parquet_variant/
decoder.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17use crate::ShortString;
18use crate::utils::{
19    array_from_slice, overflow_error, slice_from_slice_at_offset, string_from_slice,
20};
21
22use arrow_schema::ArrowError;
23use chrono::{DateTime, Duration, NaiveDate, NaiveDateTime, NaiveTime, Utc};
24use uuid::Uuid;
25
26/// The basic type of a [`Variant`] value, encoded in the first two bits of the
27/// header byte.
28///
29/// See the [Variant Encoding specification] for details
30///
31/// [`Variant`]: crate::Variant
32/// [Variant Encoding specification]: https://github.com/apache/parquet-format/blob/master/VariantEncoding.md#encoding-types
33#[derive(Debug, Clone, Copy, PartialEq)]
34pub enum VariantBasicType {
35    Primitive = 0,
36    ShortString = 1,
37    Object = 2,
38    Array = 3,
39}
40
41/// The type of [`VariantBasicType::Primitive`], for a primitive [`Variant`]
42/// value.
43///
44/// See the [Variant Encoding specification] for details
45///
46/// [`Variant`]: crate::Variant
47/// [Variant Encoding specification]: https://github.com/apache/parquet-format/blob/master/VariantEncoding.md#encoding-types
48#[derive(Debug, Clone, Copy, PartialEq)]
49pub enum VariantPrimitiveType {
50    Null = 0,
51    BooleanTrue = 1,
52    BooleanFalse = 2,
53    Int8 = 3,
54    Int16 = 4,
55    Int32 = 5,
56    Int64 = 6,
57    Double = 7,
58    Decimal4 = 8,
59    Decimal8 = 9,
60    Decimal16 = 10,
61    Date = 11,
62    TimestampMicros = 12,
63    TimestampNtzMicros = 13,
64    Float = 14,
65    Binary = 15,
66    String = 16,
67    Time = 17,
68    TimestampNanos = 18,
69    TimestampNtzNanos = 19,
70    Uuid = 20,
71}
72
73/// Extracts the basic type from a header byte
74pub(crate) fn get_basic_type(header: u8) -> VariantBasicType {
75    // See https://github.com/apache/parquet-format/blob/master/VariantEncoding.md#value-encoding
76    let basic_type = header & 0x03; // Basic type is encoded in the first 2 bits
77    match basic_type {
78        0 => VariantBasicType::Primitive,
79        1 => VariantBasicType::ShortString,
80        2 => VariantBasicType::Object,
81        3 => VariantBasicType::Array,
82        _ => {
83            //NOTE:  A 2-bit value has a max of 4 different values (0-3), hence this is unreachable as we
84            // masked `basic_type` with 0x03 above.
85            unreachable!();
86        }
87    }
88}
89
90impl TryFrom<u8> for VariantPrimitiveType {
91    type Error = ArrowError;
92
93    fn try_from(value: u8) -> Result<Self, Self::Error> {
94        match value {
95            0 => Ok(VariantPrimitiveType::Null),
96            1 => Ok(VariantPrimitiveType::BooleanTrue),
97            2 => Ok(VariantPrimitiveType::BooleanFalse),
98            3 => Ok(VariantPrimitiveType::Int8),
99            4 => Ok(VariantPrimitiveType::Int16),
100            5 => Ok(VariantPrimitiveType::Int32),
101            6 => Ok(VariantPrimitiveType::Int64),
102            7 => Ok(VariantPrimitiveType::Double),
103            8 => Ok(VariantPrimitiveType::Decimal4),
104            9 => Ok(VariantPrimitiveType::Decimal8),
105            10 => Ok(VariantPrimitiveType::Decimal16),
106            11 => Ok(VariantPrimitiveType::Date),
107            12 => Ok(VariantPrimitiveType::TimestampMicros),
108            13 => Ok(VariantPrimitiveType::TimestampNtzMicros),
109            14 => Ok(VariantPrimitiveType::Float),
110            15 => Ok(VariantPrimitiveType::Binary),
111            16 => Ok(VariantPrimitiveType::String),
112            17 => Ok(VariantPrimitiveType::Time),
113            18 => Ok(VariantPrimitiveType::TimestampNanos),
114            19 => Ok(VariantPrimitiveType::TimestampNtzNanos),
115            20 => Ok(VariantPrimitiveType::Uuid),
116            _ => Err(ArrowError::InvalidArgumentError(format!(
117                "unknown primitive type: {value}",
118            ))),
119        }
120    }
121}
122
123/// Used to unpack offset array entries such as metadata dictionary offsets or object/array value
124/// offsets. Also used to unpack object field ids. These are always derived from a two-bit
125/// `XXX_size_minus_one` field in the corresponding header byte.
126#[derive(Debug, Clone, Copy, PartialEq)]
127pub(crate) enum OffsetSizeBytes {
128    One = 1,
129    Two = 2,
130    Three = 3,
131    Four = 4,
132}
133
134impl OffsetSizeBytes {
135    /// Build from the `offset_size_minus_one` bits (see spec).
136    pub(crate) fn try_new(offset_size_minus_one: u8) -> Result<Self, ArrowError> {
137        use OffsetSizeBytes::*;
138        let result = match offset_size_minus_one {
139            0 => One,
140            1 => Two,
141            2 => Three,
142            3 => Four,
143            _ => {
144                return Err(ArrowError::InvalidArgumentError(
145                    "offset_size_minus_one must be 0–3".to_string(),
146                ));
147            }
148        };
149        Ok(result)
150    }
151
152    /// Return one unsigned little-endian value from `bytes`.
153    ///
154    /// * `bytes` – the byte buffer to index
155    /// * `index` – 0-based index into the buffer
156    ///
157    /// Each value is `self as u32` bytes wide (1, 2, 3 or 4), zero-extended to 32 bits as needed.
158    pub(crate) fn unpack_u32(&self, bytes: &[u8], index: usize) -> Result<u32, ArrowError> {
159        self.unpack_u32_at_offset(bytes, 0, index)
160    }
161
162    /// Return one unsigned little-endian value from `bytes`.
163    ///
164    /// * `bytes` – the byte buffer to index
165    /// * `byte_offset` – number of bytes to skip **before** reading the first
166    ///   value (e.g. `1` to move past a header byte).
167    /// * `offset_index` – 0-based index **after** the skipped bytes
168    ///   (`0` is the first value, `1` the next, …).
169    ///
170    /// Each value is `self as u32` bytes wide (1, 2, 3 or 4), zero-extended to 32 bits as needed.
171    pub(crate) fn unpack_u32_at_offset(
172        &self,
173        bytes: &[u8],
174        byte_offset: usize,  // how many bytes to skip
175        offset_index: usize, // which offset in an array of offsets
176    ) -> Result<u32, ArrowError> {
177        use OffsetSizeBytes::*;
178
179        // Index into the byte array:
180        // byte_offset + (*self as usize) * offset_index
181        let offset = offset_index
182            .checked_mul(*self as usize)
183            .and_then(|n| n.checked_add(byte_offset))
184            .ok_or_else(|| overflow_error("unpacking offset array value"))?;
185        let value = match self {
186            One => u8::from_le_bytes(array_from_slice(bytes, offset)?).into(),
187            Two => u16::from_le_bytes(array_from_slice(bytes, offset)?).into(),
188            Three => {
189                // Let's grab the three byte le-chunk first
190                let b3_chunks: [u8; 3] = array_from_slice(bytes, offset)?;
191                // Let's pad it and construct a padded u32 from it.
192                let mut buf = [0u8; 4];
193                buf[..3].copy_from_slice(&b3_chunks);
194                u32::from_le_bytes(buf)
195            }
196            Four => u32::from_le_bytes(array_from_slice(bytes, offset)?),
197        };
198        Ok(value)
199    }
200}
201
202/// Converts a byte buffer to offset values based on the specific offset size
203pub(crate) fn map_bytes_to_offsets(
204    buffer: &[u8],
205    offset_size: OffsetSizeBytes,
206) -> impl Iterator<Item = usize> + use<'_> {
207    buffer
208        .chunks_exact(offset_size as usize)
209        .map(move |chunk| match offset_size {
210            OffsetSizeBytes::One => chunk[0] as usize,
211            OffsetSizeBytes::Two => u16::from_le_bytes([chunk[0], chunk[1]]) as usize,
212            OffsetSizeBytes::Three => {
213                u32::from_le_bytes([chunk[0], chunk[1], chunk[2], 0]) as usize
214            }
215            OffsetSizeBytes::Four => {
216                u32::from_le_bytes([chunk[0], chunk[1], chunk[2], chunk[3]]) as usize
217            }
218        })
219}
220
221/// Extract the primitive type from a Variant value-metadata byte
222pub(crate) fn get_primitive_type(metadata: u8) -> Result<VariantPrimitiveType, ArrowError> {
223    // last 6 bits contain the primitive-type, see spec
224    VariantPrimitiveType::try_from(metadata >> 2)
225}
226
227/// Decodes an Int8 from the value section of a variant.
228pub(crate) fn decode_int8(data: &[u8]) -> Result<i8, ArrowError> {
229    Ok(i8::from_le_bytes(array_from_slice(data, 0)?))
230}
231
232/// Decodes an Int16 from the value section of a variant.
233pub(crate) fn decode_int16(data: &[u8]) -> Result<i16, ArrowError> {
234    Ok(i16::from_le_bytes(array_from_slice(data, 0)?))
235}
236
237/// Decodes an Int32 from the value section of a variant.
238pub(crate) fn decode_int32(data: &[u8]) -> Result<i32, ArrowError> {
239    Ok(i32::from_le_bytes(array_from_slice(data, 0)?))
240}
241
242/// Decodes an Int64 from the value section of a variant.
243pub(crate) fn decode_int64(data: &[u8]) -> Result<i64, ArrowError> {
244    Ok(i64::from_le_bytes(array_from_slice(data, 0)?))
245}
246
247/// Decodes a Decimal4 from the value section of a variant.
248pub(crate) fn decode_decimal4(data: &[u8]) -> Result<(i32, u8), ArrowError> {
249    let scale = u8::from_le_bytes(array_from_slice(data, 0)?);
250    let integer = i32::from_le_bytes(array_from_slice(data, 1)?);
251    Ok((integer, scale))
252}
253
254/// Decodes a Decimal8 from the value section of a variant.
255pub(crate) fn decode_decimal8(data: &[u8]) -> Result<(i64, u8), ArrowError> {
256    let scale = u8::from_le_bytes(array_from_slice(data, 0)?);
257    let integer = i64::from_le_bytes(array_from_slice(data, 1)?);
258    Ok((integer, scale))
259}
260
261/// Decodes a Decimal16 from the value section of a variant.
262pub(crate) fn decode_decimal16(data: &[u8]) -> Result<(i128, u8), ArrowError> {
263    let scale = u8::from_le_bytes(array_from_slice(data, 0)?);
264    let integer = i128::from_le_bytes(array_from_slice(data, 1)?);
265    Ok((integer, scale))
266}
267
268/// Decodes a Float from the value section of a variant.
269pub(crate) fn decode_float(data: &[u8]) -> Result<f32, ArrowError> {
270    Ok(f32::from_le_bytes(array_from_slice(data, 0)?))
271}
272
273/// Decodes a Double from the value section of a variant.
274pub(crate) fn decode_double(data: &[u8]) -> Result<f64, ArrowError> {
275    Ok(f64::from_le_bytes(array_from_slice(data, 0)?))
276}
277
278/// Decodes a Date from the value section of a variant.
279pub(crate) fn decode_date(data: &[u8]) -> Result<NaiveDate, ArrowError> {
280    let days_since_epoch = i32::from_le_bytes(array_from_slice(data, 0)?);
281    let value = DateTime::UNIX_EPOCH + Duration::days(i64::from(days_since_epoch));
282    Ok(value.date_naive())
283}
284
285/// Decodes a TimestampMicros from the value section of a variant.
286pub(crate) fn decode_timestamp_micros(data: &[u8]) -> Result<DateTime<Utc>, ArrowError> {
287    let micros_since_epoch = i64::from_le_bytes(array_from_slice(data, 0)?);
288    DateTime::from_timestamp_micros(micros_since_epoch).ok_or_else(|| {
289        ArrowError::CastError(format!(
290            "Could not cast `{micros_since_epoch}` microseconds into a DateTime<Utc>"
291        ))
292    })
293}
294
295/// Decodes a TimestampNtzMicros from the value section of a variant.
296pub(crate) fn decode_timestampntz_micros(data: &[u8]) -> Result<NaiveDateTime, ArrowError> {
297    let micros_since_epoch = i64::from_le_bytes(array_from_slice(data, 0)?);
298    DateTime::from_timestamp_micros(micros_since_epoch)
299        .ok_or_else(|| {
300            ArrowError::CastError(format!(
301                "Could not cast `{micros_since_epoch}` microseconds into a NaiveDateTime"
302            ))
303        })
304        .map(|v| v.naive_utc())
305}
306
307pub(crate) fn decode_time_ntz(data: &[u8]) -> Result<NaiveTime, ArrowError> {
308    let micros_since_epoch = u64::from_le_bytes(array_from_slice(data, 0)?);
309
310    let case_error = ArrowError::CastError(format!(
311        "Could not cast {micros_since_epoch} microseconds into a NaiveTime"
312    ));
313
314    if micros_since_epoch >= 86_400_000_000 {
315        return Err(case_error);
316    }
317
318    let nanos_since_midnight = micros_since_epoch * 1_000;
319    NaiveTime::from_num_seconds_from_midnight_opt(
320        (nanos_since_midnight / 1_000_000_000) as u32,
321        (nanos_since_midnight % 1_000_000_000) as u32,
322    )
323    .ok_or(case_error)
324}
325
326/// Decodes a TimestampNanos from the value section of a variant.
327pub(crate) fn decode_timestamp_nanos(data: &[u8]) -> Result<DateTime<Utc>, ArrowError> {
328    let nanos_since_epoch = i64::from_le_bytes(array_from_slice(data, 0)?);
329
330    // DateTime::from_timestamp_nanos would never fail
331    Ok(DateTime::from_timestamp_nanos(nanos_since_epoch))
332}
333
334/// Decodes a TimestampNtzNanos from the value section of a variant.
335pub(crate) fn decode_timestampntz_nanos(data: &[u8]) -> Result<NaiveDateTime, ArrowError> {
336    decode_timestamp_nanos(data).map(|v| v.naive_utc())
337}
338
339/// Decodes a UUID from the value section of a variant.
340pub(crate) fn decode_uuid(data: &[u8]) -> Result<Uuid, ArrowError> {
341    Uuid::from_slice(&data[0..16])
342        .map_err(|_| ArrowError::CastError(format!("Cant decode uuid from {:?}", &data[0..16])))
343}
344
345/// Decodes a Binary from the value section of a variant.
346pub(crate) fn decode_binary(data: &[u8]) -> Result<&[u8], ArrowError> {
347    let len = u32::from_le_bytes(array_from_slice(data, 0)?) as usize;
348    slice_from_slice_at_offset(data, 4, 0..len)
349}
350
351/// Decodes a long string from the value section of a variant.
352pub(crate) fn decode_long_string(data: &[u8]) -> Result<&str, ArrowError> {
353    let len = u32::from_le_bytes(array_from_slice(data, 0)?) as usize;
354    string_from_slice(data, 4, 0..len)
355}
356
357/// Decodes a short string from the value section of a variant.
358pub(crate) fn decode_short_string(
359    metadata: u8,
360    data: &[u8],
361) -> Result<ShortString<'_>, ArrowError> {
362    let len = (metadata >> 2) as usize;
363    let string = string_from_slice(data, 0, 0..len)?;
364    ShortString::try_new(string)
365}
366
367#[cfg(test)]
368mod tests {
369    use super::*;
370
371    macro_rules! test_decoder_bounds {
372        ($test_name:ident, $data:expr, $decode_fn:ident, $expected:expr) => {
373            mod $test_name {
374                use super::*;
375
376                #[test]
377                fn exact_length() {
378                    let result = $decode_fn(&$data).unwrap();
379                    assert_eq!(result, $expected);
380                }
381
382                #[test]
383                fn truncated_length() {
384                    // Remove the last byte of data so that there is not enough to decode
385                    let truncated_data = &$data[..$data.len() - 1];
386                    let result = $decode_fn(truncated_data);
387                    assert!(matches!(result, Err(ArrowError::InvalidArgumentError(_))));
388                }
389            }
390        };
391    }
392
393    mod integer {
394        use super::*;
395
396        test_decoder_bounds!(test_i8, [0x2a], decode_int8, 42);
397        test_decoder_bounds!(test_i16, [0xd2, 0x04], decode_int16, 1234);
398        test_decoder_bounds!(test_i32, [0x40, 0xe2, 0x01, 0x00], decode_int32, 123456);
399        test_decoder_bounds!(
400            test_i64,
401            [0x15, 0x81, 0xe9, 0x7d, 0xf4, 0x10, 0x22, 0x11],
402            decode_int64,
403            1234567890123456789
404        );
405    }
406
407    mod decimal {
408        use super::*;
409
410        test_decoder_bounds!(
411            test_decimal4,
412            [
413                0x02, // Scale
414                0xd2, 0x04, 0x00, 0x00, // Unscaled Value
415            ],
416            decode_decimal4,
417            (1234, 2)
418        );
419
420        test_decoder_bounds!(
421            test_decimal8,
422            [
423                0x02, // Scale
424                0xd2, 0x02, 0x96, 0x49, 0x00, 0x00, 0x00, 0x00, // Unscaled Value
425            ],
426            decode_decimal8,
427            (1234567890, 2)
428        );
429
430        test_decoder_bounds!(
431            test_decimal16,
432            [
433                0x02, // Scale
434                0xd2, 0xb6, 0x23, 0xc0, 0xf4, 0x10, 0x22, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
435                0x00, 0x00, // Unscaled Value
436            ],
437            decode_decimal16,
438            (1234567891234567890, 2)
439        );
440    }
441
442    mod float {
443        use super::*;
444
445        test_decoder_bounds!(
446            test_float,
447            [0x06, 0x2c, 0x93, 0x4e],
448            decode_float,
449            1234567890.1234
450        );
451
452        test_decoder_bounds!(
453            test_double,
454            [0xc9, 0xe5, 0x87, 0xb4, 0x80, 0x65, 0xd2, 0x41],
455            decode_double,
456            1234567890.1234
457        );
458    }
459
460    mod datetime {
461        use super::*;
462
463        test_decoder_bounds!(
464            test_date,
465            [0xe2, 0x4e, 0x0, 0x0],
466            decode_date,
467            NaiveDate::from_ymd_opt(2025, 4, 16).unwrap()
468        );
469
470        test_decoder_bounds!(
471            test_timestamp_micros,
472            [0xe0, 0x52, 0x97, 0xdd, 0xe7, 0x32, 0x06, 0x00],
473            decode_timestamp_micros,
474            NaiveDate::from_ymd_opt(2025, 4, 16)
475                .unwrap()
476                .and_hms_milli_opt(16, 34, 56, 780)
477                .unwrap()
478                .and_utc()
479        );
480
481        test_decoder_bounds!(
482            test_timestampntz_micros,
483            [0xe0, 0x52, 0x97, 0xdd, 0xe7, 0x32, 0x06, 0x00],
484            decode_timestampntz_micros,
485            NaiveDate::from_ymd_opt(2025, 4, 16)
486                .unwrap()
487                .and_hms_milli_opt(16, 34, 56, 780)
488                .unwrap()
489        );
490
491        test_decoder_bounds!(
492            test_timestamp_nanos,
493            [0x15, 0x41, 0xa2, 0x5a, 0x36, 0xa2, 0x5b, 0x18],
494            decode_timestamp_nanos,
495            NaiveDate::from_ymd_opt(2025, 8, 14)
496                .unwrap()
497                .and_hms_nano_opt(12, 33, 54, 123456789)
498                .unwrap()
499                .and_utc()
500        );
501
502        test_decoder_bounds!(
503            test_timestamp_nanos_before_epoch,
504            [0x15, 0x41, 0x52, 0xd4, 0x94, 0xe5, 0xad, 0xfa],
505            decode_timestamp_nanos,
506            NaiveDate::from_ymd_opt(1957, 11, 7)
507                .unwrap()
508                .and_hms_nano_opt(12, 33, 54, 123456789)
509                .unwrap()
510                .and_utc()
511        );
512
513        test_decoder_bounds!(
514            test_timestampntz_nanos,
515            [0x15, 0x41, 0xa2, 0x5a, 0x36, 0xa2, 0x5b, 0x18],
516            decode_timestampntz_nanos,
517            NaiveDate::from_ymd_opt(2025, 8, 14)
518                .unwrap()
519                .and_hms_nano_opt(12, 33, 54, 123456789)
520                .unwrap()
521        );
522
523        test_decoder_bounds!(
524            test_timestampntz_nanos_before_epoch,
525            [0x15, 0x41, 0x52, 0xd4, 0x94, 0xe5, 0xad, 0xfa],
526            decode_timestampntz_nanos,
527            NaiveDate::from_ymd_opt(1957, 11, 7)
528                .unwrap()
529                .and_hms_nano_opt(12, 33, 54, 123456789)
530                .unwrap()
531        );
532    }
533
534    #[test]
535    fn test_uuid() {
536        let data = [
537            0xf2, 0x4f, 0x9b, 0x64, 0x81, 0xfa, 0x49, 0xd1, 0xb7, 0x4e, 0x8c, 0x09, 0xa6, 0xe3,
538            0x1c, 0x56,
539        ];
540        let result = decode_uuid(&data).unwrap();
541        assert_eq!(
542            Uuid::parse_str("f24f9b64-81fa-49d1-b74e-8c09a6e31c56").unwrap(),
543            result
544        );
545    }
546
547    mod time {
548        use super::*;
549
550        test_decoder_bounds!(
551            test_timentz,
552            [0x53, 0x1f, 0x8e, 0xdf, 0x2, 0, 0, 0],
553            decode_time_ntz,
554            NaiveTime::from_num_seconds_from_midnight_opt(12340, 567_891_000).unwrap()
555        );
556
557        #[test]
558        fn test_decode_time_ntz_invalid() {
559            let invalid_second = u64::MAX;
560            let data = invalid_second.to_le_bytes();
561            let result = decode_time_ntz(&data);
562            assert!(matches!(result, Err(ArrowError::CastError(_))));
563        }
564    }
565
566    #[test]
567    fn test_binary_exact_length() {
568        let data = [
569            0x09, 0, 0, 0, // Length of binary data, 4-byte little-endian
570            0x03, 0x13, 0x37, 0xde, 0xad, 0xbe, 0xef, 0xca, 0xfe,
571        ];
572        let result = decode_binary(&data).unwrap();
573        assert_eq!(
574            result,
575            [0x03, 0x13, 0x37, 0xde, 0xad, 0xbe, 0xef, 0xca, 0xfe]
576        );
577    }
578
579    #[test]
580    fn test_binary_truncated_length() {
581        let data = [
582            0x09, 0, 0, 0, // Length of binary data, 4-byte little-endian
583            0x03, 0x13, 0x37, 0xde, 0xad, 0xbe, 0xef, 0xca,
584        ];
585        let result = decode_binary(&data);
586        assert!(matches!(result, Err(ArrowError::InvalidArgumentError(_))));
587    }
588
589    #[test]
590    fn test_short_string_exact_length() {
591        let data = b"Helloo";
592        let result = decode_short_string(1 | 5 << 2, data).unwrap();
593        assert_eq!(result.0, "Hello");
594    }
595
596    #[test]
597    fn test_short_string_truncated_length() {
598        let data = b"Hel";
599        let result = decode_short_string(1 | 5 << 2, data);
600        assert!(matches!(result, Err(ArrowError::InvalidArgumentError(_))));
601    }
602
603    #[test]
604    fn test_string_exact_length() {
605        let data = [
606            0x05, 0, 0, 0, // Length of string, 4-byte little-endian
607            b'H', b'e', b'l', b'l', b'o', b'o',
608        ];
609        let result = decode_long_string(&data).unwrap();
610        assert_eq!(result, "Hello");
611    }
612
613    #[test]
614    fn test_string_truncated_length() {
615        let data = [
616            0x05, 0, 0, 0, // Length of string, 4-byte little-endian
617            b'H', b'e', b'l',
618        ];
619        let result = decode_long_string(&data);
620        assert!(matches!(result, Err(ArrowError::InvalidArgumentError(_))));
621    }
622
623    #[test]
624    fn test_offset() {
625        assert_eq!(OffsetSizeBytes::try_new(0).unwrap(), OffsetSizeBytes::One);
626        assert_eq!(OffsetSizeBytes::try_new(1).unwrap(), OffsetSizeBytes::Two);
627        assert_eq!(OffsetSizeBytes::try_new(2).unwrap(), OffsetSizeBytes::Three);
628        assert_eq!(OffsetSizeBytes::try_new(3).unwrap(), OffsetSizeBytes::Four);
629
630        // everything outside 0-3 must error
631        assert!(OffsetSizeBytes::try_new(4).is_err());
632        assert!(OffsetSizeBytes::try_new(255).is_err());
633    }
634
635    #[test]
636    fn unpack_u32_all_widths() {
637        // One-byte offsets
638        let buf_one = [0x01u8, 0xAB, 0xCD];
639        assert_eq!(OffsetSizeBytes::One.unpack_u32(&buf_one, 0).unwrap(), 0x01);
640        assert_eq!(OffsetSizeBytes::One.unpack_u32(&buf_one, 2).unwrap(), 0xCD);
641
642        // Two-byte offsets (little-endian 0x1234, 0x5678)
643        let buf_two = [0x34, 0x12, 0x78, 0x56];
644        assert_eq!(
645            OffsetSizeBytes::Two.unpack_u32(&buf_two, 0).unwrap(),
646            0x1234
647        );
648        assert_eq!(
649            OffsetSizeBytes::Two.unpack_u32(&buf_two, 1).unwrap(),
650            0x5678
651        );
652
653        // Three-byte offsets (0x030201 and 0x0000FF)
654        let buf_three = [0x01, 0x02, 0x03, 0xFF, 0x00, 0x00];
655        assert_eq!(
656            OffsetSizeBytes::Three.unpack_u32(&buf_three, 0).unwrap(),
657            0x030201
658        );
659        assert_eq!(
660            OffsetSizeBytes::Three.unpack_u32(&buf_three, 1).unwrap(),
661            0x0000FF
662        );
663
664        // Four-byte offsets (0x12345678, 0x90ABCDEF)
665        let buf_four = [0x78, 0x56, 0x34, 0x12, 0xEF, 0xCD, 0xAB, 0x90];
666        assert_eq!(
667            OffsetSizeBytes::Four.unpack_u32(&buf_four, 0).unwrap(),
668            0x1234_5678
669        );
670        assert_eq!(
671            OffsetSizeBytes::Four.unpack_u32(&buf_four, 1).unwrap(),
672            0x90AB_CDEF
673        );
674    }
675
676    #[test]
677    fn unpack_u32_out_of_bounds() {
678        let tiny = [0x00u8]; // deliberately too short
679        assert!(OffsetSizeBytes::Two.unpack_u32(&tiny, 0).is_err());
680        assert!(OffsetSizeBytes::Three.unpack_u32(&tiny, 0).is_err());
681    }
682
683    #[test]
684    fn unpack_simple() {
685        let buf = [
686            0x41, // header
687            0x02, 0x00, // dictionary_size = 2
688            0x00, 0x00, // offset[0] = 0
689            0x05, 0x00, // offset[1] = 5
690            0x09, 0x00, // offset[2] = 9
691        ];
692
693        let width = OffsetSizeBytes::Two;
694
695        // dictionary_size starts immediately after the header byte
696        let dict_size = width.unpack_u32_at_offset(&buf, 1, 0).unwrap();
697        assert_eq!(dict_size, 2);
698
699        // offset array immediately follows the dictionary size
700        let first = width.unpack_u32_at_offset(&buf, 1, 1).unwrap();
701        assert_eq!(first, 0);
702
703        let second = width.unpack_u32_at_offset(&buf, 1, 2).unwrap();
704        assert_eq!(second, 5);
705
706        let third = width.unpack_u32_at_offset(&buf, 1, 3).unwrap();
707        assert_eq!(third, 9);
708
709        let err = width.unpack_u32_at_offset(&buf, 1, 4);
710        assert!(err.is_err())
711    }
712}
parquet_variant/decoder.rs

parquet_variant/
decoder.rs
