From 9f5d6b1e9dd9c275bd0ff4adb49ad2821a141c15 Mon Sep 17 00:00:00 2001
From: Tim Ruehsen <tim.ruehsen@gmx.de>
Date: Thu, 19 Jun 2014 13:15:31 +0200
Subject: [PATCH] added idn2 punycode generation as fallback for missing libicu

---
 configure.ac |  7 ++++++-
 src/psl2c.c  | 51 ++++++++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 56 insertions(+), 2 deletions(-)

diff --git a/configure.ac b/configure.ac
index 0b0527d..ca49b20 100644
--- a/configure.ac
+++ b/configure.ac
@@ -81,7 +81,7 @@ AC_ARG_WITH(libicu,
          [[#include <unicode/ustring.h>]],
          [[u_strToUTF8(NULL, 0, NULL, NULL, 0, NULL);]])],
        [HAVE_LIBICU=yes; AC_MSG_RESULT([yes]) AC_DEFINE([WITH_LIBICU], [1], [generate PSL data with IDNA2008 UTS#46 punycode])],
-       [LIBS=$OLDLIBS; AC_MSG_WARN([no working ICU unicode library was found])])
+       [LIBS=$OLDLIBS; AC_MSG_ERROR([no working ICU unicode library was found])])
 
 #    AC_SEARCH_LIBS(uidna_close, icuuc,
 #      [HAVE_LIBICU=yes; AC_DEFINE([WITH_LIBICU], [1], [generate PSL data with IDNA2008 UTS#46 punycode])],
@@ -99,6 +99,11 @@ AC_ARG_ENABLE(builtin,
   ], [
     enable_builtin=yes
     AC_DEFINE([WITH_BUILTIN], [1], [compile PSL data into library])
+    AS_IF([test $HAVE_LIBICU != yes],
+      [
+        # Check for idn2 fallback to generate punycode
+        AC_CHECK_PROG(HAVE_IDN2, idn2, yes, AC_MSG_ERROR(Cannot find required tool 'idn2' as fallback.))
+      ])
   ])
 AM_CONDITIONAL([WITH_BUILTIN], [test $enable_builtin = yes])
 
diff --git a/src/psl2c.c b/src/psl2c.c
index a270307..8ec4a3c 100644
--- a/src/psl2c.c
+++ b/src/psl2c.c
@@ -65,7 +65,7 @@ static void _print_psl_entries(FILE *fpout, const _psl_vector_t *v, const char *
 	} while (0);
 #else
 	fprintf(fpout, "/* automatically generated by psl2c (without punycode support) */\n");
-#endif
+#endif /* WITH_LIBICU */
 
 	fprintf(fpout, "static _psl_entry_t %s[] = {\n", varname);
 
@@ -79,6 +79,48 @@ static void _print_psl_entries(FILE *fpout, const _psl_vector_t *v, const char *
 	fprintf(fpout, "};\n");
 }
 
+#ifndef WITH_LIBICU
+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, n;
+
+	/* do not use 'it < v->cur' since v->cur is changed by _vector_add() ! */
+	for (it = 0, n = v->cur; it < n; it++) {
+		_psl_entry_t *e = _vector_get(v, it);
+
+		if (_str_needs_encoding(e->label_buf)) {
+			_psl_entry_t suffix, *suffixp;
+			char lookupname[64] = "";
+
+			/* this is much slower than the libidn2 API but should have no license issues */
+			FILE *pp;
+			char cmd[16 + sizeof(e->label_buf)];
+			snprintf(cmd, sizeof(cmd), "idn2 '%s'", e->label_buf);
+			if ((pp = popen(cmd, "r"))) {
+				if (fscanf(pp, "%63s", lookupname) >= 1 && strcmp(e->label_buf, lookupname)) {
+					/* fprintf(stderr, "idn2 '%s' -> '%s'\n", e->label_buf, lookupname); */
+					_suffix_init(&suffix, lookupname, strlen(lookupname));
+					suffix.wildcard = e->wildcard;
+					suffixp = _vector_get(v, _vector_add(v, &suffix));
+					suffixp->label = suffixp->label_buf; /* set label to changed address */
+				}
+				pclose(pp);
+			} else
+				fprintf(stderr, "Failed to call popen(%s, \"r\")\n", cmd);
+		}
+	}
+
+	_vector_sort(v);
+}
+#endif /* ! WITH_LIBICU */
+
 #endif /* WITH_BUILTIN */
 
 int main(int argc, const char **argv)
@@ -106,6 +148,13 @@ int main(int argc, const char **argv)
 		size_t cmdsize = 16 + strlen(argv[1]);
 		char *cmd = alloca(cmdsize), checksum[64] = "";
 
+#ifndef WITH_LIBICU
+		/* If libicu is not configured, we still need to have punycode in our built-in data. */
+		/* Else the test suite fails. */
+		_add_punycode_if_needed(psl->suffixes);
+		_add_punycode_if_needed(psl->suffix_exceptions);
+#endif
+
 		_print_psl_entries(fpout, psl->suffixes, "suffixes");
 		_print_psl_entries(fpout, psl->suffix_exceptions, "suffix_exceptions");