Commit | Line | Data |
---|---|---|
04e98a4d AD |
1 | /* dquote_static.c |
2 | * | |
3efe3cb8 | 3 | * This file contains static functions that are related to |
04e98a4d AD |
4 | * parsing double-quotish expressions, but are used in more than |
5 | * one file. | |
6 | * | |
7 | * It is currently #included by regcomp.c and toke.c. | |
8 | */ | |
9 | ||
881ffab6 | 10 | #define PERL_IN_DQUOTE_STATIC_C |
881ffab6 KW |
11 | #include "embed.h" |
12 | ||
04e98a4d AD |
13 | /* |
14 | - regcurly - a little FSA that accepts {\d+,?\d*} | |
15 | Pulled from regcomp.c. | |
16 | */ | |
04e98a4d | 17 | PERL_STATIC_INLINE I32 |
4d68ffa0 KW |
18 | S_regcurly(pTHX_ const char *s, |
19 | const bool rbrace_must_be_escaped /* Should the terminating '} be | |
20 | preceded by a backslash? This | |
21 | is an abnormal case */ | |
22 | ) | |
04e98a4d | 23 | { |
881ffab6 | 24 | PERL_ARGS_ASSERT_REGCURLY; |
04e98a4d AD |
25 | |
26 | if (*s++ != '{') | |
27 | return FALSE; | |
28 | if (!isDIGIT(*s)) | |
29 | return FALSE; | |
30 | while (isDIGIT(*s)) | |
31 | s++; | |
2ec31dd9 | 32 | if (*s == ',') { |
04e98a4d | 33 | s++; |
2ec31dd9 KW |
34 | while (isDIGIT(*s)) |
35 | s++; | |
36 | } | |
4d68ffa0 KW |
37 | |
38 | return (rbrace_must_be_escaped) | |
39 | ? *s == '\\' && *(s+1) == '}' | |
40 | : *s == '}'; | |
04e98a4d | 41 | } |
db30362b | 42 | |
68b355dd KW |
43 | /* XXX Add documentation after final interface and behavior is decided */ |
44 | /* May want to show context for error, so would pass Perl_bslash_c(pTHX_ const char* current, const char* start, const bool output_warning) | |
45 | U8 source = *current; | |
68b355dd KW |
46 | */ |
47 | ||
48 | STATIC char | |
421e43ba | 49 | S_grok_bslash_c(pTHX_ const char source, const bool output_warning) |
68b355dd KW |
50 | { |
51 | ||
52 | U8 result; | |
53 | ||
421e43ba KW |
54 | if (! isPRINT_A(source)) { |
55 | const char msg[] = "Character following \"\\c\" must be printable ASCII"; | |
56 | if (! isASCII(source)) { | |
57 | Perl_croak(aTHX_ "%s", msg); | |
58 | } | |
59 | else if (output_warning) { /* Unprintables can be removed in v5.22 */ | |
60 | Perl_ck_warner_d(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX), "%s", | |
61 | msg); | |
17a3df4c | 62 | } |
68b355dd | 63 | } |
421e43ba KW |
64 | else if (source == '{') { |
65 | assert(isPRINT_A(toCTRL('{'))); | |
66 | Perl_croak(aTHX_ "Use \"%c\" instead of \"\\c{\"", toCTRL('{')); | |
67 | } | |
68b355dd KW |
68 | |
69 | result = toCTRL(source); | |
421e43ba KW |
70 | if (output_warning && ! isCNTRL_L1(result)) { |
71 | /* We use isCNTRL_L1 above and not simply isCNTRL, because on EBCDIC | |
72 | * machines, things like \cT map into a C1 control. */ | |
4d8be631 KW |
73 | U8 clearer[3]; |
74 | U8 i = 0; | |
75 | if (! isWORDCHAR(result)) { | |
76 | clearer[i++] = '\\'; | |
77 | } | |
78 | clearer[i++] = result; | |
79 | clearer[i++] = '\0'; | |
68b355dd | 80 | |
4d8be631 KW |
81 | Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), |
82 | "\"\\c%c\" is more clearly written simply as \"%s\"", | |
83 | source, | |
84 | clearer); | |
68b355dd KW |
85 | } |
86 | ||
87 | return result; | |
88 | } | |
89 | ||
db30362b | 90 | STATIC bool |
00ce5563 | 91 | S_grok_bslash_o(pTHX_ char **s, UV *uv, const char** error_msg, |
80f4111b | 92 | const bool output_warning, const bool strict, |
17896a85 | 93 | const bool silence_non_portable, |
80f4111b | 94 | const bool UTF) |
db30362b KW |
95 | { |
96 | ||
97 | /* Documentation to be supplied when interface nailed down finally | |
98 | * This returns FALSE if there is an error which the caller need not recover | |
99 | * from; , otherwise TRUE. In either case the caller should look at *len | |
100 | * On input: | |
00ce5563 KW |
101 | * s is the address of a pointer to a NULL terminated string that begins |
102 | * with 'o', and the previous character was a backslash. At exit, *s | |
103 | * will be advanced to the byte just after those absorbed by this | |
104 | * function. Hence the caller can continue parsing from there. In | |
105 | * the case of an error, this routine has generally positioned *s to | |
106 | * point just to the right of the first bad spot, so that a message | |
107 | * that has a "<--" to mark the spot will be correctly positioned. | |
db30362b KW |
108 | * uv points to a UV that will hold the output value, valid only if the |
109 | * return from the function is TRUE | |
db30362b KW |
110 | * error_msg is a pointer that will be set to an internal buffer giving an |
111 | * error message upon failure (the return is FALSE). Untouched if | |
112 | * function succeeds | |
113 | * output_warning says whether to output any warning messages, or suppress | |
114 | * them | |
80f4111b KW |
115 | * strict is true if this should fail instead of warn if there are |
116 | * non-octal digits within the braces | |
17896a85 KW |
117 | * silence_non_portable is true if to suppress warnings about the code |
118 | * point returned being too large to fit on all platforms. | |
80f4111b | 119 | * UTF is true iff the string *s is encoded in UTF-8. |
db30362b | 120 | */ |
00ce5563 | 121 | char* e; |
db30362b KW |
122 | STRLEN numbers_len; |
123 | I32 flags = PERL_SCAN_ALLOW_UNDERSCORES | |
124 | | PERL_SCAN_DISALLOW_PREFIX | |
125 | /* XXX Until the message is improved in grok_oct, handle errors | |
126 | * ourselves */ | |
127 | | PERL_SCAN_SILENT_ILLDIGIT; | |
128 | ||
129 | PERL_ARGS_ASSERT_GROK_BSLASH_O; | |
130 | ||
131 | ||
00ce5563 KW |
132 | assert(**s == 'o'); |
133 | (*s)++; | |
db30362b | 134 | |
00ce5563 | 135 | if (**s != '{') { |
db30362b KW |
136 | *error_msg = "Missing braces on \\o{}"; |
137 | return FALSE; | |
138 | } | |
139 | ||
00ce5563 | 140 | e = strchr(*s, '}'); |
db30362b | 141 | if (!e) { |
00ce5563 | 142 | (*s)++; /* Move past the '{' */ |
b8de99ca KW |
143 | while (isOCTAL(**s)) { /* Position beyond the legal digits */ |
144 | (*s)++; | |
145 | } | |
00ce5563 | 146 | *error_msg = "Missing right brace on \\o{"; |
db30362b KW |
147 | return FALSE; |
148 | } | |
149 | ||
00ce5563 KW |
150 | (*s)++; /* Point to expected first digit (could be first byte of utf8 |
151 | sequence if not a digit) */ | |
152 | numbers_len = e - *s; | |
db30362b | 153 | if (numbers_len == 0) { |
00ce5563 | 154 | (*s)++; /* Move past the } */ |
db30362b KW |
155 | *error_msg = "Number with no digits"; |
156 | return FALSE; | |
157 | } | |
158 | ||
17896a85 KW |
159 | if (silence_non_portable) { |
160 | flags |= PERL_SCAN_SILENT_NON_PORTABLE; | |
161 | } | |
162 | ||
00ce5563 | 163 | *uv = grok_oct(*s, &numbers_len, &flags, NULL); |
db30362b KW |
164 | /* Note that if has non-octal, will ignore everything starting with that up |
165 | * to the '}' */ | |
166 | ||
80f4111b KW |
167 | if (numbers_len != (STRLEN) (e - *s)) { |
168 | if (strict) { | |
169 | *s += numbers_len; | |
170 | *s += (UTF) ? UTF8SKIP(*s) : (STRLEN) 1; | |
171 | *error_msg = "Non-octal character"; | |
172 | return FALSE; | |
173 | } | |
174 | else if (output_warning) { | |
b67d718a KW |
175 | Perl_ck_warner(aTHX_ packWARN(WARN_DIGIT), |
176 | /* diag_listed_as: Non-octal character '%c'. Resolved as "%s" */ | |
177 | "Non-octal character '%c'. Resolved as \"\\o{%.*s}\"", | |
178 | *(*s + numbers_len), | |
179 | (int) numbers_len, | |
180 | *s); | |
80f4111b | 181 | } |
db30362b KW |
182 | } |
183 | ||
00ce5563 KW |
184 | /* Return past the '}' */ |
185 | *s = e + 1; | |
186 | ||
db30362b KW |
187 | return TRUE; |
188 | } | |
189 | ||
a0481293 | 190 | PERL_STATIC_INLINE bool |
00ce5563 | 191 | S_grok_bslash_x(pTHX_ char **s, UV *uv, const char** error_msg, |
80f4111b | 192 | const bool output_warning, const bool strict, |
17896a85 | 193 | const bool silence_non_portable, |
80f4111b | 194 | const bool UTF) |
a0481293 KW |
195 | { |
196 | ||
197 | /* Documentation to be supplied when interface nailed down finally | |
198 | * This returns FALSE if there is an error which the caller need not recover | |
199 | * from; , otherwise TRUE. In either case the caller should look at *len | |
200 | * On input: | |
00ce5563 KW |
201 | * s is the address of a pointer to a NULL terminated string that begins |
202 | * with 'x', and the previous character was a backslash. At exit, *s | |
203 | * will be advanced to the byte just after those absorbed by this | |
204 | * function. Hence the caller can continue parsing from there. In | |
205 | * the case of an error, this routine has generally positioned *s to | |
206 | * point just to the right of the first bad spot, so that a message | |
207 | * that has a "<--" to mark the spot will be correctly positioned. | |
a0481293 KW |
208 | * uv points to a UV that will hold the output value, valid only if the |
209 | * return from the function is TRUE | |
a0481293 KW |
210 | * error_msg is a pointer that will be set to an internal buffer giving an |
211 | * error message upon failure (the return is FALSE). Untouched if | |
212 | * function succeeds | |
213 | * output_warning says whether to output any warning messages, or suppress | |
214 | * them | |
80f4111b KW |
215 | * strict is true if anything out of the ordinary should cause this to |
216 | * fail instead of warn or be silent. For example, it requires | |
217 | * exactly 2 digits following the \x (when there are no braces). | |
218 | * 3 digits could be a mistake, so is forbidden in this mode. | |
344451ec KW |
219 | * silence_non_portable is true if to suppress warnings about the code |
220 | * point returned being too large to fit on all platforms. | |
80f4111b | 221 | * UTF is true iff the string *s is encoded in UTF-8. |
a0481293 | 222 | */ |
00ce5563 | 223 | char* e; |
a0481293 | 224 | STRLEN numbers_len; |
344451ec | 225 | I32 flags = PERL_SCAN_DISALLOW_PREFIX; |
a0481293 KW |
226 | |
227 | PERL_ARGS_ASSERT_GROK_BSLASH_X; | |
228 | ||
4f8dbb2d | 229 | PERL_UNUSED_ARG(output_warning); |
a0481293 | 230 | |
00ce5563 KW |
231 | assert(**s == 'x'); |
232 | (*s)++; | |
a0481293 | 233 | |
80f4111b KW |
234 | if (strict) { |
235 | flags |= PERL_SCAN_SILENT_ILLDIGIT; | |
236 | } | |
237 | ||
00ce5563 | 238 | if (**s != '{') { |
80f4111b KW |
239 | STRLEN len = (strict) ? 3 : 2; |
240 | ||
00ce5563 KW |
241 | *uv = grok_hex(*s, &len, &flags, NULL); |
242 | *s += len; | |
80f4111b KW |
243 | if (strict && len != 2) { |
244 | if (len < 2) { | |
245 | *s += (UTF) ? UTF8SKIP(*s) : 1; | |
246 | *error_msg = "Non-hex character"; | |
247 | } | |
248 | else { | |
249 | *error_msg = "Use \\x{...} for more than two hex characters"; | |
250 | } | |
251 | return FALSE; | |
252 | } | |
a0481293 KW |
253 | return TRUE; |
254 | } | |
255 | ||
00ce5563 | 256 | e = strchr(*s, '}'); |
a0481293 | 257 | if (!e) { |
00ce5563 | 258 | (*s)++; /* Move past the '{' */ |
b8de99ca KW |
259 | while (isXDIGIT(**s)) { /* Position beyond the legal digits */ |
260 | (*s)++; | |
261 | } | |
a0481293 KW |
262 | /* XXX The corresponding message above for \o is just '\\o{'; other |
263 | * messages for other constructs include the '}', so are inconsistent. | |
264 | */ | |
265 | *error_msg = "Missing right brace on \\x{}"; | |
266 | return FALSE; | |
267 | } | |
268 | ||
00ce5563 KW |
269 | (*s)++; /* Point to expected first digit (could be first byte of utf8 |
270 | sequence if not a digit) */ | |
271 | numbers_len = e - *s; | |
80f4111b KW |
272 | if (numbers_len == 0) { |
273 | if (strict) { | |
274 | (*s)++; /* Move past the } */ | |
275 | *error_msg = "Number with no digits"; | |
276 | return FALSE; | |
277 | } | |
278 | return TRUE; | |
279 | } | |
280 | ||
344451ec | 281 | flags |= PERL_SCAN_ALLOW_UNDERSCORES; |
17896a85 KW |
282 | if (silence_non_portable) { |
283 | flags |= PERL_SCAN_SILENT_NON_PORTABLE; | |
284 | } | |
80f4111b | 285 | |
00ce5563 | 286 | *uv = grok_hex(*s, &numbers_len, &flags, NULL); |
a0481293 KW |
287 | /* Note that if has non-hex, will ignore everything starting with that up |
288 | * to the '}' */ | |
289 | ||
80f4111b KW |
290 | if (strict && numbers_len != (STRLEN) (e - *s)) { |
291 | *s += numbers_len; | |
292 | *s += (UTF) ? UTF8SKIP(*s) : 1; | |
293 | *error_msg = "Non-hex character"; | |
294 | return FALSE; | |
295 | } | |
296 | ||
00ce5563 KW |
297 | /* Return past the '}' */ |
298 | *s = e + 1; | |
299 | ||
a0481293 KW |
300 | return TRUE; |
301 | } | |
302 | ||
5e0a247b KW |
303 | STATIC char* |
304 | S_form_short_octal_warning(pTHX_ | |
305 | const char * const s, /* Points to first non-octal */ | |
306 | const STRLEN len /* Length of octals string, so | |
307 | (s-len) points to first | |
308 | octal */ | |
309 | ) { | |
310 | /* Return a character string consisting of a warning message for when a | |
311 | * string constant in octal is weird, like "\078". */ | |
312 | ||
313 | const char * sans_leading_zeros = s - len; | |
314 | ||
315 | PERL_ARGS_ASSERT_FORM_SHORT_OCTAL_WARNING; | |
316 | ||
317 | assert(*s == '8' || *s == '9'); | |
318 | ||
319 | /* Remove the leading zeros, retaining one zero so won't be zero length */ | |
320 | while (*sans_leading_zeros == '0') sans_leading_zeros++; | |
321 | if (sans_leading_zeros == s) { | |
322 | sans_leading_zeros--; | |
323 | } | |
324 | ||
325 | return Perl_form(aTHX_ | |
326 | "'%.*s' resolved to '\\o{%.*s}%c'", | |
327 | (int) (len + 2), s - len - 1, | |
328 | (int) (s - sans_leading_zeros), sans_leading_zeros, | |
329 | *s); | |
330 | } | |
331 | ||
04e98a4d AD |
332 | /* |
333 | * Local variables: | |
334 | * c-indentation-style: bsd | |
335 | * c-basic-offset: 4 | |
14d04a33 | 336 | * indent-tabs-mode: nil |
04e98a4d AD |
337 | * End: |
338 | * | |
14d04a33 | 339 | * ex: set ts=8 sts=4 sw=4 et: |
04e98a4d | 340 | */ |