Add tests for u8title() and fix ‘ς’ bug

4 files changed, 143 insertions, 1 deletions

diff --git a/lib/unicode/string/u8title.c b/lib/unicode/string/u8title.c
index 4545fda..01e9d2e 100644
--- a/lib/unicode/string/u8title.c
+++ b/lib/unicode/string/u8title.c
@@ -68,7 +68,7 @@ u8title(char8_t *restrict dst, size_t dstn, const char8_t *src, size_t srcn,
 		}
 
 		if (final_sigma.after == 0) {
-			rune ch;
+			rune ch = 0;
 			struct u8view cpy = {src, srcn};
 
 			do
diff --git a/test/data/TitlecaseTest b/test/data/TitlecaseTest
new file mode 100644
index 0000000..24256a5
--- /dev/null
+++ b/test/data/TitlecaseTest
@@ -0,0 +1,48 @@
+# Empty input
+;;
+
+# Short input
+a;A;
+
+# If CF_ẞ gets passed for whatever reason… don’t turn into ẞ
+ß;Ss;ẞ
+
+# Latin alphabet
+LOREM IPSUM DOLOR SIT AMET, CONSECTETUR ADIPISCING ELIT.;Lorem Ipsum Dolor Sit Amet, Consectetur Adipiscing Elit.;
+lorem ipsum dolor sit amet, consectetur adipiscing elit.;Lorem Ipsum Dolor Sit Amet, Consectetur Adipiscing Elit.;
+
+# Random punctuation and numbers
+COMPLEX-LANGUAGE AND -SCRIPT;Complex-Language And -Script;
+complex-language and -script;Complex-Language And -Script;
+
+# Greek alphabet; handle sigma properly
+ΤΟ ΓΡΆΜΜΑ ΣΊΓΜΑ ΈΧΕΙ ΔΎΟ ΠΕΖΟΎΣ ΤΎΠΟΥΣ;Το Γράμμα Σίγμα Έχει Δύο Πεζούς Τύπους;
+το γράμμα σίγμα έχει δύο πεζούς τύπους;Το Γράμμα Σίγμα Έχει Δύο Πεζούς Τύπους;
+
+# Cyrillic alphabet
+СЛАВА УКРАЇНІ ПРОТИ РОСІЙСЬКОЇ АГРЕСІЇ!;Слава Україні Проти Російської Агресії!;
+слава україні проти російської агресії!;Слава Україні Проти Російської Агресії!;
+
+# In lithuanian we need to retain the dot above ‘i’ and ‘j’ when there’s an
+# accent above the uppercased variant.  Also test with both single-codepoint
+# variants (i.e. U+00CC LATIN CAPITAL I WITH GRAVE) and variants that use
+# combining-characters.
+i̇̀;İ̀;
+i̇̀;Ì;LT
+RÀSTI, MÈSTI, KÌLO;Ràsti, Mèsti, Kìlo;
+RÀSTI, MÈSTI, KÌLO;Ràsti, Mèsti, Ki̇̀lo;LT
+
+# Croatian has 3 cases
+LJUDEVIT GAJ;Ljudevit Gaj;
+ljudevit gaj;Ljudevit Gaj;
+
+# Dutch IJ needs special handling
+ijsberg en onderzeeër in de ijssel;Ijsberg En Onderzeeër In De Ijssel;
+ijsberg en onderzeeër in de ijssel;IJsberg En Onderzeeër In De IJssel;NL
+
+# Uppercase ‘i’ to ‘İ’ in Azeri/Turkish
+istanbul’luyum;Istanbul’luyum;
+istanbul’luyum;İstanbul’luyum;AZ
+
+# Uncased language
+안녕하세요, 월드!;안녕하세요, 월드!;
diff --git a/test/run-tests b/test/run-tests
index 1d6cf56..f191c33 100755
--- a/test/run-tests
+++ b/test/run-tests
@@ -24,6 +24,7 @@ download 'auxiliary/GraphemeBreakTest.txt'
 download 'auxiliary/WordBreakTest.txt'
 
 grep '^[^#]'                         data/LowercaseTest     >lower.in
+grep '^[^#]'                         data/TitlecaseTest     >title.in
 grep '^[^#]'                         data/UppercaseTest     >upper.in
 sed -En 's/\s+//g; s/÷?#.*//g; /./p' data/GraphemeBreakTest >gnext.in
 sed -En 's/\s+//g; s/÷?#.*//g; /./p' data/WordBreakTest     >wnext.in
diff --git a/test/title-test.c b/test/title-test.c
new file mode 100644
index 0000000..bea1ce7
--- /dev/null
+++ b/test/title-test.c
@@ -0,0 +1,93 @@
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include <alloc.h>
+#include <errors.h>
+#include <macros.h>
+#include <mbstring.h>
+#include <unicode/string.h>
+
+#define TESTFILE "title.in"
+
+static bool test(const char8_t *, int);
+
+int
+main(int, char **argv)
+{
+	int rv;
+	size_t n;
+	ssize_t nr;
+	char *line;
+	FILE *fp;
+
+	rv = EXIT_SUCCESS;
+	line = nullptr;
+	mlib_setprogname(argv[0]);
+
+	if ((fp = fopen(TESTFILE, "r")) == nullptr)
+		err("fopen: %s:", TESTFILE);
+
+	for (int id = 1; (nr = getline(&line, &n, fp)) > 0; id++) {
+		if (line[nr - 1] == '\n')
+			line[--nr] = '\0';
+
+		if (!test(line, id))
+			rv = EXIT_FAILURE;
+	}
+	if (ferror(fp))
+		err("getline: %s:", TESTFILE);
+
+	free(line);
+	fclose(fp);
+	return rv;
+}
+
+bool
+test(const char8_t *line, int id)
+{
+	struct u8view before, after, flags;
+	before.p = line;
+	after.p = strchr(line, ';') + 1;
+	before.len = after.p - before.p - 1;
+	flags.p = strchr(after.p, ';') + 1;
+	after.len = flags.p - after.p - 1;
+	flags.len = strlen(flags.p);
+
+	enum caseflags cf = 0;
+	if (u8eq(U8_ARGS(flags), U8_ARGS(U8("ẞ"))))
+		cf |= CF_ẞ;
+	else if (u8eq(U8_ARGS(flags), U8_ARGS(U8("AZ"))))
+		cf |= CF_LANG_AZ;
+	else if (u8eq(U8_ARGS(flags), U8_ARGS(U8("LT"))))
+		cf |= CF_LANG_LT;
+	else if (u8eq(U8_ARGS(flags), U8_ARGS(U8("NL"))))
+		cf |= CF_LANG_NL;
+
+	char8_t *buf = bufalloc(nullptr, 1, after.len);
+	size_t bufsz = u8title(nullptr, 0, U8_ARGS(before), cf);
+	if (bufsz != after.len) {
+		warn("case %d: expected titlecased buffer size of %zu but got %zu "
+		     "(flags=‘%.*s’)",
+		     id, after.len, bufsz, U8_PRI_ARGS(flags));
+		return false;
+	}
+
+	bufsz = u8title(buf, bufsz, U8_ARGS(before), cf);
+	if (bufsz != after.len) {
+		warn("case %d: expected titlecased length of %zu but got %zu "
+		     "(flags=‘%.*s’)",
+		     id, after.len, bufsz, U8_PRI_ARGS(flags));
+		return false;
+	}
+
+	if (!memeq(buf, after.p, bufsz)) {
+		warn("case %d: expected ‘%.*s’ but got ‘%.*s’ (flags=‘%.*s’)", id,
+		     U8_PRI_ARGS(after), (int)bufsz, buf, U8_PRI_ARGS(flags));
+		return false;
+	}
+
+	free(buf);
+	return true;
+}