Regex-specific escaping (#39094)
yash/trunk/NEWS (diff) yash/trunk/NEWS.ja (diff)
yash/trunk/builtins/test.c (diff) yash/trunk/tests/bracket-y.tst (diff)| @@ -28,6 +28,9 @@ | ||
| 28 | 28 | shell is now always implicitly redirected to /dev/null, |
| 29 | 29 | regardless of whether the standard input has already been |
| 30 | 30 | redirected. |
| 31 | + = Quoted characters are now handled in (almost) the same way as | |
| 32 | + Bash in the regular expression in the "[[ word =~ regex ]]" | |
| 33 | + syntax. | |
| 31 | 34 | * The "command" built-in with the -v or -V option was printing |
| 32 | 35 | the pathnames of external commands with a redundant leading slash |
| 33 | 36 | when the current working directory is "/" or "//". |
| @@ -45,7 +48,7 @@ | ||
| 45 | 48 | character. |
| 46 | 49 | * When there are no positional parameters, the nested expansion |
| 47 | 50 | "${{@}}" now expands to nothing rather than one empty field. |
| 48 | - * Unquoted parentheses and vertical bars now can be used in a | |
| 51 | + * Unquoted parentheses and vertical bars now can be used in the | |
| 49 | 52 | regular expression in the "[[ word =~ regex ]]" syntax. |
| 50 | 53 | |
| 51 | 54 | ---------------------------------------------------------------------- |
| @@ -97,6 +97,8 @@ | ||
| 97 | 97 | __attribute__((nonnull)); |
| 98 | 98 | static wchar_t *quote_removal_for_regex(const wchar_t *s, const char *cc) |
| 99 | 99 | __attribute__((nonnull,malloc,warn_unused_result)); |
| 100 | +static const wchar_t *skip_bracket(const wchar_t *s) | |
| 101 | + __attribute__((nonnull,pure,warn_unused_result)); | |
| 100 | 102 | #endif |
| 101 | 103 | |
| 102 | 104 |
| @@ -871,19 +873,78 @@ | ||
| 871 | 873 | * string. */ |
| 872 | 874 | wchar_t *quote_removal_for_regex(const wchar_t *s, const char *cc) |
| 873 | 875 | { |
| 874 | - xwcsbuf_T result; | |
| 875 | - wb_initwithmax(&result, mul(wcslen(s), 2)); | |
| 876 | + /* First, remove quotations. */ | |
| 877 | + xwcsbuf_T tmp; | |
| 878 | + xstrbuf_T tmpcc; | |
| 879 | + size_t sizehint = wcslen(s); | |
| 880 | + wb_initwithmax(&tmp, sizehint); | |
| 881 | + sb_initwithmax(&tmpcc, sizehint); | |
| 876 | 882 | for (size_t i = 0; s[i] != L'\0'; i++) { |
| 877 | 883 | if (cc[i] & CC_QUOTATION) |
| 878 | 884 | continue; |
| 879 | - if (cc[i] & CC_QUOTED) | |
| 880 | - if (wcschr(L"^.[$()|*+?{\\", s[i]) != NULL) | |
| 885 | + wb_wccat(&tmp, s[i]); | |
| 886 | + sb_ccat(&tmpcc, cc[i]); | |
| 887 | + } | |
| 888 | + | |
| 889 | + /* Next, escape unquoted special chars outside brackets */ | |
| 890 | + xwcsbuf_T result; | |
| 891 | + wb_initwithmax(&result, sizehint); | |
| 892 | + for (size_t i = 0; tmp.contents[i] != L'\0'; ) { | |
| 893 | + if (tmpcc.contents[i] & CC_QUOTED) { | |
| 894 | + if (wcschr(L"^.[$()|*+?{\\", tmp.contents[i]) != NULL) | |
| 881 | 895 | wb_wccat(&result, L'\\'); |
| 882 | - wb_wccat(&result, s[i]); | |
| 896 | + wb_wccat(&result, tmp.contents[i++]); | |
| 897 | + } else { | |
| 898 | + if (tmp.contents[i] != L'[') { | |
| 899 | + wb_wccat(&result, tmp.contents[i++]); | |
| 900 | + } else { | |
| 901 | + const wchar_t *s2 = skip_bracket(&tmp.contents[i]); | |
| 902 | + size_t j = s2 - tmp.contents; | |
| 903 | + while (i < j) | |
| 904 | + wb_wccat(&result, tmp.contents[i++]); | |
| 905 | + } | |
| 906 | + } | |
| 883 | 907 | } |
| 908 | + | |
| 909 | + sb_destroy(&tmpcc); | |
| 910 | + wb_destroy(&tmp); | |
| 884 | 911 | return wb_towcs(&result); |
| 885 | 912 | } |
| 886 | 913 | |
| 914 | +/* Skips a bracket expression in a regular expression pattern. | |
| 915 | + * Returns a pointer to the character just after the closing L']' (or the | |
| 916 | + * terminating null character). */ | |
| 917 | +const wchar_t *skip_bracket(const wchar_t *s) | |
| 918 | +{ | |
| 919 | + assert(*s == L'['); | |
| 920 | + s++; | |
| 921 | + | |
| 922 | + if (*s == L'^') | |
| 923 | + s++; | |
| 924 | + if (*s == L']') | |
| 925 | + s++; | |
| 926 | + | |
| 927 | + while (*s != L'\0') { | |
| 928 | + if (*s == L']') | |
| 929 | + return s + 1; | |
| 930 | + if (*s++ != L'[') | |
| 931 | + continue; | |
| 932 | + | |
| 933 | + switch (*s) { | |
| 934 | + case L':': case L'.': case L'=': ; | |
| 935 | + wchar_t end[] = { *s, L']', L'\0', }; | |
| 936 | + s++; | |
| 937 | + const wchar_t *endp = wcsstr(s, end); | |
| 938 | + if (endp == NULL) | |
| 939 | + return s + wcslen(s); | |
| 940 | + s = endp + 2; | |
| 941 | + break; | |
| 942 | + } | |
| 943 | + } | |
| 944 | + | |
| 945 | + return s; | |
| 946 | +} | |
| 947 | + | |
| 887 | 948 | #endif /* YASH_ENABLE_DOUBLE_BRACKET */ |
| 888 | 949 | |
| 889 | 950 |
Fork
magicant