Parse the sections in the QM file format

Many thanks to Thiago Macieira for the extra help.

1 files changed, 72 insertions, 68 deletions

diff --git a/libappstream-glib/as-app-builder.c b/libappstream-glib/as-app-builder.c
index 8e0cea3..799e862 100644
--- a/libappstream-glib/as-app-builder.c
+++ b/libappstream-glib/as-app-builder.c
@@ -217,18 +217,27 @@ as_app_builder_search_locale_gettext (AsAppBuilderContext *ctx,
 }
 
 typedef enum {
-	AS_APP_TRANSLATION_QM_TAG_END		= 1,
+	AS_APP_TRANSLATION_QM_TAG_END		= 0x1,
 	/* SourceText16 */
-	AS_APP_TRANSLATION_QM_TAG_TRANSLATION	= 3,
+	AS_APP_TRANSLATION_QM_TAG_TRANSLATION	= 0x3,
 	/* Context16 */
-	AS_APP_TRANSLATION_QM_TAG_OBSOLETE1	= 5,
-	AS_APP_TRANSLATION_QM_TAG_SOURCE_TEXT	= 6,
-	AS_APP_TRANSLATION_QM_TAG_CONTEXT	= 7,
-	AS_APP_TRANSLATION_QM_TAG_COMMENT	= 8,
+	AS_APP_TRANSLATION_QM_TAG_OBSOLETE1	= 0x5,
+	AS_APP_TRANSLATION_QM_TAG_SOURCE_TEXT	= 0x6,
+	AS_APP_TRANSLATION_QM_TAG_CONTEXT	= 0x7,
+	AS_APP_TRANSLATION_QM_TAG_COMMENT	= 0x8,
 	/* Obsolete2 */
 	AS_APP_TRANSLATION_QM_TAG_LAST
 } AsAppBuilderQmTag;
 
+typedef enum {
+	AS_APP_TRANSLATION_QM_SECTION_CONTEXTS	= 0x2f,
+	AS_APP_TRANSLATION_QM_SECTION_HASHES	= 0x42,
+	AS_APP_TRANSLATION_QM_SECTION_MESSAGES	= 0x69,
+	AS_APP_TRANSLATION_QM_SECTION_NUMERUS	= 0x88,
+	AS_APP_TRANSLATION_QM_SECTION_DEPS	= 0x96,
+	AS_APP_TRANSLATION_QM_SECTION_LAST
+} AsAppBuilderQmSection;
+
 static guint8
 _read_uint8 (const guint8 *data, guint32 *offset)
 {
@@ -248,96 +257,44 @@ _read_uint32 (const guint8 *data, guint32 *offset)
 }
 
 /**
- * as_app_builder_parse_file_qt:
+ * as_app_builder_parse_data_qt:
  **/
-static gboolean
-as_app_builder_parse_file_qt (AsAppBuilderContext *ctx,
+static void
+as_app_builder_parse_data_qt (AsAppBuilderContext *ctx,
 			      const gchar *locale,
-			      const gchar *filename,
-			      GError **error)
+			      const guint8 *data,
+			      guint32 len)
 {
 	AsAppBuilderEntry *entry;
-	guint32 addr = 0;
-	guint32 len;
 	guint32 m = 0;
 	guint nstrings = 0;
-	g_autofree guint8 *data = NULL;
-	const guint8 qm_magic[] = {
-		0x3c, 0xb8, 0x64, 0x18, 0xca, 0xef, 0x9c, 0x95,
-		0xcd, 0x21, 0x1c, 0xbf, 0x60, 0xa1, 0xbd, 0xdd
-	};
-
-	/* load file */
-	if (!g_file_get_contents (filename, (gchar **) &data, (gsize *) &len, error))
-		return FALSE;
-
-	/* check header */
-	if (len < sizeof(qm_magic) ||
-	    memcmp (data, qm_magic, sizeof(qm_magic)) != 0) {
-		g_set_error_literal (error,
-				     AS_APP_ERROR,
-				     AS_APP_ERROR_FAILED,
-				     "file is invalid");
-		return FALSE;
-	}
-	m += sizeof(qm_magic);
-
-	/* unknown value 0x42? */
-	_read_uint8 (data, &m);
-
-	/* find offset to data table */
-	addr = _read_uint32 (data, &m);
-	m += addr;
-
-	/* unknown! */
-	_read_uint8 (data, &m);
-	_read_uint32 (data, &m);
-	//g_debug ("seeking to QM @ %x\n", m);
 
 	/* read data */
 	while (m < len) {
-		guint8 tag;
 		guint32 tag_len;
-		//g_debug ("QM @%x", m);
-		tag = _read_uint8(data, &m);
+		AsAppBuilderQmTag tag = _read_uint8(data, &m);
 		switch (tag) {
 		case AS_APP_TRANSLATION_QM_TAG_END:
-			//g_debug ("QM{END}");
 			break;
 		case AS_APP_TRANSLATION_QM_TAG_OBSOLETE1:
 			m += 4;
 			break;
 		case AS_APP_TRANSLATION_QM_TAG_TRANSLATION:
-			tag_len = _read_uint32 (data, &m);
-			if (tag_len < 0xffffffff)
-				m += tag_len;
-			//g_debug ("QM{TRANSLATION} len %i", tag_len);
-			nstrings++;
-			break;
 		case AS_APP_TRANSLATION_QM_TAG_SOURCE_TEXT:
-			tag_len = _read_uint32 (data, &m);
-			m += tag_len;
-			//g_debug ("QM{SOURCE_TEXT} len %i", tag_len);
-			break;
 		case AS_APP_TRANSLATION_QM_TAG_CONTEXT:
-			tag_len = _read_uint32 (data, &m);
-			m += tag_len;
-			//g_debug ("QM{CONTEXT} len %i", tag_len);
-			break;
 		case AS_APP_TRANSLATION_QM_TAG_COMMENT:
 			tag_len = _read_uint32 (data, &m);
-			m += tag_len;
-			//g_debug ("QM{COMMENT} len %i", tag_len);
+			if (tag_len < 0xffffffff)
+				m += tag_len;
+			if (tag == AS_APP_TRANSLATION_QM_TAG_TRANSLATION)
+				nstrings++;
 			break;
 		default:
-			//g_debug ("QM{unknown} tag kind %i", tag);
 			m = G_MAXUINT32;
 			break;
 		}
 	}
 
-//	g_debug ("for QT locale %s, nstrings=%i", locale, nstrings);
-
 	/* add new entry */
 	entry = as_app_builder_entry_new ();
 	entry->locale = g_strdup (locale);
@@ -345,6 +302,53 @@ as_app_builder_parse_file_qt (AsAppBuilderContext *ctx,
 	if (entry->nstrings > ctx->max_nstrings)
 		ctx->max_nstrings = entry->nstrings;
 	ctx->data = g_list_prepend (ctx->data, entry);
+}
+
+/**
+ * as_app_builder_parse_file_qt:
+ **/
+static gboolean
+as_app_builder_parse_file_qt (AsAppBuilderContext *ctx,
+			      const gchar *locale,
+			      const gchar *filename,
+			      GError **error)
+{
+	guint32 len;
+	guint32 m = 0;
+	g_autofree guint8 *data = NULL;
+	const guint8 qm_magic[] = {
+		0x3c, 0xb8, 0x64, 0x18, 0xca, 0xef, 0x9c, 0x95,
+		0xcd, 0x21, 0x1c, 0xbf, 0x60, 0xa1, 0xbd, 0xdd
+	};
+
+	/* load file */
+	if (!g_file_get_contents (filename, (gchar **) &data, (gsize *) &len, error))
+		return FALSE;
+
+	/* check header */
+	if (len < sizeof(qm_magic) ||
+	    memcmp (data, qm_magic, sizeof(qm_magic)) != 0) {
+		g_set_error_literal (error,
+				     AS_APP_ERROR,
+				     AS_APP_ERROR_FAILED,
+				     "file is invalid");
+		return FALSE;
+	}
+	m += sizeof(qm_magic);
+
+	/* parse each section */
+	while (m < len) {
+		AsAppBuilderQmSection section = _read_uint8(data, &m);
+		guint32 section_len = _read_uint32 (data, &m);
+		if (section == AS_APP_TRANSLATION_QM_SECTION_MESSAGES) {
+			as_app_builder_parse_data_qt (ctx,
+						      locale,
+						      data + m,
+						      section_len);
+		}
+		m += section_len;
+	}
+
 	return TRUE;
 }