Add basic support for requesting error detection; SGML scanner part missing

src/dom/scanner.c

@@ -155,7 +155,7 @@ init_dom_scanner_info(struct dom_scanner_info *scanner_info)
 void
 init_dom_scanner(struct dom_scanner *scanner, struct dom_scanner_info *scanner_info,
 		 struct dom_string *string, int state, int count_lines, int complete,
-		 int check_complete)
+		 int check_complete, int detect_errors)
 {
 	if (!scanner_info->initialized) {
 		init_dom_scanner_info(scanner_info);
@@ -173,6 +173,7 @@ init_dom_scanner(struct dom_scanner *scanner, struct dom_scanner_info *scanner_i
 	scanner->count_lines = !!count_lines;
 	scanner->incomplete = !complete;
 	scanner->check_complete = !!check_complete;
+	scanner->detect_errors = !!detect_errors;
 	scanner->lineno = scanner->count_lines;
 	scanner->info->scan(scanner);
 }

src/dom/scanner.h

@@ -20,6 +20,9 @@ struct dom_scanner_token {
 	/* Some precedence value */
 	int precedence;
 
+	/* The line number; used for error tokens */
+	unsigned int lineno;
+
 	/* The start of the token string and the token length */
 	struct dom_string string;
 };
@@ -93,7 +96,7 @@ struct dom_scanner_info {
 /* Initializes the scanner. */
 void init_dom_scanner(struct dom_scanner *scanner, struct dom_scanner_info *scanner_info,
 		      struct dom_string *string, int state, int count_lines, int complete,
-		      int check_complete);
+		      int check_complete, int detect_error);
 
 /* The number of tokens in the scanners token table:
  * At best it should be big enough to contain properties with space separated
@@ -130,6 +133,9 @@ struct dom_scanner {
 	unsigned int check_complete:1;	/* Only generate complete tokens */
 	unsigned int incomplete:1;	/* The scanned string is incomplete */
 
+	unsigned int detect_errors:1;	/* Check for markup errors */
+	unsigned int found_error;	/* Did we already report this error? */
+
 	unsigned int count_lines:1;	/* Is line counting enbaled? */
 	unsigned int lineno;		/* Line # of the last scanned token */
 

src/dom/select.c

@@ -391,7 +391,7 @@ parse_dom_select(struct dom_select *select, struct dom_stack *stack,
 	struct dom_scanner scanner;
 	struct dom_select_node sel;
 
-	init_dom_scanner(&scanner, &dom_css_scanner_info, string, 0, 0, 1, 0);
+	init_dom_scanner(&scanner, &dom_css_scanner_info, string, 0, 0, 1, 0, 0);
 
 	memset(&sel, 0, sizeof(sel));
 

src/dom/sgml/parser.c

@@ -154,6 +154,17 @@ add_sgml_node(struct dom_stack *stack, enum dom_node_type type, struct dom_scann
 
 /* SGML parser main handling: */
 
+static enum sgml_parser_code
+call_sgml_error_function(struct dom_stack *stack, struct dom_scanner_token *token)
+{
+	struct sgml_parser *parser = get_sgml_parser(stack);
+	unsigned int line = get_sgml_parser_line_number(parser);
+
+	assert(parser->error_func);
+
+	return parser->error_func(parser, &token->string, line);
+}
+
 static inline enum sgml_parser_code
 parse_sgml_attributes(struct dom_stack *stack, struct dom_scanner *scanner)
 {
@@ -217,6 +228,17 @@ parse_sgml_attributes(struct dom_stack *stack, struct dom_scanner *scanner)
 		case SGML_TOKEN_INCOMPLETE:
 			return SGML_PARSER_CODE_INCOMPLETE;
 
+		case SGML_TOKEN_ERROR:
+		{
+			enum sgml_parser_code code;
+
+			code = call_sgml_error_function(stack, token);
+			if (code != SGML_PARSER_CODE_OK)
+				return code;
+
+			skip_dom_scanner_token(scanner);
+			break;
+		}
 		default:
 			skip_dom_scanner_token(scanner);
 		}
@@ -326,7 +348,7 @@ parse_sgml_plain(struct dom_stack *stack, struct dom_scanner *scanner)
 				/* The attribute souce is complete. */
 				init_dom_scanner(&attr_scanner, &sgml_scanner_info,
 						 &token->string, SGML_STATE_ELEMENT,
-						 scanner->count_lines, 1, 0);
+						 scanner->count_lines, 1, 0, 0);
 
 				if (dom_scanner_has_tokens(&attr_scanner)) {
 					/* Ignore parser codes from this
@@ -350,6 +372,17 @@ parse_sgml_plain(struct dom_stack *stack, struct dom_scanner *scanner)
 		case SGML_TOKEN_INCOMPLETE:
 			return SGML_PARSER_CODE_INCOMPLETE;
 
+		case SGML_TOKEN_ERROR:
+		{
+			enum sgml_parser_code code;
+
+			code = call_sgml_error_function(stack, token);
+			if (code != SGML_PARSER_CODE_OK)
+				return code;
+
+			skip_dom_scanner_token(scanner);
+			break;
+		}
 		case SGML_TOKEN_SPACE:
 		case SGML_TOKEN_TEXT:
 		default:
@@ -403,11 +436,13 @@ sgml_parsing_push(struct dom_stack *stack, struct dom_node *node, void *data)
 	int count_lines = !!(parser->flags & SGML_PARSER_COUNT_LINES);
 	int complete = !!(parser->flags & SGML_PARSER_COMPLETE);
 	int incremental = !!(parser->flags & SGML_PARSER_INCREMENTAL);
+	int detect_errors = !!(parser->flags & SGML_PARSER_DETECT_ERRORS);
 
 	parsing->depth = parser->stack.depth;
 	get_dom_stack_top(&parser->stack)->immutable = 1;
 	init_dom_scanner(&parsing->scanner, &sgml_scanner_info, &node->string,
-			 SGML_STATE_TEXT, count_lines, complete, incremental);
+			 SGML_STATE_TEXT, count_lines, complete, incremental,
+			 detect_errors);
 }
 
 static void
@@ -494,6 +529,11 @@ get_sgml_parser_line_number(struct sgml_parser *parser)
 
 	assert(pstate->scanner.count_lines && pstate->scanner.lineno);
 
+	if (pstate->scanner.current
+	    && pstate->scanner.current < pstate->scanner.table + DOM_SCANNER_TOKENS
+	    && pstate->scanner.current->type == SGML_TOKEN_ERROR)
+		return pstate->scanner.current->lineno;
+
 	return pstate->scanner.lineno;
 }
 
@@ -553,6 +593,9 @@ init_sgml_parser(enum sgml_parser_type type, enum sgml_document_type doctype,
 		return NULL;
 	}
 
+	if (flags & SGML_PARSER_DETECT_ERRORS)
+		flags |= SGML_PARSER_COUNT_LINES;
+
 	parser->type  = type;
 	parser->flags = flags;
 	parser->info  = get_sgml_info(doctype);

src/dom/sgml/parser.h

@@ -7,6 +7,7 @@
 #include "dom/sgml/sgml.h"
 #include "dom/scanner.h"
 
+struct sgml_parser;
 struct string;
 struct uri;
 
@@ -26,6 +27,7 @@ enum sgml_parser_flag {
 	SGML_PARSER_COUNT_LINES	= 1,
 	SGML_PARSER_COMPLETE = 2,
 	SGML_PARSER_INCREMENTAL = 4,
+	SGML_PARSER_DETECT_ERRORS = 8,
 };
 
 struct sgml_parser_state {
@@ -48,6 +50,11 @@ enum sgml_parser_code {
 	SGML_PARSER_CODE_ERROR,
 };
 
+/* If the return code is not SGML_PARSER_CODE_OK the parsing will be ended and
+ * that code will be returned. */
+typedef enum sgml_parser_code
+(*sgml_error_T)(struct sgml_parser *, struct dom_string *, unsigned int);
+
 struct sgml_parser {
 	enum sgml_parser_type type;	/* Stream or tree */
 	enum sgml_parser_flag flags;	/* Flags that control the behaviour */
@@ -57,6 +64,8 @@ struct sgml_parser {
 	struct dom_string uri;		/* The URI of the DOM document */
 	struct dom_node *root;		/* The document root node */
 
+	sgml_error_T error_func;
+
 	struct dom_stack stack;		/* A stack for tracking parsed nodes */
 	struct dom_stack parsing;	/* Used for tracking parsing states */
 };

src/dom/sgml/scanner.h

@@ -52,6 +52,10 @@ enum sgml_token_type {
 	 * not complete. Only meaningful if scanner->complete is incomplete. */
 	SGML_TOKEN_INCOMPLETE,
 
+	/* A special token for reporting that an error in the markup was found.
+	 * Only in effect when error checking has been requested. */
+	SGML_TOKEN_ERROR,
+
 	/* Token type used internally when scanning to signal that the token
 	 * should not be recorded in the scanners token table. */
 	SGML_TOKEN_SKIP,