Jake Edge <jake@edge2.net> |
|
3 | # Copyright 21 May 2005 - (c) 2005 Jake Edge <jake@edge2.net> | |
4 | # Copyright 2005, 2006 Matt Mackall <mpm@selenic.com> |
|
4 | # Copyright 2005, 2006 Matt Mackall <mpm@selenic.com> | |
5 | # |
|
5 | # | |
6 | # This software may be used and distributed according to the terms of the |
|
6 | # This software may be used and distributed according to the terms of the | |
7 | # GNU General Public License version 2 or any later version. |
|
7 | # GNU General Public License version 2 or any later version. | |
8 |
|
8 | |||
9 | from __future__ import absolute_import |
|
9 | from __future__ import absolute_import | |
10 |
|
10 | |||
11 | import errno |
|
11 | import errno | |
12 | import socket |
|
12 | import socket | |
13 | import wsgiref.headers as wsgiheaders |
|
13 | import wsgiref.headers as wsgiheaders | |
14 | #import wsgiref.validate |
|
14 | #import wsgiref.validate | |
15 |
|
15 | |||
16 | from .common import ( |
|
16 | from .common import ( | |
17 | ErrorResponse, |
|
17 | ErrorResponse, | |
18 | statusmessage, |
|
18 | statusmessage, | |
19 | ) |
|
19 | ) | |
20 |
|
20 | |||
21 | from ..thirdparty import ( |
|
21 | from ..thirdparty import ( | |
22 | attr, |
|
22 | attr, | |
23 | ) |
|
23 | ) | |
24 | from .. import ( |
|
24 | from .. import ( | |
25 | error, |
|
25 | error, | |
26 | pycompat, |
|
26 | pycompat, | |
27 | util, |
|
27 | util, | |
28 | ) |
|
28 | ) | |
29 |
|
29 | |||
30 | class multidict(object): |
|
30 | class multidict(object): | |
31 | """A dict like object that can store multiple values for a key. |
|
31 | """A dict like object that can store multiple values for a key. | |
32 |
|
32 | |||
33 | Used to store parsed request parameters. |
|
33 | Used to store parsed request parameters. | |
34 |
|
34 | |||
35 | This is inspired by WebOb's class of the same name. |
|
35 | This is inspired by WebOb's class of the same name. | |
36 | """ |
|
36 | """ | |
37 | def __init__(self): |
|
37 | def __init__(self): | |
38 | # Stores (key, value) 2-tuples. This isn't the most efficient. But we |
|
38 | # Stores (key, value) 2-tuples. This isn't the most efficient. But we | |
39 | # don't rely on parameters that much, so it shouldn't be a perf issue. |
|
39 | # don't rely on parameters that much, so it shouldn't be a perf issue. | |
40 | # we can always add dict for fast lookups. |
|
40 | # we can always add dict for fast lookups. | |
41 | self._items = [] |
|
41 | self._items = [] | |
42 |
|
42 | |||
43 | def __getitem__(self, key): |
|
43 | def __getitem__(self, key): | |
44 | """Returns the last set value for a key.""" |
|
44 | """Returns the last set value for a key.""" | |
45 | for k, v in reversed(self._items): |
|
45 | for k, v in reversed(self._items): | |
46 | if k == key: |
|
46 | if k == key: | |
47 | return v |
|
47 | return v | |
48 |
|
48 | |||
49 | raise KeyError(key) |
|
49 | raise KeyError(key) | |
50 |
|
50 | |||
51 | def __setitem__(self, key, value): |
|
51 | def __setitem__(self, key, value): | |
52 | """Replace a values for a key with a new value.""" |
|
52 | """Replace a values for a key with a new value.""" | |
53 | try: |
|
53 | try: | |
54 | del self[key] |
|
54 | del self[key] | |
55 | except KeyError: |
|
55 | except KeyError: | |
56 | pass |
|
56 | pass | |
57 |
|
57 | |||
58 | self._items.append((key, value)) |
|
58 | self._items.append((key, value)) | |
59 |
|
59 | |||
60 | def __delitem__(self, key): |
|
60 | def __delitem__(self, key): | |
61 | """Delete all values for a key.""" |
|
61 | """Delete all values for a key.""" | |
62 | oldlen = len(self._items) |
|
62 | oldlen = len(self._items) | |
63 |
|
63 | |||
64 | self._items[:] = [(k, v) for k, v in self._items if k != key] |
|
64 | self._items[:] = [(k, v) for k, v in self._items if k != key] | |
65 |
|
65 | |||
66 | if oldlen == len(self._items): |
|
66 | if oldlen == len(self._items): | |
67 | raise KeyError(key) |
|
67 | raise KeyError(key) | |
68 |
|
68 | |||
69 | def __contains__(self, key): |
|
69 | def __contains__(self, key): | |
70 | return any(k == key for k, v in self._items) |
|
70 | return any(k == key for k, v in self._items) | |
71 |
|
71 | |||
72 | def __len__(self): |
|
72 | def __len__(self): | |
73 | return len(self._items) |
|
73 | return len(self._items) | |
74 |
|
74 | |||
75 | def get(self, key, default=None): |
|
75 | def get(self, key, default=None): | |
76 | try: |
|
76 | try: | |
77 | return self.__getitem__(key) |
|
77 | return self.__getitem__(key) | |
78 | except KeyError: |
|
78 | except KeyError: | |
79 | return default |
|
79 | return default | |
80 |
|
80 | |||
81 | def add(self, key, value): |
|
81 | def add(self, key, value): | |
82 | """Add a new value for a key. Does not replace existing values.""" |
|
82 | """Add a new value for a key. Does not replace existing values.""" | |
83 | self._items.append((key, value)) |
|
83 | self._items.append((key, value)) | |
84 |
|
84 | |||
85 | def getall(self, key): |
|
85 | def getall(self, key): | |
86 | """Obtains all values for a key.""" |
|
86 | """Obtains all values for a key.""" | |
87 | return [v for k, v in self._items if k == key] |
|
87 | return [v for k, v in self._items if k == key] | |
88 |
|
88 | |||
89 | def getone(self, key): |
|
89 | def getone(self, key): | |
90 | """Obtain a single value for a key. |
|
90 | """Obtain a single value for a key. | |
91 |
|
91 | |||
92 | Raises KeyError if key not defined or it has multiple values set. |
|
92 | Raises KeyError if key not defined or it has multiple values set. | |
93 | """ |
|
93 | """ | |
94 | vals = self.getall(key) |
|
94 | vals = self.getall(key) | |
95 |
|
95 | |||
96 | if not vals: |
|
96 | if not vals: | |
97 | raise KeyError(key) |
|
97 | raise KeyError(key) | |
98 |
|
98 | |||
99 | if len(vals) > 1: |
|
99 | if len(vals) > 1: | |
100 | raise KeyError('multiple values for %r' % key) |
|
100 | raise KeyError('multiple values for %r' % key) | |
101 |
|
101 | |||
102 | return vals[0] |
|
102 | return vals[0] | |
103 |
|
103 | |||
104 | def asdictoflists(self): |
|
104 | def asdictoflists(self): | |
105 | d = {} |
|
105 | d = {} | |
106 | for k, v in self._items: |
|
106 | for k, v in self._items: | |
107 | if k in d: |
|
107 | if k in d: | |
108 | d[k].append(v) |
|
108 | d[k].append(v) | |
109 | else: |
|
109 | else: | |
110 | d[k] = [v] |
|
110 | d[k] = [v] | |
111 |
|
111 | |||
112 | return d |
|
112 | return d | |
113 |
|
113 | |||
114 | @attr.s(frozen=True) |
|
114 | @attr.s(frozen=True) | |
115 | class parsedrequest(object): |
|
115 | class parsedrequest(object): | |
116 | """Represents a parsed WSGI request. |
|
116 | """Represents a parsed WSGI request. | |
117 |
|
117 | |||
118 | Contains both parsed parameters as well as a handle on the input stream. |
|
118 | Contains both parsed parameters as well as a handle on the input stream. | |
119 | """ |
|
119 | """ | |
120 |
|
120 | |||
121 | # Request method. |
|
121 | # Request method. | |
122 | method = attr.ib() |
|
122 | method = attr.ib() | |
123 | # Full URL for this request. |
|
123 | # Full URL for this request. | |
124 | url = attr.ib() |
|
124 | url = attr.ib() | |
125 | # URL without any path components. Just <proto>://<host><port>. |
|
125 | # URL without any path components. Just <proto>://<host><port>. | |
126 | baseurl = attr.ib() |
|
126 | baseurl = attr.ib() | |
127 | # Advertised URL. Like ``url`` and ``baseurl`` but uses SERVER_NAME instead |
|
127 | # Advertised URL. Like ``url`` and ``baseurl`` but uses SERVER_NAME instead | |
128 | # of HTTP: Host header for hostname. This is likely what clients used. |
|
128 | # of HTTP: Host header for hostname. This is likely what clients used. | |
129 | advertisedurl = attr.ib() |
|
129 | advertisedurl = attr.ib() | |
130 | advertisedbaseurl = attr.ib() |
|
130 | advertisedbaseurl = attr.ib() | |
131 | # URL scheme (part before ``://``). e.g. ``http`` or ``https``. |
|
131 | # URL scheme (part before ``://``). e.g. ``http`` or ``https``. | |
132 | urlscheme = attr.ib() |
|
132 | urlscheme = attr.ib() | |
133 | # Value of REMOTE_USER, if set, or None. |
|
133 | # Value of REMOTE_USER, if set, or None. | |
134 | remoteuser = attr.ib() |
|
134 | remoteuser = attr.ib() | |
135 | # Value of REMOTE_HOST, if set, or None. |
|
135 | # Value of REMOTE_HOST, if set, or None. | |
136 | remotehost = attr.ib() |
|
136 | remotehost = attr.ib() | |
137 | # WSGI application path. |
|
137 | # Relative WSGI application path. If defined, will begin with a | |
|
138 | # ``/``. | |||
138 | apppath = attr.ib() |
|
139 | apppath = attr.ib() | |
139 | # List of path parts to be used for dispatch. |
|
140 | # List of path parts to be used for dispatch. | |
140 | dispatchparts = attr.ib() |
|
141 | dispatchparts = attr.ib() | |
141 | # URL path component (no query string) used for dispatch. Can be |
|
142 | # URL path component (no query string) used for dispatch. Can be | |
142 | # ``None`` to signal no path component given to the request, an |
|
143 | # ``None`` to signal no path component given to the request, an | |
143 | # empty string to signal a request to the application's root URL, |
|
144 | # empty string to signal a request to the application's root URL, | |
144 | # or a string not beginning with ``/`` containing the requested |
|
145 | # or a string not beginning with ``/`` containing the requested | |
145 | # path under the application. |
|
146 | # path under the application. | |
146 | dispatchpath = attr.ib() |
|
147 | dispatchpath = attr.ib() | |
147 | # The name of the repository being accessed. |
|
148 | # The name of the repository being accessed. | |
148 | reponame = attr.ib() |
|
149 | reponame = attr.ib() | |
149 | # Raw query string (part after "?" in URL). |
|
150 | # Raw query string (part after "?" in URL). | |
150 | querystring = attr.ib() |
|
151 | querystring = attr.ib() | |
151 | # multidict of query string parameters. |
|
152 | # multidict of query string parameters. | |
152 | qsparams = attr.ib() |
|
153 | qsparams = attr.ib() | |
153 | # wsgiref.headers.Headers instance. Operates like a dict with case |
|
154 | # wsgiref.headers.Headers instance. Operates like a dict with case | |
154 | # insensitive keys. |
|
155 | # insensitive keys. | |
155 | headers = attr.ib() |
|
156 | headers = attr.ib() | |
156 | # Request body input stream. |
|
157 | # Request body input stream. | |
157 | bodyfh = attr.ib() |
|
158 | bodyfh = attr.ib() | |
158 |
|
159 | |||
159 | def parserequestfromenv(env, bodyfh, reponame=None): |
|
160 | def parserequestfromenv(env, bodyfh, reponame=None): | |
160 | """Parse URL components from environment variables. |
|
161 | """Parse URL components from environment variables. | |
161 |
|
162 | |||
162 | WSGI defines request attributes via environment variables. This function |
|
163 | WSGI defines request attributes via environment variables. This function | |
163 | parses the environment variables into a data structure. |
|
164 | parses the environment variables into a data structure. | |
164 |
|
165 | |||
165 | If ``reponame`` is defined, the leading path components matching that |
|
166 | If ``reponame`` is defined, the leading path components matching that | |
166 | string are effectively shifted from ``PATH_INFO`` to ``SCRIPT_NAME``. |
|
167 | string are effectively shifted from ``PATH_INFO`` to ``SCRIPT_NAME``. | |
167 | This simulates the world view of a WSGI application that processes |
|
168 | This simulates the world view of a WSGI application that processes | |
168 | requests from the base URL of a repo. |
|
169 | requests from the base URL of a repo. | |
169 | """ |
|
170 | """ | |
170 | # PEP-0333 defines the WSGI spec and is a useful reference for this code. |
|
171 | # PEP-0333 defines the WSGI spec and is a useful reference for this code. | |
171 |
|
172 | |||
172 | # We first validate that the incoming object conforms with the WSGI spec. |
|
173 | # We first validate that the incoming object conforms with the WSGI spec. | |
173 | # We only want to be dealing with spec-conforming WSGI implementations. |
|
174 | # We only want to be dealing with spec-conforming WSGI implementations. | |
174 | # TODO enable this once we fix internal violations. |
|
175 | # TODO enable this once we fix internal violations. | |
175 | #wsgiref.validate.check_environ(env) |
|
176 | #wsgiref.validate.check_environ(env) | |
176 |
|
177 | |||
177 | # PEP-0333 states that environment keys and values are native strings |
|
178 | # PEP-0333 states that environment keys and values are native strings | |
178 | # (bytes on Python 2 and str on Python 3). The code points for the Unicode |
|
179 | # (bytes on Python 2 and str on Python 3). The code points for the Unicode | |
179 | # strings on Python 3 must be between \00000-\000FF. We deal with bytes |
|
180 | # strings on Python 3 must be between \00000-\000FF. We deal with bytes | |
180 | # in Mercurial, so mass convert string keys and values to bytes. |
|
181 | # in Mercurial, so mass convert string keys and values to bytes. | |
181 | if pycompat.ispy3: |
|
182 | if pycompat.ispy3: | |
182 | env = {k.encode('latin-1'): v for k, v in env.iteritems()} |
|
183 | env = {k.encode('latin-1'): v for k, v in env.iteritems()} | |
183 | env = {k: v.encode('latin-1') if isinstance(v, str) else v |
|
184 | env = {k: v.encode('latin-1') if isinstance(v, str) else v | |
184 | for k, v in env.iteritems()} |
|
185 | for k, v in env.iteritems()} | |
185 |
|
186 | |||
186 | # https://www.python.org/dev/peps/pep-0333/#environ-variables defines |
|
187 | # https://www.python.org/dev/peps/pep-0333/#environ-variables defines | |
187 | # the environment variables. |
|
188 | # the environment variables. | |
188 | # https://www.python.org/dev/peps/pep-0333/#url-reconstruction defines |
|
189 | # https://www.python.org/dev/peps/pep-0333/#url-reconstruction defines | |
189 | # how URLs are reconstructed. |
|
190 | # how URLs are reconstructed. | |
190 | fullurl = env['wsgi.url_scheme'] + '://' |
|
191 | fullurl = env['wsgi.url_scheme'] + '://' | |
191 | advertisedfullurl = fullurl |
|
192 | advertisedfullurl = fullurl | |
192 |
|
193 | |||
193 | def addport(s): |
|
194 | def addport(s): | |
194 | if env['wsgi.url_scheme'] == 'https': |
|
195 | if env['wsgi.url_scheme'] == 'https': | |
195 | if env['SERVER_PORT'] != '443': |
|
196 | if env['SERVER_PORT'] != '443': | |
196 | s += ':' + env['SERVER_PORT'] |
|
197 | s += ':' + env['SERVER_PORT'] | |
197 | else: |
|
198 | else: | |
198 | if env['SERVER_PORT'] != '80': |
|
199 | if env['SERVER_PORT'] != '80': | |
199 | s += ':' + env['SERVER_PORT'] |
|
200 | s += ':' + env['SERVER_PORT'] | |
200 |
|
201 | |||
201 | return s |
|
202 | return s | |
202 |
|
203 | |||
203 | if env.get('HTTP_HOST'): |
|
204 | if env.get('HTTP_HOST'): | |
204 | fullurl += env['HTTP_HOST'] |
|
205 | fullurl += env['HTTP_HOST'] | |
205 | else: |
|
206 | else: | |
206 | fullurl += env['SERVER_NAME'] |
|
207 | fullurl += env['SERVER_NAME'] | |
207 | fullurl = addport(fullurl) |
|
208 | fullurl = addport(fullurl) | |
208 |
|
209 | |||
209 | advertisedfullurl += env['SERVER_NAME'] |
|
210 | advertisedfullurl += env['SERVER_NAME'] | |
210 | advertisedfullurl = addport(advertisedfullurl) |
|
211 | advertisedfullurl = addport(advertisedfullurl) | |
211 |
|
212 | |||
212 | baseurl = fullurl |
|
213 | baseurl = fullurl | |
213 | advertisedbaseurl = advertisedfullurl |
|
214 | advertisedbaseurl = advertisedfullurl | |
214 |
|
215 | |||
215 | fullurl += util.urlreq.quote(env.get('SCRIPT_NAME', '')) |
|
216 | fullurl += util.urlreq.quote(env.get('SCRIPT_NAME', '')) | |
216 | advertisedfullurl += util.urlreq.quote(env.get('SCRIPT_NAME', '')) |
|
217 | advertisedfullurl += util.urlreq.quote(env.get('SCRIPT_NAME', '')) | |
217 | fullurl += util.urlreq.quote(env.get('PATH_INFO', '')) |
|
218 | fullurl += util.urlreq.quote(env.get('PATH_INFO', '')) | |
218 | advertisedfullurl += util.urlreq.quote(env.get('PATH_INFO', '')) |
|
219 | advertisedfullurl += util.urlreq.quote(env.get('PATH_INFO', '')) | |
219 |
|
220 | |||
220 | if env.get('QUERY_STRING'): |
|
221 | if env.get('QUERY_STRING'): | |
221 | fullurl += '?' + env['QUERY_STRING'] |
|
222 | fullurl += '?' + env['QUERY_STRING'] | |
222 | advertisedfullurl += '?' + env['QUERY_STRING'] |
|
223 | advertisedfullurl += '?' + env['QUERY_STRING'] | |
223 |
|
224 | |||
224 | # If ``reponame`` is defined, that must be a prefix on PATH_INFO |
|
225 | # If ``reponame`` is defined, that must be a prefix on PATH_INFO | |
225 | # that represents the repository being dispatched to. When computing |
|
226 | # that represents the repository being dispatched to. When computing | |
226 | # the dispatch info, we ignore these leading path components. |
|
227 | # the dispatch info, we ignore these leading path components. | |
227 |
|
228 | |||
228 | apppath = env.get('SCRIPT_NAME', '') |
|
229 | apppath = env.get('SCRIPT_NAME', '') | |
229 |
|
230 | |||
230 | if reponame: |
|
231 | if reponame: | |
231 | repoprefix = '/' + reponame.strip('/') |
|
232 | repoprefix = '/' + reponame.strip('/') | |
232 |
|
233 | |||
233 | if not env.get('PATH_INFO'): |
|
234 | if not env.get('PATH_INFO'): | |
234 | raise error.ProgrammingError('reponame requires PATH_INFO') |
|
235 | raise error.ProgrammingError('reponame requires PATH_INFO') | |
235 |
|
236 | |||
236 | if not env['PATH_INFO'].startswith(repoprefix): |
|
237 | if not env['PATH_INFO'].startswith(repoprefix): | |
237 | raise error.ProgrammingError('PATH_INFO does not begin with repo ' |
|
238 | raise error.ProgrammingError('PATH_INFO does not begin with repo ' | |
238 | 'name: %s (%s)' % (env['PATH_INFO'], |
|
239 | 'name: %s (%s)' % (env['PATH_INFO'], | |
239 | reponame)) |
|
240 | reponame)) | |
240 |
|
241 | |||
241 | dispatchpath = env['PATH_INFO'][len(repoprefix):] |
|
242 | dispatchpath = env['PATH_INFO'][len(repoprefix):] | |
242 |
|
243 | |||
243 | if dispatchpath and not dispatchpath.startswith('/'): |
|
244 | if dispatchpath and not dispatchpath.startswith('/'): | |
244 | raise error.ProgrammingError('reponame prefix of PATH_INFO does ' |
|
245 | raise error.ProgrammingError('reponame prefix of PATH_INFO does ' | |
245 | 'not end at path delimiter: %s (%s)' % |
|
246 | 'not end at path delimiter: %s (%s)' % | |
246 | (env['PATH_INFO'], reponame)) |
|
247 | (env['PATH_INFO'], reponame)) | |
247 |
|
248 | |||
248 | apppath = apppath.rstrip('/') + repoprefix |
|
249 | apppath = apppath.rstrip('/') + repoprefix | |
249 | dispatchparts = dispatchpath.strip('/').split('/') |
|
250 | dispatchparts = dispatchpath.strip('/').split('/') | |
250 | dispatchpath = '/'.join(dispatchparts) |
|
251 | dispatchpath = '/'.join(dispatchparts) | |
251 |
|
252 | |||
252 | elif 'PATH_INFO' in env: |
|
253 | elif 'PATH_INFO' in env: | |
253 | if env['PATH_INFO'].strip('/'): |
|
254 | if env['PATH_INFO'].strip('/'): | |
254 | dispatchparts = env['PATH_INFO'].strip('/').split('/') |
|
255 | dispatchparts = env['PATH_INFO'].strip('/').split('/') | |
255 | dispatchpath = '/'.join(dispatchparts) |
|
256 | dispatchpath = '/'.join(dispatchparts) | |
256 | else: |
|
257 | else: | |
257 | dispatchparts = [] |
|
258 | dispatchparts = [] | |
258 | dispatchpath = '' |
|
259 | dispatchpath = '' | |
259 | else: |
|
260 | else: | |
260 | dispatchparts = [] |
|
261 | dispatchparts = [] | |
261 | dispatchpath = None |
|
262 | dispatchpath = None | |
262 |
|
263 | |||
263 | querystring = env.get('QUERY_STRING', '') |
|
264 | querystring = env.get('QUERY_STRING', '') | |
264 |
|
265 | |||
265 | # We store as a list so we have ordering information. We also store as |
|
266 | # We store as a list so we have ordering information. We also store as | |
266 | # a dict to facilitate fast lookup. |
|
267 | # a dict to facilitate fast lookup. | |
267 | qsparams = multidict() |
|
268 | qsparams = multidict() | |
268 | for k, v in util.urlreq.parseqsl(querystring, keep_blank_values=True): |
|
269 | for k, v in util.urlreq.parseqsl(querystring, keep_blank_values=True): | |
269 | qsparams.add(k, v) |
|
270 | qsparams.add(k, v) | |
270 |
|
271 | |||
271 | # HTTP_* keys contain HTTP request headers. The Headers structure should |
|
272 | # HTTP_* keys contain HTTP request headers. The Headers structure should | |
272 | # perform case normalization for us. We just rewrite underscore to dash |
|
273 | # perform case normalization for us. We just rewrite underscore to dash | |
273 | # so keys match what likely went over the wire. |
|
274 | # so keys match what likely went over the wire. | |
274 | headers = [] |
|
275 | headers = [] | |
275 | for k, v in env.iteritems(): |
|
276 | for k, v in env.iteritems(): | |
276 | if k.startswith('HTTP_'): |
|
277 | if k.startswith('HTTP_'): | |
277 | headers.append((k[len('HTTP_'):].replace('_', '-'), v)) |
|
278 | headers.append((k[len('HTTP_'):].replace('_', '-'), v)) | |
278 |
|
279 | |||
279 | headers = wsgiheaders.Headers(headers) |
|
280 | headers = wsgiheaders.Headers(headers) | |
280 |
|
281 | |||
281 | # This is kind of a lie because the HTTP header wasn't explicitly |
|
282 | # This is kind of a lie because the HTTP header wasn't explicitly | |
282 | # sent. But for all intents and purposes it should be OK to lie about |
|
283 | # sent. But for all intents and purposes it should be OK to lie about | |
283 | # this, since a consumer will either either value to determine how many |
|
284 | # this, since a consumer will either either value to determine how many | |
284 | # bytes are available to read. |
|
285 | # bytes are available to read. | |
285 | if 'CONTENT_LENGTH' in env and 'HTTP_CONTENT_LENGTH' not in env: |
|
286 | if 'CONTENT_LENGTH' in env and 'HTTP_CONTENT_LENGTH' not in env: | |
286 | headers['Content-Length'] = env['CONTENT_LENGTH'] |
|
287 | headers['Content-Length'] = env['CONTENT_LENGTH'] | |
287 |
|
288 | |||
288 | # TODO do this once we remove wsgirequest.inp, otherwise we could have |
|
289 | # TODO do this once we remove wsgirequest.inp, otherwise we could have | |
289 | # multiple readers from the underlying input stream. |
|
290 | # multiple readers from the underlying input stream. | |
290 | #bodyfh = env['wsgi.input'] |
|
291 | #bodyfh = env['wsgi.input'] | |
291 | #if 'Content-Length' in headers: |
|
292 | #if 'Content-Length' in headers: | |
292 | # bodyfh = util.cappedreader(bodyfh, int(headers['Content-Length'])) |
|
293 | # bodyfh = util.cappedreader(bodyfh, int(headers['Content-Length'])) | |
293 |
|
294 | |||
294 | return parsedrequest(method=env['REQUEST_METHOD'], |
|
295 | return parsedrequest(method=env['REQUEST_METHOD'], | |
295 | url=fullurl, baseurl=baseurl, |
|
296 | url=fullurl, baseurl=baseurl, | |
296 | advertisedurl=advertisedfullurl, |
|
297 | advertisedurl=advertisedfullurl, | |
297 | advertisedbaseurl=advertisedbaseurl, |
|
298 | advertisedbaseurl=advertisedbaseurl, | |
298 | urlscheme=env['wsgi.url_scheme'], |
|
299 | urlscheme=env['wsgi.url_scheme'], | |
299 | remoteuser=env.get('REMOTE_USER'), |
|
300 | remoteuser=env.get('REMOTE_USER'), | |
300 | remotehost=env.get('REMOTE_HOST'), |
|
301 | remotehost=env.get('REMOTE_HOST'), | |
301 | apppath=apppath, |
|
302 | apppath=apppath, | |
302 | dispatchparts=dispatchparts, dispatchpath=dispatchpath, |
|
303 | dispatchparts=dispatchparts, dispatchpath=dispatchpath, | |
303 | reponame=reponame, |
|
304 | reponame=reponame, | |
304 | querystring=querystring, |
|
305 | querystring=querystring, | |
305 | qsparams=qsparams, |
|
306 | qsparams=qsparams, | |
306 | headers=headers, |
|
307 | headers=headers, | |
307 | bodyfh=bodyfh) |
|
308 | bodyfh=bodyfh) | |
308 |
|
309 | |||
309 | class offsettrackingwriter(object): |
|
310 | class offsettrackingwriter(object): | |
310 | """A file object like object that is append only and tracks write count. |
|
311 | """A file object like object that is append only and tracks write count. | |
311 |
|
312 | |||
312 | Instances are bound to a callable. This callable is called with data |
|
313 | Instances are bound to a callable. This callable is called with data | |
313 | whenever a ``write()`` is attempted. |
|
314 | whenever a ``write()`` is attempted. | |
314 |
|
315 | |||
315 | Instances track the amount of written data so they can answer ``tell()`` |
|
316 | Instances track the amount of written data so they can answer ``tell()`` | |
316 | requests. |
|
317 | requests. | |
317 |
|
318 | |||
318 | The intent of this class is to wrap the ``write()`` function returned by |
|
319 | The intent of this class is to wrap the ``write()`` function returned by | |
319 | a WSGI ``start_response()`` function. Since ``write()`` is a callable and |
|
320 | a WSGI ``start_response()`` function. Since ``write()`` is a callable and | |
320 | not a file object, it doesn't implement other file object methods. |
|
321 | not a file object, it doesn't implement other file object methods. | |
321 | """ |
|
322 | """ | |
322 | def __init__(self, writefn): |
|
323 | def __init__(self, writefn): | |
323 | self._write = writefn |
|
324 | self._write = writefn | |
324 | self._offset = 0 |
|
325 | self._offset = 0 | |
325 |
|
326 | |||
326 | def write(self, s): |
|
327 | def write(self, s): | |
327 | res = self._write(s) |
|
328 | res = self._write(s) | |
328 | # Some Python objects don't report the number of bytes written. |
|
329 | # Some Python objects don't report the number of bytes written. | |
329 | if res is None: |
|
330 | if res is None: | |
330 | self._offset += len(s) |
|
331 | self._offset += len(s) | |
331 | else: |
|
332 | else: | |
332 | self._offset += res |
|
333 | self._offset += res | |
333 |
|
334 | |||
334 | def flush(self): |
|
335 | def flush(self): | |
335 | pass |
|
336 | pass | |
336 |
|
337 | |||
337 | def tell(self): |
|
338 | def tell(self): | |
338 | return self._offset |
|
339 | return self._offset | |
339 |
|
340 | |||
340 | class wsgiresponse(object): |
|
341 | class wsgiresponse(object): | |
341 | """Represents a response to a WSGI request. |
|
342 | """Represents a response to a WSGI request. | |
342 |
|
343 | |||
343 | A response consists of a status line, headers, and a body. |
|
344 | A response consists of a status line, headers, and a body. | |
344 |
|
345 | |||
345 | Consumers must populate the ``status`` and ``headers`` fields and |
|
346 | Consumers must populate the ``status`` and ``headers`` fields and | |
346 | make a call to a ``setbody*()`` method before the response can be |
|
347 | make a call to a ``setbody*()`` method before the response can be | |
347 | issued. |
|
348 | issued. | |
348 |
|
349 | |||
349 | When it is time to start sending the response over the wire, |
|
350 | When it is time to start sending the response over the wire, | |
350 | ``sendresponse()`` is called. It handles emitting the header portion |
|
351 | ``sendresponse()`` is called. It handles emitting the header portion | |
351 | of the response message. It then yields chunks of body data to be |
|
352 | of the response message. It then yields chunks of body data to be | |
352 | written to the peer. Typically, the WSGI application itself calls |
|
353 | written to the peer. Typically, the WSGI application itself calls | |
353 | and returns the value from ``sendresponse()``. |
|
354 | and returns the value from ``sendresponse()``. | |
354 | """ |
|
355 | """ | |
355 |
|
356 | |||
356 | def __init__(self, req, startresponse): |
|
357 | def __init__(self, req, startresponse): | |
357 | """Create an empty response tied to a specific request. |
|
358 | """Create an empty response tied to a specific request. | |
358 |
|
359 | |||
359 | ``req`` is a ``parsedrequest``. ``startresponse`` is the |
|
360 | ``req`` is a ``parsedrequest``. ``startresponse`` is the | |
360 | ``start_response`` function passed to the WSGI application. |
|
361 | ``start_response`` function passed to the WSGI application. | |
361 | """ |
|
362 | """ | |
362 | self._req = req |
|
363 | self._req = req | |
363 | self._startresponse = startresponse |
|
364 | self._startresponse = startresponse | |
364 |
|
365 | |||
365 | self.status = None |
|
366 | self.status = None | |
366 | self.headers = wsgiheaders.Headers([]) |
|
367 | self.headers = wsgiheaders.Headers([]) | |
367 |
|
368 | |||
368 | self._bodybytes = None |
|
369 | self._bodybytes = None | |
369 | self._bodygen = None |
|
370 | self._bodygen = None | |
370 | self._bodywillwrite = False |
|
371 | self._bodywillwrite = False | |
371 | self._started = False |
|
372 | self._started = False | |
372 | self._bodywritefn = None |
|
373 | self._bodywritefn = None | |
373 |
|
374 | |||
374 | def _verifybody(self): |
|
375 | def _verifybody(self): | |
375 | if (self._bodybytes is not None or self._bodygen is not None |
|
376 | if (self._bodybytes is not None or self._bodygen is not None | |
376 | or self._bodywillwrite): |
|
377 | or self._bodywillwrite): | |
377 | raise error.ProgrammingError('cannot define body multiple times') |
|
378 | raise error.ProgrammingError('cannot define body multiple times') | |
378 |
|
379 | |||
379 | def setbodybytes(self, b): |
|
380 | def setbodybytes(self, b): | |
380 | """Define the response body as static bytes. |
|
381 | """Define the response body as static bytes. | |
381 |
|
382 | |||
382 | The empty string signals that there is no response body. |
|
383 | The empty string signals that there is no response body. | |
383 | """ |
|
384 | """ | |
384 | self._verifybody() |
|
385 | self._verifybody() | |
385 | self._bodybytes = b |
|
386 | self._bodybytes = b | |
386 | self.headers['Content-Length'] = '%d' % len(b) |
|
387 | self.headers['Content-Length'] = '%d' % len(b) | |
387 |
|
388 | |||
388 | def setbodygen(self, gen): |
|
389 | def setbodygen(self, gen): | |
389 | """Define the response body as a generator of bytes.""" |
|
390 | """Define the response body as a generator of bytes.""" | |
390 | self._verifybody() |
|
391 | self._verifybody() | |
391 | self._bodygen = gen |
|
392 | self._bodygen = gen | |
392 |
|
393 | |||
393 | def setbodywillwrite(self): |
|
394 | def setbodywillwrite(self): | |
394 | """Signal an intent to use write() to emit the response body. |
|
395 | """Signal an intent to use write() to emit the response body. | |
395 |
|
396 | |||
396 | **This is the least preferred way to send a body.** |
|
397 | **This is the least preferred way to send a body.** | |
397 |
|
398 | |||
398 | It is preferred for WSGI applications to emit a generator of chunks |
|
399 | It is preferred for WSGI applications to emit a generator of chunks | |
399 | constituting the response body. However, some consumers can't emit |
|
400 | constituting the response body. However, some consumers can't emit | |
400 | data this way. So, WSGI provides a way to obtain a ``write(data)`` |
|
401 | data this way. So, WSGI provides a way to obtain a ``write(data)`` | |
401 | function that can be used to synchronously perform an unbuffered |
|
402 | function that can be used to synchronously perform an unbuffered | |
402 | write. |
|
403 | write. | |
403 |
|
404 | |||
404 | Calling this function signals an intent to produce the body in this |
|
405 | Calling this function signals an intent to produce the body in this | |
405 | manner. |
|
406 | manner. | |
406 | """ |
|
407 | """ | |
407 | self._verifybody() |
|
408 | self._verifybody() | |
408 | self._bodywillwrite = True |
|
409 | self._bodywillwrite = True | |
409 |
|
410 | |||
410 | def sendresponse(self): |
|
411 | def sendresponse(self): | |
411 | """Send the generated response to the client. |
|
412 | """Send the generated response to the client. | |
412 |
|
413 | |||
413 | Before this is called, ``status`` must be set and one of |
|
414 | Before this is called, ``status`` must be set and one of | |
414 | ``setbodybytes()`` or ``setbodygen()`` must be called. |
|
415 | ``setbodybytes()`` or ``setbodygen()`` must be called. | |
415 |
|
416 | |||
416 | Calling this method multiple times is not allowed. |
|
417 | Calling this method multiple times is not allowed. | |
417 | """ |
|
418 | """ | |
418 | if self._started: |
|
419 | if self._started: | |
419 | raise error.ProgrammingError('sendresponse() called multiple times') |
|
420 | raise error.ProgrammingError('sendresponse() called multiple times') | |
420 |
|
421 | |||
421 | self._started = True |
|
422 | self._started = True | |
422 |
|
423 | |||
423 | if not self.status: |
|
424 | if not self.status: | |
424 | raise error.ProgrammingError('status line not defined') |
|
425 | raise error.ProgrammingError('status line not defined') | |
425 |
|
426 | |||
426 | if (self._bodybytes is None and self._bodygen is None |
|
427 | if (self._bodybytes is None and self._bodygen is None | |
427 | and not self._bodywillwrite): |
|
428 | and not self._bodywillwrite): | |
428 | raise error.ProgrammingError('response body not defined') |
|
429 | raise error.ProgrammingError('response body not defined') | |
429 |
|
430 | |||
430 | # RFC 7232 Section 4.1 states that a 304 MUST generate one of |
|
431 | # RFC 7232 Section 4.1 states that a 304 MUST generate one of | |
431 | # {Cache-Control, Content-Location, Date, ETag, Expires, Vary} |
|
432 | # {Cache-Control, Content-Location, Date, ETag, Expires, Vary} | |
432 | # and SHOULD NOT generate other headers unless they could be used |
|
433 | # and SHOULD NOT generate other headers unless they could be used | |
433 | # to guide cache updates. Furthermore, RFC 7230 Section 3.3.2 |
|
434 | # to guide cache updates. Furthermore, RFC 7230 Section 3.3.2 | |
434 | # states that no response body can be issued. Content-Length can |
|
435 | # states that no response body can be issued. Content-Length can | |
435 | # be sent. But if it is present, it should be the size of the response |
|
436 | # be sent. But if it is present, it should be the size of the response | |
436 | # that wasn't transferred. |
|
437 | # that wasn't transferred. | |
437 | if self.status.startswith('304 '): |
|
438 | if self.status.startswith('304 '): | |
438 | # setbodybytes('') will set C-L to 0. This doesn't conform with the |
|
439 | # setbodybytes('') will set C-L to 0. This doesn't conform with the | |
439 | # spec. So remove it. |
|
440 | # spec. So remove it. | |
440 | if self.headers.get('Content-Length') == '0': |
|
441 | if self.headers.get('Content-Length') == '0': | |
441 | del self.headers['Content-Length'] |
|
442 | del self.headers['Content-Length'] | |
442 |
|
443 | |||
443 | # Strictly speaking, this is too strict. But until it causes |
|
444 | # Strictly speaking, this is too strict. But until it causes | |
444 | # problems, let's be strict. |
|
445 | # problems, let's be strict. | |
445 | badheaders = {k for k in self.headers.keys() |
|
446 | badheaders = {k for k in self.headers.keys() | |
446 | if k.lower() not in ('date', 'etag', 'expires', |
|
447 | if k.lower() not in ('date', 'etag', 'expires', | |
447 | 'cache-control', |
|
448 | 'cache-control', | |
448 | 'content-location', |
|
449 | 'content-location', | |
449 | 'vary')} |
|
450 | 'vary')} | |
450 | if badheaders: |
|
451 | if badheaders: | |
451 | raise error.ProgrammingError( |
|
452 | raise error.ProgrammingError( | |
452 | 'illegal header on 304 response: %s' % |
|
453 | 'illegal header on 304 response: %s' % | |
453 | ', '.join(sorted(badheaders))) |
|
454 | ', '.join(sorted(badheaders))) | |
454 |
|
455 | |||
455 | if self._bodygen is not None or self._bodywillwrite: |
|
456 | if self._bodygen is not None or self._bodywillwrite: | |
456 | raise error.ProgrammingError("must use setbodybytes('') with " |
|
457 | raise error.ProgrammingError("must use setbodybytes('') with " | |
457 | "304 responses") |
|
458 | "304 responses") | |
458 |
|
459 | |||
459 | # Various HTTP clients (notably httplib) won't read the HTTP response |
|
460 | # Various HTTP clients (notably httplib) won't read the HTTP response | |
460 | # until the HTTP request has been sent in full. If servers (us) send a |
|
461 | # until the HTTP request has been sent in full. If servers (us) send a | |
461 | # response before the HTTP request has been fully sent, the connection |
|
462 | # response before the HTTP request has been fully sent, the connection | |
462 | # may deadlock because neither end is reading. |
|
463 | # may deadlock because neither end is reading. | |
463 | # |
|
464 | # | |
464 | # We work around this by "draining" the request data before |
|
465 | # We work around this by "draining" the request data before | |
465 | # sending any response in some conditions. |
|
466 | # sending any response in some conditions. | |
466 | drain = False |
|
467 | drain = False | |
467 | close = False |
|
468 | close = False | |
468 |
|
469 | |||
469 | # If the client sent Expect: 100-continue, we assume it is smart enough |
|
470 | # If the client sent Expect: 100-continue, we assume it is smart enough | |
470 | # to deal with the server sending a response before reading the request. |
|
471 | # to deal with the server sending a response before reading the request. | |
471 | # (httplib doesn't do this.) |
|
472 | # (httplib doesn't do this.) | |
472 | if self._req.headers.get('Expect', '').lower() == '100-continue': |
|
473 | if self._req.headers.get('Expect', '').lower() == '100-continue': | |
473 | pass |
|
474 | pass | |
474 | # Only tend to request methods that have bodies. Strictly speaking, |
|
475 | # Only tend to request methods that have bodies. Strictly speaking, | |
475 | # we should sniff for a body. But this is fine for our existing |
|
476 | # we should sniff for a body. But this is fine for our existing | |
476 | # WSGI applications. |
|
477 | # WSGI applications. | |
477 | elif self._req.method not in ('POST', 'PUT'): |
|
478 | elif self._req.method not in ('POST', 'PUT'): | |
478 | pass |
|
479 | pass | |
479 | else: |
|
480 | else: | |
480 | # If we don't know how much data to read, there's no guarantee |
|
481 | # If we don't know how much data to read, there's no guarantee | |
481 | # that we can drain the request responsibly. The WSGI |
|
482 | # that we can drain the request responsibly. The WSGI | |
482 | # specification only says that servers *should* ensure the |
|
483 | # specification only says that servers *should* ensure the | |
483 | # input stream doesn't overrun the actual request. So there's |
|
484 | # input stream doesn't overrun the actual request. So there's | |
484 | # no guarantee that reading until EOF won't corrupt the stream |
|
485 | # no guarantee that reading until EOF won't corrupt the stream | |
485 | # state. |
|
486 | # state. | |
486 | if not isinstance(self._req.bodyfh, util.cappedreader): |
|
487 | if not isinstance(self._req.bodyfh, util.cappedreader): | |
487 | close = True |
|
488 | close = True | |
488 | else: |
|
489 | else: | |
489 | # We /could/ only drain certain HTTP response codes. But 200 and |
|
490 | # We /could/ only drain certain HTTP response codes. But 200 and | |
490 | # non-200 wire protocol responses both require draining. Since |
|
491 | # non-200 wire protocol responses both require draining. Since | |
491 | # we have a capped reader in place for all situations where we |
|
492 | # we have a capped reader in place for all situations where we | |
492 | # drain, it is safe to read from that stream. We'll either do |
|
493 | # drain, it is safe to read from that stream. We'll either do | |
493 | # a drain or no-op if we're already at EOF. |
|
494 | # a drain or no-op if we're already at EOF. | |
494 | drain = True |
|
495 | drain = True | |
495 |
|
496 | |||
496 | if close: |
|
497 | if close: | |
497 | self.headers['Connection'] = 'Close' |
|
498 | self.headers['Connection'] = 'Close' | |
498 |
|
499 | |||
499 | if drain: |
|
500 | if drain: | |
500 | assert isinstance(self._req.bodyfh, util.cappedreader) |
|
501 | assert isinstance(self._req.bodyfh, util.cappedreader) | |
501 | while True: |
|
502 | while True: | |
502 | chunk = self._req.bodyfh.read(32768) |
|
503 | chunk = self._req.bodyfh.read(32768) | |
503 | if not chunk: |
|
504 | if not chunk: | |
504 | break |
|
505 | break | |
505 |
|
506 | |||
506 | write = self._startresponse(pycompat.sysstr(self.status), |
|
507 | write = self._startresponse(pycompat.sysstr(self.status), | |
507 | self.headers.items()) |
|
508 | self.headers.items()) | |
508 |
|
509 | |||
509 | if self._bodybytes: |
|
510 | if self._bodybytes: | |
510 | yield self._bodybytes |
|
511 | yield self._bodybytes | |
511 | elif self._bodygen: |
|
512 | elif self._bodygen: | |
512 | for chunk in self._bodygen: |
|
513 | for chunk in self._bodygen: | |
513 | yield chunk |
|
514 | yield chunk | |
514 | elif self._bodywillwrite: |
|
515 | elif self._bodywillwrite: | |
515 | self._bodywritefn = write |
|
516 | self._bodywritefn = write | |
516 | else: |
|
517 | else: | |
517 | error.ProgrammingError('do not know how to send body') |
|
518 | error.ProgrammingError('do not know how to send body') | |
518 |
|
519 | |||
519 | def getbodyfile(self): |
|
520 | def getbodyfile(self): | |
520 | """Obtain a file object like object representing the response body. |
|
521 | """Obtain a file object like object representing the response body. | |
521 |
|
522 | |||
522 | For this to work, you must call ``setbodywillwrite()`` and then |
|
523 | For this to work, you must call ``setbodywillwrite()`` and then | |
523 | ``sendresponse()`` first. ``sendresponse()`` is a generator and the |
|
524 | ``sendresponse()`` first. ``sendresponse()`` is a generator and the | |
524 | function won't run to completion unless the generator is advanced. The |
|
525 | function won't run to completion unless the generator is advanced. The | |
525 | generator yields not items. The easiest way to consume it is with |
|
526 | generator yields not items. The easiest way to consume it is with | |
526 | ``list(res.sendresponse())``, which should resolve to an empty list - |
|
527 | ``list(res.sendresponse())``, which should resolve to an empty list - | |
527 | ``[]``. |
|
528 | ``[]``. | |
528 | """ |
|
529 | """ | |
529 | if not self._bodywillwrite: |
|
530 | if not self._bodywillwrite: | |
530 | raise error.ProgrammingError('must call setbodywillwrite() first') |
|
531 | raise error.ProgrammingError('must call setbodywillwrite() first') | |
531 |
|
532 | |||
532 | if not self._started: |
|
533 | if not self._started: | |
533 | raise error.ProgrammingError('must call sendresponse() first; did ' |
|
534 | raise error.ProgrammingError('must call sendresponse() first; did ' | |
534 | 'you remember to consume it since it ' |
|
535 | 'you remember to consume it since it ' | |
535 | 'is a generator?') |
|
536 | 'is a generator?') | |
536 |
|
537 | |||
537 | assert self._bodywritefn |
|
538 | assert self._bodywritefn | |
538 | return offsettrackingwriter(self._bodywritefn) |
|
539 | return offsettrackingwriter(self._bodywritefn) | |
539 |
|
540 | |||
540 | class wsgirequest(object): |
|
541 | class wsgirequest(object): | |
541 | """Higher-level API for a WSGI request. |
|
542 | """Higher-level API for a WSGI request. | |
542 |
|
543 | |||
543 | WSGI applications are invoked with 2 arguments. They are used to |
|
544 | WSGI applications are invoked with 2 arguments. They are used to | |
544 | instantiate instances of this class, which provides higher-level APIs |
|
545 | instantiate instances of this class, which provides higher-level APIs | |
545 | for obtaining request parameters, writing HTTP output, etc. |
|
546 | for obtaining request parameters, writing HTTP output, etc. | |
546 | """ |
|
547 | """ | |
547 | def __init__(self, wsgienv, start_response): |
|
548 | def __init__(self, wsgienv, start_response): | |
548 | version = wsgienv[r'wsgi.version'] |
|
549 | version = wsgienv[r'wsgi.version'] | |
549 | if (version < (1, 0)) or (version >= (2, 0)): |
|
550 | if (version < (1, 0)) or (version >= (2, 0)): | |
550 | raise RuntimeError("Unknown and unsupported WSGI version %d.%d" |
|
551 | raise RuntimeError("Unknown and unsupported WSGI version %d.%d" | |
551 | % version) |
|
552 | % version) | |
552 |
|
553 | |||
553 | inp = wsgienv[r'wsgi.input'] |
|
554 | inp = wsgienv[r'wsgi.input'] | |
554 |
|
555 | |||
555 | if r'HTTP_CONTENT_LENGTH' in wsgienv: |
|
556 | if r'HTTP_CONTENT_LENGTH' in wsgienv: | |
556 | inp = util.cappedreader(inp, int(wsgienv[r'HTTP_CONTENT_LENGTH'])) |
|
557 | inp = util.cappedreader(inp, int(wsgienv[r'HTTP_CONTENT_LENGTH'])) | |
557 | elif r'CONTENT_LENGTH' in wsgienv: |
|
558 | elif r'CONTENT_LENGTH' in wsgienv: | |
558 | inp = util.cappedreader(inp, int(wsgienv[r'CONTENT_LENGTH'])) |
|
559 | inp = util.cappedreader(inp, int(wsgienv[r'CONTENT_LENGTH'])) | |
559 |
|
560 | |||
560 | self.err = wsgienv[r'wsgi.errors'] |
|
561 | self.err = wsgienv[r'wsgi.errors'] | |
561 | self.threaded = wsgienv[r'wsgi.multithread'] |
|
562 | self.threaded = wsgienv[r'wsgi.multithread'] | |
562 | self.multiprocess = wsgienv[r'wsgi.multiprocess'] |
|
563 | self.multiprocess = wsgienv[r'wsgi.multiprocess'] | |
563 | self.run_once = wsgienv[r'wsgi.run_once'] |
|
564 | self.run_once = wsgienv[r'wsgi.run_once'] | |
564 | self.env = wsgienv |
|
565 | self.env = wsgienv | |
565 | self.req = parserequestfromenv(wsgienv, inp) |
|
566 | self.req = parserequestfromenv(wsgienv, inp) | |
566 | self.res = wsgiresponse(self.req, start_response) |
|
567 | self.res = wsgiresponse(self.req, start_response) | |
567 | self._start_response = start_response |
|
568 | self._start_response = start_response | |
568 | self.server_write = None |
|
569 | self.server_write = None | |
569 | self.headers = [] |
|
570 | self.headers = [] | |
570 |
|
571 | |||
571 | def respond(self, status, type, filename=None, body=None): |
|
572 | def respond(self, status, type, filename=None, body=None): | |
572 | if not isinstance(type, str): |
|
573 | if not isinstance(type, str): | |
573 | type = pycompat.sysstr(type) |
|
574 | type = pycompat.sysstr(type) | |
574 | if self._start_response is not None: |
|
575 | if self._start_response is not None: | |
575 | self.headers.append((r'Content-Type', type)) |
|
576 | self.headers.append((r'Content-Type', type)) | |
576 | if filename: |
|
577 | if filename: | |
577 | filename = (filename.rpartition('/')[-1] |
|
578 | filename = (filename.rpartition('/')[-1] | |
578 | .replace('\\', '\\\\').replace('"', '\\"')) |
|
579 | .replace('\\', '\\\\').replace('"', '\\"')) | |
579 | self.headers.append(('Content-Disposition', |
|
580 | self.headers.append(('Content-Disposition', | |
580 | 'inline; filename="%s"' % filename)) |
|
581 | 'inline; filename="%s"' % filename)) | |
581 | if body is not None: |
|
582 | if body is not None: | |
582 | self.headers.append((r'Content-Length', str(len(body)))) |
|
583 | self.headers.append((r'Content-Length', str(len(body)))) | |
583 |
|
584 | |||
584 | for k, v in self.headers: |
|
585 | for k, v in self.headers: | |
585 | if not isinstance(v, str): |
|
586 | if not isinstance(v, str): | |
586 | raise TypeError('header value must be string: %r' % (v,)) |
|
587 | raise TypeError('header value must be string: %r' % (v,)) | |
587 |
|
588 | |||
588 | if isinstance(status, ErrorResponse): |
|
589 | if isinstance(status, ErrorResponse): | |
589 | self.headers.extend(status.headers) |
|
590 | self.headers.extend(status.headers) | |
590 | status = statusmessage(status.code, pycompat.bytestr(status)) |
|
591 | status = statusmessage(status.code, pycompat.bytestr(status)) | |
591 | elif status == 200: |
|
592 | elif status == 200: | |
592 | status = '200 Script output follows' |
|
593 | status = '200 Script output follows' | |
593 | elif isinstance(status, int): |
|
594 | elif isinstance(status, int): | |
594 | status = statusmessage(status) |
|
595 | status = statusmessage(status) | |
595 |
|
596 | |||
596 | # Various HTTP clients (notably httplib) won't read the HTTP |
|
597 | # Various HTTP clients (notably httplib) won't read the HTTP | |
597 | # response until the HTTP request has been sent in full. If servers |
|
598 | # response until the HTTP request has been sent in full. If servers | |
598 | # (us) send a response before the HTTP request has been fully sent, |
|
599 | # (us) send a response before the HTTP request has been fully sent, | |
599 | # the connection may deadlock because neither end is reading. |
|
600 | # the connection may deadlock because neither end is reading. | |
600 | # |
|
601 | # | |
601 | # We work around this by "draining" the request data before |
|
602 | # We work around this by "draining" the request data before | |
602 | # sending any response in some conditions. |
|
603 | # sending any response in some conditions. | |
603 | drain = False |
|
604 | drain = False | |
604 | close = False |
|
605 | close = False | |
605 |
|
606 | |||
606 | # If the client sent Expect: 100-continue, we assume it is smart |
|
607 | # If the client sent Expect: 100-continue, we assume it is smart | |
607 | # enough to deal with the server sending a response before reading |
|
608 | # enough to deal with the server sending a response before reading | |
608 | # the request. (httplib doesn't do this.) |
|
609 | # the request. (httplib doesn't do this.) | |
609 | if self.env.get(r'HTTP_EXPECT', r'').lower() == r'100-continue': |
|
610 | if self.env.get(r'HTTP_EXPECT', r'').lower() == r'100-continue': | |
610 | pass |
|
611 | pass | |
611 | # Only tend to request methods that have bodies. Strictly speaking, |
|
612 | # Only tend to request methods that have bodies. Strictly speaking, | |
612 | # we should sniff for a body. But this is fine for our existing |
|
613 | # we should sniff for a body. But this is fine for our existing | |
613 | # WSGI applications. |
|
614 | # WSGI applications. | |
614 | elif self.env[r'REQUEST_METHOD'] not in (r'POST', r'PUT'): |
|
615 | elif self.env[r'REQUEST_METHOD'] not in (r'POST', r'PUT'): | |
615 | pass |
|
616 | pass | |
616 | else: |
|
617 | else: | |
617 | # If we don't know how much data to read, there's no guarantee |
|
618 | # If we don't know how much data to read, there's no guarantee | |
618 | # that we can drain the request responsibly. The WSGI |
|
619 | # that we can drain the request responsibly. The WSGI | |
619 | # specification only says that servers *should* ensure the |
|
620 | # specification only says that servers *should* ensure the | |
620 | # input stream doesn't overrun the actual request. So there's |
|
621 | # input stream doesn't overrun the actual request. So there's | |
621 | # no guarantee that reading until EOF won't corrupt the stream |
|
622 | # no guarantee that reading until EOF won't corrupt the stream | |
622 | # state. |
|
623 | # state. | |
623 | if not isinstance(self.req.bodyfh, util.cappedreader): |
|
624 | if not isinstance(self.req.bodyfh, util.cappedreader): | |
624 | close = True |
|
625 | close = True | |
625 | else: |
|
626 | else: | |
626 | # We /could/ only drain certain HTTP response codes. But 200 |
|
627 | # We /could/ only drain certain HTTP response codes. But 200 | |
627 | # and non-200 wire protocol responses both require draining. |
|
628 | # and non-200 wire protocol responses both require draining. | |
628 | # Since we have a capped reader in place for all situations |
|
629 | # Since we have a capped reader in place for all situations | |
629 | # where we drain, it is safe to read from that stream. We'll |
|
630 | # where we drain, it is safe to read from that stream. We'll | |
630 | # either do a drain or no-op if we're already at EOF. |
|
631 | # either do a drain or no-op if we're already at EOF. | |
631 | drain = True |
|
632 | drain = True | |
632 |
|
633 | |||
633 | if close: |
|
634 | if close: | |
634 | self.headers.append((r'Connection', r'Close')) |
|
635 | self.headers.append((r'Connection', r'Close')) | |
635 |
|
636 | |||
636 | if drain: |
|
637 | if drain: | |
637 | assert isinstance(self.req.bodyfh, util.cappedreader) |
|
638 | assert isinstance(self.req.bodyfh, util.cappedreader) | |
638 | while True: |
|
639 | while True: | |
639 | chunk = self.req.bodyfh.read(32768) |
|
640 | chunk = self.req.bodyfh.read(32768) | |
640 | if not chunk: |
|
641 | if not chunk: | |
641 | break |
|
642 | break | |
642 |
|
643 | |||
643 | self.server_write = self._start_response( |
|
644 | self.server_write = self._start_response( | |
644 | pycompat.sysstr(status), self.headers) |
|
645 | pycompat.sysstr(status), self.headers) | |
645 | self._start_response = None |
|
646 | self._start_response = None | |
646 | self.headers = [] |
|
647 | self.headers = [] | |
647 | if body is not None: |
|
648 | if body is not None: | |
648 | self.write(body) |
|
649 | self.write(body) | |
649 | self.server_write = None |
|
650 | self.server_write = None | |
650 |
|
651 | |||
651 | def write(self, thing): |
|
652 | def write(self, thing): | |
652 | if thing: |
|
653 | if thing: | |
653 | try: |
|
654 | try: | |
654 | self.server_write(thing) |
|
655 | self.server_write(thing) | |
655 | except socket.error as inst: |
|
656 | except socket.error as inst: | |
656 | if inst[0] != errno.ECONNRESET: |
|
657 | if inst[0] != errno.ECONNRESET: | |
657 | raise |
|
658 | raise | |
658 |
|
659 | |||
659 | def flush(self): |
|
660 | def flush(self): | |
660 | return None |
|
661 | return None | |
661 |
|
662 | |||
662 | def wsgiapplication(app_maker): |
|
663 | def wsgiapplication(app_maker): | |
663 | '''For compatibility with old CGI scripts. A plain hgweb() or hgwebdir() |
|
664 | '''For compatibility with old CGI scripts. A plain hgweb() or hgwebdir() | |
664 | can and should now be used as a WSGI application.''' |
|
665 | can and should now be used as a WSGI application.''' | |
665 | application = app_maker() |
|
666 | application = app_maker() | |
666 | def run_wsgi(env, respond): |
|
667 | def run_wsgi(env, respond): | |
667 | return application(env, respond) |
|
668 | return application(env, respond) | |
668 | return run_wsgi |
|
669 | return run_wsgi |
General Comments 0
You need to be logged in to leave comments.
Login now