From: Vsevolod Stakhov <vsevolod@rspamd.com>
Date: Mon, 4 Nov 2024 16:03:34 +0000 (+0000)
Subject: [Fix] Another story about char sign
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=3548785ebdc0e9a9aefc782c85bccb08456e8f3f;p=rspamd.git

[Fix] Another story about char sign

Ragel still produces a wrong state machine if `const char *` is used
for characters representation that are actually utf8.

This PR changes all types to `unsigned char *` for consistency.
---

diff --git a/src/ragel/smtp_addr_parser.rl b/src/ragel/smtp_addr_parser.rl
index b5b4863d3..330f3f01d 100644
--- a/src/ragel/smtp_addr_parser.rl
+++ b/src/ragel/smtp_addr_parser.rl
@@ -8,33 +8,33 @@
   action IP4_end {}
 
   action User_start {
-    addr->user = p;
+    addr->user = (const char *)p;
   }
 
   action User_end {
     if (addr->user) {
-      addr->user_len = p - addr->user;
+      addr->user_len = (const char *)p - addr->user;
     }
   }
 
   action Domain_start {
-    addr->domain = p;
+    addr->domain = (const char *)p;
   }
 
   action Domain_end {
     if (addr->domain) {
-      addr->domain_len = p - addr->domain;
+      addr->domain_len = (const char *)p - addr->domain;
     }
   }
 
   action Domain_addr_start {
-    addr->domain = p;
+    addr->domain = (const char *)p;
     addr->flags |= RSPAMD_EMAIL_ADDR_IP;
   }
 
   action Domain_addr_end {
     if (addr->domain) {
-      addr->domain_len = p - addr->domain;
+      addr->domain_len = (const char *)p - addr->domain;
     }
   }
 
@@ -64,12 +64,12 @@
   }
 
   action Addr_start {
-    addr->addr = p;
+    addr->addr = (const char *)p;
   }
 
   action Addr_end {
     if (addr->addr) {
-      addr->addr_len = p - addr->addr;
+      addr->addr_len = (const char *)p - addr->addr;
     }
   }
 
@@ -87,7 +87,7 @@
 int
 rspamd_smtp_addr_parse (const char *data, size_t len, struct rspamd_email_address *addr)
 {
-  const char *p = data, *pe = data + len, *eof;
+  const unsigned char *p = (const unsigned char *)data, *pe = (const unsigned char *)data + len, *eof;
   int cs;
 
   g_assert (addr != NULL);
diff --git a/test/lua/unit/smtp_addr.lua b/test/lua/unit/smtp_addr.lua
index 2cb7755f8..ffabd838a 100644
--- a/test/lua/unit/smtp_addr.lua
+++ b/test/lua/unit/smtp_addr.lua
@@ -5,7 +5,7 @@ context("SMTP address check functions", function()
   local ffi = require("ffi")
   local util = require("rspamd_util")
   local fun = require "fun"
-  ffi.cdef[[
+  ffi.cdef [[
   struct rspamd_email_address {
     const char *raw;
     const char *addr;
@@ -24,29 +24,30 @@ context("SMTP address check functions", function()
   ]]
 
   local cases_valid = {
-    {'<>', {addr = ''}},
-    {'<a@example.com>', {user = 'a', domain = 'example.com', addr = 'a@example.com'}},
-    {'<a-b@example.com>', {user = 'a-b', domain = 'example.com', addr = 'a-b@example.com'}},
-    {'<a-b@ex-ample.com>', {user = 'a-b', domain = 'ex-ample.com', addr = 'a-b@ex-ample.com'}},
-    {'1367=dec2a6ce-81bd-4fa9-ad02-ec5956466c04=9=1655370@example.220-volt.ru',
-     {user = '1367=dec2a6ce-81bd-4fa9-ad02-ec5956466c04=9=1655370',
-      domain = 'example.220-volt.ru',
-      addr = '1367=dec2a6ce-81bd-4fa9-ad02-ec5956466c04=9=1655370@example.220-volt.ru'}},
-    {'notification+kjdm---m7wwd@facebookmail.com', {user = 'notification+kjdm---m7wwd'}},
-    {'a@example.com', {user = 'a', domain = 'example.com', addr = 'a@example.com'}},
-    {'a+b@example.com', {user = 'a+b', domain = 'example.com', addr = 'a+b@example.com'}},
-    {'"a"@example.com', {user = 'a', domain = 'example.com', addr = 'a@example.com'}},
-    {'"a+b"@example.com', {user = 'a+b', domain = 'example.com', addr = 'a+b@example.com'}},
-    {'"<>"@example.com', {user = '<>', domain = 'example.com', addr = '<>@example.com'}},
-    {'<"<>"@example.com>', {user = '<>', domain = 'example.com', addr = '<>@example.com'}},
-    {'"\\""@example.com', {user = '"', domain = 'example.com', addr = '"@example.com'}},
-    {'"\\"abc"@example.com', {user = '"abc', domain = 'example.com', addr = '"abc@example.com'}},
-    {'<@domain1,@domain2,@domain3:abc@example.com>',
-     {user = 'abc', domain = 'example.com', addr = 'abc@example.com'}},
+    { '<>', { addr = '' } },
+    { '<a@example.com>', { user = 'a', domain = 'example.com', addr = 'a@example.com' } },
+    { '<a-b@example.com>', { user = 'a-b', domain = 'example.com', addr = 'a-b@example.com' } },
+    { '<a-b@ex-ample.com>', { user = 'a-b', domain = 'ex-ample.com', addr = 'a-b@ex-ample.com' } },
+    { '1367=dec2a6ce-81bd-4fa9-ad02-ec5956466c04=9=1655370@example.220-volt.ru',
+      { user = '1367=dec2a6ce-81bd-4fa9-ad02-ec5956466c04=9=1655370',
+        domain = 'example.220-volt.ru',
+        addr = '1367=dec2a6ce-81bd-4fa9-ad02-ec5956466c04=9=1655370@example.220-volt.ru' } },
+    { 'notification+kjdm---m7wwd@facebookmail.com', { user = 'notification+kjdm---m7wwd' } },
+    { 'a@example.com', { user = 'a', domain = 'example.com', addr = 'a@example.com' } },
+    { 'a+b@example.com', { user = 'a+b', domain = 'example.com', addr = 'a+b@example.com' } },
+    { '"a"@example.com', { user = 'a', domain = 'example.com', addr = 'a@example.com' } },
+    { '"a+b"@example.com', { user = 'a+b', domain = 'example.com', addr = 'a+b@example.com' } },
+    { '"<>"@example.com', { user = '<>', domain = 'example.com', addr = '<>@example.com' } },
+    { '<"<>"@example.com>', { user = '<>', domain = 'example.com', addr = '<>@example.com' } },
+    { '"\\""@example.com', { user = '"', domain = 'example.com', addr = '"@example.com' } },
+    { '"\\"abc"@example.com', { user = '"abc', domain = 'example.com', addr = '"abc@example.com' } },
+    { '<@domain1,@domain2,@domain3:abc@example.com>',
+      { user = 'abc', domain = 'example.com', addr = 'abc@example.com' } },
+    -- SMTP UTF8
+    { 'ñ@example.com', { user = 'ñ', domain = 'example.com' } }
 
   }
 
-
   fun.each(function(case)
     test("Parse valid smtp addr: " .. case[1], function()
       local st = ffi.C.rspamd_email_address_from_smtp(case[1], #case[1])
@@ -69,17 +70,17 @@ context("SMTP address check functions", function()
     end)
   end, cases_valid)
 
-    local cases_invalid = {
-      'a',
-      'a"b"@example.com',
-      'a"@example.com',
-      '"a@example.com',
-      '<a@example.com',
-      'a@example.com>',
-      '<a@.example.com>',
-      '<a@example.com>>',
-      '<a@example.com><>',
-    }
+  local cases_invalid = {
+    'a',
+    'a"b"@example.com',
+    'a"@example.com',
+    '"a@example.com',
+    '<a@example.com',
+    'a@example.com>',
+    '<a@.example.com>',
+    '<a@example.com>>',
+    '<a@example.com><>',
+  }
 
   fun.each(function(case)
     test("Parse invalid smtp addr: " .. case, function()
@@ -95,7 +96,7 @@ context("SMTP address check functions", function()
       local niter = 100000
       local total = 0
 
-      for i = 1,niter do
+      for i = 1, niter do
         local ncase = string.format(case, i)
         local t1 = util.get_ticks()
         local st = ffi.C.rspamd_email_address_from_smtp(ncase, #ncase)