Refactor SMS language dialect encoding / decoding

- Introduce new enum gsm_dialect instead of unsigned char arguments
- Use ISO639 3 letter codes for conversion tables
- Use a single lookup table instead of 4 different ones

diff --git a/src/util.c b/src/util.c
index 31b1e20..57e2ba2 100644 (file)
--- a/src/util.c
+++ b/src/util.c
@@ -60,7 +60,6 @@
 */
 
 #define GUND                   0xFFFF
-#define KNOWN_VARIANTS         4
 
 #define UTF8_LENGTH(c) \
        ((c) < 0x80 ? 1 : \
@@ -74,13 +73,17 @@ struct codepoint {
        unsigned short to;
 };
 
-struct single_shift_table {
-       const struct codepoint *table;
-       unsigned int len;
+struct alphabet_conversion_table {
+       const unsigned short *togsm_locking_shift;
+       const struct codepoint *togsm_single_shift;
+       unsigned int togsm_single_shift_len;
+       const struct codepoint *tounicode_locking_shift;
+       const struct codepoint *tounicode_single_shift;
+       unsigned int tounicode_single_shift_len;
 };
 
 /* GSM to Unicode extension table, for GSM sequences starting with 0x1B */
-static const struct codepoint default_ext_gsm[] = {
+static const struct codepoint def_ext_gsm[] = {
        { 0x0A, 0x000C },               /* See NOTE 3 in 23.038 */
        { 0x14, 0x005E },
        { 0x1B, 0x0020 },               /* See NOTE 1 in 23.038 */
@@ -94,7 +97,7 @@ static const struct codepoint default_ext_gsm[] = {
        { 0x65, 0x20AC }
 };
 
-static const struct codepoint default_ext_unicode[] = {
+static const struct codepoint def_ext_unicode[] = {
        { 0x000C, 0x1B0A },
        { 0x005B, 0x1B3C },
        { 0x005C, 0x1B2F },
@@ -108,7 +111,7 @@ static const struct codepoint default_ext_unicode[] = {
 };
 
 /* Appendix A.2.1. in 3GPP TS23.038, V.8.2.0 */
-static const struct codepoint turkish_ext_gsm[] = {
+static const struct codepoint tur_ext_gsm[] = {
        { 0x0A, 0x000C },               /* See NOTE 3 */
        { 0x14, 0x005E },
        { 0x1B, 0x0020 },               /* See NOTE 1 */
@@ -129,7 +132,7 @@ static const struct codepoint turkish_ext_gsm[] = {
        { 0x73, 0x015F }
 };
 
-static const struct codepoint turkish_ext_unicode[] = {
+static const struct codepoint tur_ext_unicode[] = {
        { 0x000C, 0x1B0A },
        { 0x005B, 0x1B3C },
        { 0x005C, 0x1B2F },
@@ -150,7 +153,7 @@ static const struct codepoint turkish_ext_unicode[] = {
 };
 
 /* Appendix A.2.2. in 3GPP TS23.038 V.8.2.0*/
-static const struct codepoint spanish_ext_gsm[] = {
+static const struct codepoint spa_ext_gsm[] = {
        { 0x09, 0x00E7 },
        { 0x0A, 0x000C },               /* See NOTE 3 */
        { 0x14, 0x005E },
@@ -173,7 +176,7 @@ static const struct codepoint spanish_ext_gsm[] = {
        { 0x75, 0x00FA }
 };
 
-static const struct codepoint spanish_ext_unicode[] = {
+static const struct codepoint spa_ext_unicode[] = {
        { 0x000C, 0x1B0A },
        { 0x005B, 0x1B3C },
        { 0x005C, 0x1B2F },
@@ -196,7 +199,7 @@ static const struct codepoint spanish_ext_unicode[] = {
 };
 
 /* Appendix A.2.3. in 3GPP TS23.038 V.8.2.0 */
-static const struct codepoint portuguese_ext_gsm[] = {
+static const struct codepoint por_ext_gsm[] = {
        { 0x05, 0x00EA },
        { 0x09, 0x00E7 },
        { 0x0A, 0x000C },               /* See NOTE 3 */
@@ -237,7 +240,7 @@ static const struct codepoint portuguese_ext_gsm[] = {
        { 0x7F, 0x00E2 }
 };
 
-static const struct codepoint portuguese_ext_unicode[] = {
+static const struct codepoint por_ext_unicode[] = {
        { 0x000C, 0x1B0A },
        { 0x005B, 0x1B3C },
        { 0x005C, 0x1B2F },
@@ -278,7 +281,7 @@ static const struct codepoint portuguese_ext_unicode[] = {
 };
 
 /* Used for conversion of GSM to Unicode */
-static const unsigned short default_gsm[] = {
+static const unsigned short def_gsm[] = {
        0x0040, 0x00A3, 0x0024, 0x00A5, 0x00E8, 0x00E9, 0x00F9, 0x00EC, /* 0x07 */
        0x00F2, 0x00C7, 0x000A, 0x00D8, 0x00F8, 0x000D, 0x00C5, 0x00E5, /* 0x0F */
        0x0394, 0x005F, 0x03A6, 0x0393, 0x039B, 0x03A9, 0x03A0, 0x03A8, /* 0x17 */
@@ -297,7 +300,7 @@ static const unsigned short default_gsm[] = {
        0x0078, 0x0079, 0x007A, 0x00E4, 0x00F6, 0x00F1, 0x00FC, 0x00E0  /* 0x7F */
 };
 
-static const struct codepoint default_unicode[] = {
+static const struct codepoint def_unicode[] = {
        { 0x000A, 0x0A }, { 0x000D, 0x0D }, { 0x0020, 0x20 }, { 0x0021, 0x21 },
        { 0x0022, 0x22 }, { 0x0023, 0x23 }, { 0x0024, 0x02 }, { 0x0025, 0x25 },
        { 0x0026, 0x26 }, { 0x0027, 0x27 }, { 0x0028, 0x28 }, { 0x0029, 0x29 },
@@ -333,7 +336,7 @@ static const struct codepoint default_unicode[] = {
 };
 
 /* Appendix A.3.1 in 3GPP TS23.038 */
-static const unsigned short turkish_gsm[] = {
+static const unsigned short tur_gsm[] = {
        0x0040, 0x00A3, 0x0024, 0x00A5, 0x20AC, 0x00E9, 0x00F9, 0x0131, /* 0x07 */
        0x00F2, 0x00C7, 0x000A, 0x011E, 0x011F, 0x000D, 0x00C5, 0x00E5, /* 0x0F */
        0x0394, 0x005F, 0x03A6, 0x0393, 0x039B, 0x03A9, 0x03A0, 0x03A8, /* 0x17 */
@@ -352,7 +355,7 @@ static const unsigned short turkish_gsm[] = {
        0x0078, 0x0079, 0x007A, 0x00E4, 0x00F6, 0x00F1, 0x00FC, 0x00E0  /* 0x7F */
 };
 
-static const struct codepoint turkish_unicode[] = {
+static const struct codepoint tur_unicode[] = {
        { 0x000A, 0x0A }, { 0x000D, 0x0D }, { 0x0020, 0x20 }, { 0x0021, 0x21 },
        { 0x0022, 0x22 }, { 0x0023, 0x23 }, { 0x0024, 0x02 }, { 0x0025, 0x25 },
        { 0x0026, 0x26 }, { 0x0027, 0x27 }, { 0x0028, 0x28 }, { 0x0029, 0x29 },
@@ -388,7 +391,7 @@ static const struct codepoint turkish_unicode[] = {
 };
 
 /* Appendix A.3.2 in 3GPP TS23.038 */
-static const unsigned short portuguese_gsm[] = {
+static const unsigned short por_gsm[] = {
        0x0040, 0x00A3, 0x0024, 0x00A5, 0x00EA, 0x00E9, 0x00FA, 0x00ED, /* 0x07 */
        0x00F3, 0x00E7, 0x000A, 0x00D4, 0x00F4, 0x000D, 0x00C1, 0x00E1, /* 0x0F */
        0x0394, 0x005F, 0x00AA, 0x00C7, 0x00C0, 0x221E, 0x005E, 0x005C, /* 0x17 */
@@ -407,7 +410,7 @@ static const unsigned short portuguese_gsm[] = {
        0x0078, 0x0079, 0x007A, 0x00E3, 0x00F5, 0x0060, 0x00FC, 0x00E0  /* 0x7F */
 };
 
-static const struct codepoint portuguese_unicode[] = {
+static const struct codepoint por_unicode[] = {
        { 0x000A, 0x0A }, { 0x000D, 0x0D }, { 0x0020, 0x20 }, { 0x0021, 0x21 },
        { 0x0022, 0x22 }, { 0x0023, 0x23 }, { 0x0024, 0x02 }, { 0x0025, 0x25 },
        { 0x0026, 0x26 }, { 0x0027, 0x27 }, { 0x0028, 0x28 }, { 0x0029, 0x29 },
@@ -442,32 +445,19 @@ static const struct codepoint portuguese_unicode[] = {
        { 0x00FC, 0x7E }, { 0x0394, 0x10 }, { 0x20AC, 0x18 }, { 0x221E, 0x15 }
 };
 
-static const struct single_shift_table gsm_single_shift[] = {
-       { default_ext_gsm, TABLE_SIZE(default_ext_gsm) },
-       { turkish_ext_gsm, TABLE_SIZE(turkish_ext_gsm) },
-       { spanish_ext_gsm, TABLE_SIZE(spanish_ext_gsm) },
-       { portuguese_ext_gsm, TABLE_SIZE(portuguese_ext_gsm) }
-};
-
-static const struct single_shift_table unicode_single_shift[] = {
-       { default_ext_unicode, TABLE_SIZE(default_ext_unicode) },
-       { turkish_ext_unicode, TABLE_SIZE(turkish_ext_unicode) },
-       { spanish_ext_unicode, TABLE_SIZE(spanish_ext_unicode) },
-       { portuguese_ext_unicode, TABLE_SIZE(portuguese_ext_unicode) }
-};
-
-static const unsigned short *gsm_locking_shift[] = {
-       default_gsm,
-       turkish_gsm,
-       default_gsm,
-       portuguese_gsm
-};
-
-static const struct codepoint *unicode_locking_shift[] = {
-       default_unicode,
-       turkish_unicode,
-       default_unicode,
-       portuguese_unicode
+static const struct alphabet_conversion_table alphabet_lookup[] = {
+       /* Default GSM 7 bit */
+       { def_gsm, def_ext_gsm, TABLE_SIZE(def_ext_gsm),
+               def_unicode, def_ext_unicode, TABLE_SIZE(def_ext_unicode) },
+       /* Turkish GSM dialect */
+       { tur_gsm, tur_ext_gsm, TABLE_SIZE(tur_ext_gsm),
+               tur_unicode, tur_ext_unicode, TABLE_SIZE(tur_ext_unicode) },
+       /* Spanish GSM dialect, note that this one only has extension table */
+       { def_gsm, spa_ext_gsm, TABLE_SIZE(spa_ext_gsm),
+               def_unicode, spa_ext_unicode, TABLE_SIZE(spa_ext_unicode)  },
+       /* Portuguese GSM dialect */
+       { por_gsm, por_ext_gsm, TABLE_SIZE(por_ext_gsm),
+               por_unicode, por_ext_unicode, TABLE_SIZE(por_ext_unicode) },
 };
 
 static int compare_codepoints(const void *a, const void *b)
@@ -493,40 +483,45 @@ static unsigned short codepoint_lookup(struct codepoint *key,
 static unsigned short gsm_locking_shift_lookup(unsigned char k,
                                                unsigned char lang)
 {
-       /* If language is not defined in 3GPP TS 23.038,
-        * implementations are instructed to ignore it' */
-       unsigned char variant = lang < KNOWN_VARIANTS ? lang : 0;
-
-       return gsm_locking_shift[variant][k];
+       return alphabet_lookup[lang].togsm_locking_shift[k];
 }
 
 static unsigned short gsm_single_shift_lookup(unsigned char k,
                                                unsigned char lang)
 {
        struct codepoint key = { k, 0 };
-       unsigned char variant = lang < KNOWN_VARIANTS ? lang : 0;
+       const struct codepoint *table;
+       unsigned int len;
+       
+       table = alphabet_lookup[lang].togsm_single_shift;
+       len = alphabet_lookup[lang].togsm_single_shift_len;
 
-       return codepoint_lookup(&key, gsm_single_shift[variant].table,
-                               gsm_single_shift[variant].len);
+       return codepoint_lookup(&key, table, len);
 }
 
 static unsigned short unicode_locking_shift_lookup(unsigned short k,
                                                        unsigned char lang)
 {
        struct codepoint key = { k, 0 };
-       unsigned char variant = lang < KNOWN_VARIANTS ? lang : 0;
+       const struct codepoint *table;
+       unsigned int len = 128;
 
-       return codepoint_lookup(&key, unicode_locking_shift[variant], 128);
+       table = alphabet_lookup[lang].tounicode_locking_shift;
+
+       return codepoint_lookup(&key, table, len); 
 }
 
 static unsigned short unicode_single_shift_lookup(unsigned short k,
                                                        unsigned char lang)
 {
        struct codepoint key = { k, 0 };
-       unsigned char variant = lang < KNOWN_VARIANTS ? lang : 0;
+       const struct codepoint *table;
+       unsigned int len;
 
-       return codepoint_lookup(&key, unicode_single_shift[variant].table,
-                               unicode_single_shift[variant].len);
+       table = alphabet_lookup[lang].tounicode_single_shift;
+       len = alphabet_lookup[lang].tounicode_single_shift_len;
+
+       return codepoint_lookup(&key, table, len);
 }
 
 /*!
@@ -545,14 +540,20 @@ static unsigned short unicode_single_shift_lookup(unsigned short k,
 char *convert_gsm_to_utf8_with_lang(const unsigned char *text, long len,
                                        long *items_read, long *items_written,
                                        unsigned char terminator,
-                                       unsigned char locking_lang,
-                                       unsigned char single_lang)
+                                       enum gsm_dialect locking_lang,
+                                       enum gsm_dialect single_lang)
 {
        char *res = NULL;
        char *out;
        long i = 0;
        long res_length;
 
+       if (locking_lang >= GSM_DIALECT_INVALID)
+               return NULL;
+
+       if (single_lang >= GSM_DIALECT_INVALID)
+               return NULL;
+
        if (len < 0 && !terminator)
                goto err_out;
 
@@ -626,7 +627,9 @@ char *convert_gsm_to_utf8(const unsigned char *text, long len,
 {
        return convert_gsm_to_utf8_with_lang(text, len, items_read,
                                                items_written,
-                                               terminator, 0, 0);
+                                               terminator,
+                                               GSM_DIALECT_DEFAULT,
+                                               GSM_DIALECT_DEFAULT);
 }
 
 /*!
@@ -642,8 +645,8 @@ char *convert_gsm_to_utf8(const unsigned char *text, long len,
 unsigned char *convert_utf8_to_gsm_with_lang(const char *text, long len,
                                        long *items_read, long *items_written,
                                        unsigned char terminator,
-                                       unsigned char locking_lang,
-                                       unsigned char single_lang)
+                                       enum gsm_dialect locking_lang,
+                                       enum gsm_dialect single_lang)
 {
        long nchars = 0;
        const char *in;
@@ -652,6 +655,12 @@ unsigned char *convert_utf8_to_gsm_with_lang(const char *text, long len,
        long res_len;
        long i;
 
+       if (locking_lang >= GSM_DIALECT_INVALID)
+               return NULL;
+
+       if (single_lang >= GSM_DIALECT_INVALID)
+               return NULL;
+
        in = text;
        res_len = 0;
 
@@ -730,7 +739,9 @@ unsigned char *convert_utf8_to_gsm(const char *text, long len,
 {
        return convert_utf8_to_gsm_with_lang(text, len, items_read,
                                                items_written,
-                                               terminator, 0, 0);
+                                               terminator,
+                                               GSM_DIALECT_DEFAULT,
+                                               GSM_DIALECT_DEFAULT);
 }
 
 /*!

diff --git a/src/util.h b/src/util.h
index 47aa066..6b34fa1 100644 (file)
--- a/src/util.h
+++ b/src/util.h
@@ -19,21 +19,29 @@
  *
  */
 
+enum gsm_dialect {
+       GSM_DIALECT_DEFAULT = 0,
+       GSM_DIALECT_TURKISH,
+       GSM_DIALECT_SPANISH,
+       GSM_DIALECT_PORTUGUESE,
+       GSM_DIALECT_INVALID
+};
+
 char *convert_gsm_to_utf8(const unsigned char *text, long len, long *items_read,
                                long *items_written, unsigned char terminator);
 
 char *convert_gsm_to_utf8_with_lang(const unsigned char *text, long len, long *items_read,
                                long *items_written, unsigned char terminator,
-                               unsigned char locking_shift_lang,
-                               unsigned char single_shift_lang);
+                               enum gsm_dialect locking_shift_lang,
+                               enum gsm_dialect single_shift_lang);
 
 unsigned char *convert_utf8_to_gsm(const char *text, long len, long *items_read,
                                long *items_written, unsigned char terminator);
 
 unsigned char *convert_utf8_to_gsm_with_lang(const char *text, long len, long *items_read,
                                long *items_written, unsigned char terminator,
-                               unsigned char locking_shift_lang,
-                               unsigned char single_shifth_lang);
+                               enum gsm_dialect locking_shift_lang,
+                               enum gsm_dialect single_shift_lang);
 
 unsigned char *decode_hex_own_buf(const char *in, long len, long *items_written,
                                        unsigned char terminator,