Reading parquet files with a corrupt ARROW:schema panics #2855

noselasd · 2022-10-09T22:20:15Z

Describe the bug
I've encountered a likely corrupt parquet file, containing an ARROW:schema of just 2 bytes. This panics inside arrow when
reading that file.

The issue seems to be here, which isn't verifying a minimal length:

arrow-rs/parquet/src/arrow/schema.rs

Line 106 in 2ae2309

let slice = if bytes[0..4] == [255u8; 4] {

Expected behavior

Error returned instead of panic.

* Adding len checks * Adding test case

noselasd added the bug label Oct 9, 2022

tustvold added help wanted good first issue Good for newcomers parquet Changes to the parquet crate labels Oct 10, 2022

psvri mentioned this issue Nov 17, 2022

Fix Panic on Reading Corrupt Parquet Schema (#2855) #3130

Merged

tustvold closed this as completed in #3130 Nov 21, 2022

tustvold pushed a commit that referenced this issue Nov 21, 2022

Fix Panic on Reading Corrupt Parquet Schema (#2855) (#3130)

6f8187f

* Adding len checks * Adding test case

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Reading parquet files with a corrupt ARROW:schema panics #2855

Reading parquet files with a corrupt ARROW:schema panics #2855

noselasd commented Oct 9, 2022

Reading parquet files with a corrupt ARROW:schema panics #2855

Reading parquet files with a corrupt ARROW:schema panics #2855

Comments

noselasd commented Oct 9, 2022