Generic File Source Options

These generic options/configurations are effective only when using file-based sources: parquet, orc, avro, json, csv, text.

Please note that the hierarchy of directories used in examples below are:

  1. dir1/
  2.  ├── dir2/
  3.  │    └── file2.parquet (schema: <file: string>, content: "file2.parquet")
  4.  └── file1.parquet (schema: <file, string>, content: "file1.parquet")
  5.  └── file3.json (schema: <file, string>, content: "{'file':'corrupt.json'}")

Ignore Corrupt Files

Spark allows you to use the configuration spark.sql.files.ignoreCorruptFiles or the data source option ignoreCorruptFiles to ignore corrupt files while reading data from files. When set to true, the Spark jobs will continue to run when encountering corrupted files and the contents that have been read will still be returned.

To ignore corrupt files while reading data files, you can use:

  1. // enable ignore corrupt files via the data source option
  2. // dir1/file3.json is corrupt from parquet's view
  3. val testCorruptDF0 = spark.read.option("ignoreCorruptFiles", "true").parquet(
  4.   "examples/src/main/resources/dir1/",
  5.   "examples/src/main/resources/dir1/dir2/")
  6. testCorruptDF0.show()
  7. // +-------------+
  8. // |         file|
  9. // +-------------+
  10. // |file1.parquet|
  11. // |file2.parquet|
  12. // +-------------+
  14. // enable ignore corrupt files via the configuration
  15. spark.sql("set spark.sql.files.ignoreCorruptFiles=true")
  16. // dir1/file3.json is corrupt from parquet's view
  17. val testCorruptDF1 = spark.read.parquet(
  18.   "examples/src/main/resources/dir1/",
  19.   "examples/src/main/resources/dir1/dir2/")
  20. testCorruptDF1.show()
  21. // +-------------+
  22. // |         file|
  23. // +-------------+
  24. // |file1.parquet|
  25. // |file2.parquet|
  26. // +-------------+

Find full example code at “examples/src/main/scala/org/apache/spark/examples/sql/SQLDataSourceExample.scala” in the Spark repo.

  1. // enable ignore corrupt files via the data source option
  2. // dir1/file3.json is corrupt from parquet's view
  3. Dataset<Row> testCorruptDF0 = spark.read().option("ignoreCorruptFiles", "true").parquet(
  4.     "examples/src/main/resources/dir1/",
  5.     "examples/src/main/resources/dir1/dir2/");
  6. testCorruptDF0.show();
  7. // +-------------+
  8. // |         file|
  9. // +-------------+
  10. // |file1.parquet|
  11. // |file2.parquet|
  12. // +-------------+
  14. // enable ignore corrupt files via the configuration
  15. spark.sql("set spark.sql.files.ignoreCorruptFiles=true");
  16. // dir1/file3.json is corrupt from parquet's view
  17. Dataset<Row> testCorruptDF1 = spark.read().parquet(
  18.         "examples/src/main/resources/dir1/",
  19.         "examples/src/main/resources/dir1/dir2/");
  20. testCorruptDF1.show();
  21. // +-------------+
  22. // |         file|
  23. // +-------------+
  24. // |file1.parquet|
  25. // |file2.parquet|
  26. // +-------------+

Find full example code at “examples/src/main/java/org/apache/spark/examples/sql/JavaSQLDataSourceExample.java” in the Spark repo.

  1. # enable ignore corrupt files via the data source option
  2. # dir1/file3.json is corrupt from parquet's view
  3. test_corrupt_df0 = spark.read.option("ignoreCorruptFiles", "true")\
  4.     .parquet("examples/src/main/resources/dir1/",
  5.              "examples/src/main/resources/dir1/dir2/")
  6. test_corrupt_df0.show()
  7. # +-------------+
  8. # |         file|
  9. # +-------------+
  10. # |file1.parquet|
  11. # |file2.parquet|
  12. # +-------------+
  14. # enable ignore corrupt files via the configuration
  15. spark.sql("set spark.sql.files.ignoreCorruptFiles=true")
  16. # dir1/file3.json is corrupt from parquet's view
  17. test_corrupt_df1 = spark.read.parquet("examples/src/main/resources/dir1/",
  18.                                       "examples/src/main/resources/dir1/dir2/")
  19. test_corrupt_df1.show()
  20. # +-------------+
  21. # |         file|
  22. # +-------------+
  23. # |file1.parquet|
  24. # |file2.parquet|
  25. # +-------------+

Find full example code at “examples/src/main/python/sql/datasource.py” in the Spark repo.

  1. # enable ignore corrupt files via the data source option
  2. # dir1/file3.json is corrupt from parquet's view
  3. testCorruptDF0 <- read.parquet(c("examples/src/main/resources/dir1/", "examples/src/main/resources/dir1/dir2/"), ignoreCorruptFiles = "true")
  4. head(testCorruptDF0)
  5. #            file
  6. # 1 file1.parquet
  7. # 2 file2.parquet
  9. # enable ignore corrupt files via the configuration
  10. sql("set spark.sql.files.ignoreCorruptFiles=true")
  11. # dir1/file3.json is corrupt from parquet's view
  12. testCorruptDF1 <- read.parquet(c("examples/src/main/resources/dir1/", "examples/src/main/resources/dir1/dir2/"))
  13. head(testCorruptDF1)
  14. #            file
  15. # 1 file1.parquet
  16. # 2 file2.parquet

Find full example code at “examples/src/main/r/RSparkSQLExample.R” in the Spark repo.

Ignore Missing Files

Spark allows you to use the configuration spark.sql.files.ignoreMissingFiles or the data source option ignoreMissingFiles to ignore missing files while reading data from files. Here, missing file really means the deleted file under directory after you construct the DataFrame. When set to true, the Spark jobs will continue to run when encountering missing files and the contents that have been read will still be returned.

Path Glob Filter

pathGlobFilter is used to only include files with file names matching the pattern. The syntax follows org.apache.hadoop.fs.GlobFilter. It does not change the behavior of partition discovery.

To load files with paths matching a given glob pattern while keeping the behavior of partition discovery, you can use:

  1. val testGlobFilterDF = spark.read.format("parquet")
  2.   .option("pathGlobFilter", "*.parquet") // json file should be filtered out
  3.   .load("examples/src/main/resources/dir1")
  4. testGlobFilterDF.show()
  5. // +-------------+
  6. // |         file|
  7. // +-------------+
  8. // |file1.parquet|
  9. // +-------------+

Find full example code at “examples/src/main/scala/org/apache/spark/examples/sql/SQLDataSourceExample.scala” in the Spark repo.

  1. Dataset<Row> testGlobFilterDF = spark.read().format("parquet")
  2.         .option("pathGlobFilter", "*.parquet") // json file should be filtered out
  3.         .load("examples/src/main/resources/dir1");
  4. testGlobFilterDF.show();
  5. // +-------------+
  6. // |         file|
  7. // +-------------+
  8. // |file1.parquet|
  9. // +-------------+

Find full example code at “examples/src/main/java/org/apache/spark/examples/sql/JavaSQLDataSourceExample.java” in the Spark repo.

  1. df = spark.read.load("examples/src/main/resources/dir1",
  2.                      format="parquet", pathGlobFilter="*.parquet")
  3. df.show()
  4. # +-------------+
  5. # |         file|
  6. # +-------------+
  7. # |file1.parquet|
  8. # +-------------+

Find full example code at “examples/src/main/python/sql/datasource.py” in the Spark repo.

  1. df <- read.df("examples/src/main/resources/dir1", "parquet", pathGlobFilter = "*.parquet")
  2. #            file
  3. # 1 file1.parquet

Find full example code at “examples/src/main/r/RSparkSQLExample.R” in the Spark repo.

Recursive File Lookup

recursiveFileLookup is used to recursively load files and it disables partition inferring. Its default value is false. If data source explicitly specifies the partitionSpec when recursiveFileLookup is true, exception will be thrown.

To load all files recursively, you can use:

  1. val recursiveLoadedDF = spark.read.format("parquet")
  2.   .option("recursiveFileLookup", "true")
  3.   .load("examples/src/main/resources/dir1")
  4. recursiveLoadedDF.show()
  5. // +-------------+
  6. // |         file|
  7. // +-------------+
  8. // |file1.parquet|
  9. // |file2.parquet|
  10. // +-------------+

Find full example code at “examples/src/main/scala/org/apache/spark/examples/sql/SQLDataSourceExample.scala” in the Spark repo.

  1. Dataset<Row> recursiveLoadedDF = spark.read().format("parquet")
  2.         .option("recursiveFileLookup", "true")
  3.         .load("examples/src/main/resources/dir1");
  4. recursiveLoadedDF.show();
  5. // +-------------+
  6. // |         file|
  7. // +-------------+
  8. // |file1.parquet|
  9. // |file2.parquet|
  10. // +-------------+

Find full example code at “examples/src/main/java/org/apache/spark/examples/sql/JavaSQLDataSourceExample.java” in the Spark repo.

  1. recursive_loaded_df = spark.read.format("parquet")\
  2.     .option("recursiveFileLookup", "true")\
  3.     .load("examples/src/main/resources/dir1")
  4. recursive_loaded_df.show()
  5. # +-------------+
  6. # |         file|
  7. # +-------------+
  8. # |file1.parquet|
  9. # |file2.parquet|
  10. # +-------------+

Find full example code at “examples/src/main/python/sql/datasource.py” in the Spark repo.

  1. recursiveLoadedDF <- read.df("examples/src/main/resources/dir1", "parquet", recursiveFileLookup = "true")
  2. head(recursiveLoadedDF)
  3. #            file
  4. # 1 file1.parquet
  5. # 2 file2.parquet

Find full example code at “examples/src/main/r/RSparkSQLExample.R” in the Spark repo.

Modification Time Path Filters

modifiedBefore and modifiedAfter are options that can be applied together or separately in order to achieve greater granularity over which files may load during a Spark batch query. (Note that Structured Streaming file sources don’t support these options.)

  • modifiedBefore: an optional timestamp to only include files with modification times occurring before the specified time. The provided timestamp must be in the following format: YYYY-MM-DDTHH:mm:ss (e.g. 2020-06-01T13:00:00)
  • modifiedAfter: an optional timestamp to only include files with modification times occurring after the specified time. The provided timestamp must be in the following format: YYYY-MM-DDTHH:mm:ss (e.g. 2020-06-01T13:00:00)

When a timezone option is not provided, the timestamps will be interpreted according to the Spark session timezone (spark.sql.session.timeZone).

To load files with paths matching a given modified time range, you can use:

  1. val beforeFilterDF = spark.read.format("parquet")
  2.   // Files modified before 07/01/2020 at 05:30 are allowed
  3.   .option("modifiedBefore", "2020-07-01T05:30:00")
  4.   .load("examples/src/main/resources/dir1");
  5. beforeFilterDF.show();
  6. // +-------------+
  7. // |         file|
  8. // +-------------+
  9. // |file1.parquet|
  10. // +-------------+
  11. val afterFilterDF = spark.read.format("parquet")
  12.    // Files modified after 06/01/2020 at 05:30 are allowed
  13.   .option("modifiedAfter", "2020-06-01T05:30:00")
  14.   .load("examples/src/main/resources/dir1");
  15. afterFilterDF.show();
  16. // +-------------+
  17. // |         file|
  18. // +-------------+
  19. // +-------------+

Find full example code at “examples/src/main/scala/org/apache/spark/examples/sql/SQLDataSourceExample.scala” in the Spark repo.

  1. Dataset<Row> beforeFilterDF = spark.read().format("parquet")
  2.         // Only load files modified before 7/1/2020 at 05:30
  3.         .option("modifiedBefore", "2020-07-01T05:30:00")
  4.         // Only load files modified after 6/1/2020 at 05:30
  5.         .option("modifiedAfter", "2020-06-01T05:30:00")
  6.         // Interpret both times above relative to CST timezone
  7.         .option("timeZone", "CST")
  8.         .load("examples/src/main/resources/dir1");
  9. beforeFilterDF.show();
  10. // +-------------+
  11. // |         file|
  12. // +-------------+
  13. // |file1.parquet|
  14. // +-------------+

Find full example code at “examples/src/main/java/org/apache/spark/examples/sql/JavaSQLDataSourceExample.java” in the Spark repo.

  1. # Only load files modified before 07/1/2050 @ 08:30:00
  2. df = spark.read.load("examples/src/main/resources/dir1",
  3.                      format="parquet", modifiedBefore="2050-07-01T08:30:00")
  4. df.show()
  5. # +-------------+
  6. # |         file|
  7. # +-------------+
  8. # |file1.parquet|
  9. # +-------------+
  10. # Only load files modified after 06/01/2050 @ 08:30:00
  11. df = spark.read.load("examples/src/main/resources/dir1",
  12.                      format="parquet", modifiedAfter="2050-06-01T08:30:00")
  13. df.show()
  14. # +-------------+
  15. # |         file|
  16. # +-------------+
  17. # +-------------+

Find full example code at “examples/src/main/python/sql/datasource.py” in the Spark repo.

  1. beforeDF <- read.df("examples/src/main/resources/dir1", "parquet", modifiedBefore= "2020-07-01T05:30:00")
  2. #            file
  3. # 1 file1.parquet
  4. afterDF <- read.df("examples/src/main/resources/dir1", "parquet", modifiedAfter = "2020-06-01T05:30:00")
  5. #            file

Find full example code at “examples/src/main/r/RSparkSQLExample.R” in the Spark repo.