feat(data-process): 支持单项生成五十条数据

backend/tests/test_data_process_algorithms.py

@@ -21,7 +21,6 @@ from app.modules.data_process.algorithms import (
     detect_document_structure,
     detect_pdf_document_noise,
     detect_text_format,
-    estimate_token_count,
     extract_pdf_page_texts,
     extract_structured_records,
     generate_standard_records,
@@ -748,3 +747,26 @@ def test_generate_standard_records_supports_json_qa_and_stable_variants() -> Non
         split={"train": 100, "validation": 0, "test": 0},
         split_seed="task-1",
     )
+
+
+def test_generate_standard_records_supports_fifty_unique_semantic_variants() -> None:
+    records = generate_standard_records(
+        [{"id": "preview-50", "edited_content": "问：如何操作？\n答：按步骤操作。"}],
+        qa_pairs_per_item=50,
+        semantic_enrichment=True,
+        split={"train": 100, "validation": 0, "test": 0},
+        split_seed="task-50",
+    )
+
+    assert len(records) == 50
+    assert len({record["id"] for record in records}) == 50
+    assert len({record["instruction"] for record in records}) == 50
+    assert all(record["status"] == "valid" for record in records)
+
+
+@pytest.mark.parametrize("qa_pairs_per_item", [0, 51])
+def test_generate_standard_records_rejects_out_of_range_count(
+    qa_pairs_per_item: int,
+) -> None:
+    with pytest.raises(ValueError, match=r"\[1, 50\]"):
+        generate_standard_records([], qa_pairs_per_item=qa_pairs_per_item)