Context Navigation

← Previous Change
Next Change →

parse.c

Timestamp:

05/21/06 18:16:04 (3 years ago)

Author:

krobillard

Message:

Orca - Ported parse from Thune.

Files:

: 1 modified

trunk/orca/parse.c (modified) (27 diffs)

Legend:

: Unmodified
: Added
: Removed

trunk/orca/parse.c

r154	r156
21	21	#include "os.h"
22	22	#include "ovalue.h"
23		~~#include "charset.h"~~
24	23	#include "orca_atoms.h"
	24	#include "internal.h"
	25
	26
	27	#define PARSE_EX
	28	#define orChar orInt
	29	#define UChar char
	30	#define OR_ITER_BLOCK(ita,itb,blk,scell) \
	31	ita = blk->values + scell->series.it; \
	32	itb = blk->values + blk->used;
	33
25	34
26	35
…	…
29	38	extern int orFindString( const OString* strA, int iA,
30	39	const OString* strB, int iB );
	40
	41
	42	typedef struct
	43	{
	44	uint8_t type;
	45	uint8_t flags;
	46	uint8_t datatype;
	47	uint8_t _pad;
	48	OIndex n; /* Series number */
	49	OIndex it; /* Element Iterator */
	50	OIndex end; /* Slice end */
	51	}
	52	UCellSlice;
	53
	54
	55	typedef struct
	56	{
	57	//UThread* thread;
	58	OValue* rules;
	59	UCellSlice input;
	60	OString* str;
	61	int error;
	62	int matchCase;
	63	}
	64	StringParser;
	65
	66
	67	typedef struct
	68	{
	69	//UThread* thread;
	70	OValue* rules;
	71	UCellSlice input;
	72	OString* blk;
	73	int error;
	74	}
	75	BlockParser;
31	76
32	77
…	…
48	93
49	94
50		static const uint8_t* thruChars( const OValue* bitsetVal,
51		const uint8_t* it,
52		const uint8_t* end )
	95	static int _repeatChar( OString* input, OIndex pos, int limit, int c )
53	96	{
	97	UChar* start;
	98	UChar* it;
	99	UChar* end;
	100
	101	it = input->charArray;
	102	end = it + input->used;
	103	it += pos;
	104	if( end > (it + limit) )
	105	end = it + limit;
	106
	107	start = it;
	108	while( it != end )
	109	{
	110	if( *it != c )
	111	break;
	112	++it;
	113	}
	114	return it - start;
	115	}
	116
	117
	118	static int _repeatBitset( OString* input, OIndex pos, int limit,
	119	const OValue* patc )
	120	{
	121	UChar* start;
	122	UChar* it;
	123	UChar* end;
54	124	int c;
55		OBinary* bin = orSTRING(~~bitsetVal~~);
	125	OBinary* bin = orSTRING(patc);
56	126	const uint8_t* bits = bin->byteArray;
57	127	int maxC = bin->used * 8;
58	128
	129	it = input->charArray;
	130	end = it + input->used;
	131	it += pos;
	132	if( end > (it + limit) )
	133	end = it + limit;
	134
	135	start = it;
59	136	while( it != end )
60	137	{
…	…
66	143	++it;
67	144	}
68		return it;
	145	return it - start;
69	146	}
70	147
…	…
73	150	Returns zero if matching rule not found.
74	151	*/
75		static const OValue* ~~evalParseStr( const OValue* rit, const OValue* rend~~,
76		~~OIndex strN, int~~* spos )
	152	static const OValue* _parseStr( StringParser* pe, const OValue* rit,
	153	const OValue* rend, OIndex* spos )
77	154	{
78		OString* str;
79		int atom;
80		int pos = *spos;
	155	const OBlock* cblk;
	156	const OValue* tval;
	157	int32_t repMin;
	158	int32_t repMax;
	159	OString* istr;
	160	OIndex pos = *spos;
	161
	162	istr = pe->str;
81	163
82	164	match:
…	…
87	169	{
88	170	case OT_WORD:
89		atom = orAtom(rit);
90
91		if( atom == OR_ATOM_OPT )
92		{
93		const OValue* r2 = rit + 1;
94
95		if( r2 == rend )
	171	switch( orAtom(rit) )
	172	{
	173	case OR_ATOM_OPT:
	174	++rit;
	175	repMin = 0;
	176	repMax = 1;
	177	goto repeat;
	178
	179	case OR_ATOM_ANY:
	180	++rit;
	181	repMin = 0;
	182	repMax = 0x7fffffff;
	183	goto repeat;
	184
	185	case OR_ATOM_SOME:
	186	++rit;
	187	repMin = 1;
	188	repMax = 0x7fffffff;
	189	goto repeat;
	190
	191	case OR_ATOM_BREAK:
	192	return rit;
	193
	194	case OR_ATOM_BAR:
	195	goto complete;
	196
	197	case OR_ATOM_TO:
	198	case OR_ATOM_THRU:
	199	{
	200	OAtom ratom = orAtom(rit);
	201
	202	++rit;
	203	if( rit == rend )
96	204	return 0;
97	205
98		evalParseStr( r2, r2 + 1, strN, &pos );
99		if( orErrorThrown )
100		return 0;
101		rit += 2;
102		}
103		else if( atom == OR_ATOM_ANY )
104		{
105		//until [not eval-pstr second it]
106		//it: skip it 2
107
108		const OValue* r2 = rit + 1;
109
110		if( r2 == rend )
111		return 0;
112
113		if( orIs(r2, OT_WORD) )
	206	if( orIs(rit, OT_WORD) )
114	207	{
115		OBlock* ctxBlk;
116		orWordVal( r2, ctxBlk, r2 );
117		}
118
119		if( orIs(r2, OT_BITSET) )
120		{
121		const uint8_t* it;
122		const uint8_t* end;
123
124		str = orStringPtr(strN);
125		it = str->byteArray + pos;
126		end = str->byteArray + str->used;
127
128		end = thruChars( r2, it, end );
129		if( end != it )
130		pos = end - str->byteArray;
131		rit += 2;
	208	orWordVal( rit, cblk, tval );
132	209	}
133	210	else
134	211	{
135		orError( "parse any expected bitset" );
	212	tval = rit;
	213	}
	214
	215	switch( orType(tval) )
	216	{
	217	case OT_CHAR:
	218	{
	219	int c = orChar(tval);
	220	UChar* cp = istr->charArray + pos;
	221	UChar* end = istr->charArray + pe->input.end;
	222	while( cp != end )
	223	{
	224	if( *cp == c )
	225	break;
	226	++cp;
	227	}
	228	if( cp == end )
	229	goto failed;
	230	pos = cp - istr->charArray;
	231	if( ratom == OR_ATOM_THRU )
	232	++pos;
	233	}
	234	break;
	235
	236	case OT_STRING:
	237	{
	238	OString* pat = orSTRING(tval);
	239	pos = orFindString( istr, pos,
	240	pat, tval->series.it );
	241	if( pos < 0 )
	242	goto failed;
	243	if( ratom == OR_ATOM_THRU )
	244	pos += pat->used - tval->series.it;
	245	}
	246	break;
	247
	248	case OT_BITSET:
	249	{
	250	OBinary* bin = orSTRING(tval);
	251	const uint8_t* bits = bin->byteArray;
	252	int maxC = bin->used * 8;
	253	UChar* cp = istr->charArray + pos;
	254	UChar* end = istr->charArray + pe->input.end;
	255	int c;
	256	while( cp != end )
	257	{
	258	c = *cp;
	259	if( c < maxC )
	260	{
	261	if( orBitIsSet( bits, c ) )
	262	break;
	263	}
	264	++cp;
	265	}
	266	if( cp == end )
	267	goto failed;
	268	pos = cp - istr->charArray;
	269	if( ratom == OR_ATOM_THRU )
	270	++pos;
	271	}
	272	break;
	273
	274	case OT_BLOCK:
	275	// TODO
	276	orError( PARSE_EX
	277	"to/thru block! not implemented" );
	278	pe->error = 1;
	279	return 0;
	280	}
	281	++rit;
	282	}
	283	break;
	284
	285	case OR_ATOM_SKIP:
	286	// TODO - int! skip
	287	//if( pos >= istr->used )
	288	// return 0;
	289	++rit;
	290	++pos;
	291	break;
	292
	293	//case OR_ATOM_COPY:
	294
	295	default:
	296	orWordVal( rit, cblk, tval );
	297
	298	if( orIs(tval, OT_CHAR) )
	299	goto match_char;
	300	else if( orIs(tval, OT_STRING) )
	301	goto match_string;
	302	else if( orIs(tval, OT_BLOCK) )
	303	goto match_block;
	304	else if( orIs(tval, OT_BITSET) )
	305	goto match_bitset;
	306	else
	307	{
	308	orError( PARSE_EX
	309	"parse expected char!/block!/bitset!" );
	310	pe->error = 1;
136	311	return 0;
137	312	}
138		}
139		else if( atom == OR_ATOM_SOME )
140		{
141		const OValue* found;
142		const OValue* r2end;
143		const OValue* r2 = rit + 1;
144
145		if( r2 == rend )
146		return 0;
147
148		if( orIs(r2, OT_WORD) )
149		{
150		OBlock* ctxBlk;
151		orWordVal( r2, ctxBlk, r2 );
152		}
153
154		if( orIs(r2, OT_BLOCK) )
155		{
156		OBlock* blk2;
157
158		blk2 = orBLOCK(r2);
159		r2 = blk2->values + r2->series.it;
160		r2end = blk2->values + blk2->used;
161
162		found = evalParseStr( r2, r2end, strN, &pos );
163		if( found )
164		{
165		do
166		{
167		found = evalParseStr( r2, r2end, strN, &pos );
168		if( orErrorThrown )
169		return 0;
170		}
171		while( found );
172		rit += 2;
173		}
174		else
175		{
176		goto failed_eval;
177		}
178		}
179		else if( orIs(r2, OT_BITSET) )
180		{
181		const uint8_t* it;
182		const uint8_t* end;
183
184		str = orStringPtr(strN);
185		it = str->byteArray + pos;
186		end = str->byteArray + str->used;
187
188		end = thruChars( r2, it, end );
189		if( end == it )
190		{
191		rit = nextRule( rit, rend );
192		if( ! rit )
193		return 0;
194		}
195		else
196		{
197		pos = end - str->byteArray;
198		rit += 2;
199		}
200		}
201		else
202		{
203		orError( "parse some expected block or bitset" );
204		return 0;
205		}
206		}
207		else if( atom == OR_ATOM_BREAK )
208		{
209		return rit;
210		}
211		else if( atom == OR_ATOM_BAR )
212		{
213		goto complete;
214		}
215		else if( atom == OR_ATOM_TO )
216		{
217		int si;
218		const OValue* r2 = rit + 1;
219
220		if( r2 == rend )
221		return 0;
222
223		if( ! orIsString( orType(r2) ) )
224		return 0;
225
226		si = orFindString( orStringPtr(strN), pos,
227		orSTRING(r2), r2->series.it );
228		if( si > -1 )
229		{
230		pos = si;
231		rit += 2;
232		}
233		else
234		{
235		goto failed;
236		}
237		}
238		else if( atom == OR_ATOM_SKIP )
239		{
240		//if( pos >= str->used )
241		// return 0;
242		++rit;
243		++pos;
244		}
245		/*
246		else if( atom == OR_ATOM_THRU )
247		{
248		}
249		else if( atom == OR_ATOM_SET )
250		{
251		}
252		else if( atom == OR_ATOM_COPY )
253		{
254		}
255		*/
256		else
257		{
258		const OValue* wval;
259		const OBlock* cblk;
260
261		orWordVal( rit, cblk, wval );
262
263		if( orIs(wval, OT_BLOCK) )
264		{
265		// Same as case OT_BLOCK below.
266		cblk = orBLOCK( wval );
267		wval = evalParseStr( cblk->values + wval->series.it,
268		cblk->values + cblk->used,
269		strN, &pos );
270		if( wval )
271		++rit;
272		else
273		goto failed_eval;
274		}
275		else if( orIs(wval, OT_BITSET) )
276		{
277		OString* bin = orSTRING( wval );
278		int c;
279
280		str = orStringPtr(strN);
281		c = str->charArray[ pos ];
282		if( orBitIsSet( bin->byteArray, c ) )
283		{
284		++rit;
285		++pos;
286		}
287		else
288		goto failed;
289		}
290		else
291		{
292		orError( "parse expected block or bitset" );
293		return 0;
294		}
	313	break;
295	314	}
296	315	break;
…	…
298	317	case OT_SETWORD:
299	318	{
300		OValue* wval;
301		OBlock* ctxBlk;
302
303		orWordVal( rit, ctxBlk, wval );
304
305		orSetTF( wval, OT_STRING );
306		orSetSeries( wval, strN, pos );
307
	319	OValue* cell;
	320	orWordVal( rit, cblk, cell );
308	321	++rit;
	322
	323	orSetTF( cell, OT_STRING );
	324	orSetSeries( cell, pe->input.n, pos );
309	325	}
310	326	break;
311
	327	#if 0
	328	case OT_GETWORD:
	329	{
	330	OValue* cell;
	331	orWordVal( rit, cblk, cell );
	332	++rit;
	333
	334	if( orIs(cell, OT_SLICE) && (cell->series.n == pe->input.n) )
	335	cell->slice.end = pos;
	336	}
	337	break;
	338	#endif
312	339	case OT_INTEGER:
313		{
314		const OValue* r2 = rit + 1;
315
316		if( r2 == rend )
	340	repMin = orInt(rit);
	341
	342	++rit;
	343	if( rit == rend )
317	344	return 0;
318	345
319		if( orIs(r2, OT_INTEGER) )
320		{
321		int count = 0;
322		int maxCount = orInt(r2);
323
324		++r2;
325		if( r2 == rend )
326		return 0;
327
328		while( 1 )
329		{
330		if( ! evalParseStr( r2, r2 + 1, strN, &pos ) )
331		break;
332		++count;
333		}
334
335		if( orErrorThrown )
336		return 0;
337		if( (count >= orInt(rit)) && (count <= maxCount) )
338		rit += 3;
339		else
340		goto failed;
341		}
342		}
343		break;
	346	if( orIs(rit, OT_INTEGER) )
	347	{
	348	repMax = orInt(rit);
	349	++rit;
	350	}
	351	else
	352	{
	353	repMax = repMin;
	354	}
	355	goto repeat;
344	356
345	357	case OT_CHAR:
346		str = orStringPtr(strN);
347		if( str->charArray[ pos ] == orInt(rit) )
	358	tval = rit;
	359	match_char:
	360	if( istr->charArray[ pos ] == orChar(tval) )
348	361	{
349	362	++rit;
…	…
355	368
356	369	case OT_BLOCK:
	370	tval = rit;
	371	match_block:
	372	cblk = orBLOCK( tval );
	373	tval = _parseStr( pe, cblk->values + tval->series.it,
	374	cblk->values + cblk->used, &pos );
	375	istr = pe->str;
	376	if( ! tval )
	377	goto failed_eval;
	378	++rit;
	379	break;
	380
	381	case OT_PAREN:
	382	orEvalBlock( orBlockPtr(rit->series.n), rit->series.it );
	383	if( orErrorThrown )
	384	{
	385	pe->error = 1;
	386	return 0;
	387	}
	388
	389	/* Re-aquire pointer & check if input modified. */
	390	istr = pe->str = orStringPtr( pe->input.n );
	391	if( istr->used < pe->input.end )
	392	pe->input.end = istr->used;
	393
	394	++rit;
	395	break;
	396
	397	case OT_STRING:
	398	tval = rit;
	399	match_string:
357	400	{
358		const OValue* found;
359		const OBlock* cblk;
360
361		cblk = orBLOCK( rit );
362		found = evalParseStr( cblk->values + rit->series.it,
363		cblk->values + cblk->used,
364		strN, &pos );
365		if( found )
366		++rit;
367		else
368		goto failed_eval;
369		}
370		break;
371
372		case OT_PAREN:
373		{
374		orEvalBlock( orBLOCK(rit), rit->series.it );
375		if( orErrorThrown )
376		return 0;
377		++rit;
378		}
379		break;
380
381		case OT_STRING:
382		{
383		OString* mstr = orSTRING(rit);
384
385		str = orStringPtr(strN);
	401	OString* pat = orSTRING( tval );
	402
386	403	/*
387		if( ~~use~~Case )
388		pos = orMatchStringCase(~~str, pos, mstr, rit~~->series.it);
	404	if( pe->matchCase )
	405	pos = orMatchStringCase(istr, pos, pat, tval->series.it);
389	406	else
390	407	*/
391		pos = orMatchString( ~~str, pos, mstr, rit~~->series.it );
	408	pos = orMatchString( istr, pos, pat, tval->series.it );
392	409
393	410	if( pos )
…	…
398	415