cbm-part: Fix parsing of language encodings

According to TS 23.038 the language is stored at the beginning of the
information page not in the "header".

Signed-off-by: Guido Günther <agx@sigxcpu.org>

2 files changed, 61 insertions, 4 deletions

diff --git a/src/mm-cbm-part.c b/src/mm-cbm-part.c
index ef7a7807..192bab72 100644
--- a/src/mm-cbm-part.c
+++ b/src/mm-cbm-part.c
@@ -80,6 +80,7 @@ mm_cbm_part_new_from_binary_pdu (const guint8  *pdu,
     guint16 serial, group;
     int len;
     g_autofree gchar *text = NULL;
+    gboolean has_lang = FALSE, has_7bit_lang = FALSE;
 
     cbm_part = mm_cbm_part_new ();
 
@@ -136,12 +137,10 @@ mm_cbm_part_new_from_binary_pdu (const guint8  *pdu,
     if (group == CBS_DATA_CODING_LANG_GSM7) {
         cbm_part->encoding = MM_SMS_ENCODING_GSM7;
     } else if (pdu[offset] == CBS_DATA_CODING_GSM7) {
-        PDU_SIZE_CHECK (offset + 4, "cannot skip lang");
-        offset += 3;
+        has_lang = TRUE;
         cbm_part->encoding = MM_SMS_ENCODING_GSM7;
     } else if (pdu[offset] == CBS_DATA_CODING_UCS2) {
-        PDU_SIZE_CHECK (offset + 3, "cannot skip lang");
-        offset += 2;
+        has_7bit_lang = TRUE;
         cbm_part->encoding = MM_SMS_ENCODING_UCS2;
     } else if ((group == CBS_DATA_CODING_GENERAL_CLASS) ||
                (group == CBS_DATA_CODING_GENERAL_NO_CLASS) ||
@@ -167,6 +166,14 @@ mm_cbm_part_new_from_binary_pdu (const guint8  *pdu,
     cbm_part->part_num = (pdu[offset] & 0xF0) >> 4;
     offset++;
 
+    if (has_lang) {
+        PDU_SIZE_CHECK (offset + 4, "cannot skip lang");
+        offset += 3;
+    } else if (has_7bit_lang) {
+        PDU_SIZE_CHECK (offset + 3, "cannot skip 7bit lang");
+        offset += 2;
+    }
+
     switch (cbm_part->encoding) {
     case MM_SMS_ENCODING_GSM7:
         len = ((pdu_len - offset) * 8) / 7;
diff --git a/src/tests/test-cbm-part.c b/src/tests/test-cbm-part.c
index 7981018b..6bdcebb8 100644
--- a/src/tests/test-cbm-part.c
+++ b/src/tests/test-cbm-part.c
@@ -224,6 +224,52 @@ test_cbm_nl_2023 (void)
     mm_cbm_part_free (part);
 }
 
+
+static void
+test_cbm_gsm7bit_with_lang (void)
+{
+    g_autoptr(GError) err = NULL;
+    MMCbmPart *part;
+
+    static const guint8 pdu[] = {
+        0x40, 0xC0, 0x11, 0x1F, 0x10 /* GSM 7Bit with language */, 0x13,
+        0x64 /* d */, 0x65 /* e */, 0x0D /* \r */, 0x0A, 0x0A, 0x32, 0x8B, 0x52,
+        0x2A, 0x0B, 0xE4, 0x0C, 0x52, 0x93, 0x4F, 0xE7,
+    };
+
+    part = mm_cbm_part_new_from_binary_pdu (pdu, G_N_ELEMENTS (pdu), NULL, &err);
+    g_assert_no_error (err);
+    g_assert_nonnull (part);
+
+    g_assert_cmpuint (mm_cbm_part_get_channel (part), ==, 4383);
+    g_assert_cmpuint (mm_cbm_part_get_num_parts (part), ==, 3);
+    g_assert_cmpuint (mm_cbm_part_get_part_num (part), ==, 1);
+    g_assert_cmpstr (mm_cbm_part_get_language (part), ==, "de");
+}
+
+
+static void
+test_cbm_ucs2_with_7bit_lang (void)
+{
+    g_autoptr(GError) err = NULL;
+    MMCbmPart *part;
+
+    static const guint8 pdu [] = {
+        0x63, 0x40, 0x00, 0x32, 0x11 /* UCS2 with 7Bit language */, 0x14,
+        0xF2 /* ru …*/ , 0x7A /* … in GSM 7bit */, 0x04, 0x1f, 0x04, 0x40, 0x04, 0x3e,
+    };
+
+    part = mm_cbm_part_new_from_binary_pdu (pdu, G_N_ELEMENTS (pdu), NULL, &err);
+    g_assert_no_error (err);
+    g_assert_nonnull (part);
+
+    g_assert_cmpuint (mm_cbm_part_get_channel (part), ==, 50);
+    g_assert_cmpuint (mm_cbm_part_get_num_parts (part), ==, 4);
+    g_assert_cmpuint (mm_cbm_part_get_part_num (part), ==, 1);
+    g_assert_cmpstr (mm_cbm_part_get_language (part), ==, "ru");
+}
+
+
 int main (int argc, char **argv)
 {
     setlocale (LC_ALL, "");
@@ -242,6 +288,10 @@ int main (int argc, char **argv)
     /* 2023 NL alert: */
     /* https://gitlab.freedesktop.org/mobile-broadband/ModemManager/-/issues/253#note_2192474 */
     g_test_add_func ("/MM/CBM/PDU-Parser/CBM-NL", test_cbm_nl_2023);
+    /* GSM7 bit encoding with language in the first 3 bytes of the message */
+    g_test_add_func ("/MM/CBM/PDU-Parser/has-lang", test_cbm_gsm7bit_with_lang);
+    /* UCS2 encoding with 7bit language in first 2 bytes of the message */
+    g_test_add_func ("/MM/CBM/PDU-Parser/has-7bit-lang", test_cbm_ucs2_with_7bit_lang);
 
     return g_test_run ();
 }