datamule 0.415__cp38-cp38-win_amd64.whl

datamule/portfolio.py

@@ -0,0 +1,16 @@
+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/submission.py

@@ -0,0 +1,61 @@
+from pathlib import Path
+import json
+from .document import Document
+
+class Submission:
+    def __init__(self, path):
+        self.path = Path(path)
+        self._load_metadata()
+    
+    def _load_metadata(self):
+        metadata_path = self.path / 'metadata.json'
+        with metadata_path.open('r') as f:
+            self.metadata = json.load(f)
+            
+    def keep(self, document_types):
+        """Keep files of specified document types, delete others
+        Args:
+            document_types: string or list of strings representing document types to keep
+        """
+        # Convert single string to list for consistent handling
+        if isinstance(document_types, str):
+            document_types = [document_types]
+            
+        for doc in self.metadata['documents']:
+            filename = doc.get('FILENAME')
+            if filename is None:
+                continue
+                
+            filepath = self.path / filename
+            # Delete if document type isn't in our keep list
+            if doc['TYPE'] not in document_types and filepath.exists():
+                filepath.unlink()
+                
+    def drop(self, document_types):
+        """Delete files of specified document types, keep others
+        Args:
+            document_types: string or list of strings representing document types to drop
+        """
+        # Convert single string to list for consistent handling
+        if isinstance(document_types, str):
+            document_types = [document_types]
+            
+        for doc in self.metadata['documents']:
+            filename = doc.get('FILENAME')
+            if filename is None:
+                continue
+                
+            filepath = self.path / filename
+            # Delete if document type is in our drop list
+            if doc['TYPE'] in document_types and filepath.exists():
+                filepath.unlink()
+
+    def document_type(self, document_type):
+        for doc in self.metadata['documents']:
+            if doc['TYPE'] == document_type:
+                filename = doc.get('FILENAME')
+                if filename is None:
+                    continue
+                    
+                document_path = self.path / filename
+                yield Document(doc['TYPE'], document_path)

datamule-0.415.dist-info/METADATA

@@ -0,0 +1,37 @@
+Metadata-Version: 2.1
+Name: datamule
+Version: 0.415
+Summary: Making it easier to use SEC filings.
+Home-page: https://github.com/john-friedman/datamule-python
+Author: John Friedman
+License: UNKNOWN
+Platform: UNKNOWN
+Requires-Dist: aiohttp
+Requires-Dist: aiolimiter
+Requires-Dist: tqdm
+Requires-Dist: requests
+Requires-Dist: nest-asyncio
+Requires-Dist: aiofiles
+Requires-Dist: polars
+Requires-Dist: setuptools
+Requires-Dist: selectolax
+Requires-Dist: pytz
+Requires-Dist: zstandard
+Provides-Extra: all
+Requires-Dist: pandas; extra == "all"
+Requires-Dist: psutil; extra == "all"
+Requires-Dist: flask; extra == "all"
+Requires-Dist: google-generativeai; extra == "all"
+Requires-Dist: openai; extra == "all"
+Provides-Extra: dataset_builder
+Requires-Dist: pandas; extra == "dataset-builder"
+Requires-Dist: google-generativeai; extra == "dataset-builder"
+Requires-Dist: psutil; extra == "dataset-builder"
+Provides-Extra: mulebot
+Requires-Dist: openai; extra == "mulebot"
+Provides-Extra: mulebot_server
+Requires-Dist: flask; extra == "mulebot-server"
+
+UNKNOWN
+
+

datamule-0.415.dist-info/RECORD

@@ -0,0 +1,37 @@
+datamule/__init__.py,sha256=c5Ozl19hzvyMqG75Zl3XFHIBQihxfB47uoTm7EfOOTQ,2033
+datamule/document.py,sha256=6xEaI-32AQiBxX3gZcX4Qr49bgvcvLviFAwUGpTwtr0,5273
+datamule/helper.py,sha256=8HOjB3Y7svw_zjEY-AY5JKOJ-LrBiuQMPyok3MH6CCg,4716
+datamule/monitor.py,sha256=WVds1HGV_ojYgWmo0b4Dsiv9mzZ85HHnCucH-7XoUw8,9350
+datamule/packageupdater.py,sha256=qVabK4YwSd62OjIvNY59hcQNgZCi4R2PRgCenq7THgE,9790
+datamule/portfolio.py,sha256=is7QY2QkK9S2U2yZ6INC8TmLqIy5E1i0TUAlVT-Qzhg,440
+datamule/submission.py,sha256=TNaj-epcpuV8JLRwawForQRFUAlqbyvRwAykj6xLANE,2334
+datamule/data/company_former_names.csv,sha256=zTBWdV12_JE3aROFOMrFNTHLPW_M4TDruxtl15-XfA0,714528
+datamule/data/company_metadata.csv,sha256=X7uSIwConqC0sz-moIhXIISg6FI7GLGSlvAfDDf8Sd0,3078648
+datamule/data/company_tickers.csv,sha256=ihU6aNFriN0lADloCO85Op04deFk3qVcLZ0EJhi5QVo,410362
+datamule/data/sec-glossary.csv,sha256=TPjTBVM3kyFd8xHsmihykepvKbuLAAthOfEDjh_H-Kk,251937
+datamule/data/xbrl_descriptions.csv,sha256=Hg9BOo9zSjR7Khvx0pikILcbmDK_A404dmQtWuESK4s,2631548
+datamule/dataset_builder/dataset_builder.py,sha256=h1JDzLcMKxxMcXcD24EyqjUPp78iLWCovfNLQtpwZi4,10005
+datamule/mulebot/__init__.py,sha256=GM5cTnijSSLO9GXFdsCuz5B1iwGUcxDbpoBQ6zw1Odo,30
+datamule/mulebot/helper.py,sha256=Hzzr2HReHpFe2GfpVU79EXvQFx3oL9UiwkJp--Sd1N4,1114
+datamule/mulebot/mulebot.py,sha256=wN0Tv5fvarXgk1LRCcgPhj8Xgd8uYFn-cMucxTxRsEo,5966
+datamule/mulebot/search.py,sha256=kuuPaMp2z9U8GY2LCulmqC8VFlYOLXf3bfLaBETObb0,2124
+datamule/mulebot/tools.py,sha256=6giVRck3dgHmX0w9IvTe5I2FAiaBGTYuvwH6futDaes,3022
+datamule/mulebot/mulebot_server/__init__.py,sha256=x1QhXys7BWxi2g9_ZHUYA6S6rL3VL2718x4rYtGaaIg,33
+datamule/mulebot/mulebot_server/server.py,sha256=G9Pd0mGNWU04pHc73D8rIn7PWGjoApd3BBWn3CwXkHM,3633
+datamule/mulebot/mulebot_server/static/css/minimalist.css,sha256=0oGf8p1cJFLjPUfx95rH6tbtbeUFUbZ1DhmzYkak6ro,3147
+datamule/mulebot/mulebot_server/static/scripts/artifacts.js,sha256=jvcvjRioXwKx4dJwyIrVnD0sEpdHR13DrVwfWzgcXag,2325
+datamule/mulebot/mulebot_server/static/scripts/chat.js,sha256=i9HIPcOLTuhOwT6AIxLI5FLQhMHIaDfaFSBSJCpEIlg,3199
+datamule/mulebot/mulebot_server/static/scripts/filingArtifacts.js,sha256=yVDthXjmZ9cRSeu_9mRPf8JWAWeuwvRi5Hd9o_3Puek,2089
+datamule/mulebot/mulebot_server/static/scripts/listArtifacts.js,sha256=kZeRl6ztFnOd46xNbUXIvEKYbgFcxT4ABDDmataEVpg,443
+datamule/mulebot/mulebot_server/static/scripts/main.js,sha256=GXWqjzFozNhWdc2jgh0mplIGM_FUG4LR-A8DGN5yTgg,1938
+datamule/mulebot/mulebot_server/static/scripts/prefilledPrompt.js,sha256=qRbhF3MmwmIQAV-QYxw-SOMFuFSF2tH5oBen5OYX5hQ,1119
+datamule/mulebot/mulebot_server/static/scripts/suggestions.js,sha256=vqFoHG0z0YIhmA1w2X60biKRSWlkYkXcVuAmwbvsCaE,1841
+datamule/mulebot/mulebot_server/static/scripts/tableArtifacts.js,sha256=jelibkbM9YlrlU7_-kaVisMh8OjIKWZpJQ9_FiHV4OQ,4565
+datamule/mulebot/mulebot_server/static/scripts/utils.js,sha256=pOzsJ6bGxakN6G5qUvmkZd34u0GRHFxvH3pKpUmnhMc,1282
+datamule/mulebot/mulebot_server/templates/chat-minimalist.html,sha256=mm-LomSoNcJImsFwLpFWt3kIFnxUzIPmUZ1Dt5BFN0s,7065
+datamule/parser/sgml_parsing/sgml_parser_cy.c,sha256=vxLnjpUgZ5LLvBvzYI_CZVxjpgRpulnzj3EFQG5eB8g,797203
+datamule/parser/sgml_parsing/sgml_parser_cy.cp38-win_amd64.pyd,sha256=oZADLJ-UJ07DsLNNoZVIHU2IdnSDDnNbZV8x-O4_u4Y,143360
+datamule-0.415.dist-info/METADATA,sha256=6Yn6Yt9CsgMRFqHrmi2lFnwmi6q-EAKNfNFRVYx_fKw,1089
+datamule-0.415.dist-info/WHEEL,sha256=N6sQ21sT3WcdJJLEJNx13ERof8WaAHD2AD1HUbWJ67E,100
+datamule-0.415.dist-info/top_level.txt,sha256=iOfgmtSMFVyr7JGl_bYSTDry79JbmsG4p8zKq89ktKk,9
+datamule-0.415.dist-info/RECORD,,

datamule-0.415.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: bdist_wheel (0.45.1)
+Root-Is-Purelib: false
+Tag: cp38-cp38-win_amd64
+

datamule-0.415.dist-info/top_level.txt

@@ -0,0 +1 @@
+datamule