inspect.files() and the other file metadata tables raise ArrowInvalid when a snapshot has no manifests #4269

Apache Iceberg version

0.12.0 (latest release)

Please describe the bug 🐞

Appending an empty Arrow table creates a snapshot whose manifest list has no manifests. After that, inspect.files(), data_files(), delete_files(), all_files(), all_data_files(), and all_delete_files() all raise ArrowInvalid: Must pass at least one table. I expected each of them to return an empty table with the files schema.

import tempfile

import pyarrow as pa
from pyiceberg.catalog.sql import SqlCatalog

warehouse = tempfile.mkdtemp()
catalog = SqlCatalog("t", uri=f"sqlite:///{warehouse}/catalog.db", warehouse=f"file://{warehouse}")
catalog.create_namespace("n")
df = pa.table({"x": pa.array([], pa.int64())})
table = catalog.create_table("n.t", schema=df.schema)
table.append(df)
table.inspect.files()  # ArrowInvalid: Must pass at least one table

Traceback (most recent call last):
  File "repro.py", line 12, in <module>
    table.inspect.files()  # ArrowInvalid: Must pass at least one table
    ^^^^^^^^^^^^^^^^^^^^^
  File "pyiceberg/table/inspect.py", line 900, in files
    return self._files(snapshot_id)
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "pyiceberg/table/inspect.py", line 888, in _files
    return pa.concat_tables(results)
           ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "pyarrow/table.pxi", line 6344, in pyarrow.lib.concat_tables
  File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
  File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: Must pass at least one table

overwrite() with an empty table after a non-empty append also leaves a snapshot with no manifests, and files() fails the same way.

_files reads one table per manifest and ends with pa.concat_tables(results), so an empty manifest list calls pa.concat_tables([]). _all_files does the same at L981. Both already return an empty table one level up: _files when the table has no current snapshot (L876-L877), and _all_files when the table has no snapshots (L968-L970). Could they return the same empty table when there are no manifests?

This reproduces on 0.12.0 and on main at e3564a5.

Willingness to contribute

• I can contribute a fix for this bug independently
• I would be willing to contribute a fix for this bug with guidance from the Iceberg community
• I cannot contribute a fix for this bug at this time