fn flush_block_values()

in parquet/src/encodings/encoding/mod.rs [347:411]

• 44 lines of code
• 10 McCabe index (conditional complexity)

    fn flush_block_values(&mut self) -> Result<()> {
        if self.values_in_block == 0 {
            return Ok(());
        }

        let mut min_delta = i64::MAX;
        for i in 0..self.values_in_block {
            min_delta = cmp::min(min_delta, self.deltas[i]);
        }

        // Write min delta
        self.bit_writer.put_zigzag_vlq_int(min_delta);

        // Slice to store bit width for each mini block
        let offset = self.bit_writer.skip(self.num_mini_blocks);

        for i in 0..self.num_mini_blocks {
            // Find how many values we need to encode - either block size or whatever
            // values left
            let n = cmp::min(self.mini_block_size, self.values_in_block);
            if n == 0 {
                // Decoders should be agnostic to the padding value, we therefore use 0xFF
                // when running tests. However, not all implementations may handle this correctly
                // so pad with 0 when not running tests
                let pad_value = cfg!(test).then(|| 0xFF).unwrap_or(0);
                for j in i..self.num_mini_blocks {
                    self.bit_writer.write_at(offset + j, pad_value);
                }
                break;
            }

            // Compute the max delta in current mini block
            let mut max_delta = i64::MIN;
            for j in 0..n {
                max_delta =
                    cmp::max(max_delta, self.deltas[i * self.mini_block_size + j]);
            }

            // Compute bit width to store (max_delta - min_delta)
            let bit_width =
                num_required_bits(self.subtract_u64(max_delta, min_delta)) as usize;
            self.bit_writer.write_at(offset + i, bit_width as u8);

            // Encode values in current mini block using min_delta and bit_width
            for j in 0..n {
                let packed_value = self
                    .subtract_u64(self.deltas[i * self.mini_block_size + j], min_delta);
                self.bit_writer.put_value(packed_value, bit_width);
            }

            // Pad the last block (n < mini_block_size)
            for _ in n..self.mini_block_size {
                self.bit_writer.put_value(0, bit_width);
            }

            self.values_in_block -= n;
        }

        assert_eq!(
            self.values_in_block, 0,
            "Expected 0 values in block, found {}",
            self.values_in_block
        );
        Ok(())
    }