parquet/file/page_index/
column_index.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.
17
18//! [`ColumnIndexMetaData`] structures holding decoded [`ColumnIndex`] information
19//!
20//! [`ColumnIndex`]: https://github.com/apache/parquet-format/blob/master/PageIndex.md
21//!
22
23use crate::{
24    data_type::{ByteArray, FixedLenByteArray},
25    errors::{ParquetError, Result},
26    parquet_thrift::{
27        ElementType, FieldType, ThriftCompactOutputProtocol, WriteThrift, WriteThriftField,
28    },
29};
30use std::ops::Deref;
31
32use crate::{
33    basic::BoundaryOrder,
34    data_type::{Int96, private::ParquetValueType},
35    file::page_index::index_reader::ThriftColumnIndex,
36};
37
38/// Common bits of the column index
39#[derive(Debug, Clone, PartialEq)]
40pub struct ColumnIndex {
41    pub(crate) null_pages: Vec<bool>,
42    pub(crate) boundary_order: BoundaryOrder,
43    pub(crate) null_counts: Option<Vec<i64>>,
44    pub(crate) repetition_level_histograms: Option<Vec<i64>>,
45    pub(crate) definition_level_histograms: Option<Vec<i64>>,
46}
47
48impl ColumnIndex {
49    /// Returns the number of pages
50    pub fn num_pages(&self) -> u64 {
51        self.null_pages.len() as u64
52    }
53
54    /// Returns the number of null values in the page indexed by `idx`
55    ///
56    /// Returns `None` if no null counts have been set in the index
57    pub fn null_count(&self, idx: usize) -> Option<i64> {
58        self.null_counts.as_ref().map(|nc| nc[idx])
59    }
60
61    /// Returns the repetition level histogram for the page indexed by `idx`
62    pub fn repetition_level_histogram(&self, idx: usize) -> Option<&[i64]> {
63        if let Some(rep_hists) = self.repetition_level_histograms.as_ref() {
64            let num_lvls = rep_hists.len() / self.num_pages() as usize;
65            let start = num_lvls * idx;
66            Some(&rep_hists[start..start + num_lvls])
67        } else {
68            None
69        }
70    }
71
72    /// Returns the definition level histogram for the page indexed by `idx`
73    pub fn definition_level_histogram(&self, idx: usize) -> Option<&[i64]> {
74        if let Some(def_hists) = self.definition_level_histograms.as_ref() {
75            let num_lvls = def_hists.len() / self.num_pages() as usize;
76            let start = num_lvls * idx;
77            Some(&def_hists[start..start + num_lvls])
78        } else {
79            None
80        }
81    }
82
83    /// Returns whether the page indexed by `idx` consists of all null values
84    pub fn is_null_page(&self, idx: usize) -> bool {
85        self.null_pages[idx]
86    }
87}
88
89/// Column index for primitive types
90#[derive(Debug, Clone, PartialEq)]
91pub struct PrimitiveColumnIndex<T> {
92    pub(crate) column_index: ColumnIndex,
93    pub(crate) min_values: Vec<T>,
94    pub(crate) max_values: Vec<T>,
95}
96
97impl<T: ParquetValueType> PrimitiveColumnIndex<T> {
98    pub(crate) fn try_new(
99        null_pages: Vec<bool>,
100        boundary_order: BoundaryOrder,
101        null_counts: Option<Vec<i64>>,
102        repetition_level_histograms: Option<Vec<i64>>,
103        definition_level_histograms: Option<Vec<i64>>,
104        min_bytes: Vec<&[u8]>,
105        max_bytes: Vec<&[u8]>,
106    ) -> Result<Self> {
107        let len = null_pages.len();
108
109        let mut min_values = Vec::with_capacity(len);
110        let mut max_values = Vec::with_capacity(len);
111
112        for (i, is_null) in null_pages.iter().enumerate().take(len) {
113            if !is_null {
114                let min = min_bytes[i];
115                min_values.push(T::try_from_le_slice(min)?);
116
117                let max = max_bytes[i];
118                max_values.push(T::try_from_le_slice(max)?);
119            } else {
120                // need placeholders
121                min_values.push(Default::default());
122                max_values.push(Default::default());
123            }
124        }
125
126        Ok(Self {
127            column_index: ColumnIndex {
128                null_pages,
129                boundary_order,
130                null_counts,
131                repetition_level_histograms,
132                definition_level_histograms,
133            },
134            min_values,
135            max_values,
136        })
137    }
138
139    pub(super) fn try_from_thrift(index: ThriftColumnIndex) -> Result<Self> {
140        Self::try_new(
141            index.null_pages,
142            index.boundary_order,
143            index.null_counts,
144            index.repetition_level_histograms,
145            index.definition_level_histograms,
146            index.min_values,
147            index.max_values,
148        )
149    }
150}
151
152impl<T> PrimitiveColumnIndex<T> {
153    /// Returns an array containing the min values for each page.
154    ///
155    /// Values in the returned slice are only valid if [`ColumnIndex::is_null_page()`]
156    /// is `false` for the same index.
157    pub fn min_values(&self) -> &[T] {
158        &self.min_values
159    }
160
161    /// Returns an array containing the max values for each page.
162    ///
163    /// Values in the returned slice are only valid if [`ColumnIndex::is_null_page()`]
164    /// is `false` for the same index.
165    pub fn max_values(&self) -> &[T] {
166        &self.max_values
167    }
168
169    /// Returns an iterator over the min values.
170    ///
171    /// Values may be `None` when [`ColumnIndex::is_null_page()`] is `true`.
172    pub fn min_values_iter(&self) -> impl Iterator<Item = Option<&T>> {
173        self.min_values.iter().enumerate().map(|(i, min)| {
174            if self.is_null_page(i) {
175                None
176            } else {
177                Some(min)
178            }
179        })
180    }
181
182    /// Returns an iterator over the max values.
183    ///
184    /// Values may be `None` when [`ColumnIndex::is_null_page()`] is `true`.
185    pub fn max_values_iter(&self) -> impl Iterator<Item = Option<&T>> {
186        self.max_values.iter().enumerate().map(|(i, min)| {
187            if self.is_null_page(i) {
188                None
189            } else {
190                Some(min)
191            }
192        })
193    }
194
195    /// Returns the min value for the page indexed by `idx`
196    ///
197    /// It is `None` when all values are null
198    pub fn min_value(&self, idx: usize) -> Option<&T> {
199        if self.null_pages[idx] {
200            None
201        } else {
202            Some(&self.min_values[idx])
203        }
204    }
205
206    /// Returns the max value for the page indexed by `idx`
207    ///
208    /// It is `None` when all values are null
209    pub fn max_value(&self, idx: usize) -> Option<&T> {
210        if self.null_pages[idx] {
211            None
212        } else {
213            Some(&self.max_values[idx])
214        }
215    }
216}
217
218impl<T> Deref for PrimitiveColumnIndex<T> {
219    type Target = ColumnIndex;
220
221    fn deref(&self) -> &Self::Target {
222        &self.column_index
223    }
224}
225
226impl<T: ParquetValueType> WriteThrift for PrimitiveColumnIndex<T> {
227    const ELEMENT_TYPE: ElementType = ElementType::Struct;
228    fn write_thrift<W: std::io::Write>(
229        &self,
230        writer: &mut ThriftCompactOutputProtocol<W>,
231    ) -> Result<()> {
232        self.null_pages.write_thrift_field(writer, 1, 0)?;
233
234        // need to handle min/max manually
235        let len = self.null_pages.len();
236        writer.write_field_begin(FieldType::List, 2, 1)?;
237        writer.write_list_begin(ElementType::Binary, len)?;
238        for i in 0..len {
239            let min = self.min_value(i).map(|m| m.as_bytes()).unwrap_or(&[]);
240            min.write_thrift(writer)?;
241        }
242        writer.write_field_begin(FieldType::List, 3, 2)?;
243        writer.write_list_begin(ElementType::Binary, len)?;
244        for i in 0..len {
245            let max = self.max_value(i).map(|m| m.as_bytes()).unwrap_or(&[]);
246            max.write_thrift(writer)?;
247        }
248        let mut last_field_id = self.boundary_order.write_thrift_field(writer, 4, 3)?;
249        if self.null_counts.is_some() {
250            last_field_id =
251                self.null_counts
252                    .as_ref()
253                    .unwrap()
254                    .write_thrift_field(writer, 5, last_field_id)?;
255        }
256        if self.repetition_level_histograms.is_some() {
257            last_field_id = self
258                .repetition_level_histograms
259                .as_ref()
260                .unwrap()
261                .write_thrift_field(writer, 6, last_field_id)?;
262        }
263        if self.definition_level_histograms.is_some() {
264            self.definition_level_histograms
265                .as_ref()
266                .unwrap()
267                .write_thrift_field(writer, 7, last_field_id)?;
268        }
269        writer.write_struct_end()
270    }
271}
272
273/// Column index for byte arrays (fixed length and variable)
274#[derive(Debug, Clone, PartialEq)]
275pub struct ByteArrayColumnIndex {
276    pub(crate) column_index: ColumnIndex,
277    // raw bytes for min and max values
278    pub(crate) min_bytes: Vec<u8>,
279    pub(crate) min_offsets: Vec<usize>,
280    pub(crate) max_bytes: Vec<u8>,
281    pub(crate) max_offsets: Vec<usize>,
282}
283
284impl ByteArrayColumnIndex {
285    pub(crate) fn try_new(
286        null_pages: Vec<bool>,
287        boundary_order: BoundaryOrder,
288        null_counts: Option<Vec<i64>>,
289        repetition_level_histograms: Option<Vec<i64>>,
290        definition_level_histograms: Option<Vec<i64>>,
291        min_values: Vec<&[u8]>,
292        max_values: Vec<&[u8]>,
293    ) -> Result<Self> {
294        let len = null_pages.len();
295
296        let min_len = min_values.iter().map(|&v| v.len()).sum();
297        let max_len = max_values.iter().map(|&v| v.len()).sum();
298        let mut min_bytes = vec![0u8; min_len];
299        let mut max_bytes = vec![0u8; max_len];
300
301        let mut min_offsets = vec![0usize; len + 1];
302        let mut max_offsets = vec![0usize; len + 1];
303
304        let mut min_pos = 0;
305        let mut max_pos = 0;
306
307        for (i, is_null) in null_pages.iter().enumerate().take(len) {
308            if !is_null {
309                let min = min_values[i];
310                let dst = &mut min_bytes[min_pos..min_pos + min.len()];
311                dst.copy_from_slice(min);
312                min_offsets[i] = min_pos;
313                min_pos += min.len();
314
315                let max = max_values[i];
316                let dst = &mut max_bytes[max_pos..max_pos + max.len()];
317                dst.copy_from_slice(max);
318                max_offsets[i] = max_pos;
319                max_pos += max.len();
320            } else {
321                min_offsets[i] = min_pos;
322                max_offsets[i] = max_pos;
323            }
324        }
325
326        min_offsets[len] = min_pos;
327        max_offsets[len] = max_pos;
328
329        Ok(Self {
330            column_index: ColumnIndex {
331                null_pages,
332                boundary_order,
333                null_counts,
334                repetition_level_histograms,
335                definition_level_histograms,
336            },
337            min_bytes,
338            min_offsets,
339            max_bytes,
340            max_offsets,
341        })
342    }
343
344    pub(super) fn try_from_thrift(index: ThriftColumnIndex) -> Result<Self> {
345        Self::try_new(
346            index.null_pages,
347            index.boundary_order,
348            index.null_counts,
349            index.repetition_level_histograms,
350            index.definition_level_histograms,
351            index.min_values,
352            index.max_values,
353        )
354    }
355
356    /// Returns the min value for the page indexed by `idx`
357    ///
358    /// It is `None` when all values are null
359    pub fn min_value(&self, idx: usize) -> Option<&[u8]> {
360        if self.null_pages[idx] {
361            None
362        } else {
363            let start = self.min_offsets[idx];
364            let end = self.min_offsets[idx + 1];
365            Some(&self.min_bytes[start..end])
366        }
367    }
368
369    /// Returns the max value for the page indexed by `idx`
370    ///
371    /// It is `None` when all values are null
372    pub fn max_value(&self, idx: usize) -> Option<&[u8]> {
373        if self.null_pages[idx] {
374            None
375        } else {
376            let start = self.max_offsets[idx];
377            let end = self.max_offsets[idx + 1];
378            Some(&self.max_bytes[start..end])
379        }
380    }
381
382    /// Returns an iterator over the min values.
383    ///
384    /// Values may be `None` when [`ColumnIndex::is_null_page()`] is `true`.
385    pub fn min_values_iter(&self) -> impl Iterator<Item = Option<&[u8]>> {
386        (0..self.num_pages() as usize).map(|i| {
387            if self.is_null_page(i) {
388                None
389            } else {
390                self.min_value(i)
391            }
392        })
393    }
394
395    /// Returns an iterator over the max values.
396    ///
397    /// Values may be `None` when [`ColumnIndex::is_null_page()`] is `true`.
398    pub fn max_values_iter(&self) -> impl Iterator<Item = Option<&[u8]>> {
399        (0..self.num_pages() as usize).map(|i| {
400            if self.is_null_page(i) {
401                None
402            } else {
403                self.max_value(i)
404            }
405        })
406    }
407}
408
409impl Deref for ByteArrayColumnIndex {
410    type Target = ColumnIndex;
411
412    fn deref(&self) -> &Self::Target {
413        &self.column_index
414    }
415}
416
417impl WriteThrift for ByteArrayColumnIndex {
418    const ELEMENT_TYPE: ElementType = ElementType::Struct;
419    fn write_thrift<W: std::io::Write>(
420        &self,
421        writer: &mut ThriftCompactOutputProtocol<W>,
422    ) -> Result<()> {
423        self.null_pages.write_thrift_field(writer, 1, 0)?;
424
425        // need to handle min/max manually
426        let len = self.null_pages.len();
427        writer.write_field_begin(FieldType::List, 2, 1)?;
428        writer.write_list_begin(ElementType::Binary, len)?;
429        for i in 0..len {
430            let min = self.min_value(i).unwrap_or(&[]);
431            min.write_thrift(writer)?;
432        }
433        writer.write_field_begin(FieldType::List, 3, 2)?;
434        writer.write_list_begin(ElementType::Binary, len)?;
435        for i in 0..len {
436            let max = self.max_value(i).unwrap_or(&[]);
437            max.write_thrift(writer)?;
438        }
439        let mut last_field_id = self.boundary_order.write_thrift_field(writer, 4, 3)?;
440        if self.null_counts.is_some() {
441            last_field_id =
442                self.null_counts
443                    .as_ref()
444                    .unwrap()
445                    .write_thrift_field(writer, 5, last_field_id)?;
446        }
447        if self.repetition_level_histograms.is_some() {
448            last_field_id = self
449                .repetition_level_histograms
450                .as_ref()
451                .unwrap()
452                .write_thrift_field(writer, 6, last_field_id)?;
453        }
454        if self.definition_level_histograms.is_some() {
455            self.definition_level_histograms
456                .as_ref()
457                .unwrap()
458                .write_thrift_field(writer, 7, last_field_id)?;
459        }
460        writer.write_struct_end()
461    }
462}
463
464// Macro to generate getter functions for ColumnIndexMetaData.
465macro_rules! colidx_enum_func {
466    ($self:ident, $func:ident, $arg:ident) => {{
467        match *$self {
468            Self::BOOLEAN(ref typed) => typed.$func($arg),
469            Self::INT32(ref typed) => typed.$func($arg),
470            Self::INT64(ref typed) => typed.$func($arg),
471            Self::INT96(ref typed) => typed.$func($arg),
472            Self::FLOAT(ref typed) => typed.$func($arg),
473            Self::DOUBLE(ref typed) => typed.$func($arg),
474            Self::BYTE_ARRAY(ref typed) => typed.$func($arg),
475            Self::FIXED_LEN_BYTE_ARRAY(ref typed) => typed.$func($arg),
476            _ => panic!(concat!(
477                "Cannot call ",
478                stringify!($func),
479                " on ColumnIndexMetaData::NONE"
480            )),
481        }
482    }};
483    ($self:ident, $func:ident) => {{
484        match *$self {
485            Self::BOOLEAN(ref typed) => typed.$func(),
486            Self::INT32(ref typed) => typed.$func(),
487            Self::INT64(ref typed) => typed.$func(),
488            Self::INT96(ref typed) => typed.$func(),
489            Self::FLOAT(ref typed) => typed.$func(),
490            Self::DOUBLE(ref typed) => typed.$func(),
491            Self::BYTE_ARRAY(ref typed) => typed.$func(),
492            Self::FIXED_LEN_BYTE_ARRAY(ref typed) => typed.$func(),
493            _ => panic!(concat!(
494                "Cannot call ",
495                stringify!($func),
496                " on ColumnIndexMetaData::NONE"
497            )),
498        }
499    }};
500}
501
502/// Parsed [`ColumnIndex`] information for a Parquet file.
503///
504/// See [`ParquetColumnIndex`] for more information.
505///
506/// [`ParquetColumnIndex`]: crate::file::metadata::ParquetColumnIndex
507/// [`ColumnIndex`]: https://github.com/apache/parquet-format/blob/master/PageIndex.md
508#[derive(Debug, Clone, PartialEq)]
509#[allow(non_camel_case_types)]
510pub enum ColumnIndexMetaData {
511    /// Sometimes reading page index from parquet file
512    /// will only return pageLocations without min_max index,
513    /// `NONE` represents this lack of index information
514    NONE,
515    /// Boolean type index
516    BOOLEAN(PrimitiveColumnIndex<bool>),
517    /// 32-bit integer type index
518    INT32(PrimitiveColumnIndex<i32>),
519    /// 64-bit integer type index
520    INT64(PrimitiveColumnIndex<i64>),
521    /// 96-bit integer type (timestamp) index
522    INT96(PrimitiveColumnIndex<Int96>),
523    /// 32-bit floating point type index
524    FLOAT(PrimitiveColumnIndex<f32>),
525    /// 64-bit floating point type index
526    DOUBLE(PrimitiveColumnIndex<f64>),
527    /// Byte array type index
528    BYTE_ARRAY(ByteArrayColumnIndex),
529    /// Fixed length byte array type index
530    FIXED_LEN_BYTE_ARRAY(ByteArrayColumnIndex),
531}
532
533impl ColumnIndexMetaData {
534    /// Return min/max elements inside ColumnIndex are ordered or not.
535    pub fn is_sorted(&self) -> bool {
536        // 0:UNORDERED, 1:ASCENDING ,2:DESCENDING,
537        if let Some(order) = self.get_boundary_order() {
538            order != BoundaryOrder::UNORDERED
539        } else {
540            false
541        }
542    }
543
544    /// Get boundary_order of this page index.
545    pub fn get_boundary_order(&self) -> Option<BoundaryOrder> {
546        match self {
547            Self::NONE => None,
548            Self::BOOLEAN(index) => Some(index.boundary_order),
549            Self::INT32(index) => Some(index.boundary_order),
550            Self::INT64(index) => Some(index.boundary_order),
551            Self::INT96(index) => Some(index.boundary_order),
552            Self::FLOAT(index) => Some(index.boundary_order),
553            Self::DOUBLE(index) => Some(index.boundary_order),
554            Self::BYTE_ARRAY(index) => Some(index.boundary_order),
555            Self::FIXED_LEN_BYTE_ARRAY(index) => Some(index.boundary_order),
556        }
557    }
558
559    /// Returns array of null counts, one per page.
560    ///
561    /// Returns `None` if now null counts have been set in the index
562    pub fn null_counts(&self) -> Option<&Vec<i64>> {
563        match self {
564            Self::NONE => None,
565            Self::BOOLEAN(index) => index.null_counts.as_ref(),
566            Self::INT32(index) => index.null_counts.as_ref(),
567            Self::INT64(index) => index.null_counts.as_ref(),
568            Self::INT96(index) => index.null_counts.as_ref(),
569            Self::FLOAT(index) => index.null_counts.as_ref(),
570            Self::DOUBLE(index) => index.null_counts.as_ref(),
571            Self::BYTE_ARRAY(index) => index.null_counts.as_ref(),
572            Self::FIXED_LEN_BYTE_ARRAY(index) => index.null_counts.as_ref(),
573        }
574    }
575
576    /// Returns the number of pages
577    pub fn num_pages(&self) -> u64 {
578        colidx_enum_func!(self, num_pages)
579    }
580
581    /// Returns the number of null values in the page indexed by `idx`
582    ///
583    /// Returns `None` if no null counts have been set in the index
584    pub fn null_count(&self, idx: usize) -> Option<i64> {
585        colidx_enum_func!(self, null_count, idx)
586    }
587
588    /// Returns the repetition level histogram for the page indexed by `idx`
589    pub fn repetition_level_histogram(&self, idx: usize) -> Option<&[i64]> {
590        colidx_enum_func!(self, repetition_level_histogram, idx)
591    }
592
593    /// Returns the definition level histogram for the page indexed by `idx`
594    pub fn definition_level_histogram(&self, idx: usize) -> Option<&[i64]> {
595        colidx_enum_func!(self, definition_level_histogram, idx)
596    }
597
598    /// Returns whether the page indexed by `idx` consists of all null values
599    pub fn is_null_page(&self, idx: usize) -> bool {
600        colidx_enum_func!(self, is_null_page, idx)
601    }
602}
603
604/// Provides iterators over min and max values of a [`ColumnIndexMetaData`]
605pub trait ColumnIndexIterators {
606    /// Can be one of `bool`, `i32`, `i64`, `Int96`, `f32`, `f64`, [`ByteArray`],
607    /// or [`FixedLenByteArray`]
608    type Item;
609
610    /// Return iterator over the min values for the index
611    fn min_values_iter(colidx: &ColumnIndexMetaData) -> impl Iterator<Item = Option<Self::Item>>;
612
613    /// Return iterator over the max values for the index
614    fn max_values_iter(colidx: &ColumnIndexMetaData) -> impl Iterator<Item = Option<Self::Item>>;
615}
616
617macro_rules! column_index_iters {
618    ($item: ident, $variant: ident, $conv:expr) => {
619        impl ColumnIndexIterators for $item {
620            type Item = $item;
621
622            fn min_values_iter(
623                colidx: &ColumnIndexMetaData,
624            ) -> impl Iterator<Item = Option<Self::Item>> {
625                if let ColumnIndexMetaData::$variant(index) = colidx {
626                    index.min_values_iter().map($conv)
627                } else {
628                    panic!(concat!("Wrong type for ", stringify!($item), " iterator"))
629                }
630            }
631
632            fn max_values_iter(
633                colidx: &ColumnIndexMetaData,
634            ) -> impl Iterator<Item = Option<Self::Item>> {
635                if let ColumnIndexMetaData::$variant(index) = colidx {
636                    index.max_values_iter().map($conv)
637                } else {
638                    panic!(concat!("Wrong type for ", stringify!($item), " iterator"))
639                }
640            }
641        }
642    };
643}
644
645column_index_iters!(bool, BOOLEAN, |v| v.copied());
646column_index_iters!(i32, INT32, |v| v.copied());
647column_index_iters!(i64, INT64, |v| v.copied());
648column_index_iters!(Int96, INT96, |v| v.copied());
649column_index_iters!(f32, FLOAT, |v| v.copied());
650column_index_iters!(f64, DOUBLE, |v| v.copied());
651column_index_iters!(ByteArray, BYTE_ARRAY, |v| v
652    .map(|v| ByteArray::from(v.to_owned())));
653column_index_iters!(FixedLenByteArray, FIXED_LEN_BYTE_ARRAY, |v| v
654    .map(|v| FixedLenByteArray::from(v.to_owned())));
655
656impl WriteThrift for ColumnIndexMetaData {
657    const ELEMENT_TYPE: ElementType = ElementType::Struct;
658
659    fn write_thrift<W: std::io::Write>(
660        &self,
661        writer: &mut ThriftCompactOutputProtocol<W>,
662    ) -> Result<()> {
663        match self {
664            ColumnIndexMetaData::BOOLEAN(index) => index.write_thrift(writer),
665            ColumnIndexMetaData::INT32(index) => index.write_thrift(writer),
666            ColumnIndexMetaData::INT64(index) => index.write_thrift(writer),
667            ColumnIndexMetaData::INT96(index) => index.write_thrift(writer),
668            ColumnIndexMetaData::FLOAT(index) => index.write_thrift(writer),
669            ColumnIndexMetaData::DOUBLE(index) => index.write_thrift(writer),
670            ColumnIndexMetaData::BYTE_ARRAY(index) => index.write_thrift(writer),
671            ColumnIndexMetaData::FIXED_LEN_BYTE_ARRAY(index) => index.write_thrift(writer),
672            _ => Err(general_err!("Cannot serialize NONE index")),
673        }
674    }
675}
676
677#[cfg(test)]
678mod tests {
679    use super::*;
680
681    #[test]
682    fn test_page_index_min_max_null() {
683        let column_index = PrimitiveColumnIndex {
684            column_index: ColumnIndex {
685                null_pages: vec![false],
686                boundary_order: BoundaryOrder::ASCENDING,
687                null_counts: Some(vec![0]),
688                repetition_level_histograms: Some(vec![1, 2]),
689                definition_level_histograms: Some(vec![1, 2, 3]),
690            },
691            min_values: vec![-123],
692            max_values: vec![234],
693        };
694
695        assert_eq!(column_index.min_value(0), Some(&-123));
696        assert_eq!(column_index.max_value(0), Some(&234));
697        assert_eq!(column_index.null_count(0), Some(0));
698        assert_eq!(column_index.repetition_level_histogram(0).unwrap(), &[1, 2]);
699        assert_eq!(
700            column_index.definition_level_histogram(0).unwrap(),
701            &[1, 2, 3]
702        );
703    }
704
705    #[test]
706    fn test_page_index_min_max_null_none() {
707        let column_index: PrimitiveColumnIndex<i32> = PrimitiveColumnIndex::<i32> {
708            column_index: ColumnIndex {
709                null_pages: vec![true],
710                boundary_order: BoundaryOrder::ASCENDING,
711                null_counts: Some(vec![1]),
712                repetition_level_histograms: None,
713                definition_level_histograms: Some(vec![1, 0]),
714            },
715            min_values: vec![Default::default()],
716            max_values: vec![Default::default()],
717        };
718
719        assert_eq!(column_index.min_value(0), None);
720        assert_eq!(column_index.max_value(0), None);
721        assert_eq!(column_index.null_count(0), Some(1));
722        assert_eq!(column_index.repetition_level_histogram(0), None);
723        assert_eq!(column_index.definition_level_histogram(0).unwrap(), &[1, 0]);
724    }
725
726    #[test]
727    fn test_invalid_column_index() {
728        let column_index = ThriftColumnIndex {
729            null_pages: vec![true, false],
730            min_values: vec![
731                &[],
732                &[], // this shouldn't be empty as null_pages[1] is false
733            ],
734            max_values: vec![
735                &[],
736                &[], // this shouldn't be empty as null_pages[1] is false
737            ],
738            null_counts: None,
739            repetition_level_histograms: None,
740            definition_level_histograms: None,
741            boundary_order: BoundaryOrder::UNORDERED,
742        };
743
744        let err = PrimitiveColumnIndex::<i32>::try_from_thrift(column_index).unwrap_err();
745        assert_eq!(
746            err.to_string(),
747            "Parquet error: error converting value, expected 4 bytes got 0"
748        );
749    }
750}
parquet/file/page_index/column_index.rs

parquet/file/page_index/
column_index.rs
