This is an automated email from the ASF dual-hosted git repository.
zeroshade pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/iceberg-go.git
The following commit(s) were added to refs/heads/main by this push:
new 41593a308 refactor(table)!: name path-based data file additions
explicitly (#2308)
41593a308 is described below
commit 41593a308076eb33620da234aa9bb9d36a2c468a
Author: Minh Vu <[email protected]>
AuthorDate: Mon Oct 26 19:28:24 2026 +0100
refactor(table)!: name path-based data file additions explicitly (#2308)
* refactor(table)!: name path-based data file additions explicitly
Signed-off-by: Minh Vu <[email protected]>
* docs(release): note v1 API compatibility change
* chore: sync release notes with main to resolve conflicts
* docs(release): move compatibility note to a follow-up PR
The note for this change lands in a follow-up release-notes PR so the
pre-v1 renames can merge in parallel.
---------
Signed-off-by: Minh Vu <[email protected]>
Co-authored-by: Matt Topol <[email protected]>
---
catalog/rest/rest_integration_test.go | 6 +-
catalog/rest/scan_planning_integration_test.go | 2 +-
catalog/sql/sql_integration_test.go | 2 +-
catalog/sql/sql_test.go | 2 +-
table/addfiles_merge_regression_test.go | 20 ++---
table/arrow_utils.go | 2 +-
table/arrow_utils_test.go | 4 +-
table/branch_planning_test.go | 8 +-
table/compaction/analyze_test.go | 4 +-
table/compaction/cdc_stress_test.go | 4 +-
table/compaction/pos_delete_collect_ext_test.go | 4 +-
table/compaction_file_grain_test.go | 2 +-
table/data_file_meta.go | 2 +-
table/dv_rewrite_test.go | 2 +-
table/equality_delete_commit_test.go | 10 +--
table/equality_delete_reader_test.go | 16 ++--
table/equality_delete_writer_test.go | 2 +-
table/internal/parquet_files.go | 4 +-
.../maintenance/orphan_cleanup_integration_test.go | 2 +-
table/manifest_list_partitions_regression_test.go | 10 +--
table/mor_delete_conflict_test.go | 2 +-
table/no_replay_removals_test.go | 6 +-
table/occ_scenario_test.go | 8 +-
table/read_perf_deletes_test.go | 4 +-
table/replace_files_test.go | 14 ++--
table/rewrite_data_files_test.go | 36 ++++-----
table/rewrite_files_test.go | 20 ++---
table/rewrite_manifests_test.go | 6 +-
table/row_delta_test.go | 18 ++---
table/snapshots_datafiles_internal_test.go | 2 +-
table/table_test.go | 90 +++++++++++-----------
table/transaction.go | 16 ++--
table/transaction_empty_append_test.go | 10 +--
table/transaction_internal_test.go | 8 +-
table/transaction_test.go | 2 +-
table/variant_shredded_test.go | 16 ++--
table/write_file_io_test.go | 2 +-
website/src/api.md | 2 +-
38 files changed, 187 insertions(+), 183 deletions(-)
diff --git a/catalog/rest/rest_integration_test.go
b/catalog/rest/rest_integration_test.go
index 7c4dfffb5..ea9d66cbd 100644
--- a/catalog/rest/rest_integration_test.go
+++ b/catalog/rest/rest_integration_test.go
@@ -385,7 +385,7 @@ func (s *RestIntegrationSuite) TestWriteCommitTable() {
defer mustFS(s.T(), tbl).Remove(s.ctx, pqfile)
txn := tbl.NewTransaction()
- s.Require().NoError(txn.AddFiles(s.ctx, []string{pqfile}, nil, false))
+ s.Require().NoError(txn.AddDataFilesByPath(s.ctx, []string{pqfile},
nil, false))
updated, err := txn.Commit(s.ctx)
s.Require().NoError(err)
@@ -465,10 +465,10 @@ func (s *RestIntegrationSuite) TestMultiTableCommit() {
// Build transactions that add data files to each table.
tx1 := tbl1.NewTransaction()
- s.Require().NoError(tx1.AddFiles(s.ctx, []string{pq1}, nil, false))
+ s.Require().NoError(tx1.AddDataFilesByPath(s.ctx, []string{pq1}, nil,
false))
tx2 := tbl2.NewTransaction()
- s.Require().NoError(tx2.AddFiles(s.ctx, []string{pq2}, nil, false))
+ s.Require().NoError(tx2.AddDataFilesByPath(s.ctx, []string{pq2}, nil,
false))
// Commit both atomically via MultiTableTransaction.
mtx, err := catalog.NewMultiTableTransaction(s.cat)
diff --git a/catalog/rest/scan_planning_integration_test.go
b/catalog/rest/scan_planning_integration_test.go
index 8f0e7cc94..ae508031d 100644
--- a/catalog/rest/scan_planning_integration_test.go
+++ b/catalog/rest/scan_planning_integration_test.go
@@ -268,7 +268,7 @@ func (s *RestIntegrationSuite) createScanPlanningTable(cat
*rest.Catalog, ident
s.T().Cleanup(func() { s.Require().NoError(mustFS(s.T(),
tbl).Remove(s.ctx, dataPath)) })
txn := tbl.NewTransaction()
- s.Require().NoError(txn.AddFiles(s.ctx, []string{dataPath}, nil, false))
+ s.Require().NoError(txn.AddDataFilesByPath(s.ctx, []string{dataPath},
nil, false))
updated, err := txn.Commit(s.ctx)
s.Require().NoError(err)
s.Require().NotNil(updated.CurrentSnapshot())
diff --git a/catalog/sql/sql_integration_test.go
b/catalog/sql/sql_integration_test.go
index 78ece6998..802816465 100644
--- a/catalog/sql/sql_integration_test.go
+++ b/catalog/sql/sql_integration_test.go
@@ -297,7 +297,7 @@ func (s *SQLIntegrationSuite) TestWriteCommitTable() {
}(fs, pqfile)
txn := tbl.NewTransaction()
- s.Require().NoError(txn.AddFiles(s.ctx, []string{pqfile}, nil, false))
+ s.Require().NoError(txn.AddDataFilesByPath(s.ctx, []string{pqfile},
nil, false))
updated, err := txn.Commit(s.ctx)
s.Require().NoError(err)
diff --git a/catalog/sql/sql_test.go b/catalog/sql/sql_test.go
index 7e74d8b39..7a88b76b0 100644
--- a/catalog/sql/sql_test.go
+++ b/catalog/sql/sql_test.go
@@ -2760,7 +2760,7 @@ func (s *SqliteCatalogTestSuite) TestCommitTable() {
s.EqualValues(0, tbl.Metadata().CurrentSchema().ID)
tx := tbl.NewTransaction()
- s.Require().NoError(tx.AddFiles(ctx, []string{pqfile}, nil,
false))
+ s.Require().NoError(tx.AddDataFilesByPath(ctx,
[]string{pqfile}, nil, false))
updated, err := tx.Commit(ctx)
s.Require().NoError(err)
diff --git a/table/addfiles_merge_regression_test.go
b/table/addfiles_merge_regression_test.go
index cac2a2580..0e4ee3cf1 100644
--- a/table/addfiles_merge_regression_test.go
+++ b/table/addfiles_merge_regression_test.go
@@ -17,14 +17,14 @@
package table_test
-// Regression test for: AddFiles() ignoring commit.manifest-merge.enabled.
+// Regression test for: AddDataFilesByPath() ignoring
commit.manifest-merge.enabled.
//
-// Root cause: transaction.go hardcoded .fastAppend() in AddFiles(), bypassing
+// Root cause: transaction.go hardcoded .fastAppend() in AddDataFilesByPath(),
bypassing
// appendSnapshotProducer() which reads ManifestMergeEnabledKey.
//
// Java reference: BaseTable.newAppend() always uses MergeAppend (merge
default=true).
// AddDataFiles() and Append() in Go both correctly call
appendSnapshotProducer().
-// Only AddFiles() was missing the call.
+// Only AddDataFilesByPath() was missing the call.
//
// Fix: change line 780 in transaction.go from:
//
@@ -71,10 +71,10 @@ func (c *mergeCatalog) CommitTable(_ context.Context, _
table.Identifier, _ []ta
return meta, "", nil
}
-// TestAddFilesRespectsMergeEnabled verifies that AddFiles() respects the
+// TestAddFilesRespectsMergeEnabled verifies that AddDataFilesByPath()
respects the
// commit.manifest-merge.enabled table property.
//
-// Before fix: 3 sequential AddFiles() commits produce 3 manifests because
+// Before fix: 3 sequential AddDataFilesByPath() commits produce 3 manifests
because
// fastAppend() is hardcoded and the merge property is ignored.
//
// After fix: with minCountToMerge=2, the 3rd commit triggers a merge and
@@ -131,7 +131,7 @@ func TestAddFilesRespectsMergeEnabled(t *testing.T) {
cat,
)
- // Perform 3 separate AddFiles commits — simulating the committer's
1-per-batch pattern.
+ // Perform 3 separate AddDataFilesByPath commits — simulating the
committer's 1-per-batch pattern.
// minCountToMerge=2 means the 3rd commit should trigger a merge.
const numCommits = 3
for i := range numCommits {
@@ -139,8 +139,8 @@ func TestAddFilesRespectsMergeEnabled(t *testing.T) {
writeParquet(filePath)
txn := tbl.NewTransaction()
- require.NoError(t, txn.AddFiles(ctx, []string{filePath}, nil,
false),
- "AddFiles commit %d failed", i+1)
+ require.NoError(t, txn.AddDataFilesByPath(ctx,
[]string{filePath}, nil, false),
+ "AddDataFilesByPath commit %d failed", i+1)
tbl, err = txn.Commit(ctx)
require.NoError(t, err, "Commit %d failed", i+1)
@@ -155,8 +155,8 @@ func TestAddFilesRespectsMergeEnabled(t *testing.T) {
// After fix: merge fires on commit 3 (accumulated 3 > minCount 2) → 1
merged manifest.
// Before fix: 3 manifests (fastAppend carries all forward without
merging).
assert.Equal(t, 1, len(manifests),
- "expected 1 merged manifest after %d AddFiles commits with
manifest merge enabled "+
+ "expected 1 merged manifest after %d AddDataFilesByPath commits
with manifest merge enabled "+
"(minCountToMerge=2), got %d manifests — "+
- "AddFiles() is not respecting
commit.manifest-merge.enabled",
+ "AddDataFilesByPath() is not respecting
commit.manifest-merge.enabled",
numCommits, len(manifests))
}
diff --git a/table/arrow_utils.go b/table/arrow_utils.go
index 9173bfc15..0c4752667 100644
--- a/table/arrow_utils.go
+++ b/table/arrow_utils.go
@@ -1881,7 +1881,7 @@ func filesToDataFiles(ctx context.Context, fileIO
iceio.IO, meta *MetadataBuilde
}
}()
- // Pass 0 so AddFiles does not claim sort_order_id on
files
+ // Pass 0 so AddDataFilesByPath does not claim
sort_order_id on files
// it did not write (#1184). Callers that know the
file's
// sort layout use FileToDataFile with an explicit id.
dataFiles[i] = fileToDataFile(ctx, fileIO, filePath,
currentSchema, currentSpec, 0, meta.props)
diff --git a/table/arrow_utils_test.go b/table/arrow_utils_test.go
index 7d94abd1d..621d963ad 100644
--- a/table/arrow_utils_test.go
+++ b/table/arrow_utils_test.go
@@ -3497,7 +3497,7 @@ func TestGeoTypeParquetRoundTrip(t *testing.T) {
// stored ARROW:schema, as Java/Spark writers produce. geoarrow-go maps the
// Parquet GEOMETRY logical type to geoarrow.wkb on read, so the file projects
// onto a matching geometry column. A binary table column rejects it, both on
-// read and in the FileToDataFile (AddFiles) schema check: geo types are
+// read and in the FileToDataFile (AddDataFilesByPath) schema check: geo types
are
// deliberately not promotable to or from binary (see iceberg.PromoteType and
// https://github.com/apache/iceberg-go/issues/2193), and the binary subtests
// assert that rejection.
@@ -3603,7 +3603,7 @@ func TestGeoParquetReadWithoutStoredSchema(t *testing.T) {
assert.ErrorContains(t, err, "cannot promote
geometry(srid:4326) to binary")
})
- t.Run("AddFiles schema check rejects binary column", func(t *testing.T)
{
+ t.Run("AddDataFilesByPath schema check rejects binary column", func(t
*testing.T) {
binarySchema := iceberg.NewSchema(0, iceberg.NestedField{ID: 1,
Name: "geom", Type: iceberg.PrimitiveTypes.Binary, Required: false})
_, err := table.FileToDataFile(context.Background(),
iceio.LocalFS{}, path, binarySchema, *iceberg.UnpartitionedSpec, 0, nil)
require.Error(t, err)
diff --git a/table/branch_planning_test.go b/table/branch_planning_test.go
index 35951a96d..819c494d9 100644
--- a/table/branch_planning_test.go
+++ b/table/branch_planning_test.go
@@ -430,7 +430,7 @@ func TestBranchAddDataFilesDuplicateCheckUsesBranchHead(t
*testing.T) {
})
}
-// AddFiles runs the same check from paths, through its own snapshot lookup.
+// AddDataFilesByPath runs the same check from paths, through its own snapshot
lookup.
func TestBranchAddFilesDuplicateCheckUsesBranchHead(t *testing.T) {
t.Run("rejects a file the branch already references", func(t
*testing.T) {
tbl := seedDivergentBranch(t, newBranchPlanningTable(t, nil))
@@ -439,7 +439,7 @@ func TestBranchAddFilesDuplicateCheckUsesBranchHead(t
*testing.T) {
txn, err := tbl.NewTransactionOnBranch(branchPlanningRef)
require.NoError(t, err)
- err = txn.AddFiles(t.Context(), []string{branchOnly}, nil,
false)
+ err = txn.AddDataFilesByPath(t.Context(), []string{branchOnly},
nil, false)
require.ErrorContains(t, err, "cannot add files that are
already referenced by table")
assert.Contains(t, err.Error(), branchOnly)
})
@@ -451,7 +451,7 @@ func TestBranchAddFilesDuplicateCheckUsesBranchHead(t
*testing.T) {
txn, err := tbl.NewTransactionOnBranch(branchPlanningRef)
require.NoError(t, err)
- require.NoError(t, txn.AddFiles(t.Context(),
[]string{mainOnly}, nil, false))
+ require.NoError(t, txn.AddDataFilesByPath(t.Context(),
[]string{mainOnly}, nil, false))
tbl, err = txn.Commit(t.Context())
require.NoError(t, err)
@@ -602,7 +602,7 @@ func TestBranchRowDeltaResolvesRemovedDeletesOnBranchHead(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSchema,
`[{"id":1},{"id":2},{"id":3},{"id":4},{"id":5}]`)
txn := tbl.NewTransaction()
- require.NoError(t, txn.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, txn.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = txn.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/compaction/analyze_test.go b/table/compaction/analyze_test.go
index 2b51582c3..7d0b52e10 100644
--- a/table/compaction/analyze_test.go
+++ b/table/compaction/analyze_test.go
@@ -113,7 +113,7 @@ func TestAnalyze_SmallFiles(t *testing.T) {
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -147,7 +147,7 @@ func TestAnalyze_AllOptimal(t *testing.T) {
writeTestParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data":
"hello"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/compaction/cdc_stress_test.go
b/table/compaction/cdc_stress_test.go
index bec98d901..78012de95 100644
--- a/table/compaction/cdc_stress_test.go
+++ b/table/compaction/cdc_stress_test.go
@@ -83,7 +83,7 @@ func TestCDCStress(t *testing.T) {
seedPath := tbl.Location() + "/data/seed.parquet"
writeParquet(t, seedPath, arrowSc, generateRowsJSON(baselineRows))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{seedPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{seedPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -99,7 +99,7 @@ func TestCDCStress(t *testing.T) {
fmt.Sprintf(`[{"id": %d, "data": "cdc-%d"}]`, extraID,
cycle))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/compaction/pos_delete_collect_ext_test.go
b/table/compaction/pos_delete_collect_ext_test.go
index f757e99b4..eb83d9757 100644
--- a/table/compaction/pos_delete_collect_ext_test.go
+++ b/table/compaction/pos_delete_collect_ext_test.go
@@ -49,7 +49,7 @@ func TestCollectDeadPositionDeletes(t *testing.T) {
p := tbl.Location() + fmt.Sprintf("/data/d-%d.parquet", i)
writeParquet(t, p, arrowSc, fmt.Sprintf(`[{"id": %d, "data":
"r%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(ctx, []string{p}, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(ctx, []string{p}, nil,
false))
tbl, err = tx.Commit(ctx)
require.NoError(t, err)
}
@@ -160,7 +160,7 @@ func TestDeadPositionDeleteRoundTrip(t *testing.T) {
writeParquet(t, dataPaths[i], arrowSc, fmt.Sprintf(`[{"id": %d,
"data": "r%d"}]`, i+1, i+1))
}
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(ctx, dataPaths, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(ctx, dataPaths, nil, false))
tbl, err = tx.Commit(ctx)
require.NoError(t, err)
diff --git a/table/compaction_file_grain_test.go
b/table/compaction_file_grain_test.go
index 5631297d0..ad8b929a1 100644
--- a/table/compaction_file_grain_test.go
+++ b/table/compaction_file_grain_test.go
@@ -56,7 +56,7 @@ func TestCompactionPreservesSplitFileRows(t *testing.T) {
}
tx := tbl.NewTransaction()
require.NoError(t,
tx.SetProperties(iceberg.Properties{table.ReadSplitTargetSizeKey: splitSize}))
- require.NoError(t,
tx.AddFiles(t.Context(), []string{path}, nil, false))
+ require.NoError(t,
tx.AddDataFilesByPath(t.Context(), []string{path}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
scan := tbl.Scan()
diff --git a/table/data_file_meta.go b/table/data_file_meta.go
index aabc710f0..b04bca0af 100644
--- a/table/data_file_meta.go
+++ b/table/data_file_meta.go
@@ -293,7 +293,7 @@ func DataFileFromMetadata(args DataFileArgs)
(iceberg.DataFile, error) {
statsPlan,
pathMapping,
tblutils.VariantFieldIDsFromSchema(args.Schema),
- // Metadata-only path: no file is read here, so no
arrow schema; AddFiles recomputes variant bounds in fileToDataFile.
+ // Metadata-only path: no file is read here, so no
arrow schema; AddDataFilesByPath recomputes variant bounds in fileToDataFile.
nil,
)
if stats == nil {
diff --git a/table/dv_rewrite_test.go b/table/dv_rewrite_test.go
index 028f461b6..f18ae5d02 100644
--- a/table/dv_rewrite_test.go
+++ b/table/dv_rewrite_test.go
@@ -239,7 +239,7 @@ func
TestRewriteFiles_DataSequenceNumberKeepsDVApplicableToReplacement(t *testin
oldDataPath := tbl.Location() + "/data/old-data.parquet"
writeParquetFile(t, oldDataPath, arrowSc,
`[{"id":1,"data":"a"},{"id":2,"data":"b"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{oldDataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{oldDataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/equality_delete_commit_test.go
b/table/equality_delete_commit_test.go
index fbedd9338..76e06c52f 100644
--- a/table/equality_delete_commit_test.go
+++ b/table/equality_delete_commit_test.go
@@ -61,7 +61,7 @@ func TestCommitEqualityDeletesUnpartitioned(t *testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
assertRowCount(t, tbl, 5)
@@ -190,7 +190,7 @@ func TestCommitEqualityDeletesConcurrentAppendConflict(t
*testing.T) {
{"id": 3, "data": "c"}
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -205,7 +205,7 @@ func TestCommitEqualityDeletesConcurrentAppendConflict(t
*testing.T) {
peerPath := tbl.Location() + "/data/peer.parquet"
writeParquetFile(t, peerPath, arrowSc, `[{"id": 4, "data": "d"}]`)
peerTxn := tbl.NewTransaction()
- require.NoError(t, peerTxn.AddFiles(t.Context(), []string{peerPath},
nil, false))
+ require.NoError(t, peerTxn.AddDataFilesByPath(t.Context(),
[]string{peerPath}, nil, false))
_, err = peerTxn.Commit(t.Context())
require.NoError(t, err)
@@ -241,7 +241,7 @@ func
TestCommitEqualityDeletesSnapshotIsolationAllowsConcurrentAppend(t *testing
{"id": 2, "data": "b"}
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -253,7 +253,7 @@ func
TestCommitEqualityDeletesSnapshotIsolationAllowsConcurrentAppend(t *testing
peerPath := tbl.Location() + "/data/peer.parquet"
writeParquetFile(t, peerPath, arrowSc, `[{"id": 3, "data": "c"}]`)
peerTxn := tbl.NewTransaction()
- require.NoError(t, peerTxn.AddFiles(t.Context(), []string{peerPath},
nil, false))
+ require.NoError(t, peerTxn.AddDataFilesByPath(t.Context(),
[]string{peerPath}, nil, false))
_, err = peerTxn.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/equality_delete_reader_test.go
b/table/equality_delete_reader_test.go
index 412dde78d..e2a064e0a 100644
--- a/table/equality_delete_reader_test.go
+++ b/table/equality_delete_reader_test.go
@@ -97,7 +97,7 @@ func TestEqualityDeleteReadRoundTrip(t *testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
assertRowCount(t, tbl, 5)
@@ -197,7 +197,7 @@ func
TestEqualityDeleteReadPrunesNonOverlappingDeleteFiles(t *testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(ctx, dataPaths, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(ctx, dataPaths, nil, false))
tbl, err = tx.Commit(ctx)
require.NoError(t, err)
@@ -276,7 +276,7 @@ func
TestEqualityDeleteReadResolvesRenamedDataColumnByFieldID(t *testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -400,7 +400,7 @@ func
TestEqualityDeleteReadResolvesLiteralDottedColumnName(t *testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err := tx.Commit(t.Context())
require.NoError(t, err)
@@ -447,7 +447,7 @@ func TestEqualityDeleteReadRejectsAmbiguousColumns(t
*testing.T) {
dataPath := tbl.Location() + "/data/data.parquet"
writeParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data": "one"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -709,7 +709,7 @@ func TestEqualityDeleteMultiColumnKey(t *testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -837,7 +837,7 @@ func TestEqualityDeleteNullableFastPathKeys(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc,
tt.dataJSON)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(),
[]string{dataPath}, nil, false))
+ require.NoError(t,
tx.AddDataFilesByPath(t.Context(), []string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -928,7 +928,7 @@ func TestEqualityDeleteNullableCompositeFastPathKey(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSc, tt.dataJSON)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(),
[]string{dataPath}, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/equality_delete_writer_test.go
b/table/equality_delete_writer_test.go
index c9a3df6ce..436cfcb1a 100644
--- a/table/equality_delete_writer_test.go
+++ b/table/equality_delete_writer_test.go
@@ -522,7 +522,7 @@ func TestWriteEqualityDeleteFilesCommitViaRowDelta(t
*testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
assertRowCount(t, tbl, 3)
diff --git a/table/internal/parquet_files.go b/table/internal/parquet_files.go
index 648874036..9b102b225 100644
--- a/table/internal/parquet_files.go
+++ b/table/internal/parquet_files.go
@@ -537,8 +537,8 @@ type geoColumn struct {
// Parquet drops the Statistics block (min, max, null count) for GEOMETRY and
// GEOGRAPHY columns because their sort order is undefined, so these metrics
are
// computed from the Arrow data instead. Geo files imported via
-// DataFileStatsFromMeta alone (e.g. AddFiles) get no geo null counts or
bounds,
-// which costs pruning but never gives a wrong answer.
+// DataFileStatsFromMeta alone (e.g. AddDataFilesByPath) get no geo null counts
+// or bounds, which costs pruning but never gives a wrong answer.
func collectGeoColumns(sc *arrow.Schema, colMapping map[string]int)
[]geoColumn {
var result []geoColumn
for i, f := range sc.Fields() {
diff --git a/table/maintenance/orphan_cleanup_integration_test.go
b/table/maintenance/orphan_cleanup_integration_test.go
index 2c097e654..0004f43ac 100644
--- a/table/maintenance/orphan_cleanup_integration_test.go
+++ b/table/maintenance/orphan_cleanup_integration_test.go
@@ -157,7 +157,7 @@ func (s *OrphanCleanupIntegrationSuite)
createTableWithData() *table.Table {
// Commit the data file to the table
txn := tbl.NewTransaction()
- s.Require().NoError(txn.AddFiles(s.ctx, []string{dataFile}, nil, false))
+ s.Require().NoError(txn.AddDataFilesByPath(s.ctx, []string{dataFile},
nil, false))
updatedTable, err := txn.Commit(s.ctx)
s.Require().NoError(err)
diff --git a/table/manifest_list_partitions_regression_test.go
b/table/manifest_list_partitions_regression_test.go
index 56c3afa4a..90ad2b1fb 100644
--- a/table/manifest_list_partitions_regression_test.go
+++ b/table/manifest_list_partitions_regression_test.go
@@ -17,13 +17,13 @@
package table_test
-// End-to-end regression test for issue #1309: AddFiles() after a prior
Delete()
+// End-to-end regression test for issue #1309: AddDataFilesByPath() after a
prior Delete()
// produces a manifest list whose `partitions` field serializes as Avro null,
// which Redshift Spectrum rejects ("Wrong type in Avro file. Field:
partitions.
// Expected: 12 (union). Got: 7 (null)").
//
// The trigger is the inheritance path: Delete() uses the mergeOverwrite
producer
-// and the following AddFiles() uses fastAppend, which inherits the delete
+// and the following AddDataFilesByPath() uses fastAppend, which inherits the
delete
// snapshot's manifests unchanged. When the manifest list is rewritten, an
// inherited manifest whose `partitions` summaries are present-but-empty (an
// unpartitioned table has zero field summaries) round-trips as a non-nil
pointer
@@ -149,13 +149,13 @@ func
TestManifestListPartitionsPresentAfterDeleteThenAddFiles(t *testing.T) {
tbl, err = tbl.Delete(ctx, iceberg.EqualTo(iceberg.Reference("id"),
int64(2)), nil)
require.NoError(t, err)
- // Step 3 — AddFiles (fastAppend) inherits the delete snapshot's
manifests.
+ // Step 3 — AddDataFilesByPath (fastAppend) inherits the delete
snapshot's manifests.
// This is the commit whose manifest list #1309 reports as unreadable.
extraPath := location + "/data-extra.parquet"
writeInt64Parquet(t, fs, extraPath, arrowSchema, []int64{4, 5})
txn := tbl.NewTransaction()
- require.NoError(t, txn.AddFiles(ctx, []string{extraPath}, nil, false))
+ require.NoError(t, txn.AddDataFilesByPath(ctx, []string{extraPath},
nil, false))
tbl, err = txn.Commit(ctx)
require.NoError(t, err)
@@ -187,7 +187,7 @@ func
TestManifestListPartitionsPresentAfterDeleteThenAddFiles(t *testing.T) {
"iceberg-go must read the appended rows back with their real
values")
}
-// writeInt64Parquet writes a single-column ("id") Parquet file so AddFiles
has a
+// writeInt64Parquet writes a single-column ("id") Parquet file so
AddDataFilesByPath has a
// real file to ingest.
func writeInt64Parquet(t *testing.T, fs iceio.LocalFS, path string,
arrowSchema *arrow.Schema, ids []int64) {
t.Helper()
diff --git a/table/mor_delete_conflict_test.go
b/table/mor_delete_conflict_test.go
index 84778c39b..17adaa519 100644
--- a/table/mor_delete_conflict_test.go
+++ b/table/mor_delete_conflict_test.go
@@ -140,7 +140,7 @@ func
TestMergeOnReadDeleteOfDataFileAddedInSameTransaction(t *testing.T) {
writeRowDeltaDataFile(t, tbl, path, 11, 12)
txn := tbl.NewTransaction()
- require.NoError(t, txn.AddFiles(ctx, []string{path}, nil, false))
+ require.NoError(t, txn.AddDataFilesByPath(ctx, []string{path}, nil,
false))
require.NoError(t, txn.Delete(ctx,
iceberg.EqualTo(iceberg.Reference("id"), int64(11)), nil))
tbl, err := txn.Commit(ctx)
require.NoError(t, err)
diff --git a/table/no_replay_removals_test.go b/table/no_replay_removals_test.go
index 34e3b181b..36dbf2fd6 100644
--- a/table/no_replay_removals_test.go
+++ b/table/no_replay_removals_test.go
@@ -153,7 +153,7 @@ func TestReplaceFilesWithRemovalsFailsInsteadOfReplaying(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSc,
`[{"id": 1, "data": "alpha"}, {"id": 2, "data": "beta"}, {"id":
3, "data": "gamma"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -250,7 +250,7 @@ func TestReplaceFilesDataOnlyStillReplays(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc,
`[{"id": 1, "data": "alpha"}, {"id": 2, "data": "beta"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
// Overwrites validate added data files under the default
// serializable isolation; relax to snapshot so the peer's append is
// not rejected as a semantic conflict and the replay itself is
@@ -285,7 +285,7 @@ func TestReplaceFilesDataOnlyStillReplays(t *testing.T) {
peerPath := tbl.Location() + "/data/data-peer.parquet"
writeParquetFile(t, peerPath, arrowSc, `[{"id": 100, "data": "peer"}]`)
peerTxn := tbl.NewTransaction()
- require.NoError(t, peerTxn.AddFiles(t.Context(), []string{peerPath},
nil, false))
+ require.NoError(t, peerTxn.AddDataFilesByPath(t.Context(),
[]string{peerPath}, nil, false))
_, err = peerTxn.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/occ_scenario_test.go b/table/occ_scenario_test.go
index 452924a33..ff3b68a28 100644
--- a/table/occ_scenario_test.go
+++ b/table/occ_scenario_test.go
@@ -497,7 +497,7 @@ func (s *OCCScenarioTestSuite)
TestReplaceFilesNoResurrectionAfterConflict() {
`[{"id":1,"val":"a"},{"id":2,"val":"b"},{"id":3,"val":"c"}]`)
tx := baseTbl.NewTransaction()
- s.Require().NoError(tx.AddFiles(s.ctx, []string{file0Path}, nil, false))
+ s.Require().NoError(tx.AddDataFilesByPath(s.ctx, []string{file0Path},
nil, false))
_, err = tx.Commit(s.ctx)
s.Require().NoError(err)
metaBase := cat.current
@@ -511,7 +511,7 @@ func (s *OCCScenarioTestSuite)
TestReplaceFilesNoResurrectionAfterConflict() {
writeParquetFile(s.T(), peerPath, arrowSc, `[{"id":4,"val":"d"}]`)
ptx := peerTbl.NewTransaction()
- s.Require().NoError(ptx.AddFiles(s.ctx, []string{peerPath}, nil, false))
+ s.Require().NoError(ptx.AddDataFilesByPath(s.ctx, []string{peerPath},
nil, false))
_, err = ptx.Commit(s.ctx)
s.Require().NoError(err)
metaPeer := peerCat.current
@@ -589,7 +589,7 @@ func (s *OCCScenarioTestSuite)
TestReplaceFilesNoResurrectionAfterTwoConflicts()
`[{"id":1,"val":"a"},{"id":2,"val":"b"},{"id":3,"val":"c"}]`)
tx := baseTbl.NewTransaction()
- s.Require().NoError(tx.AddFiles(s.ctx, []string{file0Path}, nil, false))
+ s.Require().NoError(tx.AddDataFilesByPath(s.ctx, []string{file0Path},
nil, false))
_, err = tx.Commit(s.ctx)
s.Require().NoError(err)
metaBase := cat.current
@@ -607,7 +607,7 @@ func (s *OCCScenarioTestSuite)
TestReplaceFilesNoResurrectionAfterTwoConflicts()
peer++
ptx := peerTbl.NewTransaction()
- s.Require().NoError(ptx.AddFiles(s.ctx, []string{peerPath},
nil, false))
+ s.Require().NoError(ptx.AddDataFilesByPath(s.ctx,
[]string{peerPath}, nil, false))
_, err := ptx.Commit(s.ctx)
s.Require().NoError(err)
diff --git a/table/read_perf_deletes_test.go b/table/read_perf_deletes_test.go
index 350edeed1..f586e8220 100644
--- a/table/read_perf_deletes_test.go
+++ b/table/read_perf_deletes_test.go
@@ -61,7 +61,7 @@ func TestReadPerfUnderDeletePressure(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc, generateRowsJSON(numRows))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -210,7 +210,7 @@ func setupTableWithDeletes(tb testing.TB, numRows,
numDeletes int) *table.Table
ctx := context.Background()
tx := tbl.NewTransaction()
- require.NoError(tb, tx.AddFiles(ctx, []string{dataPath}, nil, false))
+ require.NoError(tb, tx.AddDataFilesByPath(ctx, []string{dataPath}, nil,
false))
tbl, err = tx.Commit(ctx)
require.NoError(tb, err)
diff --git a/table/replace_files_test.go b/table/replace_files_test.go
index a8e2e62fc..172de05ad 100644
--- a/table/replace_files_test.go
+++ b/table/replace_files_test.go
@@ -123,7 +123,7 @@ func TestReplaceFiles_DataAndDeleteFiles(t *testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
assertRowCount(t, tbl, 3)
@@ -164,7 +164,7 @@ func TestReplaceFiles_DataAndDeleteFiles(t *testing.T) {
{"id": 3, "data": "gamma"}
]`)
- // Build new DataFile directly (not via AddFiles which would commit it)
+ // Build new DataFile directly (not via AddDataFilesByPath which would
commit it)
newDataFileBuilder, err := iceberg.NewDataFileBuilder(
*iceberg.UnpartitionedSpec, iceberg.EntryContentData,
newDataPath, iceberg.ParquetFile, nil, nil, nil, 2, 512)
@@ -200,7 +200,7 @@ func
TestReplaceFilesWithDeleteFilesPreservesDataSequenceNumber(t *testing.T) {
dataPath := tbl.Location() + "/data/data.parquet"
writeParquetFile(t, dataPath, arrowSc,
`[{"id":1,"data":"a"},{"id":2,"data":"b"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -497,7 +497,7 @@ func
TestReplaceFilesWithDeleteFilesRejectsPartialPositionDeleteToDVRewrite(t *t
dataPath := tbl.Location() + "/data/data.parquet"
writeParquetFile(t, dataPath, arrowSc,
`[{"id":1,"data":"a"},{"id":2,"data":"b"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -802,7 +802,7 @@ func
TestReplaceFiles_DelegatesToReplaceDataFilesWhenNoDeleteFiles(t *testing.T)
writeParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data": "hello"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -842,7 +842,7 @@ func TestReplaceFiles_ValidationErrors(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data": "hello"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -968,7 +968,7 @@ func seedDataFilesWithPositionDeletes(t *testing.T)
(*table.Table, []table.FileS
}
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), dataPaths, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(), dataPaths, nil,
false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/rewrite_data_files_test.go b/table/rewrite_data_files_test.go
index 242fe6436..0ba78aa92 100644
--- a/table/rewrite_data_files_test.go
+++ b/table/rewrite_data_files_test.go
@@ -195,7 +195,7 @@ func TestRewriteDataFiles_SmallFiles(t *testing.T) {
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -241,7 +241,7 @@ func TestRewriteDataFiles_WithPositionDeletes(t *testing.T)
{
`[{"id": %d, "data": "a"}, {"id": %d, "data": "b"}]`,
i*2+1, i*2+2))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -327,7 +327,7 @@ func TestExecuteCompactionGroup_TargetFileSizeForwarded(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -388,7 +388,7 @@ func
TestExecuteCompactionGroup_ParquetRowGroupLimitForwarded(t *testing.T) {
dataPath := tbl.Location() + "/data/rg-limit.parquet"
writeParquetFile(t, dataPath, arrowSc, rows.String())
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -443,7 +443,7 @@ func TestExecuteCompactionGroup_ScanConcurrencyForwarded(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -484,7 +484,7 @@ func TestRewriteDataFiles_GroupOptionsForwarded(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -549,7 +549,7 @@ func TestRewriteDataFiles_PartialProgress(t *testing.T) {
fmt.Sprintf(`[{"id": %d, "data": "row"}]`, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -594,7 +594,7 @@ func
TestRewriteDataFiles_PartialProgressMaxCommitsProcessesAllGroups(t *testing
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row"}]`, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -634,7 +634,7 @@ func
TestRewriteDataFiles_PartialProgressRejectsCommittedParentReuse(t *testing.
dataPath := tbl.Location() + "/data/reuse.parquet"
writeParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data": "row"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -679,7 +679,7 @@ func
TestRewriteDataFiles_PartialProgressStopsOnUnknownCommitState(t *testing.T)
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row"}, {"id": %d,
"data": "row"}]`, i*2+1, i*2+2))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -718,7 +718,7 @@ func
TestRewriteDataFiles_PartialProgressRetainsSharedPositionDeleteAfterFailure
`[{"id": %d, "data": "a"}, {"id": %d, "data": "b"}]`,
i*2+1, i*2+2))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -802,7 +802,7 @@ func
TestRewriteDataFiles_PartialProgressCleansOutputsBeforeCommit(t *testing.T)
`[{"id": %d, "data": "row"}]`, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -985,7 +985,7 @@ func TestRewriteDataFiles_ContextCancellation(t *testing.T)
{
fmt.Sprintf(`[{"id": %d, "data": "row"}]`, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -1023,7 +1023,7 @@ func TestRewriteDataFiles_DeadEqualityDeletesDropped(t
*testing.T) {
`[{"id": %d, "data": "a"}, {"id": %d, "data": "b"}]`,
i*2+1, i*2+2))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -1108,7 +1108,7 @@ func
TestRewriteDataFiles_PartialRewritePreservesEqDelete(t *testing.T) {
`[{"id": %d, "data": "a"}]`, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -1120,7 +1120,7 @@ func
TestRewriteDataFiles_PartialRewritePreservesEqDelete(t *testing.T) {
preserveDataPath := tbl.Location() + "/data/preserve.parquet"
writeParquetFile(t, preserveDataPath, arrowSc, `[{"id": 50, "data":
"preserved"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{preserveDataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{preserveDataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -1193,7 +1193,7 @@ func
TestRewriteDataFiles_PreserveDeadEqualityDeletesOptOut(t *testing.T) {
`[{"id": %d, "data": "a"}]`, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -1299,7 +1299,7 @@ func
TestRewriteDataFiles_PartialProgressPreservesEqDeletes(t *testing.T) {
`[{"id": %d, "data": "a"}]`, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
diff --git a/table/rewrite_files_test.go b/table/rewrite_files_test.go
index 5b6d21d5d..fb4189f4c 100644
--- a/table/rewrite_files_test.go
+++ b/table/rewrite_files_test.go
@@ -268,7 +268,7 @@ func TestRewriteFiles_Commit_SingleShot(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "seed"}]`, i+1))
seedTx := tbl.NewTransaction()
- require.NoError(t, seedTx.AddFiles(t.Context(),
[]string{dataPath}, nil, false))
+ require.NoError(t, seedTx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = seedTx.Commit(t.Context())
require.NoError(t, err)
}
@@ -311,7 +311,7 @@ func
TestRewriteFiles_PreservesIndependentPositionDeleteSequences(t *testing.T)
writeParquetFile(t, dataPath, arrowSc,
`[{"id": 1, "data": "a"}, {"id": 2, "data": "b"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -427,7 +427,7 @@ func TestRewriteFiles_DeleteFile_RoutesByContentType(t
*testing.T) {
dataPath := tbl.Location() + "/data/seed.parquet"
writeParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data": "seed"}]`)
seedTx := tbl.NewTransaction()
- require.NoError(t, seedTx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, seedTx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = seedTx.Commit(t.Context())
require.NoError(t, err)
@@ -462,7 +462,7 @@ func TestRewriteFiles_DeleteFile_RoutesByContentType(t
*testing.T) {
// tombstone handling: a rewrite that removes a data file leaves a
// DELETED manifest entry for its path on the branch head. That entry
// records a removal — the path is no longer referenced by the table —
-// so a subsequent AddFiles for the same path must not report a false
+// so a subsequent AddDataFilesByPath for the same path must not report a false
// "already referenced" conflict.
func TestAddFiles_ReAddsPathDeletedByRewrite(t *testing.T) {
tbl := newRewriteTestTable(t)
@@ -472,7 +472,7 @@ func TestAddFiles_ReAddsPathDeletedByRewrite(t *testing.T) {
dataPath := tbl.Location() + "/data/seed.parquet"
writeParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data": "seed"}]`)
seedTx := tbl.NewTransaction()
- require.NoError(t, seedTx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, seedTx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = seedTx.Commit(t.Context())
require.NoError(t, err)
@@ -488,7 +488,7 @@ func TestAddFiles_ReAddsPathDeletedByRewrite(t *testing.T) {
require.NoError(t, err)
readdTx := tbl.NewTransaction()
- require.NoError(t, readdTx.AddFiles(t.Context(), []string{dataPath},
nil, false),
+ require.NoError(t, readdTx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false),
"a path whose only head entry is a DELETED tombstone is not
referenced by the table")
tbl, err = readdTx.Commit(t.Context())
require.NoError(t, err)
@@ -557,7 +557,7 @@ func TestRewriteFiles_DistributedEquivalence(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -631,7 +631,7 @@ func TestRewriteFiles_DropsSafePositionDeletes(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSc, fmt.Sprintf(
`[{"id": %d, "data": "a"}, {"id": %d, "data": "b"}]`,
i*2+1, i*2+2))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -735,7 +735,7 @@ func TestRewriteFiles_RejectsConcurrentEqDelete(t
*testing.T) {
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
@@ -865,7 +865,7 @@ func
TestRewriteFiles_RejectsConcurrentPartitionScopedPosDelete(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc,
fmt.Sprintf(`[{"id": %d, "data": "row-%d"}]`, i+1, i+1))
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
}
diff --git a/table/rewrite_manifests_test.go b/table/rewrite_manifests_test.go
index 963830206..398e73c9c 100644
--- a/table/rewrite_manifests_test.go
+++ b/table/rewrite_manifests_test.go
@@ -102,7 +102,7 @@ func appendSeparateManifests(t *testing.T, ctx
context.Context, tbl *table.Table
writeOneRowParquet(t, fs, filePath)
txn := tbl.NewTransaction()
- require.NoError(t, txn.AddFiles(ctx, []string{filePath}, nil,
false))
+ require.NoError(t, txn.AddDataFilesByPath(ctx,
[]string{filePath}, nil, false))
tbl, err = txn.Commit(ctx)
require.NoError(t, err)
}
@@ -664,7 +664,7 @@ func TestRewriteManifestsSpecIDFilter(t *testing.T) {
specOneFile := dir + "/spec1-0.parquet"
writeOneRowParquet(t, fs, specOneFile)
txn = tbl.NewTransaction()
- require.NoError(t, txn.AddFiles(ctx,
[]string{specOneFile}, nil, false))
+ require.NoError(t, txn.AddDataFilesByPath(ctx,
[]string{specOneFile}, nil, false))
tbl, err = txn.Commit(ctx)
require.NoError(t, err)
@@ -908,7 +908,7 @@ func appendOne(t *testing.T, ctx context.Context, tbl
*table.Table, fs iceio.Wri
writeOneRowParquet(t, fs, path)
txn := tbl.NewTransaction()
- require.NoError(t, txn.AddFiles(ctx, []string{path}, nil, false))
+ require.NoError(t, txn.AddDataFilesByPath(ctx, []string{path}, nil,
false))
out, err := txn.Commit(ctx)
require.NoError(t, err)
diff --git a/table/row_delta_test.go b/table/row_delta_test.go
index 86d025758..4ea780580 100644
--- a/table/row_delta_test.go
+++ b/table/row_delta_test.go
@@ -573,7 +573,7 @@ func TestRowDeltaIntegrationPosDeleteRoundTrip(t
*testing.T) {
]`)
tx := tbl.NewTransaction()
- err = tx.AddFiles(t.Context(), []string{dataPath}, nil, false)
+ err = tx.AddDataFilesByPath(t.Context(), []string{dataPath}, nil, false)
require.NoError(t, err)
tbl, err = tx.Commit(t.Context())
@@ -747,7 +747,7 @@ func newTableWithLiveDV(t *testing.T) (*table.Table,
string, string, iceberg.Dat
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -780,7 +780,7 @@ func newTableWithSharedPuffinDVs(t *testing.T)
(*table.Table, string, string, st
writeParquetFile(t, pathB, arrowSc, `[{"id": 3, "data": "gamma"},
{"id": 4, "data": "delta"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{pathA, pathB},
nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(), []string{pathA,
pathB}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -859,7 +859,7 @@ func TestRowDeltaDVSupersessionSingleSnapshot(t *testing.T)
{
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
assertRowCount(t, tbl, 5)
@@ -1359,7 +1359,7 @@ func TestRowDeltaRemoveDeletesStaleReference(t
*testing.T) {
writeParquetFile(t, pathB, arrowSc, `[{"id": 3, "data":
"gamma"}, {"id": 4, "data": "delta"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{pathA,
pathB}, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{pathA, pathB}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -1551,7 +1551,7 @@ func TestRowDeltaRemoveDeletesUnknownFile(t *testing.T) {
writeParquetFile(t, realDataPath, arrowSc, `[{"id": 1, "data":
"alpha"}]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(),
[]string{realDataPath}, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{realDataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -1771,7 +1771,7 @@ func TestRowDeltaFastAppendRejectsDuplicateAddedDVs(t
*testing.T) {
]`)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -1827,7 +1827,7 @@ func TestRowDeltaRejectsConcurrentFirstDVOnRetry(t
*testing.T) {
]`)
tx := baseTbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
_, err = tx.Commit(t.Context())
require.NoError(t, err)
metaBase := baseCat.current
@@ -1903,7 +1903,7 @@ func seedRowDeltaDataFile(t *testing.T, tbl *table.Table,
path string, ids ...in
writeRowDeltaDataFile(t, tbl, path, ids...)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{path}, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(), []string{path},
nil, false))
tbl, err := tx.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/snapshots_datafiles_internal_test.go
b/table/snapshots_datafiles_internal_test.go
index fdd8a8432..1dd7c6103 100644
--- a/table/snapshots_datafiles_internal_test.go
+++ b/table/snapshots_datafiles_internal_test.go
@@ -93,7 +93,7 @@ func writeSnapshotTestManifestList(t *testing.T, fs
iceio.WriteFileIO, dir strin
// Why: a manifest entry with status DELETED is a tombstone recording a
// removal, not a file reachable from the snapshot. Yielding it would make
// existence and duplicate checks (ReplaceFiles' belong-to-table check,
-// AddFiles' duplicate walk, classifyFilesForDeletions) treat a file
+// AddDataFilesByPath' duplicate walk, classifyFilesForDeletions) treat a file
// deleted by the snapshot as still live. ADDED and EXISTING entries are
// both live and must keep flowing through.
// Condition: a snapshot whose data and delete manifests each carry one
diff --git a/table/table_test.go b/table/table_test.go
index 1467d2d97..be0a8a2ce 100644
--- a/table/table_test.go
+++ b/table/table_test.go
@@ -556,7 +556,7 @@ func (t *TableWritingTestSuite) TestAddFilesUnpartitioned()
{
}
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, files, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files, nil, false))
stagedTbl, err := tx.StagedTable()
t.Require().NoError(err)
@@ -605,7 +605,7 @@ func (t *TableWritingTestSuite) TestAddFilesFileNotFound() {
files = append(files,
t.location+"/unpartitioned_file_not_found/unknown.parquet")
tx := tbl.NewTransaction()
- err := tx.AddFiles(t.ctx, files, nil, false)
+ err := tx.AddDataFilesByPath(t.ctx, files, nil, false)
t.Error(err)
t.ErrorIs(err, fs.ErrNotExist)
}
@@ -624,7 +624,7 @@ func (t *TableWritingTestSuite)
TestAddFilesUnpartitionedHasFieldIDs() {
}
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, files, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files, nil, false))
stagedTbl, err := tx.StagedTable()
t.Require().NoError(err)
@@ -679,7 +679,7 @@ func (t *TableWritingTestSuite)
TestAddFilesFieldIDsWithDifferentNames() {
}
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, files, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files, nil, false))
stagedTbl, err := tx.StagedTable()
t.Require().NoError(err)
@@ -735,7 +735,7 @@ func (t *TableWritingTestSuite)
TestAddFilesFailsSchemaMismatch() {
files := []string{filePath}
tx := tbl.NewTransaction()
- err = tx.AddFiles(t.ctx, files, nil, false)
+ err = tx.AddDataFilesByPath(t.ctx, files, nil, false)
t.Error(err)
t.EqualError(err, `error encountered during file conversion: invalid
schema: mismatch in fields:
Table Field Requested Field
@@ -775,7 +775,7 @@ func (t *TableWritingTestSuite)
TestAddFilesPartitionedTable() {
}
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, files, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files, nil, false))
stagedTbl, err := tx.StagedTable()
t.Require().NoError(err)
@@ -832,7 +832,7 @@ func (t *TableWritingTestSuite)
TestAddFilesToBucketPartitionedTableFails() {
}
tx := tbl.NewTransaction()
- err := tx.AddFiles(t.ctx, files, nil, false)
+ err := tx.AddDataFilesByPath(t.ctx, files, nil, false)
t.Error(err)
t.ErrorContains(err, "cannot infer partition value from parquet
metadata for a non-linear partition field: baz_bucket_3 with transform
bucket[3]")
}
@@ -860,7 +860,7 @@ func (t *TableWritingTestSuite)
TestAddFilesToPartitionedTableFailsLowerAndUpper
}
tx := tbl.NewTransaction()
- err := tx.AddFiles(t.ctx, files, nil, false)
+ err := tx.AddDataFilesByPath(t.ctx, files, nil, false)
t.Error(err)
t.ErrorContains(err, "cannot infer partition value from parquet
metadata as there is more than one value for partition field: baz. (low: 123,
high: 124)")
}
@@ -898,7 +898,7 @@ func (t *TableWritingTestSuite)
TestAddFilesWithLargeAndRegular() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), filePathLarge,
arrTableLarge)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{filePath,
filePathLarge}, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, []string{filePath,
filePathLarge}, nil, false))
scan, err := tx.Scan(table.WithOptions(iceberg.Properties{
table.ScanOptionArrowUseLargeTypes: "true",
@@ -921,7 +921,7 @@ func (t *TableWritingTestSuite)
TestScanWithOptionsAliasing() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), filePath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{filePath}, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, []string{filePath},
nil, false))
opts := iceberg.Properties{
table.ScanOptionArrowUseLargeTypes: "false",
@@ -946,7 +946,7 @@ func (t *TableWritingTestSuite) TestAddFilesValidUpcast() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), filePath,
t.arrTablePromotedTypes)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{filePath}, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, []string{filePath},
nil, false))
scan, err := tx.Scan()
t.Require().NoError(err)
@@ -994,7 +994,7 @@ func (t *TableWritingTestSuite)
TestAddFilesSubsetOfSchema() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), filePath,
withoutCol)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{filePath}, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, []string{filePath},
nil, false))
scan, err := tx.Scan()
t.Require().NoError(err)
@@ -1029,9 +1029,9 @@ func (t *TableWritingTestSuite)
TestAddFilesDuplicateFilesInFilePaths() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), filePath,
t.arrTbl)
tx := tbl.NewTransaction()
- err := tx.AddFiles(t.ctx, []string{filePath, filePath}, nil, false)
+ err := tx.AddDataFilesByPath(t.ctx, []string{filePath, filePath}, nil,
false)
t.Error(err)
- t.ErrorContains(err, "file paths must be unique for AddFiles")
+ t.ErrorContains(err, "file paths must be unique for AddDataFilesByPath")
}
func (t *TableWritingTestSuite) TestAddFilesReferencedByCurrentSnapshot() {
@@ -1049,9 +1049,9 @@ func (t *TableWritingTestSuite)
TestAddFilesReferencedByCurrentSnapshot() {
}
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, files, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files, nil, false))
- err := tx.AddFiles(t.ctx, files[len(files)-1:], nil, false)
+ err := tx.AddDataFilesByPath(t.ctx, files[len(files)-1:], nil, false)
t.Error(err)
t.ErrorContains(err, "cannot add files that are already referenced by
table, files:")
}
@@ -1071,9 +1071,9 @@ func (t *TableWritingTestSuite)
TestAddFilesReferencedCurrentSnapshotIgnoreDupli
}
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, files, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files, nil, false))
- t.Require().NoError(tx.AddFiles(t.ctx, files[len(files)-1:], nil, true))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files[len(files)-1:],
nil, true))
staged, err := tx.StagedTable()
t.Require().NoError(err)
@@ -1139,7 +1139,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesByPath() {
)
for i := range 5 {
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(ctx, files[i:i+1], nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(ctx, files[i:i+1],
nil, false))
tbl, err = tx.Commit(ctx)
t.Require().NoError(err)
}
@@ -1438,7 +1438,7 @@ func (t *TableWritingTestSuite) TestReplaceDataFiles() {
}
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, files, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx, files, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1521,7 +1521,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesValidatesPartitionSpecID() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1551,7 +1551,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesValidatesPartitionData() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1686,7 +1686,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesDuplicateAddPaths() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1708,7 +1708,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesDuplicateDeletePaths() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1732,7 +1732,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesAddAlreadyReferenced() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath2,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath1,
existingPath2}, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath1, existingPath2}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1753,7 +1753,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesDeleteNotInTable() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1776,7 +1776,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesNilDeleteFile() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1797,7 +1797,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesNilAddFile() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1832,7 +1832,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesInvalidAddContentType() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1893,7 +1893,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesByPathDuplicateDeletePaths()
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1914,7 +1914,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesByPathDuplicateAddPaths() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1937,7 +1937,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesByPathAddAlreadyReferenced()
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath2,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath1,
existingPath2}, nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath1, existingPath2}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -1955,7 +1955,7 @@ func (t *TableWritingTestSuite)
TestReplaceDataFilesByPathDeleteNotInTable() {
t.writeParquet(mustFS(t.T(), tbl).(iceio.WriteFileIO), existingPath,
t.arrTbl)
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(t.ctx, []string{existingPath}, nil,
false))
+ t.Require().NoError(tx.AddDataFilesByPath(t.ctx,
[]string{existingPath}, nil, false))
tbl, err := tx.Commit(t.ctx)
t.Require().NoError(err)
@@ -2013,7 +2013,7 @@ func (t *TableWritingTestSuite) TestExpireSnapshots() {
for i := range 5 {
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(ctx, files[i:i+1], nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(ctx, files[i:i+1],
nil, false))
tbl, err = tx.Commit(ctx)
t.Require().NoError(err)
}
@@ -2064,7 +2064,7 @@ func (t *TableWritingTestSuite)
TestExpireSnapshotsNoOpWhenNothingToExpire() {
// Create 3 snapshots
for i := range 3 {
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(ctx, files[i:i+1], nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(ctx, files[i:i+1],
nil, false))
tbl, err = tx.Commit(ctx)
t.Require().NoError(err)
}
@@ -2130,7 +2130,7 @@ func (t *TableWritingTestSuite)
TestExpireSnapshotsUsesTableProperties() {
for i := range 5 {
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(ctx, files[i:i+1], nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(ctx, files[i:i+1],
nil, false))
tbl, err = tx.Commit(ctx)
t.Require().NoError(err)
}
@@ -2182,7 +2182,7 @@ func (t *TableWritingTestSuite)
TestExpireSnapshotsWithMissingParent() {
// Create 5 snapshots, each one with a parent pointing to the previous
for i := range 5 {
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(ctx, files[i:i+1], nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(ctx, files[i:i+1],
nil, false))
tbl, err = tx.Commit(ctx)
t.Require().NoError(err)
}
@@ -2302,7 +2302,7 @@ func (t *TableWritingTestSuite)
TestExpireSnapshotsRejectsOnRefRollback() {
// Create 5 snapshots
for i := range 5 {
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(ctx, files[i:i+1], nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(ctx, files[i:i+1],
nil, false))
tbl, err = tx.Commit(ctx)
t.Require().NoError(err)
}
@@ -2372,7 +2372,7 @@ func (t *TableWritingTestSuite)
TestExpireSnapshotsRejectsOnRefUpdate() {
// Create 3 snapshots
for i := range 3 {
tx := tbl.NewTransaction()
- t.Require().NoError(tx.AddFiles(ctx, files[i:i+1], nil, false))
+ t.Require().NoError(tx.AddDataFilesByPath(ctx, files[i:i+1],
nil, false))
tbl, err = tx.Commit(ctx)
t.Require().NoError(err)
}
@@ -3152,13 +3152,13 @@ func (t *TableWritingTestSuite)
TestDeleteOldMetadataLogsErrorOnFileNotFound() {
// transaction 1 to create metadata file
tx := tbl.NewTransaction()
- tx.AddFiles(ctx, files[0:1], nil, false)
+ tx.AddDataFilesByPath(ctx, files[0:1], nil, false)
tbl_new, err := tx.Commit(ctx)
t.Require().NoError(err)
// transaction 2 to add files
tx_new := tbl_new.NewTransaction()
- tx_new.AddFiles(ctx, files[1:2], nil, false)
+ tx_new.AddDataFilesByPath(ctx, files[1:2], nil, false)
_, err = tx_new.Commit(ctx)
t.Require().NoError(err)
@@ -3205,7 +3205,7 @@ func (t *TableWritingTestSuite)
TestDeleteOldMetadataNoErrorLogsOnFileFound() {
// transaction 1 to create metadata file
tx := tbl.NewTransaction()
- tx.AddFiles(ctx, files[0:1], nil, false)
+ tx.AddDataFilesByPath(ctx, files[0:1], nil, false)
tbl_new, err := tx.Commit(ctx)
t.Require().NoError(err)
@@ -3217,7 +3217,7 @@ func (t *TableWritingTestSuite)
TestDeleteOldMetadataNoErrorLogsOnFileFound() {
// transaction 2 to add files
tx_new := tbl_new.NewTransaction()
- tx_new.AddFiles(ctx, files[1:2], nil, false)
+ tx_new.AddDataFilesByPath(ctx, files[1:2], nil, false)
_, err = tx_new.Commit(ctx)
t.Require().NoError(err)
@@ -3602,7 +3602,7 @@ func TestWritesToUnknownTransformSpecFail(t *testing.T) {
location :=
filepath.ToSlash(strings.ReplaceAll(t.TempDir(), "#", ""))
ctx := context.Background()
- t.Run("AddFiles", func(t *testing.T) {
+ t.Run("AddDataFilesByPath", func(t *testing.T) {
tbl := newTable(t, location, formatVersion,
unknownSpec)
arrTable, err :=
array.TableFromJSON(memory.DefaultAllocator, arrSchema, []string{
`[{"foo": true, "bar": "bar_string",
"baz": 1, "qux": "2024-03-07"}]`,
@@ -3615,7 +3615,7 @@ func TestWritesToUnknownTransformSpecFail(t *testing.T) {
require.NoError(t, err)
require.NoError(t, pqarrow.WriteTable(arrTable,
fo, arrTable.NumRows(), nil, pqarrow.DefaultWriterProps()))
- err = tbl.NewTransaction().AddFiles(ctx,
[]string{filePath}, nil, false)
+ err =
tbl.NewTransaction().AddDataFilesByPath(ctx, []string{filePath}, nil, false)
require.ErrorIs(t, err,
iceberg.ErrInvalidTransform)
assert.ErrorContains(t, err,
"custom_transform[42]")
})
diff --git a/table/transaction.go b/table/transaction.go
index b82477946..ca194e2fd 100644
--- a/table/transaction.go
+++ b/table/transaction.go
@@ -964,7 +964,7 @@ func (t *Transaction) ReplaceDataFilesByPath(ctx
context.Context, filesToDelete,
}
if len(filesToDelete) == 0 {
if len(filesToAdd) > 0 {
- return t.AddFiles(ctx, filesToAdd, snapshotProps, false)
+ return t.AddDataFilesByPath(ctx, filesToAdd,
snapshotProps, false)
}
}
@@ -1575,7 +1575,7 @@ func (t *Transaction) ensureNameMapping() error {
// This is useful for clients who have already constructed DataFile objects
with metadata,
// avoiding the need to read files to extract schema and statistics.
//
-// Unlike AddFiles, this method does not read files from storage. It validates
only metadata
+// Unlike AddDataFilesByPath, this method does not read files from storage. It
validates only metadata
// that can be checked without opening files (for example spec-id and
partition field IDs).
//
// Each DataFile may target any partition spec registered in the table
metadata,
@@ -2247,15 +2247,15 @@ func (t *Transaction) replaceFiles(ctx context.Context,
dataFilesToDelete, dataF
return nil
}
-type AddFilesOption func(addFilesOp *addFilesOperation)
+type AddDataFilesByPathOption func(addFilesOp *addFilesOperation)
type addFilesOperation struct {
concurrency int
}
-// WithAddFilesConcurrency overwrites the default concurrency for add files
operation.
+// WithAddDataFilesByPathConcurrency overrides the concurrency used by
AddDataFilesByPath.
// Default: runtime.GOMAXPROCS(0)
-func WithAddFilesConcurrency(concurrency int) AddFilesOption {
+func WithAddDataFilesByPathConcurrency(concurrency int)
AddDataFilesByPathOption {
return func(op *addFilesOperation) {
if concurrency > 0 {
op.concurrency = concurrency
@@ -2263,7 +2263,9 @@ func WithAddFilesConcurrency(concurrency int)
AddFilesOption {
}
}
-func (t *Transaction) AddFiles(ctx context.Context, filePaths []string,
snapshotProps iceberg.Properties, ignoreDuplicates bool, opts
...AddFilesOption) error {
+// AddDataFilesByPath adds data files at the provided paths to the table.
+// Unlike AddDataFiles, it opens each file to read its schema and statistics.
+func (t *Transaction) AddDataFilesByPath(ctx context.Context, filePaths
[]string, snapshotProps iceberg.Properties, ignoreDuplicates bool, opts
...AddDataFilesByPathOption) error {
meta, err := t.txnMeta()
if err != nil {
return err
@@ -2278,7 +2280,7 @@ func (t *Transaction) AddFiles(ctx context.Context,
filePaths []string, snapshot
set := make(map[string]struct{}, len(filePaths))
for _, filePath := range filePaths {
if _, ok := set[filePath]; ok {
- return errors.New("file paths must be unique for
AddFiles")
+ return errors.New("file paths must be unique for
AddDataFilesByPath")
}
set[filePath] = struct{}{}
}
diff --git a/table/transaction_empty_append_test.go
b/table/transaction_empty_append_test.go
index edf5227ec..b0907b2bd 100644
--- a/table/transaction_empty_append_test.go
+++ b/table/transaction_empty_append_test.go
@@ -34,10 +34,10 @@ import (
// such commits for metadata-only bookkeeping (e.g. recording ingestion
// progress in snapshot or table properties) that must still serialize
// through the producer's AssertRefSnapshotID requirement, which only
-// engages when a snapshot moves the head. AddFiles with an empty path
+// engages when a snapshot moves the head. AddDataFilesByPath with an empty
path
// list must keep committing an empty append rather than degrading to a
// silent no-op.
-// Condition: AddFiles with no paths on an empty table, then again on
+// Condition: AddDataFilesByPath with no paths on an empty table, then again on
// the resulting table after real data was added.
// Assertion: each commit creates a new "append" snapshot chained to its
// parent, the branch head moves, and existing data files stay reachable.
@@ -62,7 +62,7 @@ func TestAddFilesEmptyCommitsEmptySnapshot(t *testing.T) {
// Empty append on an empty table creates the branch head.
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), nil, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(), nil, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -80,14 +80,14 @@ func TestAddFilesEmptyCommitsEmptySnapshot(t *testing.T) {
writeParquetFile(t, dataPath, arrowSc, `[{"id": 1, "data": "alpha"}]`)
txData := tbl.NewTransaction()
- require.NoError(t, txData.AddFiles(t.Context(), []string{dataPath},
nil, false))
+ require.NoError(t, txData.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = txData.Commit(t.Context())
require.NoError(t, err)
withData := tbl.CurrentSnapshot()
require.NotNil(t, withData)
txEmpty := tbl.NewTransaction()
- require.NoError(t, txEmpty.AddFiles(t.Context(), nil, nil, false))
+ require.NoError(t, txEmpty.AddDataFilesByPath(t.Context(), nil, nil,
false))
tbl, err = txEmpty.Commit(t.Context())
require.NoError(t, err)
diff --git a/table/transaction_internal_test.go
b/table/transaction_internal_test.go
index 2c28305e7..587f3caf4 100644
--- a/table/transaction_internal_test.go
+++ b/table/transaction_internal_test.go
@@ -1169,7 +1169,9 @@ func
TestTransactionEntryPointsRejectNilAndBrokenTransaction(t *testing.T) {
{"AppendTable", func(txn *Transaction) error { return
txn.AppendTable(ctx, nil, 0, nil) }},
{"Append", func(txn *Transaction) error { return
txn.Append(ctx, nil, nil) }},
{"AddDataFiles", func(txn *Transaction) error { return
txn.AddDataFiles(ctx, []iceberg.DataFile{dataFile}, nil) }},
- {"AddFiles", func(txn *Transaction) error { return
txn.AddFiles(ctx, []string{"file://a.parquet"}, nil, false) }},
+ {"AddDataFilesByPath", func(txn *Transaction) error {
+ return txn.AddDataFilesByPath(ctx,
[]string{"file://a.parquet"}, nil, false)
+ }},
{"ReplaceDataFilesByPath", func(txn *Transaction) error {
return txn.ReplaceDataFilesByPath(ctx, []string{"a"},
[]string{"b"}, nil)
}},
@@ -2170,7 +2172,7 @@ func TestTableCommitWithSnapshotProducerPinsBaseHead(t
*testing.T) {
tbl, _, _ := newProducerAssertRefTable(t)
seed := tbl.NewTransaction()
- require.NoError(t, seed.AddFiles(t.Context(), nil, nil, false))
+ require.NoError(t, seed.AddDataFilesByPath(t.Context(), nil, nil,
false))
tbl, err := seed.Commit(t.Context())
require.NoError(t, err)
@@ -2178,7 +2180,7 @@ func TestTableCommitWithSnapshotProducerPinsBaseHead(t
*testing.T) {
require.NotNil(t, baseHead)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), nil, nil, false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(), nil, nil, false))
tc, err := tx.TableCommit()
require.NoError(t, err)
diff --git a/table/transaction_test.go b/table/transaction_test.go
index b3c02361f..cf323caa1 100644
--- a/table/transaction_test.go
+++ b/table/transaction_test.go
@@ -155,7 +155,7 @@ func (s *SparkIntegrationTestSuite) TestAddFile() {
}
tx := tbl.NewTransaction()
- err = tx.AddFiles(s.ctx, []string{filename}, nil, false)
+ err = tx.AddDataFilesByPath(s.ctx, []string{filename}, nil, false)
s.Require().NoError(err)
tbl, err = tx.Commit(s.ctx)
diff --git a/table/variant_shredded_test.go b/table/variant_shredded_test.go
index babc68060..d1bc9c231 100644
--- a/table/variant_shredded_test.go
+++ b/table/variant_shredded_test.go
@@ -72,7 +72,7 @@ func TestShreddedVariantTableScan(t *testing.T) {
)
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -185,7 +185,7 @@ func TestShreddedVariantScanNulls(t *testing.T) {
&rowDeltaCatalog{metadata: meta})
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -427,7 +427,7 @@ func scanShreddedFixture(t *testing.T, fixture string)
arrow.Table {
return scanned
}
-// scanShreddedFixtureErr returns the error from AddFiles or the scan instead
of
+// scanShreddedFixtureErr returns the error from AddDataFilesByPath or the
scan instead of
// failing, for the rejection cases.
func scanShreddedFixtureErr(t *testing.T, fixture string) error {
t.Helper()
@@ -467,7 +467,7 @@ func scanShreddedFixtureRaw(t *testing.T, fixture string)
(arrow.Table, error) {
&rowDeltaCatalog{metadata: meta})
tx := tbl.NewTransaction()
- if err := tx.AddFiles(t.Context(), []string{dataPath}, nil, false); err
!= nil {
+ if err := tx.AddDataFilesByPath(t.Context(), []string{dataPath}, nil,
false); err != nil {
return nil, err
}
tbl, err = tx.Commit(t.Context())
@@ -573,7 +573,7 @@ func writeFullyShreddedVariantFile(t *testing.T, path
string, fieldID, nRows int
writeVariantColumnParquet(t, path, fieldID, shreddedType, bldr)
}
-// TestShreddedVariantAddFilesBounds pins that AddFiles (which opens the file
and
+// TestShreddedVariantAddFilesBounds pins that AddDataFilesByPath (which opens
the file and
// recomputes) emits correct shredded variant child bounds under the parent
field id.
func TestShreddedVariantAddFilesBounds(t *testing.T) {
location := filepath.ToSlash(t.TempDir())
@@ -591,7 +591,7 @@ func TestShreddedVariantAddFilesBounds(t *testing.T) {
func(ctx context.Context) (iceio.IO, error) { return
iceio.LocalFS{}, nil }, &rowDeltaCatalog{metadata: meta})
tx := tbl.NewTransaction()
- require.NoError(t, tx.AddFiles(t.Context(), []string{dataPath}, nil,
false))
+ require.NoError(t, tx.AddDataFilesByPath(t.Context(),
[]string{dataPath}, nil, false))
tbl, err = tx.Commit(t.Context())
require.NoError(t, err)
@@ -614,8 +614,8 @@ func TestShreddedVariantAddFilesBounds(t *testing.T) {
return append(append([]byte{}, v.Metadata().Bytes()...),
v.Bytes()...)
}
- assert.Equal(t, buildObj(0, "row-A"),
tasks[0].File.LowerBoundValues()[1], "AddFiles lower bound")
- assert.Equal(t, buildObj(nData-1, "row-E"),
tasks[0].File.UpperBoundValues()[1], "AddFiles upper bound")
+ assert.Equal(t, buildObj(0, "row-A"),
tasks[0].File.LowerBoundValues()[1], "AddDataFilesByPath lower bound")
+ assert.Equal(t, buildObj(nData-1, "row-E"),
tasks[0].File.UpperBoundValues()[1], "AddDataFilesByPath upper bound")
}
// setupVariantScanTable writes a shredded variant table from rows and returns
a closure reading the "id" column of a filtered scan.
diff --git a/table/write_file_io_test.go b/table/write_file_io_test.go
index 13e3af85b..76dc39ff9 100644
--- a/table/write_file_io_test.go
+++ b/table/write_file_io_test.go
@@ -158,7 +158,7 @@ func TestWritePathsReturnErrorForReadOnlyIO(t *testing.T) {
{
name: "add files",
run: func(ctx context.Context, tbl *Table) error {
- return tbl.NewTransaction().AddFiles(ctx,
[]string{
+ return
tbl.NewTransaction().AddDataFilesByPath(ctx, []string{
"file:///tmp/read-only-write/add-files.parquet",
}, nil, false)
},
diff --git a/website/src/api.md b/website/src/api.md
index 82ccc145a..b52509255 100644
--- a/website/src/api.md
+++ b/website/src/api.md
@@ -354,7 +354,7 @@ When you already have data files (e.g. produced by another
writer), register the
```go
txn := tbl.NewTransaction()
-err := txn.AddFiles(ctx, []string{
+err := txn.AddDataFilesByPath(ctx, []string{
"s3://my-bucket/sales/orders/data/file-1.parquet",
"s3://my-bucket/sales/orders/data/file-2.parquet",
}, nil /* snapshot props */, false /* ignoreDuplicates */)
