tipi 0.42 → 0.43
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +5 -0
- data/Gemfile +1 -1
- data/Gemfile.lock +14 -19
- data/benchmarks/bm_http1_parser.rb +44 -20
- data/df/server.rb +2 -0
- data/df/server_utils.rb +11 -14
- data/lib/tipi/digital_fabric/service.rb +4 -0
- data/lib/tipi/http1_adapter.rb +5 -4
- data/lib/tipi/version.rb +1 -1
- data/test/helper.rb +0 -1
- data/test/test_request.rb +1 -1
- data/tipi.gemspec +3 -5
- metadata +26 -47
- data/ext/tipi/extconf.rb +0 -13
- data/ext/tipi/http1_parser.c +0 -823
- data/ext/tipi/http1_parser.h +0 -18
- data/ext/tipi/tipi_ext.c +0 -5
- data/security/http1.rb +0 -12
- data/test/test_http1_parser.rb +0 -586
data/ext/tipi/extconf.rb
DELETED
@@ -1,13 +0,0 @@
|
|
1
|
-
# frozen_string_literal: true
|
2
|
-
|
3
|
-
require 'rubygems'
|
4
|
-
require 'mkmf'
|
5
|
-
|
6
|
-
require_relative '../../security/http1'
|
7
|
-
|
8
|
-
$CFLAGS << " -Wno-format-security"
|
9
|
-
CONFIG['optflags'] << ' -fno-strict-aliasing' unless RUBY_PLATFORM =~ /mswin/
|
10
|
-
Tipi::HTTP1_LIMITS.each { |k, v| $defs << "-D#{k.upcase}=#{v}" }
|
11
|
-
|
12
|
-
dir_config 'tipi_ext'
|
13
|
-
create_makefile 'tipi_ext'
|
data/ext/tipi/http1_parser.c
DELETED
@@ -1,823 +0,0 @@
|
|
1
|
-
#include "ruby.h"
|
2
|
-
#include "http1_parser.h"
|
3
|
-
|
4
|
-
// Security-related limits are defined in security/http1.rb and injected as
|
5
|
-
// defines in extconf.rb
|
6
|
-
|
7
|
-
#define INITIAL_BUFFER_SIZE 4096
|
8
|
-
#define BUFFER_TRIM_MIN_LEN 4096
|
9
|
-
#define BUFFER_TRIM_MIN_POS 2048
|
10
|
-
#define MAX_HEADERS_READ_LENGTH 4096
|
11
|
-
#define MAX_BODY_READ_LENGTH (1 << 20) // 1MB
|
12
|
-
|
13
|
-
#define BODY_READ_MODE_UNKNOWN -2
|
14
|
-
#define BODY_READ_MODE_CHUNKED -1
|
15
|
-
|
16
|
-
ID ID_arity;
|
17
|
-
ID ID_backend_read;
|
18
|
-
ID ID_backend_recv;
|
19
|
-
ID ID_call;
|
20
|
-
ID ID_downcase;
|
21
|
-
ID ID_eq;
|
22
|
-
ID ID_parser_read_method;
|
23
|
-
ID ID_read;
|
24
|
-
ID ID_readpartial;
|
25
|
-
ID ID_to_i;
|
26
|
-
|
27
|
-
static VALUE mPolyphony = Qnil;
|
28
|
-
static VALUE cError;
|
29
|
-
|
30
|
-
VALUE NUM_max_headers_read_length;
|
31
|
-
VALUE NUM_buffer_start;
|
32
|
-
VALUE NUM_buffer_end;
|
33
|
-
|
34
|
-
VALUE STR_pseudo_method;
|
35
|
-
VALUE STR_pseudo_path;
|
36
|
-
VALUE STR_pseudo_protocol;
|
37
|
-
VALUE STR_pseudo_rx;
|
38
|
-
|
39
|
-
VALUE STR_chunked;
|
40
|
-
VALUE STR_content_length;
|
41
|
-
VALUE STR_transfer_encoding;
|
42
|
-
|
43
|
-
VALUE SYM_backend_read;
|
44
|
-
VALUE SYM_backend_recv;
|
45
|
-
|
46
|
-
enum read_method {
|
47
|
-
method_readpartial, // receiver.readpartial (Polyphony-specific)
|
48
|
-
method_backend_read, // Polyphony.backend_read (Polyphony-specific)
|
49
|
-
method_backend_recv, // Polyphony.backend_recv (Polyphony-specific)
|
50
|
-
method_call // receiver.call(len) (Universal)
|
51
|
-
};
|
52
|
-
|
53
|
-
typedef struct parser {
|
54
|
-
VALUE io;
|
55
|
-
VALUE buffer;
|
56
|
-
VALUE headers;
|
57
|
-
int pos;
|
58
|
-
int current_request_rx;
|
59
|
-
|
60
|
-
enum read_method read_method;
|
61
|
-
int body_read_mode;
|
62
|
-
int body_left;
|
63
|
-
int request_completed;
|
64
|
-
} Parser_t;
|
65
|
-
|
66
|
-
VALUE cParser = Qnil;
|
67
|
-
|
68
|
-
static void Parser_mark(void *ptr) {
|
69
|
-
Parser_t *parser = ptr;
|
70
|
-
rb_gc_mark(parser->io);
|
71
|
-
rb_gc_mark(parser->buffer);
|
72
|
-
rb_gc_mark(parser->headers);
|
73
|
-
}
|
74
|
-
|
75
|
-
static void Parser_free(void *ptr) {
|
76
|
-
xfree(ptr);
|
77
|
-
}
|
78
|
-
|
79
|
-
static size_t Parser_size(const void *ptr) {
|
80
|
-
return sizeof(Parser_t);
|
81
|
-
}
|
82
|
-
|
83
|
-
static const rb_data_type_t Parser_type = {
|
84
|
-
"Parser",
|
85
|
-
{Parser_mark, Parser_free, Parser_size,},
|
86
|
-
0, 0, 0
|
87
|
-
};
|
88
|
-
|
89
|
-
static VALUE Parser_allocate(VALUE klass) {
|
90
|
-
Parser_t *parser;
|
91
|
-
|
92
|
-
parser = ALLOC(Parser_t);
|
93
|
-
return TypedData_Wrap_Struct(klass, &Parser_type, parser);
|
94
|
-
}
|
95
|
-
|
96
|
-
#define GetParser(obj, parser) \
|
97
|
-
TypedData_Get_Struct((obj), Parser_t, &Parser_type, (parser))
|
98
|
-
|
99
|
-
enum read_method detect_read_method(VALUE io) {
|
100
|
-
if (rb_respond_to(io, ID_parser_read_method)) {
|
101
|
-
if (mPolyphony == Qnil)
|
102
|
-
mPolyphony = rb_const_get(rb_cObject, rb_intern("Polyphony"));
|
103
|
-
VALUE method = rb_funcall(io, ID_parser_read_method, 0);
|
104
|
-
if (method == SYM_backend_read) return method_backend_read;
|
105
|
-
if (method == SYM_backend_recv) return method_backend_recv;
|
106
|
-
return method_readpartial;
|
107
|
-
}
|
108
|
-
else if (rb_respond_to(io, ID_call)) {
|
109
|
-
return method_call;
|
110
|
-
}
|
111
|
-
else
|
112
|
-
rb_raise(rb_eRuntimeError, "Provided reader should be a callable or respond to #__parser_read_method__");
|
113
|
-
}
|
114
|
-
|
115
|
-
VALUE Parser_initialize(VALUE self, VALUE io) {
|
116
|
-
Parser_t *parser;
|
117
|
-
GetParser(self, parser);
|
118
|
-
|
119
|
-
parser->io = io;
|
120
|
-
parser->buffer = rb_str_new_literal("");
|
121
|
-
parser->headers = Qnil;
|
122
|
-
parser->pos = 0;
|
123
|
-
|
124
|
-
// pre-allocate the buffer
|
125
|
-
rb_str_modify_expand(parser->buffer, INITIAL_BUFFER_SIZE);
|
126
|
-
|
127
|
-
parser->read_method = detect_read_method(io);
|
128
|
-
parser->body_read_mode = BODY_READ_MODE_UNKNOWN;
|
129
|
-
parser->body_left = 0;
|
130
|
-
return self;
|
131
|
-
}
|
132
|
-
|
133
|
-
////////////////////////////////////////////////////////////////////////////////
|
134
|
-
|
135
|
-
#define str_downcase(str) (rb_funcall((str), ID_downcase, 0))
|
136
|
-
|
137
|
-
#define FILL_BUFFER_OR_GOTO_EOF(state) { if (!fill_buffer(state)) goto eof; }
|
138
|
-
|
139
|
-
#define BUFFER_POS(state) ((state)->parser->pos)
|
140
|
-
#define BUFFER_LEN(state) ((state)->len)
|
141
|
-
#define BUFFER_CUR(state) ((state)->ptr[(state)->parser->pos])
|
142
|
-
#define BUFFER_AT(state, pos) ((state)->ptr[pos])
|
143
|
-
#define BUFFER_PTR(state, pos) ((state)->ptr + pos)
|
144
|
-
#define BUFFER_STR(state, pos, len) (rb_utf8_str_new((state)->ptr + pos, len))
|
145
|
-
|
146
|
-
#define INC_BUFFER_POS(state) { \
|
147
|
-
BUFFER_POS(state)++; \
|
148
|
-
if (BUFFER_POS(state) == BUFFER_LEN(state)) FILL_BUFFER_OR_GOTO_EOF(state); \
|
149
|
-
}
|
150
|
-
|
151
|
-
#define INC_BUFFER_POS_NO_FILL(state) BUFFER_POS(state)++;
|
152
|
-
|
153
|
-
#define INC_BUFFER_POS_UTF8(state, len) { \
|
154
|
-
unsigned char c = BUFFER_CUR(state); \
|
155
|
-
if ((c & 0xf0) == 0xf0) { \
|
156
|
-
while (BUFFER_LEN(state) - BUFFER_POS(state) < 4) FILL_BUFFER_OR_GOTO_EOF(state); \
|
157
|
-
BUFFER_POS(state) += 4; \
|
158
|
-
len += 4; \
|
159
|
-
} \
|
160
|
-
else if ((c & 0xe0) == 0xe0) { \
|
161
|
-
while (BUFFER_LEN(state) - BUFFER_POS(state) < 3) FILL_BUFFER_OR_GOTO_EOF(state); \
|
162
|
-
BUFFER_POS(state) += 3; \
|
163
|
-
len += 3; \
|
164
|
-
} \
|
165
|
-
else if ((c & 0xc0) == 0xc0) { \
|
166
|
-
while (BUFFER_LEN(state) - BUFFER_POS(state) < 2) FILL_BUFFER_OR_GOTO_EOF(state); \
|
167
|
-
BUFFER_POS(state) += 2; \
|
168
|
-
len += 2; \
|
169
|
-
} \
|
170
|
-
else { \
|
171
|
-
BUFFER_POS(state)++; \
|
172
|
-
len ++; \
|
173
|
-
if (BUFFER_POS(state) == BUFFER_LEN(state)) FILL_BUFFER_OR_GOTO_EOF(state); \
|
174
|
-
} \
|
175
|
-
}
|
176
|
-
|
177
|
-
#define INIT_PARSER_STATE(state) { \
|
178
|
-
(state)->len = RSTRING_LEN((state)->parser->buffer); \
|
179
|
-
if (BUFFER_POS(state) == BUFFER_LEN(state)) \
|
180
|
-
FILL_BUFFER_OR_GOTO_EOF(state) \
|
181
|
-
else \
|
182
|
-
(state)->ptr = RSTRING_PTR((state)->parser->buffer); \
|
183
|
-
}
|
184
|
-
|
185
|
-
#define RAISE_BAD_REQUEST(msg) rb_raise(cError, msg)
|
186
|
-
|
187
|
-
#define SET_HEADER_VALUE_FROM_BUFFER(state, headers, key, pos, len) { \
|
188
|
-
VALUE value = BUFFER_STR(state, pos, len); \
|
189
|
-
rb_hash_aset(headers, key, value); \
|
190
|
-
RB_GC_GUARD(value); \
|
191
|
-
}
|
192
|
-
|
193
|
-
#define SET_HEADER_DOWNCASE_VALUE_FROM_BUFFER(state, headers, key, pos, len) { \
|
194
|
-
VALUE value = str_downcase(BUFFER_STR(state, pos, len)); \
|
195
|
-
rb_hash_aset(headers, key, value); \
|
196
|
-
RB_GC_GUARD(value); \
|
197
|
-
}
|
198
|
-
|
199
|
-
#define CONSUME_CRLF(state) { \
|
200
|
-
INC_BUFFER_POS(state); \
|
201
|
-
if (BUFFER_CUR(state) != '\n') goto bad_request; \
|
202
|
-
INC_BUFFER_POS(state); \
|
203
|
-
}
|
204
|
-
|
205
|
-
#define CONSUME_CRLF_NO_FILL(state) { \
|
206
|
-
INC_BUFFER_POS(state); \
|
207
|
-
if (BUFFER_CUR(state) != '\n') goto bad_request; \
|
208
|
-
INC_BUFFER_POS_NO_FILL(state); \
|
209
|
-
}
|
210
|
-
|
211
|
-
#define GLOBAL_STR(v, s) v = rb_str_new_literal(s); rb_global_variable(&v)
|
212
|
-
|
213
|
-
struct parser_state {
|
214
|
-
struct parser *parser;
|
215
|
-
char *ptr;
|
216
|
-
int len;
|
217
|
-
};
|
218
|
-
|
219
|
-
////////////////////////////////////////////////////////////////////////////////
|
220
|
-
|
221
|
-
static inline VALUE io_read_call(VALUE io, VALUE maxlen, VALUE buf, VALUE buf_pos) {
|
222
|
-
VALUE result = rb_funcall(io, ID_call, 1, maxlen);
|
223
|
-
if (result == Qnil) return Qnil;
|
224
|
-
|
225
|
-
if (buf_pos == NUM_buffer_start) rb_str_set_len(buf, 0);
|
226
|
-
rb_str_append(buf, result);
|
227
|
-
RB_GC_GUARD(result);
|
228
|
-
return buf;
|
229
|
-
}
|
230
|
-
|
231
|
-
static inline VALUE parser_io_read(Parser_t *parser, VALUE maxlen, VALUE buf, VALUE buf_pos) {
|
232
|
-
switch (parser->read_method) {
|
233
|
-
case method_backend_read:
|
234
|
-
return rb_funcall(mPolyphony, ID_backend_read, 5, parser->io, buf, maxlen, Qfalse, buf_pos);
|
235
|
-
case method_backend_recv:
|
236
|
-
return rb_funcall(mPolyphony, ID_backend_recv, 4, parser->io, buf, maxlen, buf_pos);
|
237
|
-
case method_readpartial:
|
238
|
-
return rb_funcall(parser->io, ID_readpartial, 4, maxlen, buf, buf_pos, Qfalse);
|
239
|
-
case method_call:
|
240
|
-
return io_read_call(parser->io, maxlen, buf, buf_pos);
|
241
|
-
default:
|
242
|
-
return Qnil;
|
243
|
-
}
|
244
|
-
}
|
245
|
-
|
246
|
-
static inline int fill_buffer(struct parser_state *state) {
|
247
|
-
VALUE ret = parser_io_read(state->parser, NUM_max_headers_read_length, state->parser->buffer, NUM_buffer_end);
|
248
|
-
if (ret == Qnil) return 0;
|
249
|
-
|
250
|
-
state->parser->buffer = ret;
|
251
|
-
int len = RSTRING_LEN(state->parser->buffer);
|
252
|
-
int read_bytes = len - state->len;
|
253
|
-
if (!read_bytes) return 0;
|
254
|
-
|
255
|
-
state->ptr = RSTRING_PTR(state->parser->buffer);
|
256
|
-
state->len = len;
|
257
|
-
return read_bytes;
|
258
|
-
}
|
259
|
-
|
260
|
-
static inline void buffer_trim(struct parser_state *state) {
|
261
|
-
int len = RSTRING_LEN(state->parser->buffer);
|
262
|
-
int pos = state->parser->pos;
|
263
|
-
int left = len - pos;
|
264
|
-
|
265
|
-
// The buffer is trimmed only if length and position thresholds are passed,
|
266
|
-
// *and* position is past the halfway point.
|
267
|
-
if (len < BUFFER_TRIM_MIN_LEN ||
|
268
|
-
pos < BUFFER_TRIM_MIN_POS ||
|
269
|
-
left >= pos) return;
|
270
|
-
|
271
|
-
if (left > 0) {
|
272
|
-
char *ptr = RSTRING_PTR(state->parser->buffer);
|
273
|
-
memcpy(ptr, ptr + pos, left);
|
274
|
-
}
|
275
|
-
rb_str_set_len(state->parser->buffer, left);
|
276
|
-
state->parser->pos = 0;
|
277
|
-
}
|
278
|
-
|
279
|
-
static inline void str_append_from_buffer(VALUE str, char *ptr, int len) {
|
280
|
-
int str_len = RSTRING_LEN(str);
|
281
|
-
rb_str_modify_expand(str, len);
|
282
|
-
memcpy(RSTRING_PTR(str) + str_len, ptr, len);
|
283
|
-
rb_str_set_len(str, str_len + len);
|
284
|
-
}
|
285
|
-
|
286
|
-
////////////////////////////////////////////////////////////////////////////////
|
287
|
-
|
288
|
-
static inline int parse_method(struct parser_state *state, VALUE headers) {
|
289
|
-
int pos = BUFFER_POS(state);
|
290
|
-
int len = 0;
|
291
|
-
|
292
|
-
while (1) {
|
293
|
-
switch (BUFFER_CUR(state)) {
|
294
|
-
case ' ':
|
295
|
-
if (len < 1 || len > MAX_METHOD_LENGTH) goto bad_request;
|
296
|
-
INC_BUFFER_POS(state);
|
297
|
-
goto done;
|
298
|
-
case '\r':
|
299
|
-
case '\n':
|
300
|
-
goto bad_request;
|
301
|
-
default:
|
302
|
-
INC_BUFFER_POS(state);
|
303
|
-
len++;
|
304
|
-
// INC_BUFFER_POS_UTF8(state, len);
|
305
|
-
if (len > MAX_METHOD_LENGTH) goto bad_request;
|
306
|
-
}
|
307
|
-
}
|
308
|
-
done:
|
309
|
-
SET_HEADER_DOWNCASE_VALUE_FROM_BUFFER(state, headers, STR_pseudo_method, pos, len);
|
310
|
-
return 1;
|
311
|
-
bad_request:
|
312
|
-
RAISE_BAD_REQUEST("Invalid method");
|
313
|
-
eof:
|
314
|
-
return 0;
|
315
|
-
}
|
316
|
-
|
317
|
-
static int parse_request_target(struct parser_state *state, VALUE headers) {
|
318
|
-
while (BUFFER_CUR(state) == ' ') INC_BUFFER_POS(state);
|
319
|
-
int pos = BUFFER_POS(state);
|
320
|
-
int len = 0;
|
321
|
-
while (1) {
|
322
|
-
switch (BUFFER_CUR(state)) {
|
323
|
-
case ' ':
|
324
|
-
if (len < 1 || len > MAX_PATH_LENGTH) goto bad_request;
|
325
|
-
INC_BUFFER_POS(state);
|
326
|
-
goto done;
|
327
|
-
case '\r':
|
328
|
-
case '\n':
|
329
|
-
goto bad_request;
|
330
|
-
default:
|
331
|
-
INC_BUFFER_POS(state);
|
332
|
-
len++;
|
333
|
-
// INC_BUFFER_POS_UTF8(state, len);
|
334
|
-
if (len > MAX_PATH_LENGTH) goto bad_request;
|
335
|
-
}
|
336
|
-
}
|
337
|
-
done:
|
338
|
-
SET_HEADER_VALUE_FROM_BUFFER(state, headers, STR_pseudo_path, pos, len);
|
339
|
-
return 1;
|
340
|
-
bad_request:
|
341
|
-
RAISE_BAD_REQUEST("Invalid request target");
|
342
|
-
eof:
|
343
|
-
return 0;
|
344
|
-
}
|
345
|
-
|
346
|
-
// case-insensitive compare
|
347
|
-
#define CMP_CI(state, down, up) ((BUFFER_CUR(state) == down) || (BUFFER_CUR(state) == up))
|
348
|
-
|
349
|
-
static int parse_protocol(struct parser_state *state, VALUE headers) {
|
350
|
-
while (BUFFER_CUR(state) == ' ') INC_BUFFER_POS(state);
|
351
|
-
int pos = BUFFER_POS(state);
|
352
|
-
int len = 0;
|
353
|
-
|
354
|
-
if (CMP_CI(state, 'H', 'h')) INC_BUFFER_POS(state) else goto bad_request;
|
355
|
-
if (CMP_CI(state, 'T', 't')) INC_BUFFER_POS(state) else goto bad_request;
|
356
|
-
if (CMP_CI(state, 'T', 't')) INC_BUFFER_POS(state) else goto bad_request;
|
357
|
-
if (CMP_CI(state, 'P', 'p')) INC_BUFFER_POS(state) else goto bad_request;
|
358
|
-
if (BUFFER_CUR(state) == '/') INC_BUFFER_POS(state) else goto bad_request;
|
359
|
-
if (BUFFER_CUR(state) == '1') INC_BUFFER_POS(state) else goto bad_request;
|
360
|
-
len = 6;
|
361
|
-
while (1) {
|
362
|
-
switch (BUFFER_CUR(state)) {
|
363
|
-
case '\r':
|
364
|
-
CONSUME_CRLF(state);
|
365
|
-
goto done;
|
366
|
-
case '\n':
|
367
|
-
INC_BUFFER_POS(state);
|
368
|
-
goto done;
|
369
|
-
case '.':
|
370
|
-
INC_BUFFER_POS(state);
|
371
|
-
char c = BUFFER_CUR(state);
|
372
|
-
if (c == '0' || c == '1') {
|
373
|
-
INC_BUFFER_POS(state);
|
374
|
-
len += 2;
|
375
|
-
continue;
|
376
|
-
}
|
377
|
-
goto bad_request;
|
378
|
-
default:
|
379
|
-
goto bad_request;
|
380
|
-
}
|
381
|
-
}
|
382
|
-
done:
|
383
|
-
if (len < 6 || len > 8) goto bad_request;
|
384
|
-
SET_HEADER_DOWNCASE_VALUE_FROM_BUFFER(state, headers, STR_pseudo_protocol, pos, len);
|
385
|
-
return 1;
|
386
|
-
bad_request:
|
387
|
-
RAISE_BAD_REQUEST("Invalid protocol");
|
388
|
-
eof:
|
389
|
-
return 0;
|
390
|
-
}
|
391
|
-
|
392
|
-
int parse_request_line(struct parser_state *state, VALUE headers) {
|
393
|
-
if (!parse_method(state, headers)) goto eof;
|
394
|
-
if (!parse_request_target(state, headers)) goto eof;
|
395
|
-
if (!parse_protocol(state, headers)) goto eof;
|
396
|
-
|
397
|
-
return 1;
|
398
|
-
eof:
|
399
|
-
return 0;
|
400
|
-
}
|
401
|
-
|
402
|
-
static inline int parse_header_key(struct parser_state *state, VALUE *key) {
|
403
|
-
int pos = BUFFER_POS(state);
|
404
|
-
int len = 0;
|
405
|
-
|
406
|
-
while (1) {
|
407
|
-
switch (BUFFER_CUR(state)) {
|
408
|
-
case ' ':
|
409
|
-
goto bad_request;
|
410
|
-
case ':':
|
411
|
-
if (len < 1 || len > MAX_HEADER_KEY_LENGTH)
|
412
|
-
goto bad_request;
|
413
|
-
INC_BUFFER_POS(state);
|
414
|
-
goto done;
|
415
|
-
case '\r':
|
416
|
-
if (BUFFER_POS(state) > pos) goto bad_request;
|
417
|
-
CONSUME_CRLF_NO_FILL(state);
|
418
|
-
goto done;
|
419
|
-
case '\n':
|
420
|
-
if (BUFFER_POS(state) > pos) goto bad_request;
|
421
|
-
|
422
|
-
INC_BUFFER_POS_NO_FILL(state);
|
423
|
-
goto done;
|
424
|
-
default:
|
425
|
-
INC_BUFFER_POS(state);
|
426
|
-
len++;
|
427
|
-
// INC_BUFFER_POS_UTF8(state, len);
|
428
|
-
if (len > MAX_HEADER_KEY_LENGTH) goto bad_request;
|
429
|
-
}
|
430
|
-
}
|
431
|
-
done:
|
432
|
-
if (len == 0) return -1;
|
433
|
-
(*key) = str_downcase(BUFFER_STR(state, pos, len));
|
434
|
-
return 1;
|
435
|
-
bad_request:
|
436
|
-
RAISE_BAD_REQUEST("Invalid header key");
|
437
|
-
eof:
|
438
|
-
return 0;
|
439
|
-
}
|
440
|
-
|
441
|
-
static inline int parse_header_value(struct parser_state *state, VALUE *value) {
|
442
|
-
while (BUFFER_CUR(state) == ' ') INC_BUFFER_POS(state);
|
443
|
-
|
444
|
-
int pos = BUFFER_POS(state);
|
445
|
-
int len = 0;
|
446
|
-
|
447
|
-
while (1) {
|
448
|
-
switch (BUFFER_CUR(state)) {
|
449
|
-
case '\r':
|
450
|
-
CONSUME_CRLF(state);
|
451
|
-
goto done;
|
452
|
-
case '\n':
|
453
|
-
INC_BUFFER_POS(state);
|
454
|
-
goto done;
|
455
|
-
default:
|
456
|
-
INC_BUFFER_POS_UTF8(state, len);
|
457
|
-
if (len > MAX_HEADER_VALUE_LENGTH) goto bad_request;
|
458
|
-
}
|
459
|
-
}
|
460
|
-
done:
|
461
|
-
if (len < 1 || len > MAX_HEADER_VALUE_LENGTH) goto bad_request;
|
462
|
-
(*value) = BUFFER_STR(state, pos, len);
|
463
|
-
return 1;
|
464
|
-
bad_request:
|
465
|
-
RAISE_BAD_REQUEST("Invalid header value");
|
466
|
-
eof:
|
467
|
-
return 0;
|
468
|
-
}
|
469
|
-
|
470
|
-
static inline int parse_header(struct parser_state *state, VALUE headers) {
|
471
|
-
VALUE key, value;
|
472
|
-
|
473
|
-
switch (parse_header_key(state, &key)) {
|
474
|
-
case -1: return -1;
|
475
|
-
case 0: goto eof;
|
476
|
-
}
|
477
|
-
|
478
|
-
if (!parse_header_value(state, &value)) goto eof;
|
479
|
-
|
480
|
-
VALUE existing = rb_hash_aref(headers, key);
|
481
|
-
if (existing != Qnil) {
|
482
|
-
if (TYPE(existing) != T_ARRAY) {
|
483
|
-
existing = rb_ary_new3(2, existing, value);
|
484
|
-
rb_hash_aset(headers, key, existing);
|
485
|
-
}
|
486
|
-
else
|
487
|
-
rb_ary_push(existing, value);
|
488
|
-
}
|
489
|
-
else
|
490
|
-
rb_hash_aset(headers, key, value);
|
491
|
-
|
492
|
-
RB_GC_GUARD(existing);
|
493
|
-
RB_GC_GUARD(key);
|
494
|
-
RB_GC_GUARD(value);
|
495
|
-
return 1;
|
496
|
-
eof:
|
497
|
-
return 0;
|
498
|
-
}
|
499
|
-
|
500
|
-
VALUE Parser_parse_headers(VALUE self) {
|
501
|
-
struct parser_state state;
|
502
|
-
GetParser(self, state.parser);
|
503
|
-
state.parser->headers = rb_hash_new();
|
504
|
-
|
505
|
-
buffer_trim(&state);
|
506
|
-
int initial_pos = state.parser->pos;
|
507
|
-
INIT_PARSER_STATE(&state);
|
508
|
-
state.parser->current_request_rx = 0;
|
509
|
-
|
510
|
-
if (!parse_request_line(&state, state.parser->headers)) goto eof;
|
511
|
-
|
512
|
-
int header_count = 0;
|
513
|
-
while (1) {
|
514
|
-
if (header_count > MAX_HEADER_COUNT) RAISE_BAD_REQUEST("Too many headers");
|
515
|
-
switch (parse_header(&state, state.parser->headers)) {
|
516
|
-
case -1: goto done; // empty header => end of headers
|
517
|
-
case 0: goto eof;
|
518
|
-
}
|
519
|
-
header_count++;
|
520
|
-
}
|
521
|
-
eof:
|
522
|
-
state.parser->headers = Qnil;
|
523
|
-
done:
|
524
|
-
state.parser->body_read_mode = BODY_READ_MODE_UNKNOWN;
|
525
|
-
int read_bytes = BUFFER_POS(&state) - initial_pos;
|
526
|
-
|
527
|
-
state.parser->current_request_rx += read_bytes;
|
528
|
-
if (state.parser->headers != Qnil)
|
529
|
-
rb_hash_aset(state.parser->headers, STR_pseudo_rx, INT2NUM(read_bytes));
|
530
|
-
return state.parser->headers;
|
531
|
-
}
|
532
|
-
|
533
|
-
////////////////////////////////////////////////////////////////////////////////
|
534
|
-
|
535
|
-
static inline int str_to_int(VALUE value, const char *error_msg) {
|
536
|
-
char *ptr = RSTRING_PTR(value);
|
537
|
-
int len = RSTRING_LEN(value);
|
538
|
-
int int_value = 0;
|
539
|
-
|
540
|
-
while (len) {
|
541
|
-
char c = *ptr;
|
542
|
-
if ((c >= '0') && (c <= '9'))
|
543
|
-
int_value = int_value * 10 + (c - '0');
|
544
|
-
else
|
545
|
-
RAISE_BAD_REQUEST(error_msg);
|
546
|
-
len--;
|
547
|
-
ptr++;
|
548
|
-
}
|
549
|
-
|
550
|
-
return int_value;
|
551
|
-
}
|
552
|
-
|
553
|
-
VALUE read_body_with_content_length(Parser_t *parser, int read_entire_body, int buffered_only) {
|
554
|
-
if (parser->body_left <= 0) return Qnil;
|
555
|
-
|
556
|
-
VALUE body = Qnil;
|
557
|
-
|
558
|
-
int len = RSTRING_LEN(parser->buffer);
|
559
|
-
int pos = parser->pos;
|
560
|
-
|
561
|
-
if (pos < len) {
|
562
|
-
int available = len - pos;
|
563
|
-
if (available > parser->body_left) available = parser->body_left;
|
564
|
-
body = rb_str_new(RSTRING_PTR(parser->buffer) + pos, available);
|
565
|
-
parser->pos += available;
|
566
|
-
parser->current_request_rx += available;
|
567
|
-
parser->body_left -= available;
|
568
|
-
if (!parser->body_left) parser->request_completed = 1;
|
569
|
-
}
|
570
|
-
else {
|
571
|
-
body = Qnil;
|
572
|
-
len = 0;
|
573
|
-
}
|
574
|
-
if (buffered_only) return body;
|
575
|
-
|
576
|
-
while (parser->body_left) {
|
577
|
-
int maxlen = parser->body_left <= MAX_BODY_READ_LENGTH ? parser->body_left : MAX_BODY_READ_LENGTH;
|
578
|
-
VALUE tmp_buf = parser_io_read(parser, INT2NUM(maxlen), Qnil, NUM_buffer_start);
|
579
|
-
if (tmp_buf == Qnil) goto eof;
|
580
|
-
if (body != Qnil)
|
581
|
-
rb_str_append(body, tmp_buf);
|
582
|
-
else
|
583
|
-
body = tmp_buf;
|
584
|
-
int read_bytes = RSTRING_LEN(tmp_buf);
|
585
|
-
parser->current_request_rx += read_bytes;
|
586
|
-
parser->body_left -= read_bytes;
|
587
|
-
if (!parser->body_left) parser->request_completed = 1;
|
588
|
-
RB_GC_GUARD(tmp_buf);
|
589
|
-
if (!read_entire_body) goto done;
|
590
|
-
}
|
591
|
-
done:
|
592
|
-
rb_hash_aset(parser->headers, STR_pseudo_rx, INT2NUM(parser->current_request_rx));
|
593
|
-
RB_GC_GUARD(body);
|
594
|
-
return body;
|
595
|
-
eof:
|
596
|
-
RAISE_BAD_REQUEST("Incomplete body");
|
597
|
-
}
|
598
|
-
|
599
|
-
int chunked_encoding_p(VALUE transfer_encoding) {
|
600
|
-
if (transfer_encoding == Qnil) return 0;
|
601
|
-
return rb_funcall(str_downcase(transfer_encoding), ID_eq, 1, STR_chunked) == Qtrue;
|
602
|
-
}
|
603
|
-
|
604
|
-
int parse_chunk_size(struct parser_state *state, int *chunk_size) {
|
605
|
-
int len = 0;
|
606
|
-
int value = 0;
|
607
|
-
int initial_pos = BUFFER_POS(state);
|
608
|
-
|
609
|
-
while (1) {
|
610
|
-
char c = BUFFER_CUR(state);
|
611
|
-
if ((c >= '0') && (c <= '9')) value = (value << 4) + (c - '0');
|
612
|
-
else if ((c >= 'a') && (c <= 'f')) value = (value << 4) + (c - 'a' + 10);
|
613
|
-
else if ((c >= 'A') && (c <= 'F')) value = (value << 4) + (c - 'A' + 10);
|
614
|
-
else switch (c) {
|
615
|
-
case '\r':
|
616
|
-
CONSUME_CRLF_NO_FILL(state);
|
617
|
-
goto done;
|
618
|
-
case '\n':
|
619
|
-
INC_BUFFER_POS_NO_FILL(state);
|
620
|
-
goto done;
|
621
|
-
default:
|
622
|
-
goto bad_request;
|
623
|
-
}
|
624
|
-
INC_BUFFER_POS(state);
|
625
|
-
len++;
|
626
|
-
if (len >= MAX_CHUNKED_ENCODING_CHUNK_SIZE_LENGTH) goto bad_request;
|
627
|
-
}
|
628
|
-
done:
|
629
|
-
if (len == 0) goto bad_request;
|
630
|
-
(*chunk_size) = value;
|
631
|
-
state->parser->current_request_rx += BUFFER_POS(state) - initial_pos;
|
632
|
-
return 1;
|
633
|
-
bad_request:
|
634
|
-
RAISE_BAD_REQUEST("Invalid chunk size");
|
635
|
-
eof:
|
636
|
-
return 0;
|
637
|
-
}
|
638
|
-
|
639
|
-
int read_body_chunk_with_chunked_encoding(struct parser_state *state, VALUE *body, int chunk_size, int buffered_only) {
|
640
|
-
int len = RSTRING_LEN(state->parser->buffer);
|
641
|
-
int pos = state->parser->pos;
|
642
|
-
int left = chunk_size;
|
643
|
-
|
644
|
-
if (pos < len) {
|
645
|
-
int available = len - pos;
|
646
|
-
if (available > left) available = left;
|
647
|
-
if (*body != Qnil)
|
648
|
-
str_append_from_buffer(*body, RSTRING_PTR(state->parser->buffer) + pos, available);
|
649
|
-
else
|
650
|
-
*body = rb_str_new(RSTRING_PTR(state->parser->buffer) + pos, available);
|
651
|
-
state->parser->pos += available;
|
652
|
-
state->parser->current_request_rx += available;
|
653
|
-
left -= available;
|
654
|
-
}
|
655
|
-
if (buffered_only) return 1;
|
656
|
-
|
657
|
-
while (left) {
|
658
|
-
int maxlen = left <= MAX_BODY_READ_LENGTH ? left : MAX_BODY_READ_LENGTH;
|
659
|
-
|
660
|
-
VALUE tmp_buf = parser_io_read(state->parser, INT2NUM(maxlen), Qnil, NUM_buffer_start);
|
661
|
-
if (tmp_buf == Qnil) goto eof;
|
662
|
-
if (*body != Qnil)
|
663
|
-
rb_str_append(*body, tmp_buf);
|
664
|
-
else
|
665
|
-
*body = tmp_buf;
|
666
|
-
int read_bytes = RSTRING_LEN(tmp_buf);
|
667
|
-
state->parser->current_request_rx += read_bytes;
|
668
|
-
left -= read_bytes;
|
669
|
-
RB_GC_GUARD(tmp_buf);
|
670
|
-
}
|
671
|
-
return 1;
|
672
|
-
eof:
|
673
|
-
return 0;
|
674
|
-
}
|
675
|
-
|
676
|
-
static inline int parse_chunk_postfix(struct parser_state *state) {
|
677
|
-
int initial_pos = BUFFER_POS(state);
|
678
|
-
if (initial_pos == BUFFER_LEN(state)) FILL_BUFFER_OR_GOTO_EOF(state);
|
679
|
-
switch (BUFFER_CUR(state)) {
|
680
|
-
case '\r':
|
681
|
-
CONSUME_CRLF_NO_FILL(state);
|
682
|
-
goto done;
|
683
|
-
case '\n':
|
684
|
-
INC_BUFFER_POS_NO_FILL(state);
|
685
|
-
goto done;
|
686
|
-
default:
|
687
|
-
goto bad_request;
|
688
|
-
}
|
689
|
-
done:
|
690
|
-
state->parser->current_request_rx += BUFFER_POS(state) - initial_pos;
|
691
|
-
return 1;
|
692
|
-
bad_request:
|
693
|
-
RAISE_BAD_REQUEST("Invalid protocol");
|
694
|
-
eof:
|
695
|
-
return 0;
|
696
|
-
}
|
697
|
-
|
698
|
-
VALUE read_body_with_chunked_encoding(Parser_t *parser, int read_entire_body, int buffered_only) {
|
699
|
-
struct parser_state state;
|
700
|
-
state.parser = parser;
|
701
|
-
buffer_trim(&state);
|
702
|
-
INIT_PARSER_STATE(&state);
|
703
|
-
VALUE body = Qnil;
|
704
|
-
|
705
|
-
while (1) {
|
706
|
-
int chunk_size = 0;
|
707
|
-
if (BUFFER_POS(&state) == BUFFER_LEN(&state)) FILL_BUFFER_OR_GOTO_EOF(&state);
|
708
|
-
if (!parse_chunk_size(&state, &chunk_size)) goto bad_request;
|
709
|
-
|
710
|
-
if (chunk_size) {
|
711
|
-
if (!read_body_chunk_with_chunked_encoding(&state, &body, chunk_size, buffered_only)) goto bad_request;
|
712
|
-
}
|
713
|
-
else parser->request_completed = 1;
|
714
|
-
|
715
|
-
if (!parse_chunk_postfix(&state)) goto bad_request;
|
716
|
-
if (!chunk_size || !read_entire_body) goto done;
|
717
|
-
}
|
718
|
-
bad_request:
|
719
|
-
RAISE_BAD_REQUEST("Malformed request body");
|
720
|
-
eof:
|
721
|
-
RAISE_BAD_REQUEST("Incomplete request body");
|
722
|
-
done:
|
723
|
-
rb_hash_aset(parser->headers, STR_pseudo_rx, INT2NUM(state.parser->current_request_rx));
|
724
|
-
RB_GC_GUARD(body);
|
725
|
-
return body;
|
726
|
-
}
|
727
|
-
|
728
|
-
static inline void detect_body_read_mode(Parser_t *parser) {
|
729
|
-
VALUE content_length = rb_hash_aref(parser->headers, STR_content_length);
|
730
|
-
if (content_length != Qnil) {
|
731
|
-
int int_content_length = str_to_int(content_length, "Invalid content length");
|
732
|
-
if (int_content_length < 0) RAISE_BAD_REQUEST("Invalid body content length");
|
733
|
-
parser->body_read_mode = parser->body_left = int_content_length;
|
734
|
-
parser->request_completed = 0;
|
735
|
-
return;
|
736
|
-
}
|
737
|
-
|
738
|
-
VALUE transfer_encoding = rb_hash_aref(parser->headers, STR_transfer_encoding);
|
739
|
-
if (chunked_encoding_p(transfer_encoding)) {
|
740
|
-
parser->body_read_mode = BODY_READ_MODE_CHUNKED;
|
741
|
-
parser->request_completed = 0;
|
742
|
-
return;
|
743
|
-
}
|
744
|
-
parser->request_completed = 1;
|
745
|
-
|
746
|
-
}
|
747
|
-
|
748
|
-
static inline VALUE read_body(VALUE self, int read_entire_body, int buffered_only) {
|
749
|
-
Parser_t *parser;
|
750
|
-
GetParser(self, parser);
|
751
|
-
|
752
|
-
if (parser->body_read_mode == BODY_READ_MODE_UNKNOWN)
|
753
|
-
detect_body_read_mode(parser);
|
754
|
-
|
755
|
-
if (parser->body_read_mode == BODY_READ_MODE_CHUNKED)
|
756
|
-
return read_body_with_chunked_encoding(parser, read_entire_body, buffered_only);
|
757
|
-
return read_body_with_content_length(parser, read_entire_body, buffered_only);
|
758
|
-
}
|
759
|
-
|
760
|
-
VALUE Parser_read_body(VALUE self) {
|
761
|
-
return read_body(self, 1, 0);
|
762
|
-
}
|
763
|
-
|
764
|
-
VALUE Parser_read_body_chunk(VALUE self, VALUE buffered_only) {
|
765
|
-
return read_body(self, 0, buffered_only == Qtrue);
|
766
|
-
}
|
767
|
-
|
768
|
-
VALUE Parser_complete_p(VALUE self) {
|
769
|
-
Parser_t *parser;
|
770
|
-
GetParser(self, parser);
|
771
|
-
|
772
|
-
if (parser->body_read_mode == BODY_READ_MODE_UNKNOWN)
|
773
|
-
detect_body_read_mode(parser);
|
774
|
-
|
775
|
-
return parser->request_completed ? Qtrue : Qfalse;
|
776
|
-
}
|
777
|
-
|
778
|
-
void Init_HTTP1_Parser() {
|
779
|
-
VALUE mTipi;
|
780
|
-
VALUE cHTTP1Parser;
|
781
|
-
|
782
|
-
mTipi = rb_define_module("Tipi");
|
783
|
-
cHTTP1Parser = rb_define_class_under(mTipi, "HTTP1Parser", rb_cObject);
|
784
|
-
rb_define_alloc_func(cHTTP1Parser, Parser_allocate);
|
785
|
-
|
786
|
-
cError = rb_define_class_under(cHTTP1Parser, "Error", rb_eRuntimeError);
|
787
|
-
|
788
|
-
// backend methods
|
789
|
-
rb_define_method(cHTTP1Parser, "initialize", Parser_initialize, 1);
|
790
|
-
rb_define_method(cHTTP1Parser, "parse_headers", Parser_parse_headers, 0);
|
791
|
-
rb_define_method(cHTTP1Parser, "read_body", Parser_read_body, 0);
|
792
|
-
rb_define_method(cHTTP1Parser, "read_body_chunk", Parser_read_body_chunk, 1);
|
793
|
-
rb_define_method(cHTTP1Parser, "complete?", Parser_complete_p, 0);
|
794
|
-
|
795
|
-
ID_arity = rb_intern("arity");
|
796
|
-
ID_backend_read = rb_intern("backend_read");
|
797
|
-
ID_backend_recv = rb_intern("backend_recv");
|
798
|
-
ID_call = rb_intern("call");
|
799
|
-
ID_downcase = rb_intern("downcase");
|
800
|
-
ID_eq = rb_intern("==");
|
801
|
-
ID_parser_read_method = rb_intern("__parser_read_method__");
|
802
|
-
ID_read = rb_intern("read");
|
803
|
-
ID_readpartial = rb_intern("readpartial");
|
804
|
-
ID_to_i = rb_intern("to_i");
|
805
|
-
|
806
|
-
NUM_max_headers_read_length = INT2NUM(MAX_HEADERS_READ_LENGTH);
|
807
|
-
NUM_buffer_start = INT2NUM(0);
|
808
|
-
NUM_buffer_end = INT2NUM(-1);
|
809
|
-
|
810
|
-
GLOBAL_STR(STR_pseudo_method, ":method");
|
811
|
-
GLOBAL_STR(STR_pseudo_path, ":path");
|
812
|
-
GLOBAL_STR(STR_pseudo_protocol, ":protocol");
|
813
|
-
GLOBAL_STR(STR_pseudo_rx, ":rx");
|
814
|
-
|
815
|
-
GLOBAL_STR(STR_chunked, "chunked");
|
816
|
-
GLOBAL_STR(STR_content_length, "content-length");
|
817
|
-
GLOBAL_STR(STR_transfer_encoding, "transfer-encoding");
|
818
|
-
|
819
|
-
SYM_backend_read = ID2SYM(ID_backend_read);
|
820
|
-
SYM_backend_recv = ID2SYM(ID_backend_recv);
|
821
|
-
|
822
|
-
rb_global_variable(&mTipi);
|
823
|
-
}
|