Task 2.2: Use the COPY INTO command to incrementally load files from cloud object storage (ADLS/S3/GCS) into Unity Catalog–governed tables.

• AFiles loaded on an earlier day are skipped when it runs again

  Correct

  COPY INTO is retryable and idempotent: files in the source location that have already been loaded are skipped on later runs, so each daily run loads only the new files.

• BA file rewritten in place with corrected rows is loaded again

  Incorrect

  A file that was already loaded is skipped even if it has been modified since it was loaded, so the corrected rows are not picked up by a normal run.

• CWith COPY_OPTIONS ('force' = 'true'), it loads files already loaded

  Correct

  The force copy option disables idempotency, so files are loaded whether or not they were loaded before.

• DWith VALIDATE ALL, it writes the rows that parse and reports the rest

  Incorrect

  VALIDATE checks whether the data parses, matches the schema and meets the constraints, but the validated data is not written to the table.

• EFiles in another format under the prefix are read in their own format

  Incorrect

  Files in the source location must have the format named in FILEFORMAT; COPY INTO does not switch readers file by file.

COPY INTO remembers the files it has loaded from the source location and skips them on later runs, even when a file has changed since. To load files again, the force copy option turns that idempotency off. VALIDATE checks data without writing it, and every file is read with the format that FILEFORMAT names.

• AREAD FILES on the volume main.landing.raw

  Incorrect

  READ FILES is the privilege for paths in an external location; a Unity Catalog volume uses volume privileges.

• BWRITE VOLUME on the volume main.landing.raw

  Incorrect

  WRITE VOLUME allows adding or changing files in the volume; COPY INTO only needs to read them.

• CREAD VOLUME on the volume main.landing.raw

  Correct

  To load from a Unity Catalog volume, COPY INTO requires READ VOLUME on the volume, which applies to all directories under it.

• DCREATE EXTERNAL TABLE on the external location behind the volume

  Incorrect

  That privilege is for creating external tables in an external location; it does not grant reading a volume's files.

COPY INTO needs read access to its source: READ VOLUME for a volume, or READ FILES for an external location path, plus the usage privileges on the parents.

• AFORMAT_OPTIONS ('header' = 'true', 'sep' = ';')

  Correct

  FORMAT_OPTIONS passes reader options to the CSV data source: header uses the first line as column names, and sep sets the field separator.

• BCOPY_OPTIONS ('header' = 'true', 'sep' = ';')

  Incorrect

  COPY_OPTIONS controls the COPY INTO operation itself, with options such as force and mergeSchema; reader options for the file format belong in FORMAT_OPTIONS.

• CFORMAT_OPTIONS ('header' = 'true', 'lineSep' = ';')

  Incorrect

  lineSep defines the line separator, so each semicolon would end a record while the fields would still be split on the default comma.

• DFORMAT_OPTIONS ('header' = 'false', 'sep' = ';')

  Incorrect

  With header set to false, the first line is read as data rather than as column names, so the header row is loaded as a record.

COPY INTO hands the options in FORMAT_OPTIONS to the Apache Spark reader of the chosen format, so CSV options such as header and sep go there. The CSV separator defaults to a comma and header defaults to false, so a headered, semicolon-separated file needs both set.

• ACOPY INTO main.bronze.reviews FROM the source path, letting COPY INTO create the table

  Incorrect

  COPY INTO needs an existing Delta table as its target; it does not create the table.

• BCREATE TABLE main.bronze.reviews listing each column as STRING, then COPY INTO it

  Incorrect

  This means writing the column list by hand, which the engineer wants to avoid.

• CCREATE TABLE IF NOT EXISTS main.bronze.reviews, then INSERT INTO it from the JSON files

  Incorrect

  INSERT INTO and MERGE INTO are not supported for writing to a schemaless placeholder table.

• DCREATE TABLE IF NOT EXISTS, then COPY INTO it with mergeSchema in both option clauses

  Correct

  An empty placeholder Delta table gets its schema inferred during COPY INTO when mergeSchema is set in both options.

A schemaless placeholder table plus mergeSchema lets COPY INTO infer the schema on the first load. The table becomes queryable after COPY INTO writes data to it.

• ACOPY INTO t FROM 's3://landing/orders' FILEFORMAT = JSON FORMAT_OPTIONS ('inferSchema' = 'true')

  Incorrect

  Schema inference picks types from the data; it neither forces DECIMAL(10,2) nor adds the file path.

• BCOPY INTO t FROM (SELECT CAST(amount AS DECIMAL(10,2)) AS amount, _metadata.file_path AS source_file FROM 's3://landing/orders') FILEFORMAT = JSON

  Correct

  The SELECT form of the source lets you cast columns and add expressions such as the hidden _metadata.file_path column during the load.

• CCOPY INTO t FROM 's3://landing/orders' FILEFORMAT = JSON COPY_OPTIONS ('mergeSchema' = 'true')

  Incorrect

  mergeSchema concerns the table's columns, not the values; it neither casts amount nor records the file path.

• DCOPY INTO t FROM 's3://landing/orders' FILEFORMAT = JSON, followed by an UPDATE that casts amount

  Incorrect

  A second statement rewrites the data after loading, and the file path is no longer available to it.

COPY INTO accepts a SELECT over the source, so you can cast, rename and add columns, including file metadata from _metadata, as part of the load.