aboutsummaryrefslogtreecommitdiff
path: root/sql/core
Commit message (Collapse)AuthorAgeFilesLines
* [SPARK-14335][SQL] Describe function command returns wrong outputYong Tang2016-04-091-13/+31
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? …because some of built-in functions are not in function registry. This fix tries to fix issues in `describe function` command where some of the outputs still shows Hive's function because some built-in functions are not in FunctionRegistry. The following built-in functions have been added to FunctionRegistry: ``` - ! * / & % ^ + < <= <=> = == > >= | ~ and in like not or rlike when ``` The following listed functions are not added, but hard coded in `commands.scala` (hvanhovell): ``` != <> between case ``` Below are the existing result of the above functions that have not been added: ``` spark-sql> describe function `!=`; Function: <> Class: org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPNotEqual Usage: a <> b - Returns TRUE if a is not equal to b ``` ``` spark-sql> describe function `<>`; Function: <> Class: org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPNotEqual Usage: a <> b - Returns TRUE if a is not equal to b ``` ``` spark-sql> describe function `between`; Function: between Class: org.apache.hadoop.hive.ql.udf.generic.GenericUDFBetween Usage: between a [NOT] BETWEEN b AND c - evaluate if a is [not] in between b and c ``` ``` spark-sql> describe function `case`; Function: case Class: org.apache.hadoop.hive.ql.udf.generic.GenericUDFCase Usage: CASE a WHEN b THEN c [WHEN d THEN e]* [ELSE f] END - When a = b, returns c; when a = d, return e; else return f ``` ## How was this patch tested? Existing tests passed. Additional test cases added. Author: Yong Tang <yong.tang.github@outlook.com> Closes #12128 from yongtang/SPARK-14335.
* Revert "[SPARK-14419] [SQL] Improve HashedRelation for key fit within Long"Davies Liu2016-04-098-633/+346
| | | | This reverts commit 90c0a04506a4972b7a2ac2b7dda0c5f8509a6e2f.
* [SPARK-14496][SQL] fix some javadoc typosbomeng2016-04-091-1/+1
| | | | | | | | | | | | | ## What changes were proposed in this pull request? Minor issues. Found 2 typos while browsing the code. ## How was this patch tested? None. Author: bomeng <bmeng@us.ibm.com> Closes #12264 from bomeng/SPARK-14496.
* [SPARK-14419] [SQL] Improve HashedRelation for key fit within LongDavies Liu2016-04-098-346/+633
| | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? Currently, we use java HashMap for HashedRelation if the key could fit within a Long. The java HashMap and CompactBuffer are not memory efficient, the memory used by them is also accounted accurately. This PR introduce a LongToUnsafeRowMap (similar to BytesToBytesMap) for better memory efficiency and performance. ## How was this patch tested? Updated existing tests. Author: Davies Liu <davies@databricks.com> Closes #12190 from davies/long_map2.
* [SPARK-14451][SQL] Move encoder definition into Aggregator interfaceReynold Xin2016-04-094-76/+102
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? When we first introduced Aggregators, we required the user of Aggregators to (implicitly) specify the encoders. It would actually make more sense to have the encoders be specified by the implementation of Aggregators, since each implementation should have the most state about how to encode its own data type. Note that this simplifies the Java API because Java users no longer need to explicitly specify encoders for aggregators. ## How was this patch tested? Updated unit tests. Author: Reynold Xin <rxin@databricks.com> Closes #12231 from rxin/SPARK-14451.
* [SPARK-14482][SQL] Change default Parquet codec from gzip to snappyReynold Xin2016-04-084-33/+65
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? Based on our tests, gzip decompression is very slow (< 100MB/s), making queries decompression bound. Snappy can decompress at ~ 500MB/s on a single core. This patch changes the default compression codec for Parquet output from gzip to snappy, and also introduces a ParquetOptions class to be more consistent with other data sources (e.g. CSV, JSON). ## How was this patch tested? Should be covered by existing unit tests. Author: Reynold Xin <rxin@databricks.com> Closes #12256 from rxin/SPARK-14482.
* [SPARK-14498][ML][PYTHON][SQL] Many cleanups to ML and ML-related docsJoseph K. Bradley2016-04-081-0/+4
| | | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? Cleanups to documentation. No changes to code. * GBT docs: Move Scala doc for private object GradientBoostedTrees to public docs for GBTClassifier,Regressor * GLM regParam: needs doc saying it is for L2 only * TrainValidationSplitModel: add .. versionadded:: 2.0.0 * Rename “_transformer_params_from_java” to “_transfer_params_from_java” * LogReg Summary classes: “probability” col should not say “calibrated” * LR summaries: coefficientStandardErrors —> document that intercept stderr comes last. Same for t,p-values * approxCountDistinct: Document meaning of “rsd" argument. * LDA: note which params are for online LDA only ## How was this patch tested? Doc build Author: Joseph K. Bradley <joseph@databricks.com> Closes #12266 from jkbradley/ml-doc-cleanups.
* [SPARK-14454] Better exception handling while marking tasks as failedSameer Agarwal2016-04-081-37/+34
| | | | | | | | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This patch adds support for better handling of exceptions inside catch blocks if the code within the block throws an exception. For instance here is the code in a catch block before this change in `WriterContainer.scala`: ```scala logError("Aborting task.", cause) // call failure callbacks first, so we could have a chance to cleanup the writer. TaskContext.get().asInstanceOf[TaskContextImpl].markTaskFailed(cause) if (currentWriter != null) { currentWriter.close() } abortTask() throw new SparkException("Task failed while writing rows.", cause) ``` If `markTaskFailed` or `currentWriter.close` throws an exception, we currently lose the original cause. This PR fixes this problem by implementing a utility function `Utils.tryWithSafeCatch` that suppresses (`Throwable.addSuppressed`) the exception that are thrown within the catch block and rethrowing the original exception. ## How was this patch tested? No new functionality added Author: Sameer Agarwal <sameer@databricks.com> Closes #12234 from sameeragarwal/fix-exception.
* [SPARK-14394][SQL] Generate AggregateHashMap class for LongTypes during ↵Sameer Agarwal2016-04-082-3/+210
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | TungstenAggregate codegen ## What changes were proposed in this pull request? This PR adds support for generating the `AggregateHashMap` class in `TungstenAggregate` if the aggregate group by keys/value are of `LongType`. Note that currently this generate aggregate is not actually used. NB: This currently only supports `LongType` keys/values (please see `isAggregateHashMapSupported` in `TungstenAggregate`) and will be generalized to other data types in a subsequent PR. ## How was this patch tested? Manually inspected the generated code. This is what the generated map looks like for 2 keys: ```java /* 068 */ public class agg_GeneratedAggregateHashMap { /* 069 */ private org.apache.spark.sql.execution.vectorized.ColumnarBatch batch; /* 070 */ private int[] buckets; /* 071 */ private int numBuckets; /* 072 */ private int maxSteps; /* 073 */ private int numRows = 0; /* 074 */ private org.apache.spark.sql.types.StructType schema = /* 075 */ new org.apache.spark.sql.types.StructType() /* 076 */ .add("k1", org.apache.spark.sql.types.DataTypes.LongType) /* 077 */ .add("k2", org.apache.spark.sql.types.DataTypes.LongType) /* 078 */ .add("sum", org.apache.spark.sql.types.DataTypes.LongType); /* 079 */ /* 080 */ public agg_GeneratedAggregateHashMap(int capacity, double loadFactor, int maxSteps) { /* 081 */ assert (capacity > 0 && ((capacity & (capacity - 1)) == 0)); /* 082 */ this.maxSteps = maxSteps; /* 083 */ numBuckets = (int) (capacity / loadFactor); /* 084 */ batch = org.apache.spark.sql.execution.vectorized.ColumnarBatch.allocate(schema, /* 085 */ org.apache.spark.memory.MemoryMode.ON_HEAP, capacity); /* 086 */ buckets = new int[numBuckets]; /* 087 */ java.util.Arrays.fill(buckets, -1); /* 088 */ } /* 089 */ /* 090 */ public agg_GeneratedAggregateHashMap() { /* 091 */ new agg_GeneratedAggregateHashMap(1 << 16, 0.25, 5); /* 092 */ } /* 093 */ /* 094 */ public org.apache.spark.sql.execution.vectorized.ColumnarBatch.Row findOrInsert(long agg_key, long agg_key1) { /* 095 */ long h = hash(agg_key, agg_key1); /* 096 */ int step = 0; /* 097 */ int idx = (int) h & (numBuckets - 1); /* 098 */ while (step < maxSteps) { /* 099 */ // Return bucket index if it's either an empty slot or already contains the key /* 100 */ if (buckets[idx] == -1) { /* 101 */ batch.column(0).putLong(numRows, agg_key); /* 102 */ batch.column(1).putLong(numRows, agg_key1); /* 103 */ batch.column(2).putLong(numRows, 0); /* 104 */ buckets[idx] = numRows++; /* 105 */ return batch.getRow(buckets[idx]); /* 106 */ } else if (equals(idx, agg_key, agg_key1)) { /* 107 */ return batch.getRow(buckets[idx]); /* 108 */ } /* 109 */ idx = (idx + 1) & (numBuckets - 1); /* 110 */ step++; /* 111 */ } /* 112 */ // Didn't find it /* 113 */ return null; /* 114 */ } /* 115 */ /* 116 */ private boolean equals(int idx, long agg_key, long agg_key1) { /* 117 */ return batch.column(0).getLong(buckets[idx]) == agg_key && batch.column(1).getLong(buckets[idx]) == agg_key1; /* 118 */ } /* 119 */ /* 120 */ // TODO: Improve this Hash Function /* 121 */ private long hash(long agg_key, long agg_key1) { /* 122 */ return agg_key ^ agg_key1; /* 123 */ } /* 124 */ /* 125 */ } ``` Author: Sameer Agarwal <sameer@databricks.com> Closes #12161 from sameeragarwal/tungsten-aggregate.
* [SPARK-14448] Improvements to ColumnVectortedyu2016-04-082-22/+36
| | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? In this PR, two changes are proposed for ColumnVector : 1. ColumnVector should be declared as implementing AutoCloseable - it already has close() method 2. In OnHeapColumnVector#reserveInternal(), we only need to allocate new array when existing array is null or the length of existing array is shorter than the newCapacity. ## How was this patch tested? Existing unit tests. Author: tedyu <yuzhihong@gmail.com> Closes #12225 from tedyu/master.
* [SPARK-14189][SQL] JSON data sources find compatible types even if inferred ↵hyukjinkwon2016-04-082-0/+30
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | decimal type is not capable of the others ## What changes were proposed in this pull request? https://issues.apache.org/jira/browse/SPARK-14189 When inferred types in the same field during finding compatible `DataType`, are `IntegralType` and `DecimalType` but `DecimalType` is not capable of the given `IntegralType`, JSON data source simply fails to find a compatible type resulting in `StringType`. This can be observed when `prefersDecimal` is enabled. ```scala def mixedIntegerAndDoubleRecords: RDD[String] = sqlContext.sparkContext.parallelize( """{"a": 3, "b": 1.1}""" :: """{"a": 3.1, "b": 1}""" :: Nil) val jsonDF = sqlContext.read .option("prefersDecimal", "true") .json(mixedIntegerAndDoubleRecords) .printSchema() ``` - **Before** ``` root |-- a: string (nullable = true) |-- b: string (nullable = true) ``` - **After** ``` root |-- a: decimal(21, 1) (nullable = true) |-- b: decimal(21, 1) (nullable = true) ``` (Note that integer is inferred as `LongType` which becomes `DecimalType(20, 0)`) ## How was this patch tested? unit tests were used and style tests by `dev/run_tests`. Author: hyukjinkwon <gurwls223@gmail.com> Closes #11993 from HyukjinKwon/SPARK-14189.
* [SPARK-14103][SQL] Parse unescaped quotes in CSV data source.hyukjinkwon2016-04-084-1/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This PR resolves the problem during parsing unescaped quotes in input data. For example, currently the data below: ``` "a"b,ccc,ddd e,f,g ``` produces a data below: - **Before** ```bash ["a"b,ccc,ddd[\n]e,f,g] <- as a value. ``` - **After** ```bash ["a"b], [ccc], [ddd] [e], [f], [g] ``` This PR bumps up the Univocity parser's version. This was fixed in `2.0.2`, https://github.com/uniVocity/univocity-parsers/issues/60. ## How was this patch tested? Unit tests in `CSVSuite` and `sbt/sbt scalastyle`. Author: hyukjinkwon <gurwls223@gmail.com> Closes #12226 from HyukjinKwon/SPARK-14103-quote.
* Replace getLocalizedMessage with just normal toString in exception handling ↵Reynold Xin2016-04-071-1/+1
| | | | in WriterContainer.
* [SPARK-14270][SQL] whole stage codegen support for typed filterWenchen Fan2016-04-076-12/+174
| | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? We implement typed filter by `MapPartitions`, which doesn't work well with whole stage codegen. This PR use `Filter` to implement typed filter and we can get the whole stage codegen support for free. This PR also introduced `DeserializeToObject` and `SerializeFromObject`, to seperate serialization logic from object operator, so that it's eaiser to write optimization rules for adjacent object operators. ## How was this patch tested? existing tests. Author: Wenchen Fan <wenchen@databricks.com> Closes #12061 from cloud-fan/whole-stage-codegen.
* [SPARK-14410][SQL] Push functions existence check into catalogAndrew Or2016-04-074-47/+40
| | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This is a followup to #12117 and addresses some of the TODOs introduced there. In particular, the resolution of database is now pushed into session catalog, which knows about the current database. Further, the logic for checking whether a function exists is pushed into the external catalog. No change in functionality is expected. ## How was this patch tested? `SessionCatalogSuite`, `DDLSuite` Author: Andrew Or <andrew@databricks.com> Closes #12198 from andrewor14/function-exists.
* [SPARK-12740] [SPARK-13932] support grouping()/grouping_id() in having/order ↵Davies Liu2016-04-071-0/+82
| | | | | | | | | | | | | | | | | | | | | | clause ## What changes were proposed in this pull request? This PR brings the support of using grouping()/grouping_id() in HAVING/ORDER BY clause. The resolved grouping()/grouping_id() will be replaced by unresolved "spark_gropuing_id" virtual attribute, then resolved by ResolveMissingAttribute. This PR also fix the HAVING clause that access a grouping column that is not presented in SELECT clause, for example: ```sql select count(1) from (select 1 as a) t group by a having a > 0 ``` ## How was this patch tested? Add new tests. Author: Davies Liu <davies@databricks.com> Closes #12235 from davies/grouping_having.
* [SPARK-14456][SQL][MINOR] Remove unused variables and logics in DataSourceKousuke Saruta2016-04-071-10/+0
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? In DataSource#write method, the variables `dataSchema` and `equality`, and related logics are no longer used. Let's remove them. ## How was this patch tested? Existing tests. Author: Kousuke Saruta <sarutak@oss.nttdata.co.jp> Closes #12237 from sarutak/SPARK-14456.
* [SQL][TESTS] Fix for flaky test in ContinuousQueryManagerSuiteTathagata Das2016-04-071-2/+2
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? The timeouts were lower the other timeouts in the test. Other tests were stable over the last month. ## How was this patch tested? Jenkins tests. Author: Tathagata Das <tathagata.das1565@gmail.com> Closes #12219 from tdas/flaky-test-fix.
* [SPARK-10063][SQL] Remove DirectParquetOutputCommitterReynold Xin2016-04-074-157/+5
| | | | | | | | | | | | ## What changes were proposed in this pull request? This patch removes DirectParquetOutputCommitter. This was initially created by Databricks as a faster way to write Parquet data to S3. However, given how the underlying S3 Hadoop implementation works, this committer only works when there are no failures. If there are multiple attempts of the same task (e.g. speculation or task failures or node failures), the output data can be corrupted. I don't think this performance optimization outweighs the correctness issue. ## How was this patch tested? Removed the related tests also. Author: Reynold Xin <rxin@databricks.com> Closes #12229 from rxin/SPARK-10063.
* [SPARK-14452][SQL] Explicit APIs in Scala for specifying encodersReynold Xin2016-04-071-9/+9
| | | | | | | | | | | | ## What changes were proposed in this pull request? The Scala Dataset public API currently only allows users to specify encoders through SQLContext.implicits. This is OK but sometimes people want to explicitly get encoders without a SQLContext (e.g. Aggregator implementations). This patch adds public APIs to Encoders class for getting Scala encoders. ## How was this patch tested? None - I will update test cases once https://github.com/apache/spark/pull/12231 is merged. Author: Reynold Xin <rxin@databricks.com> Closes #12232 from rxin/SPARK-14452.
* [SPARK-12610][SQL] Left Anti JoinHerman van Hovell2016-04-068-97/+201
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | ### What changes were proposed in this pull request? This PR adds support for `LEFT ANTI JOIN` to Spark SQL. A `LEFT ANTI JOIN` is the exact opposite of a `LEFT SEMI JOIN` and can be used to identify rows in one dataset that are not in another dataset. Note that `nulls` on the left side of the join cannot match a row on the right hand side of the join; the result is that left anti join will always select a row with a `null` in one or more of its keys. We currently add support for the following SQL join syntax: SELECT * FROM tbl1 A LEFT ANTI JOIN tbl2 B ON A.Id = B.Id Or using a dataframe: tbl1.as("a").join(tbl2.as("b"), $"a.id" === $"b.id", "left_anti) This PR provides serves as the basis for implementing `NOT EXISTS` and `NOT IN (...)` correlated sub-queries. It would also serve as good basis for implementing an more efficient `EXCEPT` operator. The PR has been (losely) based on PR's by both davies (https://github.com/apache/spark/pull/10706) and chenghao-intel (https://github.com/apache/spark/pull/10563); credit should be given where credit is due. This PR adds supports for `LEFT ANTI JOIN` to `BroadcastHashJoin` (including codegeneration), `ShuffledHashJoin` and `BroadcastNestedLoopJoin`. ### How was this patch tested? Added tests to `JoinSuite` and ported `ExistenceJoinSuite` from https://github.com/apache/spark/pull/10563. cc davies chenghao-intel rxin Author: Herman van Hovell <hvanhovell@questtec.nl> Closes #12214 from hvanhovell/SPARK-12610.
* [SPARK-12555][SQL] Result should not be corrupted after input columns are ↵Luciano Resende2016-04-071-0/+19
| | | | | | | | | | reordered This PR add test case described in SPARK-12555 to validate that correct data is returned when input data is reordered and to avoid future regressions. Author: Luciano Resende <lresende@apache.org> Closes #11623 from lresende/SPARK-12555.
* [SPARK-14436][SQL] Make JavaDatasetAggregatorSuiteBase public.Marcelo Vanzin2016-04-062-53/+83
| | | | | | | | | | Without this, unit tests that extend that class fail for me locally on maven, because JUnit tries to run methods in that class and gets an IllegalAccessError. Author: Marcelo Vanzin <vanzin@cloudera.com> Closes #12212 from vanzin/SPARK-14436.
* [SPARK-14444][BUILD] Add a new scalastyle `NoScalaDoc` to prevent ↵Dongjoon Hyun2016-04-064-11/+11
| | | | | | | | | | | | | | | | | | | | | ScalaDoc-style multiline comments ## What changes were proposed in this pull request? According to the [Spark Code Style Guide](https://cwiki.apache.org/confluence/display/SPARK/Spark+Code+Style+Guide#SparkCodeStyleGuide-Indentation), this PR adds a new scalastyle rule to prevent the followings. ``` /** In Spark, we don't use the ScalaDoc style so this * is not correct. */ ``` ## How was this patch tested? Pass the Jenkins tests (including `lint-scala`). Author: Dongjoon Hyun <dongjoon@apache.org> Closes #12221 from dongjoon-hyun/SPARK-14444.
* [SPARK-14224] [SPARK-14223] [SPARK-14310] [SQL] fix RowEncoder and parquet ↵Davies Liu2016-04-0612-224/+253
| | | | | | | | | | | | | | | | | | | | | reader for wide table ## What changes were proposed in this pull request? 1) fix the RowEncoder for wide table (many columns) by splitting the generate code into multiple functions. 2) Separate DataSourceScan as RowDataSourceScan and BatchedDataSourceScan 3) Disable the returning columnar batch in parquet reader if there are many columns. 4) Added a internal config for maximum number of fields (nested) columns supported by whole stage codegen. Closes #12098 ## How was this patch tested? Add a tests for table with 1000 columns. Author: Davies Liu <davies@databricks.com> Closes #12047 from davies/many_columns.
* [SPARK-14382][SQL] QueryProgress should be post after committedOffsets is ↵Shixiong Zhu2016-04-062-12/+6
| | | | | | | | | | | | | | | | | | updated ## What changes were proposed in this pull request? Make sure QueryProgress is post after committedOffsets is updated. If QueryProgress is post before committedOffsets is updated, the listener may see a wrong sinkStatus (created from committedOffsets). See https://amplab.cs.berkeley.edu/jenkins/job/spark-master-test-maven-hadoop-2.2/644/testReport/junit/org.apache.spark.sql.util/ContinuousQueryListenerSuite/single_listener/ for an example of the failure. ## How was this patch tested? Existing unit tests. Author: Shixiong Zhu <shixiong@databricks.com> Closes #12155 from zsxwing/SPARK-14382.
* [SPARK-14320][SQL] Make ColumnarBatch.Row mutableSameer Agarwal2016-04-065-8/+135
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? In order to leverage a data structure like `AggregateHashMap` (https://github.com/apache/spark/pull/12055) to speed up aggregates with keys, we need to make `ColumnarBatch.Row` mutable. ## How was this patch tested? Unit test in `ColumnarBatchSuite`. Also, tested via `BenchmarkWholeStageCodegen`. Author: Sameer Agarwal <sameer@databricks.com> Closes #12103 from sameeragarwal/mutable-row.
* [SPARK-14383][SQL] missing "|" in the g4 filebomeng2016-04-061-0/+7
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? A very trivial one. It missed "|" between DISTRIBUTE and UNSET. ## How was this patch tested? I do not think it is really needed. Author: bomeng <bmeng@us.ibm.com> Closes #12156 from bomeng/SPARK-14383.
* [SPARK-14429][SQL] Improve LIKE pattern in "SHOW TABLES / FUNCTIONS LIKE ↵bomeng2016-04-061-7/+5
| | | | | | | | | | | | | | | | | | | | | | <pattern>" DDL LIKE <pattern> is commonly used in SHOW TABLES / FUNCTIONS etc DDL. In the pattern, user can use `|` or `*` as wildcards. 1. Currently, we used `replaceAll()` to replace `*` with `.*`, but the replacement was scattered in several places; I have created an utility method and use it in all the places; 2. Consistency with Hive: the pattern is case insensitive in Hive and white spaces will be trimmed, but current pattern matching does not do that. For example, suppose we have tables (t1, t2, t3), `SHOW TABLES LIKE ' T* ' ` will list all the t-tables. Please use Hive to verify it. 3. Combined with `|`, the result will be sorted. For pattern like `' B*|a* '`, it will list the result in a-b order. I've made some changes to the utility method to make sure we will get the same result as Hive does. A new method was created in StringUtil and test cases were added. andrewor14 Author: bomeng <bmeng@us.ibm.com> Closes #12206 from bomeng/SPARK-14429.
* [SPARK-14288][SQL] Memory Sink for streamingMichael Armbrust2016-04-065-18/+159
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This PR exposes the internal testing `MemorySink` though the data source API. This will allow users to easily test streaming applications in the Spark shell or other local tests. Usage: ```scala inputStream.write .format("memory") .queryName("memStream") .startStream() // Now you can query the result of the stream here. sqlContext.table("memStream") ``` The most complicated part of the logic is choosing the checkpoint directory. There are a few requirements we are attempting to satisfy here: - when working in the shell locally, it should just work with no extra configuration. - when working on a cluster you should be able to make it easily create the checkpoint on a distributed file system so you can test aggregation (state checkpoints are also stored in this directory and must be accessible from workers). - it should be clear that you can't resume since the data is just in memory. The chosen algorithm proceeds as follows: - the user gives a checkpoint directory, use it - if the conf has a checkpoint location, use `$location/$queryName` - if neither, create a local directory - always check to make sure there are no offsets written to the directory Author: Michael Armbrust <michael@databricks.com> Closes #12119 from marmbrus/memorySink.
* [SPARK-14396][SQL] Throw Exceptions for DDLs of Partitioned Viewsgatorsmile2016-04-052-36/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | #### What changes were proposed in this pull request? Because the concept of partitioning is associated with physical tables, we disable all the supports of partitioned views, which are defined in the following three commands in [Hive DDL Manual](https://cwiki.apache.org/confluence/display/Hive/LanguageManual+DDL#LanguageManualDDL-Create/Drop/AlterView): ``` ALTER VIEW view DROP [IF EXISTS] PARTITION spec1[, PARTITION spec2, ...]; ALTER VIEW view ADD [IF NOT EXISTS] PARTITION spec; CREATE VIEW [IF NOT EXISTS] [db_name.]view_name [(column_name [COMMENT column_comment], ...) ] [COMMENT view_comment] [TBLPROPERTIES (property_name = property_value, ...)] AS SELECT ...; ``` An exception is thrown when users issue any of these three DDL commands. #### How was this patch tested? Added test cases for parsing create view and changed the existing test cases to verify if the exceptions are thrown. Author: gatorsmile <gatorsmile@gmail.com> Author: xiaoli <lixiao1983@gmail.com> Author: Xiao Li <xiaoli@Xiaos-MacBook-Pro.local> Closes #12169 from gatorsmile/viewPartition.
* [SPARK-14128][SQL] Alter table DDL followupAndrew Or2016-04-052-5/+21
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? This is just a followup to #12121, which implemented the alter table DDLs using the `SessionCatalog`. Specially, this corrects the behavior of setting the location of a datasource table. For datasource tables, we need to set the `locationUri` in addition to the `path` entry in the serde properties. Additionally, changing the location of a datasource table partition is not allowed. ## How was this patch tested? `DDLSuite` Author: Andrew Or <andrew@databricks.com> Closes #12186 from andrewor14/alter-table-ddl-followup.
* [SPARK-14296][SQL] whole stage codegen support for Dataset.mapWenchen Fan2016-04-067-23/+186
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? This PR adds a new operator `MapElements` for `Dataset.map`, it's a 1-1 mapping and is easier to adapt to whole stage codegen framework. ## How was this patch tested? new test in `WholeStageCodegenSuite` Author: Wenchen Fan <wenchen@databricks.com> Closes #12087 from cloud-fan/map.
* [SPARK-14359] Unit tests for java 8 lambda syntax with typed aggregatesEric Liang2016-04-051-41/+45
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? Adds unit tests for java 8 lambda syntax with typed aggregates as a follow-up to #12168 ## How was this patch tested? Unit tests. Author: Eric Liang <ekl@databricks.com> Closes #12181 from ericl/sc-2794-2.
* [SPARK-529][SQL] Modify SQLConf to use new config API from core.Marcelo Vanzin2016-04-054-453/+366
| | | | | | | | | | | | Because SQL keeps track of all known configs, some customization was needed in SQLConf to allow that, since the core API does not have that feature. Tested via existing (and slightly updated) unit tests. Author: Marcelo Vanzin <vanzin@cloudera.com> Closes #11570 from vanzin/SPARK-529-sql.
* [SPARK-14411][SQL] Add a note to warn that onQueryProgress is asynchronousShixiong Zhu2016-04-051-2/+10
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? onQueryProgress is asynchronous so the user may see some future status of `ContinuousQuery`. This PR just updated comments to warn it. ## How was this patch tested? Only updated comments. Author: Shixiong Zhu <shixiong@databricks.com> Closes #12180 from zsxwing/ContinuousQueryListener-doc.
* [SPARK-14129][SPARK-14128][SQL] Alter table DDL commandsAndrew Or2016-04-054-295/+499
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? In Spark 2.0, we want to handle the most common `ALTER TABLE` commands ourselves instead of passing the entire query text to Hive. This is done using the new `SessionCatalog` API introduced recently. The commands supported in this patch include: ``` ALTER TABLE ... RENAME TO ... ALTER TABLE ... SET TBLPROPERTIES ... ALTER TABLE ... UNSET TBLPROPERTIES ... ALTER TABLE ... SET LOCATION ... ALTER TABLE ... SET SERDE ... ``` The commands we explicitly do not support are: ``` ALTER TABLE ... CLUSTERED BY ... ALTER TABLE ... SKEWED BY ... ALTER TABLE ... NOT CLUSTERED ALTER TABLE ... NOT SORTED ALTER TABLE ... NOT SKEWED ALTER TABLE ... NOT STORED AS DIRECTORIES ``` For these we throw exceptions complaining that they are not supported. ## How was this patch tested? `DDLSuite` Author: Andrew Or <andrew@databricks.com> Closes #12121 from andrewor14/alter-table-ddl.
* [SPARK-14402][SQL] initcap UDF doesn't match Hive/Oracle behavior in ↵Dongjoon Hyun2016-04-051-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | lowercasing rest of string ## What changes were proposed in this pull request? Current, SparkSQL `initCap` is using `toTitleCase` function. However, `UTF8String.toTitleCase` implementation changes only the first letter and just copy the other letters: e.g. sParK --> SParK. This is the correct implementation `toTitleCase`. ``` hive> select initcap('sParK'); Spark ``` ``` scala> sql("select initcap('sParK')").head res0: org.apache.spark.sql.Row = [SParK] ``` This PR updates the implementation of `initcap` using `toLowerCase` and `toTitleCase`. ## How was this patch tested? Pass the Jenkins tests (including new testcase). Author: Dongjoon Hyun <dongjoon@apache.org> Closes #12175 from dongjoon-hyun/SPARK-14402.
* [SPARK-14353] Dataset Time Window `window` API for Python, and SQLBurak Yavuz2016-04-052-6/+60
| | | | | | | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? The `window` function was added to Dataset with [this PR](https://github.com/apache/spark/pull/12008). This PR adds the Python, and SQL, API for this function. With this PR, SQL, Java, and Scala will share the same APIs as in users can use: - `window(timeColumn, windowDuration)` - `window(timeColumn, windowDuration, slideDuration)` - `window(timeColumn, windowDuration, slideDuration, startTime)` In Python, users can access all APIs above, but in addition they can do - In Python: `window(timeColumn, windowDuration, startTime=...)` that is, they can provide the startTime without providing the `slideDuration`. In this case, we will generate tumbling windows. ## How was this patch tested? Unit tests + manual tests Author: Burak Yavuz <brkyvz@gmail.com> Closes #12136 from brkyvz/python-windows.
* [SPARK-14123][SPARK-14384][SQL] Handle CreateFunction/DropFunctionYin Huai2016-04-0510-50/+194
| | | | | | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This PR implements CreateFunction and DropFunction commands. Besides implementing these two commands, we also change how to manage functions. Here are the main changes. * `FunctionRegistry` will be a container to store all functions builders and it will not actively load any functions. Because of this change, we do not need to maintain a separate registry for HiveContext. So, `HiveFunctionRegistry` is deleted. * SessionCatalog takes care the job of loading a function if this function is not in the `FunctionRegistry` but its metadata is stored in the external catalog. For this case, SessionCatalog will (1) load the metadata from the external catalog, (2) load all needed resources (i.e. jars and files), (3) create a function builder based on the function definition, (4) register the function builder in the `FunctionRegistry`. * A `UnresolvedGenerator` is created. So, the parser will not need to call `FunctionRegistry` directly during parsing, which is not a good time to create a Hive UDTF. In the analysis phase, we will resolve `UnresolvedGenerator`. This PR is based on viirya's https://github.com/apache/spark/pull/12036/ ## How was this patch tested? Existing tests and new tests. ## TODOs [x] Self-review [x] Cleanup [x] More tests for create/drop functions (we need to more tests for permanent functions). [ ] File JIRAs for all TODOs [x] Standardize the error message when a function does not exist. Author: Yin Huai <yhuai@databricks.com> Author: Liang-Chi Hsieh <simonh@tw.ibm.com> Closes #12117 from yhuai/function.
* [SPARK-14257][SQL] Allow multiple continuous queries to be started from the ↵Shixiong Zhu2016-04-0510-26/+118
| | | | | | | | | | | | | | | | same DataFrame ## What changes were proposed in this pull request? Make StreamingRelation store the closure to create the source in StreamExecution so that we can start multiple continuous queries from the same DataFrame. ## How was this patch tested? `test("DataFrame reuse")` Author: Shixiong Zhu <shixiong@databricks.com> Closes #12049 from zsxwing/df-reuse.
* [SPARK-14349][SQL] Issue Error Messages for Unsupported Operators/DML/DDL in ↵gatorsmile2016-04-051-0/+23
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | SQL Context. #### What changes were proposed in this pull request? Currently, the weird error messages are issued if we use Hive Context-only operations in SQL Context. For example, - When calling `Drop Table` in SQL Context, we got the following message: ``` Expected exception org.apache.spark.sql.catalyst.parser.ParseException to be thrown, but java.lang.ClassCastException was thrown. ``` - When calling `Script Transform` in SQL Context, we got the message: ``` assertion failed: No plan for ScriptTransformation [key#9,value#10], cat, [tKey#155,tValue#156], null +- LogicalRDD [key#9,value#10], MapPartitionsRDD[3] at beforeAll at BeforeAndAfterAll.scala:187 ``` Updates: Based on the investigation from hvanhovell , the root cause is `visitChildren`, which is the default implementation. It always returns the result of the last defined context child. After merging the code changes from hvanhovell , it works! Thank you hvanhovell ! #### How was this patch tested? A few test cases are added. Not sure if the same issue exist for the other operators/DDL/DML. hvanhovell Author: gatorsmile <gatorsmile@gmail.com> Author: xiaoli <lixiao1983@gmail.com> Author: Herman van Hovell <hvanhovell@questtec.nl> Author: Xiao Li <xiaoli@Xiaos-MacBook-Pro.local> Closes #12134 from gatorsmile/hiveParserCommand.
* [SPARK-14348][SQL] Support native execution of SHOW TBLPROPERTIES commandDilip Biswal2016-04-053-8/+81
| | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This PR adds Native execution of SHOW TBLPROPERTIES command. Command Syntax: ``` SQL SHOW TBLPROPERTIES table_name[(property_key_literal)] ``` ## How was this patch tested? Tests added in HiveComandSuiie and DDLCommandSuite Author: Dilip Biswal <dbiswal@us.ibm.com> Closes #12133 from dilipbiswal/dkb_show_tblproperties.
* [SPARK-14359] Create built-in functions for typed aggregates in JavaEric Liang2016-04-053-0/+124
| | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This adds the corresponding Java static functions for built-in typed aggregates already exposed in Scala. ## How was this patch tested? Unit tests. rxin Author: Eric Liang <ekl@databricks.com> Closes #12168 from ericl/sc-2794.
* [SPARK-14287] isStreaming method for DatasetBurak Yavuz2016-04-042-0/+33
| | | | | | | | | | | | | | | | | | With the addition of StreamExecution (ContinuousQuery) to Datasets, data will become unbounded. With unbounded data, the execution of some methods and operations will not make sense, e.g. `Dataset.count()`. A simple API is required to check whether the data in a Dataset is bounded or unbounded. This will allow users to check whether their Dataset is in streaming mode or not. ML algorithms may check if the data is unbounded and throw an exception for example. The implementation of this method is simple, however naming it is the challenge. Some possible names for this method are: - isStreaming - isContinuous - isBounded - isUnbounded I've gone with `isStreaming` for now. We can change it before Spark 2.0 if we decide to come up with a different name. For that reason I've marked it as `Experimental` Author: Burak Yavuz <brkyvz@gmail.com> Closes #12080 from brkyvz/is-streaming.
* [SPARK-14259] [SQL] Merging small files together based on the cost of openingDavies Liu2016-04-043-19/+21
| | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This PR basically re-do the things in #12068 but with a different model, which should work better in case of small files with different sizes. ## How was this patch tested? Updated existing tests. Ran a query on thousands of partitioned small files locally, with all default settings (the cost to open a file should be over estimated), the durations of tasks become smaller and smaller, which is good (the last few tasks will be shortest). Author: Davies Liu <davies@databricks.com> Closes #12095 from davies/file_cost.
* [SPARK-14334] [SQL] add toLocalIterator for Dataset/DataFrameDavies Liu2016-04-044-11/+60
| | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? RDD.toLocalIterator() could be used to fetch one partition at a time to reduce the memory usage. Right now, for Dataset/Dataframe we have to use df.rdd.toLocalIterator, which is super slow also requires lots of memory (because of the Java serializer or even Kyro serializer). This PR introduce an optimized toLocalIterator for Dataset/DataFrame, which is much faster and requires much less memory. For a partition with 5 millions rows, `df.rdd.toIterator` took about 100 seconds, but df.toIterator took less than 7 seconds. For 10 millions row, rdd.toIterator will crash (not enough memory) with 4G heap, but df.toLocalIterator could finished in 12 seconds. The JDBC server has been updated to use DataFrame.toIterator. ## How was this patch tested? Existing tests. Author: Davies Liu <davies@databricks.com> Closes #12114 from davies/local_iterator.
* [SPARK-12981] [SQL] extract Pyhton UDF in physical planDavies Liu2016-04-046-69/+55
| | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? Currently we extract Python UDFs into a special logical plan EvaluatePython in analyzer, But EvaluatePython is not part of catalyst, many rules have no knowledge of it , which will break many things (for example, filter push down or column pruning). We should treat Python UDFs as normal expressions, until we want to evaluate in physical plan, we could extract them in end of optimizer, or physical plan. This PR extract Python UDFs in physical plan. Closes #10935 ## How was this patch tested? Added regression tests. Author: Davies Liu <davies@databricks.com> Closes #12127 from davies/py_udf.
* [SPARK-14176][SQL] Add DataFrameWriter.trigger to set the stream batch periodShixiong Zhu2016-04-049-13/+413
| | | | | | | | | | | | | | ## What changes were proposed in this pull request? Add a processing time trigger to control the batch processing speed ## How was this patch tested? Unit tests Author: Shixiong Zhu <shixiong@databricks.com> Closes #11976 from zsxwing/trigger.
* [SPARK-14137] [SQL] Cleanup hash joinDavies Liu2016-04-046-401/+268
| | | | | | | | | | | | | | | | | | | | | ## What changes were proposed in this pull request? This PR did a few cleanup on HashedRelation and HashJoin: 1) Merge HashedRelation and UniqueHashedRelation together 2) Return an iterator from HashedRelation, so we donot need a create many UnsafeRow objects. 3) Return a copy of HashedRelation for thread-safety in BroadcastJoin, so we can re-use the UnafeRow objects. 4) Cleanup HashJoin, share most of the code between BroadcastHashJoin and ShuffleHashJoin 5) Removed UniqueLongHashedRelation, which will be replaced by LongUnsafeMap (another PR). 6) Update benchmark, before this patch, the selectivity of joins are too high. ## How was this patch tested? Existing tests. Author: Davies Liu <davies@databricks.com> Closes #12102 from davies/cleanup_hash.
generated by cgit v1.2.3 (git 2.39.1) at 2024-08-18 09:12:32 +0000