diff options
| author | Tomas Heinrich <heinrich.tomas@gmail.com> | 2010-01-04 16:21:31 +0100 |
|---|---|---|
| committer | Denys Vlasenko <vda.linux@googlemail.com> | 2010-01-04 16:21:31 +0100 |
| commit | d2b1ba6fdee59645763e915ade3ec55e67d5839a (patch) | |
| tree | 9b1a97996255647452851908dc76e4e91a57420b | |
| parent | 4928f3b90b3925e6f3cc234df48e46f88fc5689b (diff) | |
| download | busybox-w32-d2b1ba6fdee59645763e915ade3ec55e67d5839a.tar.gz busybox-w32-d2b1ba6fdee59645763e915ade3ec55e67d5839a.tar.bz2 busybox-w32-d2b1ba6fdee59645763e915ade3ec55e67d5839a.zip | |
[un]expand: unicode support
function old new delta
expand_main 633 663 +30
Signed-off-by: Tomas Heinrich <heinrich.tomas@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| -rw-r--r-- | coreutils/expand.c | 51 | ||||
| -rwxr-xr-x | testsuite/expand.tests | 6 | ||||
| -rwxr-xr-x | testsuite/unexpand.tests | 3 |
3 files changed, 42 insertions, 18 deletions
diff --git a/coreutils/expand.c b/coreutils/expand.c index 7137c3b89..e08eb1d0a 100644 --- a/coreutils/expand.c +++ b/coreutils/expand.c | |||
| @@ -20,8 +20,8 @@ | |||
| 20 | * | 20 | * |
| 21 | * Caveat: this versions of expand and unexpand don't accept tab lists. | 21 | * Caveat: this versions of expand and unexpand don't accept tab lists. |
| 22 | */ | 22 | */ |
| 23 | |||
| 24 | #include "libbb.h" | 23 | #include "libbb.h" |
| 24 | #include "unicode.h" | ||
| 25 | 25 | ||
| 26 | enum { | 26 | enum { |
| 27 | OPT_INITIAL = 1 << 0, | 27 | OPT_INITIAL = 1 << 0, |
| @@ -30,35 +30,37 @@ enum { | |||
| 30 | }; | 30 | }; |
| 31 | 31 | ||
| 32 | #if ENABLE_EXPAND | 32 | #if ENABLE_EXPAND |
| 33 | static void expand(FILE *file, int tab_size, unsigned opt) | 33 | static void expand(FILE *file, unsigned tab_size, unsigned opt) |
| 34 | { | 34 | { |
| 35 | char *line; | 35 | char *line; |
| 36 | 36 | ||
| 37 | tab_size = -tab_size; | ||
| 38 | |||
| 39 | while ((line = xmalloc_fgets(file)) != NULL) { | 37 | while ((line = xmalloc_fgets(file)) != NULL) { |
| 40 | int pos; | ||
| 41 | unsigned char c; | 38 | unsigned char c; |
| 42 | char *ptr = line; | 39 | char *ptr; |
| 40 | char *ptr_strbeg; | ||
| 43 | 41 | ||
| 44 | goto start; | 42 | ptr = ptr_strbeg = line; |
| 45 | while ((c = *ptr) != '\0') { | 43 | while ((c = *ptr) != '\0') { |
| 46 | if ((opt & OPT_INITIAL) && !isblank(c)) { | 44 | if ((opt & OPT_INITIAL) && !isblank(c)) { |
| 47 | fputs(ptr, stdout); | 45 | /* not space or tab */ |
| 48 | break; | 46 | break; |
| 49 | } | 47 | } |
| 50 | ptr++; | ||
| 51 | if (c == '\t') { | 48 | if (c == '\t') { |
| 52 | c = ' '; | 49 | unsigned len; |
| 53 | while (++pos < 0) | 50 | *ptr = '\0'; |
| 54 | bb_putchar(c); | 51 | # if ENABLE_FEATURE_ASSUME_UNICODE |
| 55 | } | 52 | len = bb_mbstrlen(ptr_strbeg); |
| 56 | bb_putchar(c); | 53 | # else |
| 57 | if (++pos >= 0) { | 54 | len = ptr - ptr_strbeg; |
| 58 | start: | 55 | # endif |
| 59 | pos = tab_size; | 56 | len = tab_size - (len % tab_size); |
| 57 | /*while (ptr[1] == '\t') { ptr++; len += tab_size; } - can handle many tabs at once */ | ||
| 58 | printf("%s%*s", ptr_strbeg, len, ""); | ||
| 59 | ptr_strbeg = ptr + 1; | ||
| 60 | } | 60 | } |
| 61 | ptr++; | ||
| 61 | } | 62 | } |
| 63 | fputs(ptr_strbeg, stdout); | ||
| 62 | free(line); | 64 | free(line); |
| 63 | } | 65 | } |
| 64 | } | 66 | } |
| @@ -75,6 +77,7 @@ static void unexpand(FILE *file, unsigned tab_size, unsigned opt) | |||
| 75 | 77 | ||
| 76 | while (*ptr) { | 78 | while (*ptr) { |
| 77 | unsigned n; | 79 | unsigned n; |
| 80 | unsigned len; | ||
| 78 | 81 | ||
| 79 | while (*ptr == ' ') { | 82 | while (*ptr == ' ') { |
| 80 | column++; | 83 | column++; |
| @@ -97,8 +100,19 @@ static void unexpand(FILE *file, unsigned tab_size, unsigned opt) | |||
| 97 | } | 100 | } |
| 98 | n = strcspn(ptr, "\t "); | 101 | n = strcspn(ptr, "\t "); |
| 99 | printf("%*s%.*s", column, "", n, ptr); | 102 | printf("%*s%.*s", column, "", n, ptr); |
| 103 | # if ENABLE_FEATURE_ASSUME_UNICODE | ||
| 104 | { | ||
| 105 | char c; | ||
| 106 | c = ptr[n]; | ||
| 107 | ptr[n] = '\0'; | ||
| 108 | len = bb_mbstrlen(ptr); | ||
| 109 | ptr[n] = c; | ||
| 110 | } | ||
| 111 | # else | ||
| 112 | len = n; | ||
| 113 | # endif | ||
| 100 | ptr += n; | 114 | ptr += n; |
| 101 | column = (column + n) % tab_size; | 115 | column = (column + len) % tab_size; |
| 102 | } | 116 | } |
| 103 | free(line); | 117 | free(line); |
| 104 | } | 118 | } |
| @@ -130,6 +144,7 @@ int expand_main(int argc UNUSED_PARAM, char **argv) | |||
| 130 | "all\0" No_argument "a" | 144 | "all\0" No_argument "a" |
| 131 | ; | 145 | ; |
| 132 | #endif | 146 | #endif |
| 147 | check_unicode_in_env(); | ||
| 133 | 148 | ||
| 134 | if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e')) { | 149 | if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e')) { |
| 135 | IF_FEATURE_EXPAND_LONG_OPTIONS(applet_long_options = expand_longopts); | 150 | IF_FEATURE_EXPAND_LONG_OPTIONS(applet_long_options = expand_longopts); |
diff --git a/testsuite/expand.tests b/testsuite/expand.tests index 57498a201..996631450 100755 --- a/testsuite/expand.tests +++ b/testsuite/expand.tests | |||
| @@ -12,4 +12,10 @@ testing "expand" \ | |||
| 12 | "" \ | 12 | "" \ |
| 13 | "\t12345678\t12345678\n" \ | 13 | "\t12345678\t12345678\n" \ |
| 14 | 14 | ||
| 15 | testing "expand with unicode characher 0x394" \ | ||
| 16 | "expand" \ | ||
| 17 | "Δ 12345ΔΔΔ 12345678\n" \ | ||
| 18 | "" \ | ||
| 19 | "Δ\t12345ΔΔΔ\t12345678\n" \ | ||
| 20 | |||
| 15 | exit $FAILCOUNT | 21 | exit $FAILCOUNT |
diff --git a/testsuite/unexpand.tests b/testsuite/unexpand.tests index 5c82a1e50..5c2a29b5f 100755 --- a/testsuite/unexpand.tests +++ b/testsuite/unexpand.tests | |||
| @@ -27,4 +27,7 @@ testing "unexpand case 6" "unexpand" \ | |||
| 27 | testing "unexpand case 7" "unexpand" \ | 27 | testing "unexpand case 7" "unexpand" \ |
| 28 | "123\t 45678\n" "" "123 \t 45678\n" \ | 28 | "123\t 45678\n" "" "123 \t 45678\n" \ |
| 29 | 29 | ||
| 30 | testing "unexpand with unicode characher 0x394" "unexpand" \ | ||
| 31 | "1ΔΔΔ5\t99999\n" "" "1ΔΔΔ5 99999\n" \ | ||
| 32 | |||
| 30 | exit $FAILCOUNT | 33 | exit $FAILCOUNT |