[GitHub] [spark] dongjoon-hyun commented on a change in pull request #27366: [SPARK-30648][SQL] Support filters pushdown in JSON datasource

[GitBox <gi...@apache.org>]

dongjoon-hyun commented on a change in pull request #27366:
URL: https://github.com/apache/spark/pull/27366#discussion_r455159539



##########
File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/json/JsonFilters.scala
##########
@@ -0,0 +1,157 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql.catalyst.json
+
+import org.apache.spark.sql.catalyst.{InternalRow, StructFilters}
+import org.apache.spark.sql.catalyst.expressions._
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.sql.sources
+import org.apache.spark.sql.types.StructType
+
+/**
+ * The class provides API for applying pushed down source filters to rows with
+ * a struct schema parsed from JSON records. The class should be used in this way:
+ * 1. Before processing of the next row, `JacksonParser` (parser for short) resets the internal
+ *    state of `JsonFilters` by calling the `reset()` method.
+ * 2. The parser reads JSON fields one-by-one in streaming fashion. It converts an incoming
+ *    field value to the desired type from the schema. After that, it sets the value to an instance
+ *    of `InternalRow` at the position according to the schema. Order of parsed JSON fields can
+ *    be different from the order in the schema.
+ * 3. Per every JSON field of the top-level JSON object, the parser calls `skipRow` by passing
+ *    an `InternalRow` in which some of fields can be already set, and the position of the JSON
+ *    field according to the schema.
+ *    3.1 `skipRow` finds a group of predicates that refers to this JSON field.
+ *    3.2 Per each predicate from the group, `skipRow` decrements its reference counter.
+ *    3.2.1 If predicate reference counter becomes 0, it means that all predicate attributes have
+ *          been already set in the internal row, and the predicate can be applied to it. `skipRow`
+ *          invokes the predicate for the row.
+ *    3.3 `skipRow` applies predicates until one of them returns `false`. In that case, the method
+ *        returns `true` to the parser.
+ *    3.4 If all predicates with zero reference counter return `true`, the final result of
+ *        the method is `false` which tells the parser to not skip the row.
+ * 4. If the parser gets `true` from `JsonFilters.skipRow`, it must not call the method anymore
+ *    for this internal row, and should go the step 1.
+ *
+ * `JsonFilters` assumes that:
+ *   - `reset()` is called before any `skipRow()` calls for new row.
+ *   - `skipRow()` can be called for any valid index of the struct fields,
+ *      and in any order.
+ *   - After `skipRow()` returns `true`, the internal state of `JsonFilters` can be inconsistent,
+ *     so, `skipRow()` must not be called for the current row anymore without `reset()`.

Review comment:
       The above three assumption looks like a global assumption for all `StructFilters`. Also, some of them is already described in `StructFilters` function description. If we have multiple places, it becomes outdated easily. Shall we remove here and mention the global rule in `StructFilters` class description?




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org