refuse to transliterate Hindi-specific characters

compat.h

@@ -10,4 +10,8 @@
 #include <getopt.h>
 #include <errno.h>
 
+enum err {
+	EHINDI = 1
+};
+
 #endif /* __COMPAT_H */

main.c

@@ -68,18 +68,42 @@ static void error(const char *msg, ...)
 	va_end(params);
 }
 
-static char *process_input(const char *input, unsigned int flags)
+static int process_input(const char *input, char **out, unsigned int flags)
 {
 	if (flags & FLAG_REVERSE)
-		return transliterate_latin_to_devanagari(input);
+		return transliterate_latin_to_devanagari(input, out);
 
 	if (flags & FLAG_VELTHUIS)
-		return encode_velthuis_to_iast_punctation(input);
+		return encode_velthuis_to_iast_punctation(input, out);
 
 	if (flags & FLAG_CZECH)
-		return transcript_devanagari_to_czech(input);
+		return transcript_devanagari_to_czech(input, out);
 
-	return transliterate_devanagari_to_latin(input);
+	return transliterate_devanagari_to_latin(input, out);
+}
+
+static int process_string(const char *input, unsigned int flags)
+{
+	char *output;
+	int ret;
+
+	ret = process_input(input, &output, flags);
+
+	switch (ret) {
+	case 0:
+		fprintf(stdout, "%s", output);
+		break;
+	case EHINDI:
+		error("the input text is Hindi");
+		break;
+	default:
+		error("unexpected error");
+		break;
+	}
+
+	free(output);
+
+	return ret;
 }
 
 #define CHUNKSIZE 1024
@@ -135,7 +159,7 @@ int main(int argc, const char **argv)
 	const char *files[argc];
 	unsigned int nfiles = 0;
 	unsigned int flags = 0;
-	char *input, *output;
+	char *input;
 
 	if (argc == 1) {
 		print_usage();
@@ -178,10 +202,11 @@ int main(int argc, const char **argv)
 	while (optind < argc) {
 		const char *arg = argv[optind++];
 
-		output = process_input(arg, flags);
-		fprintf(stdout, "%s\n", output);
-		free(output);
-	}
+		retval = process_string(arg, flags);
+		if (retval != 0)
+			return retval;
+		putchar('\n');
+ 	}
 
 	for (i = 0; i < nfiles; i++) {
 		if (strcmp(files[i], "-") == 0) {
@@ -194,11 +219,10 @@ int main(int argc, const char **argv)
 			error("failed to read file '%s'.", files[i]);
 			return retval;
 		}
-
-		output = process_input(input, flags);
-		fprintf(stdout, "%s", output);
-		free(output);
+		retval = process_string(input, flags);
 		free(input);
+		if (retval != 0)
+			return retval;
 	}
 
 	return 0;

tests/transcript.c

@@ -1,10 +1,15 @@
 #include "test.h"
 #include "transcript.h"
+#include "../compat.h"
 #include "../transcription.h"
 
 static void test_transcript(const char *devanagari, const char *latin)
 {
-	char *czech = transcript_devanagari_to_czech(devanagari);
+	char *czech;
+	int ret;
+
+	ret = transcript_devanagari_to_czech(devanagari, &czech);
+	ck_assert_int_eq(0, ret);
 	ck_assert_str_eq(latin, czech);
 	free(czech);
 }
@@ -27,7 +32,18 @@ START_TEST(test_transcript_devanagari_to_czech)
 }
 END_TEST
 
+START_TEST(test_transcript_detect_hindi)
+{
+	char *czech;
+	int ret;
+
+	ret = transcript_devanagari_to_czech("लड़की", &czech);
+	ck_assert_int_eq(EHINDI, ret);
+}
+END_TEST
+
 void register_transcript_tests(TCase *test_case)
 {
 	tcase_add_test(test_case, test_transcript_devanagari_to_czech);
+	tcase_add_test(test_case, test_transcript_detect_hindi);
 }

tests/translit.c

@@ -1,11 +1,16 @@
 #include "test.h"
 #include "translit.h"
+#include "../compat.h"
 #include "../transliteration.h"
 
 static void test_transliterate_devanagari_to_latin(const char *devanagari,
 						   const char *latin)
 {
-	char *str = transliterate_devanagari_to_latin(devanagari);
+	char *str;
+	int ret;
+
+	ret = transliterate_devanagari_to_latin(devanagari, &str);
+	ck_assert_int_eq(0, ret);
 	ck_assert_str_eq(latin, str);
 	free(str);
 }
@@ -13,7 +18,10 @@ static void test_transliterate_devanagari_to_latin(const char *devanagari,
 static void test_transliterate_latin_to_devanagari(const char *latin,
 						   const char *devanagari)
 {
-	char *str = transliterate_latin_to_devanagari(latin);
+	char *str;
+	int ret;
+
+	ret = transliterate_latin_to_devanagari(latin, &str);
 	ck_assert_str_eq(devanagari, str);
 	free(str);
 }
@@ -65,10 +73,19 @@ START_TEST(test_translit_candrabindu)
 }
 END_TEST
 
+START_TEST(test_translit_detect_hindi)
+{
+	char *hindi = NULL;
+	int ret = transliterate_devanagari_to_latin("लड़की", &hindi);
+	ck_assert_int_eq(EHINDI, ret);
+}
+END_TEST
+
 void register_translit_tests(TCase *test_case)
 {
 	tcase_add_test(test_case, test_translit_devanagari_to_latin);
 	tcase_add_test(test_case, test_translit_latin_to_devanagari);
 	tcase_add_test(test_case, test_translit_lla_sylable);
 	tcase_add_test(test_case, test_translit_candrabindu);
+	tcase_add_test(test_case, test_translit_detect_hindi);
 }

tests/velthuis.c

@@ -4,7 +4,8 @@
 
 static void test_encoding(const char *in, const char *expected)
 {
-	char *iast = encode_velthuis_to_iast_punctation(in);
+	char *iast;
+	encode_velthuis_to_iast_punctation(in, &iast);
 	ck_assert_str_eq(expected, iast);
 	free(iast);
 }

transcription.c

@@ -6,8 +6,10 @@
 #include "iast-czech.h"
 #include "utf8.h"
 
-#define CHUNKSIZE 1024
 #define SCHWA_CHARACTER 'a'
+#define VIRAMA    0x094d
+#define NUKTA     0x093c
+#define CHUNKSIZE 1024
 
 static inline int is_consonant(unsigned int c)
 {
@@ -58,7 +60,7 @@ static void end_of_word_filter(char *latin, unsigned int *pos,
 
 		/* remove singular nominative suffix */
 		len = utf8_char_length(c);
-		if (prev == 0x094d && *(latin + *pos - 1 - len) == 'm') {
+		if (prev == VIRAMA && *(latin + *pos - 1 - len) == 'm') {
 			memmove(latin + *pos - 1 - len, latin + *pos - len, c);
 			*pos = *pos - 1;
 		}
@@ -77,7 +79,7 @@ static struct translit_letter *letter_by_code(struct translit_letter *table,
 	return NULL;
 }
 
-char *transcript_devanagari_to_czech(const char *devanagari)
+int transcript_devanagari_to_czech(const char *devanagari, char **ret)
 {
 	struct translit_letter *table, *letter;
 	unsigned int c, prev = 0, alloc = 0, done = 0, len;
@@ -98,6 +100,11 @@ char *transcript_devanagari_to_czech(const char *devanagari)
 
 		nasal_consonants_filter(latin, &done, prev, c);
 
+		if (c == NUKTA) {
+			*ret = NULL;
+			return EHINDI;
+		}
+
 		letter = letter_by_code(table, c);
 		if (letter) {
 			switch (letter->type) {
@@ -107,11 +114,10 @@ char *transcript_devanagari_to_czech(const char *devanagari)
 				*(latin + done++) = SCHWA_CHARACTER;
 				break;
 			case VOWEL_SIGN:
-				if (done)
+				if (done) {
+					/* delete the inherent schwa */
 					done--;
-				strcpy(latin + done, letter->data);
-				done += strlen(letter->data);
-				break;
+				}
 			default:
 				strcpy(latin + done, letter->data);
 				done += strlen(letter->data);
@@ -132,5 +138,7 @@ char *transcript_devanagari_to_czech(const char *devanagari)
 
 	*(latin + done - 1) = '\0';
 
-	return latin;
+	*ret = latin;
+
+	return 0;
 }

transcription.h

@@ -3,6 +3,6 @@
 #ifndef __TRANSCRIPTION_H
 #define __TRANSCRIPTION_H
 
-char *transcript_devanagari_to_czech(const char *devanagari);
+int transcript_devanagari_to_czech(const char *devanagari, char **ret);
 
 #endif /* __TRANSCRIPTION_H */

transliteration.c

@@ -6,7 +6,8 @@
 #include "utf8.h"
 
 #define SCHWA_CHARACTER 'a'
-#define VIRAMA 0x094d
+#define VIRAMA    0x094d
+#define NUKTA     0x093c
 #define CHUNKSIZE 1024
 
 static struct translit_letter *letter_by_code(struct translit_letter *table,
@@ -21,7 +22,7 @@ static struct translit_letter *letter_by_code(struct translit_letter *table,
 	return NULL;
 }
 
-char *transliterate_devanagari_to_latin(const char *devanagari)
+int transliterate_devanagari_to_latin(const char *devanagari, char **ret)
 {
 	struct translit_letter *table, *letter;
 	unsigned int c, alloc = 0, done = 0, len;
@@ -40,6 +41,11 @@ char *transliterate_devanagari_to_latin(const char *devanagari)
 		len = utf8_char_length(c);
 		src += len;
 
+		if (c == NUKTA) {
+			*ret = NULL;
+			return EHINDI;
+		}
+
 		letter = letter_by_code(table, c);
 		if (letter) {
 			switch (letter->type) {
@@ -49,11 +55,10 @@ char *transliterate_devanagari_to_latin(const char *devanagari)
 				*(latin + done++) = SCHWA_CHARACTER;
 				break;
 			case VOWEL_SIGN:
-				if (done)
+				if (done) {
+					/* delete the inherent schwa */
 					done--;
-				strcpy(latin + done, letter->data);
-				done += strlen(letter->data);
-				break;
+				}
 			default:
 				strcpy(latin + done, letter->data);
 				done += strlen(letter->data);
@@ -68,7 +73,9 @@ char *transliterate_devanagari_to_latin(const char *devanagari)
 			break;
 	}
 
-	return latin;
+	*ret = latin;
+
+	return 0;
 }
 
 static struct translit_letter *letter_by_data(struct translit_letter *table,
@@ -98,7 +105,7 @@ static struct translit_letter *vowel_sign_by_data(struct translit_letter *table,
 	return NULL;
 }
 
-char *transliterate_latin_to_devanagari(const char *latin)
+int transliterate_latin_to_devanagari(const char *latin, char **ret)
 {
 	struct translit_letter *table, *letter, *next;
 	unsigned int alloc = 0, done = 0, len;
@@ -165,5 +172,7 @@ encode_vowel_modifier:
 	if (devanagari)
 		devanagari[done] = '\0';
 
-	return devanagari;
+	*ret = devanagari;
+
+	return 0;
 }

transliteration.h

@@ -22,8 +22,8 @@ struct translit_context {
 	struct translit_letter *table;
 };
 
-char *transliterate_devanagari_to_latin(const char *text);
-char *transliterate_latin_to_devanagari(const char *text);
+int transliterate_devanagari_to_latin(const char *text, char **ret);
+int transliterate_latin_to_devanagari(const char *text, char **ret);
 
 static inline int is_devanagari(unsigned int code)
 {

velthuis.c

@@ -55,7 +55,7 @@ static const struct encoder_tuple *find_tuple(const char *text)
 	return NULL;
 }
 
-char *encode_velthuis_to_iast_punctation(const char *text)
+int encode_velthuis_to_iast_punctation(const char *text, char **out)
 {
 	const char *str = text, *end = str + strlen(str);
 	const struct encoder_tuple *tuple;
@@ -79,5 +79,7 @@ char *encode_velthuis_to_iast_punctation(const char *text)
 		}
 	}
 
-	return buf;
+	*out = buf;
+
+	return 0;
 }

velthuis.h

@@ -3,6 +3,6 @@
 #ifndef __VELTHUIS_H
 #define __VELTHUIS_H
 
-char *encode_velthuis_to_iast_punctation(const char *text);
+int encode_velthuis_to_iast_punctation(const char *text, char **out);
 
 #endif /* __VELTHUIS_H */