[Feature] Add C->Lua URL filter consultation infrastructure

- Add rspamd_url_lua_consult() helper function in url.c
- Function calls lua_url_filter.filter_url_string() from C
- Returns ACCEPT/SUSPICIOUS/REJECT to guide C parser
- Add filter_url_string() function in lua_url_filter.lua
- Validates URL strings and rejects obvious garbage
- Checks: length, @ count, user field size, control chars, UTF-8
- Add TODO comment at oversized user field check (line 1204)
- Infrastructure ready, needs lua_State plumbing through call chain
- This completes the two-level architecture design

diff --git a/lualib/lua_url_filter.lua b/lualib/lua_url_filter.lua
index adb7fac1f48fbd008b032abf7ca5ba2b9193fc55..98ed571c95bd78d6168bd422656e2ad6c7e132e7 100644 (file)
--- a/lualib/lua_url_filter.lua
+++ b/lualib/lua_url_filter.lua
@@ -177,4 +177,54 @@ function exports.register_custom_filter(name, func)
   return true
 end
 
+-- Function called from C parser when encountering suspicious URL patterns
+-- This is called DURING parsing when C is unsure how to proceed
+-- @param url_str: URL string fragment (may be partial URL being parsed)
+-- @param flags: Current parsing flags from C
+-- @return 0=ACCEPT (continue), 1=SUSPICIOUS (mark obscured), 2=REJECT (abort)
+function exports.filter_url_string(url_str, flags)
+  if not url_str or #url_str == 0 then
+    return exports.ACCEPT
+  end
+
+  -- Quick rejection of obviously malicious patterns
+  if #url_str > 2048 then
+    return exports.REJECT -- Absurdly long URL
+  end
+
+  -- Count @ signs (excessive indicates obfuscation)
+  local at_count = select(2, url_str:gsub("@", ""))
+  if at_count > 20 then
+    return exports.REJECT -- Too many @ signs
+  end
+
+  -- Check for extremely long user field
+  local user = url_str:match("^[^:/@]*://([^:/@]+)@") or url_str:match("^([^@]+)@")
+  if user then
+    if #user > 512 then
+      return exports.REJECT -- Absurdly long user field
+    elseif #user > 128 then
+      return exports.SUSPICIOUS -- Long user field, mark for inspection
+    end
+  end
+
+  -- Check for null bytes or other control characters (except tab/newline)
+  if url_str:find("[\0-\8\11-\31\127]") then
+    return exports.REJECT -- Control characters
+  end
+
+  -- Basic UTF-8 validation (reject obviously broken)
+  local ok = pcall(function()
+    -- Try to iterate UTF-8 codepoints
+    for _ in url_str:gmatch("[%z\1-\127\194-\244][\128-\191]*") do
+    end
+  end)
+  if not ok then
+    return exports.REJECT -- Invalid UTF-8
+  end
+
+  -- Allow through - looks reasonable enough to continue parsing
+  return exports.ACCEPT
+end
+
 return exports

diff --git a/src/libserver/url.c b/src/libserver/url.c
index 6add598a43703ed86a36c9f706b96602297d3aa6..5a6a9f55096ff6758d97da8dc7c5104c144c0560 100644 (file)
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -22,6 +22,7 @@
 #include "multipattern.h"
 #include "contrib/uthash/utlist.h"
 #include "contrib/http-parser/http_parser.h"
+#include "lua/lua_common.h"
 #include <unicode/utf8.h>
 #include <unicode/uchar.h>
 #include <unicode/usprep.h>
@@ -1200,7 +1201,10 @@ rspamd_web_parse(struct http_parser_url *u, const char *str, gsize len,
                        }
                        else if (p - c > max_email_user) {
                                /* Allow oversized user fields but mark them - fixes #5731 */
-                               /* Don't fail completely, just mark with flag and continue */
+                               /* TODO: Call rspamd_url_lua_consult(pool, c, p-c, *flags, L) here
+                                * to ask Lua if we should continue parsing this URL.
+                                * Returns: 0=continue, 1=mark obscured, 2=abort (goto out)
+                                * Challenge: need lua_State *L passed through call chain */
                                *flags |= RSPAMD_URL_FLAG_HAS_USER;
                                /* Continue parsing - the Lua plugin will handle scoring */
                        }
@@ -2189,6 +2193,64 @@ rspamd_url_remove_dots(struct rspamd_url *uri)
        return ret;
 }
 
+/**
+ * Consult Lua filter when C parser encounters suspicious/ambiguous URL patterns
+ * This is called DURING parsing when C is unsure how to proceed
+ * @param pool Memory pool
+ * @param url_str URL string fragment being examined
+ * @param len Length of the fragment
+ * @param flags Current URL parsing flags
+ * @param L Lua state (may be NULL)
+ * @return 0=ACCEPT (continue), 1=SUSPICIOUS (mark obscured), 2=REJECT (abort)
+ */
+static int
+rspamd_url_lua_consult(rspamd_mempool_t *pool,
+                                          const char *url_str,
+                                          gsize len,
+                                          unsigned int flags,
+                                          lua_State *L)
+{
+       int result = 0; /* Default: ACCEPT */
+       int err_idx, ret;
+
+       if (!L) {
+               return 0; /* No Lua available, accept */
+       }
+
+       lua_pushcfunction(L, &rspamd_lua_traceback);
+       err_idx = lua_gettop(L);
+
+       /* Try to load lua_url_filter.filter_url_string function */
+       if (!rspamd_lua_require_function(L, "lua_url_filter", "filter_url_string")) {
+               lua_pop(L, 1); /* Remove error handler */
+               return 0;      /* Filter not available, accept */
+       }
+
+       /* Push arguments: url_string, flags */
+       lua_pushlstring(L, url_str, len);
+       lua_pushinteger(L, flags);
+
+       /* Call filter_url_string(url_str, flags) */
+       if ((ret = lua_pcall(L, 2, 1, err_idx)) != 0) {
+               msg_err("cannot call lua_url_filter.filter_url_string: %s",
+                               lua_isstring(L, -1) ? lua_tostring(L, -1) : "unknown error");
+               lua_pop(L, 2); /* Error + error handler */
+               return 0;      /* On error, accept */
+       }
+
+       /* Get result */
+       if (lua_isnumber(L, -1)) {
+               result = lua_tointeger(L, -1);
+               /* Clamp to valid range */
+               if (result < 0) result = 0;
+               if (result > 2) result = 2;
+       }
+
+       lua_pop(L, 2); /* Result + error handler */
+
+       return result;
+}
+
 enum uri_errno
 rspamd_url_parse(struct rspamd_url *uri,
                                 char *uristring, gsize len,