Support UTF-8 SMS submission using UCS-2

README.md

@@ -3,7 +3,7 @@ SMS Tool for 3G/4G/5G modem
 
 * sms tool for various of 3g/4g modem
 * read sms as raw pdu or decoded text
-* support ucs2 decoding
+* support ucs2 decoding and UTF-8/UCS-2 sending
 
 Usage:
 ----------------

pdu_lib/Makefile

@@ -15,7 +15,11 @@ ucs2_to_utf8:
 pdu_decoder: pdu.o pdu_decoder.o ucs2_to_utf8
 	$(CC) $(CFLAGS) ucs2_to_utf8.o pdu.o pdu_decoder.o -lm -o pdu_decoder
 clean:
-	rm -f *.o pdu_decoder
-test: clean ALL
+	rm -f *.o pdu_decoder pdu_test
+test: clean ALL pdu_test
 	echo "0891683108501405F8240BA10156686616F60008414090912385235C6D4191CF6C4752A860015BC67801FF1A00350030003900360036FF0C4EB2FF0C8BB05F9762BD59566BCF592990FD8981676554E6FF0C611F89C9597D76848BDD63A883507ED960A87684670B53CBFF019884795D60A84E2D5956FF01"|./pdu_decoder
 	echo "0891683108501405F8640BA10156686616F6000841400100957423830608048A3002026B21767B5F556D4191CF6C475373900100356D4191CF5E01FF0853EF63620035004D6D4191CFFF09FF0C731B623394FE63A5FF1A0068007400740070003A002F002F007300680061006B0065002E00730064002E006300680069006E0061006D006F00620069006C0065002E0063006F006D30025C714E1C79FB52A8"|./pdu_decoder
+	./pdu_test
+
+pdu_test: pdu.o pdu_test.o
+	$(CC) $(CFLAGS) pdu.o pdu_test.o -lm -o pdu_test

pdu_lib/pdu.c

@@ -30,6 +30,7 @@ enum {
 	SMS_SUBMIT              = 0x11,
 
 	SMS_MAX_7BIT_TEXT_LENGTH  = 160,
+	SMS_MAX_UCS2_TEXT_LENGTH  = 70,
 };
 
 // Swap decimal digits of a number (e.g. 12 -> 21).
@@ -208,7 +209,7 @@ static const int latin1_to_gsm7bits[256] = {
 };
 
 static int
-AsciiToG7bit(const char* buffer, int buffer_length, unsigned char* output_buffer)
+AsciiToG7bit(const char* buffer, int buffer_length, char* output_buffer)
 {
 	int i, j, val;
 
@@ -247,7 +248,7 @@ EncodePhoneNumber(const char* phone_number, unsigned char* output_buffer, int bu
 	int i = 0;
 	for (; i < phone_number_length; ++i) {
 
-		if (phone_number[i] < '0' && phone_number[i] > '9')
+		if (phone_number[i] < '0' || phone_number[i] > '9')
 			return -1;
 
 		if (i % 2 == 0) {
@@ -262,6 +263,67 @@ EncodePhoneNumber(const char* phone_number, unsigned char* output_buffer, int bu
 	return output_buffer_length;
 }
 
+/*
+ * Convert a valid UTF-8 string to UCS-2 (UTF-16BE without surrogate pairs).
+ * SMS UCS-2 user data is limited to 140 octets for a single message.
+ */
+static int
+Utf8ToUcs2(const char* input, unsigned char* output_buffer, int buffer_size)
+{
+	const unsigned char* p = (const unsigned char*)input;
+	int output_buffer_length = 0;
+
+	while (*p) {
+		unsigned int codepoint;
+
+		if (*p < 0x80) {
+			codepoint = *p++;
+		} else if (*p >= 0xC2 && *p <= 0xDF) {
+			if ((p[1] & 0xC0) != 0x80)
+				return -1;
+			codepoint = ((*p & 0x1F) << 6) | (p[1] & 0x3F);
+			p += 2;
+		} else if (*p >= 0xE0 && *p <= 0xEF) {
+			if ((p[1] & 0xC0) != 0x80 || (p[2] & 0xC0) != 0x80)
+				return -1;
+			if ((*p == 0xE0 && p[1] < 0xA0) ||
+			    (*p == 0xED && p[1] >= 0xA0))
+				return -1;
+			codepoint = ((*p & 0x0F) << 12) |
+				    ((p[1] & 0x3F) << 6) | (p[2] & 0x3F);
+			p += 3;
+		} else {
+			/* UCS-2 cannot represent four-byte UTF-8 code points. */
+			return -1;
+		}
+
+		if (output_buffer_length + 2 > buffer_size)
+			return -1;
+		output_buffer[output_buffer_length++] = codepoint >> 8;
+		output_buffer[output_buffer_length++] = codepoint & 0xFF;
+	}
+
+	return output_buffer_length;
+}
+
+static int
+NormalizePhoneNumber(const char* phone_number, const char** digits, int* international)
+{
+	if (!phone_number || !*phone_number)
+		return -1;
+
+	*international = phone_number[0] == '+';
+	*digits = phone_number + *international;
+	if (!**digits)
+		return -1;
+
+	const size_t length = strlen(*digits);
+	if (length > 255)
+		return -1;
+
+	return length;
+}
+
 // Decode a digit based phone number for SMS based format.
 static int
 DecodePhoneNumber(const unsigned char* buffer, int phone_number_length, char* output_phone_number)
@@ -282,18 +344,23 @@ int
 pdu_encode(const char* service_center_number, const char* phone_number, const char* sms_text,
 	   unsigned char* output_buffer, int buffer_size)
 {	
-	if (buffer_size < 2)
+	if (!phone_number || !sms_text || !output_buffer || buffer_size < 2)
 		return -1;
 
 	int output_buffer_length = 0;
 
 	// 1. Set SMS center number.
 	int length = 0;
 	if (service_center_number && strlen(service_center_number) > 0) {
+		const char* service_center_digits;
+		int service_center_international;
+		if (NormalizePhoneNumber(service_center_number, &service_center_digits,
+					 &service_center_international) < 0)
+			return -1;
 		output_buffer[1] = TYPE_OF_ADDRESS_INTERNATIONAL_PHONE;
-		length = EncodePhoneNumber(service_center_number,
+		length = EncodePhoneNumber(service_center_digits,
 					   output_buffer + 2, buffer_size - 2);
-		if (length < 0 && length >= 254)
+		if (length < 0 || length >= 254)
 			return -1;
 		length++;  // Add type of address.
 	}
@@ -307,29 +374,60 @@ pdu_encode(const char* service_center_number, const char* phone_number, const ch
 	output_buffer[output_buffer_length++] = 0x00;  // Message reference.
 
 	// 3. Set phone number.
-	output_buffer[output_buffer_length] = strlen(phone_number);
+	const char* phone_number_digits;
+	int phone_number_international;
+	const int phone_number_length = NormalizePhoneNumber(phone_number,
+							    &phone_number_digits,
+							    &phone_number_international);
+	if (phone_number_length < 0)
+		return -1;
+	output_buffer[output_buffer_length] = phone_number_length;
 
-	if (strlen(phone_number) < 6) {
+	if (!phone_number_international && phone_number_length < 6) {
 		output_buffer[output_buffer_length + 1] = TYPE_OF_ADDRESS_UNKNOWN;
 	} else {
 		output_buffer[output_buffer_length + 1] = TYPE_OF_ADDRESS_INTERNATIONAL_PHONE;
 	}
 
-	length = EncodePhoneNumber(phone_number,
+	length = EncodePhoneNumber(phone_number_digits,
 				   output_buffer + output_buffer_length + 2,
 				   buffer_size - output_buffer_length - 2);
+	if (length < 0)
+		return -1;
 	output_buffer_length += length + 2;
 	if (output_buffer_length + 4 > buffer_size)
 		return -1;  // Check if it has space for four more bytes.
 
 
+	/* Use UCS-2 whenever the UTF-8 input is not pure ASCII. */
+	int use_ucs2 = 0;
+	for (const unsigned char* p = (const unsigned char*)sms_text; *p; ++p) {
+		if (*p & 0x80) {
+			use_ucs2 = 1;
+			break;
+		}
+	}
+
 	// 4. Protocol identifiers.
 	output_buffer[output_buffer_length++] = 0x00;  // TP-PID: Protocol identifier.
-	output_buffer[output_buffer_length++] = 0x00;  // TP-DCS: Data coding scheme.
+	output_buffer[output_buffer_length++] = use_ucs2 ? 0x08 : 0x00;
 	output_buffer[output_buffer_length++] = 0xB0;  // TP-VP: Validity: 10 days
 
 	// 5. SMS message.
 	int sms_text_length = strlen(sms_text);
+	if (use_ucs2) {
+		unsigned char sms_text_ucs2[2 * SMS_MAX_UCS2_TEXT_LENGTH];
+		sms_text_length = Utf8ToUcs2(sms_text, sms_text_ucs2,
+					      sizeof(sms_text_ucs2));
+		if (sms_text_length < 0 ||
+		    output_buffer_length + 1 + sms_text_length > buffer_size)
+			return -1;
+		output_buffer[output_buffer_length++] = sms_text_length;
+		memcpy(output_buffer + output_buffer_length, sms_text_ucs2,
+		       sms_text_length);
+		return output_buffer_length + sms_text_length;
+	}
+
 	char sms_text_7bit[2*SMS_MAX_7BIT_TEXT_LENGTH];
 	sms_text_length = AsciiToG7bit(sms_text, sms_text_length, sms_text_7bit);
 	if (sms_text_length > SMS_MAX_7BIT_TEXT_LENGTH)
@@ -445,4 +543,3 @@ int pdu_decode(const unsigned char* buffer, int buffer_length,
 
 	return output_sms_text_length;
 }
-

pdu_lib/pdu_test.c

@@ -0,0 +1,64 @@
+#include "pdu.h"
+
+#include <stdio.h>
+#include <string.h>
+
+static int expect_pdu(const char* name, const char* number, const char* text,
+		      const unsigned char* expected, size_t expected_length)
+{
+	unsigned char actual[SMS_MAX_PDU_LENGTH];
+	int actual_length = pdu_encode("", number, text, actual, sizeof(actual));
+
+	if (actual_length != (int)expected_length ||
+	    memcmp(actual, expected, expected_length) != 0) {
+		fprintf(stderr, "%s: encoded PDU does not match\n", name);
+		return 1;
+	}
+	return 0;
+}
+
+int main(void)
+{
+	static const unsigned char ascii_pdu[] = {
+		0x00, 0x11, 0x00, 0x0A, 0x91, 0x21, 0x43, 0x65, 0x87, 0x09,
+		0x00, 0x00, 0xB0, 0x05, 0xE8, 0x32, 0x9B, 0xFD, 0x06,
+	};
+	static const unsigned char ucs2_pdu[] = {
+		0x00, 0x11, 0x00, 0x0B, 0x91, 0x97, 0x50, 0x25, 0x00, 0x53,
+		0xF3, 0x00, 0x08, 0xB0, 0x2C, 0x04, 0x14, 0x04, 0x38, 0x04,
+		0x3C, 0x04, 0x30, 0x00, 0x2C, 0x00, 0x20, 0x04, 0x3F, 0x04,
+		0x38, 0x04, 0x48, 0x04, 0x35, 0x04, 0x42, 0x00, 0x20, 0x04,
+		0x42, 0x04, 0x35, 0x04, 0x31, 0x04, 0x35, 0x00, 0x20, 0x00,
+		0x4E, 0x00, 0x4C, 0x00, 0x36, 0x00, 0x37, 0x00, 0x38,
+	};
+	unsigned char pdu[SMS_MAX_PDU_LENGTH];
+	char long_ucs2[71 * 2 + 1];
+	int failed = 0;
+
+	failed |= expect_pdu("ASCII", "+1234567890", "hello",
+			     ascii_pdu, sizeof(ascii_pdu));
+	failed |= expect_pdu("UCS-2", "+79055200353",
+			     "Дима, пишет тебе NL678",
+			     ucs2_pdu, sizeof(ucs2_pdu));
+
+	if (pdu_encode("", "+12x34", "test", pdu, sizeof(pdu)) >= 0) {
+		fprintf(stderr, "invalid destination was accepted\n");
+		failed = 1;
+	}
+	if (pdu_encode("", "+1234", "\xF0\x9F\x98\x80", pdu,
+		       sizeof(pdu)) >= 0) {
+		fprintf(stderr, "non-UCS-2 code point was accepted\n");
+		failed = 1;
+	}
+	for (int i = 0; i < 71; ++i) {
+		long_ucs2[2 * i] = (char)0xC2;
+		long_ucs2[2 * i + 1] = (char)0xA3;
+	}
+	long_ucs2[sizeof(long_ucs2) - 1] = '\0';
+	if (pdu_encode("", "+1234", long_ucs2, pdu, sizeof(pdu)) >= 0) {
+		fprintf(stderr, "overlong UCS-2 message was accepted\n");
+		failed = 1;
+	}
+
+	return failed;
+}

sms_main.c

@@ -194,8 +194,6 @@ int main(int argc, char* argv[])
 	{
 		if(argc < 3)
 			usage();
-		if(strlen(argv[2]) > 160)
-			fprintf(stderr,"sms message too long: '%s'\n", argv[2]);
 	}else if (!strcmp("delete",argv[0]))
 	{
 		if(argc < 2)
@@ -222,20 +220,26 @@ int main(int argc, char* argv[])
 	// open the port
 
 	port = open(dev, O_RDWR|O_NONBLOCK|O_NOCTTY);
-	if (port < 0)
-		fprintf(stderr,"open(%s)\n", dev);
+	if (port < 0) {
+		fprintf(stderr, "open(%s): %s\n", dev, strerror(errno));
+		return 1;
+	}
 	setserial(baudrate);
 	atexit(resetserial);
 
 	close(port);
 	port = open(dev, O_RDWR|O_NOCTTY);
-	if (port < 0)
-		fprintf(stderr,"reopen(%s)\n", dev);
+	if (port < 0) {
+		fprintf(stderr, "reopen(%s): %s\n", dev, strerror(errno));
+		return 1;
+	}
 
 	FILE* pf = fdopen(port, "w");
 	FILE* pfi = fdopen(port, "r");
-	if (!pf || ! pfi)
-		fprintf(stderr,"open port failed\n");
+	if (!pf || !pfi) {
+		fprintf(stderr, "fdopen(%s): %s\n", dev, strerror(errno));
+		return 1;
+	}
 	if(setvbuf(pf, NULL, _IOLBF, 0))
 	{
 		fprintf(stderr, "failed to make serial port linebuffered\n");
@@ -245,8 +249,11 @@ int main(int argc, char* argv[])
 	if (!strcmp("send", argv[0]))
 	{
 		int pdu_len = pdu_encode("", argv[1], argv[2], pdu, sizeof(pdu));
-		if (pdu_len < 0)
-			fprintf(stderr,"error encoding to PDU: %s \"%s\n", argv[1], argv[2]);
+		if (pdu_len < 0) {
+			fprintf(stderr, "error encoding to PDU: %s \"%s\"\n",
+				argv[1], argv[2]);
+			return 1;
+		}
 
 		const int pdu_len_except_smsc = pdu_len - 1 - pdu[0];
 		snprintf(cmdstr, sizeof(cmdstr), "AT+CMGS=%d\r\n", pdu_len_except_smsc);
@@ -256,36 +263,51 @@ int main(int argc, char* argv[])
 			sprintf(pdustr+2*i, "%02X", pdu[i]);
 		sprintf(pdustr+2*i, "%c\r\n", 0x1A);   // End PDU mode with Ctrl-Z.
 
-		fputs("AT+CMGF=0\r\n", pf);
+		alarm(30);
+		if (fputs("AT+CMGF=0\r\n", pf) == EOF)
+			return 1;
+		int pdu_mode_ready = 0;
 		while(fgets(buf, sizeof(buf), pfi)) {
-			if(starts_with("OK", buf))
+			if(starts_with("OK", buf)) {
+				pdu_mode_ready = 1;
 				break;
+			}
+			if(starts_with("ERROR", buf) || starts_with("+CMS ERROR:", buf)) {
+				fprintf(stderr, "failed to enable PDU mode: %s", buf);
+				return 1;
+			}
 		}
-		fputs(cmdstr, pf);
+		if (!pdu_mode_ready) {
+			fprintf(stderr, "no response while enabling PDU mode\n");
+			return 1;
+		}
+		if (fputs(cmdstr, pf) == EOF)
+			return 1;
 		sleep(1);
-		fputs(pdustr, pf);
+		if (fputs(pdustr, pf) == EOF)
+			return 1;
 
-		alarm(5);
 		errno = 0;
 
 		while(fgets(buf, sizeof(buf), pfi))
 		{
 			if(starts_with("+CMGS:", buf))
 			{
-				printf("sms sent sucessfully: %s", buf + 7);
+				alarm(0);
+				printf("sms sent successfully: %s", buf + 7);
 				return 0;
 			} else if(starts_with("+CMS ERROR:", buf))
 			{
 				fprintf(stderr,"sms not sent, code: %s\n", buf + 11);
+				return 1;
 			} else if(starts_with("ERROR", buf))
 			{
 				fprintf(stderr,"sms not sent, command error\n");
-			} else if(starts_with("OK", buf))
-			{
-				return 0;
+				return 1;
 			}
 		}
-		fprintf(stderr,"reading port\n");
+		fprintf(stderr, "reading port: %s\n", strerror(errno));
+		return 1;
 	}
 
 	if (!strcmp("recv", argv[0]))