Add num2text

Signed-off-by: Mattias Andrée <maandree@kth.se>

5 files changed, 170 insertions, 4 deletions

diff --git a/Makefile b/Makefile
index eb187ed..ce1193a 100644
--- a/Makefile
+++ b/Makefile
@@ -18,6 +18,7 @@ LANG =\
 
 CMD =\
 	card2ord\
+	num2text\
 	numtext-strip
 
 LIB_OBJ =\
diff --git a/TODO b/TODO
index 8e94350..52c52b5 100644
--- a/TODO
+++ b/TODO
@@ -1,5 +1,4 @@
 Add [libnumtext_]text2num: text to numerals, with analysis (possible flags from num2text), use support mixed in numerals
-Add num2text: conversion utility that wraps libnumtext_num2text
 Add opposite of libnumtext_remove_separators and numtext-strip
 Add man pages and README
 Add IPA output support?
diff --git a/common.h b/common.h
index 1404b35..e51652e 100644
--- a/common.h
+++ b/common.h
@@ -63,6 +63,7 @@ void process_options(char **optss, size_t n_optss, struct option *options, uint3
 
 int numtext_strip_main(int argc, char *argv[]);
 int card2ord_main(int argc, char *argv[]);
+int num2text_main(int argc, char *argv[]);
 
 
 ssize_t libnumtext_num2text_swedish__(struct common_num2text_params *params, const char *num, size_t num_len, uint32_t flags);
diff --git a/num2text.c b/num2text.c
new file mode 100644
index 0000000..d9d4046
--- /dev/null
+++ b/num2text.c
@@ -0,0 +1,160 @@
+/* See LICENSE file for copyright and license details. */
+#define LIBSIMPLY_CONFIG_MULTICALL_BINARY
+#include "common.h"
+#include <libsimple-arg.h>
+
+USAGE("-l language [-o options ...] [cardinal ...]");
+
+
+static enum libnumtext_language lang;
+static uint32_t flags = 0;
+
+
+static struct option swedish_options[] = {
+
+	{"[type=]c[ard[inal]]",        "type=cardinal",          LIBNUMTEXT_N2T_SWEDISH_CARDINAL,              UINT32_C(0x00000001)},
+	{NULL,                         "type=card",              LIBNUMTEXT_N2T_SWEDISH_CARDINAL,              UINT32_C(0x00000001)},
+	{NULL,                         "type=c",                 LIBNUMTEXT_N2T_SWEDISH_CARDINAL,              UINT32_C(0x00000001)},
+	{NULL,                         "cardinal",               LIBNUMTEXT_N2T_SWEDISH_CARDINAL,              UINT32_C(0x00000001)},
+	{NULL,                         "card",                   LIBNUMTEXT_N2T_SWEDISH_CARDINAL,              UINT32_C(0x00000001)},
+	{NULL,                         "c",                      LIBNUMTEXT_N2T_SWEDISH_CARDINAL,              UINT32_C(0x00000001)},
+	{"[type=]o[rd[inal]]",         "type=ordinal",           LIBNUMTEXT_N2T_SWEDISH_ORDINAL,               UINT32_C(0x00000001)},
+	{NULL,                         "type=ord",               LIBNUMTEXT_N2T_SWEDISH_ORDINAL,               UINT32_C(0x00000001)},
+	{NULL,                         "type=o",                 LIBNUMTEXT_N2T_SWEDISH_ORDINAL,               UINT32_C(0x00000001)},
+	{NULL,                         "ordinal",                LIBNUMTEXT_N2T_SWEDISH_ORDINAL,               UINT32_C(0x00000001)},
+	{NULL,                         "ord",                    LIBNUMTEXT_N2T_SWEDISH_ORDINAL,               UINT32_C(0x00000001)},
+	{NULL,                         "o",                      LIBNUMTEXT_N2T_SWEDISH_ORDINAL,               UINT32_C(0x00000001)},
+
+	{"numerator|denominator=n[o]", "denominator=no",         LIBNUMTEXT_N2T_SWEDISH_NUMERATOR,             UINT32_C(0x00000002)},
+	{NULL,                         "denominator=n",          LIBNUMTEXT_N2T_SWEDISH_NUMERATOR,             UINT32_C(0x00000002)},
+	{NULL,                         "numerator",              LIBNUMTEXT_N2T_SWEDISH_NUMERATOR,             UINT32_C(0x00000002)},
+	{"denominator[=y[es]]",        "denominator=yes",        LIBNUMTEXT_N2T_SWEDISH_DENOMINATOR,           UINT32_C(0x00000002)},
+	{NULL,                         "denominator=y",          LIBNUMTEXT_N2T_SWEDISH_DENOMINATOR,           UINT32_C(0x00000002)},
+	{NULL,                         "denominator",            LIBNUMTEXT_N2T_SWEDISH_DENOMINATOR,           UINT32_C(0x00000002)},
+
+	{"singular|plural=n[o]",       "plural=no",              LIBNUMTEXT_N2T_SWEDISH_SINGULAR_FORM,         UINT32_C(0x00000004)},
+	{NULL,                         "plural=n",               LIBNUMTEXT_N2T_SWEDISH_SINGULAR_FORM,         UINT32_C(0x00000004)},
+	{NULL,                         "singular",               LIBNUMTEXT_N2T_SWEDISH_SINGULAR_FORM,         UINT32_C(0x00000004)},
+	{"plural[=y[es]]",             "plural=yes",             LIBNUMTEXT_N2T_SWEDISH_PLURAL_FORM,           UINT32_C(0x00000004)},
+	{NULL,                         "plural=y",               LIBNUMTEXT_N2T_SWEDISH_PLURAL_FORM,           UINT32_C(0x00000004)},
+	{NULL,                         "plural",                 LIBNUMTEXT_N2T_SWEDISH_PLURAL_FORM,           UINT32_C(0x00000004)},
+
+	{"indefinite|definite=n[o]",   "definite=no",            LIBNUMTEXT_N2T_SWEDISH_INDEFINITE_FORM,       UINT32_C(0x00000008)},
+	{NULL,                         "definite=n",             LIBNUMTEXT_N2T_SWEDISH_INDEFINITE_FORM,       UINT32_C(0x00000008)},
+	{NULL,                         "indefinite",             LIBNUMTEXT_N2T_SWEDISH_INDEFINITE_FORM,       UINT32_C(0x00000008)},
+	{"definite[=y[es]]",           "definite=yes",           LIBNUMTEXT_N2T_SWEDISH_DEFINITE_FORM,         UINT32_C(0x00000008)},
+	{NULL,                         "definite=y",             LIBNUMTEXT_N2T_SWEDISH_DEFINITE_FORM,         UINT32_C(0x00000008)},
+	{NULL,                         "definite",               LIBNUMTEXT_N2T_SWEDISH_DEFINITE_FORM,         UINT32_C(0x00000008)},
+
+	{"gender=c[ommon]|u|t",        "gender=common",          LIBNUMTEXT_N2T_SWEDISH_COMMON_GENDER,         UINT32_C(0x00000030)},
+	{NULL,                         "gender=c",               LIBNUMTEXT_N2T_SWEDISH_COMMON_GENDER,         UINT32_C(0x00000030)},
+	{NULL,                         "gender=u",               LIBNUMTEXT_N2T_SWEDISH_COMMON_GENDER,         UINT32_C(0x00000030)},
+	{NULL,                         "gender=t",               LIBNUMTEXT_N2T_SWEDISH_COMMON_GENDER,         UINT32_C(0x00000030)},
+	{"gender=n[euter]",            "gender=neuter",          LIBNUMTEXT_N2T_SWEDISH_NEUTER_GENDER,         UINT32_C(0x00000030)},
+	{NULL,                         "gender=n",               LIBNUMTEXT_N2T_SWEDISH_NEUTER_GENDER,         UINT32_C(0x00000030)},
+	{"gender=m[asculine]",         "gender=masculine",       LIBNUMTEXT_N2T_SWEDISH_MASCULINE_GENDER,      UINT32_C(0x00000030)},
+	{NULL,                         "gender=m",               LIBNUMTEXT_N2T_SWEDISH_MASCULINE_GENDER,      UINT32_C(0x00000030)},
+	{"gender=f[eminine]",          "gender=feminine",        LIBNUMTEXT_N2T_SWEDISH_FEMININE_GENDER,       UINT32_C(0x00000030)},
+	{NULL,                         "gender=f",               LIBNUMTEXT_N2T_SWEDISH_FEMININE_GENDER,       UINT32_C(0x00000030)},
+
+	{"one=e[xplicit]",             "one=explicit",           LIBNUMTEXT_N2T_SWEDISH_EXPLICIT_ONE,          UINT32_C(0x00000040)},
+	{NULL,                         "one=e",                  LIBNUMTEXT_N2T_SWEDISH_EXPLICIT_ONE,          UINT32_C(0x00000040)},
+	{"one=i[mplicit]",             "one=implicit",           LIBNUMTEXT_N2T_SWEDISH_IMPLICIT_ONE,          UINT32_C(0x00000040)},
+	{NULL,                         "one=i",                  LIBNUMTEXT_N2T_SWEDISH_IMPLICIT_ONE,          UINT32_C(0x00000040)},
+
+	{"hyphenated=n[o]",            "hyphenated=no",          LIBNUMTEXT_N2T_SWEDISH_NOT_HYPHENATED,        UINT32_C(0x00000080)},
+	{NULL,                         "hyphenated=n",           LIBNUMTEXT_N2T_SWEDISH_NOT_HYPHENATED,        UINT32_C(0x00000080)},
+	{"hyphenated[=y[es]]",         "hyphenated=yes",         LIBNUMTEXT_N2T_SWEDISH_HYPHENATED,            UINT32_C(0x00000080)},
+	{NULL,                         "hyphenated=y",           LIBNUMTEXT_N2T_SWEDISH_HYPHENATED,            UINT32_C(0x00000080)},
+	{NULL,                         "hyphenated",             LIBNUMTEXT_N2T_SWEDISH_HYPHENATED,            UINT32_C(0x00000080)},
+
+	{"case=l[ower]",               "case=lower",             LIBNUMTEXT_N2T_SWEDISH_LOWER_CASE,            UINT32_C(0x00000300)},
+	{NULL,                         "case=l",                 LIBNUMTEXT_N2T_SWEDISH_LOWER_CASE,            UINT32_C(0x00000300)},
+	{"case=u[pper]",               "case=upper",             LIBNUMTEXT_N2T_SWEDISH_UPPER_CASE,            UINT32_C(0x00000300)},
+	{NULL,                         "case=u",                 LIBNUMTEXT_N2T_SWEDISH_UPPER_CASE,            UINT32_C(0x00000300)},
+	{"case=m[oney]",               "case=money",             LIBNUMTEXT_N2T_SWEDISH_MONEY_CASE,            UINT32_C(0x00000300)},
+	{NULL,                         "case=m",                 LIBNUMTEXT_N2T_SWEDISH_MONEY_CASE,            UINT32_C(0x00000300)},
+	{"case=s[entence]",            "case=sentence",          LIBNUMTEXT_N2T_SWEDISH_SENTENCE_CASE,         UINT32_C(0x00000300)},
+	{NULL,                         "case=s",                 LIBNUMTEXT_N2T_SWEDISH_SENTENCE_CASE,         UINT32_C(0x00000300)},
+
+	{"hyphenation=n[o[ne]]",       "hyphenation=none",       LIBNUMTEXT_N2T_SWEDISH_NO_HYPHENATION,        UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=no",         LIBNUMTEXT_N2T_SWEDISH_NO_HYPHENATION,        UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=n",          LIBNUMTEXT_N2T_SWEDISH_NO_HYPHENATION,        UINT32_C(0x00000C00)},
+	{"hyphenation=c[omponent[s]]", "hyphenation=components", LIBNUMTEXT_N2T_SWEDISH_COMPONENT_HYPHENATION, UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=component",  LIBNUMTEXT_N2T_SWEDISH_COMPONENT_HYPHENATION, UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=c",          LIBNUMTEXT_N2T_SWEDISH_COMPONENT_HYPHENATION, UINT32_C(0x00000C00)},
+	{"hyphenation=sy[llable[s]]",  "hyphenation=syllables",  LIBNUMTEXT_N2T_SWEDISH_SYLLABLE_HYPHENATION,  UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=syllable",   LIBNUMTEXT_N2T_SWEDISH_SYLLABLE_HYPHENATION,  UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=sy",         LIBNUMTEXT_N2T_SWEDISH_SYLLABLE_HYPHENATION,  UINT32_C(0x00000C00)},
+	{"hyphenation=se[condary]|2",  "hyphenation=secondary",  LIBNUMTEXT_N2T_SWEDISH_SECONDARY_HYPHENATION, UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=se",         LIBNUMTEXT_N2T_SWEDISH_SECONDARY_HYPHENATION, UINT32_C(0x00000C00)},
+	{NULL,                         "hyphenation=2",          LIBNUMTEXT_N2T_SWEDISH_SECONDARY_HYPHENATION, UINT32_C(0x00000C00)},
+
+	{"triplets=r[educed]",         "triplets=reduced",       LIBNUMTEXT_N2T_SWEDISH_REDUCED_TRIPLETS,      UINT32_C(0x00003000)},
+	{NULL,                         "triplets=r",             LIBNUMTEXT_N2T_SWEDISH_REDUCED_TRIPLETS,      UINT32_C(0x00003000)},
+	{"triplets=l[atex]",           "triplets=latex",         LIBNUMTEXT_N2T_SWEDISH_LATEX_TRIPLETS,        UINT32_C(0x00003000)},
+	{NULL,                         "triplets=l",             LIBNUMTEXT_N2T_SWEDISH_LATEX_TRIPLETS,        UINT32_C(0x00003000)},
+	{"triplets=e[xplicit]",        "triplets=explicit",      LIBNUMTEXT_N2T_SWEDISH_EXPLICIT_TRIPLETS,     UINT32_C(0x00003000)},
+	{NULL,                         "triplets=e",             LIBNUMTEXT_N2T_SWEDISH_EXPLICIT_TRIPLETS,     UINT32_C(0x00003000)},
+
+	{NULL, NULL, 0, 0}
+};
+
+static struct option *options[] = {
+	[LIBNUMTEXT_SWEDISH] = swedish_options
+};
+
+
+static ssize_t
+process(char *outbuf, size_t outbuf_size, const char *num, size_t num_len)
+{
+	ssize_t ret;
+	ret = libnumtext_num2text(outbuf, outbuf_size, num, num_len, lang, flags);
+	if (ret < 0)
+		fprintf(stderr, "%s: libnumtext_num2text %s: %s\n", argv0, num, strerror(errno));
+	return ret;
+}
+
+
+int
+num2text_main(int argc, char *argv[])
+{
+	int have_lang = 0;
+	char **optionses;
+	size_t n_optionses = 0;
+	uint32_t flag_mask;
+
+	optionses = calloc((size_t)argc, sizeof(optionses));
+	if (!optionses) {
+		fprintf(stderr, "%s: calloc %zu %zu: %s\n", argv0, (size_t)argc, sizeof(optionses), strerror(errno));
+		return 1;
+	}
+
+	ARGBEGIN {
+	case 'l':
+		if (!get_language(ARG(), &lang, &have_lang))
+			usage();
+		break;
+	case 'o':
+		optionses[n_optionses++] = ARG();
+		break;
+	default:
+		usage();
+	} ARGEND;
+
+	if (!have_lang)
+		usage();
+
+	process_options(optionses, n_optionses, options[lang], &flags, &flag_mask);
+	free(optionses);
+
+	if (lang == LIBNUMTEXT_SWEDISH) {
+		if ((flag_mask & UINT32_C(0x0000000C)) && !(flags & UINT32_C(0x00000003))) {
+			fprintf(stderr, "%s: singular/plural and indefinite/definite"
+			                " cannot be used with numerator cardinals\n", argv0);
+			return 1;
+		}
+	}
+
+	return run(argc, argv, process);
+}
diff --git a/numtext.c b/numtext.c
index 3b78756..18d6c2c 100644
--- a/numtext.c
+++ b/numtext.c
@@ -20,6 +20,8 @@ main(int argc, char *argv[])
 		return numtext_strip_main(argc, argv);
 	else if (!strcmp(argv0, "card2ord"))
 		return card2ord_main(argc, argv);
+	else if (!strcmp(argv0, "num2text"))
+		return num2text_main(argc, argv);
 
 	fprintf(stderr, "%s: not a recognised command for numtext multicall binary\n", argv[0]);
 	return 1;
@@ -145,6 +147,7 @@ get_language(const char *arg, enum libnumtext_language *langp, int *have_langp)
 static char *
 process_option(char *opt, struct option *options, uint32_t *flagsp, uint32_t *maskedp)
 {
+	uint32_t flag, mask;
 	size_t len;
 
 	if (opt[0] == '?' && (!opt[1] || opt[1] == ',')) {
@@ -158,12 +161,14 @@ process_option(char *opt, struct option *options, uint32_t *flagsp, uint32_t *ma
 	for (; options->option; options++) {
 		len = strlen(options->option);
 		if (!strncmp(opt, options->option, len) && (!opt[len] || opt[len] == ',')) {
-			if (options->mask & *maskedp) {
+			flag = options->flag;
+			mask = options->mask ? options->mask : options->flag;
+			if ((mask & *maskedp) && (*flagsp & mask) != flag) {
 				fprintf(stderr, "%s: option conflicts with previously specified option: %.*s\n", argv0, (int)len, opt);
 				exit(1);
 			}
-			*flagsp |= options->flag;
-			*maskedp |= options->mask;
+			*flagsp |= flag;
+			*maskedp |= mask;
 			return &opt[len];
 		}
 	}