|
| 1 | +use std::sync::Arc; |
| 2 | + |
| 3 | +use async_trait::async_trait; |
| 4 | + |
| 5 | +use datafusion::{ |
| 6 | + arrow::{ |
| 7 | + array::{Array, ArrayRef, BooleanBuilder, Int32Builder, StringBuilder, UInt32Builder}, |
| 8 | + datatypes::{DataType, Field, Schema}, |
| 9 | + record_batch::RecordBatch, |
| 10 | + }, |
| 11 | + datasource::{datasource::TableProviderFilterPushDown, TableProvider}, |
| 12 | + error::Result, |
| 13 | + logical_plan::Expr, |
| 14 | + physical_plan::{memory::MemoryExec, ExecutionPlan}, |
| 15 | +}; |
| 16 | + |
| 17 | +use crate::compile::engine::information_schema::postgres::PG_NAMESPACE_CATALOG_OID; |
| 18 | + |
| 19 | +struct PgCollation { |
| 20 | + oid: u32, |
| 21 | + collname: &'static str, |
| 22 | + collnamespace: u32, |
| 23 | + collowner: u32, |
| 24 | + collprovider: String, |
| 25 | + collisdeterministic: bool, |
| 26 | + collencoding: i32, |
| 27 | + collcollate: Option<String>, |
| 28 | + collctype: Option<String>, |
| 29 | + // Column `colliculocale` renamed to `colllocale` since PostgreSQL 17. |
| 30 | + // Support both columns for backward-compatibility. |
| 31 | + // Reference: https://pgpedia.info/p/pg_collation.html |
| 32 | + colliculocale: Option<String>, |
| 33 | + colllocale: Option<String>, |
| 34 | + collicurules: Option<String>, |
| 35 | + collversion: Option<String>, |
| 36 | +} |
| 37 | + |
| 38 | +struct PgCatalogCollationBuilder { |
| 39 | + oid: UInt32Builder, |
| 40 | + collname: StringBuilder, |
| 41 | + collnamespace: UInt32Builder, |
| 42 | + collowner: UInt32Builder, |
| 43 | + collprovider: StringBuilder, |
| 44 | + collisdeterministic: BooleanBuilder, |
| 45 | + collencoding: Int32Builder, |
| 46 | + colliculocale: StringBuilder, |
| 47 | + collcollate: StringBuilder, |
| 48 | + collctype: StringBuilder, |
| 49 | + colllocale: StringBuilder, |
| 50 | + collicurules: StringBuilder, |
| 51 | + collversion: StringBuilder, |
| 52 | +} |
| 53 | + |
| 54 | +impl PgCatalogCollationBuilder { |
| 55 | + fn new() -> Self { |
| 56 | + let capacity = 5; |
| 57 | + Self { |
| 58 | + oid: UInt32Builder::new(capacity), |
| 59 | + collname: StringBuilder::new(capacity), |
| 60 | + collnamespace: UInt32Builder::new(capacity), |
| 61 | + collowner: UInt32Builder::new(capacity), |
| 62 | + collprovider: StringBuilder::new(capacity), |
| 63 | + collisdeterministic: BooleanBuilder::new(capacity), |
| 64 | + collencoding: Int32Builder::new(capacity), |
| 65 | + colliculocale: StringBuilder::new(capacity), |
| 66 | + collcollate: StringBuilder::new(capacity), |
| 67 | + collctype: StringBuilder::new(capacity), |
| 68 | + colllocale: StringBuilder::new(capacity), |
| 69 | + collicurules: StringBuilder::new(capacity), |
| 70 | + collversion: StringBuilder::new(capacity), |
| 71 | + } |
| 72 | + } |
| 73 | + fn add_collation(&mut self, coll: &PgCollation) { |
| 74 | + self.oid.append_value(coll.oid).unwrap(); |
| 75 | + self.collname.append_value(coll.collname).unwrap(); |
| 76 | + self.collnamespace.append_value(coll.collnamespace).unwrap(); |
| 77 | + self.collowner.append_value(coll.collowner).unwrap(); |
| 78 | + self.collprovider |
| 79 | + .append_value(coll.collprovider.clone()) |
| 80 | + .unwrap(); |
| 81 | + self.collisdeterministic |
| 82 | + .append_value(coll.collisdeterministic) |
| 83 | + .unwrap(); |
| 84 | + self.collencoding.append_value(coll.collencoding).unwrap(); |
| 85 | + self.colliculocale |
| 86 | + .append_option(coll.colliculocale.clone()) |
| 87 | + .unwrap(); |
| 88 | + self.collcollate |
| 89 | + .append_option(coll.collcollate.clone()) |
| 90 | + .unwrap(); |
| 91 | + self.collctype |
| 92 | + .append_option(coll.collctype.clone()) |
| 93 | + .unwrap(); |
| 94 | + self.collicurules |
| 95 | + .append_option(coll.collicurules.clone()) |
| 96 | + .unwrap(); |
| 97 | + self.colllocale |
| 98 | + .append_option(coll.colllocale.clone()) |
| 99 | + .unwrap(); |
| 100 | + self.collversion |
| 101 | + .append_option(coll.collversion.clone()) |
| 102 | + .unwrap(); |
| 103 | + } |
| 104 | + |
| 105 | + fn finish(mut self) -> Vec<Arc<dyn Array>> { |
| 106 | + let columns: Vec<Arc<dyn Array>> = vec![ |
| 107 | + Arc::new(self.oid.finish()), |
| 108 | + Arc::new(self.collname.finish()), |
| 109 | + Arc::new(self.collnamespace.finish()), |
| 110 | + Arc::new(self.collowner.finish()), |
| 111 | + Arc::new(self.collprovider.finish()), |
| 112 | + Arc::new(self.collisdeterministic.finish()), |
| 113 | + Arc::new(self.collencoding.finish()), |
| 114 | + Arc::new(self.colliculocale.finish()), |
| 115 | + Arc::new(self.collcollate.finish()), |
| 116 | + Arc::new(self.collctype.finish()), |
| 117 | + Arc::new(self.collicurules.finish()), |
| 118 | + Arc::new(self.colllocale.finish()), |
| 119 | + Arc::new(self.collversion.finish()), |
| 120 | + ]; |
| 121 | + columns |
| 122 | + } |
| 123 | +} |
| 124 | + |
| 125 | +pub struct PgCatalogCollationProvider { |
| 126 | + data: Arc<Vec<ArrayRef>>, |
| 127 | +} |
| 128 | + |
| 129 | +impl PgCatalogCollationProvider { |
| 130 | + pub fn new() -> Self { |
| 131 | + // See https://github.com/postgres/postgres/blob/REL_16_4/src/include/catalog/pg_collation.h |
| 132 | + let mut builder = PgCatalogCollationBuilder::new(); |
| 133 | + |
| 134 | + // Initial contents of the pg_collation system catalog. |
| 135 | + // See https://github.com/postgres/postgres/blob/REL_16_4/src/include/catalog/pg_collation.dat |
| 136 | + builder.add_collation(&PgCollation { |
| 137 | + oid: 100, |
| 138 | + collname: "default", |
| 139 | + collnamespace: PG_NAMESPACE_CATALOG_OID, |
| 140 | + collowner: 10, |
| 141 | + collprovider: "d".to_string(), |
| 142 | + collisdeterministic: true, |
| 143 | + collencoding: -1, |
| 144 | + collcollate: None, |
| 145 | + colliculocale: None, |
| 146 | + collctype: None, |
| 147 | + colllocale: None, |
| 148 | + collicurules: None, |
| 149 | + collversion: None, |
| 150 | + }); |
| 151 | + builder.add_collation(&PgCollation { |
| 152 | + oid: 950, |
| 153 | + collname: "C", |
| 154 | + collnamespace: PG_NAMESPACE_CATALOG_OID, |
| 155 | + collowner: 10, |
| 156 | + collprovider: "c".to_string(), |
| 157 | + collisdeterministic: true, |
| 158 | + collencoding: -1, |
| 159 | + collcollate: Some("C".to_string()), |
| 160 | + colliculocale: Some("C".to_string()), |
| 161 | + collctype: Some("C".to_string()), |
| 162 | + colllocale: None, |
| 163 | + collicurules: None, |
| 164 | + collversion: None, |
| 165 | + }); |
| 166 | + builder.add_collation(&PgCollation { |
| 167 | + oid: 951, |
| 168 | + collname: "POSIX", |
| 169 | + collnamespace: PG_NAMESPACE_CATALOG_OID, |
| 170 | + collowner: 10, |
| 171 | + collprovider: "c".to_string(), |
| 172 | + collisdeterministic: true, |
| 173 | + collencoding: -1, |
| 174 | + collcollate: Some("POSIX".to_string()), |
| 175 | + colliculocale: Some("POSIX".to_string()), |
| 176 | + collctype: Some("POSIX".to_string()), |
| 177 | + colllocale: None, |
| 178 | + collicurules: None, |
| 179 | + collversion: None, |
| 180 | + }); |
| 181 | + builder.add_collation(&PgCollation { |
| 182 | + oid: 962, |
| 183 | + collname: "usc_basic", |
| 184 | + collnamespace: PG_NAMESPACE_CATALOG_OID, |
| 185 | + collowner: 10, |
| 186 | + collprovider: "c".to_string(), |
| 187 | + collisdeterministic: true, |
| 188 | + collencoding: 6, |
| 189 | + collcollate: Some("C".to_string()), |
| 190 | + colliculocale: Some("C".to_string()), |
| 191 | + collctype: Some("C".to_string()), |
| 192 | + colllocale: None, |
| 193 | + collicurules: None, |
| 194 | + collversion: None, |
| 195 | + }); |
| 196 | + builder.add_collation(&PgCollation { |
| 197 | + oid: 963, |
| 198 | + collname: "unicode", |
| 199 | + collnamespace: PG_NAMESPACE_CATALOG_OID, |
| 200 | + collowner: 10, |
| 201 | + collprovider: "i".to_string(), |
| 202 | + collisdeterministic: true, |
| 203 | + collencoding: -1, |
| 204 | + collcollate: None, |
| 205 | + colliculocale: None, |
| 206 | + collctype: None, |
| 207 | + colllocale: Some("und".to_string()), |
| 208 | + collicurules: None, |
| 209 | + collversion: Some("153.121".to_string()), |
| 210 | + }); |
| 211 | + Self { |
| 212 | + data: Arc::new(builder.finish()), |
| 213 | + } |
| 214 | + } |
| 215 | +} |
| 216 | + |
| 217 | +#[async_trait] |
| 218 | +impl TableProvider for PgCatalogCollationProvider { |
| 219 | + fn as_any(&self) -> &dyn std::any::Any { |
| 220 | + self |
| 221 | + } |
| 222 | + fn schema(&self) -> datafusion::arrow::datatypes::SchemaRef { |
| 223 | + Arc::new(Schema::new(vec![ |
| 224 | + Field::new("oid", DataType::UInt32, false), |
| 225 | + Field::new("collname", DataType::Utf8, false), |
| 226 | + Field::new("collnamespace", DataType::UInt32, false), |
| 227 | + ])) |
| 228 | + } |
| 229 | + async fn scan( |
| 230 | + &self, |
| 231 | + projection: &Option<Vec<usize>>, |
| 232 | + _filters: &[Expr], |
| 233 | + // limit can be used to reduce the amount scanned |
| 234 | + // from the datasource as a performance optimization. |
| 235 | + // If set, it contains the amount of rows needed by the `LogicalPlan`, |
| 236 | + // The datasource should return *at least* this number of rows if available. |
| 237 | + _limit: Option<usize>, |
| 238 | + ) -> Result<Arc<dyn ExecutionPlan>> { |
| 239 | + let batch = RecordBatch::try_new(self.schema(), self.data.to_vec())?; |
| 240 | + Ok(Arc::new(MemoryExec::try_new( |
| 241 | + &[vec![batch]], |
| 242 | + self.schema(), |
| 243 | + projection.clone(), |
| 244 | + )?)) |
| 245 | + } |
| 246 | + fn supports_filter_pushdown(&self, _filter: &Expr) -> Result<TableProviderFilterPushDown> { |
| 247 | + Ok(TableProviderFilterPushDown::Unsupported) |
| 248 | + } |
| 249 | +} |
0 commit comments