vlastavesely/sanskrit-iast
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

velthuis: allow to precisely define vowels clusters

Browse source
This commit is contained in:
Vlasta Vesely 2021-12-28 21:23:49 +01:00
parent ccf4994273
commit cd8ef0a12b
No known key found for this signature in database
GPG key ID: EB0E649DC0DFCC22
2 changed files with 44 additions and 34 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

72
tests/velthuis.c
View file

@ -2,56 +2,64 @@
#include "velthuis.h" #include "velthuis.h"
#include "../velthuis.h" #include "../velthuis.h"
static void test_encoding_to_iast(const char *in, const char *expected) #define ZWJ "\u200d"
{ #define ZWNJ "\u200c"
char *out;
encode_velthuis_to_iast(in, &out); static void test_velthuis(const char *velthuis, const char *iast)
ck_assert_str_eq(expected, out); {
free(out); char *a, *b;
int ret;
ret = encode_velthuis_to_iast(velthuis, &a);
ck_assert_int_eq(0, ret);
ck_assert_str_eq(iast, a);
ret = encode_iast_to_velthuis(a, &b);
ck_assert_int_eq(0, ret);
ck_assert_str_eq(velthuis, b);
free(a);
free(b);
} }
static void test_encoding_to_velthuis(const char *in, const char *expected) static void test_velthuis_oneway(const char *velthuis, const char *iast)
{ {
char *out; char *str;
int ret;
encode_iast_to_velthuis(in, &out); ret = encode_velthuis_to_iast(velthuis, &str);
ck_assert_str_eq(expected, out); ck_assert_int_eq(0, ret);
free(out); ck_assert_str_eq(iast, str);
free(str);
} }
START_TEST(test_encode_velthuis_to_iast) START_TEST(test_velthuis_encoding)
{ {
test_encoding_to_iast("sa.msk.rtam", "saṃskṛtam"); test_velthuis("sa.msk.rtam", "saṃskṛtam");
test_encoding_to_iast("yoga.h", "yogaḥ"); test_velthuis("yoga.h", "yogaḥ");
test_encoding_to_iast("tantra\"saastram", "tantraśāstram"); test_velthuis("tantra\"saastram", "tantraśāstram");
test_encoding_to_iast("Aa AA - II Ii - .RR .Rr", "Ā Ā - Ī Ī - Ṝ Ṝ"); test_velthuis("Aa Ii .Rr", "Ā Ī Ṝ");
test_encoding_to_iast("atha prathamo.adhyaaya.h", "atha prathamo'dhyāyaḥ"); test_velthuis("atha prathamo.adhyaaya.h", "atha prathamo'dhyāyaḥ");
}
END_TEST
START_TEST(test_encode_iast_to_velthuis)
{
test_encoding_to_velthuis("saṃskṛtam", "sa.msk.rtam");
test_encoding_to_velthuis("tantraśāstram", "tantra\"saastram");
test_encoding_to_velthuis("Ā - Ī - Ṝ", "Aa - Ii - .Rr");
test_encoding_to_velthuis("atha prathamo'dhyāyaḥ", "atha prathamo.adhyaaya.h");
} }
END_TEST END_TEST
START_TEST(test_encode_zwnj_and_zwj) START_TEST(test_encode_zwnj_and_zwj)
{ {
test_encoding_to_iast("ka+i", "ka\u200di"); test_velthuis("ka+i", "ka"ZWJ"i");
test_encoding_to_velthuis("ka\u200di", "ka+i"); test_velthuis("ka-i", "ka"ZWNJ"i");
}
END_TEST
test_encoding_to_iast("ka_i", "ka\u200ci"); START_TEST(test_encode_oneway)
test_encoding_to_velthuis("ka\u200ci", "ka_i"); {
test_velthuis_oneway("puuu puu{}u pu{}uu", "pūu pūu puū");
} }
END_TEST END_TEST
void register_velthuis_encoder_tests(TCase *test_case) void register_velthuis_encoder_tests(TCase *test_case)
{ {
tcase_add_test(test_case, test_encode_velthuis_to_iast); tcase_add_test(test_case, test_velthuis_encoding);
tcase_add_test(test_case, test_encode_iast_to_velthuis);
tcase_add_test(test_case, test_encode_zwnj_and_zwj); tcase_add_test(test_case, test_encode_zwnj_and_zwj);
tcase_add_test(test_case, test_encode_oneway);
} }

4
velthuis.c
View file

@ -35,7 +35,7 @@ static const struct encoder_tuple table[] = {
{".s", "\u1e63"}, {".S", "\u1e62"}, {".s", "\u1e63"}, {".S", "\u1e62"},
{"/", "m\u0310"}, {".a", "'"}, {"/", "m\u0310"}, {".a", "'"},
{"_", "\u200c"}, /* ZWNJ */ {"-", "\u200c"}, /* ZWNJ */
{"+", "\u200d"} /* ZWJ */ {"+", "\u200d"} /* ZWJ */
}; };
@ -82,6 +82,8 @@ int encode_velthuis_to_iast(const char *text, char **out)
sprintf(dest, "%s", tuple->to); sprintf(dest, "%s", tuple->to);
str += strlen(tuple->from); str += strlen(tuple->from);
dest += strlen(tuple->to); dest += strlen(tuple->to);
} else if (strncmp(str, "{}", 2) == 0) {
str += 2;
} else { } else {
sprintf(dest, "%c", *str); sprintf(dest, "%c", *str);
str++; str++;