diff options
Diffstat (limited to 'text/test-text.c')
| -rw-r--r-- | text/test-text.c | 32 |
1 files changed, 27 insertions, 5 deletions
diff --git a/text/test-text.c b/text/test-text.c index 9c7af2d..702ea94 100644 --- a/text/test-text.c +++ b/text/test-text.c @@ -9,6 +9,7 @@ #include <stdio.h> #include <stdlib.h> #include <string.h> +#include <unistd.h> enum { MAX = 512 }; static unsigned failures; @@ -127,7 +128,7 @@ static size_t bidi_character_test(const char *path) { return cases; } -/* NormalizationTest.txt (through bzcat): for each line c1;c2;c3;c4;c5, +/* NormalizationTest.txt (as it is, or through bzcat as Debian keeps it): for each line c1;c2;c3;c4;c5, * NFC(c1) = NFC(c2) = NFC(c3) = c2, NFC(c4) = NFC(c5) = c4, NFD(c1) = * NFD(c2) = NFD(c3) = c3 and NFD(c4) = NFD(c5) = c5; and every code point * part 1 does not list is its own NFC and NFD. */ @@ -150,8 +151,10 @@ static bool normalizes(size_t (*form)(const uint32_t *, size_t, uint32_t *), con static size_t normalization_test(const char *path, size_t *invariant) { char command[1200]; + size_t length = strlen(path); + bool compressed = length > 4 && !strcmp(path + length - 4, ".bz2"); snprintf(command, sizeof(command), "bzcat '%s'", path); - FILE *in = popen(command, "r"); + FILE *in = compressed ? popen(command, "r") : fopen(path, "r"); if (!in) { perror(path); exit(2); } static bool listed[0x110000]; char line[4096]; @@ -184,7 +187,8 @@ static size_t normalization_test(const char *path, size_t *invariant) { } if (!ok) fail(path, number, "a form differs"); } - pclose(in); + if (compressed) pclose(in); + else fclose(in); *invariant = 0; for (uint32_t code = 0; code < 0x110000; ++code) { if (listed[code] || (code >= 0xD800 && code <= 0xDFFF)) continue; @@ -277,12 +281,30 @@ static size_t segment_test(const char *path, bool (*segment)(const uint32_t *, s int main(int argc, char **argv) { const char *ucd = argc > 1 ? argv[1] : "/usr/share/unicode"; - char path[1024]; + char path[1024], line[256], version[32] = ""; + /* The tests of another Unicode than the tables' would fail for that alone. */ + snprintf(path, sizeof(path), "%s/ReadMe.txt", ucd); + FILE *readme = fopen(path, "r"); + while (readme && fgets(line, sizeof(line), readme) && !version[0]) + for (const char *at = line; *at && !version[0]; ++at) { /* the first x.y.z, as text/unicode.py reads it */ + unsigned major, minor, update; + int used = 0; + if ((at == line || at[-1] < '0' || at[-1] > '9') && *at >= '0' && *at <= '9' && + sscanf(at, "%u.%u.%u%n", &major, &minor, &update, &used) == 3 && used) + snprintf(version, sizeof(version), "%u.%u.%u", major, minor, update); + } + if (readme) fclose(readme); + if (version[0] && strcmp(version, vega_unicode_version())) { + printf("{\"skipped\":\"the UCD at %s is %s, the tables %s (text/upstream.py fetches theirs)\"}\n", ucd, + version, vega_unicode_version()); + return 0; + } snprintf(path, sizeof(path), "%s/BidiTest.txt", ucd); size_t bidi = bidi_test(path); snprintf(path, sizeof(path), "%s/BidiCharacterTest.txt", ucd); size_t characters = bidi_character_test(path), invariant; - snprintf(path, sizeof(path), "%s/NormalizationTest.txt.bz2", ucd); + snprintf(path, sizeof(path), "%s/NormalizationTest.txt", ucd); + if (access(path, R_OK)) snprintf(path, sizeof(path), "%s/NormalizationTest.txt.bz2", ucd); size_t normalized = normalization_test(path, &invariant); snprintf(path, sizeof(path), "%s/auxiliary/LineBreakTest.txt", ucd); size_t line_breaks = line_break_test(path); |