| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2023-09-02 | 1.8 kB | |
| v0.13.0 source code.tar.gz | 2023-09-02 | 605.3 kB | |
| v0.13.0 source code.zip | 2023-09-02 | 644.3 kB | |
| Totals: 3 Items | 1.3 MB | 0 | |
Changes
Core
- Add
datapipe.metastore.TransformMetaTable. Now each transform gets it's own meta table that tracks status of each transformation - Generalize
BatchTransformandDatatableBatchTransformthroughBaseBatchTransformStep - Add
transform_keysto*BatchTransform - Move changed idx computation out of
DataStoretoBaseBatchTransformStep - Add column
priorityto transform meta table, sort work by priority - Switch from vanilla
tqdmtotqdm_loggablefor better display in logs TableStoreFiledirconstructor accepts new argumentfsspec_kwargs- Add
filters,order_by,orderarguments to*BatchTransformStep - Add magic injection of
ds,idx,run_configto transform function via parameters introspection toBatchTransform - Add magic
dsinject intoBatchGenerate - Split
core_stepsintostep.batch_transform,step.batch_generate,step.datatable_transform,step.update_external_table - Move
metatable.MetaTabletodatatable - Enable WAL mode for sqlite database by default
CLI
- Add
step reset-metadataCLI command - Add
step fill-metadataCLI command that populates transform meta-table with all indices to process - Add
step run-idxCLI command - CLI
step run_changelistcommand accepts new argument--chunk-size - New CLI command
table migrate_transform_tablesfor0.13migration - Add
--start-stepparameter tostep run-changelistCLI - Move
--executorparameter fromdatapipe steptodatapipecommand
Execution
- Executors:
datapipe.executor.SingleThreadExecutor,datapipe.executor.ray.RayExecutor
Bugfixes
- Fix
QdrantStore.read_rowswhen no idx is specified - Fix
RedisStoreserialization for Ray