path: root/old/test/filtering.2.patch

diff -r e820d7cb2be7 src/cache.c
--- a/src/cache.c	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/cache.c	Mon Dec 28 01:54:09 2009 +0000
@@ -937,7 +937,7 @@
          NewUrl = a_Url_new(URL_STR_(entry->Location), URL_STR_(entry->Url));
          if (entry->Flags & CA_TempRedirect)
             a_Url_set_flags(NewUrl, URL_FLAGS(NewUrl) | URL_E2EQuery);
-         a_Nav_push(bw, NewUrl);
+         a_Nav_push(bw, NewUrl, entry->Url);
          a_Url_free(NewUrl);
       } else {
          /* Sub entity redirection (most probably an image) */
diff -r e820d7cb2be7 src/capi.c
--- a/src/capi.c	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/capi.c	Mon Dec 28 01:54:09 2009 +0000
@@ -15,6 +15,7 @@
  * to get the requests served. Kind of a broker.
  */
 
+#include <ctype.h>    /* isalpha */
 #include <string.h>
 
 #include "msg.h"
@@ -313,6 +314,86 @@
 }
 
 /*
+ * When dillo wants to open an URL, this can be either due to user action
+ * (e.g., typing in an URL, clicking a link), or automatic (HTTP header
+ * indicates redirection, META HTML tag with refresh attribute and 0 delay,
+ * and images and stylesheets on an HTML page when autoloading is enabled).
+ *
+ * For a user request, the action will be permitted.
+ * For an automatic request, permission to load depends on the filter set
+ * by the user.
+ */ 
+static bool_t Capi_filters_allow(const DilloUrl *wanted,
+                                 const DilloUrl *requester)
+{
+   bool_t ret;
+
+   if (requester == NULL) {
+      /* request made by user */
+      ret = TRUE;
+   } else {
+      switch (prefs.filter_auto_requests) {
+         case PREFS_FILTER_SAME_HOST:
+            ret = dStrcasecmp(URL_HOST(wanted), URL_HOST(requester)) == 0;
+            break;
+         case PREFS_FILTER_SAME_DOMAIN:
+         {
+            const char *req_host = URL_HOST(requester),
+                       *want_host = URL_HOST(wanted);
+            uint_t req_hostlen = strlen(req_host);
+ 
+            MSG("Capi_filters_allow:\n%s\n%s\n", req_host, want_host);
+            if (req_hostlen > 0 && !isalpha(req_host[req_hostlen - 1])) {
+               /* for IP addrs, match fully */
+               ret = dStrcasecmp(req_host, want_host) == 0;
+            } else {
+               /* For hostnames, let's say for now that if you have two dots,
+                * we'll ignore everything through the first dot, e.g.,
+                * 'www.dillo.org' => 'dillo.org'
+                *
+                * TODO: it sounds like requiring a third dot would be best for
+                * ai,au,bd,bh,ck,eg,et,fk,il,in,kh,kr,mk,mt,na,np,nz,pg,pk,qa,
+                * sa,sb,sg,sv,ua,ug,uk,uy,vn,za,zw, name
+                */
+               uint_t want_hostlen;
+               char *dot = strchr(req_host, '.');
+ 
+               if (dot && strchr(dot + 1, '.')) {
+                  req_hostlen -= dot + 1 - req_host;
+                  req_host = dot + 1;
+               }
+               want_hostlen = strlen(want_host);
+ 
+               if (want_hostlen < req_hostlen) {
+                  ret = FALSE;
+               } else {
+                  if (want_hostlen > req_hostlen &&
+                     want_host[want_hostlen - req_hostlen - 1] == '.') {
+                     /* If the longer, leading portion of the host ends with
+                      * a dot, trim it off, e.g., 'hg.dillo.org' => 'dillo.org'
+                      * but 'hgdillo.org' would be unchanged.
+                      */
+                     want_host += want_hostlen - req_hostlen;
+                  }
+                  ret = dStrcasecmp(req_host, want_host) == 0;
+               }
+            }
+            if (ret)
+               MSG("ALLOW\n");
+            else
+               MSG("BLOCK\n");
+            break;
+         }
+         case PREFS_FILTER_ALLOW_ALL:
+         default:
+            ret = TRUE;
+            break;
+      }
+   }
+   return ret;
+}
+
+/*
  * Most used function for requesting a URL.
  * TODO: clean up the ad-hoc bindings with an API that allows dynamic
  *       addition of new plugins.
@@ -328,6 +409,9 @@
    const char *scheme = URL_SCHEME(web->url);
    int safe = 0, ret = 0, use_cache = 0;
 
+   dReturn_val_if_fail((a_Capi_get_flags(web->url) & CAPI_IsCached) ||
+                       Capi_filters_allow(web->url, web->requester), 0);
+
    /* reload test */
    reload = (!(a_Capi_get_flags(web->url) & CAPI_IsCached) ||
              (URL_FLAGS(web->url) & URL_E2EQuery));
diff -r e820d7cb2be7 src/html.cc
--- a/src/html.cc	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/html.cc	Mon Dec 28 01:54:09 2009 +0000
@@ -105,8 +105,8 @@
                                   const char *attrname,
                                   int tag_parsing_flags);
 static int Html_write_raw(DilloHtml *html, char *buf, int bufsize, int Eof);
-static void Html_load_image(BrowserWindow *bw, DilloUrl *url,
-                            DilloImage *image);
+static bool Html_load_image(BrowserWindow *bw, DilloUrl *url,
+                            const DilloUrl *requester, DilloImage *image);
 static void Html_callback(int Op, CacheClient_t *Client);
 static void Html_tag_cleanup_at_close(DilloHtml *html, int TagIdx);
 
@@ -654,12 +654,21 @@
 {
    dReturn_if_fail (bw->nav_expecting == FALSE);
 
+   /* If the user asked for a specific URL, the user (NULL) is the requester,
+    * but if the user just asked for all URLs, use the page URL as the
+    * requester. If the possible patterns become more complex, it might be
+    * good to have the caller supply the requester instead.
+    */
+   const DilloUrl *requester = pattern ? NULL : this->page_url;
+
    for (int i = 0; i < images->size(); i++) {
       if (images->get(i)->image) {
          if ((!pattern) || (!a_Url_cmp(images->get(i)->url, pattern))) {
-            Html_load_image(bw, images->get(i)->url, images->get(i)->image);
-            a_Image_unref (images->get(i)->image);
-            images->get(i)->image = NULL;  // web owns it now
+            if (Html_load_image(bw, images->get(i)->url, requester,
+                                images->get(i)->image)) {
+               a_Image_unref (images->get(i)->image);
+               images->get(i)->image = NULL;  // web owns it now
+            }
          }
       }
    }
@@ -2086,9 +2095,10 @@
 
    load_now = prefs.load_images ||
               (a_Capi_get_flags_with_redirection(url) & CAPI_IsCached);
-   Html_add_new_htmlimage(html, &url, load_now ? NULL : Image);
+   bool loading = false;
    if (load_now)
-      Html_load_image(html->bw, url, Image);
+      loading = Html_load_image(html->bw, url, html->page_url, Image);
+   Html_add_new_htmlimage(html, &url, loading ? NULL : Image);
 
    dFree(tooltip_str);
    dFree(width_ptr);
@@ -2100,13 +2110,13 @@
 /*
  * Tell cache to retrieve image
  */
-static void Html_load_image(BrowserWindow *bw, DilloUrl *url,
-                            DilloImage *Image)
+static bool Html_load_image(BrowserWindow *bw, DilloUrl *url,
+                            const DilloUrl *requester, DilloImage *Image)
 {
    DilloWeb *Web;
    int ClientKey;
    /* Fill a Web structure for the cache query */
-   Web = a_Web_new(url);
+   Web = a_Web_new(url, requester);
    Web->bw = bw;
    Web->Image = Image;
    a_Image_ref(Image);
@@ -2116,6 +2126,7 @@
       a_Bw_add_client(bw, ClientKey, 0);
       a_Bw_add_url(bw, url);
    }
+   return ClientKey != 0;
 }
 
 /*
@@ -2924,7 +2935,7 @@
    } else {
       /* Fill a Web structure for the cache query */
       int ClientKey;
-      DilloWeb *Web = a_Web_new(url);
+      DilloWeb *Web = a_Web_new(url, html->page_url);
       Web->bw = html->bw;
       if ((ClientKey = a_Capi_open_url(Web, Html_css_load_callback, NULL))) {
          ++html->bw->NumPendingStyleSheets;
diff -r e820d7cb2be7 src/nav.c
--- a/src/nav.c	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/nav.c	Mon Dec 28 01:54:09 2009 +0000
@@ -191,7 +191,8 @@
  * This function requests the page's root-URL; images and related stuff
  * are fetched directly by the HTML module.
  */
-static void Nav_open_url(BrowserWindow *bw, const DilloUrl *url, int offset)
+static void Nav_open_url(BrowserWindow *bw, const DilloUrl *url,
+                         const DilloUrl *requester, int offset)
 {
    DilloUrl *old_url;
    bool_t MustLoad, ForceReload, Repush, IgnoreScroll;
@@ -232,7 +233,7 @@
 
       // a_Menu_pagemarks_new(bw);
 
-      Web = a_Web_new(url);
+      Web = a_Web_new(url, requester);
       Web->bw = bw;
       Web->flags |= WEB_RootUrl;
       if ((ClientKey = a_Capi_open_url(Web, NULL, NULL)) != 0) {
@@ -341,7 +342,8 @@
  * - Set bw to expect the URL data
  * - Ask the cache to feed back the requested URL (via Nav_open_url)
  */
-void a_Nav_push(BrowserWindow *bw, const DilloUrl *url)
+void a_Nav_push(BrowserWindow *bw, const DilloUrl *url,
+                                   const DilloUrl *requester)
 {
    dReturn_if_fail (bw != NULL);
 
@@ -353,7 +355,7 @@
    a_Nav_cancel_expect(bw);
    bw->nav_expect_url = a_Url_dup(url);
    bw->nav_expecting = TRUE;
-   Nav_open_url(bw, url, 0);
+   Nav_open_url(bw, url, requester, 0);
 }
 
 /*
@@ -370,7 +372,7 @@
       a_Url_set_flags(url, URL_FLAGS(url) | URL_ReloadFromCache);
       bw->nav_expect_url = a_Url_dup(url);
       bw->nav_expecting = TRUE;
-      Nav_open_url(bw, url, 0);
+      Nav_open_url(bw, url, NULL, 0);
       a_Url_free(url);
    }
 }
@@ -407,7 +409,7 @@
 
    if (bw->meta_refresh_status == 2) {
       Nav_stack_move_ptr(bw, -1);
-      a_Nav_push(bw, bw->meta_refresh_url);
+      a_Nav_push(bw, bw->meta_refresh_url,a_History_get_url(NAV_TOP_UIDX(bw)));
    }
    a_Url_free(bw->meta_refresh_url);
    bw->meta_refresh_url = NULL;
@@ -441,7 +443,7 @@
    a_Nav_cancel_expect(bw);
    if (--idx >= 0){
       a_UIcmd_set_msg(bw, "");
-      Nav_open_url(bw, a_History_get_url(NAV_UIDX(bw,idx)), -1);
+      Nav_open_url(bw, a_History_get_url(NAV_UIDX(bw,idx)), NULL, -1);
    }
 }
 
@@ -455,7 +457,7 @@
    a_Nav_cancel_expect(bw);
    if (++idx < a_Nav_stack_size(bw)) {
       a_UIcmd_set_msg(bw, "");
-      Nav_open_url(bw, a_History_get_url(NAV_UIDX(bw,idx)), +1);
+      Nav_open_url(bw, a_History_get_url(NAV_UIDX(bw,idx)), NULL, +1);
    }
 }
 
@@ -464,7 +466,7 @@
  */
 void a_Nav_home(BrowserWindow *bw)
 {
-   a_Nav_push(bw, prefs.home);
+   a_Nav_push(bw, prefs.home, NULL);
 }
 
 /*
@@ -496,7 +498,7 @@
          a_Url_set_flags(r_url, URL_FLAGS(r_url) & ~URL_SpamSafe);
          bw->nav_expect_url = r_url;
          bw->nav_expecting = TRUE;
-         Nav_open_url(bw, r_url, 0);
+         Nav_open_url(bw, r_url, NULL, 0);
       }
    }
 }
@@ -523,7 +525,7 @@
       a_UIcmd_open_url_nw(bw, a_History_get_url(NAV_UIDX(bw,idx)));
    } else {
       a_Nav_cancel_expect(bw);
-      Nav_open_url(bw, a_History_get_url(NAV_UIDX(bw,idx)), offset);
+      Nav_open_url(bw, a_History_get_url(NAV_UIDX(bw,idx)), NULL, offset);
       a_UIcmd_set_buttons_sens(bw);
    }
 }
@@ -560,7 +562,7 @@
 void a_Nav_save_url(BrowserWindow *bw,
                     const DilloUrl *url, const char *filename)
 {
-   DilloWeb *Web = a_Web_new(url);
+   DilloWeb *Web = a_Web_new(url, NULL);
    Web->bw = bw;
    Web->filename = dStrdup(filename);
    Web->flags |= WEB_Download;
diff -r e820d7cb2be7 src/nav.h
--- a/src/nav.h	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/nav.h	Mon Dec 28 01:54:09 2009 +0000
@@ -14,7 +14,8 @@
 #endif /* __cplusplus */
 
 void a_Nav_redirection0(BrowserWindow *bw, const DilloUrl *new_url);
-void a_Nav_push(BrowserWindow *bw, const DilloUrl *url);
+void a_Nav_push(BrowserWindow *bw, const DilloUrl *url,
+                const DilloUrl *requester);
 void a_Nav_repush(BrowserWindow *bw);
 void a_Nav_back(BrowserWindow *bw);
 void a_Nav_forw(BrowserWindow *bw);
diff -r e820d7cb2be7 src/prefs.c
--- a/src/prefs.c	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/prefs.c	Mon Dec 28 01:54:09 2009 +0000
@@ -39,6 +39,7 @@
    prefs.buffered_drawing = 1;
    prefs.contrast_visited_color = TRUE;
    prefs.enterpress_forces_submit = FALSE;
+   prefs.filter_auto_requests = PREFS_FILTER_ALLOW_ALL;
    prefs.focus_new_tab = TRUE;
    prefs.font_cursive = dStrdup(PREFS_FONT_CURSIVE);
    prefs.font_factor = 1.0;
diff -r e820d7cb2be7 src/prefs.h
--- a/src/prefs.h	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/prefs.h	Mon Dec 28 01:54:09 2009 +0000
@@ -26,6 +26,10 @@
 /* Panel sizes */
 enum { P_tiny = 0, P_small, P_medium, P_large };
 
+enum {PREFS_FILTER_ALLOW_ALL,
+      PREFS_FILTER_SAME_HOST,
+      PREFS_FILTER_SAME_DOMAIN};
+
 typedef struct _DilloPrefs DilloPrefs;
 
 struct _DilloPrefs {
@@ -69,6 +73,7 @@
    bool_t load_images;
    bool_t load_stylesheets;
    bool_t parse_embedded_css;
+   int filter_auto_requests;
    int32_t buffered_drawing;
    char *font_serif;
    char *font_sans_serif;
diff -r e820d7cb2be7 src/prefsparser.cc
--- a/src/prefsparser.cc	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/prefsparser.cc	Mon Dec 28 01:54:09 2009 +0000
@@ -26,6 +26,7 @@
    PREFS_INT32,
    PREFS_DOUBLE,
    PREFS_GEOMETRY,
+   PREFS_FILTER,
    PREFS_PANEL_SIZE
 } PrefType_t;
 
@@ -50,6 +51,7 @@
       { "contrast_visited_color", &prefs.contrast_visited_color, PREFS_BOOL },
       { "enterpress_forces_submit", &prefs.enterpress_forces_submit,
         PREFS_BOOL },
+      { "filter_auto_requests", &prefs.filter_auto_requests, PREFS_FILTER },
       { "focus_new_tab", &prefs.focus_new_tab, PREFS_BOOL },
       { "font_cursive", &prefs.font_cursive, PREFS_STRING },
       { "font_factor", &prefs.font_factor, PREFS_DOUBLE },
@@ -136,6 +138,17 @@
       a_Misc_parse_geometry(value, &prefs.xpos, &prefs.ypos,
                             &prefs.width, &prefs.height);
       break;
+   case PREFS_FILTER:
+      if (!dStrcasecmp(value, "same_host"))
+         prefs.filter_auto_requests = PREFS_FILTER_SAME_HOST;
+      else if (!dStrcasecmp(value, "same_domain"))
+         prefs.filter_auto_requests = PREFS_FILTER_SAME_DOMAIN;
+      else {
+         if (dStrcasecmp(value, "allow_all"))
+            MSG_WARN("prefs: unrecognized value for filter_auto_requests\n");
+         prefs.filter_auto_requests = PREFS_FILTER_ALLOW_ALL;
+      }
+      break;
    case PREFS_PANEL_SIZE:
       if (!dStrcasecmp(value, "tiny"))
          prefs.panel_size = P_tiny;
diff -r e820d7cb2be7 src/uicmd.cc
--- a/src/uicmd.cc	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/uicmd.cc	Mon Dec 28 01:54:09 2009 +0000
@@ -616,7 +616,7 @@
  */
 void a_UIcmd_open_url(BrowserWindow *bw, const DilloUrl *url)
 {
-   a_Nav_push(bw, url);
+   a_Nav_push(bw, url, NULL);
    a_UIcmd_focus_main_area(bw);
 }
 
@@ -626,7 +626,7 @@
     * Location if we don't yet have an URL, main otherwise.
     */
    if (url) {
-      a_Nav_push(new_bw, url);
+      a_Nav_push(new_bw, url, NULL);
       BW2UI(new_bw)->focus_main();
    } else {
       BW2UI(new_bw)->focus_location();
@@ -840,7 +840,7 @@
 
    if (name) {
       url = a_Url_new(name, "file:");
-      a_Nav_push((BrowserWindow*)vbw, url);
+      a_Nav_push((BrowserWindow*)vbw, url, NULL);
       a_Url_free(url);
       dFree(name);
    }
@@ -931,7 +931,7 @@
 void a_UIcmd_book(void *vbw)
 {
    DilloUrl *url = a_Url_new("dpi:/bm/", NULL);
-   a_Nav_push((BrowserWindow*)vbw, url);
+   a_Nav_push((BrowserWindow*)vbw, url, NULL);
    a_Url_free(url);
 }
 
diff -r e820d7cb2be7 src/web.cc
--- a/src/web.cc	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/web.cc	Mon Dec 28 01:54:09 2009 +0000
@@ -103,12 +103,13 @@
 /*
  * Allocate and set safe values for a DilloWeb structure
  */
-DilloWeb* a_Web_new(const DilloUrl *url)
+DilloWeb* a_Web_new(const DilloUrl *url, const DilloUrl *requester)
 {
    DilloWeb *web= dNew(DilloWeb, 1);
 
    _MSG(" a_Web_new: ValidWebs ==> %d\n", dList_length(ValidWebs));
    web->url = a_Url_dup(url);
+   web->requester = a_Url_dup(requester);
    web->bw = NULL;
    web->flags = 0;
    web->Image = NULL;
@@ -136,6 +137,7 @@
 {
    if (!web) return;
    a_Url_free(web->url);
+   a_Url_free(web->requester);
    a_Image_unref(web->Image);
    dFree(web->filename);
    dList_remove(ValidWebs, (void *)web);
diff -r e820d7cb2be7 src/web.hh
--- a/src/web.hh	Sun Dec 27 20:35:17 2009 +0000
+++ b/src/web.hh	Mon Dec 28 01:54:09 2009 +0000
@@ -22,6 +22,8 @@
 
 struct _DilloWeb {
   DilloUrl *url;              /* Requested URL */
+  DilloUrl *requester;        /* URL that caused this request, or
+                               * NULL if user-initiated. */
   BrowserWindow *bw;          /* The requesting browser window [reference] */
   int flags;                  /* Additional info */
 
@@ -34,7 +36,7 @@
 };
 
 void a_Web_init(void);
-DilloWeb* a_Web_new (const DilloUrl* url);
+DilloWeb* a_Web_new (const DilloUrl* url, const DilloUrl *requester);
 int a_Web_valid(DilloWeb *web);
 void a_Web_free (DilloWeb*);
 int a_Web_dispatch_by_type (const char *Type, DilloWeb *web,