ja*: Fix preserve logic to check for any header in fingerprint group

plugins/experimental/ja4_fingerprint/plugin.cc

@@ -31,11 +31,13 @@
 #include <openssl/ssl.h>
 
 #include <arpa/inet.h>
+#include <getopt.h>
 #include <netinet/in.h>
 
 #include <cstddef>
 #include <cstdint>
 #include <cstdio>
+#include <cstring>
 #include <memory>
 #include <string>
 #include <string_view>
@@ -79,8 +81,37 @@ constexpr int          SSL_SUCCESS{1};
 
 DbgCtl dbg_ctl{PLUGIN_NAME};
 
+int global_preserve_enabled{0};
+
 } // end anonymous namespace
 
+static bool
+read_config_option(int argc, char const *argv[], int &preserve)
+{
+  const struct option longopts[] = {
+    {"preserve", no_argument, &preserve, 1},
+    {nullptr,    0,           nullptr,   0}
+  };
+
+  optind = 0;
+  int opt{0};
+  while ((opt = getopt_long(argc, const_cast<char *const *>(argv), "", longopts, nullptr)) >= 0) {
+    switch (opt) {
+    case '?':
+      Dbg(dbg_ctl, "Unrecognized command argument.");
+    case 0:
+    case -1:
+      break;
+    default:
+      Dbg(dbg_ctl, "Unexpected options error.");
+      return false;
+    }
+  }
+
+  Dbg(dbg_ctl, "JA4 preserve is %s", (preserve == 1) ? "enabled" : "disabled");
+  return true;
+}
+
 static int *
 get_user_arg_index()
 {
@@ -112,12 +143,16 @@ make_word(unsigned char lowbyte, unsigned char highbyte)
 }
 
 void
-TSPluginInit(int /* argc ATS_UNUSED */, char const ** /* argv ATS_UNUSED */)
+TSPluginInit(int argc, char const **argv)
 {
   if (!register_plugin()) {
     TSError("[%s] Failed to register.", PLUGIN_NAME);
     return;
   }
+  if (!read_config_option(argc, argv, global_preserve_enabled)) {
+    TSError("[%s] Failed to parse options.", PLUGIN_NAME);
+    return;
+  }
   reserve_user_arg();
   if (!create_log_file()) {
     TSError("[%s] Failed to create log.", PLUGIN_NAME);
@@ -334,12 +369,36 @@ handle_read_request_hdr(TSCont cont, TSEvent event, void *edata)
   return TS_SUCCESS;
 }
 
+// Check if a header field exists in the request.
+static bool
+header_exists(TSMBuffer bufp, TSMLoc hdr_loc, char const *field, int field_len)
+{
+  TSMLoc loc = TSMimeHdrFieldFind(bufp, hdr_loc, field, field_len);
+  if (loc != TS_NULL_MLOC) {
+    TSHandleMLocRelease(bufp, hdr_loc, loc);
+    return true;
+  }
+  return false;
+}
+
 void
 append_JA4_headers(TSCont /* cont ATS_UNUSED */, TSHttpTxn txnp, std::string const *fingerprint)
 {
   TSMBuffer bufp;
   TSMLoc    hdr_loc;
-  if (TS_SUCCESS == TSHttpTxnClientReqGet(txnp, &bufp, &hdr_loc)) {
+  if (TS_SUCCESS != TSHttpTxnClientReqGet(txnp, &bufp, &hdr_loc)) {
+    Dbg(dbg_ctl, "Failed to get headers.");
+    return;
+  }
+
+  // When preserve is enabled, check if ANY JA4 header exists. If so, skip
+  // adding ALL JA4 headers to avoid mismatched fingerprint data when requests
+  // traverse multiple proxies.
+  bool const ja4_header_exists = header_exists(bufp, hdr_loc, "ja4", 3) ||
+                                 header_exists(bufp, hdr_loc, JA4_VIA_HEADER.data(), static_cast<int>(JA4_VIA_HEADER.length()));
+  bool const skip_ja4_headers = global_preserve_enabled && ja4_header_exists;
+
+  if (!skip_ja4_headers) {
     append_to_field(bufp, hdr_loc, "ja4", 3, fingerprint->data(), fingerprint->size());
 
     TSMgmtString proxy_name = nullptr;
@@ -351,9 +410,6 @@ append_JA4_headers(TSCont /* cont ATS_UNUSED */, TSHttpTxn txnp, std::string con
     append_to_field(bufp, hdr_loc, JA4_VIA_HEADER.data(), static_cast<int>(JA4_VIA_HEADER.length()), proxy_name,
                     static_cast<int>(std::strlen(proxy_name)));
     TSfree(proxy_name);
-
-  } else {
-    Dbg(dbg_ctl, "Failed to get headers.");
   }
 
   TSHandleMLocRelease(bufp, TS_NULL_MLOC, hdr_loc);

plugins/ja3_fingerprint/ja3_fingerprint.cc

@@ -194,6 +194,18 @@ append_to_field(TSMBuffer bufp, TSMLoc hdr_loc, const char *field, int field_len
   TSHandleMLocRelease(bufp, hdr_loc, target);
 }
 
+// Check if a header field exists in the request.
+static bool
+header_exists(TSMBuffer bufp, TSMLoc hdr_loc, char const *field, int field_len)
+{
+  TSMLoc loc = TSMimeHdrFieldFind(bufp, hdr_loc, field, field_len);
+  if (loc != TS_NULL_MLOC) {
+    TSHandleMLocRelease(bufp, hdr_loc, loc);
+    return true;
+  }
+  return false;
+}
+
 static ja3_data *
 create_ja3_data(TSVConn const ssl_vc)
 {
@@ -258,23 +270,31 @@ modify_ja3_headers(TSCont contp, TSHttpTxn txnp, ja3_data const *ja3_vconn_data)
     TSAssert(TS_SUCCESS == TSHttpTxnServerReqGet(txnp, &bufp, &hdr_loc));
   }
 
-  TSMgmtString proxy_name = nullptr;
-  if (TS_SUCCESS != TSMgmtStringGet("proxy.config.proxy_name", &proxy_name)) {
-    TSError("[%s] Failed to get proxy name for %s, set 'proxy.config.proxy_name' in records.config", PLUGIN_NAME,
-            JA3_VIA_HEADER.data());
-    proxy_name = TSstrdup("unknown");
-  }
-  append_to_field(bufp, hdr_loc, JA3_VIA_HEADER.data(), static_cast<int>(JA3_VIA_HEADER.length()), proxy_name,
-                  static_cast<int>(std::strlen(proxy_name)), preserve_flag);
-  TSfree(proxy_name);
+  // When preserve is enabled, check if ANY JA3 header exists. If so, skip
+  // adding ALL JA3 headers to avoid mismatched fingerprint data when requests
+  // traverse multiple proxies.
+  bool const ja3_header_exists = header_exists(bufp, hdr_loc, JA3_VIA_HEADER.data(), static_cast<int>(JA3_VIA_HEADER.length())) ||
+                                 header_exists(bufp, hdr_loc, "x-ja3-sig", 9) || header_exists(bufp, hdr_loc, "x-ja3-raw", 9);
+  bool const skip_ja3_headers = preserve_flag && ja3_header_exists;
+
+  if (!skip_ja3_headers) {
+    TSMgmtString proxy_name = nullptr;
+    if (TS_SUCCESS != TSMgmtStringGet("proxy.config.proxy_name", &proxy_name)) {
+      TSError("[%s] Failed to get proxy name for %s, set 'proxy.config.proxy_name' in records.config", PLUGIN_NAME,
+              JA3_VIA_HEADER.data());
+      proxy_name = TSstrdup("unknown");
+    }
+    append_to_field(bufp, hdr_loc, JA3_VIA_HEADER.data(), static_cast<int>(JA3_VIA_HEADER.length()), proxy_name,
+                    static_cast<int>(std::strlen(proxy_name)), false);
 
-  // Add JA3 md5 fingerprints
-  append_to_field(bufp, hdr_loc, "x-ja3-sig", 9, ja3_vconn_data->md5_string, 32, preserve_flag);
+    // Add JA3 md5 fingerprints.
+    append_to_field(bufp, hdr_loc, "x-ja3-sig", 9, ja3_vconn_data->md5_string, 32, false);
 
-  // If raw string is configured, added JA3 raw string to header as well
-  if (raw_flag) {
-    append_to_field(bufp, hdr_loc, "x-ja3-raw", 9, ja3_vconn_data->ja3_string.data(), ja3_vconn_data->ja3_string.size(),
-                    preserve_flag);
+    // If raw string is configured, add JA3 raw string to header as well.
+    if (raw_flag) {
+      append_to_field(bufp, hdr_loc, "x-ja3-raw", 9, ja3_vconn_data->ja3_string.data(), ja3_vconn_data->ja3_string.size(), false);
+    }
+    TSfree(proxy_name);
   }
   TSHandleMLocRelease(bufp, TS_NULL_MLOC, hdr_loc);
 

tests/gold_tests/pluginTest/ja3_fingerprint/ja3_fingerprint.test.py

@@ -81,13 +81,15 @@ def _configure_server(self, tr: 'TestRun') -> None:
         self._server.Streams.All += Testers.ContainsExpression("https-request", "Verify the HTTPS request was received.")
         self._server.Streams.All += Testers.ContainsExpression("http2-request", "Verify the HTTP/2 request was received.")
         if not self._test_remap:
-            # Verify --preserve worked.
-            self._server.Streams.All += Testers.ContainsExpression("x-ja3-raw: .*,", "Verify the new raw header was added.")
+            # The first request has no existing JA3 headers, so headers are added.
             self._server.Streams.All += Testers.ContainsExpression(
-                "x-ja3-raw: first-signature", "Verify the already-existing raw header was preserved.")
-            self._server.Streams.All += Testers.ExcludesExpression(
-                "x-ja3-raw: first-signature;", "Verify no extra values were added due to preserve.")
+                "x-ja3-raw: .*,", "Verify the new raw header was added.", reflags=re.IGNORECASE)
             self._server.Streams.All += Testers.ContainsExpression("x-ja3-via: test.proxy.com", "The x-ja3-via string was added.")
+            # The second request has existing JA3 headers. With --preserve,
+            # no new JA3 headers are added (including x-ja3-sig). The replay
+            # file verifies x-ja3-sig is absent for the http2 transaction.
+            self._server.Streams.All += Testers.ContainsExpression(
+                "x-ja3-raw: first-signature", "Verify the already-existing raw header was preserved.", reflags=re.IGNORECASE)
 
     def _configure_trafficserver(self) -> None:
         """Configure Traffic Server to be used in the test."""
@@ -189,8 +191,10 @@ def _verify_internal_headers(self) -> None:
 
         if self._modify_incoming:
             p.Streams.All += "modify-incoming-proxy.gold"
+        elif self._test_remap:
+            p.Streams.All += "modify-sent-proxy-remap.gold"
         else:
-            p.Streams.All += "modify-sent-proxy.gold"
+            p.Streams.All += "modify-sent-proxy-global.gold"
 
 
 JA3FingerprintTest(test_remap=False, modify_incoming=False)

tests/gold_tests/pluginTest/ja3_fingerprint/ja3_fingerprint_global.replay.yaml

@@ -86,13 +86,14 @@ sessions:
       content:
         size: 399
 
+    # With --preserve and any JA3 header present, no new headers are added.
     proxy-request:
       headers:
         fields:
         - [ x-request, { value: 'http2-request', as: equal } ]
         - [ x-ja3-via, { value: 'first-via', as: equal } ]
-        - [ X-JA3-Sig, { as: present } ]
-        - [ X-JA3-Raw, { as: present } ]
+        - [ x-ja3-raw, { value: 'first-signature', as: equal } ]
+        - [ x-ja3-sig, { as: absent } ]
 
     server-response:
       headers:

tests/gold_tests/pluginTest/ja3_fingerprint/modify-incoming-proxy.gold

@@ -2,5 +2,6 @@
 -- State Machine Id``
 POST /some/path/http2``
 ``
-x-ja3-sig: ``
+x-ja3-raw: first-signature``
+x-ja3-via: first-via``
 ``

tests/gold_tests/pluginTest/ja3_fingerprint/modify-sent-proxy-global.gold

@@ -0,0 +1,7 @@
++++++++++ Proxy's Request after hooks +++++++++
+-- State Machine Id``
+POST /some/path/http2``
+``
+x-ja3-raw: first-signature``
+x-ja3-via: first-via``
+``

tests/gold_tests/pluginTest/ja3_fingerprint/modify-sent-proxy.gold → tests/gold_tests/pluginTest/ja3_fingerprint/modify-sent-proxy-remap.gold

@@ -3,4 +3,5 @@
 POST /some/path/http2``
 ``
 x-ja3-sig: ``
+x-ja3-via: ``
 ``

tests/gold_tests/pluginTest/ja4_fingerprint/ja4_fingerprint.replay.yaml

@@ -18,6 +18,8 @@ meta:
   version: "1.0"
 
 sessions:
+
+# Session 1: No pre-existing JA4 headers - new headers should be added.
 - protocol:
   - name: http
     version: 1
@@ -33,11 +35,12 @@ sessions:
         fields:
         - [ Connection, keep-alive ]
         - [ Content-Length, 0 ]
+        - [ uuid, no-existing-headers ]
 
     proxy-request:
       headers:
         fields:
-        - [ ja4, { as: contains } ]
+        - [ ja4, { as: present } ]
         - [ x-ja4-via, { value: 'test.proxy.com', as: equal } ]
 
     server-response:
@@ -46,3 +49,70 @@ sessions:
       content:
         encoding: plain
         data: Yay!
+
+# Session 2: Pre-existing JA4 headers - with preserve, no new headers added.
+- protocol:
+  - name: http
+    version: 1
+  - name: tcp
+  - name: ip
+
+  transactions:
+  - client-request:
+      method: "GET"
+      version: "1.1"
+      url: /resource-with-headers
+      headers:
+        fields:
+        - [ Connection, keep-alive ]
+        - [ Content-Length, 0 ]
+        - [ uuid, existing-ja4-headers ]
+        - [ ja4, upstream-fingerprint ]
+        - [ x-ja4-via, upstream.proxy.com ]
+
+    # With --preserve and existing JA4 headers, no new headers should be added.
+    proxy-request:
+      headers:
+        fields:
+        - [ ja4, { value: 'upstream-fingerprint', as: equal } ]
+        - [ x-ja4-via, { value: 'upstream.proxy.com', as: equal } ]
+
+    server-response:
+      status: 200
+      reason: OK
+      content:
+        encoding: plain
+        data: Preserved!
+
+# Session 3: Only x-ja4-via exists - should still trigger preserve for all.
+- protocol:
+  - name: http
+    version: 1
+  - name: tcp
+  - name: ip
+
+  transactions:
+  - client-request:
+      method: "GET"
+      version: "1.1"
+      url: /resource-via-only
+      headers:
+        fields:
+        - [ Connection, keep-alive ]
+        - [ Content-Length, 0 ]
+        - [ uuid, existing-via-only ]
+        - [ x-ja4-via, upstream.proxy.com ]
+
+    # With --preserve and only x-ja4-via present, no JA4 headers should be added.
+    proxy-request:
+      headers:
+        fields:
+        - [ ja4, { as: absent } ]
+        - [ x-ja4-via, { value: 'upstream.proxy.com', as: equal } ]
+
+    server-response:
+      status: 200
+      reason: OK
+      content:
+        encoding: plain
+        data: Via only!