7 files changed, 81 insertions, 7 deletions

diff --git a/include/hubbub/errors.h b/include/hubbub/errors.h
index fe369dc..9199d09 100644
--- a/include/hubbub/errors.h
+++ b/include/hubbub/errors.h
@@ -14,6 +14,7 @@ typedef enum hubbub_error {
 	HUBBUB_OK               = 0,
 	HUBBUB_OOD		= 1, /**< Out of data */
 	HUBBUB_REPROCESS	= 2,
+	HUBBUB_ENCODINGCHANGE	= 3,
 
 	HUBBUB_NOMEM            = 5,
 	HUBBUB_BADPARM          = 6,
diff --git a/include/hubbub/functypes.h b/include/hubbub/functypes.h
index d41f6bd..8d26926 100644
--- a/include/hubbub/functypes.h
+++ b/include/hubbub/functypes.h
@@ -121,5 +121,10 @@ typedef int (*hubbub_tree_add_attributes)(void *ctx, void *node,
  */
 typedef int (*hubbub_tree_set_quirks_mode)(void *ctx, hubbub_quirks_mode mode);
 
+/**
+ * Type of encoding change notification function
+ */
+typedef int (*hubbub_tree_encoding_change)(void *ctx, uint16_t mibenum);
+
 #endif
 
diff --git a/include/hubbub/tree.h b/include/hubbub/tree.h
index 7e2e11f..0a286a0 100644
--- a/include/hubbub/tree.h
+++ b/include/hubbub/tree.h
@@ -30,6 +30,7 @@ typedef struct hubbub_tree_handler {
 	hubbub_tree_form_associate form_associate;
 	hubbub_tree_add_attributes add_attributes;
 	hubbub_tree_set_quirks_mode set_quirks_mode;
+	hubbub_tree_encoding_change encoding_change;
 	void *ctx;
 } hubbub_tree_handler;
 
diff --git a/src/charset/detect.c b/src/charset/detect.c
index 3809770..7d3459f 100644
--- a/src/charset/detect.c
+++ b/src/charset/detect.c
@@ -21,8 +21,6 @@ static uint16_t hubbub_charset_read_bom(const uint8_t *data, size_t len);
 static uint16_t hubbub_charset_scan_meta(const uint8_t *data, size_t len);
 static uint16_t hubbub_charset_parse_attributes(const uint8_t **pos,
 		const uint8_t *end);
-static uint16_t hubbub_charset_parse_content(const uint8_t *value,
-		uint32_t valuelen);
 static bool hubbub_charset_get_attribute(const uint8_t **data,
 		const uint8_t *end,
 		const uint8_t **name, uint32_t *namelen,
diff --git a/src/charset/detect.h b/src/charset/detect.h
index 807f374..cb837d0 100644
--- a/src/charset/detect.h
+++ b/src/charset/detect.h
@@ -16,5 +16,9 @@
 parserutils_error hubbub_charset_extract(const uint8_t *data, size_t len,
 		uint16_t *mibenum, uint32_t *source);
 
+/* Parse a Content-Type string for an encoding */
+uint16_t hubbub_charset_parse_content(const uint8_t *value,
+                uint32_t valuelen);
+
 #endif
 
diff --git a/src/treebuilder/in_head.c b/src/treebuilder/in_head.c
index 034be21..ddb7453 100644
--- a/src/treebuilder/in_head.c
+++ b/src/treebuilder/in_head.c
@@ -8,10 +8,76 @@
 #include <assert.h>
 #include <string.h>
 
+#include <parserutils/charset/mibenum.h>
+
 #include "treebuilder/modes.h"
 #include "treebuilder/internal.h"
 #include "treebuilder/treebuilder.h"
+
+#include "charset/detect.h"
+
 #include "utils/utils.h"
+#include "utils/string.h"
+
+
+/**
+ * Process a <meta> tag as if "in head".
+ *
+ * \param treebuilder	The treebuilder instance
+ * \param token        The token to process
+ */
+static hubbub_error process_meta_in_head(hubbub_treebuilder *treebuilder,
+		const hubbub_token *token)
+{
+	insert_element_no_push(treebuilder, &token->data.tag);
+
+	/** \todo ack sc flag */
+
+#if 0
+	if (confidence == certain)
+		return HUBBUB_OK;
+#endif
+
+	uint16_t charset_enc = 0;
+	uint16_t content_type_enc = 0;
+
+	for (size_t i = 0; i < token->data.tag.n_attributes; i++) {
+		hubbub_attribute *attr = &token->data.tag.attributes[i];
+
+		if (hubbub_string_match(attr->name.ptr, attr->name.len,
+				(const uint8_t *) "charset",
+				SLEN("charset")) == true) {
+			/* Extract charset */
+			charset_enc = parserutils_charset_mibenum_from_name(
+					(const char *) attr->value.ptr,
+					attr->value.len);
+		} else if (hubbub_string_match(attr->name.ptr, attr->name.len,
+				(const uint8_t *) "content",
+				SLEN("content")) == true) {
+			/* Extract charset from Content-Type */
+			content_type_enc = hubbub_charset_parse_content(
+					attr->value.ptr, attr->value.len);
+		}
+	}
+
+	if (charset_enc != 0) {
+		if (treebuilder->tree_handler->encoding_change) {
+			treebuilder->tree_handler->encoding_change(
+					treebuilder->tree_handler->ctx,
+					charset_enc);
+		}
+		return HUBBUB_ENCODINGCHANGE;
+	} else if (content_type_enc != 0) {
+		if (treebuilder->tree_handler->encoding_change) {
+			treebuilder->tree_handler->encoding_change(
+					treebuilder->tree_handler->ctx,
+					content_type_enc);
+		}
+		return HUBBUB_ENCODINGCHANGE;
+	}
+
+	return HUBBUB_OK;
+}
 
 
 
@@ -101,11 +167,7 @@ hubbub_error handle_in_head(hubbub_treebuilder *treebuilder,
 
 			/** \todo ack sc flag */
 		} else if (type == META) {
-			insert_element_no_push(treebuilder, &token->data.tag);
-
-			/** \todo ack sc flag */
-
-			/** \todo detect charset */
+			err = process_meta_in_head(treebuilder, token);
 		} else if (type == TITLE) {
 			parse_generic_rcdata(treebuilder, token, true);
 		} else if (type == NOFRAMES || type == STYLE) {
diff --git a/src/utils/errors.c b/src/utils/errors.c
index 9b9dfdb..7c6b5ea 100644
--- a/src/utils/errors.c
+++ b/src/utils/errors.c
@@ -29,6 +29,9 @@ const char *hubbub_error_to_string(hubbub_error error)
 	case HUBBUB_REPROCESS:
 		result = "Internal (reprocess token)";
 		break;
+	case HUBBUB_ENCODINGCHANGE:
+		result = "Encoding of document has changed";
+		break;
 	case HUBBUB_NOMEM:
 		result = "Insufficient memory";
 		break;