xref: /plan9/sys/src/cmd/gs/src/slzwe.c (revision 593dc095aefb2a85c828727bbfa9da139a49bdf4)
1 /* Copyright (C) 1993 Aladdin Enterprises.  All rights reserved.
2 
3   This file is part of Aladdin Ghostscript.
4 
5   Aladdin Ghostscript is distributed with NO WARRANTY OF ANY KIND.  No author
6   or distributor accepts any responsibility for the consequences of using it,
7   or for whether it serves any particular purpose or works at all, unless he
8   or she says so in writing.  Refer to the Aladdin Ghostscript Free Public
9   License (the "License") for full details.
10 
11   Every copy of Aladdin Ghostscript must include a copy of the License,
12   normally in a plain ASCII text file named PUBLIC.  The License grants you
13   the right to copy, modify and redistribute Aladdin Ghostscript, but only
14   under certain conditions described in the License.  Among other things, the
15   License requires that the copyright notice and this notice be preserved on
16   all copies.
17 */
18 
19 /* $Id: slzwe.c,v 1.4 2005/04/04 23:00:24 igor Exp $ */
20 /* LZW encoding filter */
21 #include "stdio_.h"	/* includes std.h */
22 #include "gdebug.h"
23 #include "strimpl.h"
24 #include "slzwx.h"
25 
26 /********************************************************/
27 /* LZW routines are based on:				*/
28 /* Dr. Dobbs Journal --- Oct. 1989. 			*/
29 /* Article on LZW Data Compression by Mark R. Nelson 	*/
30 /********************************************************/
31 
32 /* Define the special codes */
33 #define code_reset 256
34 #define code_eod 257
35 #define code_0 258			/* first assignable code */
36 
37 /* Import the stream closing procedure */
38 extern stream_proc_release(s_LZW_release);
39 
40 typedef struct lzw_encode_s {
41 	byte datum;			/* last byte of this code */
42 	ushort prefix;			/* code for prefix of this code */
43 } lzw_encode;
44 
45 #define encode_max 4095		/* max # of codes, must be */
46 				/* > code_0 and <= 4095 */
47 #define hash_size (encode_max + encode_max / 4)
48 
49 struct lzw_encode_table_s {
50 	lzw_encode encode[encode_max];
51 	ushort hashed[hash_size];
52 };
53 gs_private_st_simple(st_lzwe_table, lzw_encode_table, "lzw_encode_table");
54 
55 /* Hashing function */
56 #define encode_hash(code, chr)\
57   ((uint)((code) * 59 + (chr) * ((hash_size / 256) | 1)) % hash_size)
58 
59 /* Internal routine to put a code into the output buffer. */
60 /* Let S = ss->code_size, M = ss->next_code, N = 1 << M. */
61 /* Relevant invariants: 9 <= S <= 12; N / 2 <= M < N; 0 <= code < N; */
62 /* 1 <= ss->bits_left <= 8; only the rightmost (8 - ss->bits_left) */
63 /* bits of ss->bits contain valid data. */
64 private byte *
lzw_put_code(register stream_LZW_state * ss,byte * q,uint code)65 lzw_put_code(register stream_LZW_state *ss, byte *q, uint code)
66 {	uint size = ss->code_size;
67 	byte cb = (ss->bits << ss->bits_left) +
68 		(code >> (size - ss->bits_left));
69 	if_debug2('W', "[w]writing 0x%x,%d\n", code, ss->code_size);
70 	*++q = cb;
71 	if ( (ss->bits_left += 8 - size) <= 0 )
72 	{	*++q = code >> -ss->bits_left;
73 		ss->bits_left += 8;
74 	}
75 	ss->bits = code;
76 	return q;
77 }
78 
79 /* Internal routine to reset the encoding table */
80 private void
lzw_reset_encode(stream_LZW_state * ss)81 lzw_reset_encode(stream_LZW_state *ss)
82 {	register int c;
83 	lzw_encode_table *table = ss->table.encode;
84 	ss->next_code = code_0;
85 	ss->code_size = 9;
86 	ss->prev_code = code_eod;
87 	for ( c = 0; c < hash_size; c++ )
88 		table->hashed[c] = code_eod;
89 	for ( c = 0; c < 256; c++ )
90 	{	lzw_encode *ec = &table->encode[c];
91 		register ushort *tc = &table->hashed[encode_hash(code_eod, c)];
92 		while ( *tc != code_eod )
93 		  if ( ++tc == &table->hashed[hash_size] )
94 		    tc = &table->hashed[0];
95 		*tc = c;
96 		ec->datum = c, ec->prefix = code_eod;
97 	}
98 	table->encode[code_eod].prefix = code_reset;	/* guarantee no match */
99 }
100 
101 #define ss ((stream_LZW_state *)st)
102 
103 /* Initialize LZWEncode filter */
104 private int
s_LZWE_init(stream_state * st)105 s_LZWE_init(stream_state *st)
106 {	ss->bits_left = 8;
107 	ss->table.encode = gs_alloc_struct(st->memory,
108 			lzw_encode_table, &st_lzwe_table, "LZWEncode init");
109 	if ( ss->table.encode == 0 )
110 		return ERRC;		/****** WRONG ******/
111 	ss->first = true;
112 	lzw_reset_encode(ss);
113 	return 0;
114 }
115 
116 /* Process a buffer */
117 private int
s_LZWE_process(stream_state * st,stream_cursor_read * pr,stream_cursor_write * pw,bool last)118 s_LZWE_process(stream_state *st, stream_cursor_read *pr,
119   stream_cursor_write *pw, bool last)
120 {	register const byte *p = pr->ptr;
121 	const byte *rlimit = pr->limit;
122 	register byte *q = pw->ptr;
123 	byte *wlimit = pw->limit;
124 	int code = ss->prev_code;
125 	lzw_encode_table *table = ss->table.encode;
126 	ushort *table_end = &table->hashed[hash_size];
127 	int status = 0;
128 	int limit_code;
129 #define set_limit_code()\
130   limit_code = (1 << ss->code_size) - ss->EarlyChange;\
131   if ( limit_code > encode_max ) limit_code = encode_max
132 	set_limit_code();
133 	if ( ss->first )
134 	{	/* Emit the initial reset code. */
135 		if ( wlimit - q < 2 )
136 			return 1;
137 		q = lzw_put_code(ss, q, code_reset);
138 		ss->first = false;
139 	}
140 	while ( p < rlimit )
141 	{	byte c = p[1];
142 		ushort *tp;
143 		for ( tp = &table->hashed[encode_hash(code, c)]; ; )
144 		{	lzw_encode *ep = &table->encode[*tp];
145 			if ( ep->prefix == code && ep->datum == c )
146 			{	code = *tp;
147 				p++;
148 				break;
149 			}
150 			else if ( *tp != code_eod )
151 			{	if ( ++tp == table_end )
152 				  tp = &table->hashed[0]; /* wrap around */
153 			}
154 			else
155 			{	/* end of recognized sequence */
156 				if ( wlimit - q <= 4 )
157 				{	status = 1;
158 					goto out;
159 				}
160 				q = lzw_put_code(ss, q, code);
161 				if ( ss->next_code == limit_code )
162 				{	/* Reached power of 2 or limit. */
163 					/* Determine which. */
164 					if ( ss->next_code == encode_max )
165 					{	q = lzw_put_code(ss, q, code_reset);
166 						lzw_reset_encode(ss);
167 						set_limit_code();
168 						goto cx;
169 					}
170 					ss->code_size++;
171 					set_limit_code();
172 				}
173 				if_debug3('W', "[W]encoding 0x%x=0x%x+%c\n",
174 					  ss->next_code, code, c);
175 				*tp = ss->next_code++;
176 				ep = &table->encode[*tp];
177 				ep->datum = c;
178 				ep->prefix = code;
179 cx:				code = code_eod;
180 				break;
181 			}
182 		}
183 	}
184 	if ( last && status == 0 )
185 	{	if ( wlimit - q < 4 )
186 			status = 1;
187 		else
188 		  {	if ( code != code_eod )
189 			  {	q = lzw_put_code(ss, q, code);	/* put out final code */
190 			  }
191 			q = lzw_put_code(ss, q, code_eod);
192 			if ( ss->bits_left < 8 )
193 			  *++q = ss->bits << ss->bits_left;  /* final byte */
194 		  }
195 	}
196 out:	ss->prev_code = code;
197 	pr->ptr = p;
198 	pw->ptr = q;
199 	return status;
200 }
201 
202 #undef ss
203 
204 /* Stream template */
205 const stream_template s_LZWE_template =
206 {	&st_LZW_state, s_LZWE_init, s_LZWE_process, 1, 4, s_LZW_release,
207 	s_LZW_set_defaults
208 };
209