PARQUET-1534: [parquet-cli] Argument error: Illegal character in opaque part at index 2 on Windows

[masayuki038]

Some commands of parquet-cli don't work on Windows.

$ java -cp target/classes;target/dependency/* org.apache.parquet.cli.Main schema C:\Users\masayuki\Downloads\test.parquet                                                               Argument error: Illegal character in opaque part at index 2: C:\Users\masayuki\Downloads\test.parquet          
                                    
$ java -cp target/classes;target/dependency/* org.apache.parquet.cli.Main csv-schema C:\Users\masayuki\Downloads\test.csv --record-name Test                                              
Argument error: Illegal character in opaque part at index 2: C:\Users\masayuki\Downloads\test.csv              

$ java -cp target/classes;target/dependency/* org.apache.parquet.cli.Main convert-csv C:\Users\masayuki\Downlo ads\test.csv -o C:\Users\masayuki\Downloads\test-csv.parquet
Argument error: Illegal character in opaque part at index 2: C:\Users\masayuki\Downloads\test.csv

$ java -cp target/classes;target/dependency/* org.apache.parquet.cli.Main convert C:\Users\masayuki\Downloads\ userdata1.avro -o C:\Users\masayuki\Downloads\userdata1.parquet
Argument error: Illegal character in opaque part at index 2: C:\Users\masayuki\Downloads\userdata1.avro

$ java -cp target/classes;target/dependency/* org.apache.parquet.cli.Main to-avro C:\Users\masayuki\Downloads\ test.parquet -o C:\Users\masayuki\Downloads\test.avro
Argument error: Illegal character in opaque part at index 2: C:\Users\masayuki\Downloads\test.parquet

$ java -cp target/classes;target/dependency/* org.apache.parquet.cli.Main cat C:\Users\masayuki\Downloads\test.parquet
Argument error: Illegal character in opaque part at index 2: C:\Users\masayuki\Downloads\test.parquet

$ java -cp target/classes;target/dependency/* org.apache.parquet.cli.Main head -n 10 C:\Users\masayuki\Downloa ds\test.parquet
Argument error: Illegal character in opaque part at index 2: C:\Users\masayuki\Downloads\test.parquet

Calling BaseCommand#qualifiedURI with Windows file path, java.net.URI.create throws IllegalArgumentException and the command execution is aborted.

This PR fixes the issue. With this PR, parquet-cli will work all commands on Windows.

parquet-cli has no tests. I created PR to add simple tests for each command at #625. Could you review that first?

[gszadovszky]

Could you fix the test failures?

[masayuki038]

@gszadovszky I fixed the test failures. Could you review this?

[masayuki038]

Could anyone merge or review this?

[rdblue]

This is a behavior change that breaks the stated contract of this method: @throws IOException if there is an error creating a qualified URI.

In addition, the tests do not validate this change. What happens to invalid URIs after this change? Are characters automatically escaped by using Path? What about a resource URI with an unescaped character?

It is fine to change the method contract, but the new behavior must be as well understood as the previous behavior and it should be tested. The tests in this PR check, for example, that the file name has not changed and don't exercise the error handling code at all.

[masayuki038]

@rdblue Thanks for your comment.

This is a behavior change that breaks the stated contract of this method: @throws IOException if there is an error creating a qualified URI.

I don't agree. URI.create and fileURI.getScheme() don't throw IOException and URISyntaxException is not a subclass of IOException.

• https://docs.oracle.com/javase/8/docs/api/java/net/URI.html#create-java.lang.String-
• https://docs.oracle.com/javase/8/docs/api/java/net/URISyntaxException.html

In addition, the tests do not validate this change. What happens to invalid URIs after this change?

The difference from URI.create and new URI is only whether the exception to throw is the IllegalArgumentException or URISyntaxException. The reason to use new URI is that catching with IllegalArgumentException may catch an unintended exception.

https://hg.openjdk.java.net/jdk8u/jdk8u/jdk/file/2d4edd21bb38/src/share/classes/java/net/URI.java#l848