实现对Chunk类型的应答内容的解析 #2

3 files changed, 127 insertions, 9 deletions

diff --git a/src/http1.cc b/src/http1.cc
index ac02b58..2011bd6 100644
--- a/src/http1.cc
+++ b/src/http1.cc
@@ -670,17 +670,22 @@ int Http1ResponseParserCallbacks::CallbackOnMessageComplete(http_parser * parser
 
 int Http1ResponseParserCallbacks::CallbackOnChunkHeader(http_parser * parser)
 {
+    auto * resp_ptr = __response_this_ptr(parser);
+    resp_ptr->__set_section_state(resp_ptr->kSectionBody, resp_ptr->kStateReading);
+
     /* On chunk begin, setup chunk tag */
-    __response_this_ptr(parser)->body_is_chunk = true;
-    __response_this_ptr(parser)->body_chunk_id++;
+    resp_ptr->body_is_chunk = true;
+    resp_ptr->body_chunk_id++;
 
     /* Drop old body content */
-    __response_this_ptr(parser)->body_content_ = nullptr;
+    resp_ptr->body_content_ = nullptr;
     return 0;
 }
 
 int Http1ResponseParserCallbacks::CallbackOnChunkComplete(http_parser * parser)
 {
+    auto * resp_ptr = __response_this_ptr(parser);
+    resp_ptr->__set_section_state(resp_ptr->kSectionBody, resp_ptr->kStateComplete);
     return 0;
 }
 
@@ -848,28 +853,36 @@ int Http1Connection::on_connection_read_response(pxy_conn_ctx_t * conn_ctx, pxy_
         conn_ctx->passthrough = 1;
         return 0;
     }
+    else if (forward_len == 0)
+    {
+        return 0;
+    }
 
     if (response.SectionState(response.kSectionHeader) == response.kStateComplete)
     {
         http_session.CallRequestHeaderCallback();
-        goto __forward;
     }
 
     if (response.SectionState(response.kSectionBody) == response.kStateComplete)
     {
         http_session.CallResponseBodyCallback();
-        goto __forward;
     }
 
+    /* 如果有任何一部分为Reading，说明该部分数据还为到来，暂时不转发 */
+    if (response.SectionState(response.kSectionHeader) == response.kStateReading ||
+        response.SectionState(response.kSectionBody) == response.kStateReading)
+    {
+        return 0;
+    }
+
+    /* 转发缓存中的数据，如果该Response处理结束，本Session处理完毕，销毁Session */
+    evbuffer_add_buffer(upstream_evbuf, response.StolenEvBuf().release());
     if (response.SectionState(response.kSectionMessage) == response.kStateComplete)
     {
         drop_last_session();
     }
 
     return 0;
-
-__forward:
-    evbuffer_add_buffer(upstream_evbuf, response.StolenEvBuf().release());
 }
 
 HttpSession & Http1Connection::create_new_session()
diff --git a/src/httpscan.cc b/src/httpscan.cc
index 6366f85..29ee773 100644
--- a/src/httpscan.cc
+++ b/src/httpscan.cc
@@ -232,14 +232,109 @@ void HttpScanSession::ScanRequestBody(HttpSession * http_session_ctx)
 
 void HttpScanSession::ScanResponseHeader(HttpSession * http_session_ctx)
 {
+    auto & response = http_session_ctx->response();
+
+    /* 扫描应答头部 */
+    auto scan_result = scan_headers(response.cHeaders(), httpscan_module_ref_.table_id_ctrl_http_res_hdr);
+
+    /* Hit */
+    if (scan_result == scan_result_t::kScanResultHit)
+    {
+        http_session_ctx->SetResponseHeaderTag(http_session_ctx->kCallbackTagRepeat);
+        return hit_config_and_do_action(http_session_ctx);
+    }
+    /* Error */
+    else if (scan_result == scan_result_t::kScanResultError)
+    {
+        return hit_scan_error();
+    }
+
+    /* Not Hit */
     return;
 }
 
 void HttpScanSession::ScanResponseBody(HttpSession * http_session_ctx)
 {
+    auto & response = http_session_ctx->response();
+
+    /* Body Content and length, prepare for maat */
+    const char * body_content_raw = response.Body()->data();
+    size_t body_content_length = response.Body()->size();
+
+    auto scan_result = scan_body(body_content_raw, body_content_length,
+        httpscan_module_ref_.table_id_ctrl_http_res_body);
+
+    /* Hit */
+    if (scan_result == scan_result_t::kScanResultHit)
+    {
+        http_session_ctx->SetResponseBodyTag(http_session_ctx->kCallbackTagRepeat);
+        return hit_config_and_do_action(http_session_ctx);
+    }
+    /* Error */
+    else if (scan_result == scan_result_t::kScanResultError)
+    {
+        return hit_scan_error();
+    }
+
     return;
 }
 
+HttpScanSession::scan_result_t HttpScanSession::scan_headers(const HttpHeaders & c_headers, int table_id)
+{
+    scan_result_t scan_result = scan_result_t::kScanResultNotHit;
+
+    c_headers.ForEachHeader([this, &scan_result, table_id]
+        (const std::string & field, const std::string & value) -> bool
+    {
+        int ret = Maat_set_scan_status(httpscan_module_ref_.maat_feather_ref, &maat_scan_mid_,
+            MAAT_SET_SCAN_DISTRICT, field.c_str(), (int) field.length());
+
+        if (ret < 0)
+        {
+            scan_result = scan_result_t::kScanResultError;
+            return false;
+        }
+
+        int __dummy[MAAT_SCAN_RESULT_];
+
+        nr_maat_scan_result_ = Maat_full_scan_string(httpscan_module_ref_.maat_feather_ref,
+            table_id, MAAT_DEFAULT_CHARSET_, value.c_str(), (int) value.length(), maat_scan_result_, __dummy,
+            MAAT_SCAN_RESULT_, &maat_scan_mid_, 0);
+
+        if (nr_maat_scan_result_ > 0)
+        {
+            scan_result = scan_result_t::kScanResultHit;
+            return false;
+        }
+        else if (nr_maat_scan_result_ == -1)
+        {
+            scan_result = scan_result_t::kScanResultError;
+            return false;
+        }
+
+        return true;
+    });
+
+    return scan_result;
+}
+
+HttpScanSession::scan_result_t HttpScanSession::scan_body(const char * data, size_t len, int table_id)
+{
+    /* Dummy For maat */
+    int __dummy[MAAT_SCAN_RESULT_];
+
+    nr_maat_scan_result_ = Maat_full_scan_string(httpscan_module_ref_.maat_feather_ref,
+        table_id, CHARSET_UTF8, data, (int) len, maat_scan_result_, __dummy, MAAT_SCAN_RESULT_, &maat_scan_mid_, 0);
+
+    if (nr_maat_scan_result_ > 0)
+        return scan_result_t::kScanResultHit;
+
+    else if (nr_maat_scan_result_ == -1)
+        return scan_result_t::kScanResultError;
+
+    return scan_result_t::kScanResultNotHit;
+}
+
 void HttpScanSession::hit_config_and_do_action(HttpSession * session)
 {
     /* 判断命中数量，若为多命中，选择优先级最高的动作执行 */
@@ -300,4 +395,4 @@ void HttpScanSession::hit_config_and_do_action(HttpSession * session)
 void HttpScanSession::hit_scan_error()
 {
     return;
-}
\ No newline at end of file
+}
diff --git a/src/httpscan.h b/src/httpscan.h
index 6b3fd1f..e3891ce 100644
--- a/src/httpscan.h
+++ b/src/httpscan.h
@@ -68,6 +68,16 @@ public:
     void ScanResponseBody(HttpSession *http_session_ctx);
 
 private:
+    enum class scan_result_t
+    {
+        kScanResultNotHit = 0,
+        kScanResultHit = 1,
+        kScanResultError = -1
+    };
+
+    scan_result_t scan_headers(const HttpHeaders & c_headers, int table_id);
+    scan_result_t scan_body(const char * data, size_t len, int table_id);
+
     void hit_config_and_do_action(HttpSession *session);
     void hit_scan_error();