lance_file/versions/v2_3/
mod.rs1use std::{
7 collections::BTreeMap,
8 sync::{
9 Arc,
10 atomic::{AtomicBool, Ordering},
11 },
12};
13
14use bytes::Bytes;
15use lance_core::{
16 Error, Result,
17 datatypes::{Field, Schema},
18};
19use lance_encoding::{
20 compression_config::CompressionParams,
21 encoder::{
22 ColumnIndexSequence, EncodedBatch, FieldEncoder, FieldEncodingContext,
23 FieldEncodingStrategy,
24 structural::{
25 PrimitiveFieldEncoding, PrimitivePageEncoding, try_create_binary_blob, try_create_list,
26 try_create_map, try_create_struct, try_create_structural_blob,
27 try_create_structural_fixed_size_list,
28 },
29 },
30};
31use lance_io::traits::Writer as ObjectWriter;
32
33use crate::{
34 reader::{ReadProjection, structural},
35 writer::FileWriterOptions,
36};
37
38mod compression;
39mod reader;
40mod writer;
41
42pub(crate) use reader::{
43 decode_column_metadata, finish_metadata, finish_metadata_index, validate_global_buffers,
44};
45pub use reader::{
46 projection_from_column_names, projection_from_field_ids, projection_from_whole_schema,
47};
48
49pub(crate) fn read_projection() -> Arc<dyn ReadProjection> {
50 structural::read_projection(reader::decode_column)
51}
52pub use writer::Writer;
53
54static WARNED_ON_UNSTABLE_FORMAT: AtomicBool = AtomicBool::new(false);
55
56pub fn physical_column_count(field: &Field) -> usize {
58 structural::physical_column_count(field)
59}
60
61pub fn data_file_columns(schema: &Schema) -> (Vec<i32>, Vec<i32>) {
63 structural::data_file_columns(schema)
64}
65
66pub(super) fn field_id_to_column_index(schema: &Schema) -> BTreeMap<u32, u32> {
67 structural::field_id_to_column_index(schema)
68}
69
70#[derive(Debug)]
71struct FieldStrategy {
72 primitive: PrimitiveFieldEncoding,
73}
74
75impl FieldEncodingStrategy for FieldStrategy {
76 fn create_field_encoder(
77 &self,
78 field: &Field,
79 column_index: &mut ColumnIndexSequence,
80 context: &FieldEncodingContext<'_>,
81 ) -> Result<Box<dyn FieldEncoder>> {
82 if let Some(encoder) =
83 try_create_binary_blob(&self.primitive, field, column_index, context)?
84 {
85 return Ok(encoder);
86 }
87 if let Some(encoder) =
88 try_create_structural_blob(&self.primitive, field, column_index, context)?
89 {
90 return Ok(encoder);
91 }
92 if field.is_blob() {
93 return Err(Error::invalid_input_source(
94 format!(
95 "Blob encoding is not available for field '{}' with data type {}",
96 field.name,
97 field.data_type()
98 )
99 .into(),
100 ));
101 }
102 if let Some(encoder) = try_create_map(field, column_index, context)? {
103 return Ok(encoder);
104 }
105 if let Some(encoder) = try_create_structural_fixed_size_list(field, column_index, context)?
106 {
107 return Ok(encoder);
108 }
109 if let Some(encoder) = self.primitive.try_create(field, column_index, context)? {
110 return Ok(encoder);
111 }
112 if let Some(encoder) = try_create_list(field, column_index, context)? {
113 return Ok(encoder);
114 }
115 if let Some(encoder) = try_create_struct(field, column_index, context)? {
116 return Ok(encoder);
117 }
118 Err(Error::not_supported_source(
119 format!(
120 "Lance v2.3 has no field encoding for '{}' with data type {}",
121 field.name,
122 field.data_type()
123 )
124 .into(),
125 ))
126 }
127}
128
129pub fn encoding_strategy(params: CompressionParams) -> Arc<dyn FieldEncodingStrategy> {
131 let compression = Arc::new(compression::Strategy::new(params));
132 Arc::new(FieldStrategy {
133 primitive: PrimitiveFieldEncoding::new([
134 PrimitivePageEncoding::sparse(compression.clone()),
135 PrimitivePageEncoding::constant(),
136 PrimitivePageEncoding::dense_u32(compression),
137 ]),
138 })
139}
140
141fn warn_unstable_format() {
142 if WARNED_ON_UNSTABLE_FORMAT
143 .compare_exchange(false, true, Ordering::Relaxed, Ordering::Relaxed)
144 .is_ok()
145 {
146 log::warn!(
147 "You have requested an unstable format version. Files written with this format version may not be readable in the future! This is a development feature and should only be used for experimentation and never for production data."
148 );
149 }
150}
151
152pub fn create_writer(
154 object_writer: Box<dyn ObjectWriter>,
155 schema: Schema,
156 options: FileWriterOptions,
157) -> Result<Writer> {
158 warn_unstable_format();
159 Writer::try_new(object_writer, schema, options)
160}
161
162pub fn create_writer_with_compression(
164 object_writer: Box<dyn ObjectWriter>,
165 schema: Schema,
166 options: FileWriterOptions,
167 compression: CompressionParams,
168) -> Result<Writer> {
169 warn_unstable_format();
170 Writer::try_new_with_compression(object_writer, schema, options, compression)
171}
172
173pub fn create_lazy_writer(
175 object_writer: Box<dyn ObjectWriter>,
176 options: FileWriterOptions,
177) -> Writer {
178 warn_unstable_format();
179 Writer::new_lazy(object_writer, options)
180}
181
182pub fn create_lazy_writer_with_compression(
184 object_writer: Box<dyn ObjectWriter>,
185 options: FileWriterOptions,
186 compression: CompressionParams,
187) -> Writer {
188 warn_unstable_format();
189 Writer::new_lazy_with_compression(object_writer, options, compression)
190}
191
192pub fn encode_self_described_batch(batch: &EncodedBatch) -> Result<Bytes> {
194 writer::concat_lance_footer(batch, true)
195}
196
197pub fn encode_mini_batch(batch: &EncodedBatch) -> Result<Bytes> {
199 writer::concat_lance_footer(batch, false)
200}