o ȳgz_@sdZddlmZddlZddlmZmZmZmZm Z m Z m Z m Z ddl mZmZmZmZddlmZerd?ZdXd@dAZ dQdYdCdDZe  dZd[dEdFZdS)\CassandraDatabaseu$Apache Cassandra® database wrapper.NsessionOptional[Session]exclude_tablesOptional[List[str]]include_tablescassio_init_kwargsOptional[Dict[str, Any]]cCs<|||}|s td||_t|_|pg|_|pg|_dS)Nz+Session not provided and cannot be resolved)_resolve_session ValueError_sessionIGNORED_KEYSPACES_exclude_keyspaces_exclude_tables_include_tables)selfrrrr r$r*l/var/www/html/chatdoc2/venv/lib/python3.10/site-packages/langchain_community/utilities/cassandra_database.py__init__s  zCassandraDatabase.__init__allquerystrfetchkwargsrreturn&Union[list, Dict[str, Any], ResultSet]cKsV|dkr |j|fi|S|dkr|j|fi|S|dkr'|j|fi|Std)z+Execute a CQL query and return the results.r-onecursorz8Fetch parameter must be either 'one', 'all', or 'cursor') fetch_all fetch_one_fetchr#)r)r.r0r1r*r*r+run-szCassandraDatabase.runrcKs ||d}|jj|fi|S)NSELECT) _validate_cqlr$execute)r)r.r1 clean_queryr*r*r+r8=s zCassandraDatabase._fetchlistcKst|j|fi|SN)r>r8)r)r.r1r*r*r+r6AszCassandraDatabase.fetch_allDict[str, Any]cKs&|j|fi|}|r|SiSr?)r8r4_asdict)r)r.r1resultr*r*r+r7DszCassandraDatabase.fetch_onekeyspace List[Table]cCs ||g}||vr||SgS)z1Get the Table objects for the specified keyspace.)_resolve_schema)r)rCschemar*r*r+get_keyspace_tablesHs z%CassandraDatabase.get_keyspace_tablestable predicatelimitintcCs^d|d|}|r|d|7}|r|d|7}|d7}||}ddd|D}|S) zbsz3CassandraDatabase.get_table_data..)r6join)r)rCrHrIrJr.rBdatar*r*r+get_table_dataSs z CassandraDatabase.get_table_datacCs|}dd|iS)z4Return db context that you may want in agent prompt. keyspaces, )_fetch_keyspacesrR)r)rUr*r*r+ get_contexteszCassandraDatabase.get_contexttablesOptional[List[Table]]cCsZ|s||}|r+d|d}|r%|D]}||jddd7}|d7}q|S|d7}|SdS)a Generates a markdown representation of the schema for a specific keyspace by iterating over all tables within that keyspace and calling their as_markdown method. Args: keyspace: The name of the keyspace to generate markdown documentation for. tables: list of tables in the keyspace; it will be resolved if not provided. Returns: A string containing the markdown representation of the specified keyspace schema. z ## Keyspace:  F)include_keyspace header_levelzNo tables present in keyspace )rG as_markdown)r)rCrYoutputrHr*r*r+format_keyspace_to_markdownjs   z-CassandraDatabase.format_keyspace_to_markdowncCs8|}d}|D]\}}||||d7}q |S)a Generates a markdown representation of the schema for all keyspaces and tables within the CassandraDatabase instance. This method utilizes the format_keyspace_to_markdown method to create markdown sections for each keyspace, assembling them into a comprehensive schema document. Iterates through each keyspace in the database, utilizing format_keyspace_to_markdown to generate markdown for each keyspace's schema, including details of its tables. These sections are concatenated to form a single markdown document that represents the schema of the entire database or the subset of keyspaces that have been resolved in this instance. Returns: A markdown string that documents the schema of all resolved keyspaces and their tables within this CassandraDatabase instance. This includes keyspace names, table names, comments, columns, partition keys, clustering keys, and indexes for each table. z# Cassandra Database Schema r[)rEitemsrb)r)rFrarCrYr*r*r+format_schema_to_markdowns z+CassandraDatabase.format_schema_to_markdownr:cqltypecCsdg}|r||vrtd|d||}||s,td|d|d}tdd|}td d|}d|vrGtd |S) a Validates a CQL query string for basic formatting and safety checks. Ensures that `cql` starts with the specified type (e.g., SELECT) and does not contain content that could indicate CQL injection vulnerabilities. Args: cql: The CQL query string to be validated. type: The expected starting keyword of the query, used to verify that the query begins with the correct operation type (e.g., "SELECT", "UPDATE"). Defaults to "SELECT". Returns: The trimmed and validated CQL query string without a trailing semicolon. Raises: ValueError: If the value of `type` is not supported DatabaseError: If `cql` is considered unsafe r:zUnsupported CQL type: z1. Supported types: zCQL must start with rLrMz'.*?'r_z".*?"zPotentially unsafe CQL, as it contains a ; at a place other than the end or within quotation marks.)upperr#strip startswith DatabaseErrorrstripresub)r)rerfSUPPORTED_TYPES cql_trimmed cql_sanitizedr*r*r+r;s$ zCassandraDatabase._validate_cqlrU List[str]cCs`|d}g}|D]$}t|tsq |d}|r!||vr!||q |s-||jvr-||q |S)a Fetches a list of keyspace names from the Cassandra database. The list can be filtered by a provided list of keyspace names or by excluding predefined keyspaces. Args: keyspaces: A list of keyspace names to specifically include. If provided and not empty, the method returns only the keyspaces present in this list. If not provided or empty, the method returns all keyspaces except those specified in the _exclude_keyspaces attribute. Returns: A list of keyspace names according to the filtering criteria. z1SELECT keyspace_name FROM system_schema.keyspaces keyspace_name)r6 isinstancerappendr&)r)rU all_keyspacesfiltered_keyspacesksrrr*r*r+rWs    z"CassandraDatabase._fetch_keyspacescCs$ddd|D}|d|dS)NrVcSsg|]}d|dqS)'r*)rOrwr*r*r+ sz.z WHERE keyspace_name IN ())rR)r)r.rUkeyspace_in_clauser*r*r+_format_keyspace_querysz(CassandraDatabase._format_keyspace_querycC|d|}||S)aFetches tables schema data, filtered by a list of keyspaces. This method allows for efficiently fetching schema information for multiple keyspaces in a single operation, enabling applications to programmatically analyze or document the database schema. Args: keyspaces: A list of keyspace names from which to fetch tables schema data. Returns: Dictionaries of table details (keyspace name, table name, and comment). zDSELECT keyspace_name, table_name, comment FROM system_schema.tablesr|r6r)rU tables_queryr*r*r+_fetch_tables_datas   z$CassandraDatabase._fetch_tables_datacCr})aFetches columns schema data, filtered by a list of keyspaces. This method allows for efficiently fetching schema information for multiple keyspaces in a single operation, enabling applications to programmatically analyze or document the database schema. Args: keyspaces: A list of keyspace names from which to fetch tables schema data. Returns: Dictionaries of column details (keyspace name, table name, column name, type, kind, and position). z SELECT keyspace_name, table_name, column_name, type, kind, clustering_order, position FROM system_schema.columns r~rr*r*r+_fetch_columns_data   z%CassandraDatabase._fetch_columns_datacCr})a Fetches indexes schema data, filtered by a list of keyspaces. This method allows for efficiently fetching schema information for multiple keyspaces in a single operation, enabling applications to programmatically analyze or document the database schema. Args: keyspaces: A list of keyspace names from which to fetch tables schema data. Returns: Dictionaries of index details (keyspace name, table name, index name, kind, and options). z SELECT keyspace_name, table_name, index_name, kind, options FROM system_schema.indexes r~rr*r*r+_fetch_indexes_data&rz%CassandraDatabase._fetch_indexes_dataDict[str, List[Table]]c s|s|}||}||}||}i}|D]_}|j|j|j}|jr-|jvr-q|jr6|jvr6qfdd|D}fdd|D} fdd|D} fdd|D} t ||| | | d} |vrqg|<| | q|S)a? Efficiently fetches and organizes Cassandra table schema information, such as comments, columns, and indexes, into a dictionary mapping keyspace names to lists of Table objects. Args: keyspaces: An optional list of keyspace names from which to fetch tables schema data. Returns: A dictionary with keyspace names as keys and lists of Table objects as values, where each Table object is populated with schema details appropriate for its keyspace and table name. cs,g|]}|jkr|jkr|j|jfqSr*)rr table_name column_namerfrOcrCrr*r+rybs  z5CassandraDatabase._resolve_schema..cs0g|]}|jdkr|jkr|jkr|jqS) partition_key)kindrrrrrrr*r+ryhs   cs6g|]}|jdkr|jkr|jkr|j|jfqS) clustering)rrrrrclustering_orderrrr*r+ryps    cs0g|]}|jkr|jkr|j|j|jfqSr*)rrr index_nameroptionsrrr*r+ryxs rCrcommentcolumns partitionrindexes) rWrrrrrrrr(r'Tablert) r)rU tables_data columns_data indexes_data keyspace_dict table_datar table_columnspartition_keysclustering_keys table_indexes table_objr*rr+rE=sN        z!CassandraDatabase._resolve_schemacCsv|r|Szddl}Wn tytdw|j}|r|S|r9t|tr5|jdi||j}|StddS)a% Attempts to resolve and return a Session object for use in database operations. This function follows a specific order of precedence to determine the appropriate session to use: 1. `session` parameter if given, 2. Existing `cassio` session, 3. A new `cassio` session derived from `cassio_init_kwargs`, 4. `None` Args: session: An optional session to use directly. cassio_init_kwargs: An optional dictionary of keyword arguments to `cassio`. Returns: The resolved session object if successful, or `None` if the session cannot be resolved. Raises: ValueError: If `cassio_init_kwargs` is provided but is not a dictionary of keyword arguments. rNzBcassio package not found, please install with `pip install cassio`z/cassio_init_kwargs must be a keyword dictionaryr*) cassio.config ImportErrorr#configresolve_sessionrsdictinitcheck_resolve_session)rr cassiosr*r*r+r"s&     z"CassandraDatabase._resolve_session)NNNN)rrrrrrr r!)r-)r.r/r0r/r1rr2r3)r.r/r1rr2r)r.r/r1rr2r>)r.r/r1rr2r@)rCr/r2rD) rCr/rHr/rIr/rJrKr2r/)r2r@r?)rCr/rYrZr2r/)r2r/)r:)rer/rfr/r2r/)rUrr2rq)r.r/rUrqr2r/)rUrqr2r>)rUrr2r)NN)rrr r!r2r)__name__ __module__ __qualname____doc__r,r9r8r6r7rGrTrXrbrdr;rWr|rrrrE staticmethodr"r*r*r*r+rs<        3 "    Qrcs"eZdZdZdfdd ZZS)rjzuException raised for errors in the database schema. Attributes: message -- explanation of the error messager/cs||_t|jdSr?)rsuperr,)r)r __class__r*r+r,szDatabaseError.__init__)rr/)rrrrr, __classcell__r*r*rr+rjsrjc@seZdZUded< ded< dZded< eedZded <eedZd ed <eedZ ded <eedZ d ed<e ddZ e ddd(ddZed)ddZ d*d+dd Zed,d!d"Zed-d$d%Zed.d&d'ZdS)/rr/rCrN Optional[str]r)default_factoryzList[Tuple[str, str]]rrqrrList[Tuple[str, str, str]]rT)frozenafter)moder2rcCs |jstd|jstd|S)Nz*non-empty column list for must be providedz)non-empty partition list must be provided)rr#r)r)r*r*r+check_required_fieldss zTable.check_required_fieldsdbrc Cs>||||\}}}||||||||||||||dS)Nr)_resolve_columns_resolve_comment_resolve_indexes)clsrCrrrrrr*r*r+ from_databases  zTable.from_databaser]boolr^ Optional[int]c CsHd}|dur|d|d7}|d|jd7}|r#|d|jd7}|jr/|d|jd7}|d 7}|jD]\}}|d |d |d 7}q6|d d|jd 7}|d7}|jrg}|jD]\}}|dkrn||q^||d|q^|dd|d 7}|j r|d7}|j D]\}} } |d |d| d| d7}q|S)a Generates a Markdown representation of the Cassandra table schema, allowing for customizable header levels for the table name section. Args: include_keyspace: If True, includes the keyspace in the output. Defaults to True. header_level: Specifies the markdown header level for the table name. If None, the table name is included without a header. Defaults to None (no header level). Returns: A string in Markdown format detailing the table name (with optional header level), keyspace (optional), comment, columns, partition keys, clustering keys (with optional clustering order), and indexes. r_N# z Table Name: rNz - Keyspace: z - Comment: z - Columns z - z (z) z- Partition Keys: (rVz- Clustering Keys: none(z - Indexes z : kind=z , options=) rrCrrrRrrlowerrtr) r)r]r^racolumnrf cluster_listrnamerrr*r*r+r`s2  zTable.as_markdowncCsP|jd|d|ddd}t|tr|d}|r|SdStdt|j)NzbSELECT comment FROM system_schema.tables WHERE keyspace_name = 'z%' AND table_name = '';r4)r0rzBUnexpected result type from db.run: )r9rsrgetr#rfr)rCrrrBrr*r*r+r3s$  zTable._resolve_comment>Tuple[List[Tuple[str, str]], List[str], List[Tuple[str, str]]]c Csg}g}g}|d|d|d}t|tstd|D]9}t|ts%q||d|df|ddkrB||d|d fq|dd krV||d|d |d fqd d t|dddD}dd t|dddD} ||| fS)NzSELECT column_name, type, kind, clustering_order, position FROM system_schema.columns WHERE keyspace_name = 'z0' AND table_name = 'r6Expected a sequence of dictionaries from 'run' method.rrfrrpositionrrcSsg|]\}}|qSr*r*)rOr_r*r*r+ryhsz*Table._resolve_columns..cS|dS)Nr*xr*r*r+iz(Table._resolve_columns..)keycSsg|] \}}}||fqSr*r*)rOrrrr*r*r+rylscSr)Nr*rr*r*r+ror)r9rsr TypeErrorrrtsorted) rCrrrpartition_info cluster_inforesultsrPrclusterr*r*r+rKs@     zTable._resolve_columnscCszg}|d|d|d}t|tstd|D]!}t|ts!q|d}t|ts.t|}||d|d|fq|S)NzSELECT index_name, kind, options FROM system_schema.indexes WHERE keyspace_name = 'z-' AND table_name = 'rrrrr)r9rsrrrr/rt)rCrrrrrP index_optionsr*r*r+rus$   zTable._resolve_indexes)r2r)rCr/rr/rrr2r)TN)r]rr^rr2r/)rCr/rr/rrr2r)rCr/rr/rrr2r)rCr/rr/rrr2r)rrr__annotations__rr r>rrrrr model_configrr classmethodrr`rrrrr*r*r*r+rs4     4  )r)r __future__rrltypingrrrrrrr r pydanticr r r rtyping_extensionsrcassandra.clusterrrr%r Exceptionrjrr*r*r*r+s (  2