[perf] fix some minor bugs in json-lines processing

Author: tstack

src/log_format.cc

@@ -1138,11 +1138,12 @@ struct json_log_userdata {
                 == field_frag)
         {
             auto retval
-                = format->elf_value_def_read_order[this->jlu_read_order_index++]
+                = format->elf_value_def_read_order[this->jlu_read_order_index]
                       .second;
             if (retval != nullptr) {
                 this->jlu_precision += 1;
             }
+            this->jlu_read_order_index += 1;
             return retval;
         }
 
@@ -1407,7 +1408,9 @@ json_array_end(void* ctx)
 {
     auto* ypc = (yajlpp_parse_context*) ctx;
     auto* jlu = (json_log_userdata*) ypc->ypc_userdata;
-    const auto* vd = jlu->get_field_def(ypc);
+    const auto* vd = ypc->ypc_path_index_stack.size() > 1
+        ? jlu->get_field_def(ypc)
+        : nullptr;
 
     if (ypc->ypc_path_index_stack.size() == 1 || vd != nullptr) {
         intern_string_t field_name;
@@ -1438,7 +1441,9 @@ read_array_end(void* ctx)
 {
     auto* ypc = (yajlpp_parse_context*) ctx;
     auto* jlu = (json_log_userdata*) ypc->ypc_userdata;
-    const auto* vd = jlu->get_field_def(ypc);
+    const auto* vd = ypc->ypc_path_index_stack.size() > 1
+        ? jlu->get_field_def(ypc)
+        : nullptr;
 
     if (ypc->ypc_path_index_stack.size() == 1 || vd != nullptr) {
         const intern_string_t field_name = ypc->get_path_fragment_i(0);

src/log_format_ext.hh

@@ -343,7 +343,11 @@ public:
     value_defs_state elf_specialized_value_defs_state;
 
     std::vector<std::shared_ptr<value_def>> elf_value_def_order;
-    robin_hood::unordered_map<string_fragment, value_def*, frag_hasher>
+    robin_hood::unordered_map<string_fragment,
+                              value_def*,
+                              frag_hasher,
+                              std::equal_to<string_fragment>,
+                              50>
         elf_value_def_frag_map;
     std::vector<std::pair<string_fragment, value_def*>>
         elf_value_def_read_order;

src/log_vtab_impl.cc

@@ -2529,11 +2529,6 @@ vt_update(sqlite3_vtab* tab,
             argc, argv, 2 + vt->footer_index(log_footer_columns::annotations));
         auto log_opid = from_sqlite<std::optional<string_fragment>>()(
             argc, argv, 2 + vt->footer_index(log_footer_columns::opid));
-        const auto log_user_opid
-            = from_sqlite<std::optional<string_fragment>>()(
-                argc,
-                argv,
-                2 + vt->footer_index(log_footer_columns::user_opid));
         bookmark_metadata tmp_bm;
         parsed_tags tmp_tags;
 

src/yajl/yajl_encode.c

@@ -122,15 +122,13 @@ void yajl_string_decode(yajl_buf buf, const unsigned char * str,
     props->has_ansi = 0;
     props->line_feeds = 0;
     while (end < len) {
-        if (str[end] == '\x1b') {
-            props->has_ansi = 1;
-            end++;
-        } else if (str[end] == '/' || str[end] == '#' || str[end] == '~') {
+        if (str[end] == '/' || str[end] == '#' || str[end] == '~') {
             props->ptr_escapes += 1;
             end++;
         } else if (str[end] == '\\') {
             char utf8Buf[5];
             const char * unescaped = "?";
+            unsigned int unescapedLen = 1;
             yajl_buf_append(buf, str + beg, end - beg);
             switch (str[++end]) {
                 case 'r': unescaped = "\r"; break;
@@ -157,7 +155,6 @@ void yajl_string_decode(yajl_buf buf, const unsigned char * str,
                                    | (surrogate & 0x3FF));
                             end += 5;
                         } else {
-                            unescaped = "?";
                             break;
                         }
                     } else if (codepoint == 0x1b) {
@@ -168,6 +165,10 @@ void yajl_string_decode(yajl_buf buf, const unsigned char * str,
 
                     Utf32toUtf8(codepoint, utf8Buf);
                     unescaped = utf8Buf;
+                    if (codepoint < 0x80) unescapedLen = 1;
+                    else if (codepoint < 0x800) unescapedLen = 2;
+                    else if (codepoint < 0x10000) unescapedLen = 3;
+                    else unescapedLen = 4;
 
                     if (codepoint == 0) {
                         yajl_buf_append(buf, unescaped, 1);
@@ -180,7 +181,7 @@ void yajl_string_decode(yajl_buf buf, const unsigned char * str,
                 default:
                     assert("this should never happen" == NULL);
             }
-            yajl_buf_append(buf, unescaped, (unsigned int)strlen(unescaped));
+            yajl_buf_append(buf, unescaped, unescapedLen);
             beg = ++end;
         } else {
             end++;

src/yajl/yajl_parser.c

@@ -384,7 +384,7 @@ yajl_do_parse(yajl_handle hand, const unsigned char * jsonText,
             yajl_need_key(hand->lexer, 1);
             tok = yajl_lex_lex(hand->lexer, jsonText, jsonTextLen,
                                offset, &buf, &bufLen, &str_props);
-            yajl_need_key(hand->lexer, 1);
+            yajl_need_key(hand->lexer, 0);
             switch (tok) {
                 case yajl_tok_eof:
                     return yajl_status_ok;