1use std::collections::HashMap;
19use std::sync::Arc;
20
21use arrow_array::RecordBatch;
22use arrow_array::builder::{
23 BooleanBuilder, GenericListBuilder, ListBuilder, PrimitiveBuilder, StringBuilder, StructBuilder,
24};
25use arrow_array::types::{Int32Type, Int64Type};
26use arrow_schema::{DataType, Field, Fields};
27use futures::{StreamExt, stream};
28
29use crate::arrow::schema_to_arrow_schema;
30use crate::scan::ArrowRecordBatchStream;
31use crate::spec::{Datum, FieldSummary, ListType, NestedField, PrimitiveType, StructType, Type};
32use crate::table::Table;
33use crate::{Error, ErrorKind, Result};
34
35pub struct ManifestsTable<'a> {
37 table: &'a Table,
38}
39
40impl<'a> ManifestsTable<'a> {
41 pub fn new(table: &'a Table) -> Self {
43 Self { table }
44 }
45
46 pub fn schema(&self) -> crate::spec::Schema {
48 let fields = vec![
49 NestedField::new(14, "content", Type::Primitive(PrimitiveType::Int), true),
50 NestedField::new(1, "path", Type::Primitive(PrimitiveType::String), true),
51 NestedField::new(2, "length", Type::Primitive(PrimitiveType::Long), true),
52 NestedField::new(
53 3,
54 "partition_spec_id",
55 Type::Primitive(PrimitiveType::Int),
56 true,
57 ),
58 NestedField::new(
59 4,
60 "added_snapshot_id",
61 Type::Primitive(PrimitiveType::Long),
62 true,
63 ),
64 NestedField::new(
65 5,
66 "added_data_files_count",
67 Type::Primitive(PrimitiveType::Int),
68 true,
69 ),
70 NestedField::new(
71 6,
72 "existing_data_files_count",
73 Type::Primitive(PrimitiveType::Int),
74 true,
75 ),
76 NestedField::new(
77 7,
78 "deleted_data_files_count",
79 Type::Primitive(PrimitiveType::Int),
80 true,
81 ),
82 NestedField::new(
83 15,
84 "added_delete_files_count",
85 Type::Primitive(PrimitiveType::Int),
86 true,
87 ),
88 NestedField::new(
89 16,
90 "existing_delete_files_count",
91 Type::Primitive(PrimitiveType::Int),
92 true,
93 ),
94 NestedField::new(
95 17,
96 "deleted_delete_files_count",
97 Type::Primitive(PrimitiveType::Int),
98 true,
99 ),
100 NestedField::new(
101 8,
102 "partition_summaries",
103 Type::List(ListType {
104 element_field: Arc::new(NestedField::new(
105 9,
106 "item",
107 Type::Struct(StructType::new(vec![
108 Arc::new(NestedField::new(
109 10,
110 "contains_null",
111 Type::Primitive(PrimitiveType::Boolean),
112 true,
113 )),
114 Arc::new(NestedField::new(
115 11,
116 "contains_nan",
117 Type::Primitive(PrimitiveType::Boolean),
118 false,
119 )),
120 Arc::new(NestedField::new(
121 12,
122 "lower_bound",
123 Type::Primitive(PrimitiveType::String),
124 false,
125 )),
126 Arc::new(NestedField::new(
127 13,
128 "upper_bound",
129 Type::Primitive(PrimitiveType::String),
130 false,
131 )),
132 ])),
133 true,
134 )),
135 }),
136 true,
137 ),
138 ];
139
140 crate::spec::Schema::builder()
141 .with_fields(fields.into_iter().map(|f| f.into()))
142 .build()
143 .unwrap()
144 }
145
146 pub async fn scan(&self) -> Result<ArrowRecordBatchStream> {
148 let schema = schema_to_arrow_schema(&self.schema())?;
149
150 let mut content = PrimitiveBuilder::<Int32Type>::new();
151 let mut path = StringBuilder::new();
152 let mut length = PrimitiveBuilder::<Int64Type>::new();
153 let mut partition_spec_id = PrimitiveBuilder::<Int32Type>::new();
154 let mut added_snapshot_id = PrimitiveBuilder::<Int64Type>::new();
155 let mut added_data_files_count = PrimitiveBuilder::<Int32Type>::new();
156 let mut existing_data_files_count = PrimitiveBuilder::<Int32Type>::new();
157 let mut deleted_data_files_count = PrimitiveBuilder::<Int32Type>::new();
158 let mut added_delete_files_count = PrimitiveBuilder::<Int32Type>::new();
159 let mut existing_delete_files_count = PrimitiveBuilder::<Int32Type>::new();
160 let mut deleted_delete_files_count = PrimitiveBuilder::<Int32Type>::new();
161 let mut partition_summaries = self.partition_summary_builder()?;
162
163 if let Some(snapshot) = self.table.metadata().current_snapshot() {
164 let manifest_list = self.table.manifest_list_reader(snapshot).load().await?;
165 for manifest in manifest_list.entries() {
166 content.append_value(manifest.content as i32);
167 path.append_value(manifest.manifest_path.clone());
168 length.append_value(manifest.manifest_length);
169 partition_spec_id.append_value(manifest.partition_spec_id);
170 added_snapshot_id.append_value(manifest.added_snapshot_id);
171 added_data_files_count.append_value(manifest.added_files_count.unwrap_or(0) as i32);
172 existing_data_files_count
173 .append_value(manifest.existing_files_count.unwrap_or(0) as i32);
174 deleted_data_files_count
175 .append_value(manifest.deleted_files_count.unwrap_or(0) as i32);
176 added_delete_files_count
177 .append_value(manifest.added_files_count.unwrap_or(0) as i32);
178 existing_delete_files_count
179 .append_value(manifest.existing_files_count.unwrap_or(0) as i32);
180 deleted_delete_files_count
181 .append_value(manifest.deleted_files_count.unwrap_or(0) as i32);
182
183 let spec = self
184 .table
185 .metadata()
186 .partition_spec_by_id(manifest.partition_spec_id)
187 .ok_or_else(|| {
188 Error::new(
189 ErrorKind::DataInvalid,
190 format!(
191 "Partition spec {} for manifest {} is not in table metadata",
192 manifest.partition_spec_id, manifest.manifest_path
193 ),
194 )
195 })?;
196 let spec_struct = spec.partition_type(self.table.metadata().current_schema())?;
197 self.append_partition_summaries(
198 &mut partition_summaries,
199 manifest.partitions.as_deref().unwrap_or(&[]),
200 spec_struct,
201 );
202 }
203 }
204
205 let batch = RecordBatch::try_new(Arc::new(schema), vec![
206 Arc::new(content.finish()),
207 Arc::new(path.finish()),
208 Arc::new(length.finish()),
209 Arc::new(partition_spec_id.finish()),
210 Arc::new(added_snapshot_id.finish()),
211 Arc::new(added_data_files_count.finish()),
212 Arc::new(existing_data_files_count.finish()),
213 Arc::new(deleted_data_files_count.finish()),
214 Arc::new(added_delete_files_count.finish()),
215 Arc::new(existing_delete_files_count.finish()),
216 Arc::new(deleted_delete_files_count.finish()),
217 Arc::new(partition_summaries.finish()),
218 ])?;
219 Ok(stream::iter(vec![Ok(batch)]).boxed())
220 }
221
222 fn partition_summary_builder(&self) -> Result<GenericListBuilder<i32, StructBuilder>> {
223 let schema = schema_to_arrow_schema(&self.schema())?;
224 let partition_summary_fields =
225 match schema.field_with_name("partition_summaries")?.data_type() {
226 DataType::List(list_type) => match list_type.data_type() {
227 DataType::Struct(fields) => fields.to_vec(),
228 _ => unreachable!(),
229 },
230 _ => unreachable!(),
231 };
232
233 let partition_summaries = ListBuilder::new(StructBuilder::from_fields(
234 Fields::from(partition_summary_fields.clone()),
235 0,
236 ))
237 .with_field(Arc::new(
238 Field::new_struct("item", partition_summary_fields, false).with_metadata(
239 HashMap::from([("PARQUET:field_id".to_string(), "9".to_string())]),
240 ),
241 ));
242
243 Ok(partition_summaries)
244 }
245
246 fn append_partition_summaries(
247 &self,
248 builder: &mut GenericListBuilder<i32, StructBuilder>,
249 partitions: &[FieldSummary],
250 partition_struct: StructType,
251 ) {
252 let partition_summaries_builder = builder.values();
253 for (summary, field) in partitions.iter().zip(partition_struct.fields()) {
254 partition_summaries_builder
255 .field_builder::<BooleanBuilder>(0)
256 .unwrap()
257 .append_value(summary.contains_null);
258 partition_summaries_builder
259 .field_builder::<BooleanBuilder>(1)
260 .unwrap()
261 .append_option(summary.contains_nan);
262
263 partition_summaries_builder
264 .field_builder::<StringBuilder>(2)
265 .unwrap()
266 .append_option(summary.lower_bound.as_ref().map(|v| {
267 Datum::try_from_bytes(v, field.field_type.as_primitive_type().unwrap().clone())
268 .unwrap()
269 .to_string()
270 }));
271 partition_summaries_builder
272 .field_builder::<StringBuilder>(3)
273 .unwrap()
274 .append_option(summary.upper_bound.as_ref().map(|v| {
275 Datum::try_from_bytes(v, field.field_type.as_primitive_type().unwrap().clone())
276 .unwrap()
277 .to_string()
278 }));
279 partition_summaries_builder.append(true);
280 }
281 builder.append(true);
282 }
283}
284
285#[cfg(test)]
286mod tests {
287 use std::sync::Arc;
288
289 use expect_test::expect;
290 use futures::TryStreamExt;
291
292 use crate::scan::tests::TableTestFixture;
293 use crate::spec::TableMetadata;
294 use crate::test_utils::check_record_batches;
295
296 #[tokio::test]
297 async fn test_manifests_table() {
298 let mut fixture = TableTestFixture::new();
299 fixture.setup_manifest_files().await;
300
301 let record_batch = fixture.table.inspect().manifests().scan().await.unwrap();
302
303 check_record_batches(
304 record_batch.try_collect::<Vec<_>>().await.unwrap(),
305 expect![[r#"
306 Field { "content": Int32, metadata: {"PARQUET:field_id": "14"} },
307 Field { "path": Utf8, metadata: {"PARQUET:field_id": "1"} },
308 Field { "length": Int64, metadata: {"PARQUET:field_id": "2"} },
309 Field { "partition_spec_id": Int32, metadata: {"PARQUET:field_id": "3"} },
310 Field { "added_snapshot_id": Int64, metadata: {"PARQUET:field_id": "4"} },
311 Field { "added_data_files_count": Int32, metadata: {"PARQUET:field_id": "5"} },
312 Field { "existing_data_files_count": Int32, metadata: {"PARQUET:field_id": "6"} },
313 Field { "deleted_data_files_count": Int32, metadata: {"PARQUET:field_id": "7"} },
314 Field { "added_delete_files_count": Int32, metadata: {"PARQUET:field_id": "15"} },
315 Field { "existing_delete_files_count": Int32, metadata: {"PARQUET:field_id": "16"} },
316 Field { "deleted_delete_files_count": Int32, metadata: {"PARQUET:field_id": "17"} },
317 Field { "partition_summaries": List(non-null Struct("contains_null": non-null Boolean, metadata: {"PARQUET:field_id": "10"}, "contains_nan": Boolean, metadata: {"PARQUET:field_id": "11"}, "lower_bound": Utf8, metadata: {"PARQUET:field_id": "12"}, "upper_bound": Utf8, metadata: {"PARQUET:field_id": "13"}), metadata: {"PARQUET:field_id": "9"}), metadata: {"PARQUET:field_id": "8"} }"#]],
318 expect![[r#"
319 content: PrimitiveArray<Int32>
320 [
321 0,
322 ],
323 path: (skipped),
324 length: (skipped),
325 partition_spec_id: PrimitiveArray<Int32>
326 [
327 0,
328 ],
329 added_snapshot_id: PrimitiveArray<Int64>
330 [
331 3055729675574597004,
332 ],
333 added_data_files_count: PrimitiveArray<Int32>
334 [
335 1,
336 ],
337 existing_data_files_count: PrimitiveArray<Int32>
338 [
339 1,
340 ],
341 deleted_data_files_count: PrimitiveArray<Int32>
342 [
343 1,
344 ],
345 added_delete_files_count: PrimitiveArray<Int32>
346 [
347 1,
348 ],
349 existing_delete_files_count: PrimitiveArray<Int32>
350 [
351 1,
352 ],
353 deleted_delete_files_count: PrimitiveArray<Int32>
354 [
355 1,
356 ],
357 partition_summaries: ListArray
358 [
359 StructArray
360 -- validity:
361 [
362 valid,
363 ]
364 [
365 -- child 0: "contains_null" (Boolean)
366 BooleanArray
367 [
368 false,
369 ]
370 -- child 1: "contains_nan" (Boolean)
371 BooleanArray
372 [
373 false,
374 ]
375 -- child 2: "lower_bound" (Utf8)
376 StringArray
377 [
378 "100",
379 ]
380 -- child 3: "upper_bound" (Utf8)
381 StringArray
382 [
383 "300",
384 ]
385 ],
386 ]"#]],
387 &["path", "length"],
388 Some("path"),
389 );
390 }
391
392 #[tokio::test]
393 async fn test_manifests_table_with_dropped_partition_source_column() {
394 let mut fixture = TableTestFixture::new();
395 fixture.setup_manifest_files().await;
396
397 let mut metadata = serde_json::to_value(fixture.table.metadata()).unwrap();
401 let current_schema_id = metadata["current-schema-id"].clone();
402 let schemas = metadata["schemas"].as_array_mut().unwrap();
403 for schema in schemas {
404 if schema["schema-id"] == current_schema_id {
405 let fields = schema["fields"].as_array_mut().unwrap();
406 fields.retain(|field| field["id"] != 1);
407 let identifier_ids = schema["identifier-field-ids"].as_array_mut().unwrap();
408 identifier_ids.retain(|id| *id != 1);
409 }
410 }
411
412 metadata["partition-specs"]
413 .as_array_mut()
414 .unwrap()
415 .push(serde_json::json!({"spec-id": 1, "fields": []}));
416 metadata["default-spec-id"] = serde_json::json!(1);
417
418 let metadata: TableMetadata = serde_json::from_value(metadata).unwrap();
419 let table = fixture.table.clone().with_metadata(Arc::new(metadata));
420
421 match table.inspect().manifests().scan().await {
422 Err(err) => assert!(err.to_string().contains("No column with source column id")),
423 Ok(_) => panic!("expected scan to fail for a dropped partition source column"),
424 }
425 }
426}