Add env flag SMS_CP, and transliterate too complex characters.

1 files changed, 17 insertions, 2 deletions

diff --git a/localmarkdown2sms b/localmarkdown2sms
index 896b34f..7293583 100755
--- a/localmarkdown2sms
+++ b/localmarkdown2sms
@@ -28,7 +28,8 @@ use Log::Log4perl qw(:easy);
 use File::Spec;
 use File::Slurp;
 use Time::Duration::Parse;
-use Encode;
+use Text::Unidecode;
+use Encode 2.12 qw(encode decode_utf8 _utf8_off); # need v2.12 to support coderef
 use LWP::UserAgent;
 use URI::Escape;
 use Proc::Daemon;
@@ -41,6 +42,7 @@ my $sms_pw = $ENV{SMS_PW} || "foobar";
 my $sms_phone = $ENV{SMS_PHONE};
 my $sms_smsc = $ENV{SMS_SMSC};
 my $sms_msgtag = $ENV{SMS_MSGTAG} || "text";
+my $sms_cp = $ENV{SMS_CP} || "utf8";
 my $path = $ENV{mdpath};
 
 my (%file, %delay, %reply);
@@ -172,6 +174,19 @@ sub tidymsg {
 	return $msg;
 }
 
+# based on Text::Unidecode bug#8017: http://rt.cpan.org/Ticket/Display.html?id=8017#txn-322351
+sub transliterate {
+	my ($tocharset, $string) = @_;
+
+	my $res = encode($tocharset, decode_utf8($string), sub {
+		my $ascii = unidecode(chr $_[0]);
+		_utf8_off($ascii);
+		$ascii;
+		});
+
+	return $res;
+}
+
 sub sendmsg {
 	my ($phone, $desc, $msg) = @_;
 
@@ -184,7 +199,7 @@ sub sendmsg {
 			. '&to=' . uri_escape($phone);
 		$url .= '&from=' . uri_escape($sms_phone) if ($sms_phone);
 		$url .= '&smsc=' . uri_escape($sms_smsc) if ($sms_smsc);
-		$url .= '&' . $sms_msgtag . '=' . uri_escape(encode("cp1252", $msg));
+		$url .= '&' . $sms_msgtag . '=' . uri_escape(transliterate($sms_cp, $msg));
 		my $response = $ua->request(HTTP::Request->new('GET', $url));
 		unless ($response->is_success) {
 			ERROR $response->status_line;