fix(cpp): preserve table time statistics in filtered metadata

cpp/src/file/tsfile_io_reader.cc

@@ -399,16 +399,14 @@ int TsFileIOReader::get_device_timeseries_meta_without_chunk_meta(
     }
     std::shared_ptr<IMetaIndexEntry> meta_index_entry;
     int64_t end_offset;
-    std::vector<std::pair<std::shared_ptr<IMetaIndexEntry>, int64_t>>
-        meta_index_entry_list;
+    // Keep the shared time index in filtered metadata so table statistics
+    // can derive null counts from the footer.
     if (RET_FAIL(load_device_index_entry(
             std::make_shared<DeviceIDComparable>(device_id), meta_index_entry,
             end_offset))) {
-    } else if (RET_FAIL(load_all_measurement_index_entry(
-                   meta_index_entry->get_offset(), end_offset, pa,
-                   meta_index_entry_list))) {
-    } else if (RET_FAIL(do_load_all_timeseries_index(meta_index_entry_list, pa,
-                                                     timeseries_indexs))) {
+    } else if (RET_FAIL(get_device_timeseries_meta_by_offset(
+                   meta_index_entry->get_offset(), end_offset,
+                   timeseries_indexs, pa))) {
     }
     return ret;
 }

cpp/test/reader/tsfile_reader_test.cc

@@ -826,6 +826,13 @@ TEST_F(TsFileReaderTest, GetTimeseriesMetadataTableModelTypeAndDeviceFilter) {
     auto selected_list = selected_meta.begin()->second;
     std::unordered_map<std::string, TSDataType> type_by_measurement;
     for (const auto& index : selected_list) {
+        auto* aligned =
+            dynamic_cast<storage::AlignedTimeseriesIndex*>(index.get());
+        ASSERT_NE(aligned, nullptr)
+            << "filtered metadata must preserve the shared time index";
+        ASSERT_NE(aligned->time_ts_idx_, nullptr);
+        ASSERT_NE(aligned->time_ts_idx_->get_statistic(), nullptr);
+        EXPECT_EQ(aligned->time_ts_idx_->get_statistic()->get_count(), 5);
         type_by_measurement[index->get_measurement_name().to_std_string()] =
             index->get_data_type();
     }

cpp/test/tools/command_e2e_test.cc

@@ -165,11 +165,55 @@ TEST(CliE2E, StatsReportsCountAndTimeRange) {
               std::string::npos)
         << out.str();
     EXPECT_NE(out.str().find("table,table1,id1_field_1,id2_field_2,s1,INT64,"
-                             "5,\\N,0,4,0,40,0,40,\\N,statistics"),
+                             "5,0,0,4,0,40,0,40,\\N,statistics"),
               std::string::npos)
         << out.str();
 }
 
+TEST(CliE2E, StatsReportsNumericNullCountForTableFields) {
+    // Filtered metadata must preserve the shared time statistic so the
+    // footer's row count and each field's non-null count determine null_count.
+    const std::string csv_path =
+        tsfile_cli_test::unique_temp_path("tsfile_cli_null_count", ".csv");
+    const std::string out_path =
+        tsfile_cli_test::unique_temp_path("tsfile_cli_null_count", ".tsfile");
+    {
+        std::ofstream csv(csv_path.c_str(), std::ios::binary);
+        ASSERT_TRUE(csv.good());
+        csv << "time,site,temp,note\n"
+               "1000,beijing,20,n0\n"
+               "2000,shanghai,21,n1\n"
+               "3000,beijing,22,\\N\n"
+               "4000,shanghai,23,n3\n";
+    }
+    std::ostringstream wout, werr;
+    ASSERT_EQ(
+        tsfile_cli::run_cli({"write", "--table", "sensors", "--tag", "site",
+                             "STRING", "--field", "temp", "FLOAT", "--field",
+                             "note", "TEXT", "-i", csv_path, "-o", out_path},
+                            wout, werr),
+        0)
+        << werr.str();
+
+    std::ostringstream out, err;
+    ASSERT_EQ(tsfile_cli::run_cli({"stats", "-t", "sensors", "-m", "note", "-f",
+                                   "csv", out_path},
+                                  out, err),
+              0)
+        << err.str();
+    EXPECT_TRUE(err.str().empty()) << err.str();
+    EXPECT_EQ(out.str(),
+              "model,object,tag.site,field,data_type,non_null_count,null_count,"
+              "min_time,max_time,min,max,first,last,sum,stats_source\n"
+              "table,sensors,beijing,note,TEXT,1,1,1000,1000,\\N,\\N,n0,n0,"
+              "\\N,statistics\n"
+              "table,sensors,shanghai,note,TEXT,2,0,2000,4000,\\N,\\N,n1,n3,"
+              "\\N,statistics\n");
+
+    std::remove(csv_path.c_str());
+    std::remove(out_path.c_str());
+}
+
 TEST(CliE2E, HeadProjectsAndLimits) {
     Fixture f;
     std::ostringstream out;
@@ -692,7 +736,7 @@ TEST(CliE2E, MetadataTableFilterIsCaseInsensitive) {
                             stats_out, stats_err),
         0);
     EXPECT_NE(stats_out.str().find("table,table1,id1_field_1,id2_field_2,s1,"
-                                   "INT64,5,\\N,0,4,0,40,0,40,\\N,statistics"),
+                                   "INT64,5,0,0,4,0,40,0,40,\\N,statistics"),
               std::string::npos)
         << stats_out.str();
 }

cpp/test/tools/golden/table_stats_csv.txt

@@ -1,5 +1,5 @@
 model,object,tag.site,tag.rack,tag.sensor,field,data_type,non_null_count,null_count,min_time,max_time,min,max,first,last,sum,stats_source
-table,table1,\N,\N,\N,value,INT64,20,\N,0,19,0,19,0,19,\N,statistics
+table,table1,\N,\N,\N,value,INT64,20,0,0,19,0,19,0,19,\N,statistics
 table,table2,\N,rack0,sensor0,reading,DOUBLE,0,1,0,0,\N,\N,\N,\N,\N,scan
 table,table2,"",rack1,sensor1,reading,DOUBLE,1,0,1,1,1.25,1.25,1.25,1.25,1.25,statistics
 table,table2,null,rack2,sensor2,reading,DOUBLE,1,0,2,2,2.5,2.5,2.5,2.5,2.5,statistics

cpp/test/tools/golden/table_stats_ndjson.txt

@@ -1,4 +1,4 @@
-{"model":"table","object":"table1","tag.site":null,"tag.rack":null,"tag.sensor":null,"field":"value","data_type":"INT64","non_null_count":"20","null_count":null,"min_time":"0","max_time":"19","min":"0","max":"19","first":"0","last":"19","sum":null,"stats_source":"statistics"}
+{"model":"table","object":"table1","tag.site":null,"tag.rack":null,"tag.sensor":null,"field":"value","data_type":"INT64","non_null_count":"20","null_count":"0","min_time":"0","max_time":"19","min":"0","max":"19","first":"0","last":"19","sum":null,"stats_source":"statistics"}
 {"model":"table","object":"table2","tag.site":null,"tag.rack":"rack0","tag.sensor":"sensor0","field":"reading","data_type":"DOUBLE","non_null_count":"0","null_count":"1","min_time":"0","max_time":"0","min":null,"max":null,"first":null,"last":null,"sum":null,"stats_source":"scan"}
 {"model":"table","object":"table2","tag.site":"","tag.rack":"rack1","tag.sensor":"sensor1","field":"reading","data_type":"DOUBLE","non_null_count":"1","null_count":"0","min_time":"1","max_time":"1","min":1.25,"max":1.25,"first":1.25,"last":1.25,"sum":1.25,"stats_source":"statistics"}
 {"model":"table","object":"table2","tag.site":"null","tag.rack":"rack2","tag.sensor":"sensor2","field":"reading","data_type":"DOUBLE","non_null_count":"1","null_count":"0","min_time":"2","max_time":"2","min":2.5,"max":2.5,"first":2.5,"last":2.5,"sum":2.5,"stats_source":"statistics"}

cpp/test/tools/golden/table_stats_table.txt

@@ -1,5 +1,5 @@
 model  object  tag.site  tag.rack  tag.sensor  field    data_type  non_null_count  null_count  min_time  max_time  min    max    first  last   sum    stats_source
-table  table1                                  value    INT64      20                          0         19        0      19     0      19            statistics
+table  table1                                  value    INT64      20              0           0         19        0      19     0      19            statistics
 table  table2            rack0     sensor0     reading  DOUBLE     0               1           0         0                                            scan
 table  table2            rack1     sensor1     reading  DOUBLE     1               0           1         1         1.25   1.25   1.25   1.25   1.25   statistics
 table  table2  null      rack2     sensor2     reading  DOUBLE     1               0           2         2         2.5    2.5    2.5    2.5    2.5    statistics