1 | #include <assert.h> |
2 | #include <stdio.h> |
3 | #include <string.h> |
4 | |
5 | #include "buffer.h" |
6 | #include "houdini.h" |
7 | #include "utf8.h" |
8 | #include "entities.inc" |
9 | |
10 | /* Binary tree lookup code for entities added by JGM */ |
11 | |
12 | static const unsigned char *S_lookup(int i, int low, int hi, |
13 | const unsigned char *s, int len) { |
14 | int j; |
15 | int cmp = |
16 | strncmp((const char *)s, (const char *)cmark_entities[i].entity, len); |
17 | if (cmp == 0 && cmark_entities[i].entity[len] == 0) { |
18 | return (const unsigned char *)cmark_entities[i].bytes; |
19 | } else if (cmp <= 0 && i > low) { |
20 | j = i - ((i - low) / 2); |
21 | if (j == i) |
22 | j -= 1; |
23 | return S_lookup(j, low, i - 1, s, len); |
24 | } else if (cmp > 0 && i < hi) { |
25 | j = i + ((hi - i) / 2); |
26 | if (j == i) |
27 | j += 1; |
28 | return S_lookup(j, i + 1, hi, s, len); |
29 | } else { |
30 | return NULL; |
31 | } |
32 | } |
33 | |
34 | static const unsigned char *S_lookup_entity(const unsigned char *s, int len) { |
35 | return S_lookup(CMARK_NUM_ENTITIES / 2, 0, CMARK_NUM_ENTITIES - 1, s, len); |
36 | } |
37 | |
38 | bufsize_t houdini_unescape_ent(cmark_strbuf *ob, const uint8_t *src, |
39 | bufsize_t size) { |
40 | bufsize_t i = 0; |
41 | |
42 | if (size >= 3 && src[0] == '#') { |
43 | int codepoint = 0; |
44 | int num_digits = 0; |
45 | int max_digits = 7; |
46 | |
47 | if (_isdigit(src[1])) { |
48 | for (i = 1; i < size && _isdigit(src[i]); ++i) { |
49 | codepoint = (codepoint * 10) + (src[i] - '0'); |
50 | |
51 | if (codepoint >= 0x110000) { |
52 | // Keep counting digits but |
53 | // avoid integer overflow. |
54 | codepoint = 0x110000; |
55 | } |
56 | } |
57 | |
58 | num_digits = i - 1; |
59 | max_digits = 7; |
60 | } |
61 | |
62 | else if (src[1] == 'x' || src[1] == 'X') { |
63 | for (i = 2; i < size && _isxdigit(src[i]); ++i) { |
64 | codepoint = (codepoint * 16) + ((src[i] | 32) % 39 - 9); |
65 | |
66 | if (codepoint >= 0x110000) { |
67 | // Keep counting digits but |
68 | // avoid integer overflow. |
69 | codepoint = 0x110000; |
70 | } |
71 | } |
72 | |
73 | num_digits = i - 2; |
74 | max_digits = 6; |
75 | } |
76 | |
77 | if (num_digits >= 1 && num_digits <= max_digits && |
78 | i < size && src[i] == ';') { |
79 | if (codepoint == 0 || (codepoint >= 0xD800 && codepoint < 0xE000) || |
80 | codepoint >= 0x110000) { |
81 | codepoint = 0xFFFD; |
82 | } |
83 | cmark_utf8proc_encode_char(codepoint, ob); |
84 | return i + 1; |
85 | } |
86 | } |
87 | |
88 | else { |
89 | if (size > CMARK_ENTITY_MAX_LENGTH) |
90 | size = CMARK_ENTITY_MAX_LENGTH; |
91 | |
92 | for (i = CMARK_ENTITY_MIN_LENGTH; i < size; ++i) { |
93 | if (src[i] == ' ') |
94 | break; |
95 | |
96 | if (src[i] == ';') { |
97 | const unsigned char *entity = S_lookup_entity(src, i); |
98 | |
99 | if (entity != NULL) { |
100 | cmark_strbuf_puts(ob, (const char *)entity); |
101 | return i + 1; |
102 | } |
103 | |
104 | break; |
105 | } |
106 | } |
107 | } |
108 | |
109 | return 0; |
110 | } |
111 | |
112 | int houdini_unescape_html(cmark_strbuf *ob, const uint8_t *src, |
113 | bufsize_t size) { |
114 | bufsize_t i = 0, org, ent; |
115 | |
116 | while (i < size) { |
117 | org = i; |
118 | while (i < size && src[i] != '&') |
119 | i++; |
120 | |
121 | if (likely(i > org)) { |
122 | if (unlikely(org == 0)) { |
123 | if (i >= size) |
124 | return 0; |
125 | |
126 | cmark_strbuf_grow(ob, HOUDINI_UNESCAPED_SIZE(size)); |
127 | } |
128 | |
129 | cmark_strbuf_put(ob, src + org, i - org); |
130 | } |
131 | |
132 | /* escaping */ |
133 | if (i >= size) |
134 | break; |
135 | |
136 | i++; |
137 | |
138 | ent = houdini_unescape_ent(ob, src + i, size - i); |
139 | i += ent; |
140 | |
141 | /* not really an entity */ |
142 | if (ent == 0) |
143 | cmark_strbuf_putc(ob, '&'); |
144 | } |
145 | |
146 | return 1; |
147 | } |
148 | |
149 | void houdini_unescape_html_f(cmark_strbuf *ob, const uint8_t *src, |
150 | bufsize_t size) { |
151 | if (!houdini_unescape_html(ob, src, size)) |
152 | cmark_strbuf_put(ob, src, size); |
153 | } |
154 | |