fix: Change strategy back to auto from fast (#152)

* Change strategy back to auto

Author: Jason Scheirer

pipeline-notebooks/pipeline-general.ipynb

@@ -772,7 +772,7 @@
     "        # since fast api might sent the wrong one.\n",
     "        file_content_type = \"application/x-ole-storage\"\n",
     "    \n",
-    "    strategy = (m_strategy[0] if len(m_strategy) else 'fast').lower()\n",
+    "    strategy = (m_strategy[0] if len(m_strategy) else 'auto').lower()\n",
     "    strategies = ['fast', 'hi_res', 'auto', 'ocr_only']\n",
     "    if strategy not in strategies:\n",
     "        raise HTTPException(\n",

prepline_general/api/general.py

@@ -236,7 +236,7 @@ def pipeline_api(
         # since fast api might sent the wrong one.
         file_content_type = "application/x-ole-storage"
 
-    strategy = (m_strategy[0] if len(m_strategy) else "fast").lower()
+    strategy = (m_strategy[0] if len(m_strategy) else "auto").lower()
     strategies = ["fast", "hi_res", "auto", "ocr_only"]
     if strategy not in strategies:
         raise HTTPException(

scripts/smoketest.py

@@ -15,7 +15,7 @@
 def send_document(
     filename,
     content_type,
-    strategy="fast",
+    strategy="auto",
     output_format="application/json",
     pdf_infer_table_structure="false",
 ):
@@ -110,14 +110,21 @@ def test_strategy_performance():
     performance_ratio = 4
     test_file = Path("sample-docs") / "layout-parser-paper.pdf"
 
-    start_time = time.time()
+    start_time = time.monotonic()
     response = send_document(test_file, content_type="application/pdf", strategy="hi_res")
-    hi_res_time = time.time() - start_time
+    hi_res_time = time.monotonic() - start_time
     assert response.status_code == 200
 
-    start_time = time.time()
+    start_time = time.monotonic()
+    response = send_document(test_file, content_type="application/pdf", strategy="auto")
+    auto_time = time.monotonic() - start_time
+    assert response.status_code == 200
+
+    assert hi_res_time > performance_ratio * auto_time
+
+    start_time = time.monotonic()
     response = send_document(test_file, content_type="application/pdf", strategy="fast")
-    fast_time = time.time() - start_time
+    fast_time = time.monotonic() - start_time
     assert response.status_code == 200
 
     assert hi_res_time > performance_ratio * fast_time