[project @ 2005-04-16 05:09:32 by jmb]

Split out UTF-8 handling functions.
Submit URL-encoded forms in sensible encoding:
  * First entry in accept-charset list, if present
  * Document encoding, otherwise
  We may want to explicitly look for UTF-8, to save converting.
Convert cnv_str_local_enc/cnv_local_enc_str to use iconv (they're now veneers for utf8_[to/from]_enc).
Provide mechanism for looking up local system charset (derived from system alphabet, under RISC OS)

svn path=/import/netsurf/; revision=1647

3 files changed, 59 insertions, 12 deletions

diff --git a/render/box_construct.c b/render/box_construct.c
index 9723b4de1..88a432e67 100644
--- a/render/box_construct.c
+++ b/render/box_construct.c
@@ -1575,7 +1575,7 @@ bool box_iframe(BOX_SPECIAL_PARAMS)
 
 bool box_form(BOX_SPECIAL_PARAMS)
 {
-	char *action, *method, *enctype;
+	char *action, *method, *enctype, *charset;
 	form_method fmethod;
 	struct form *form;
 
@@ -1598,9 +1598,35 @@ bool box_form(BOX_SPECIAL_PARAMS)
 		xmlFree(method);
 	}
 
-	form = form_new(action, fmethod);
+	/* acceptable encoding(s) for form data */
+	if ((charset = (char *) xmlGetProp(n, (const xmlChar *) "accept-charset"))) {
+		char *comma = strchr(charset, ',');
+		if (!comma)
+			/* only one => use it */
+			comma = strdup(charset);
+		else
+			/* multiple => use first */
+			comma = strndup(charset, comma - charset);
+
+		xmlFree(charset);
+		charset = comma;
+	}
+	else if (content->data.html.encoding)
+		/* none specified => try document encoding */
+		charset = strdup(content->data.html.encoding);
+	else
+		/* none specified and no document encoding => 8859-1 */
+		charset = strdup("ISO-8859-1");
+
+	if (!charset) {
+		xmlFree(action);
+		return false;
+	}
+
+	form = form_new(action, fmethod, charset);
 	if (!form) {
 		xmlFree(action);
+		free(charset);
 		return false;
 	}
 	form->prev = content->data.html.forms;
diff --git a/render/form.c b/render/form.c
index 559f7e3e2..1cc7d0840 100644
--- a/render/form.c
+++ b/render/form.c
@@ -17,21 +17,22 @@
 #include "netsurf/render/box.h"
 #include "netsurf/render/form.h"
 #include "netsurf/utils/log.h"
+#include "netsurf/utils/utf8.h"
 #include "netsurf/utils/utils.h"
 
 
 static char *form_textarea_value(struct form_control *textarea);
 
-
 /**
  * Create a struct form.
  *
  * \param  action  URL to submit form to, used directly (not copied)
  * \param  method  method and enctype
+ * \param  charset characterset of form (not copied)
  * \return  a new structure, or 0 on memory exhaustion
  */
 
-struct form *form_new(char *action, form_method method)
+struct form *form_new(char *action, form_method method, char *charset)
 {
 	struct form *form;
 
@@ -40,6 +41,7 @@ struct form *form_new(char *action, form_method method)
 		return 0;
 	form->action = action;
 	form->method = method;
+	form->charset = charset;
 	form->controls = 0;
 	form->last_control = 0;
 	form->prev = 0;
@@ -465,15 +467,15 @@ char *form_textarea_value(struct form_control *textarea)
 /**
  * Encode controls using application/x-www-form-urlencoded.
  *
+ * \param  form  form to which successful controls relate
  * \param  control  linked list of form_successful_control
  * \return  URL-encoded form, or 0 on memory exhaustion
- *
- * \todo  encoding conversion
  */
 
-char *form_url_encode(struct form_successful_control *control)
+char *form_url_encode(struct form *form,
+		struct form_successful_control *control)
 {
-	char *name, *value;
+	char *name, *value, *n_temp, *v_temp;
 	char *s = malloc(1), *s2;
 	unsigned int len = 0, len1;
 
@@ -482,11 +484,26 @@ char *form_url_encode(struct form_successful_control *control)
 	s[0] = 0;
 
 	for (; control; control = control->next) {
-		name = curl_escape(control->name, 0);
-		value = curl_escape(control->value, 0);
+		n_temp = utf8_to_enc(control->name, form->charset, 0);
+		if (!n_temp) {
+			free(s);
+			return 0;
+		}
+		v_temp = utf8_to_enc(control->value, form->charset, 0);
+		if (!v_temp) {
+			free(n_temp);
+			free(s);
+			return 0;
+		}
+		name = curl_escape(n_temp, 0);
+		value = curl_escape(v_temp, 0);
 		len1 = len + strlen(name) + strlen(value) + 2;
 		s2 = realloc(s, len1 + 1);
 		if (!s2) {
+			curl_free(value);
+			curl_free(name);
+			free(v_temp);
+			free(n_temp);
 			free(s);
 			return 0;
 		}
@@ -495,6 +512,8 @@ char *form_url_encode(struct form_successful_control *control)
 		len = len1;
 		curl_free(name);
 		curl_free(value);
+		free(v_temp);
+		free(n_temp);
 	}
 	if (len)
 		s[len - 1] = 0;
diff --git a/render/form.h b/render/form.h
index ceddd96f8..c1e01ae0a 100644
--- a/render/form.h
+++ b/render/form.h
@@ -31,6 +31,7 @@ typedef enum {
 struct form {
 	char *action;				/**< URL to submit to. */
 	form_method method;			/**< Method and enctype. */
+	char *charset;				/**< Charset to submit form in */
 	struct form_control *controls;		/**< Linked list of controls. */
 	struct form_control *last_control;	/**< Last control in list. */
 	struct form *prev;			/**< Previous form in doc. */
@@ -101,7 +102,7 @@ struct form_successful_control {
 	struct form_successful_control *next;	/**< Next in linked list. */
 };
 
-struct form *form_new(char *action, form_method method);
+struct form *form_new(char *action, form_method method, char *charset);
 struct form_control *form_new_control(form_control_type type);
 void form_add_control(struct form *form, struct form_control *control);
 void form_free_control(struct form_control *control);
@@ -110,7 +111,8 @@ bool form_add_option(struct form_control *control, char *value, char *text,
 bool form_successful_controls(struct form *form,
 		struct form_control *submit_button,
 		struct form_successful_control **successful_controls);
-char *form_url_encode(struct form_successful_control *control);
+char *form_url_encode(struct form *form,
+		struct form_successful_control *control);
 void form_free_successful(struct form_successful_control *control);
 
 #endif