Tags: datastax/astrapy
Tags
Last minor improvements for 2.1.0rc1 prerelease (#368) * remove redundant serdes options in tests * reduced unused clutter in settings serdes options in unit tests * reduced unused clutter in settings serdes options in integration tests * adjust a silly typo in the udt async IT * ensure binary-encoded vectors are tested in all places for tables * all of the collection write path uses binary-encoded vectors now. Added IT * Added support for a 'definition' parameter in all create-index table methods, +full IT * version changed to 2.1.0rc1 * type-correct create-index-call-pattern tests now * other minor test-related fixes and improvements
Astrapy 2.0.0-preview: Tables and much more (#326) * WIP on tables, descriptor refactor, dse compose * database table methods: list/listnames sync+async done * first table (unit) tests * covering vector types in table defs * wip on tolerant parse/coerce of table options * table crd in place * table info structures handle unsupported columns * just introduced scalar-type enum * check for ok:1 in all database class table/coll CD ops * table index descriptor classes * create (regular/vector) index; drop index * better, shorter, privacy-oriented repr for Full/APIOptions/db classes * more streamlined and secure hiding of tokens in repr and such * repr for apicommander * add support for sourceModel in vector index options * add items to the cql type enum * most table vanity methods * finished table vanity methods; started refactor exceptions * binary is handled at the payload coded layer * removed the pointless {ok:1} returns * remove check_exists and CollectionAlreadyExistsException * rationalize table test dir; some table DML+testing * fix timeout s-vs-ms bugs, simplify timeout internal management, unflaky timeout tests * collection find_one uses API findOne command * WIP table generics * wip generics of collection * tiny edit * brought typing in all tests; add 1/4 typing-specific test * add test for typing collections+tables * rename the default Row and Document types * started cursor2 * cursor2 rewritten and wired to collection find. distinct todo * distinct+in cursors for collections + test/ruff * refine cursor timeouts and replace CursorIsStartedException -> CursorException * for_each, map (also rename overall_max_time_ms -> overall_timeout_ms) * test of timeouts in cursor 'reduce' operations * add typing test for map, etc * table cursors, table's find and distinct methods * improve class hierarchy for table column descriptors * table converters, core logic and most converters * TableSet class and its usage * TableMap and its usage * apitimestamp class, loss/y api optins known to payload conv, support in collections read/write * TableDuration type and its wiring * use stdlib ipaddress for 'inet' CQL type * TableDate class and its use in converters * TableTime type and its usage * DataAPITimestamp covers the stringy rfc3339 formats; used in table converters * DataAPIVector basics * binary encoding of vectors in collections + api options to control it * renamed wire-format options * limit vector bin-enc for collections to insertions * fix tests with the new tolerant-on-insert policy for types * guard against empty-keyspace DBs without 'keyspace' and 'keyspaces' fields on devops responses * add .region property to databases; optimize vector-path in coll pre/post process; adjust dbInfo tests * move the pre-post-processing logic out of commander, into collections (tables todo) * renamed option to 'unroll_iterables_to_list' and used in the general collection preproc route * rename tpostprocessors; nested nan in lists/maps and its test * table type-driven preprocessing wip (DataAPITimestamp missing + tests) * most pre/post processing for tables (except using the custom dt option); missing wiring * full table pre-post-processors, testing and use of options * sensible repr and str for DataAPIVector * converter agent for tables, with caching * unexpected response exceptions renamed; result classes split T/C; full wiring of pre-post-processing for tables * rename WireFormatOptions->SerdesOptions; better timeout messages that always give the number in ms * info methods of async classes are async; adjust collections test re: datetime/DataAPITimestamp; made custom types hashable * WIP on surfacing original timeouts * wip 2 on surfacing the nominal timeout values * surfacing the timeouts as the user configured them * cursors' for_each and to_list error on closed cursors * adjust tests to fetch_db_info signature; suspend dropIndex (the command is missing somehow) * cursor to_list and for_each will error if cursor is closed * equality method on CollectionDescriptor * mostly reworked exception tree; added InvalidEnvironmentException * rearrange and rename collection-specific exceptions * rearranged exception tree * final arrangement of exceptions all in astrapy.exceptions * cursor.has_next can start iteration on IDLE cursor * less unwieldy repr for XInsertManyResult if many items * cursors don't accept project after map is set * if/not/exists option and all the create/drop table/index commands ready * renamed all TableX custom data types to DataAPIX * drop_table_index now a Database method * delete_many for tables; fixes and int.tests for distinct; int.tests for delete_many/one * full options to (table)find_one + testing * wip on the decimal management in jsonencoding * full feature for the decimal-first flag in collections + unit tests * integration testing of the decimal encoding and related options for collections * unit test of table codec paths for decimal/number support * remove CollectionNotFoundException; cleaned out a forgotten table drop_index method * table deletion methods return None; removed associated XResult classes and related exceptions * alterTable and tests * revert to common CumulativeOperationException; remove Table{DeleteMany|UpdateMany}Exception and Table{Delete|Update}Result * rename test modules tableX to dataapiX; factored the 'c' format for duration logic in a separate module * wip DataAPIDuration format flip * DataAPIDuration switched to the ISO-8601-with-signum notation * table's count_documents, estimated_document_count methods + aliases + unit test * removed outdated 'astrapy_latest_interface.py' * table update_one method and test * table insert_many sync + tests * table insert_many, async + integration tests * fix Decimal schema-hashing bug; added IT for update_one * align cursor class names * more terse repr of collections and tables * improve error messages for multi-error DataAPIResponseExceptions * cursors, reverted to .data_source common property for tables/collections * v2.0 diagrams (abstractions, exceptions) * DataAPIClient init and spawner methods all accept api_options besides top named params * database's get_database_admin accepts admin_api_options parameter * removed unnecessary token coercion in tests * uniform api_options to all copy/clone/conversion methods * remove useless token coercion steps * minor (typing and typos) * Table.alter() supports typing and returns the newly-typed table * Table.alter() accepts a dict which it tries to parse into an Operation * cursors' get_sort_vector behaves as dictated by serdes options for tables/collections * include_similarity full support * support for source_model in create collection * add tests for paginated table fnid; rename SortDocuments to SortMode (with alias) * switch to the 6-tier timeout options * duration, use c-format in write path for lossless (for now) * remove row-named count methods * schema-based fillers for missing columns in row, * table's list_indexes and list_index_names methods + test * remove cursor .keyspace property * database 'command' method can target no-keyspace; update param names and streamlined logic * fixed some tests after the api and client minor changes * all tests (except countDocuments) passing again * rename max_time_ms to timeout_ms consistently * refactored the per-method timeout override chains in a zero-respecting manner * reinstated use of command() in a test * DataAPITimestamp has friendlier repr * adjusted the year/sign part of DataAPITimestamp parsing and stringifying * tiny fix to timestamp parsing (+0000123 allowed, 0000123 no for year) * adjusted DataAPIDate parsing and to_string to respect sign-of-year constraints as the API * add test for the timeout overrides including passing zero * added timeout-override and timeout-zero test for async collection * tables, int. test restructure+enrich wip * wip for tuple right typing in ids * proper type handling in insertedIds (dict+tuple) and its testing * a mostly complete integration test suite for tables * DataAPIMap and DataAPISet __eq__ method behaves like regular maps and sets; completed read-then-write-all-types tests * bugfix: token passthrough in the async_fetch_database_info * standard table vectorize integration tests * include sort vector, include similarity; covered by table vectorize tests * more table IT coverage * bigfix: equality of DataAPIMap * I test of table command * some more find IT for tables * increase default per-req timeout to 10s * tables, full IT coverage of find * full findOne IT for tables * full integration test coverage of table cursors * wip3 naming timeouts in errors * wip1 fine-align timeouts * wip2 fine-align timeouts * full alignment to timeout management * split Create- vs List- TableIndexDefinition and TableIndexDescriptor * index creation/list final API for tables * CollectionOptions -> CollectionDefinition class rename * from_dict -> _from_dict method consistently * cursors, for_each and to_list start where consumtion stopped * callback to for_each can stop early by returning False * added a tiny bit of testing to table cursors with foreach * added all collection cursor integration tests * table class docstrings * made some internal-method param required; remove any check/error on include_sort_vector and include_similarity * docstring fixes; in part. move tokens out of client (unless admin example docstrings) * table index creation new signature (for real) * first quickstart and some more timezonecare in the readme * table create_index has optional options param * better readme quickstarts (t/c both) * table.py docstrings: create_index, create_vector_index, list_index_names, list_indexes, alter (+async) * table docstring wip: halfway throug find * minor parameters to table-creation structures made optional; more table docstrings * more docstrings to table.py * sync table docstrings tier0 done * full tier-0 table.py docstring, incl async * more standardization of table.py docstrings * TableVectorIndexOptions can be coerced from None; create_vector_index has optional options; minor docs improvement * more docstring improvement around tables * final datetime/timezone serdes options, completed * adjusted tests in the wake of datetime tzinfo rework * systematc, consistent rewrite of the options to copy and spawn methods for the various abstractions * fix clipped datetime diagram picture * conftest param naming update * minor IT adjustments * create/get[table/coll] docstrings done, vectorize IT improved * adapt a vectorize test to DataAPIVector * wip on createCollection arg unflattening * create_collection signature rework + fluent interface, completed * table creation docstring order fix * column_type may be omitted for vector column type descriptor * param name change in builder method; better repr for DataAPIMap * AsyncDatabase: get_table and get_collection methods are not async functions anymore * environment locked in APIOptions * TODO cleanup in tests * api_options: all docstrings; fixed a couple of spawn/api_option parameter names here and there * datetime serdes example in readme * a small apioptions example in readme * minor api-option docstring changes * docstring for date and vector * docstring all of the custom data types * more docstrings * rename CursorState->FindCursorState; most cursor docstrings * full cursor docstrings * simplified drop_table and drop_collection to only accept the string name of the target to drop * wip typing finds * add type parameter to find methods for returned items * more docstring; all collection vector writes abide by binary_encode_vectors setting * disable (for now) collection vector binary encoding altogether * all docstrings in table.py and database_info.py * docstring for collection.py * all database.py docstrings done * table descriptors, docstrings ok * all docstrings done * primitive->scalar terminology consistently for columns * rewritten CHANGES file * mark version as preview; upgrade hcd docker-compose data api image * comment a stupid flaky test * another tricky async timeout test gone * another tricky async timeout test gone 2 * disabled most of the timeout tests with httpsserver * removed the alias SortMode->SortDocuments * builder interface renamed to .builder()---.build() pattern * renamed TableScalarColumnType -> ColumnType * wip toward proper list/create split for collection def/descr * split list/create table descriptors for real
More deprecations and several fixes for v1.5.1 (#318) * resilience against devops api not returning keyspace/keyspaces in the GET databases/ID endpoint * changesfile * Consistent handling of deletedCount=-1 from the API (always returned as-is) * bump to 1.5.1 + deprecate all set_caller calls * remove ragstack detection * deprecate caller_name/_version in favor of callers everywhere; tests to augment still * callers is Sequence; adjusted test_admin_conversions * adapted test_collections x2 * split caller_x vs callers in all unit tests * callers deprecation and identity tests completed * docstring examples ditch caller_name/caller_version altogether * list_databases + async have options and pagination * remove stray comment * docstr and py3.8 compliant type * reshuffled client.py to endpoint-first/then id + code optimization * better handling of no-endpoint, no-id init params for client.py * WIP on reshuffling admin which retain id/region as a second option * admin reshuffled: get_db_admin, get[async]database allow endpt AND id * endpoint-first flows, remove constructor unwanted params; logic cleanup and adjust unit tests * client.py deprecates use of id in get[async]database * all tests pass * rework cursors: state is enum; aligned state names * deprecation of Cursor.retrieved * cursor: deprecate 'collection' property for 'data_source' * changesfile * take cassio dependency out (which also removes numpy)
Deprecation of all things "namespace" (#316) * logging WARNINGs from Data API * deprecate bulk_write * keyspace method introduced + deprecation of namespace-same-named ones * property-adjusted testing in idiomatic * all namespace-methods deprecated and replicated with keyspace * introduce update_db_keyspace parameter, deprecate update_db_namespace + tests * fixed the deprecation of 'namespace' when it is a property * fix/simplify stacklevel logic for deprecation utilities * first sweep to rename internal variables * second sweep to rename internal variables * exceptions, constants and a few renames in tests * DatabaseInfo and CollectionInfo double-attributes * keyspace as parameter to all methods, with namespace as deprecated alias * all 'namespace=' renamed * swept/cleaned docstrings and stray references * most of the test assets cleaned as well * adapt to nulls-vs-absent ambiguity in findEmbProv responses * ensure forced dimensions overcome whether defaultValue=null for dimension in findEmvProv response * bump to 1.5.0
PreviousNext