Python code coverage for Modules/cjkcodecs/_codecs_hk.c

#	count	content
1	n/a	/*
2	n/a	* _codecs_hk.c: Codecs collection for encodings from Hong Kong
3	n/a	*
4	n/a	* Written by Hye-Shik Chang <perky@FreeBSD.org>
5	n/a	*/
6	n/a
7	n/a	#define USING_IMPORTED_MAPS
8	n/a
9	n/a	#include "cjkcodecs.h"
10	n/a	#include "mappings_hk.h"
11	n/a
12	n/a	/*
13	n/a	* BIG5HKSCS codec
14	n/a	*/
15	n/a
16	n/a	static const encode_map *big5_encmap = NULL;
17	n/a	static const decode_map *big5_decmap = NULL;
18	n/a
19	n/a	CODEC_INIT(big5hkscs)
20	n/a	{
21	n/a	static int initialized = 0;
22	n/a
23	n/a	if (!initialized && IMPORT_MAP(tw, big5, &big5_encmap, &big5_decmap))
24	n/a	return -1;
25	n/a	initialized = 1;
26	n/a	return 0;
27	n/a	}
28	n/a
29	n/a	/*
30	n/a	* There are four possible pair unicode -> big5hkscs maps as in HKSCS 2004:
31	n/a	* U+00CA U+0304 -> 8862 (U+00CA alone is mapped to 8866)
32	n/a	* U+00CA U+030C -> 8864
33	n/a	* U+00EA U+0304 -> 88a3 (U+00EA alone is mapped to 88a7)
34	n/a	* U+00EA U+030C -> 88a5
35	n/a	* These are handled by not mapping tables but a hand-written code.
36	n/a	*/
37	n/a	static const DBCHAR big5hkscs_pairenc_table[4] = {0x8862, 0x8864, 0x88a3, 0x88a5};
38	n/a
39	n/a	ENCODER(big5hkscs)
40	n/a	{
41	n/a	while (*inpos < inlen) {
42	n/a	Py_UCS4 c = INCHAR1;
43	n/a	DBCHAR code;
44	n/a	Py_ssize_t insize;
45	n/a
46	n/a	if (c < 0x80) {
47	n/a	REQUIRE_OUTBUF(1);
48	n/a	**outbuf = (unsigned char)c;
49	n/a	NEXT(1, 1);
50	n/a	continue;
51	n/a	}
52	n/a
53	n/a	insize = 1;
54	n/a	REQUIRE_OUTBUF(2);
55	n/a
56	n/a	if (c < 0x10000) {
57	n/a	if (TRYMAP_ENC(big5hkscs_bmp, code, c)) {
58	n/a	if (code == MULTIC) {
59	n/a	Py_UCS4 c2;
60	n/a	if (inlen - *inpos >= 2)
61	n/a	c2 = INCHAR2;
62	n/a	else
63	n/a	c2 = 0;
64	n/a
65	n/a	if (inlen - *inpos >= 2 &&
66	n/a	((c & 0xffdf) == 0x00ca) &&
67	n/a	((c2 & 0xfff7) == 0x0304)) {
68	n/a	code = big5hkscs_pairenc_table[
69	n/a	((c >> 4) \|
70	n/a	(c2 >> 3)) & 3];
71	n/a	insize = 2;
72	n/a	}
73	n/a	else if (inlen - *inpos < 2 &&
74	n/a	!(flags & MBENC_FLUSH))
75	n/a	return MBERR_TOOFEW;
76	n/a	else {
77	n/a	if (c == 0xca)
78	n/a	code = 0x8866;
79	n/a	else /* c == 0xea */
80	n/a	code = 0x88a7;
81	n/a	}
82	n/a	}
83	n/a	}
84	n/a	else if (TRYMAP_ENC(big5, code, c))
85	n/a	;
86	n/a	else
87	n/a	return 1;
88	n/a	}
89	n/a	else if (c < 0x20000)
90	n/a	return insize;
91	n/a	else if (c < 0x30000) {
92	n/a	if (TRYMAP_ENC(big5hkscs_nonbmp, code, c & 0xffff))
93	n/a	;
94	n/a	else
95	n/a	return insize;
96	n/a	}
97	n/a	else
98	n/a	return insize;
99	n/a
100	n/a	OUTBYTE1(code >> 8);
101	n/a	OUTBYTE2(code & 0xFF);
102	n/a	NEXT(insize, 2);
103	n/a	}
104	n/a
105	n/a	return 0;
106	n/a	}
107	n/a
108	n/a	#define BH2S(c1, c2) (((c1) - 0x87) * (0xfe - 0x40 + 1) + ((c2) - 0x40))
109	n/a
110	n/a	DECODER(big5hkscs)
111	n/a	{
112	n/a	while (inleft > 0) {
113	n/a	unsigned char c = INBYTE1;
114	n/a	Py_UCS4 decoded;
115	n/a
116	n/a	if (c < 0x80) {
117	n/a	OUTCHAR(c);
118	n/a	NEXT_IN(1);
119	n/a	continue;
120	n/a	}
121	n/a
122	n/a	REQUIRE_INBUF(2);
123	n/a
124	n/a	if (0xc6 > c \|\| c > 0xc8 \|\| (c < 0xc7 && INBYTE2 < 0xa1)) {
125	n/a	if (TRYMAP_DEC(big5, decoded, c, INBYTE2)) {
126	n/a	OUTCHAR(decoded);
127	n/a	NEXT_IN(2);
128	n/a	continue;
129	n/a	}
130	n/a	}
131	n/a
132	n/a	if (TRYMAP_DEC(big5hkscs, decoded, c, INBYTE2))
133	n/a	{
134	n/a	int s = BH2S(c, INBYTE2);
135	n/a	const unsigned char *hintbase;
136	n/a
137	n/a	assert(0x87 <= c && c <= 0xfe);
138	n/a	assert(0x40 <= INBYTE2 && INBYTE2 <= 0xfe);
139	n/a
140	n/a	if (BH2S(0x87, 0x40) <= s && s <= BH2S(0xa0, 0xfe)) {
141	n/a	hintbase = big5hkscs_phint_0;
142	n/a	s -= BH2S(0x87, 0x40);
143	n/a	}
144	n/a	else if (BH2S(0xc6,0xa1) <= s && s <= BH2S(0xc8,0xfe)){
145	n/a	hintbase = big5hkscs_phint_12130;
146	n/a	s -= BH2S(0xc6, 0xa1);
147	n/a	}
148	n/a	else if (BH2S(0xf9,0xd6) <= s && s <= BH2S(0xfe,0xfe)){
149	n/a	hintbase = big5hkscs_phint_21924;
150	n/a	s -= BH2S(0xf9, 0xd6);
151	n/a	}
152	n/a	else
153	n/a	return MBERR_INTERNAL;
154	n/a
155	n/a	if (hintbase[s >> 3] & (1 << (s & 7))) {
156	n/a	OUTCHAR(decoded \| 0x20000);
157	n/a	NEXT_IN(2);
158	n/a	}
159	n/a	else {
160	n/a	OUTCHAR(decoded);
161	n/a	NEXT_IN(2);
162	n/a	}
163	n/a	continue;
164	n/a	}
165	n/a
166	n/a	switch ((c << 8) \| INBYTE2) {
167	n/a	case 0x8862: OUTCHAR2(0x00ca, 0x0304); break;
168	n/a	case 0x8864: OUTCHAR2(0x00ca, 0x030c); break;
169	n/a	case 0x88a3: OUTCHAR2(0x00ea, 0x0304); break;
170	n/a	case 0x88a5: OUTCHAR2(0x00ea, 0x030c); break;
171	n/a	default: return 1;
172	n/a	}
173	n/a
174	n/a	NEXT_IN(2); /* all decoded code points are pairs, above. */
175	n/a	}
176	n/a
177	n/a	return 0;
178	n/a	}
179	n/a
180	n/a
181	n/a	BEGIN_MAPPINGS_LIST
182	n/a	MAPPING_DECONLY(big5hkscs)
183	n/a	MAPPING_ENCONLY(big5hkscs_bmp)
184	n/a	MAPPING_ENCONLY(big5hkscs_nonbmp)
185	n/a	END_MAPPINGS_LIST
186	n/a
187	n/a	BEGIN_CODECS_LIST
188	n/a	CODEC_STATELESS_WINIT(big5hkscs)
189	n/a	END_CODECS_LIST
190	n/a
191	n/a	I_AM_A_MODULE_FOR(hk)