Strip and collapse whitespace when gathering html option values.

5 files changed, 207 insertions, 3 deletions

diff --git a/src/core/document.c b/src/core/document.c
index 22f08f1..a78cde9 100644
--- a/src/core/document.c
+++ b/src/core/document.c
@@ -147,6 +147,14 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		return err;
 	}
 
+	err = dom_string_create_interned((const uint8_t *) "script",
+			SLEN("script"), &doc->script_string);
+	if (err != DOM_NO_ERR) {
+		dom_string_unref(doc->id_name);
+		dom_string_unref(doc->class_string);
+		return err;
+	}
+
 	/* Intern the empty string. The use of a space in the constant
 	 * is to prevent the compiler warning about an empty string.
 	 */
@@ -155,6 +163,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 	if (err != DOM_NO_ERR) {
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -165,6 +174,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		dom_string_unref(doc->_memo_empty);
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -176,6 +186,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		dom_string_unref(doc->_memo_empty);
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -188,6 +199,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		dom_string_unref(doc->_memo_empty);
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -201,6 +213,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		dom_string_unref(doc->_memo_empty);
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -215,6 +228,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		dom_string_unref(doc->_memo_empty);
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -230,6 +244,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		dom_string_unref(doc->_memo_empty);
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -246,6 +261,7 @@ dom_exception _dom_document_initialise(dom_document *doc,
 		dom_string_unref(doc->_memo_empty);
 		dom_string_unref(doc->id_name);
 		dom_string_unref(doc->class_string);
+		dom_string_unref(doc->script_string);
 		return err;
 	}
 
@@ -283,6 +299,7 @@ bool _dom_document_finalise(dom_document *doc)
 		dom_string_unref(doc->id_name);
 
 	dom_string_unref(doc->class_string);
+	dom_string_unref(doc->script_string);
 	dom_string_unref(doc->_memo_empty);
 	dom_string_unref(doc->_memo_domnodeinserted);
 	dom_string_unref(doc->_memo_domnoderemoved);
diff --git a/src/core/document.h b/src/core/document.h
index de49cf2..2837893 100644
--- a/src/core/document.h
+++ b/src/core/document.h
@@ -54,6 +54,8 @@ struct dom_document {
 
 	dom_string *class_string;	/**< The string "class". */
 
+	dom_string *script_string;	/**< The string "script". */
+
 	dom_document_event_internal dei;
 			/**< The DocumentEvent interface */
 	dom_document_quirks_mode quirks;
diff --git a/src/core/string.c b/src/core/string.c
index 9ba3576..9df2cd3 100644
--- a/src/core/string.c
+++ b/src/core/string.c
@@ -271,8 +271,8 @@ bool dom_string_caseless_isequal(const dom_string *s1, const dom_string *s2)
 			is2->type == DOM_STRING_INTERNED) {
 		bool match;
 
-		if (lwc_string_caseless_isequal(is1->data.intern, is2->data.intern, 
-                                                &match) != lwc_error_ok) 
+		if (lwc_string_caseless_isequal(is1->data.intern,
+				is2->data.intern, &match) != lwc_error_ok)
 			return false;
 
 		return match;
@@ -1018,3 +1018,83 @@ dom_string_tolower(dom_string *source, bool ascii_only, dom_string **lower)
 	return exc;
 }
 
+/* exported function documented in string.h */
+dom_exception dom_string_whitespace_op(dom_string *s,
+		enum dom_whitespace_op op, dom_string **ret)
+{
+	const uint8_t *src_text = (const uint8_t *) dom_string_data(s);
+	size_t len = dom_string_byte_length(s);
+	const uint8_t *src_pos;
+	const uint8_t *src_end;
+	dom_exception exc;
+	uint8_t *temp_pos;
+	uint8_t *temp;
+
+	if (len == 0) {
+		*ret = dom_string_ref(s);
+	}
+
+	temp = malloc(len);
+	if (temp == NULL) {
+		return DOM_NO_MEM_ERR;
+	}
+
+	src_pos = src_text;
+	src_end = src_text + len;
+	temp_pos = temp;
+
+	if (op & DOM_WHITESPACE_STRIP_LEADING) {
+		while (src_pos < src_end) {
+			if (*src_pos == ' '  || *src_pos == '\t' ||
+			    *src_pos == '\n' || *src_pos == '\r' ||
+			    *src_pos == '\f')
+				src_pos++;
+			else
+				break;
+		}
+	}
+
+	while (src_pos < src_end) {
+		if ((op & DOM_WHITESPACE_COLLAPSE) &&
+				(*src_pos == ' ' || *src_pos == '\t' ||
+				*src_pos == '\n' || *src_pos == '\r' ||
+				*src_pos == '\f')) {
+			/* Got a whitespace character */
+			do {
+				/* Skip all adjacent whitespace */
+				src_pos++;
+			} while (src_pos < src_end &&
+					(*src_pos == ' ' || *src_pos == '\t' ||
+					*src_pos == '\n' || *src_pos == '\r' ||
+					*src_pos == '\f'));
+			/* Gets replaced with single space in output */
+			*temp_pos++ = ' ';
+		} else {
+			/* Otherwise, copy to output */
+			*temp_pos++ = *src_pos++;
+		}
+	}
+
+	if (op & DOM_WHITESPACE_STRIP_TRAILING) {
+		if (temp_pos > temp) {
+			temp_pos--;
+			if (*temp_pos != ' ')
+				temp_pos++;
+		}
+	}
+
+	/* New length */
+	len = temp_pos - temp;
+
+	/* Make new string */
+	if (((dom_string_internal *) s)->type == DOM_STRING_CDATA) {
+		exc = dom_string_create(temp, len, ret);
+	} else {
+		exc = dom_string_create_interned(temp, len, ret);
+	}
+
+	free(temp);
+
+	return exc;
+}
+
diff --git a/src/core/string.h b/src/core/string.h
index cbf7d36..9fca1fa 100644
--- a/src/core/string.h
+++ b/src/core/string.h
@@ -14,5 +14,28 @@
 /* Map the lwc_error to dom_exception */
 dom_exception _dom_exception_from_lwc_error(lwc_error err);
 
+enum dom_whitespace_op {
+	DOM_WHITESPACE_STRIP_LEADING	= (1 << 0),
+	DOM_WHITESPACE_STRIP_TRAILING	= (1 << 1),
+	DOM_WHITESPACE_STRIP		= DOM_WHITESPACE_STRIP_LEADING |
+					  DOM_WHITESPACE_STRIP_TRAILING,
+	DOM_WHITESPACE_COLLAPSE		= (1 << 2),
+	DOM_WHITESPACE_STRIP_COLLAPSE	= DOM_WHITESPACE_STRIP |
+					  DOM_WHITESPACE_COLLAPSE
+};
+
+/** Perform whitespace operations on given string
+ *
+ * \param s	Given string
+ * \param op	Whitespace operation(s) to perform
+ * \param ret	New string with whitespace ops performed.  Caller owns ref
+ *
+ * \return DOM_NO_ERR on success.
+ *
+ * \note Right now, will return DOM_NOT_SUPPORTED_ERR if ascii_only is false.
+ */
+dom_exception dom_string_whitespace_op(dom_string *s,
+		enum dom_whitespace_op op, dom_string **ret);
+
 #endif
 
diff --git a/src/html/html_option_element.c b/src/html/html_option_element.c
index 1584bac..b133deb 100644
--- a/src/html/html_option_element.c
+++ b/src/html/html_option_element.c
@@ -8,6 +8,7 @@
 #include <assert.h>
 #include <stdlib.h>
 
+#include <dom/dom.h>
 #include <dom/html/html_option_element.h>
 #include <dom/html/html_select_element.h>
 
@@ -153,6 +154,86 @@ dom_exception dom_html_option_element_set_default_selected(
 }
 
 /**
+ * Helper for dom_html_option_element_get_text
+ */
+static dom_exception dom_html_option_element_get_text_node(
+	dom_node_internal *n, dom_string **text)
+{
+	dom_string *node_name = NULL;
+	dom_string *node_ns = NULL;
+	dom_document *owner = NULL;
+	dom_string *str = NULL;
+	dom_string *ret = NULL;
+	dom_exception exc;
+
+	*text = NULL;
+
+	assert(n->owner != NULL);
+	owner = n->owner;
+
+	for (n = n->first_child; n != NULL; n = n->next) {
+		/* Skip irrelevent node types */
+		if (n->type == DOM_COMMENT_NODE ||
+		    n->type == DOM_PROCESSING_INSTRUCTION_NODE)
+			continue;
+
+		if (n->type == DOM_ELEMENT_NODE) {
+			/* Skip script elements with html or svg namespace */
+			exc = dom_node_get_local_name(n, &node_name);
+			if (exc != DOM_NO_ERR)
+				return exc;
+			if (dom_string_caseless_isequal(node_name,
+					owner->script_string)) {
+				exc = dom_node_get_namespace(n, &node_ns);
+				if (exc != DOM_NO_ERR) {
+					dom_string_unref(node_name);
+					return exc;
+				}
+				if (dom_string_caseless_isequal(node_ns,
+						dom_namespaces[
+							DOM_NAMESPACE_HTML]) ||
+				    dom_string_caseless_isequal(node_ns,
+						dom_namespaces[
+							DOM_NAMESPACE_SVG])) {
+					dom_string_unref(node_name);
+					dom_string_unref(node_ns);
+					continue;
+				}
+				dom_string_unref(node_ns);
+			}
+			dom_string_unref(node_name);
+
+			/* Get text inside child node 'n' */
+			dom_html_option_element_get_text_node(n,
+					(str == NULL) ? &str : &ret);
+		} else {
+			/* Handle other nodes with their get_text_content
+			 * specialisation */
+			dom_node_get_text_content(n,
+					(str == NULL) ? &str : &ret);
+		}
+
+		/* If we already have text, concatenate it */
+		if (ret != NULL) {
+			dom_string *new_str;
+			dom_string_concat(str, ret, &new_str);
+			dom_string_unref(str);
+			dom_string_unref(ret);
+			str = new_str;
+		}
+	}
+
+	/* Strip and collapse whitespace */
+	if (str != NULL) {
+		dom_string_whitespace_op(str,
+				DOM_WHITESPACE_STRIP_COLLAPSE, text);
+		dom_string_unref(str);
+	}
+
+	return DOM_NO_ERR;
+}
+
+/**
  * Get the text contained in the option
  *
  * \param option  The dom_html_option_element object
@@ -162,7 +243,8 @@ dom_exception dom_html_option_element_set_default_selected(
 dom_exception dom_html_option_element_get_text(
 	dom_html_option_element *option, dom_string **text)
 {
-	return dom_node_get_text_content(option, text);
+	return dom_html_option_element_get_text_node(
+			(dom_node_internal *) option, text);
 }
 
 /**