datamule 0.418__cp311-cp311-macosx_10_9_universal2.whl → 0.420__cp311-cp311-macosx_10_9_universal2.whl

datamule/portfolio.py

@@ -1,16 +1,21 @@
 from pathlib import Path
+from tqdm import tqdm
+from concurrent.futures import ProcessPoolExecutor
 from .submission import Submission
 
 class Portfolio:
     def __init__(self, path):
         self.path = Path(path)
-        self.submissions = []
-        
-        # Load all subdirectories as submissions
-        for folder in self.path.iterdir():
-            if folder.is_dir():
-                self.submissions.append(Submission(folder))
-
+        folders = [f for f in self.path.iterdir() if f.is_dir()]
+        print(f"Loading {len(folders)} submissions")
+        # Load submissions in parallel
+        with ProcessPoolExecutor() as executor:
+            # Show progress while loading
+            self.submissions = list(tqdm(
+                executor.map(Submission, folders),
+                total=len(folders),
+                desc="Loading submissions"  
+            ))
 
     def __iter__(self):
         return iter(self.submissions)

datamule/submission.py

@@ -51,8 +51,14 @@ class Submission:
                 filepath.unlink()
 
     def document_type(self, document_type):
+        # Convert single document type to list for consistent handling
+        if isinstance(document_type, str):
+            document_types = [document_type]
+        else:
+            document_types = document_type
+
         for doc in self.metadata['documents']:
-            if doc['TYPE'] == document_type:
+            if doc['TYPE'] in document_types:
                 filename = doc.get('FILENAME')
                 if filename is None:
                     continue

{datamule-0.418.dist-info → datamule-0.420.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: datamule
-Version: 0.418
+Version: 0.420
 Summary: Making it easier to use SEC filings.
 Home-page: https://github.com/john-friedman/datamule-python
 Author: John Friedman
@@ -16,11 +16,11 @@ Requires-Dist: selectolax
 Requires-Dist: pytz
 Requires-Dist: zstandard
 Provides-Extra: all
+Requires-Dist: flask; extra == "all"
 Requires-Dist: psutil; extra == "all"
-Requires-Dist: pandas; extra == "all"
 Requires-Dist: openai; extra == "all"
 Requires-Dist: google-generativeai; extra == "all"
-Requires-Dist: flask; extra == "all"
+Requires-Dist: pandas; extra == "all"
 Provides-Extra: dataset_builder
 Requires-Dist: pandas; extra == "dataset-builder"
 Requires-Dist: google-generativeai; extra == "dataset-builder"

{datamule-0.418.dist-info → datamule-0.420.dist-info}/RECORD

@@ -3,8 +3,8 @@ datamule/document.py,sha256=Yn8UqUjKwYPE29MrMjreHK_HY9eTqOSjPyM5B1VBrHQ,5144
 datamule/helper.py,sha256=tr3AQWus9dHNZFKpLSglWjcb8zmm5qDXjOWACMhvMxQ,4594
 datamule/monitor.py,sha256=mRaM8v5NgcMF9DJ1s_YBzucjrbr-3yFwW422MVml-_Q,9114
 datamule/packageupdater.py,sha256=rw4hTnQGjZttee8QmTadg9vlhEjVp_dwdSFv8uQNvss,9584
-datamule/portfolio.py,sha256=piKibT_xZ9vZDpep1xS69eQceRrdq8FKqeRDf_-aQ38,425
-datamule/submission.py,sha256=aTpXH79vgVXzvpja6Kax-RdtyXiR7S9BwwkpMFmq7rA,2274
+datamule/portfolio.py,sha256=Qd_j3gGUi95Ad0g-PpjfJ3ozg-92VohMvCOvcEb0Kbw,710
+datamule/submission.py,sha256=LlwHJL3CIMHOlaHqCfDH7UBiKCFu0tPYV-g92uPvfcY,2492
 datamule/data/company_former_names.csv,sha256=HE9cAv-_QKFX6jT-_-D0rHmaDyQuAzL4MJwank5O1U8,706380
 datamule/data/company_metadata.csv,sha256=yPovrCVjYwLWTU_hBUFJymp8iNO0NBYuq_QwOkRLoN8,3068599
 datamule/data/company_tickers.csv,sha256=GW6lOP54RiGJCx-d9N5jEBy7tGVgU3zI-5xHJXrZfSI,400363
@@ -36,7 +36,6 @@ datamule/parser/document_parsing/basic_10q_parser.py,sha256=ccQc3pwBqevDb6-vBwEE
 datamule/parser/document_parsing/basic_13d_parser.py,sha256=loJC97H_ccu_hWMhgNt5tvGZnN3--7tsqZxzBnWB_FY,1528
 datamule/parser/document_parsing/basic_13g_parser.py,sha256=sWg83-QTAzUDNs45iWtpxnMxQgtC3zJlFj0R9ybZpNI,1631
 datamule/parser/document_parsing/basic_8k_parser.py,sha256=inCSmlH_BkLK0Lkvt0kZ6EUJ0nijul_RkdXzccyOmRI,2466
-datamule/parser/document_parsing/company_concepts_parser.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datamule/parser/document_parsing/form_d_parser.py,sha256=dWlGeVZRzh0kfT3gVMC8eyqeQORdVV3r8KXUwEqAW3s,2036
 datamule/parser/document_parsing/generalized_item_parser.py,sha256=67_DFb1BQbMmdHefEgoCPlEoiUT0zyxh3eBNJpjGXUk,2616
 datamule/parser/document_parsing/generalized_xml_parser.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -48,8 +47,8 @@ datamule/parser/document_parsing/n_port_p_parser.py,sha256=GmmQFkCZt57WikUZ5Daht
 datamule/parser/document_parsing/sec_parser.py,sha256=AS8H4h1sfUAdWP2gotULcjbylsYN_nHgTfkeVRyENPo,2716
 datamule/parser/document_parsing/sgml_parser.py,sha256=tC1cL3cdVQPWbc9QtoRUYSo2wRuYNaglFaCmP57oEfA,3317
 datamule/parser/sgml_parsing/sgml_parser_cy.c,sha256=UwXSuLGSBLKfO5bM7xTzjFLnGDV3-NNnCjgUCPAY1gk,796689
-datamule/parser/sgml_parsing/sgml_parser_cy.cpython-311-darwin.so,sha256=zRsW_WYAuzA942b6X9fUg8lQB5b-gVVS_7FO0g-5pZ4,362376
-datamule-0.418.dist-info/METADATA,sha256=qDLN8uEVIL0x3kUHnfs3lENMLdOK7ky0dA7IT3TMuvs,1007
-datamule-0.418.dist-info/WHEEL,sha256=vrdZjlh4exCGuOeHFVGYgW_9eYEDoByDCp4vVTWiG0w,115
-datamule-0.418.dist-info/top_level.txt,sha256=iOfgmtSMFVyr7JGl_bYSTDry79JbmsG4p8zKq89ktKk,9
-datamule-0.418.dist-info/RECORD,,
+datamule/parser/sgml_parsing/sgml_parser_cy.cpython-311-darwin.so,sha256=hds9ZxW6dUjdyxCvqlPmuDbgRpR5Q8xGJ53aeb7seiE,362376
+datamule-0.420.dist-info/METADATA,sha256=yl7VYR4JZIqpbkLIUYZXM4DrFs1NpKdvaffNIVYg2xo,1007
+datamule-0.420.dist-info/WHEEL,sha256=vrdZjlh4exCGuOeHFVGYgW_9eYEDoByDCp4vVTWiG0w,115
+datamule-0.420.dist-info/top_level.txt,sha256=iOfgmtSMFVyr7JGl_bYSTDry79JbmsG4p8zKq89ktKk,9
+datamule-0.420.dist-info/RECORD,,