From 2f6cc15cdb3d64135b29cfd5ee376a5a03ebbee7 Mon Sep 17 00:00:00 2001 From: Nobuyoshi Nakada Date: Sat, 29 Jun 2019 10:10:17 +0900 Subject: Fixed String#grapheme_clusters with wide encodings * string.c (get_reg_grapheme_cluster): make regexp from properly encoded sources fro wide-char encodings. [Bug #15965] * regparse.c (node_extended_grapheme_cluster): suppress false duplicated range warning for the time being. --- regparse.c | 6 ++++++ string.c | 25 +++++++++++++++++++++++-- test/ruby/test_string.rb | 15 +++++++++++---- 3 files changed, 40 insertions(+), 6 deletions(-) diff --git a/regparse.c b/regparse.c index 5e51e39502..574a07e05d 100644 --- a/regparse.c +++ b/regparse.c @@ -5961,6 +5961,10 @@ node_extended_grapheme_cluster(Node** np, ScanEnv* env) if (ONIGENC_MBC_MINLEN(env->enc) > 1) { /* UTF-16/UTF-32 */ BBuf *inverted_buf = NULL; + /* TODO: fix false warning */ + const int dup_not_warned = env->warnings_flag | ~ONIG_SYN_WARN_CC_DUP; + env->warnings_flag |= ONIG_SYN_WARN_CC_DUP; + /* Start with a positive buffer and invert at the end. * Otherwise, adding single-character ranges work the wrong way. */ R_ERR(add_property_to_cc(cc, "Grapheme_Cluster_Break=Control", 0, env)); @@ -5968,6 +5972,8 @@ node_extended_grapheme_cluster(Node** np, ScanEnv* env) R_ERR(add_code_range(&(cc->mbuf), env, 0x000D, 0x000D)); /* LF */ R_ERR(not_code_range_buf(env->enc, cc->mbuf, &inverted_buf, env)); cc->mbuf = inverted_buf; /* TODO: check what to do with buffer before inversion */ + + env->warnings_flag &= dup_not_warned; /* TODO: fix false warning */ } else { R_ERR(add_property_to_cc(cc, "Grapheme_Cluster_Break=Control", 1, env)); diff --git a/string.c b/string.c index 8ac33e3e52..d85f3425bf 100644 --- a/string.c +++ b/string.c @@ -8593,9 +8593,30 @@ get_reg_grapheme_cluster(rb_encoding *enc) reg_grapheme_cluster = reg_grapheme_cluster_utf8; } if (!reg_grapheme_cluster) { - const OnigUChar source[] = "\\X"; + const OnigUChar source_ascii[] = "\\X"; OnigErrorInfo einfo; - int r = onig_new(®_grapheme_cluster, source, source + sizeof(source) - 1, + const OnigUChar *source = source_ascii; + size_t source_len = sizeof(source_ascii) - 1; + switch (encidx) { +#define CHARS_16BE(x) (OnigUChar)((x)>>8), (OnigUChar)(x) +#define CHARS_16LE(x) (OnigUChar)(x), (OnigUChar)((x)>>8) +#define CHARS_32BE(x) CHARS_16BE((x)>>16), CHARS_16BE(x) +#define CHARS_32LE(x) CHARS_16LE(x), CHARS_16LE((x)>>16) +#define CASE_UTF(e) \ + case ENCINDEX_UTF_##e: { \ + static const OnigUChar source_UTF_##e[] = {CHARS_##e('\\'), CHARS_##e('X')}; \ + source = source_UTF_##e; \ + source_len = sizeof(source_UTF_##e); \ + break; \ + } + CASE_UTF(16BE); CASE_UTF(16LE); CASE_UTF(32BE); CASE_UTF(32LE); +#undef CASE_UTF +#undef CHARS_16BE +#undef CHARS_16LE +#undef CHARS_32BE +#undef CHARS_32LE + } + int r = onig_new(®_grapheme_cluster, source, source + source_len, ONIG_OPTION_DEFAULT, enc, OnigDefaultSyntax, &einfo); if (r) { UChar message[ONIG_MAX_ERROR_MESSAGE_LEN]; diff --git a/test/ruby/test_string.rb b/test/ruby/test_string.rb index 8eb806c032..507e067a0d 100644 --- a/test/ruby/test_string.rb +++ b/test/ruby/test_string.rb @@ -1036,13 +1036,20 @@ CODE "\u{1F468 200D 1F393}", "\u{1F46F 200D 2642 FE0F}", "\u{1f469 200d 2764 fe0f 200d 1f469}", - ].each do |g| + ].product([Encoding::UTF_8, *WIDE_ENCODINGS]) do |g, enc| + g = g.encode(enc) assert_equal [g], g.grapheme_clusters - assert_predicate g.dup.taint.grapheme_clusters[0], :tainted? + assert_predicate g.taint.grapheme_clusters[0], :tainted? end - assert_equal ["\u000A", "\u0324"], "\u{a 324}".grapheme_clusters - assert_equal ["\u000D", "\u0324"], "\u{d 324}".grapheme_clusters + [ + "\u{a 324}", + "\u{d 324}", + "abc", + ].product([Encoding::UTF_8, *WIDE_ENCODINGS]) do |g, enc| + g = g.encode(enc) + assert_equal g.chars, g.grapheme_clusters + end assert_equal ["a", "b", "c"], "abc".b.grapheme_clusters if ENUMERATOR_WANTARRAY -- cgit v1.2.3