Libraryless. Click here for Pure Java version (4412L/27K).
1 | static Object unstructure(String text) {
|
2 | ret unstructure(text, false); |
3 | } |
4 | |
5 | static Object unstructure(String text, final boolean allDynamic) {
|
6 | ret unstructure(text, allDynamic, null); |
7 | } |
8 | |
9 | static int structure_internStringsLongerThan = 50; |
10 | static int unstructure_unquoteBufSize = 100; |
11 | |
12 | static int unstructure_tokrefs; // stats |
13 | |
14 | abstract sclass unstructure_Receiver {
|
15 | abstract void set(O o); |
16 | } |
17 | |
18 | // classFinder: func(name) -> class (optional) |
19 | static Object unstructure(String text, boolean allDynamic, |
20 | O classFinder) {
|
21 | if (text == null) ret null; |
22 | ret unstructure_tok(javaTokC_noMLS_iterator(text), allDynamic, classFinder); |
23 | } |
24 | |
25 | static O unstructure_reader(BufferedReader reader) {
|
26 | ret unstructure_tok(javaTokC_noMLS_onReader(reader), false, null); |
27 | } |
28 | |
29 | static O unstructure_tok(final Producer<S> tok, final boolean allDynamic, final O _classFinder) {
|
30 | final boolean debug = unstructure_debug; |
31 | |
32 | final class X {
|
33 | int i = -1; |
34 | final O classFinder = _classFinder != null ? _classFinder : _defaultClassFinder(); |
35 | new HashMap<Integer, O> refs; |
36 | new HashMap<Integer, O> tokrefs; |
37 | new HashSet<S> concepts; |
38 | new HashMap<S, Class> classesMap; |
39 | new L<Runnable> stack; |
40 | S curT; |
41 | char[] unquoteBuf = new char[unstructure_unquoteBufSize]; |
42 | |
43 | S unquote(S s) {
|
44 | ret unquoteUsingCharArray(s, unquoteBuf); |
45 | } |
46 | |
47 | // look at current token |
48 | S t() {
|
49 | ret curT; |
50 | } |
51 | |
52 | // get current token, move to next |
53 | S tpp() {
|
54 | S t = curT; |
55 | consume(); |
56 | ret t; |
57 | } |
58 | |
59 | void parse(final unstructure_Receiver out) {
|
60 | S t = t(); |
61 | |
62 | int refID = 0; |
63 | if (structure_isMarker(t, 0, l(t))) {
|
64 | refID = parseInt(t.substring(1)); |
65 | consume(); |
66 | } |
67 | final int _refID = refID; |
68 | |
69 | // if (debug) print("parse: " + quote(t));
|
70 | |
71 | final int tokIndex = i; |
72 | parse_inner(refID, tokIndex, new unstructure_Receiver {
|
73 | void set(O o) {
|
74 | if (_refID != 0) |
75 | refs.put(_refID, o); |
76 | if (o != null) |
77 | tokrefs.put(tokIndex, o); |
78 | out.set(o); |
79 | } |
80 | }); |
81 | } |
82 | |
83 | void parse_inner(int refID, int tokIndex, final unstructure_Receiver out) {
|
84 | S t = t(); |
85 | |
86 | // if (debug) print("parse_inner: " + quote(t));
|
87 | |
88 | Class c = classesMap.get(t); |
89 | if (c == null) {
|
90 | if (t.startsWith("\"")) {
|
91 | S s = internIfLongerThan(unquote(tpp()), structure_internStringsLongerThan); |
92 | out.set(s); ret; |
93 | } |
94 | |
95 | if (t.startsWith("'")) {
|
96 | out.set(unquoteCharacter(tpp())); ret; |
97 | } |
98 | if (t.equals("bigint")) {
|
99 | out.set(parseBigInt()); ret; |
100 | } |
101 | if (t.equals("d")) {
|
102 | out.set(parseDouble()); ret; |
103 | } |
104 | if (t.equals("fl")) {
|
105 | out.set(parseFloat()); ret; |
106 | } |
107 | if (t.equals("sh")) {
|
108 | consume(); |
109 | t = tpp(); |
110 | if (t.equals("-")) {
|
111 | t = tpp(); |
112 | out.set((short) (-parseInt(t)); ret; |
113 | } |
114 | out.set((short) parseInt(t)); ret; |
115 | } |
116 | if (t.equals("-")) {
|
117 | consume(); |
118 | t = tpp(); |
119 | out.set(isLongConstant(t) ? (O) (-parseLong(t)) : (O) (-parseInt(t))); ret; |
120 | } |
121 | if (isInteger(t) || isLongConstant(t)) {
|
122 | consume(); |
123 | //if (debug) print("isLongConstant " + quote(t) + " => " + isLongConstant(t));
|
124 | if (isLongConstant(t)) {
|
125 | out.set(parseLong(t)); ret; |
126 | } |
127 | long l = parseLong(t); |
128 | bool isInt = l == (int) l; |
129 | ifdef unstructure_debug |
130 | print("l=" + l + ", isInt: " + isInt);
|
131 | endifdef |
132 | out.set(isInt ? (O) Integer.valueOf((int) l) : (O) Long.valueOf(l)); ret; |
133 | } |
134 | if (t.equals("false") || t.equals("f")) {
|
135 | consume(); out.set(false); ret; |
136 | } |
137 | if (t.equals("true") || t.equals("t")) {
|
138 | consume(); out.set(true); ret; |
139 | } |
140 | if (t.equals("-")) {
|
141 | consume(); |
142 | t = tpp(); |
143 | out.set(isLongConstant(t) ? (O) (-parseLong(t)) : (O) (-parseInt(t))); ret; |
144 | } |
145 | if (isInteger(t) || isLongConstant(t)) {
|
146 | consume(); |
147 | //if (debug) print("isLongConstant " + quote(t) + " => " + isLongConstant(t));
|
148 | if (isLongConstant(t)) {
|
149 | out.set(parseLong(t)); ret; |
150 | } |
151 | long l = parseLong(t); |
152 | bool isInt = l == (int) l; |
153 | ifdef unstructure_debug |
154 | print("l=" + l + ", isInt: " + isInt);
|
155 | endifdef |
156 | out.set(isInt ? (O) Integer.valueOf((int) l) : (O) Long.valueOf(l)); ret; |
157 | } |
158 | |
159 | if (t.equals("File")) {
|
160 | consume(); |
161 | File f = new File(unquote(tpp())); |
162 | out.set(f); ret; |
163 | } |
164 | |
165 | if (t.startsWith("r") && isInteger(t.substring(1))) {
|
166 | consume(); |
167 | int ref = Integer.parseInt(t.substring(1)); |
168 | O o = refs.get(ref); |
169 | if (o == null) |
170 | fail("unsatisfied back reference " + ref);
|
171 | out.set(o); ret; |
172 | } |
173 | |
174 | if (t.startsWith("t") && isInteger(t.substring(1))) {
|
175 | consume(); |
176 | int ref = Integer.parseInt(t.substring(1)); |
177 | O o = tokrefs.get(ref); |
178 | if (o == null) |
179 | fail("unsatisfied token reference " + ref + " at " + tokIndex);
|
180 | out.set(o); ret; |
181 | } |
182 | |
183 | if (t.equals("hashset")) ret with parseHashSet(out);
|
184 | if (t.equals("lhs")) ret with parseLinkedHashSet(out);
|
185 | if (t.equals("treeset")) ret with parseTreeSet(out);
|
186 | if (t.equals("ciset")) ret with parseCISet(out);
|
187 | |
188 | if (eqOneOf(t, "hashmap", "hm")) {
|
189 | consume(); |
190 | parseMap(new HashMap, out); |
191 | ret; |
192 | } |
193 | if (t.equals("lhm")) {
|
194 | consume(); |
195 | parseMap(new LinkedHashMap, out); |
196 | ret; |
197 | } |
198 | if (t.equals("tm")) {
|
199 | consume(); |
200 | parseMap(new TreeMap, out); |
201 | ret; |
202 | } |
203 | if (t.equals("cimap")) {
|
204 | consume(); |
205 | parseMap(ciMap(), out); |
206 | ret; |
207 | } |
208 | |
209 | if (t.equals("ll")) {
|
210 | consume(); |
211 | ret with parseList(new LinkedList, out); |
212 | } |
213 | |
214 | if (t.equals("syncLL")) {
|
215 | consume(); |
216 | ret with parseList(synchroLinkedList(), out); |
217 | } |
218 | |
219 | if (t.equals("sync")) {
|
220 | consume(); |
221 | if (t().equals("tm")) {
|
222 | consume(); |
223 | ret with parseMap(synchronizedTreeMap(), out); |
224 | } |
225 | if (t().equals("["))
|
226 | ret with parseList(synchroList(), out); |
227 | ret with parseMap(synchronizedMap(), out); |
228 | } |
229 | if (t.equals("{")) {
|
230 | parseMap(out); ret; |
231 | } |
232 | if (t.equals("[")) {
|
233 | this.parseList(new ArrayList, out); ret; |
234 | } |
235 | if (t.equals("bitset")) {
|
236 | parseBitSet(out); ret; |
237 | } |
238 | if (t.equals("array") || t.equals("intarray")) {
|
239 | parseArray(out); ret; |
240 | } |
241 | if (t.equals("ba")) {
|
242 | consume(); |
243 | S hex = unquote(tpp()); |
244 | out.set(hexToBytes(hex)); ret; |
245 | } |
246 | if (t.equals("boolarray")) {
|
247 | consume(); |
248 | int n = parseInt(tpp()); |
249 | S hex = unquote(tpp()); |
250 | out.set(boolArrayFromBytes(hexToBytes(hex), n)); ret; |
251 | } |
252 | if (t.equals("class")) {
|
253 | out.set(parseClass()); ret; |
254 | } |
255 | if (t.equals("l")) {
|
256 | parseLisp(out); ret; |
257 | } |
258 | if (t.equals("null")) {
|
259 | consume(); out.set(null); ret; |
260 | } |
261 | |
262 | if (eq(t, "c")) {
|
263 | consume(); |
264 | t = t(); |
265 | assertTrue(isJavaIdentifier(t)); |
266 | concepts.add(t); |
267 | } |
268 | |
269 | // custom deserialization |
270 | if (eq(t, "cu")) {
|
271 | consume(); |
272 | t = tpp(); |
273 | assertTrue(isJavaIdentifier(t)); |
274 | S fullClassName = "main$" + t; |
275 | c = classFinder != null ? (Class) callF(classFinder, fullClassName) : findClass_fullName(fullClassName); |
276 | if (c == null) fail("Class not found: " + fullClassName);
|
277 | O o = nuEmptyObject(c); |
278 | out.set(o); |
279 | parse(new unstructure_Receiver {
|
280 | void set(O value) {
|
281 | call(o, "_deserialize", value); |
282 | } |
283 | }); |
284 | ret; |
285 | } |
286 | } |
287 | |
288 | if (eq(t, "j")) {
|
289 | consume("j");
|
290 | out.set(parseJava()); ret; |
291 | } |
292 | |
293 | if (c == null && !isJavaIdentifier(t)) |
294 | throw new RuntimeException("Unknown token " + (i+1) + ": " + t);
|
295 | |
296 | // any other class name (or package name) |
297 | consume(); |
298 | S className, fullClassName; |
299 | |
300 | // Is it a package name? |
301 | if (eq(t(), ".")) {
|
302 | consume(); |
303 | className = fullClassName = t + "." + assertIdentifier(tpp()); |
304 | } else {
|
305 | className = t; |
306 | fullClassName = "main$" + t; |
307 | } |
308 | |
309 | if (c == null) {
|
310 | // First, find class |
311 | if (allDynamic) c = null; |
312 | else c = classFinder != null ? (Class) callF(classFinder, fullClassName) : findClass_fullName(fullClassName); |
313 | if (c != null) |
314 | classesMap.put(className, c); |
315 | } |
316 | |
317 | // Check if it has an outer reference |
318 | bool hasBracket = eq(t(), "(");
|
319 | if (hasBracket) consume(); |
320 | bool hasOuter = hasBracket && eq(t(), "this$1"); |
321 | |
322 | DynamicObject dO = null; |
323 | O o = null; |
324 | fS thingName = t; |
325 | if (c != null) {
|
326 | o = hasOuter ? nuStubInnerObject(c, classFinder) : nuEmptyObject(c); |
327 | if (o instanceof DynamicObject) dO = (DynamicObject) o; |
328 | } else {
|
329 | if (concepts.contains(t) && (c = findClass("Concept")) != null)
|
330 | o = dO = (DynamicObject) nuEmptyObject(c); |
331 | else |
332 | dO = new DynamicObject; |
333 | dO.className = className; |
334 | ifdef unstructure_debug |
335 | print("Made dynamic object " + t + " " + shortClassName(dO));
|
336 | endifdef |
337 | } |
338 | |
339 | // Save in references list early because contents of object |
340 | // might link back to main object |
341 | |
342 | if (refID != 0) |
343 | refs.put(refID, o != null ? o : dO); |
344 | tokrefs.put(tokIndex, o != null ? o : dO); |
345 | |
346 | // NOW parse the fields! |
347 | |
348 | final new LinkedHashMap<S, O> fields; // preserve order |
349 | final O _o = o; |
350 | final DynamicObject _dO = dO; |
351 | if (hasBracket) {
|
352 | stack.add(r {
|
353 | if (eq(t(), ")")) {
|
354 | consume(")");
|
355 | objRead(_o, _dO, fields); |
356 | out.set(_o != null ? _o : _dO); |
357 | } else {
|
358 | final S key = unquote(tpp()); |
359 | if (!eq(tpp(), "=")) |
360 | fail("= expected, got " + t() + " after " + quote(key) + " in object " + thingName /*+ " " + sfu(fields)*/);
|
361 | stack.add(this); |
362 | parse(new unstructure_Receiver {
|
363 | void set(O value) {
|
364 | fields.put(key, value); |
365 | if (eq(t(), ",")) consume(); |
366 | } |
367 | }); |
368 | } |
369 | }); |
370 | } else {
|
371 | objRead(o, dO, fields); |
372 | out.set(o != null ? o : dO); |
373 | } |
374 | } |
375 | |
376 | void objRead(O o, DynamicObject dO, Map<S, O> fields) {
|
377 | ifdef unstructure_debug |
378 | print("objRead " + className(o) + " " + className(dO) + " " + struct(fields));
|
379 | endifdef |
380 | if (o != null) |
381 | if (dO != null) {
|
382 | ifdef unstructure_debug |
383 | printStructure("setOptAllDyn", fields);
|
384 | endifdef |
385 | setOptAllDyn(dO, fields); |
386 | } else {
|
387 | setOptAll_pcall(o, fields); |
388 | ifdef unstructure_debug |
389 | print("objRead now: " + struct(o));
|
390 | endifdef |
391 | } |
392 | else for (Map.Entry<S, O> e : fields.entrySet()) |
393 | setDynObjectValue(dO, intern(e.getKey()), e.getValue()); |
394 | |
395 | if (o != null) |
396 | pcallOpt_noArgs(o, "_doneLoading"); |
397 | } |
398 | |
399 | void parseSet(final Set set, final unstructure_Receiver out) {
|
400 | this.parseList(new ArrayList, new unstructure_Receiver {
|
401 | void set(O o) {
|
402 | set.addAll((L) o); |
403 | out.set(set); |
404 | } |
405 | }); |
406 | } |
407 | |
408 | void parseLisp(final unstructure_Receiver out) {
|
409 | ifclass Lisp |
410 | consume("l");
|
411 | consume("(");
|
412 | final new ArrayList list; |
413 | stack.add(r {
|
414 | if (eq(t(), ")")) {
|
415 | consume(")");
|
416 | out.set(Lisp((S) list.get(0), subList(list, 1))); |
417 | } else {
|
418 | stack.add(this); |
419 | parse(new unstructure_Receiver {
|
420 | void set(O o) {
|
421 | list.add(o); |
422 | if (eq(t(), ",")) consume(); |
423 | } |
424 | }); |
425 | } |
426 | }); |
427 | if (false) // skip fail line |
428 | endif |
429 | |
430 | fail("class Lisp not included");
|
431 | } |
432 | |
433 | void parseBitSet(final unstructure_Receiver out) {
|
434 | consume("bitset");
|
435 | consume("{");
|
436 | final new BitSet bs; |
437 | stack.add(r {
|
438 | if (eq(t(), "}")) {
|
439 | consume("}");
|
440 | out.set(bs); |
441 | } else {
|
442 | stack.add(this); |
443 | parse(new unstructure_Receiver {
|
444 | void set(O o) {
|
445 | bs.set((Integer) o); |
446 | if (eq(t(), ",")) consume(); |
447 | } |
448 | }); |
449 | } |
450 | }); |
451 | } |
452 | |
453 | void parseList(final L list, final unstructure_Receiver out) {
|
454 | consume("[");
|
455 | stack.add(r {
|
456 | if (eq(t(), "]")) {
|
457 | consume("]");
|
458 | out.set(list); |
459 | } else {
|
460 | stack.add(this); |
461 | parse(new unstructure_Receiver {
|
462 | void set(O o) {
|
463 | //if (debug) print("List element type: " + getClassName(o));
|
464 | list.add(o); |
465 | if (eq(t(), ",")) consume(); |
466 | } |
467 | }); |
468 | } |
469 | }); |
470 | } |
471 | |
472 | void parseArray(final unstructure_Receiver out) {
|
473 | final S type = tpp(); |
474 | consume("{");
|
475 | final List list = new ArrayList; |
476 | |
477 | stack.add(r {
|
478 | if (eq(t(), "}")) {
|
479 | consume("}");
|
480 | out.set(type.equals("intarray") ? toIntArray(list) : list.toArray());
|
481 | } else {
|
482 | stack.add(this); |
483 | parse(new unstructure_Receiver {
|
484 | void set(O o) {
|
485 | list.add(o); |
486 | if (eq(t(), ",")) consume(); |
487 | } |
488 | }); |
489 | } |
490 | }); |
491 | } |
492 | |
493 | Object parseClass() {
|
494 | consume("class");
|
495 | consume("(");
|
496 | S name = unquote(tpp()); |
497 | consume(")");
|
498 | Class c = allDynamic ? null : classFinder != null ? (Class) callF(classFinder, name) : findClass_fullName(name); |
499 | if (c != null) ret c; |
500 | new DynamicObject dO; |
501 | dO.className = "java.lang.Class"; |
502 | name = dropPrefix("main$", name);
|
503 | dO.fieldValues.put("name", name);
|
504 | ret dO; |
505 | } |
506 | |
507 | Object parseBigInt() {
|
508 | consume("bigint");
|
509 | consume("(");
|
510 | S val = tpp(); |
511 | if (eq(val, "-")) |
512 | val = "-" + tpp(); |
513 | consume(")");
|
514 | ret new BigInteger(val); |
515 | } |
516 | |
517 | Object parseDouble() {
|
518 | consume("d");
|
519 | consume("(");
|
520 | S val = unquote(tpp()); |
521 | consume(")");
|
522 | ret Double.parseDouble(val); |
523 | } |
524 | |
525 | Object parseFloat() {
|
526 | consume("fl");
|
527 | S val; |
528 | if (eq(t(), "(")) {
|
529 | consume("(");
|
530 | val = unquote(tpp()); |
531 | consume(")");
|
532 | } else {
|
533 | val = unquote(tpp()); |
534 | } |
535 | ret Float.parseFloat(val); |
536 | } |
537 | |
538 | void parseHashSet(unstructure_Receiver out) {
|
539 | consume("hashset");
|
540 | parseSet(new HashSet, out); |
541 | } |
542 | |
543 | void parseLinkedHashSet(unstructure_Receiver out) {
|
544 | consume("lhs");
|
545 | parseSet(new LinkedHashSet, out); |
546 | } |
547 | |
548 | void parseTreeSet(unstructure_Receiver out) {
|
549 | consume("treeset");
|
550 | parseSet(new TreeSet, out); |
551 | } |
552 | |
553 | void parseCISet(unstructure_Receiver out) {
|
554 | consume("ciset");
|
555 | parseSet(ciSet(), out); |
556 | } |
557 | |
558 | void parseMap(unstructure_Receiver out) {
|
559 | parseMap(new TreeMap, out); |
560 | } |
561 | |
562 | O parseJava() {
|
563 | S j = unquote(tpp()); |
564 | new Matches m; |
565 | if (jmatch("java.awt.Color[r=*,g=*,b=*]", j, m))
|
566 | ret nuObject("java.awt.Color", parseInt($1), parseInt($2), parseInt($3));
|
567 | else {
|
568 | warn("Unknown Java object: " + j);
|
569 | null; |
570 | } |
571 | } |
572 | |
573 | void parseMap(final Map map, final unstructure_Receiver out) {
|
574 | consume("{");
|
575 | stack.add(new Runnable {
|
576 | bool v; |
577 | O key; |
578 | |
579 | public void run() {
|
580 | if (v) {
|
581 | v = false; |
582 | stack.add(this); |
583 | if (!eq(tpp(), "=")) |
584 | fail("= expected, got " + t() + " in map of size " + l(map));
|
585 | |
586 | parse(new unstructure_Receiver {
|
587 | void set(O value) {
|
588 | map.put(key, value); |
589 | ifdef unstructure_debug |
590 | print("parseMap: Got value " + getClassName(value) + ", next token: " + quote(t()));
|
591 | endifdef |
592 | if (eq(t(), ",")) consume(); |
593 | } |
594 | }); |
595 | } else {
|
596 | if (eq(t(), "}")) {
|
597 | consume("}");
|
598 | out.set(map); |
599 | } else {
|
600 | v = true; |
601 | stack.add(this); |
602 | parse(new unstructure_Receiver {
|
603 | void set(O o) {
|
604 | key = o; |
605 | } |
606 | }); |
607 | } |
608 | } // if v else |
609 | } // run() |
610 | }); |
611 | } |
612 | |
613 | /*void parseSub(unstructure_Receiver out) {
|
614 | int n = l(stack); |
615 | parse(out); |
616 | while (l(stack) > n) |
617 | stack |
618 | }*/ |
619 | |
620 | void consume() { curT = tok.next(); ++i; }
|
621 | |
622 | void consume(S s) {
|
623 | if (!eq(t(), s)) {
|
624 | /*S prevToken = i-1 >= 0 ? tok.get(i-1) : ""; |
625 | S nextTokens = join(tok.subList(i, Math.min(i+2, tok.size()))); |
626 | fail(quote(s) + " expected: " + prevToken + " " + nextTokens + " (" + i + "/" + tok.size() + ")");*/
|
627 | fail(quote(s) + " expected, got " + quote(t())); |
628 | } |
629 | consume(); |
630 | } |
631 | |
632 | void parse_x(unstructure_Receiver out) {
|
633 | consume(); // get first token |
634 | parse(out); |
635 | while (nempty(stack)) |
636 | popLast(stack).run(); |
637 | } |
638 | } |
639 | |
640 | Bool b = DynamicObject_loading!; |
641 | DynamicObject_loading.set(true); |
642 | try {
|
643 | final new Var v; |
644 | new X x; |
645 | x.parse_x(new unstructure_Receiver {
|
646 | void set(O o) { v.set(o); }
|
647 | }); |
648 | unstructure_tokrefs = x.tokrefs.size(); |
649 | ret v.get(); |
650 | } finally {
|
651 | DynamicObject_loading.set(b); |
652 | } |
653 | } |
654 | |
655 | static boolean unstructure_debug; |
Began life as a copy of #1020754
download show line numbers debug dex old transpilations
Travelled to 6 computer(s): bhatertpkbcr, mqqgnosmbjvj, pyentgdyhuwx, pzhvpgtvlbxg, tvejysmllsmz, vouqrxazstgt
No comments. add comment
| Snippet ID: | #1024869 |
| Snippet name: | unstructure (v13, with custom serialization) |
| Eternal ID of this version: | #1024869/8 |
| Text MD5: | a8b25b5081794630c6a6df8550b9d820 |
| Transpilation MD5: | 508b5140cc5f308027e5f73cad5897b1 |
| Author: | stefan |
| Category: | javax |
| Type: | JavaX fragment (include) |
| Public (visible to everyone): | Yes |
| Archived (hidden from active list): | No |
| Created/modified: | 2019-08-28 12:35:42 |
| Source code size: | 19117 bytes / 655 lines |
| Pitched / IR pitched: | No / No |
| Views / Downloads: | 236 / 363 |
| Version history: | 7 change(s) |
| Referenced in: | [show references] |