mkarbo commented on code in PR #7535:
URL: https://github.com/apache/arrow-rs/pull/7535#discussion_r2104376735
##########
parquet-variant/src/variant.rs:
##########
@@ -18,158 +109,223 @@ impl<'m> VariantMetadata<'m> {
self.bytes
}
+ pub fn try_new(bytes: &'m [u8]) -> Result<Self, ArrowError> {
+ let header = VariantMetadataHeader::try_new(bytes)?;
+ // Offset 1, index 0 because first element after header is dictionary
size
+ let dict_size = header.offset_size.unpack_usize(bytes, 1, 0)?;
+
+ // TODO: Refactor, add test for validation
+ let valid = (0..=dict_size)
+ .map(|i| header.offset_size.unpack_usize(bytes, 1, i + 1))
+ .scan(0, |prev, cur| {
+ let Ok(cur_offset) = cur else {
+ return Some(false);
+ };
+ // Skip the first offset, which is always 0
+ if *prev == 0 {
+ *prev = cur_offset;
+ return Some(true);
+ }
+
+ let valid = cur_offset > *prev;
+ *prev = cur_offset;
+ Some(valid)
+ })
+ .all(|valid| valid);
+
+ if !valid {
+ return Err(ArrowError::InvalidArgumentError(
+ "Offsets are not monotonically increasing".to_string(),
+ ));
+ }
+ Ok(Self {
+ bytes,
+ header,
+ dict_size,
+ })
+ }
+
/// Whether the dictionary keys are sorted and unique
pub fn is_sorted(&self) -> bool {
- todo!()
+ self.header.is_sorted
}
- /// Get the dict length
- pub fn dict_len(&self) -> Result<usize, ArrowError> {
- let end_location = self.offset_size()? as usize + 1;
- if self.bytes.len() < end_location {
- let err_str = format!(
- "Invalid metadata bytes, must have at least length {} but has
length {}",
- &end_location,
- self.bytes.len()
- );
- return Err(ArrowError::InvalidArgumentError(err_str));
- }
- let dict_len_bytes = &self.bytes[1..end_location];
- let dict_len =
usize::from_le_bytes(dict_len_bytes.try_into().map_err(|e| {
- ArrowError::InvalidArgumentError(format!(
- "Unable to convert dictionary_size bytes into usize: {}",
- e,
- ))
- })?);
- Ok(dict_len)
+ /// Get the dictionary size
+ pub fn dictionary_size(&self) -> usize {
+ self.dict_size
}
- pub fn version(&self) -> usize {
- todo!()
+ pub fn version(&self) -> u8 {
+ self.header.version
}
- /// Get the offset by index
- pub fn get_offset_by(&self, index: usize) -> Result<usize, ArrowError> {
- todo!()
+ /// Get the offset by key-index
+ pub fn get_offset_by(&self, index: usize) -> Result<Range<usize>,
ArrowError> {
+ // TODO: Should we memoize the offsets? There could be thousands of
them (https://github.com/apache/arrow-rs/pull/7535#discussion_r2101351294)
+ if index >= self.dict_size {
+ return Err(ArrowError::InvalidArgumentError(format!(
+ "Index {} out of bounds for dictionary of length {}",
+ index, self.dict_size
+ )));
+ }
+
+ // Skipping the header byte (setting byte_offset = 1) and the
dictionary_size (setting offset_index +1)
+ // TODO: Validate size before looking up?
+ // TODO: Fix location / bytes here, the index is wrong.
Review Comment:
The comment was outdated :)
--
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.
To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
