buffer.gno
8.39 Kb ยท 462 lines
1package json
2
3import (
4 "errors"
5 "io"
6
7 "gno.land/p/demo/ufmt"
8)
9
10type buffer struct {
11 data []byte
12 length int
13 index int
14
15 last States
16 state States
17 class Classes
18}
19
20// newBuffer creates a new buffer with the given data
21func newBuffer(data []byte) *buffer {
22 return &buffer{
23 data: data,
24 length: len(data),
25 last: GO,
26 state: GO,
27 }
28}
29
30// first retrieves the first non-whitespace (or other escaped) character in the buffer.
31func (b *buffer) first() (byte, error) {
32 for ; b.index < b.length; b.index++ {
33 c := b.data[b.index]
34
35 if !(c == whiteSpace || c == carriageReturn || c == newLine || c == tab) {
36 return c, nil
37 }
38 }
39
40 return 0, io.EOF
41}
42
43// current returns the byte of the current index.
44func (b *buffer) current() (byte, error) {
45 if b.index >= b.length {
46 return 0, io.EOF
47 }
48
49 return b.data[b.index], nil
50}
51
52// next moves to the next byte and returns it.
53func (b *buffer) next() (byte, error) {
54 b.index++
55 return b.current()
56}
57
58// step just moves to the next position.
59func (b *buffer) step() error {
60 _, err := b.next()
61 return err
62}
63
64// move moves the index by the given position.
65func (b *buffer) move(pos int) error {
66 newIndex := b.index + pos
67
68 if newIndex > b.length {
69 return io.EOF
70 }
71
72 b.index = newIndex
73
74 return nil
75}
76
77// slice returns the slice from the current index to the given position.
78func (b *buffer) slice(pos int) ([]byte, error) {
79 end := b.index + pos
80
81 if end > b.length {
82 return nil, io.EOF
83 }
84
85 return b.data[b.index:end], nil
86}
87
88// sliceFromIndices returns a slice of the buffer's data starting from 'start' up to (but not including) 'stop'.
89func (b *buffer) sliceFromIndices(start, stop int) []byte {
90 if start > b.length {
91 start = b.length
92 }
93
94 if stop > b.length {
95 stop = b.length
96 }
97
98 return b.data[start:stop]
99}
100
101// skip moves the index to skip the given byte.
102func (b *buffer) skip(bs byte) error {
103 for b.index < b.length {
104 if b.data[b.index] == bs && !b.backslash() {
105 return nil
106 }
107
108 b.index++
109 }
110
111 return io.EOF
112}
113
114// skipAndReturnIndex moves the buffer index forward by one and returns the new index.
115func (b *buffer) skipAndReturnIndex() (int, error) {
116 err := b.step()
117 if err != nil {
118 return 0, err
119 }
120
121 return b.index, nil
122}
123
124// skipUntil moves the buffer index forward until it encounters a byte contained in the endTokens set.
125func (b *buffer) skipUntil(endTokens map[byte]bool) (int, error) {
126 for b.index < b.length {
127 currentByte, err := b.current()
128 if err != nil {
129 return b.index, err
130 }
131
132 // Check if the current byte is in the set of end tokens.
133 if _, exists := endTokens[currentByte]; exists {
134 return b.index, nil
135 }
136
137 b.index++
138 }
139
140 return b.index, io.EOF
141}
142
143// significantTokens is a map where the keys are the significant characters in a JSON path.
144// The values in the map are all true, which allows us to use the map as a set for quick lookups.
145var significantTokens = [256]bool{
146 dot: true, // access properties of an object
147 dollarSign: true, // root object
148 atSign: true, // current object
149 bracketOpen: true, // start of an array index or filter expression
150 bracketClose: true, // end of an array index or filter expression
151}
152
153// filterTokens stores the filter expression tokens.
154var filterTokens = [256]bool{
155 aesterisk: true, // wildcard
156 andSign: true,
157 orSign: true,
158}
159
160// skipToNextSignificantToken advances the buffer index to the next significant character.
161// Significant characters are defined based on the JSON path syntax.
162func (b *buffer) skipToNextSignificantToken() {
163 for b.index < b.length {
164 current := b.data[b.index]
165
166 if significantTokens[current] {
167 break
168 }
169
170 b.index++
171 }
172}
173
174// backslash checks to see if the number of backslashes before the current index is odd.
175//
176// This is used to check if the current character is escaped. However, unlike the "unescape" function,
177// "backslash" only serves to check the number of backslashes.
178func (b *buffer) backslash() bool {
179 if b.index == 0 {
180 return false
181 }
182
183 count := 0
184 for i := b.index - 1; ; i-- {
185 if b.data[i] != backSlash {
186 break
187 }
188
189 count++
190
191 if i == 0 {
192 break
193 }
194 }
195
196 return count%2 != 0
197}
198
199// numIndex holds a map of valid numeric characters
200var numIndex = [256]bool{
201 '0': true,
202 '1': true,
203 '2': true,
204 '3': true,
205 '4': true,
206 '5': true,
207 '6': true,
208 '7': true,
209 '8': true,
210 '9': true,
211 '.': true,
212 'e': true,
213 'E': true,
214}
215
216// pathToken checks if the current token is a valid JSON path token.
217func (b *buffer) pathToken() error {
218 var stack []byte
219
220 inToken := false
221 inNumber := false
222 first := b.index
223
224 for b.index < b.length {
225 c := b.data[b.index]
226
227 switch {
228 case c == doubleQuote || c == singleQuote:
229 inToken = true
230 if err := b.step(); err != nil {
231 return errors.New("error stepping through buffer")
232 }
233
234 if err := b.skip(c); err != nil {
235 return errUnmatchedQuotePath
236 }
237
238 if b.index >= b.length {
239 return errUnmatchedQuotePath
240 }
241
242 case c == bracketOpen || c == parenOpen:
243 inToken = true
244 stack = append(stack, c)
245
246 case c == bracketClose || c == parenClose:
247 inToken = true
248 if len(stack) == 0 || (c == bracketClose && stack[len(stack)-1] != bracketOpen) || (c == parenClose && stack[len(stack)-1] != parenOpen) {
249 return errUnmatchedParenthesis
250 }
251
252 stack = stack[:len(stack)-1]
253
254 case pathStateContainsValidPathToken(c):
255 inToken = true
256
257 case c == plus || c == minus:
258 if inNumber || (b.index > 0 && numIndex[b.data[b.index-1]]) {
259 inToken = true
260 } else if !inToken && (b.index+1 < b.length && numIndex[b.data[b.index+1]]) {
261 inToken = true
262 inNumber = true
263 } else if !inToken {
264 return errInvalidToken
265 }
266
267 default:
268 if len(stack) != 0 || inToken {
269 inToken = true
270 } else {
271 goto end
272 }
273 }
274
275 b.index++
276 }
277
278end:
279 if len(stack) != 0 {
280 return errUnmatchedParenthesis
281 }
282
283 if first == b.index {
284 return errors.New("no token found")
285 }
286
287 if inNumber && !numIndex[b.data[b.index-1]] {
288 inNumber = false
289 }
290
291 return nil
292}
293
294func pathStateContainsValidPathToken(c byte) bool {
295 if significantTokens[c] {
296 return true
297 }
298
299 if filterTokens[c] {
300 return true
301 }
302
303 if numIndex[c] {
304 return true
305 }
306
307 if 'A' <= c && c <= 'Z' || 'a' <= c && c <= 'z' {
308 return true
309 }
310
311 return false
312}
313
314func (b *buffer) numeric(token bool) error {
315 if token {
316 b.last = GO
317 }
318
319 for ; b.index < b.length; b.index++ {
320 b.class = b.getClasses(doubleQuote)
321 if b.class == __ {
322 return errInvalidToken
323 }
324
325 b.state = StateTransitionTable[b.last][b.class]
326 if b.state == __ {
327 if token {
328 break
329 }
330
331 return errInvalidToken
332 }
333
334 if b.state < __ {
335 return nil
336 }
337
338 if b.state < MI || b.state > E3 {
339 return nil
340 }
341
342 b.last = b.state
343 }
344
345 if b.last != ZE && b.last != IN && b.last != FR && b.last != E3 {
346 return errInvalidToken
347 }
348
349 return nil
350}
351
352func (b *buffer) getClasses(c byte) Classes {
353 if b.data[b.index] >= 128 {
354 return C_ETC
355 }
356
357 if c == singleQuote {
358 return QuoteAsciiClasses[b.data[b.index]]
359 }
360
361 return AsciiClasses[b.data[b.index]]
362}
363
364func (b *buffer) getState() States {
365 b.last = b.state
366
367 b.class = b.getClasses(doubleQuote)
368 if b.class == __ {
369 return __
370 }
371
372 b.state = StateTransitionTable[b.last][b.class]
373
374 return b.state
375}
376
377// string parses a string token from the buffer.
378func (b *buffer) string(search byte, token bool) error {
379 if token {
380 b.last = GO
381 }
382
383 for ; b.index < b.length; b.index++ {
384 b.class = b.getClasses(search)
385
386 if b.class == __ {
387 return errInvalidToken
388 }
389
390 b.state = StateTransitionTable[b.last][b.class]
391 if b.state == __ {
392 return errInvalidToken
393 }
394
395 if b.state < __ {
396 break
397 }
398
399 b.last = b.state
400 }
401
402 return nil
403}
404
405func (b *buffer) word(bs []byte) error {
406 var c byte
407
408 max := len(bs)
409 index := 0
410
411 for ; b.index < b.length && index < max; b.index++ {
412 c = b.data[b.index]
413
414 if c != bs[index] {
415 return errInvalidToken
416 }
417
418 index++
419 if index >= max {
420 break
421 }
422 }
423
424 if index != max {
425 return errInvalidToken
426 }
427
428 return nil
429}
430
431func numberKind2f64(value any) (result float64, err error) {
432 switch typed := value.(type) {
433 case float64:
434 result = typed
435 case float32:
436 result = float64(typed)
437 case int:
438 result = float64(typed)
439 case int8:
440 result = float64(typed)
441 case int16:
442 result = float64(typed)
443 case int32:
444 result = float64(typed)
445 case int64:
446 result = float64(typed)
447 case uint:
448 result = float64(typed)
449 case uint8:
450 result = float64(typed)
451 case uint16:
452 result = float64(typed)
453 case uint32:
454 result = float64(typed)
455 case uint64:
456 result = float64(typed)
457 default:
458 err = ufmt.Errorf("invalid number type: %T", value)
459 }
460
461 return
462}
