unstructure (v8, virtual stack, dev.) [1005658]

static Object unstructure(String text) {
  ret unstructure(text, false);
}

static Object unstructure(String text, final boolean allDynamic) {
  ret unstructure(text, allDynamic, null);
}

static int structure_internStringsLongerThan = 50;

// classFinder: func(name) -> class (optional)
static Object unstructure(String text, final boolean allDynamic,
  final O classFinder) {
  if (text == null) ret null;
  final L<S> tok = javaTokC(text);
  final boolean debug = unstructure_debug;

  class X {
    int i = 0;
    new HashMap<Integer, O> refs;
    new HashMap<Integer, O> tokrefs;
    new HashSet<S> concepts;
    new HashMap<S, Class> classesMap;
    new L<Runnable> stack;

    Object parse() {
      String t = tok.get(i);
      
      int refID = 0;
      if (structure_isMarker(t, 0, l(t))) {
        refID = parseInt(t.substring(1));
        i++;
      }
      
      // if (debug) print("parse: " + quote(t));
      
      int tokIndex = i;  
      O o = parse_inner(refID, tokIndex);
      if (refID != 0)
        refs.put(refID, o);
      if (o != null)
        tokrefs.put(tokIndex, o);
      ret o;
    }
    
    O parse_inner(int refID, int tokIndex) {
      String t = tok.get(i);
      
      // if (debug) print("parse_inner: " + quote(t));
      
      Class c = classesMap.get(t);
      if (c == null) {
        if (t.startsWith("\"")) {
          String s = internIfLongerThan(unquote(tok.get(i)), structure_internStringsLongerThan);
          i++;
          return s;
        }
        
        if (t.startsWith("'"))
          ret unquoteCharacter(tok.get(i++));
        if (t.equals("bigint"))
          return parseBigInt();
        if (t.equals("d"))
          return parseDouble();
        if (t.equals("fl"))
          return parseFloat();
        if (t.equals("false") || t.equals("f")) {
          i++; return false;
        }
        if (t.equals("true") || t.equals("t")) {
          i++; return true;
        }
        if (t.equals("-")) {
          t = tok.get(i+1);
          i += 2;
          ret isLongConstant(t) ? (O) (-parseLong(t)) : (O) (-parseInt(t));
        }
        if (isInteger(t) || isLongConstant(t)) {
          i++;
          //if (debug) print("isLongConstant " + quote(t) + " => " + isLongConstant(t));
          if (isLongConstant(t)) ret parseLong(t);
          long l = parseLong(t);
          bool isInt = l == (int) l;
          if (debug)
            print("l=" + l + ", isInt: " + isInt);
          ret isInt ? (O) new Integer((int) l) : (O) new Long(l);
        }
        
        if (t.equals("File")) {
          File f = new File(unquote(tok.get(i+1)));
          i += 2;
          ret f;
        }
        
        if (t.startsWith("r") && isInteger(t.substring(1))) {
          i++;
          int ref = Integer.parseInt(t.substring(1));
          O o = refs.get(ref);
          if (o == null)
            print("Warning: unsatisfied back reference " + ref);
          ret o;
        }
      
        if (t.startsWith("t") && isInteger(t.substring(1))) {
          i++;
          int ref = Integer.parseInt(t.substring(1));
          O o = tokrefs.get(ref);
          if (o == null)
            print("Warning: unsatisfied token reference " + ref);
          ret o;
        }
        
        if (t.equals("hashset"))
          return parseHashSet();
        if (t.equals("treeset"))
          return parseTreeSet();
        if (eqOneOf(t, "hashmap", "hm"))
          return parseHashMap();
        if (t.equals("{"))
          return parseMap();
        if (t.equals("["))
          return parseList();
        if (t.equals("bitset"))
          return parseBitSet();
        if (t.equals("array") || t.equals("intarray"))
          return parseArray();
        if (t.equals("ba")) {
          S hex = unquote(tok.get(i+1));
          i += 2;
          ret hexToBytes(hex);
        }
        if (t.equals("boolarray")) {
          int n = parseInt(tok.get(i+1));
          S hex = unquote(tok.get(i+2));
          i += 6;
          ret boolArrayFromBytes(hexToBytes(hex), n);
        }
        if (t.equals("class"))
          return parseClass();
        if (t.equals("l"))
          return parseLisp();
        if (t.equals("null")) {
          i++; return null;
        }
        
        /* in dev.
        if (!allDynamic && t.equals("run")) {
          S snippetID = unquote(t.get(i+1));
          i += 2;
          run(
        }
        */
        
        if (eq(t, "c")) {
          consume("c");
          t = tok.get(i);
          assertTrue(isJavaIdentifier(t));
          concepts.add(t);
        }
      }
      
      if (c == null && !isJavaIdentifier(t))
        throw new RuntimeException("Unknown token " + (i+1) + ": " + t);

      // any other class name
      if (c == null) {
        // First, find class
        if (allDynamic) c = null;
        else if (classFinder != null)
          c = (Class) callF(classFinder, t);
        else
          c = findClass(t);
        if (c != null)
          classesMap.put(t, c);
      }
          
      // Check if it has an outer reference
      i++;
      bool hasOuter = eq(get(tok, i), "(") && eq(get(tok, i+1), "this$1");
      
      DynamicObject dO = null;
      O o = null;
      if (c != null)
        o = hasOuter ? nuStubInnerObject(c) : nuEmptyObject(c);
      else {
        if (concepts.contains(t) && (c = findClass("Concept")) != null)
          o = dO = (DynamicObject) nuEmptyObject(c);
        else
          dO = new DynamicObject;
        dO.className = t;
        if (debug) print("Made dynamic object " + t + " " + shortClassName(dO));
      }
      
      // Save in references list early because contents of object
      // might link back to main object
      
      if (refID != 0)
        refs.put(refID, o != null ? o : dO);
      tokrefs.put(tokIndex, o != null ? o : dO);
      
      // NOW parse the fields!
      
      final new HashMap<S, O> fields;
      final O _o = o;
      final DynamicObject _dO = dO;
      if (i < tok.size() && tok.get(i).equals("(")) {
        consume("(");
        stack.add(r {
          if (tok.get(i).equals(")")) {
            consume(")");
            objRead(_o, _dO, fields);
          } else {
            S key = unquote(tok.get(i++));
            consume("=");
            stack.add(this);
            O value = parse();
            fields.put(key, value);
            if (tok.get(i).equals(",")) i++;
          }
        });
      } else
        objRead(o, dO, fields);

      return o != null ? o : dO;
    }
    
    void objRead(O o, DynamicObject dO, Map<S, O> fields) {
      if (o != null)
        if (dO != null) {
          if (debug)
            printStructure("setOptAllDyn", fields);
          setOptAllDyn(dO, fields);
        } else
          setOptAll(o, fields);
      else for (S field : keys(fields))
        dO.fieldValues.put(field.intern(), fields.get(field));

      if (o != null)
        pcallOpt_noArgs(o, "_doneLoading");
    }
    
    Object parseSet(Set set) {
      set.addAll((L) parseList());
      return set;
    }
    
    Object parseLisp() {
      consume("l");
      consume("(");
      new ArrayList list;
      stack.add(r {
        if (tok.get(i).equals(")"))
          consume(")");
        else {
          stack.add(this);
          list.add(parse());
          if (tok.get(i).equals(",")) i++;
        }
      });
      return newObject("main$Lisp", (S) list.get(0), subList(list, 1));
    }
    
    Object parseBitSet() {
      consume("bitset");
      consume("{");
      new BitSet bs;
      stack.add(r {
        if (tok.get(i).equals("}"))
          consume("}");
        else {
          stack.add(this);
          bs.set((Integer) parse());
          if (tok.get(i).equals(",")) i++;
        }
      });
      return bs;
    }
    
    Object parseList() {
      consume("[");
      new ArrayList list;
      stack.add(r {
        if (tok.get(i).equals("]"))
          consume("]");
        else {
          stack.add(this);
          O o = parse();
          //if (debug) print("List element type: " + getClassName(o));
          list.add(o);
          if (tok.get(i).equals(",")) i++;
        }
      });
      return list;
    }
    
    Object parseArray() {
      S type = tok.get(i);
      i++;
      consume("{");
      final List list = new ArrayList;
      
      // TODO!!
      
      stack.add(r {
        if (tok.get(i).equals("}"))
          consume("}");
        else {
          stack.add(this);
          list.add(parse());
          if (tok.get(i).equals(",")) i++;
        }
      });
      if (type.equals("intarray"))
        return toIntArray(list);
      return list.toArray();
    }
    
    Object parseClass() {
      consume("class");
      consume("(");
      S name = tok.get(i);
      i++;
      consume(")");
      Class c = allDynamic ? null : findClass(name);
      if (c != null) ret c;
      new DynamicObject dO;
      dO.className = "java.lang.Class";
      dO.fieldValues.put("name", name);
      ret dO;
    }
    
    Object parseBigInt() {
      consume("bigint");
      consume("(");
      S val = tok.get(i);
      i++;
      if (eq(val, "-")) {
        val = "-" + tok.get(i);
        i++;
      }
      consume(")");
      ret new BigInteger(val);
    }
    
    Object parseDouble() {
      consume("d");
      consume("(");
      S val = unquote(tok.get(i));
      i++;
      consume(")");
      ret Double.parseDouble(val);
    }
    
    Object parseFloat() {
      consume("fl");
      S val;
      if (eq(tok.get(i), "(")) {
        consume("(");
        val = unquote(tok.get(i));
        i++;
        consume(")");
      } else {
        val = unquote(tok.get(i));
        i++;
      }
      ret Float.parseFloat(val);
    }
    
    Object parseHashMap() {
      i++;
      return parseMap(new HashMap);
    }
    
    Object parseHashSet() {
      consume("hashset");
      return parseSet(new HashSet);
    }
    
    Object parseTreeSet() {
      consume("treeset");
      return parseSet(new TreeSet);
    }
    
    Object parseMap() {
      return parseMap(new TreeMap);
    }
    
    Object parseMap(Map map) {
      consume("{");
      while (!tok.get(i).equals("}")) {
        O key = parse();
        consume("=");
        Object value = parse();
        map.put(key, value);
        if (tok.get(i).equals(",")) i++;
      }
      consume("}");
      return map;
    }
    
    void consume(String s) {
      if (!tok.get(i).equals(s)) {
        S prevToken = i-1 >= 0 ? tok.get(i-1) : "";
        S nextTokens = join(tok.subList(i, Math.min(i+2, tok.size())));
        fail(quote(s) + " expected: " + prevToken + " " + nextTokens + " (" + i + "/" + tok.size() + ")");
      }
      i++;
    }
    
    O parse_x() {
      O o = parse();
      while (nempty(stack))
        popLast(stack).run();
      ret o;
    }
  }
  
  ret new X().parse_x();
}

static boolean unstructure_debug;

Travelled to 14 computer(s): aoiabmzegqzx, bhatertpkbcr, cbybwowwnfue, cfunsshuasjs, ddnzoavkxhuk, gwrvuhgaqvyk, ishqpsrjomds, lpdgvwnxivlt, mqqgnosmbjvj, pyentgdyhuwx, pzhvpgtvlbxg, tslmcundralx, tvejysmllsmz, vouqrxazstgt

Snippet ID:	#1005658
Snippet name:	unstructure (v8, virtual stack, dev.)
Eternal ID of this version:	#1005658/1
Text MD5:	c838217897ac97113ac26a743137c1e8
Author:	stefan
Category:	javax
Type:	JavaX fragment (include)
Public (visible to everyone):	Yes
Archived (hidden from active list):	No
Created/modified:	2016-12-13 15:56:28
Source code size:	11328 bytes / 421 lines
Pitched / IR pitched:	No / No
Views / Downloads:	776 / 779
Referenced in:	[show references]

< > BotCompany Repo | #1005658 // unstructure (v8, virtual stack, dev.)

JavaX fragment (include)

Author comment

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
11	// classFinder: func(name) -> class (optional)
12	static Object unstructure(String text, final boolean allDynamic,
13	final O classFinder) {
14	if (text == null) ret null;
15	final L<S> tok = javaTokC(text);
16	final boolean debug = unstructure_debug;
17
18	class X {
19	int i = 0;
20	new HashMap<Integer, O> refs;
21	new HashMap<Integer, O> tokrefs;
22	new HashSet<S> concepts;
23	new HashMap<S, Class> classesMap;
24	new L<Runnable> stack;
25
26	Object parse() {
27	String t = tok.get(i);
28
29	int refID = 0;
30	if (structure_isMarker(t, 0, l(t))) {
31	refID = parseInt(t.substring(1));
32	i++;
33	}
34
35	// if (debug) print("parse: " + quote(t));
36
37	int tokIndex = i;
38	O o = parse_inner(refID, tokIndex);
39	if (refID != 0)
40	refs.put(refID, o);
41	if (o != null)
42	tokrefs.put(tokIndex, o);
43	ret o;
44	}
45
46	O parse_inner(int refID, int tokIndex) {
47	String t = tok.get(i);
48
49	// if (debug) print("parse_inner: " + quote(t));
50
51	Class c = classesMap.get(t);
52	if (c == null) {
53	if (t.startsWith("\"")) {
54	String s = internIfLongerThan(unquote(tok.get(i)), structure_internStringsLongerThan);
55	i++;
56	return s;
57	}
58
59	if (t.startsWith("'"))
60	ret unquoteCharacter(tok.get(i++));
61	if (t.equals("bigint"))
62	return parseBigInt();
63	if (t.equals("d"))
64	return parseDouble();
65	if (t.equals("fl"))
66	return parseFloat();
67	if (t.equals("false") \|\| t.equals("f")) {
68	i++; return false;
69	}
70	if (t.equals("true") \|\| t.equals("t")) {
71	i++; return true;
72	}
73	if (t.equals("-")) {
74	t = tok.get(i+1);
75	i += 2;
76	ret isLongConstant(t) ? (O) (-parseLong(t)) : (O) (-parseInt(t));
77	}
78	if (isInteger(t) \|\| isLongConstant(t)) {
79	i++;
80	//if (debug) print("isLongConstant " + quote(t) + " => " + isLongConstant(t));
81	if (isLongConstant(t)) ret parseLong(t);
82	long l = parseLong(t);
83	bool isInt = l == (int) l;
84	if (debug)
85	print("l=" + l + ", isInt: " + isInt);
86	ret isInt ? (O) new Integer((int) l) : (O) new Long(l);
87	}
88
89	if (t.equals("File")) {
90	File f = new File(unquote(tok.get(i+1)));
91	i += 2;
92	ret f;
93	}
94
95	if (t.startsWith("r") && isInteger(t.substring(1))) {
96	i++;
97	int ref = Integer.parseInt(t.substring(1));
98	O o = refs.get(ref);
99	if (o == null)
100	print("Warning: unsatisfied back reference " + ref);
101	ret o;
102	}
103
104	if (t.startsWith("t") && isInteger(t.substring(1))) {
105	i++;
106	int ref = Integer.parseInt(t.substring(1));
107	O o = tokrefs.get(ref);
108	if (o == null)
109	print("Warning: unsatisfied token reference " + ref);
110	ret o;
111	}
112
113	if (t.equals("hashset"))
114	return parseHashSet();
115	if (t.equals("treeset"))
116	return parseTreeSet();
117	if (eqOneOf(t, "hashmap", "hm"))
118	return parseHashMap();
119	if (t.equals("{"))
120	return parseMap();
121	if (t.equals("["))
122	return parseList();
123	if (t.equals("bitset"))
124	return parseBitSet();
125	if (t.equals("array") \|\| t.equals("intarray"))
126	return parseArray();
127	if (t.equals("ba")) {
128	S hex = unquote(tok.get(i+1));
129	i += 2;
130	ret hexToBytes(hex);
131	}
132	if (t.equals("boolarray")) {
133	int n = parseInt(tok.get(i+1));
134	S hex = unquote(tok.get(i+2));
135	i += 6;
136	ret boolArrayFromBytes(hexToBytes(hex), n);
137	}
138	if (t.equals("class"))
139	return parseClass();
140	if (t.equals("l"))
141	return parseLisp();
142	if (t.equals("null")) {
143	i++; return null;
144	}
145
146	/* in dev.
147	if (!allDynamic && t.equals("run")) {
148	S snippetID = unquote(t.get(i+1));
149	i += 2;
150	run(
151	}
152	*/
153
154	if (eq(t, "c")) {
155	consume("c");
156	t = tok.get(i);
157	assertTrue(isJavaIdentifier(t));
158	concepts.add(t);
159	}
160	}
161
162	if (c == null && !isJavaIdentifier(t))
163	throw new RuntimeException("Unknown token " + (i+1) + ": " + t);
164
165	// any other class name
166	if (c == null) {
167	// First, find class
168	if (allDynamic) c = null;
169	else if (classFinder != null)
170	c = (Class) callF(classFinder, t);
171	else
172	c = findClass(t);
173	if (c != null)
174	classesMap.put(t, c);
175	}
176
177	// Check if it has an outer reference
178	i++;
179	bool hasOuter = eq(get(tok, i), "(") && eq(get(tok, i+1), "this$1");
180
181	DynamicObject dO = null;
182	O o = null;
183	if (c != null)
184	o = hasOuter ? nuStubInnerObject(c) : nuEmptyObject(c);
185	else {
186	if (concepts.contains(t) && (c = findClass("Concept")) != null)
187	o = dO = (DynamicObject) nuEmptyObject(c);
188	else
189	dO = new DynamicObject;
190	dO.className = t;
191	if (debug) print("Made dynamic object " + t + " " + shortClassName(dO));
192	}
193
194	// Save in references list early because contents of object
195	// might link back to main object
196
197	if (refID != 0)
198	refs.put(refID, o != null ? o : dO);
199	tokrefs.put(tokIndex, o != null ? o : dO);
200
201	// NOW parse the fields!
202
203	final new HashMap<S, O> fields;
204	final O _o = o;
205	final DynamicObject _dO = dO;
206	if (i < tok.size() && tok.get(i).equals("(")) {
207	consume("(");
208	stack.add(r {
209	if (tok.get(i).equals(")")) {
210	consume(")");
211	objRead(_o, _dO, fields);
212	} else {
213	S key = unquote(tok.get(i++));
214	consume("=");
215	stack.add(this);
216	O value = parse();
217	fields.put(key, value);
218	if (tok.get(i).equals(",")) i++;
219	}
220	});
221	} else
222	objRead(o, dO, fields);
223
224	return o != null ? o : dO;
225	}
226
227	void objRead(O o, DynamicObject dO, Map<S, O> fields) {
228	if (o != null)
229	if (dO != null) {
230	if (debug)
231	printStructure("setOptAllDyn", fields);
232	setOptAllDyn(dO, fields);
233	} else
234	setOptAll(o, fields);
235	else for (S field : keys(fields))
236	dO.fieldValues.put(field.intern(), fields.get(field));
237
238	if (o != null)
239	pcallOpt_noArgs(o, "_doneLoading");
240	}
241
242	Object parseSet(Set set) {
243	set.addAll((L) parseList());
244	return set;
245	}
246
247	Object parseLisp() {
248	consume("l");
249	consume("(");
250	new ArrayList list;
251	stack.add(r {
252	if (tok.get(i).equals(")"))
253	consume(")");
254	else {
255	stack.add(this);
256	list.add(parse());
257	if (tok.get(i).equals(",")) i++;
258	}
259	});
260	return newObject("main$Lisp", (S) list.get(0), subList(list, 1));
261	}
262
263	Object parseBitSet() {
264	consume("bitset");
265	consume("{");
266	new BitSet bs;
267	stack.add(r {
268	if (tok.get(i).equals("}"))
269	consume("}");
270	else {
271	stack.add(this);
272	bs.set((Integer) parse());
273	if (tok.get(i).equals(",")) i++;
274	}
275	});
276	return bs;
277	}
278
279	Object parseList() {
280	consume("[");
281	new ArrayList list;
282	stack.add(r {
283	if (tok.get(i).equals("]"))
284	consume("]");
285	else {
286	stack.add(this);
287	O o = parse();
288	//if (debug) print("List element type: " + getClassName(o));
289	list.add(o);
290	if (tok.get(i).equals(",")) i++;
291	}
292	});
293	return list;
294	}
295
296	Object parseArray() {
297	S type = tok.get(i);
298	i++;
299	consume("{");
300	final List list = new ArrayList;
301
302	// TODO!!
303
304	stack.add(r {
305	if (tok.get(i).equals("}"))
306	consume("}");
307	else {
308	stack.add(this);
309	list.add(parse());
310	if (tok.get(i).equals(",")) i++;
311	}
312	});
313	if (type.equals("intarray"))
314	return toIntArray(list);
315	return list.toArray();
316	}
317
318	Object parseClass() {
319	consume("class");
320	consume("(");
321	S name = tok.get(i);
322	i++;
323	consume(")");
324	Class c = allDynamic ? null : findClass(name);
325	if (c != null) ret c;
326	new DynamicObject dO;
327	dO.className = "java.lang.Class";
328	dO.fieldValues.put("name", name);
329	ret dO;
330	}
331
332	Object parseBigInt() {
333	consume("bigint");
334	consume("(");
335	S val = tok.get(i);
336	i++;
337	if (eq(val, "-")) {
338	val = "-" + tok.get(i);
339	i++;
340	}
341	consume(")");
342	ret new BigInteger(val);
343	}
344
345	Object parseDouble() {
346	consume("d");
347	consume("(");
348	S val = unquote(tok.get(i));
349	i++;
350	consume(")");
351	ret Double.parseDouble(val);
352	}
353
354	Object parseFloat() {
355	consume("fl");
356	S val;
357	if (eq(tok.get(i), "(")) {
358	consume("(");
359	val = unquote(tok.get(i));
360	i++;
361	consume(")");
362	} else {
363	val = unquote(tok.get(i));
364	i++;
365	}
366	ret Float.parseFloat(val);
367	}
368
369	Object parseHashMap() {
370	i++;
371	return parseMap(new HashMap);
372	}
373
374	Object parseHashSet() {
375	consume("hashset");
376	return parseSet(new HashSet);
377	}
378
379	Object parseTreeSet() {
380	consume("treeset");
381	return parseSet(new TreeSet);
382	}
383
384	Object parseMap() {
385	return parseMap(new TreeMap);
386	}
387
388	Object parseMap(Map map) {
389	consume("{");
390	while (!tok.get(i).equals("}")) {
391	O key = parse();
392	consume("=");
393	Object value = parse();
394	map.put(key, value);
395	if (tok.get(i).equals(",")) i++;
396	}
397	consume("}");
398	return map;
399	}
400
401	void consume(String s) {
402	if (!tok.get(i).equals(s)) {
403	S prevToken = i-1 >= 0 ? tok.get(i-1) : "";
404	S nextTokens = join(tok.subList(i, Math.min(i+2, tok.size())));
405	fail(quote(s) + " expected: " + prevToken + " " + nextTokens + " (" + i + "/" + tok.size() + ")");
406	}
407	i++;
408	}
409
410	O parse_x() {
411	O o = parse();
412	while (nempty(stack))
413	popLast(stack).run();
414	ret o;
415	}
416	}
417
418	ret new X().parse_x();
419	}
420
421	static boolean unstructure_debug;