Added add-files option to CLI.

Author: subkanthi

pyiceberg/cli/console.py

@@ -179,6 +179,59 @@ def files(ctx: Context, identifier: str, history: bool) -> None:
     output.files(catalog_table, history)
 
 
+@run.command("add-files")
+@click.argument("identifier")
+@click.argument("file_paths", nargs=-1)
+@click.option("--branch", default=None, help="Branch to add files to (default: main).")
+@click.option(
+    "--no-check-duplicates",
+    is_flag=True,
+    help="Skip check for files already referenced by the table.",
+)
+@click.option(
+    "--property",
+    "-p",
+    "properties",
+    multiple=True,
+    help="Snapshot property key=value (repeatable).",
+)
+@click.pass_context
+@catch_exception()
+def add_files(
+    ctx: Context,
+    identifier: str,
+    file_paths: tuple[str, ...],
+    branch: str | None,
+    no_check_duplicates: bool,
+    properties: tuple[str, ...],
+) -> None:
+    """Add one or more data files to the table by path."""
+    if not file_paths:
+        raise click.UsageError("At least one file path is required.")
+
+    catalog, output = _catalog_and_output(ctx)
+
+    snapshot_properties: dict[str, str] = {}
+    for prop in properties:
+        if "=" not in prop:
+            raise click.UsageError(f"Property must be in key=value form, got: {prop!r}")
+        key, _, value = prop.partition("=")
+        snapshot_properties[key] = value
+
+    file_paths_list = []
+    for item in file_paths:
+        file_paths_list.append(item)
+
+    table = catalog.load_table(identifier)
+    table.add_files(
+        file_paths=file_paths_list,
+        branch=branch or MAIN_BRANCH,
+        snapshot_properties=snapshot_properties,
+        check_duplicate_files=not no_check_duplicates,
+    )
+    output.text(f"Added {len(file_paths)} file(s) to {identifier}")
+
+
 @run.command("delete-files")
 @click.argument("identifier")
 @click.argument("file_paths", nargs=-1)

tests/cli/test_console.py

@@ -1031,6 +1031,105 @@ def test_log_level_cli_overrides_env(mocker: MockFixture) -> None:
     assert call_kwargs["level"] == logging.ERROR
 
 
+def test_add_files_invokes_table_api(catalog: InMemoryCatalog, mocker: MockFixture) -> None:
+    catalog.create_namespace(TEST_TABLE_NAMESPACE)
+    catalog.create_table(
+        identifier=TEST_TABLE_IDENTIFIER,
+        schema=TEST_TABLE_SCHEMA,
+        partition_spec=TEST_TABLE_PARTITION_SPEC,
+    )
+    mock_table = MagicMock()
+    mocker.patch.object(catalog, "load_table", return_value=mock_table)
+
+    runner = CliRunner()
+    result = runner.invoke(
+        run,
+        [
+            "add-files",
+            "default.my_table",
+            "s3://bucket/path/file.parquet",
+            "--branch",
+            "main",
+            "--property",
+            "k1=v1",
+            "-p",
+            "k2=v2",
+        ],
+    )
+
+    if result.exit_code != 0:
+        out = (result.output or "") + (getattr(result, "stderr", "") or "")
+        raise AssertionError(f"exit_code={result.exit_code} output/stderr: {out!r}")
+    mock_table.add_files.assert_called_once()
+    call_kwargs = mock_table.add_files.call_args[1]
+    assert call_kwargs["file_paths"] == ["s3://bucket/path/file.parquet"]
+    assert call_kwargs["branch"] == "main"
+    assert call_kwargs["snapshot_properties"] == {"k1": "v1", "k2": "v2"}
+    assert call_kwargs["check_duplicate_files"] is True
+
+
+def test_add_files_requires_at_least_one_path(catalog: InMemoryCatalog) -> None:
+    catalog.create_namespace(TEST_TABLE_NAMESPACE)
+    catalog.create_table(
+        identifier=TEST_TABLE_IDENTIFIER,
+        schema=TEST_TABLE_SCHEMA,
+        partition_spec=TEST_TABLE_PARTITION_SPEC,
+    )
+    runner = CliRunner()
+    result = runner.invoke(run, ["add-files", "default.my_table"])
+    assert result.exit_code != 0
+    out = (result.output or "") + (getattr(result, "stderr", "") or "")
+    assert "file path" in out.lower() or "required" in out.lower()
+
+
+def test_add_files_invalid_property_format(catalog: InMemoryCatalog, mocker: MockFixture) -> None:
+    catalog.create_namespace(TEST_TABLE_NAMESPACE)
+    catalog.create_table(
+        identifier=TEST_TABLE_IDENTIFIER,
+        schema=TEST_TABLE_SCHEMA,
+        partition_spec=TEST_TABLE_PARTITION_SPEC,
+    )
+    runner = CliRunner()
+    result = runner.invoke(
+        run,
+        ["add-files", "default.my_table", "s3://bucket/file.parquet", "--property", "invalid_no_equals"],
+    )
+    assert result.exit_code != 0
+    out = (result.output or "") + (getattr(result, "stderr", "") or "")
+    assert "key=value" in out or "invalid_no_equals" in out
+
+
+def test_add_files_table_does_not_exist(catalog: InMemoryCatalog) -> None:
+    catalog.create_namespace(TEST_TABLE_NAMESPACE)
+    runner = CliRunner()
+    result = runner.invoke(run, ["add-files", "default.doesnotexist", "s3://bucket/file.parquet"])
+    assert result.exit_code != 0
+    out = (result.output or "") + (getattr(result, "stderr", "") or "")
+    assert "default.doesnotexist" in out and ("Table does not exist" in out or "does not exist" in out)
+
+
+def test_add_files_no_check_duplicates_flag(catalog: InMemoryCatalog, mocker: MockFixture) -> None:
+    catalog.create_namespace(TEST_TABLE_NAMESPACE)
+    catalog.create_table(
+        identifier=TEST_TABLE_IDENTIFIER,
+        schema=TEST_TABLE_SCHEMA,
+        partition_spec=TEST_TABLE_PARTITION_SPEC,
+    )
+    mock_table = MagicMock()
+    mocker.patch.object(catalog, "load_table", return_value=mock_table)
+
+    runner = CliRunner()
+    result = runner.invoke(
+        run,
+        ["add-files", "default.my_table", "s3://bucket/file.parquet", "--no-check-duplicates"],
+    )
+
+    assert result.exit_code == 0
+    mock_table.add_files.assert_called_once()
+    call_kwargs = mock_table.add_files.call_args[1]
+    assert call_kwargs["check_duplicate_files"] is False
+
+
 def test_delete_files_invalid_property_format(catalog: InMemoryCatalog, mocker: MockFixture) -> None:
     catalog.create_namespace(TEST_TABLE_NAMESPACE)
     catalog.create_table(