aboutsummaryrefslogtreecommitdiffstats
path: root/library/cpp/unicode/punycode/punycode.h
diff options
context:
space:
mode:
authorDevtools Arcadia <arcadia-devtools@yandex-team.ru>2022-02-07 18:08:42 +0300
committerDevtools Arcadia <arcadia-devtools@mous.vla.yp-c.yandex.net>2022-02-07 18:08:42 +0300
commit1110808a9d39d4b808aef724c861a2e1a38d2a69 (patch)
treee26c9fed0de5d9873cce7e00bc214573dc2195b7 /library/cpp/unicode/punycode/punycode.h
downloadydb-1110808a9d39d4b808aef724c861a2e1a38d2a69.tar.gz
intermediate changes
ref:cde9a383711a11544ce7e107a78147fb96cc4029
Diffstat (limited to 'library/cpp/unicode/punycode/punycode.h')
-rw-r--r--library/cpp/unicode/punycode/punycode.h46
1 files changed, 46 insertions, 0 deletions
diff --git a/library/cpp/unicode/punycode/punycode.h b/library/cpp/unicode/punycode/punycode.h
new file mode 100644
index 0000000000..af4acc25c1
--- /dev/null
+++ b/library/cpp/unicode/punycode/punycode.h
@@ -0,0 +1,46 @@
+#pragma once
+
+#include <util/generic/string.h>
+#include <util/generic/strbuf.h>
+#include <util/generic/yexception.h>
+
+// Simplified arcadia wrappers for contrib/libs/libidn/
+
+// Raw strings encoder/decoder: does not prepend with ACE prefix ("xn--"),
+// does not limit input length. Throws TPunycodeError on any internal error.
+// Returned strbuf points to @out data.
+TStringBuf WideToPunycode(const TWtringBuf& in, TString& out);
+TWtringBuf PunycodeToWide(const TStringBuf& in, TUtf16String& out);
+
+inline TString WideToPunycode(const TWtringBuf& in) {
+ TString out;
+ WideToPunycode(in, out);
+ return out;
+}
+
+inline TUtf16String PunycodeToWide(const TStringBuf& in) {
+ TUtf16String out;
+ PunycodeToWide(in, out);
+ return out;
+}
+
+// Encode a sequence of point-separated domain labels
+// into a sequence of corresponding punycode labels.
+// Labels containing non-ASCII characters are prefixed with ACE prefix ("xn--").
+// Limits maximal encoded domain label length to IDNA_LABEL_MAX_LENGTH (255 by default).
+// Throws TPunycodeError on failure.
+TString HostNameToPunycode(const TWtringBuf& unicodeHost);
+TUtf16String PunycodeToHostName(const TStringBuf& punycodeHost);
+
+// Robust versions: on failure return original input, converted to/from UTF8
+TString ForceHostNameToPunycode(const TWtringBuf& unicodeHost);
+TUtf16String ForcePunycodeToHostName(const TStringBuf& punycodeHost);
+
+// True if @host looks like punycode domain label sequence,
+// containing at least one ACE-prefixed label.
+// Note that this function does not check all requied IDNA constraints
+// (max label length, empty non-root domains, etc.)
+bool CanBePunycodeHostName(const TStringBuf& host);
+
+class TPunycodeError: public yexception {
+};