1 #include "command.h"
2 #include "debug.h"
3 #include "editor.h"
4 #include "env.h"
5 #include "error.h"
6 #include "util/ascii.h"
7 #include "util/ptr-array.h"
8 #include "util/str-util.h"
9 #include "util/string.h"
10 #include "util/utf8.h"
11 #include "util/xmalloc.h"
12
parse_sq(const char * cmd,size_t len,String * buf)13 static size_t parse_sq(const char *cmd, size_t len, String *buf)
14 {
15 size_t pos = 0;
16 char ch = '\0';
17 while (pos < len) {
18 ch = cmd[pos];
19 if (ch == '\'') {
20 break;
21 }
22 pos++;
23 }
24 string_add_buf(buf, cmd, pos);
25 if (ch == '\'') {
26 pos++;
27 }
28 return pos;
29 }
30
unicode_escape(const char * str,size_t count,String * buf)31 static size_t unicode_escape(const char *str, size_t count, String *buf)
32 {
33 if (count == 0) {
34 return 0;
35 }
36
37 CodePoint u = 0;
38 size_t i;
39 for (i = 0; i < count; i++) {
40 int x = hex_decode(str[i]);
41 if (x < 0) {
42 break;
43 }
44 u = u << 4 | x;
45 }
46 if (u_is_unicode(u)) {
47 string_add_ch(buf, u);
48 }
49 return i;
50 }
51
min(size_t a,size_t b)52 static size_t min(size_t a, size_t b)
53 {
54 return (a < b) ? a : b;
55 }
56
parse_dq(const char * cmd,size_t len,String * buf)57 static size_t parse_dq(const char *cmd, size_t len, String *buf)
58 {
59 size_t pos = 0;
60 while (pos < len) {
61 unsigned char ch = cmd[pos++];
62
63 if (ch == '"') {
64 break;
65 }
66
67 if (ch == '\\' && pos < len) {
68 ch = cmd[pos++];
69 switch (ch) {
70 case 'a': ch = '\a'; break;
71 case 'b': ch = '\b'; break;
72 case 'f': ch = '\f'; break;
73 case 'n': ch = '\n'; break;
74 case 'r': ch = '\r'; break;
75 case 't': ch = '\t'; break;
76 case 'v': ch = '\v'; break;
77 case '\\':
78 case '"':
79 break;
80 case 'x':
81 if (pos < len) {
82 const int x1 = hex_decode(cmd[pos]);
83 if (x1 >= 0 && ++pos < len) {
84 const int x2 = hex_decode(cmd[pos]);
85 if (x2 >= 0) {
86 pos++;
87 ch = x1 << 4 | x2;
88 break;
89 }
90 }
91 }
92 continue;
93 case 'u':
94 pos += unicode_escape(cmd + pos, min(4, len - pos), buf);
95 continue;
96 case 'U':
97 pos += unicode_escape(cmd + pos, min(8, len - pos), buf);
98 continue;
99 default:
100 string_add_byte(buf, '\\');
101 break;
102 }
103 }
104
105 string_add_byte(buf, ch);
106 }
107
108 return pos;
109 }
110
parse_var(const char * cmd,size_t len,String * buf)111 static size_t parse_var(const char *cmd, size_t len, String *buf)
112 {
113 if (len == 0 || !is_alpha_or_underscore(cmd[0])) {
114 return 0;
115 }
116
117 size_t n = 1;
118 while (n < len && is_alnum_or_underscore(cmd[n])) {
119 n++;
120 }
121
122 char *name = xstrcut(cmd, n);
123 char *value;
124 if (expand_builtin_env(name, &value)) {
125 if (value != NULL) {
126 string_add_str(buf, value);
127 free(value);
128 }
129 } else {
130 const char *val = getenv(name);
131 if (val != NULL) {
132 string_add_str(buf, val);
133 }
134 }
135
136 free(name);
137 return n;
138 }
139
parse_command_arg(const char * cmd,size_t len,bool tilde)140 char *parse_command_arg(const char *cmd, size_t len, bool tilde)
141 {
142 String buf;
143 size_t pos = 0;
144
145 if (tilde && len >= 2 && cmd[0] == '~' && cmd[1] == '/') {
146 const size_t home_dir_len = strlen(editor.home_dir);
147 buf = string_new(len + home_dir_len);
148 string_add_buf(&buf, editor.home_dir, home_dir_len);
149 string_add_byte(&buf, '/');
150 pos += 2;
151 } else {
152 buf = string_new(len);
153 }
154
155 while (pos < len) {
156 char ch = cmd[pos++];
157 switch (ch) {
158 case '\t':
159 case '\n':
160 case '\r':
161 case ' ':
162 case ';':
163 goto end;
164 case '\'':
165 pos += parse_sq(cmd + pos, len - pos, &buf);
166 break;
167 case '"':
168 pos += parse_dq(cmd + pos, len - pos, &buf);
169 break;
170 case '$':
171 pos += parse_var(cmd + pos, len - pos, &buf);
172 break;
173 case '\\':
174 if (pos == len) {
175 goto end;
176 }
177 ch = cmd[pos++];
178 // Fallthrough
179 default:
180 string_add_byte(&buf, ch);
181 break;
182 }
183 }
184
185 end:
186 return string_steal_cstring(&buf);
187 }
188
find_end(const char * cmd,const size_t startpos,CommandParseError * err)189 size_t find_end(const char *cmd, const size_t startpos, CommandParseError *err)
190 {
191 size_t pos = startpos;
192 while (1) {
193 switch (cmd[pos++]) {
194 case '\'':
195 while (1) {
196 if (cmd[pos] == '\'') {
197 pos++;
198 break;
199 }
200 if (cmd[pos] == '\0') {
201 *err = CMDERR_UNCLOSED_SINGLE_QUOTE;
202 return 0;
203 }
204 pos++;
205 }
206 break;
207 case '"':
208 while (1) {
209 if (cmd[pos] == '"') {
210 pos++;
211 break;
212 }
213 if (cmd[pos] == '\0') {
214 *err = CMDERR_UNCLOSED_DOUBLE_QUOTE;
215 return 0;
216 }
217 if (cmd[pos++] == '\\') {
218 if (cmd[pos] == '\0') {
219 goto unexpected_eof;
220 }
221 pos++;
222 }
223 }
224 break;
225 case '\\':
226 if (cmd[pos] == '\0') {
227 goto unexpected_eof;
228 }
229 pos++;
230 break;
231 case '\0':
232 case '\t':
233 case '\n':
234 case '\r':
235 case ' ':
236 case ';':
237 return pos - 1;
238 }
239 }
240 BUG("Unexpected break of outer loop");
241 unexpected_eof:
242 *err = CMDERR_UNEXPECTED_EOF;
243 return 0;
244 }
245
parse_commands(PointerArray * array,const char * cmd,CommandParseError * err)246 bool parse_commands(PointerArray *array, const char *cmd, CommandParseError *err)
247 {
248 size_t pos = 0;
249
250 while (1) {
251 while (ascii_isspace(cmd[pos])) {
252 pos++;
253 }
254
255 if (cmd[pos] == '\0') {
256 break;
257 }
258
259 if (cmd[pos] == ';') {
260 ptr_array_add(array, NULL);
261 pos++;
262 continue;
263 }
264
265 size_t end = find_end(cmd, pos, err);
266 if (*err != CMDERR_NONE) {
267 return false;
268 }
269
270 ptr_array_add(array, parse_command_arg(cmd + pos, end - pos, true));
271 pos = end;
272 }
273 ptr_array_add(array, NULL);
274 return true;
275 }
276
command_parse_error_to_string(CommandParseError err)277 const char *command_parse_error_to_string(CommandParseError err)
278 {
279 switch (err) {
280 case CMDERR_UNCLOSED_SINGLE_QUOTE:
281 return "Missing '";
282 case CMDERR_UNCLOSED_DOUBLE_QUOTE:
283 return "Missing \"";
284 case CMDERR_UNEXPECTED_EOF:
285 return "Unexpected EOF";
286 case CMDERR_NONE:
287 break;
288 }
289 return NULL;
290 }
291