[HTTP_DECODER]bugfix for gzip decompress

3 files changed, 88 insertions, 149 deletions

diff --git a/src/http_decoder/http_content_decompress.c b/src/http_decoder/http_content_decompress.c
index f57d4cf..b887487 100644
--- a/src/http_decoder/http_content_decompress.c
+++ b/src/http_decoder/http_content_decompress.c
@@ -148,15 +148,14 @@ http_content_decompress_write_zlib(struct http_content_decompress *decompress,
     z_stream *z_stream_ptr = decompress->z_stream_ptr;
     z_stream_ptr->avail_in = (unsigned int)indata_len;
     z_stream_ptr->next_in = (unsigned char *)indata;
+    z_stream_ptr->avail_out = (unsigned int)decompress->buffer_size;
+    z_stream_ptr->next_out = (unsigned char *)decompress->buffer;
 
     *outdata = NULL;
     *outdata_len = 0;
 
     int ret = 0;
     do {
-        z_stream_ptr->avail_out = (unsigned int)decompress->buffer_size;
-        z_stream_ptr->next_out = (unsigned char *)decompress->buffer;
-
         ret = inflate(z_stream_ptr, Z_NO_FLUSH);
         if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT
             || ret == Z_DATA_ERROR || ret == Z_MEM_ERROR) {
@@ -166,23 +165,24 @@ http_content_decompress_write_zlib(struct http_content_decompress *decompress,
 
         size_t have = decompress->buffer_size - z_stream_ptr->avail_out;
         if (have > 0) {
-            if (NULL == *outdata) {
-                // http_decoder_log(DEBUG, "%s alloc outbuffer %zu bytes", 
-                //                  http_content_encoding_int2str(decompress->encoding),
-                //                  have);
-                *outdata = safe_dup(decompress->buffer, have);
-                *outdata_len = have;
-            } else {
-                // http_decoder_log(DEBUG, "%s realloc outbuffer %zu bytes", 
-                //                  http_content_encoding_int2str(decompress->encoding), 
-                //                  *outdata_len + have + 1);
-                *outdata = REALLOC(char, *outdata, *outdata_len + have + 1);
-                memcpy(*outdata + *outdata_len, decompress->buffer, have);
+            if (0 == z_stream_ptr->avail_out) {
+                decompress->buffer_size += have;
+                decompress->buffer = REALLOC(char, decompress->buffer,
+                                             decompress->buffer_size);
+                *outdata = decompress->buffer;
                 *outdata_len = *outdata_len + have;
-                memset(*outdata + *outdata_len, 0, 1);
+                http_decoder_log(DEBUG, "%s realloc outbuffer %zu bytes", 
+                                 http_content_encoding_int2str(decompress->encoding), 
+                                 decompress->buffer_size);
+                z_stream_ptr->avail_out = have;
+                z_stream_ptr->next_out = (unsigned char *)decompress->buffer + 
+                                         (decompress->buffer_size - have);
+            } else {
+                *outdata = decompress->buffer;
+                *outdata_len = have;
             }
         }
-    } while (0 == z_stream_ptr->avail_out);
+    } while (z_stream_ptr->avail_in != 0);
 
     return 0;
 }
@@ -192,37 +192,32 @@ http_content_decompress_write_br(struct http_content_decompress *decompress,
                                  const char *indata, size_t indata_len,
                                  char **outdata, size_t *outdata_len)
 {
-    size_t available_out;
     size_t available_in = indata_len;
     const unsigned char *next_in = (const unsigned char *)indata;
-    unsigned char *next_out;
+    size_t available_out = decompress->buffer_size;
+    unsigned char *next_out = (unsigned char *)decompress->buffer;
 
     *outdata = NULL;
     *outdata_len = 0;
 
     int ret;
     for (;;) {
-        available_out = decompress->buffer_size;
-        next_out = (unsigned char *)decompress->buffer;
-
         ret = BrotliDecoderDecompressStream(decompress->br_state, &available_in, 
                                             &next_in, &available_out, &next_out, 0);
         size_t have = decompress->buffer_size - available_out;
         if (have > 0) {
-            if (NULL == *outdata) {
-                // http_decoder_log(DEBUG, "%s alloc outbuffer %zu bytes",
-                //                 http_content_encoding_int2str(decompress->encoding),
-                //                 have);
-                *outdata = safe_dup(decompress->buffer, have);
-                *outdata_len = have;
-            } else {
-                // http_decoder_log(DEBUG, "%s realloc outbuffer %zu bytes",
-                //                  http_content_encoding_int2str(decompress->encoding),
-                //                  *outdata_len + have + 1);
-                *outdata = REALLOC(char, *outdata, *outdata_len + have + 1);
-                memcpy(*outdata + *outdata_len, decompress->buffer, have);
+            if (0 == available_out) {
+                decompress->buffer_size += have;
+                decompress->buffer = REALLOC(char, decompress->buffer,
+                                             decompress->buffer_size);
+                *outdata = decompress->buffer;
                 *outdata_len = *outdata_len + have;
-                memset(*outdata + *outdata_len, 0, 1);
+                available_out = have;
+                next_out = (unsigned char *)decompress->buffer +
+                           (decompress->buffer_size - have);
+            } else {
+                *outdata = decompress->buffer;
+                *outdata_len = have;
             }
         }
 
diff --git a/src/http_decoder/http_decoder_half.c b/src/http_decoder/http_decoder_half.c
index 40e5d01..d97cd25 100644
--- a/src/http_decoder/http_decoder_half.c
+++ b/src/http_decoder/http_decoder_half.c
@@ -26,7 +26,8 @@ struct http_decoder_half_data {
     int status_code;
 
     enum http_content_encoding content_encoding;
-    char *decompress_body;
+    struct http_content_decompress *decompress;
+    char *ref_decompress_body;
     size_t decompress_body_len;
 };
 
@@ -44,7 +45,7 @@ struct http_decoder_half {
     long long trans_counter;
 };
 
-//#define HTTP_DECODER_DEBUG
+// #define HTTP_DECODER_DEBUG
 #ifdef HTTP_DECODER_DEBUG
 static void printf_debug_info(const char *desc, const char *at, size_t length)
 {
@@ -68,9 +69,6 @@ http_decoder_half_data_decompress(struct http_decoder_half_data *data)
 {
     assert(data);
 
-    FREE(data->decompress_body);
-    data->decompress_body_len = 0;
-
     if (data->content_encoding == HTTP_CONTENT_ENCODING_NONE) {
         return;
     }
@@ -81,15 +79,19 @@ http_decoder_half_data_decompress(struct http_decoder_half_data *data)
         return;
     }
 
-    struct http_content_decompress *decompress =
-        http_content_decompress_create(data->content_encoding);
-    assert(decompress);
-    if (http_content_decompress_write(decompress, raw_body.str, raw_body.str_len,
-                                      &data->decompress_body,
+    if (NULL == data->decompress) {
+        data->decompress = http_content_decompress_create(data->content_encoding);
+    }
+        
+    assert(data->decompress);
+    if (http_content_decompress_write(data->decompress, raw_body.str,
+                                      raw_body.str_len,
+                                      &data->ref_decompress_body,
                                       &data->decompress_body_len) == -1) {
         // log error
+        http_content_decompress_destroy(data->decompress);
+        data->decompress = NULL;
     }
-    http_content_decompress_destroy(decompress);
 }
 
 /* Possible return values 0, -1, `HPE_PAUSED` */
@@ -125,62 +127,20 @@ static int on_message_complete(llhttp_t *http)
         container_of(http, struct http_decoder_half, parser);
     assert(half);
 
-    // if enable  cache_body, trigger body_data/body_end event
-    // if disable cache_body, trigger body_end event
-    if (half->is_cache_body
-        || (half->ref_data->content_encoding != HTTP_CONTENT_ENCODING_NONE)) {
-        if (http_decoder_table_state(half->ref_data->table, HTTP_ITEM_BODY)
-            == STRING_STATE_CACHE) {
-            http_decoder_table_commit(half->ref_data->table, HTTP_ITEM_BODY);
-            http_decoder_half_data_decompress(half->ref_data);
-        }
-
-        if (half->parser.type == HTTP_REQUEST) {
-            if (half->event == HTTP_EVENT_REQ_BODY_BEGIN) {
-                half->event = HTTP_EVENT_REQ_BODY_DATA;
-                if (half->http_ev_cb != NULL) {
-                    half->http_ev_cb(half->event, &half->ref_data,
-                                     half->http_ev_ctx);
-                }
-
-                half->event = HTTP_EVENT_REQ_BODY_END;
-                if (half->http_ev_cb != NULL) {
-                    half->http_ev_cb(half->event, &half->ref_data,
-                                     half->http_ev_ctx);
-                }
-            }
-        } else {
-            if (half->event == HTTP_EVENT_RES_BODY_BEGIN) {
-                half->event = HTTP_EVENT_RES_BODY_DATA;
-                if (half->http_ev_cb != NULL) {
-                    half->http_ev_cb(half->event, &half->ref_data,
-                                     half->http_ev_ctx);
-                }
-
-                half->event = HTTP_EVENT_RES_BODY_END;
-                if (half->http_ev_cb != NULL) {
-                    half->http_ev_cb(half->event, &half->ref_data,
-                                     half->http_ev_ctx);
-                }
+    if (half->parser.type == HTTP_REQUEST) {
+        if (half->event == HTTP_EVENT_REQ_BODY_DATA) {
+            half->event = HTTP_EVENT_REQ_BODY_END;
+            if (half->http_ev_cb != NULL) {
+                half->http_ev_cb(half->event, &half->ref_data,
+                                 half->http_ev_ctx);
             }
         }
     } else {
-        // no cache body
-        if (half->parser.type == HTTP_REQUEST) {
-            if (half->event == HTTP_EVENT_REQ_BODY_DATA) {
-                half->event = HTTP_EVENT_REQ_BODY_END;
-                if (half->http_ev_cb != NULL) {
-                    half->http_ev_cb(half->event, &half->ref_data,
-                                     half->http_ev_ctx);
-                }
-            }
-        } else {
-            if (half->event == HTTP_EVENT_RES_BODY_DATA) {
-                half->event = HTTP_EVENT_RES_BODY_END;
-                if (half->http_ev_cb != NULL) {
-                    half->http_ev_cb(half->event, &half->ref_data,
-                                     half->http_ev_ctx);
-                }
+        if (half->event == HTTP_EVENT_RES_BODY_DATA) {
+            half->event = HTTP_EVENT_RES_BODY_END;
+            if (half->http_ev_cb != NULL) {
+                half->http_ev_cb(half->event, &half->ref_data,
+                                 half->http_ev_ctx);
             }
         }
     }
@@ -510,37 +470,30 @@ static int on_body(llhttp_t *http, const char *at, size_t length)
         }
     }
 
-    // if enable  cache_body, trigger body_data event on_message_complete
-    // if disable cache_body, trigger body_data event on_body
-    if (half->is_cache_body ||
-        (half->ref_data->content_encoding != HTTP_CONTENT_ENCODING_NONE)) {
-        http_decoder_table_refer(half->ref_data->table, HTTP_ITEM_BODY,
-                                 at, length);
-        http_decoder_table_cache(half->ref_data->table, HTTP_ITEM_BODY);
-    } else {
-        if (half->ref_data != NULL) {
-            if (http_decoder_table_state(half->ref_data->table, HTTP_ITEM_BODY)
-                == STRING_STATE_COMMIT) {
-                http_decoder_table_reset(half->ref_data->table, HTTP_ITEM_BODY);
-            }
-
-            http_decoder_table_refer(half->ref_data->table, HTTP_ITEM_BODY,
-                                     at, length);
-            http_decoder_table_commit(half->ref_data->table, HTTP_ITEM_BODY);
+    if (half->ref_data != NULL) {
+        if (http_decoder_table_state(half->ref_data->table, HTTP_ITEM_BODY) ==
+            STRING_STATE_COMMIT) {
+            http_decoder_table_reset(half->ref_data->table, HTTP_ITEM_BODY);
         }
 
-        if (half->parser.type == HTTP_REQUEST) {
-            half->event = HTTP_EVENT_REQ_BODY_DATA;
-            if (half->http_ev_cb) {
-                half->http_ev_cb(half->event, &half->ref_data,
-                                 half->http_ev_ctx);
-            }
-        } else {
-            half->event = HTTP_EVENT_RES_BODY_DATA;
-            if (half->http_ev_cb) {
-                half->http_ev_cb(half->event, &half->ref_data,
-                                 half->http_ev_ctx);
-            }
+        http_decoder_table_refer(half->ref_data->table, HTTP_ITEM_BODY, at,
+                                 length);
+        http_decoder_table_commit(half->ref_data->table, HTTP_ITEM_BODY);
+    }
+
+    if (half->ref_data->content_encoding != HTTP_CONTENT_ENCODING_NONE) {
+        http_decoder_half_data_decompress(half->ref_data);
+    }
+
+    if (half->parser.type == HTTP_REQUEST) {
+        half->event = HTTP_EVENT_REQ_BODY_DATA;
+        if (half->http_ev_cb) {
+            half->http_ev_cb(half->event, &half->ref_data, half->http_ev_ctx);
+        }
+    } else {
+        half->event = HTTP_EVENT_RES_BODY_DATA;
+        if (half->http_ev_cb) {
+            half->http_ev_cb(half->event, &half->ref_data, half->http_ev_ctx);
         }
     }
 
@@ -733,7 +686,7 @@ struct http_decoder_half_data *http_decoder_half_data_new()
     data->status_code = -1;
 
     data->content_encoding = HTTP_CONTENT_ENCODING_NONE;
-    data->decompress_body = NULL;
+    data->ref_decompress_body = NULL;
     data->decompress_body_len = 0;
 
     return data;
@@ -750,8 +703,9 @@ void http_decoder_half_data_free(struct http_decoder_half_data *data)
         data->table = NULL;
     }
 
-    if (data->decompress_body != NULL) {
-        FREE(data->decompress_body);
+    if (data->decompress != NULL) {
+        http_content_decompress_destroy(data->decompress);
+        data->decompress = NULL;
     }
 
     FREE(data);
@@ -834,7 +788,7 @@ int http_decoder_half_data_get_decompress_body(struct http_decoder_half_data *da
     }
 
 
-    body->str = data->decompress_body;
+    body->str = data->ref_decompress_body;
     body->str_len = data->decompress_body_len;
     return 0;
 }
diff --git a/test/http_decoder/http_decoder_gtest.cpp b/test/http_decoder/http_decoder_gtest.cpp
index 725f42d..852844e 100644
--- a/test/http_decoder/http_decoder_gtest.cpp
+++ b/test/http_decoder/http_decoder_gtest.cpp
@@ -40,7 +40,7 @@ int g_res_exdata_idx = 0;
 int g_topic_id = 0;
 
 
-#if 0
+#if 1
 void output_http_req_line(struct http_request_line *req_line)
 {
     char tmp_str[MAX_KEY_STR_LEN] = {0};
@@ -74,28 +74,16 @@ void output_http_header(struct http_header *header)
 #endif
 
 void
-output_http_header(struct http_header *header, int res)
-{
-    char key[MAX_KEY_STR_LEN] = {0};
-
-    snprintf(key, header->key.str_len + 1, "%s", header->key.str);
-    if (0 == res) {
-        printf("<ouput_http_req_header> header:%s\n", key);
-    } else {
-        printf("<ouput_http_res_header> header:%s\n", key);
-    }
-    
-}
-
-void
 output_http_body(struct hstring *body, int decompress_flag)
 {
     int counter = 0;
 
     if (1 == decompress_flag) {
-        printf("\n\n----------------decompress body---------------\n");
+        printf("\n\n----------------decompress body len:%zu---------------\n",
+              body->str_len);
     } else {
-        printf("\n\n----------------raw body---------------\n");
+        printf("\n\n----------------raw body len:%zu---------------\n",
+               body->str_len);
     }
 
     for (size_t i = 0; i < body->str_len; i++) {
@@ -207,6 +195,7 @@ next:
     case HTTP_MESSAGE_REQ_HEADER:
         while (http_message_request_header_next(msg, &header) > 0) {
             http_header_to_json(json, &header);
+            // output_http_header(&header);
         }
         g_header_count = 1;
         break;
@@ -224,6 +213,7 @@ next:
     case HTTP_MESSAGE_RES_HEADER:
         while (http_message_response_header_next(msg, &header) > 0) {
             http_header_to_json(json, &header);
+            // output_http_header(&header);
         }
         g_header_count = 1;
         break;