blob: b7de0a89c462af605f889bc46ce165e5d4238add [file] [log] [blame]
mpagenkoaf801632020-07-03 10:00:42 +00001//
2// Copyright (c) 2011-2019 Canonical Ltd
3// Copyright (c) 2006-2010 Kirill Simonov
4//
5// Permission is hereby granted, free of charge, to any person obtaining a copy of
6// this software and associated documentation files (the "Software"), to deal in
7// the Software without restriction, including without limitation the rights to
8// use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
9// of the Software, and to permit persons to whom the Software is furnished to do
10// so, subject to the following conditions:
11//
12// The above copyright notice and this permission notice shall be included in all
13// copies or substantial portions of the Software.
14//
15// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21// SOFTWARE.
22
Holger Hildebrandtda7758b2020-03-16 11:30:03 +000023package yaml
24
25import (
26 "io"
27)
28
29// Set the reader error and return 0.
30func yaml_parser_set_reader_error(parser *yaml_parser_t, problem string, offset int, value int) bool {
31 parser.error = yaml_READER_ERROR
32 parser.problem = problem
33 parser.problem_offset = offset
34 parser.problem_value = value
35 return false
36}
37
38// Byte order marks.
39const (
40 bom_UTF8 = "\xef\xbb\xbf"
41 bom_UTF16LE = "\xff\xfe"
42 bom_UTF16BE = "\xfe\xff"
43)
44
45// Determine the input stream encoding by checking the BOM symbol. If no BOM is
46// found, the UTF-8 encoding is assumed. Return 1 on success, 0 on failure.
47func yaml_parser_determine_encoding(parser *yaml_parser_t) bool {
48 // Ensure that we had enough bytes in the raw buffer.
49 for !parser.eof && len(parser.raw_buffer)-parser.raw_buffer_pos < 3 {
50 if !yaml_parser_update_raw_buffer(parser) {
51 return false
52 }
53 }
54
55 // Determine the encoding.
56 buf := parser.raw_buffer
57 pos := parser.raw_buffer_pos
58 avail := len(buf) - pos
59 if avail >= 2 && buf[pos] == bom_UTF16LE[0] && buf[pos+1] == bom_UTF16LE[1] {
60 parser.encoding = yaml_UTF16LE_ENCODING
61 parser.raw_buffer_pos += 2
62 parser.offset += 2
63 } else if avail >= 2 && buf[pos] == bom_UTF16BE[0] && buf[pos+1] == bom_UTF16BE[1] {
64 parser.encoding = yaml_UTF16BE_ENCODING
65 parser.raw_buffer_pos += 2
66 parser.offset += 2
67 } else if avail >= 3 && buf[pos] == bom_UTF8[0] && buf[pos+1] == bom_UTF8[1] && buf[pos+2] == bom_UTF8[2] {
68 parser.encoding = yaml_UTF8_ENCODING
69 parser.raw_buffer_pos += 3
70 parser.offset += 3
71 } else {
72 parser.encoding = yaml_UTF8_ENCODING
73 }
74 return true
75}
76
77// Update the raw buffer.
78func yaml_parser_update_raw_buffer(parser *yaml_parser_t) bool {
79 size_read := 0
80
81 // Return if the raw buffer is full.
82 if parser.raw_buffer_pos == 0 && len(parser.raw_buffer) == cap(parser.raw_buffer) {
83 return true
84 }
85
86 // Return on EOF.
87 if parser.eof {
88 return true
89 }
90
91 // Move the remaining bytes in the raw buffer to the beginning.
92 if parser.raw_buffer_pos > 0 && parser.raw_buffer_pos < len(parser.raw_buffer) {
93 copy(parser.raw_buffer, parser.raw_buffer[parser.raw_buffer_pos:])
94 }
95 parser.raw_buffer = parser.raw_buffer[:len(parser.raw_buffer)-parser.raw_buffer_pos]
96 parser.raw_buffer_pos = 0
97
98 // Call the read handler to fill the buffer.
99 size_read, err := parser.read_handler(parser, parser.raw_buffer[len(parser.raw_buffer):cap(parser.raw_buffer)])
100 parser.raw_buffer = parser.raw_buffer[:len(parser.raw_buffer)+size_read]
101 if err == io.EOF {
102 parser.eof = true
103 } else if err != nil {
104 return yaml_parser_set_reader_error(parser, "input error: "+err.Error(), parser.offset, -1)
105 }
106 return true
107}
108
109// Ensure that the buffer contains at least `length` characters.
110// Return true on success, false on failure.
111//
112// The length is supposed to be significantly less that the buffer size.
113func yaml_parser_update_buffer(parser *yaml_parser_t, length int) bool {
114 if parser.read_handler == nil {
115 panic("read handler must be set")
116 }
117
118 // [Go] This function was changed to guarantee the requested length size at EOF.
119 // The fact we need to do this is pretty awful, but the description above implies
mpagenkoaf801632020-07-03 10:00:42 +0000120 // for that to be the case, and there are tests
Holger Hildebrandtda7758b2020-03-16 11:30:03 +0000121
122 // If the EOF flag is set and the raw buffer is empty, do nothing.
123 if parser.eof && parser.raw_buffer_pos == len(parser.raw_buffer) {
124 // [Go] ACTUALLY! Read the documentation of this function above.
125 // This is just broken. To return true, we need to have the
126 // given length in the buffer. Not doing that means every single
127 // check that calls this function to make sure the buffer has a
128 // given length is Go) panicking; or C) accessing invalid memory.
129 //return true
130 }
131
132 // Return if the buffer contains enough characters.
133 if parser.unread >= length {
134 return true
135 }
136
137 // Determine the input encoding if it is not known yet.
138 if parser.encoding == yaml_ANY_ENCODING {
139 if !yaml_parser_determine_encoding(parser) {
140 return false
141 }
142 }
143
144 // Move the unread characters to the beginning of the buffer.
145 buffer_len := len(parser.buffer)
146 if parser.buffer_pos > 0 && parser.buffer_pos < buffer_len {
147 copy(parser.buffer, parser.buffer[parser.buffer_pos:])
148 buffer_len -= parser.buffer_pos
149 parser.buffer_pos = 0
150 } else if parser.buffer_pos == buffer_len {
151 buffer_len = 0
152 parser.buffer_pos = 0
153 }
154
155 // Open the whole buffer for writing, and cut it before returning.
156 parser.buffer = parser.buffer[:cap(parser.buffer)]
157
158 // Fill the buffer until it has enough characters.
159 first := true
160 for parser.unread < length {
161
162 // Fill the raw buffer if necessary.
163 if !first || parser.raw_buffer_pos == len(parser.raw_buffer) {
164 if !yaml_parser_update_raw_buffer(parser) {
165 parser.buffer = parser.buffer[:buffer_len]
166 return false
167 }
168 }
169 first = false
170
171 // Decode the raw buffer.
172 inner:
173 for parser.raw_buffer_pos != len(parser.raw_buffer) {
174 var value rune
175 var width int
176
177 raw_unread := len(parser.raw_buffer) - parser.raw_buffer_pos
178
179 // Decode the next character.
180 switch parser.encoding {
181 case yaml_UTF8_ENCODING:
182 // Decode a UTF-8 character. Check RFC 3629
183 // (http://www.ietf.org/rfc/rfc3629.txt) for more details.
184 //
185 // The following table (taken from the RFC) is used for
186 // decoding.
187 //
188 // Char. number range | UTF-8 octet sequence
189 // (hexadecimal) | (binary)
190 // --------------------+------------------------------------
191 // 0000 0000-0000 007F | 0xxxxxxx
192 // 0000 0080-0000 07FF | 110xxxxx 10xxxxxx
193 // 0000 0800-0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx
194 // 0001 0000-0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
195 //
196 // Additionally, the characters in the range 0xD800-0xDFFF
197 // are prohibited as they are reserved for use with UTF-16
198 // surrogate pairs.
199
200 // Determine the length of the UTF-8 sequence.
201 octet := parser.raw_buffer[parser.raw_buffer_pos]
202 switch {
203 case octet&0x80 == 0x00:
204 width = 1
205 case octet&0xE0 == 0xC0:
206 width = 2
207 case octet&0xF0 == 0xE0:
208 width = 3
209 case octet&0xF8 == 0xF0:
210 width = 4
211 default:
212 // The leading octet is invalid.
213 return yaml_parser_set_reader_error(parser,
214 "invalid leading UTF-8 octet",
215 parser.offset, int(octet))
216 }
217
218 // Check if the raw buffer contains an incomplete character.
219 if width > raw_unread {
220 if parser.eof {
221 return yaml_parser_set_reader_error(parser,
222 "incomplete UTF-8 octet sequence",
223 parser.offset, -1)
224 }
225 break inner
226 }
227
228 // Decode the leading octet.
229 switch {
230 case octet&0x80 == 0x00:
231 value = rune(octet & 0x7F)
232 case octet&0xE0 == 0xC0:
233 value = rune(octet & 0x1F)
234 case octet&0xF0 == 0xE0:
235 value = rune(octet & 0x0F)
236 case octet&0xF8 == 0xF0:
237 value = rune(octet & 0x07)
238 default:
239 value = 0
240 }
241
242 // Check and decode the trailing octets.
243 for k := 1; k < width; k++ {
244 octet = parser.raw_buffer[parser.raw_buffer_pos+k]
245
246 // Check if the octet is valid.
247 if (octet & 0xC0) != 0x80 {
248 return yaml_parser_set_reader_error(parser,
249 "invalid trailing UTF-8 octet",
250 parser.offset+k, int(octet))
251 }
252
253 // Decode the octet.
254 value = (value << 6) + rune(octet&0x3F)
255 }
256
257 // Check the length of the sequence against the value.
258 switch {
259 case width == 1:
260 case width == 2 && value >= 0x80:
261 case width == 3 && value >= 0x800:
262 case width == 4 && value >= 0x10000:
263 default:
264 return yaml_parser_set_reader_error(parser,
265 "invalid length of a UTF-8 sequence",
266 parser.offset, -1)
267 }
268
269 // Check the range of the value.
270 if value >= 0xD800 && value <= 0xDFFF || value > 0x10FFFF {
271 return yaml_parser_set_reader_error(parser,
272 "invalid Unicode character",
273 parser.offset, int(value))
274 }
275
276 case yaml_UTF16LE_ENCODING, yaml_UTF16BE_ENCODING:
277 var low, high int
278 if parser.encoding == yaml_UTF16LE_ENCODING {
279 low, high = 0, 1
280 } else {
281 low, high = 1, 0
282 }
283
284 // The UTF-16 encoding is not as simple as one might
285 // naively think. Check RFC 2781
286 // (http://www.ietf.org/rfc/rfc2781.txt).
287 //
288 // Normally, two subsequent bytes describe a Unicode
289 // character. However a special technique (called a
290 // surrogate pair) is used for specifying character
291 // values larger than 0xFFFF.
292 //
293 // A surrogate pair consists of two pseudo-characters:
294 // high surrogate area (0xD800-0xDBFF)
295 // low surrogate area (0xDC00-0xDFFF)
296 //
297 // The following formulas are used for decoding
298 // and encoding characters using surrogate pairs:
299 //
300 // U = U' + 0x10000 (0x01 00 00 <= U <= 0x10 FF FF)
301 // U' = yyyyyyyyyyxxxxxxxxxx (0 <= U' <= 0x0F FF FF)
302 // W1 = 110110yyyyyyyyyy
303 // W2 = 110111xxxxxxxxxx
304 //
305 // where U is the character value, W1 is the high surrogate
306 // area, W2 is the low surrogate area.
307
308 // Check for incomplete UTF-16 character.
309 if raw_unread < 2 {
310 if parser.eof {
311 return yaml_parser_set_reader_error(parser,
312 "incomplete UTF-16 character",
313 parser.offset, -1)
314 }
315 break inner
316 }
317
318 // Get the character.
319 value = rune(parser.raw_buffer[parser.raw_buffer_pos+low]) +
320 (rune(parser.raw_buffer[parser.raw_buffer_pos+high]) << 8)
321
322 // Check for unexpected low surrogate area.
323 if value&0xFC00 == 0xDC00 {
324 return yaml_parser_set_reader_error(parser,
325 "unexpected low surrogate area",
326 parser.offset, int(value))
327 }
328
329 // Check for a high surrogate area.
330 if value&0xFC00 == 0xD800 {
331 width = 4
332
333 // Check for incomplete surrogate pair.
334 if raw_unread < 4 {
335 if parser.eof {
336 return yaml_parser_set_reader_error(parser,
337 "incomplete UTF-16 surrogate pair",
338 parser.offset, -1)
339 }
340 break inner
341 }
342
343 // Get the next character.
344 value2 := rune(parser.raw_buffer[parser.raw_buffer_pos+low+2]) +
345 (rune(parser.raw_buffer[parser.raw_buffer_pos+high+2]) << 8)
346
347 // Check for a low surrogate area.
348 if value2&0xFC00 != 0xDC00 {
349 return yaml_parser_set_reader_error(parser,
350 "expected low surrogate area",
351 parser.offset+2, int(value2))
352 }
353
354 // Generate the value of the surrogate pair.
355 value = 0x10000 + ((value & 0x3FF) << 10) + (value2 & 0x3FF)
356 } else {
357 width = 2
358 }
359
360 default:
361 panic("impossible")
362 }
363
364 // Check if the character is in the allowed range:
365 // #x9 | #xA | #xD | [#x20-#x7E] (8 bit)
366 // | #x85 | [#xA0-#xD7FF] | [#xE000-#xFFFD] (16 bit)
367 // | [#x10000-#x10FFFF] (32 bit)
368 switch {
369 case value == 0x09:
370 case value == 0x0A:
371 case value == 0x0D:
372 case value >= 0x20 && value <= 0x7E:
373 case value == 0x85:
374 case value >= 0xA0 && value <= 0xD7FF:
375 case value >= 0xE000 && value <= 0xFFFD:
376 case value >= 0x10000 && value <= 0x10FFFF:
377 default:
378 return yaml_parser_set_reader_error(parser,
379 "control characters are not allowed",
380 parser.offset, int(value))
381 }
382
383 // Move the raw pointers.
384 parser.raw_buffer_pos += width
385 parser.offset += width
386
387 // Finally put the character into the buffer.
388 if value <= 0x7F {
389 // 0000 0000-0000 007F . 0xxxxxxx
390 parser.buffer[buffer_len+0] = byte(value)
391 buffer_len += 1
392 } else if value <= 0x7FF {
393 // 0000 0080-0000 07FF . 110xxxxx 10xxxxxx
394 parser.buffer[buffer_len+0] = byte(0xC0 + (value >> 6))
395 parser.buffer[buffer_len+1] = byte(0x80 + (value & 0x3F))
396 buffer_len += 2
397 } else if value <= 0xFFFF {
398 // 0000 0800-0000 FFFF . 1110xxxx 10xxxxxx 10xxxxxx
399 parser.buffer[buffer_len+0] = byte(0xE0 + (value >> 12))
400 parser.buffer[buffer_len+1] = byte(0x80 + ((value >> 6) & 0x3F))
401 parser.buffer[buffer_len+2] = byte(0x80 + (value & 0x3F))
402 buffer_len += 3
403 } else {
404 // 0001 0000-0010 FFFF . 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
405 parser.buffer[buffer_len+0] = byte(0xF0 + (value >> 18))
406 parser.buffer[buffer_len+1] = byte(0x80 + ((value >> 12) & 0x3F))
407 parser.buffer[buffer_len+2] = byte(0x80 + ((value >> 6) & 0x3F))
408 parser.buffer[buffer_len+3] = byte(0x80 + (value & 0x3F))
409 buffer_len += 4
410 }
411
412 parser.unread++
413 }
414
415 // On EOF, put NUL into the buffer and return.
416 if parser.eof {
417 parser.buffer[buffer_len] = 0
418 buffer_len++
419 parser.unread++
420 break
421 }
422 }
423 // [Go] Read the documentation of this function above. To return true,
424 // we need to have the given length in the buffer. Not doing that means
425 // every single check that calls this function to make sure the buffer
426 // has a given length is Go) panicking; or C) accessing invalid memory.
427 // This happens here due to the EOF above breaking early.
428 for buffer_len < length {
429 parser.buffer[buffer_len] = 0
430 buffer_len++
431 }
432 parser.buffer = parser.buffer[:buffer_len]
433 return true
434}