datamule 0.417__tar.gz → 0.420__tar.gz

{datamule-0.417 → datamule-0.420}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: datamule
-Version: 0.417
+Version: 0.420
 Summary: Making it easier to use SEC filings.
 Home-page: https://github.com/john-friedman/datamule-python
 Author: John Friedman

{datamule-0.417 → datamule-0.420}/datamule/__init__.py

@@ -12,12 +12,21 @@ def __getattr__(name):
     elif name == 'Parser':
         from .parser.document_parsing.sec_parser import Parser
         return Parser
+    elif name == 'Monitor':
+        from .monitor import Monitor
+        return Monitor
+    elif name == 'PackageUpdater':
+        from .packageupdater import PackageUpdater
+        return PackageUpdater
     elif name == 'Submission':
         from .submission import Submission
         return Submission
     elif name == 'Portfolio':
         from .portfolio import Portfolio
         return Portfolio
+    elif name == 'Document':
+        from .document import Document
+        return Document
     elif name == "parse_sgml_submission":
         from .parser.sgml_parsing.sgml_parser_cy import parse_sgml_submission
         return parse_sgml_submission

datamule-0.420/datamule/portfolio.py

@@ -0,0 +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)
+        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-0.417 → datamule-0.420}/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-0.417 → datamule-0.420}/datamule.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: datamule
-Version: 0.417
+Version: 0.420
 Summary: Making it easier to use SEC filings.
 Home-page: https://github.com/john-friedman/datamule-python
 Author: John Friedman

{datamule-0.417 → datamule-0.420}/datamule.egg-info/SOURCES.txt

@@ -42,7 +42,6 @@ datamule/parser/document_parsing/basic_10q_parser.py
 datamule/parser/document_parsing/basic_13d_parser.py
 datamule/parser/document_parsing/basic_13g_parser.py
 datamule/parser/document_parsing/basic_8k_parser.py
-datamule/parser/document_parsing/company_concepts_parser.py
 datamule/parser/document_parsing/form_d_parser.py
 datamule/parser/document_parsing/generalized_item_parser.py
 datamule/parser/document_parsing/generalized_xml_parser.py

{datamule-0.417 → datamule-0.420}/setup.py

@@ -55,7 +55,7 @@ extras["all"] = list(all_dependencies)
 setup(
     name="datamule",
     author="John Friedman",
-    version="0.417",
+    version="0.420",
     description="Making it easier to use SEC filings.",
     packages=find_namespace_packages(include=['datamule*']),
     url="https://github.com/john-friedman/datamule-python",

datamule-0.417/datamule/portfolio.py

@@ -1,16 +0,0 @@
-from pathlib import Path
-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))
-
-
-    def __iter__(self):
-        return iter(self.submissions)

{datamule-0.417 → datamule-0.420}/datamule.egg-info/requires.txt

@@ -12,10 +12,10 @@ zstandard
 
 [all]
 openai
-psutil
+flask
 google-generativeai
 pandas
-flask
+psutil
 
 [dataset_builder]
 pandas