Merge pull request #163 from tombom3000000/fix_for_windows

Updated to work with threading on windows when running as a main scrit instead of importing as a module (main module protection is required). Also updated tests with new URL to pass, but these may change as you have stated.

Author: khoivan88

find_sds/find_sds.py

@@ -83,21 +83,27 @@ def find_sds(cas_list: List[str], download_path: str = None, pool_size: int = 10
 
     download_result = []
     try:
-        # # Using multithreading
+        # # Using multiprocessing (need to protect on Windows)
         if not debug:
-            with Pool(pool_size) as p:
-                download_result = p.map(partial(
-                                        download_sds,
-                                        download_path=download_path),
-                                    to_be_downloaded)
+            # On Windows, multiprocessing requires the main module to be protected
+            if __name__ == '__main__':
+                with Pool(pool_size) as p:
+                    download_result = p.map(partial(
+                                            download_sds,
+                                            download_path=download_path),
+                                        to_be_downloaded)
+            else:
+                # Fall back to sequential processing if not in main
+                download_result = []
+                for cas_nr in to_be_downloaded:
+                    download_result.append(download_sds(cas_nr=cas_nr, download_path=download_path))
         else:
             download_result = []
             for cas_nr in to_be_downloaded:
                 download_result.append(download_sds(cas_nr=cas_nr, download_path=download_path))
     except Exception as error:
         # if debug:
-        traceback_str = ''.join(traceback.format_exception(etype=type(error), value=error, tb=error.__traceback__))
-        print(traceback_str)
+        traceback.print_exc()
 
 
     # Step 2: print out summary

tests/unit/test_extract_url.py

@@ -69,7 +69,7 @@ def test_extract_url_from_fisher_with_exception(monkeypatch, cas_nr, expect):
             # 'ChemicalSafety',
             # 'http://sds.chemicalsafety.com/sds/pda/msds/getpdf.ashx?action=msdsdocument&auth=200C200C200C200C2008207A200D2078200C200C200C200C200C200C200C200C200C2008&param1=ZmRwLjFfMTQ2NzY2MDNORQ==&unique='
             # 'ChemScene LLC', 'https://file.ambeed.com/static/upload/prosds/am/178/SDS-A177089.pdf',
-            'Ambeed, Inc.', 'https://file.ambeed.com/static/upload/prosds/am/178/SDS-A177089.pdf',
+            'ChemScene LLC', 'https://file.chemscene.com/pdf/UsaMSDS/MSDSUSACS-0030746.pdf',
             )
         ),
         ('1450-76-6', (
@@ -109,8 +109,8 @@ def test_extract_url_from_chemicalsafety_with_exception(monkeypatch, cas_nr, exp
 @pytest.mark.parametrize(
     "cas_nr, expect", [
         ('623-51-8', (
-            None,
-            None
+            'Fluorochem', 
+            'https://7128445.app.netsuite.com/core/media/media.nl?id=8540237&c=7128445&h=jbNYp3s7X1x7ed2QDCYb7_vQCj1Ae8VUCUBjXHjCnNLQCwhZ&_xt=.pdf'
             )
         ),
         ('28697-53-2', (
@@ -120,7 +120,7 @@ def test_extract_url_from_chemicalsafety_with_exception(monkeypatch, cas_nr, exp
         ),
         ('1450-76-6', (
             'Fluorochem',
-            'https://7128445.app.netsuite.com/core/media/media.nl?id=3274568&c=7128445&h=JZhDK4ckHBy0gdR1VbKyhzkmtZYBQzSKpmFRM33N7hUNvc4D&_xt=.pdf'
+            'https://7128445.app.netsuite.com/core/media/media.nl?id=8259398&c=7128445&h=NWks4Z9GW9rIsd-QxKkLp5Mn0dhDNmooZYDjFD36i0k3CnCp&_xt=.pdf'
             )
         ),
         ('491588-98-8', (