fix(output): align TSV rows to block headers

tests/integration_tests.py

@@ -240,3 +240,32 @@ def test_convert_preserves_explicit_empty_string_fields():
     assert row[header.index("watermark")] == ""
     assert row[header.index("fieldType")] == "text"
     assert row[header.index("displayOrder")] == "1"
+
+
+def test_convert_aligns_rows_to_header_order(tmp_path):
+    """Ensure item key order does not shift values into wrong TSV columns."""
+    input_file = "tests/valid/mixed_key_order.yml"
+    output_file = tmp_path / "mixed_key_order.tsv"
+    runner = CliRunner()
+
+    result = runner.invoke(
+        yml2block.__main__.main,
+        ["convert", input_file, "-o", str(output_file)],
+    )
+
+    assert result.exit_code == 0, result.output
+    lines = output_file.read_text().splitlines()
+    header_idx = next(
+        idx for idx, line in enumerate(lines) if line.startswith("#datasetField")
+    )
+    header = lines[header_idx].split("\t")
+    row = lines[header_idx + 2].split("\t")
+
+    assert row[header.index("name")] == "Second"
+    assert row[header.index("watermark")] == "Second watermark"
+    assert row[header.index("fieldType")] == "textbox"
+
+    expected_width = len(header)
+    assert expected_width > 0
+    for line in lines:
+        assert len(line.split("\t")) == expected_width

tests/valid/mixed_key_order.yml

@@ -0,0 +1,35 @@
+---
+metadataBlock:
+  - name: MixedOrder
+    displayName: Mixed Order
+datasetField:
+  - name: First
+    title: First
+    description: First description
+    watermark: First watermark
+    fieldType: text
+    displayOrder: 1
+    advancedSearchField: true
+    allowControlledVocabulary: false
+    allowmultiples: false
+    facetable: false
+    displayoncreate: true
+    required: true
+    parent:
+    metadatablock_id: MixedOrder
+    termURI: https://example.org/first
+  - fieldType: textbox
+    watermark: Second watermark
+    name: Second
+    title: Second
+    description: Second description
+    displayOrder: 2
+    displayFormat:
+    advancedSearchField: false
+    allowControlledVocabulary: false
+    allowmultiples: false
+    facetable: false
+    displayoncreate: false
+    required: false
+    parent:
+    metadatablock_id: MixedOrder

yml2block/output.py

@@ -15,12 +15,20 @@ def write_metadata_block(yml_metadata, output_path, longest_line, verbose):
             block_lines = []
 
             for block_line in content:
-                new_line = [""]
-
-                for key, entry in block_line.items():
+                for key in block_line.keys():
                     if key not in block_headers:
                         block_headers.append(key)
 
+            for block_line in content:
+                new_line = [""]
+
+                for key in block_headers:
+                    if key not in block_line:
+                        new_line.append("")
+                        continue
+
+                    entry = block_line[key]
+
                     value = entry.value
                     # TODO: Consider screening for True, False, None
                     # before and replace them.

yml2block/validation.py

@@ -85,11 +85,17 @@ def validate_entry(yaml_chunk, tsv_keyword, lint_conf, verbose):
             lint = lint_conf.get(lint)
             violations.extend(lint(item, tsv_keyword))
 
-        # Compute the highest number of columns in the block
-        row_length = (
-            len(item.keys()) if tsv_keyword == "metadataBlock" else len(item.keys()) + 1
-        )
-        longest_row = max(longest_row, row_length)
+    # Compute the highest number of columns in the block
+    block_headers = []
+    for item in yaml_chunk:
+        for key in item.keys():
+            if key not in block_headers:
+                block_headers.append(key)
+
+    row_length = (
+        len(block_headers) if tsv_keyword == "metadataBlock" else len(block_headers) + 1
+    )
+    longest_row = max(longest_row, row_length)
 
     if verbose and len(violations) == 0:
         print("SUCCESS!" if verbose == 1 else "SUCCESS!\n")