Column headers only as headers

This fixes a bug where column headers were being repeated for each batch in the CSV output.
sul-dlss-labs · Jan 11, 2024 · e10a0af · e10a0af
1 parent 40a000f
commit e10a0af
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 2 deletions.
diff --git a/marctable/utils.py b/marctable/utils.py
@@ -31,6 +31,7 @@ def to_csv(
     first_batch = True
     for df in dataframe_iter(marc_input, rules=rules, batch=batch):
         df.to_csv(csv_output, header=first_batch, index=False)
+        first_batch = False
 
 
 def to_jsonl(

diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "marctable"
-version = "0.3.1"
+version = "0.3.2"
 description = "Convert MARC to CSV and Parquet"
 authors = ["Ed Summers <ehs@pobox.com>"]
 license = "Apache"

diff --git a/test_marctable.py b/test_marctable.py
@@ -133,7 +133,7 @@ def test_to_csv() -> None:
         open("test-data/utf8.marc", "rb"), open("test-data/utf8.csv", "w"), batch=1000
     )
     df = pandas.read_csv("test-data/utf8.csv")
-    assert len(df) == 10622
+    assert len(df) == 10612
     assert len(df.columns) == 215
     assert (
         df.iloc[0]["F245"]