Change the pre-format event so that hooks are given the cache entry

instead of the URI, content, and length of the entry. Change the hooks
to use add_fragment. This should fix the memory leakage when multiple
hooks change the same document, closing bug 703.

doc/events.txt

@@ -217,17 +217,16 @@ Triggered When:
 
 Arguments:
 
- unsigned char **html, int *html_len, struct session *ses, unsigned char *url
+ struct cache_entry *cached
 
 Description:
 
- Makes it possible to fix up bad HTML code, remove tags etc.  The HTML source
- is changed by making @html point to the new source.  If @html is changed the
- event propagation should be ended and @html_len should be updated to the new
- length of the document content.
- Possible values for @html includes:
- - new document content in a dynamically allocated string; or
- - NULL to keep the content unchanged.
+ Makes it possible to fix up bad HTML code, remove tags etc. The parameter
+ cached is guaranteed to have a single fragment. The HTML source is changed
+ by replacing this fragment:
+
+	add_fragment(cached, 0, new_string, new_len);
+	normalize_cache_entry(cached, new_len);
 
 -------------------------------------------------------------------------------
 Name:		quit

src/scripting/guile/hooks.c

@@ -8,7 +8,9 @@
 
 #include "elinks.h"
 
+#include "cache/cache.h"
 #include "main/event.h"
+#include "protocol/uri.h"
 #include "scripting/guile/hooks.h"
 #include "session/session.h"
 #include "util/string.h"
@@ -116,27 +118,29 @@ script_hook_follow_url(va_list ap, void *data)
 static enum evhook_status
 script_hook_pre_format_html(va_list ap, void *data)
 {
-	unsigned char **html = va_arg(ap, unsigned char **);
-	int *html_len = va_arg(ap, int *);
 	struct session *ses = va_arg(ap, struct session *);
-	unsigned char *url = va_arg(ap, unsigned char *);
+	struct cache_entry *cached = va_arg(ap, struct cache_entry *);
+	struct fragment *fragment = get_cache_fragment(cached);
+	unsigned char *url = struri(cached->uri);
+	int len;
 	SCM proc;
 	SCM x;
 
-	evhook_use_params(html && html_len && ses && url);
+	evhook_use_params(ses && cached);
 
-	if (*html == NULL || *html_len == 0) return EVENT_HOOK_STATUS_NEXT;
+	if (!cached->length || !*fragment->data) return EVENT_HOOK_STATUS_NEXT;
 
 	proc = get_guile_hook("%pre-format-html-hook");
 	if (SCM_FALSEP(proc)) return EVENT_HOOK_STATUS_NEXT;
 
 	x = scm_call_2(SCM_VARIABLE_REF(proc), scm_makfrom0str(url),
-		       scm_mem2string(*html, *html_len));
+		       scm_mem2string(fragment->data, fragment->length));
 
 	if (!SCM_STRINGP(x)) return EVENT_HOOK_STATUS_NEXT;
 
-	*html_len = SCM_STRING_LENGTH(x);
-	*html = memacpy(SCM_STRING_UCHARS(x), *html_len);
+	len = SCM_STRING_LENGTH(x);
+	add_fragment(cached, 0, SCM_STRING_UCHARS(x), len);
+	normalize_cache_entry(cached, len);
 
 	return EVENT_HOOK_STATUS_NEXT;
 }

src/scripting/lua/hooks.c

@@ -6,6 +6,7 @@
 
 #include "elinks.h"
 
+#include "cache/cache.h"
 #include "main/event.h"
 #include "protocol/uri.h"
 #include "scripting/lua/core.h"
@@ -114,13 +115,13 @@ static enum evhook_status
 script_hook_pre_format_html(va_list ap, void *data)
 {
 	lua_State *L = lua_state;
-	unsigned char **html = va_arg(ap, unsigned char **);
-	int *html_len = va_arg(ap, int *);
 	struct session *ses = va_arg(ap, struct session *);
-	unsigned char *url = va_arg(ap, unsigned char *);
+	struct cache_entry *cached = va_arg(ap, struct cache_entry *);
+	struct fragment *fragment = get_cache_fragment(cached);
+	unsigned char *url = struri(cached->uri);
 	int err;
 
-	if (*html == NULL || *html_len == 0) return EVENT_HOOK_STATUS_NEXT;
+	if (!cached->length || !*fragment->data) return EVENT_HOOK_STATUS_NEXT;
 
 	lua_getglobal(L, "pre_format_html_hook");
 	if (lua_isnil(L, -1)) {
@@ -130,7 +131,7 @@ script_hook_pre_format_html(va_list ap, void *data)
 	}
 
 	lua_pushstring(L, url);
-	lua_pushlstring(L, *html, *html_len);
+	lua_pushlstring(L, fragment->data, fragment->length);
 
 	if (prepare_lua(ses)) return EVENT_HOOK_STATUS_NEXT;
 
@@ -139,8 +140,10 @@ script_hook_pre_format_html(va_list ap, void *data)
 	if (err) return EVENT_HOOK_STATUS_NEXT;
 
 	if (lua_isstring(L, -1)) {
-		*html_len = lua_strlen(L, -1);
-		*html = memacpy((unsigned char *) lua_tostring(L, -1), *html_len);
+		int len = lua_strlen(L, -1);
+
+		add_fragment(cached, 0, (unsigned char *) lua_tostring(L, -1), len);
+		normalize_cache_entry(cached, len);
 	} else if (!lua_isnil(L, -1)) {
 		alert_lua_error("pre_format_html_hook must return a string or nil");
 	}

src/scripting/perl/hooks.c

@@ -6,6 +6,7 @@
 
 #include "elinks.h"
 
+#include "cache/cache.h"
 #include "main/event.h"
 #include "protocol/uri.h"
 #include "scripting/perl/core.h"
@@ -118,8 +119,8 @@ script_hook_follow_url(va_list ap, void *data)
 }
 
 static inline void
-do_script_hook_pre_format_html(unsigned char *url, unsigned char **html,
-			       int *html_len)
+do_script_hook_pre_format_html(unsigned char *url, struct cache_entry *cached,
+			       struct fragment *fragment)
 {
 	int count;
 	dSP;	/* Keep in variables declaration block. */
@@ -129,7 +130,7 @@ do_script_hook_pre_format_html(unsigned char *url, unsigned char **html,
 
 	PUSHMARK(SP);
 	my_XPUSHs(url, strlen(url));
-	my_XPUSHs(*html, *html_len);
+	my_XPUSHs(fragment->data, fragment->length);
 	PUTBACK;
 
 	count = call_pv("pre_format_html_hook", G_EVAL | G_SCALAR);
@@ -140,8 +141,10 @@ do_script_hook_pre_format_html(unsigned char *url, unsigned char **html,
 		unsigned char *new_html = POPpx;
 
 		if (new_html) {
-			*html_len = n_a;
-			*html = memacpy(new_html, *html_len);
+			int len = n_a;
+
+			add_fragment(cached, 0, new_html, len);
+			normalize_cache_entry(cached, len);
 		}
 	}
 
@@ -153,13 +156,13 @@ do_script_hook_pre_format_html(unsigned char *url, unsigned char **html,
 static enum evhook_status
 script_hook_pre_format_html(va_list ap, void *data)
 {
-	unsigned char **html = va_arg(ap, unsigned char **);
-	int *html_len = va_arg(ap, int *);
 	struct session *ses = va_arg(ap, struct session *);
-	unsigned char *url = va_arg(ap, unsigned char *);
+	struct cache_entry *cached = va_arg(ap, struct cache_entry *);
+	struct fragment *fragment = get_cache_fragment(cached);
+	unsigned char *url = struri(cached->uri);
 
-	if (my_perl && ses && url && *html && *html_len)
-		do_script_hook_pre_format_html(url, html, html_len);
+	if (my_perl && ses && url && cached->length && *fragment->data)
+		do_script_hook_pre_format_html(url, cached, fragment);
 
 	return EVENT_HOOK_STATUS_NEXT;
 }

src/scripting/python/hooks.c

@@ -8,6 +8,7 @@
 
 #include "elinks.h"
 
+#include "cache/cache.h"
 #include "main/event.h"
 #include "protocol/uri.h"
 #include "scripting/python/hooks.h"
@@ -100,22 +101,23 @@ script_hook_follow_url(va_list ap, void *data)
 }
 
 static void
-do_script_hook_pre_format_html(unsigned char *url, unsigned char **html,
-			       int *html_len)
+do_script_hook_pre_format_html(unsigned char *url, struct cache_entry *cached,
+			       struct fragment *fragment)
 {
 	PyObject *pFunc = PyDict_GetItemString(pDict, "pre_format_html_hook");
 
 	if (pFunc && PyCallable_Check(pFunc)) {
-		PyObject *pValue = PyObject_CallFunction(pFunc, "ss", url, *html);
+		PyObject *pValue = PyObject_CallFunction(pFunc, "ss", url,
+		                                         fragment->data);
 
 		if (pValue && (pValue != Py_None)) {
 			const unsigned char *str = PyString_AsString(pValue);
 
 			if (str) {
-				*html_len = PyString_Size(pValue); /* strlen(str); */
-				*html = memacpy((unsigned char *)str, *html_len);
-				/* Isn't a memleak here? --witekfl */
-				if (!*html) *html_len = 0;
+				int len = PyString_Size(pValue); /* strlen(str); */
+
+				add_fragment(cached, 0, (unsigned char *) str, len);
+				normalize_cache_entry(cached, len);
 			}
 			Py_DECREF(pValue);
 		} else {
@@ -130,13 +132,13 @@ do_script_hook_pre_format_html(unsigned char *url, unsigned char **html,
 static enum evhook_status
 script_hook_pre_format_html(va_list ap, void *data)
 {
-	unsigned char **html = va_arg(ap, unsigned char **);
-	int *html_len = va_arg(ap, int *);
 	struct session *ses = va_arg(ap, struct session *);
-	unsigned char *url = va_arg(ap, unsigned char *);
+	struct cache_entry *cached = va_arg(ap, struct cache_entry *);
+	struct fragment *fragment = get_cache_fragment(cached);
+	unsigned char *url = struri(cached->uri);
 
-	if (pDict && ses && url && *html && *html_len)
-		do_script_hook_pre_format_html(url, html, html_len);
+	if (pDict && ses && url && cached->length && *fragment->data)
+		do_script_hook_pre_format_html(url, cached, fragment);
 
 	return EVENT_HOOK_STATUS_NEXT;
 }

src/scripting/ruby/hooks.c

@@ -8,6 +8,7 @@
 
 #include "elinks.h"
 
+#include "cache/cache.h"
 #include "main/event.h"
 #include "protocol/uri.h"
 #include "scripting/ruby/core.h"
@@ -144,22 +145,22 @@ script_hook_follow_url(va_list ap, void *data)
 static enum evhook_status
 script_hook_pre_format_html(va_list ap, void *data)
 {
-	unsigned char **html = va_arg(ap, unsigned char **);
-	int *html_len = va_arg(ap, int *);
 	struct session *ses = va_arg(ap, struct session *);
-	unsigned char *url = va_arg(ap, unsigned char *);
+	struct cache_entry *cached = va_arg(ap, struct cache_entry *);
+	struct fragment *fragment = get_cache_fragment(cached);
+	unsigned char *url = struri(cached->uri);
 	int error;
 	VALUE args[2];
 	VALUE result;
 
-	evhook_use_params(url && ses);
+	evhook_use_params(ses, cached);
 
-	if (*html == NULL || *html_len == 0)
+	if (!cached->length || !*fragment->data)
 		return EVENT_HOOK_STATUS_NEXT;
 
 	args[0] = rb_str_new2(url);
 	/* FIXME: Use html_len */
-	args[1] = rb_str_new2(*html);
+	args[1] = rb_str_new2(fragment->data);
 
 	result = erb_protected_method_call("pre_format_html_hook", 2, args, &error);
 	if (error) {
@@ -170,13 +171,11 @@ script_hook_pre_format_html(va_list ap, void *data)
 	switch (rb_type(result)) {
 	case T_STRING:
 	{
-		unsigned char *new_html;
+		int len = RSTRING(result)->len;
+
+		add_fragment(cached, 0, RSTRING(result)->ptr, len);
+		normalize_cache_entry(cached, len);
 
-		new_html = memacpy(RSTRING(result)->ptr, RSTRING(result)->len);
-		if (new_html) {
-			*html_len = RSTRING(result)->len;
-			*html = new_html;
-		}
 		break;
 	}
 	case T_NIL:

src/scripting/see/hooks.c

@@ -8,6 +8,7 @@
 
 #include "elinks.h"
 
+#include "cache/cache.h"
 #include "main/event.h"
 #include "protocol/uri.h"
 #include "scripting/see/core.h"
@@ -156,21 +157,21 @@ script_hook_follow_url(va_list ap, void *data)
 static enum evhook_status
 script_hook_pre_format_html(va_list ap, void *data)
 {
-	unsigned char **html = va_arg(ap, unsigned char **);
-	int *html_len = va_arg(ap, int *);
 	struct session *ses = va_arg(ap, struct session *);
-	unsigned char *url = va_arg(ap, unsigned char *);
+	struct cache_entry *cached = va_arg(ap, struct cache_entry *);
+	struct fragment *fragment = get_cache_fragment(cached);
+	unsigned char *url = struri(cached->uri);
 	struct SEE_interpreter *see = &see_interpreter;
 	struct SEE_value args_[2], *args[2] = { &args_[0], &args_[1] };
 	struct SEE_value result;
 
-	evhook_use_params(url && ses);
+	evhook_use_params(ses, cached);
 
-	if (*html == NULL || *html_len == 0)
+	if (!cached->length || !*fragment->data)
 		return EVENT_HOOK_STATUS_NEXT;
 
 	SEE_SET_STRING(args[0], SEE_string_sprintf(see, "%s", url));
-	SEE_SET_STRING(args[1], SEE_string_sprintf(see, "%s", *html));
+	SEE_SET_STRING(args[1], SEE_string_sprintf(see, "%s", fragment->data));
 
 	if (!call_see_hook(see, ses, "pre_format_html", args, sizeof_array(args), &result))
 		return EVENT_HOOK_STATUS_NEXT;
@@ -181,8 +182,8 @@ script_hook_pre_format_html(va_list ap, void *data)
 		struct string new_html;
 
 		if (convert_see_string(&new_html, result.u.string)) {
-			mem_free_set(html, new_html.source);
-			*html_len = new_html.length;
+			add_fragment(cached, 0, new_html.source, new_html.length);
+			normalize_cache_entry(cached, new_html.length);
 		}
 		break;
 	}

src/session/session.c

@@ -500,8 +500,6 @@ static void
 maybe_pre_format_html(struct cache_entry *cached, struct session *ses)
 {
 	struct fragment *fragment;
-	unsigned char *src;
-	int len;
 	static int pre_format_html_event = EVENT_NONE;
 
 	if (!cached || cached->preformatted)
@@ -513,19 +511,11 @@ maybe_pre_format_html(struct cache_entry *cached, struct session *ses)
 	/* We cannot do anything if the data are fragmented. */
 	if (!list_is_singleton(cached->frag)) return;
 
-	src = fragment->data;
-	len = fragment->length;
-
 	set_event_id(pre_format_html_event, "pre-format-html");
-	trigger_event(pre_format_html_event, &src, &len, ses, struri(cached->uri));
+	trigger_event(pre_format_html_event, ses, cached);
 
-	if (src && src != fragment->data) {
-		add_fragment(cached, 0, src, len);
-		normalize_cache_entry(cached, len);
-		mem_free(src);
-	}
-
-	/* XXX: Keep after normalize_cache_entry()! */
+	/* XXX: Keep this after the trigger_event, because hooks might call
+	 * normalize_cache_entry()! */
 	cached->preformatted = 1;
 }
 #endif