added IDNA2008 punycode support for psl_inline_is_public()

src/Makefile.am

@@ -21,8 +21,8 @@ libpsl_inline_@LIBPSL_API_VERSION@_la_LDFLAGS = -version-info $(LIBPSL_SO_VERSIO
 
 noinst_PROGRAMS = psl2c
 psl2c_SOURCES = psl2c.c
-psl2c_CPPFLAGS = -I$(top_srcdir)/include
+psl2c_CPPFLAGS = -I$(top_srcdir)/include -D _GNU_SOURCE
-#psl2c_LDADD = libpsl-@LIBPSL_API_VERSION@.la
+psl2c_LDADD = -lidn2
 #psl2c_LDFLAGS = -static
 
 # Build rule for suffix.c

src/psl-inline.c

@@ -44,7 +44,7 @@
 
 typedef struct {
 	char
-		label_buf[42];
+		label_buf[48];
 	const char *
 		label;
 	unsigned short
@@ -102,15 +102,15 @@ int psl_inline_is_public(const char *domain)
 		if (*p == '.')
 			suffix.nlabels++;
 
-	// if domain has enough labels, it won't match
+	// if domain has enough labels, it is public
 	rule = &suffixes[0];
 	if (!rule || rule->nlabels < suffix.nlabels - 1)
-		return 0;
+		return 1;
 
 	rule = bsearch(&suffix, suffixes, countof(suffixes), sizeof(suffixes[0]), (int(*)(const void *, const void *))_suffix_compare);
 	if (rule) {
 		// definitely a match, no matter if the found rule is a wildcard or not
-		return 1;
+		return 0;
 	}
 
 	label_bak = suffix.label;
@@ -130,14 +130,14 @@ int psl_inline_is_public(const char *domain)
 				suffix.nlabels++;
 
 				if (bsearch(&suffix, suffix_exceptions, countof(suffix_exceptions), sizeof(suffix_exceptions[0]), (int(*)(const void *, const void *))_suffix_compare))
-					return 0; // found an exception, so 'domain' is not a public suffix
+					return 1; // found an exception, so 'domain' is public
-
-				return 1;
-			}
-		}
-	}
 
 				return 0;
+			}
+		}
+	}
+
+	return 1;
 }
 
 /* does not include exceptions */

src/psl.c

@@ -44,7 +44,7 @@
 
 typedef struct {
 	char
-		label_buf[42];
+		label_buf[48];
 	const char *
 		label;
 	unsigned short
@@ -187,7 +187,7 @@ static void _suffix_init(_psl_entry_t *suffix, const char *rule, size_t length)
 
 	if (length >= sizeof(suffix->label_buf) - 1) {
 		suffix->nlabels = 0;
-		fprintf(stderr, _("Suffix rule too long (ignored): %s\n"), rule);
+		fprintf(stderr, _("Suffix rule too long (%zd, ignored): %s\n"), length, rule);
 		return;
 	}
 

src/psl2c.c

@@ -28,7 +28,10 @@
 # include <config.h>
 #endif
 
-// # include <idn2.h>
+//#ifdef WITH_LIBIDN2
+#	include <idn2.h>
+//#endif
+
 #include "psl.c"
 
 static void _print_psl_entries(_psl_vector_t *v, const char *varname)
@@ -43,22 +46,44 @@ static void _print_psl_entries(_psl_vector_t *v, const char *varname)
 
 		printf("\t{ \"%s\", NULL, %hd, %hhd, %hhd },\n",
 			e->label_buf, e->length, e->nlabels, e->wildcard);
-/*
-		if (str_needs_encoding(e->label_buf)) {
-			char *asc = NULL;
-			int rc;
-
-			if ((rc = idn2_lookup_u8((uint8_t *)e->label_buf, (uint8_t **)&asc, 0)) == IDN2_OK) {
-				fprintf(stderr, "idn2 '%s' -> '%s'\n", e->label_buf, asc);
-			} else
-				fprintf(stderr, "toASCII(%s) failed (%d): %s\n", e->label_buf, rc, idn2_strerror(rc));
-		}
-*/
 	}
 
 	printf("};\n");
 }
 
+static int _str_needs_encoding(const char *s)
+{
+	while (*s > 0) s++;
+
+	return !!*s;
+}
+
+static void _add_punycode_if_needed(_psl_vector_t *v)
+{
+	int it;
+
+	for (it = 0; it < v->cur; it++) {
+		_psl_entry_t *e = _vector_get(v, it);
+
+		if (_str_needs_encoding(e->label_buf)) {
+			_psl_entry_t suffix;
+			char *asc = NULL;
+			int rc;
+
+
+			if ((rc = idn2_lookup_u8((uint8_t *)e->label_buf, (uint8_t **)&asc, 0)) == IDN2_OK) {
+				fprintf(stderr, "idn2 '%s' -> '%s'\n", e->label_buf, asc);
+				_suffix_init(&suffix, asc, strlen(asc));
+				suffix.wildcard = e->wildcard;
+				_vector_add(v, &suffix);
+			} else
+				fprintf(stderr, "toASCII(%s) failed (%d): %s\n", e->label_buf, rc, idn2_strerror(rc));
+		}
+	}
+
+	_vector_sort(v);
+}
+
 // int main(int argc, const char **argv)
 int main(void)
 {
@@ -67,6 +92,9 @@ int main(void)
 	if (!(psl = psl_load_fp(stdin)))
 		return 1;
 
+	_add_punycode_if_needed(psl->suffixes);
+	_add_punycode_if_needed(psl->suffix_exceptions);
+
 	_print_psl_entries(psl->suffixes, "suffixes");
 	_print_psl_entries(psl->suffix_exceptions, "suffix_exceptions");
 

tests/test-is-public-inline.c

@@ -62,8 +62,8 @@ static void test_psl(void)
 		{ "www.xxx.ck", 1 },
 		{ "\345\225\206\346\240\207", 0 }, // xn--czr694b oder 商标
 		{ "www.\345\225\206\346\240\207", 1 },
-//		{ "xn--czr694b", 1 },
+		{ "xn--czr694b", 0 },
-//		{ "www.xn--czr694b", 1 },
+		{ "www.xn--czr694b", 1 },
 	};
 	unsigned it;
 
@@ -79,7 +79,7 @@ static void test_psl(void)
 			ok++;
 		} else {
 			failed++;
-			printf("psl_is_tld(%s)=%d (expected %d)\n", t->domain, result, t->result);
+			printf("psl_is_public(%s)=%d (expected %d)\n", t->domain, result, t->result);
 		}
 	}
 

tests/test-is-public.c

@@ -42,6 +42,8 @@ static int
 
 static void test_psl(void)
 {
+	// punycode generation: idn 商标
+	// octal code generation: echo -n "商标" | od -b
 	static const struct test_data {
 		const char
 			*domain;
@@ -58,6 +60,8 @@ static void test_psl(void)
 		{ "abc.www.ck", 1 },
 		{ "xxx.ck", 0 },
 		{ "www.xxx.ck", 1 },
+		{ "\345\225\206\346\240\207", 0 }, // xn--czr694b oder 商标
+		{ "www.\345\225\206\346\240\207", 1 },
 	};
 	unsigned it;
 	psl_ctx_t *psl;
@@ -74,7 +78,7 @@ static void test_psl(void)
 			ok++;
 		} else {
 			failed++;
-			printf("psl_is_tld(%s)=%d (expected %d)\n", t->domain, result, t->result);
+			printf("psl_is_public(%s)=%d (expected %d)\n", t->domain, result, t->result);
 		}
 	}