summaryrefslogtreecommitdiffstats
path: root/Parser/parser.c
blob: 686161b2bbfebfa453fbe51e811b013c4fa85455 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418

/* Parser implementation */

/* For a description, see the comments at end of this file */

/* XXX To do: error recovery */

#include "Python.h"
#include "pgenheaders.h"
#include "token.h"
#include "grammar.h"
#include "node.h"
#include "parser.h"
#include "errcode.h"


#ifdef Py_DEBUG
extern int Py_DebugFlag;
#define D(x) if (!Py_DebugFlag); else x
#else
#define D(x)
#endif


/* STACK DATA TYPE */

static void s_reset(stack *);

static void
s_reset(stack *s)
{
	s->s_top = &s->s_base[MAXSTACK];
}

#define s_empty(s) ((s)->s_top == &(s)->s_base[MAXSTACK])

static int
s_push(register stack *s, dfa *d, node *parent)
{
	register stackentry *top;
	if (s->s_top == s->s_base) {
		fprintf(stderr, "s_push: parser stack overflow\n");
		return E_NOMEM;
	}
	top = --s->s_top;
	top->s_dfa = d;
	top->s_parent = parent;
	top->s_state = 0;
	return 0;
}

#ifdef Py_DEBUG

static void
s_pop(register stack *s)
{
	if (s_empty(s))
		Py_FatalError("s_pop: parser stack underflow -- FATAL");
	s->s_top++;
}

#else /* !Py_DEBUG */

#define s_pop(s) (s)->s_top++

#endif


/* PARSER CREATION */

parser_state *
PyParser_New(grammar *g, int start)
{
	parser_state *ps;
	
	if (!g->g_accel)
		PyGrammar_AddAccelerators(g);
	ps = PyMem_NEW(parser_state, 1);
	if (ps == NULL)
		return NULL;
	ps->p_grammar = g;
#if 0 /* future keyword */
	ps->p_generators = 0;
#endif
	ps->p_tree = PyNode_New(start);
	if (ps->p_tree == NULL) {
		PyMem_DEL(ps);
		return NULL;
	}
	s_reset(&ps->p_stack);
	(void) s_push(&ps->p_stack, PyGrammar_FindDFA(g, start), ps->p_tree);
	return ps;
}

void
PyParser_Delete(parser_state *ps)
{
	/* NB If you want to save the parse tree,
	   you must set p_tree to NULL before calling delparser! */
	PyNode_Free(ps->p_tree);
	PyMem_DEL(ps);
}


/* PARSER STACK OPERATIONS */

static int
shift(register stack *s, int type, char *str, int newstate, int lineno)
{
	int err;
	assert(!s_empty(s));
	err = PyNode_AddChild(s->s_top->s_parent, type, str, lineno);
	if (err)
		return err;
	s->s_top->s_state = newstate;
	return 0;
}

static int
push(register stack *s, int type, dfa *d, int newstate, int lineno)
{
	int err;
	register node *n;
	n = s->s_top->s_parent;
	assert(!s_empty(s));
	err = PyNode_AddChild(n, type, (char *)NULL, lineno);
	if (err)
		return err;
	s->s_top->s_state = newstate;
	return s_push(s, d, CHILD(n, NCH(n)-1));
}


/* PARSER PROPER */

static int
classify(parser_state *ps, int type, char *str)
{
	grammar *g = ps->p_grammar;
	register int n = g->g_ll.ll_nlabels;
	
	if (type == NAME) {
		register char *s = str;
		register label *l = g->g_ll.ll_label;
		register int i;
		for (i = n; i > 0; i--, l++) {
			if (l->lb_type == NAME && l->lb_str != NULL &&
					l->lb_str[0] == s[0] &&
					strcmp(l->lb_str, s) == 0) {
#if 0 /* future keyword */
				if (!ps->p_generators &&
				    s[0] == 'y' &&
				    strcmp(s, "yield") == 0)
					break; /* not a keyword */
#endif
				D(printf("It's a keyword\n"));
				return n - i;
			}
		}
	}
	
	{
		register label *l = g->g_ll.ll_label;
		register int i;
		for (i = n; i > 0; i--, l++) {
			if (l->lb_type == type && l->lb_str == NULL) {
				D(printf("It's a token we know\n"));
				return n - i;
			}
		}
	}
	
	D(printf("Illegal token\n"));
	return -1;
}

#if 0 /* future keyword */
static void
future_hack(parser_state *ps)
{
	node *n = ps->p_stack.s_top->s_parent;
	node *ch;
	int i;

	if (strcmp(STR(CHILD(n, 0)), "from") != 0)
		return;
	ch = CHILD(n, 1);
	if (strcmp(STR(CHILD(ch, 0)), "__future__") != 0)
		return;
	for (i = 3; i < NCH(n); i += 2) {
		ch = CHILD(n, i);
		if (NCH(ch) >= 1 && TYPE(CHILD(ch, 0)) == NAME &&
		    strcmp(STR(CHILD(ch, 0)), "generators") == 0) {
			ps->p_generators = 1;
			break;
		}
	}
}
#endif /* future keyword */

int
PyParser_AddToken(register parser_state *ps, register int type, char *str,
	          int lineno, int *expected_ret)
{
	register int ilabel;
	int err;
	
	D(printf("Token %s/'%s' ... ", _PyParser_TokenNames[type], str));
	
	/* Find out which label this token is */
	ilabel = classify(ps, type, str);
	if (ilabel < 0)
		return E_SYNTAX;
	
	/* Loop until the token is shifted or an error occurred */
	for (;;) {
		/* Fetch the current dfa and state */
		register dfa *d = ps->p_stack.s_top->s_dfa;
		register state *s = &d->d_state[ps->p_stack.s_top->s_state];
		
		D(printf(" DFA '%s', state %d:",
			d->d_name, ps->p_stack.s_top->s_state));
		
		/* Check accelerator */
		if (s->s_lower <= ilabel && ilabel < s->s_upper) {
			register int x = s->s_accel[ilabel - s->s_lower];
			if (x != -1) {
				if (x & (1<<7)) {
					/* Push non-terminal */
					int nt = (x >> 8) + NT_OFFSET;
					int arrow = x & ((1<<7)-1);
					dfa *d1 = PyGrammar_FindDFA(
						ps->p_grammar, nt);
					if ((err = push(&ps->p_stack, nt, d1,
						arrow, lineno)) > 0) {
						D(printf(" MemError: push\n"));
						return err;
					}
					D(printf(" Push ...\n"));
					continue;
				}
				
				/* Shift the token */
				if ((err = shift(&ps->p_stack, type, str,
						x, lineno)) > 0) {
					D(printf(" MemError: shift.\n"));
					return err;
				}
				D(printf(" Shift.\n"));
				/* Pop while we are in an accept-only state */
				while (s = &d->d_state
						[ps->p_stack.s_top->s_state],
					s->s_accept && s->s_narcs == 1) {
					D(printf("  DFA '%s', state %d: "
						 "Direct pop.\n",
						 d->d_name,
						 ps->p_stack.s_top->s_state));
#if 0 /* future keyword */
					if (d->d_name[0] == 'i' &&
					    strcmp(d->d_name,
						   "import_stmt") == 0)
						future_hack(ps);
#endif
					s_pop(&ps->p_stack);
					if (s_empty(&ps->p_stack)) {
						D(printf("  ACCEPT.\n"));
						return E_DONE;
					}
					d = ps->p_stack.s_top->s_dfa;
				}
				return E_OK;
			}
		}
		
		if (s->s_accept) {
#if 0 /* future keyword */
			if (d->d_name[0] == 'i' &&
			    strcmp(d->d_name, "import_stmt") == 0)
				future_hack(ps);
#endif
			/* Pop this dfa and try again */
			s_pop(&ps->p_stack);
			D(printf(" Pop ...\n"));
			if (s_empty(&ps->p_stack)) {
				D(printf(" Error: bottom of stack.\n"));
				return E_SYNTAX;
			}
			continue;
		}
		
		/* Stuck, report syntax error */
		D(printf(" Error.\n"));
		if (expected_ret) {
			if (s->s_lower == s->s_upper - 1) {
				/* Only one possible expected token */
				*expected_ret = ps->p_grammar->
				    g_ll.ll_label[s->s_lower].lb_type;
			}
			else 
		        	*expected_ret = -1;
		}
		return E_SYNTAX;
	}
}


#ifdef Py_DEBUG

/* DEBUG OUTPUT */

void
dumptree(grammar *g, node *n)
{
	int i;
	
	if (n == NULL)
		printf("NIL");
	else {
		label l;
		l.lb_type = TYPE(n);
		l.lb_str = STR(n);
		printf("%s", PyGrammar_LabelRepr(&l));
		if (ISNONTERMINAL(TYPE(n))) {
			printf("(");
			for (i = 0; i < NCH(n); i++) {
				if (i > 0)
					printf(",");
				dumptree(g, CHILD(n, i));
			}
			printf(")");
		}
	}
}

void
showtree(grammar *g, node *n)
{
	int i;
	
	if (n == NULL)
		return;
	if (ISNONTERMINAL(TYPE(n))) {
		for (i = 0; i < NCH(n); i++)
			showtree(g, CHILD(n, i));
	}
	else if (ISTERMINAL(TYPE(n))) {
		printf("%s", _PyParser_TokenNames[TYPE(n)]);
		if (TYPE(n) == NUMBER || TYPE(n) == NAME)
			printf("(%s)", STR(n));
		printf(" ");
	}
	else
		printf("? ");
}

void
printtree(parser_state *ps)
{
	if (Py_DebugFlag) {
		printf("Parse tree:\n");
		dumptree(ps->p_grammar, ps->p_tree);
		printf("\n");
		printf("Tokens:\n");
		showtree(ps->p_grammar, ps->p_tree);
		printf("\n");
	}
	printf("Listing:\n");
	PyNode_ListTree(ps->p_tree);
	printf("\n");
}

#endif /* Py_DEBUG */

/*

Description
-----------

The parser's interface is different than usual: the function addtoken()
must be called for each token in the input.  This makes it possible to
turn it into an incremental parsing system later.  The parsing system
constructs a parse tree as it goes.

A parsing rule is represented as a Deterministic Finite-state Automaton
(DFA).  A node in a DFA represents a state of the parser; an arc represents
a transition.  Transitions are either labeled with terminal symbols or
with non-terminals.  When the parser decides to follow an arc labeled
with a non-terminal, it is invoked recursively with the DFA representing
the parsing rule for that as its initial state; when that DFA accepts,
the parser that invoked it continues.  The parse tree constructed by the
recursively called parser is inserted as a child in the current parse tree.

The DFA's can be constructed automatically from a more conventional
language description.  An extended LL(1) grammar (ELL(1)) is suitable.
Certain restrictions make the parser's life easier: rules that can produce
the empty string should be outlawed (there are other ways to put loops
or optional parts in the language).  To avoid the need to construct
FIRST sets, we can require that all but the last alternative of a rule
(really: arc going out of a DFA's state) must begin with a terminal
symbol.

As an example, consider this grammar:

expr:	term (OP term)*
term:	CONSTANT | '(' expr ')'

The DFA corresponding to the rule for expr is:

------->.---term-->.------->
	^          |
	|          |
	\----OP----/

The parse tree generated for the input a+b is:

(expr: (term: (NAME: a)), (OP: +), (term: (NAME: b)))

*/