]>
Commit | Line | Data |
---|---|---|
53e1b683 | 1 | /* SPDX-License-Identifier: LGPL-2.1+ */ |
08bcebf3 | 2 | |
11c3a366 TA |
3 | #include <errno.h> |
4 | #include <stddef.h> | |
08bcebf3 LP |
5 | #include <string.h> |
6 | ||
11c3a366 | 7 | #include "macro.h" |
07630cea | 8 | #include "string-util.h" |
08bcebf3 LP |
9 | #include "xml.h" |
10 | ||
11 | enum { | |
bcf3295d | 12 | STATE_NULL, |
08bcebf3 LP |
13 | STATE_TEXT, |
14 | STATE_TAG, | |
15 | STATE_ATTRIBUTE, | |
16 | }; | |
17 | ||
bcf3295d LP |
18 | static void inc_lines(unsigned *line, const char *s, size_t n) { |
19 | const char *p = s; | |
20 | ||
21 | if (!line) | |
22 | return; | |
23 | ||
24 | for (;;) { | |
25 | const char *f; | |
26 | ||
27 | f = memchr(p, '\n', n); | |
28 | if (!f) | |
29 | return; | |
30 | ||
31 | n -= (f - p) + 1; | |
32 | p = f + 1; | |
33 | (*line)++; | |
34 | } | |
35 | } | |
36 | ||
08bcebf3 LP |
37 | /* We don't actually do real XML here. We only read a simplistic |
38 | * subset, that is a bit less strict that XML and lacks all the more | |
39 | * complex features, like entities, or namespaces. However, we do | |
40 | * support some HTML5-like simplifications */ | |
41 | ||
bcf3295d | 42 | int xml_tokenize(const char **p, char **name, void **state, unsigned *line) { |
08bcebf3 LP |
43 | const char *c, *e, *b; |
44 | char *ret; | |
45 | int t; | |
46 | ||
47 | assert(p); | |
48 | assert(*p); | |
49 | assert(name); | |
50 | assert(state); | |
51 | ||
52 | t = PTR_TO_INT(*state); | |
53 | c = *p; | |
54 | ||
bcf3295d LP |
55 | if (t == STATE_NULL) { |
56 | if (line) | |
57 | *line = 1; | |
58 | t = STATE_TEXT; | |
59 | } | |
60 | ||
08bcebf3 LP |
61 | for (;;) { |
62 | if (*c == 0) | |
63 | return XML_END; | |
64 | ||
65 | switch (t) { | |
66 | ||
67 | case STATE_TEXT: { | |
68 | int x; | |
69 | ||
70 | e = strchrnul(c, '<'); | |
71 | if (e > c) { | |
72 | /* More text... */ | |
73 | ret = strndup(c, e - c); | |
74 | if (!ret) | |
75 | return -ENOMEM; | |
76 | ||
bcf3295d LP |
77 | inc_lines(line, c, e - c); |
78 | ||
08bcebf3 LP |
79 | *name = ret; |
80 | *p = e; | |
81 | *state = INT_TO_PTR(STATE_TEXT); | |
82 | ||
83 | return XML_TEXT; | |
84 | } | |
85 | ||
86 | assert(*e == '<'); | |
87 | b = c + 1; | |
88 | ||
89 | if (startswith(b, "!--")) { | |
90 | /* A comment */ | |
91 | e = strstr(b + 3, "-->"); | |
92 | if (!e) | |
93 | return -EINVAL; | |
94 | ||
bcf3295d LP |
95 | inc_lines(line, b, e + 3 - b); |
96 | ||
08bcebf3 LP |
97 | c = e + 3; |
98 | continue; | |
99 | } | |
100 | ||
101 | if (*b == '?') { | |
102 | /* Processing instruction */ | |
103 | ||
104 | e = strstr(b + 1, "?>"); | |
105 | if (!e) | |
106 | return -EINVAL; | |
107 | ||
bcf3295d LP |
108 | inc_lines(line, b, e + 2 - b); |
109 | ||
08bcebf3 LP |
110 | c = e + 2; |
111 | continue; | |
112 | } | |
113 | ||
114 | if (*b == '!') { | |
115 | /* DTD */ | |
116 | ||
117 | e = strchr(b + 1, '>'); | |
118 | if (!e) | |
119 | return -EINVAL; | |
120 | ||
bcf3295d LP |
121 | inc_lines(line, b, e + 1 - b); |
122 | ||
08bcebf3 LP |
123 | c = e + 1; |
124 | continue; | |
125 | } | |
126 | ||
127 | if (*b == '/') { | |
128 | /* A closing tag */ | |
129 | x = XML_TAG_CLOSE; | |
130 | b++; | |
131 | } else | |
132 | x = XML_TAG_OPEN; | |
133 | ||
134 | e = strpbrk(b, WHITESPACE "/>"); | |
135 | if (!e) | |
136 | return -EINVAL; | |
137 | ||
138 | ret = strndup(b, e - b); | |
139 | if (!ret) | |
140 | return -ENOMEM; | |
141 | ||
142 | *name = ret; | |
143 | *p = e; | |
144 | *state = INT_TO_PTR(STATE_TAG); | |
145 | ||
146 | return x; | |
147 | } | |
148 | ||
149 | case STATE_TAG: | |
150 | ||
151 | b = c + strspn(c, WHITESPACE); | |
152 | if (*b == 0) | |
153 | return -EINVAL; | |
154 | ||
bcf3295d LP |
155 | inc_lines(line, c, b - c); |
156 | ||
08bcebf3 LP |
157 | e = b + strcspn(b, WHITESPACE "=/>"); |
158 | if (e > b) { | |
159 | /* An attribute */ | |
160 | ||
161 | ret = strndup(b, e - b); | |
162 | if (!ret) | |
163 | return -ENOMEM; | |
164 | ||
165 | *name = ret; | |
166 | *p = e; | |
167 | *state = INT_TO_PTR(STATE_ATTRIBUTE); | |
168 | ||
169 | return XML_ATTRIBUTE_NAME; | |
170 | } | |
171 | ||
172 | if (startswith(b, "/>")) { | |
173 | /* An empty tag */ | |
174 | ||
175 | *name = NULL; /* For empty tags we return a NULL name, the caller must be prepared for that */ | |
176 | *p = b + 2; | |
177 | *state = INT_TO_PTR(STATE_TEXT); | |
178 | ||
179 | return XML_TAG_CLOSE_EMPTY; | |
180 | } | |
181 | ||
182 | if (*b != '>') | |
183 | return -EINVAL; | |
184 | ||
185 | c = b + 1; | |
186 | t = STATE_TEXT; | |
187 | continue; | |
188 | ||
189 | case STATE_ATTRIBUTE: | |
190 | ||
191 | if (*c == '=') { | |
192 | c++; | |
193 | ||
4c701096 | 194 | if (IN_SET(*c, '\'', '\"')) { |
08bcebf3 LP |
195 | /* Tag with a quoted value */ |
196 | ||
197 | e = strchr(c+1, *c); | |
198 | if (!e) | |
199 | return -EINVAL; | |
200 | ||
bcf3295d LP |
201 | inc_lines(line, c, e - c); |
202 | ||
08bcebf3 LP |
203 | ret = strndup(c+1, e - c - 1); |
204 | if (!ret) | |
205 | return -ENOMEM; | |
206 | ||
207 | *name = ret; | |
208 | *p = e + 1; | |
209 | *state = INT_TO_PTR(STATE_TAG); | |
210 | ||
211 | return XML_ATTRIBUTE_VALUE; | |
212 | ||
213 | } | |
214 | ||
215 | /* Tag with a value without quotes */ | |
216 | ||
217 | b = strpbrk(c, WHITESPACE ">"); | |
218 | if (!b) | |
219 | b = c; | |
220 | ||
221 | ret = strndup(c, b - c); | |
222 | if (!ret) | |
223 | return -ENOMEM; | |
224 | ||
225 | *name = ret; | |
226 | *p = b; | |
227 | *state = INT_TO_PTR(STATE_TAG); | |
228 | return XML_ATTRIBUTE_VALUE; | |
229 | } | |
230 | ||
231 | t = STATE_TAG; | |
232 | continue; | |
233 | } | |
234 | ||
235 | } | |
236 | ||
237 | assert_not_reached("Bad state"); | |
238 | } |