encoding.c: Don't crash if NULL is passed as the string to the decode APIs.

src/libespeak-ng/encoding.c

@@ -615,6 +615,12 @@ string_decoder_getc_auto(espeak_ng_TEXT_DECODER *decoder)
 	return c;
 }
 
+static uint32_t
+null_decoder_getc(espeak_ng_TEXT_DECODER *decoder)
+{
+	return 0;
+}
+
 typedef struct
 {
 	uint32_t (*get)(espeak_ng_TEXT_DECODER *decoder);
@@ -678,10 +684,10 @@ text_decoder_decode_string(espeak_ng_TEXT_DECODER *decoder,
 	if (enc->get == NULL)
 		return ENS_UNKNOWN_TEXT_ENCODING;
 
-	decoder->get = enc->get;
+	decoder->get = string ? enc->get : null_decoder_getc;
 	decoder->codepage = enc->codepage;
 	decoder->current = (const uint8_t *)string;
-	decoder->end = (const uint8_t *)(string + length);
+	decoder->end = (const uint8_t *)(string ? string + length : string);
 	return ENS_OK;
 }
 
@@ -698,10 +704,10 @@ text_decoder_decode_string_auto(espeak_ng_TEXT_DECODER *decoder,
 	if (enc->get == NULL)
 		return ENS_UNKNOWN_TEXT_ENCODING;
 
-	decoder->get = string_decoder_getc_auto;
+	decoder->get = string ? string_decoder_getc_auto : null_decoder_getc;
 	decoder->codepage = enc->codepage;
 	decoder->current = (const uint8_t *)string;
-	decoder->end = (const uint8_t *)(string + length);
+	decoder->end = (const uint8_t *)(string ? string + length : string);
 	return ENS_OK;
 }
 
@@ -710,10 +716,10 @@ text_decoder_decode_wstring(espeak_ng_TEXT_DECODER *decoder,
                             const wchar_t *string,
                             int length)
 {
-	decoder->get = string_decoder_getc_wchar;
+	decoder->get = string ? string_decoder_getc_wchar : null_decoder_getc;
 	decoder->codepage = NULL;
 	decoder->current = (const uint8_t *)string;
-	decoder->end = (const uint8_t *)(string + length);
+	decoder->end = (const uint8_t *)(string ? string + length : string);
 	return ENS_OK;
 }
 

tests/encoding.c

@@ -728,6 +728,22 @@ test_iso_10646_ucs_2_encoding()
 	destroy_text_decoder(decoder);
 }
 
+void
+test_char_decoder()
+{
+	printf("testing char decoder\n");
+
+	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
+
+	// null string
+	assert(text_decoder_decode_string(decoder, NULL, 5, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
+	assert(text_decoder_eof(decoder) == 1);
+	assert(text_decoder_getc(decoder) == 0);
+	assert(text_decoder_eof(decoder) == 1);
+
+	destroy_text_decoder(decoder);
+}
+
 void
 test_wchar_decoder()
 {
@@ -735,6 +751,12 @@ test_wchar_decoder()
 
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
+	// null string
+	assert(text_decoder_decode_wstring(decoder, NULL, 5) == ENS_OK);
+	assert(text_decoder_eof(decoder) == 1);
+	assert(text_decoder_getc(decoder) == 0);
+	assert(text_decoder_eof(decoder) == 1);
+
 	assert(text_decoder_decode_wstring(decoder, L"aG\xA0\x2045", 4) == ENS_OK);
 	assert(text_decoder_eof(decoder) == 0);
 	assert(text_decoder_getc(decoder) == 'a');
@@ -756,6 +778,12 @@ test_auto_decoder()
 
 	espeak_ng_TEXT_DECODER *decoder = create_text_decoder();
 
+	// null string
+	assert(text_decoder_decode_string_auto(decoder, NULL, 5, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
+	assert(text_decoder_eof(decoder) == 1);
+	assert(text_decoder_getc(decoder) == 0);
+	assert(text_decoder_eof(decoder) == 1);
+
 	// UTF-8
 	assert(text_decoder_decode_string_auto(decoder, "aG\xC2\xA0 ", 5, ESPEAKNG_ENCODING_ISO_8859_1) == ENS_OK);
 	assert(text_decoder_eof(decoder) == 0);
@@ -813,6 +841,7 @@ main(int argc, char **argv)
 	test_utf_8_encoding();
 	test_iso_10646_ucs_2_encoding();
 
+	test_char_decoder();
 	test_wchar_decoder();
 	test_auto_decoder();