datamule 1.2.2__py3-none-any.whl → 1.2.3__py3-none-any.whl

datamule/submission.py

@@ -17,6 +17,7 @@ class Submission:
             self.path = None
             self.metadata, raw_documents = parse_sgml_submission_into_memory(sgml_content)
 
+            # code dupe
             self.accession = self.metadata['accession-number']
             self.filing_date= f"{self.metadata['filing-date'][:4]}-{self.metadata['filing-date'][4:6]}-{self.metadata['filing-date'][6:8]}"
     
@@ -43,6 +44,11 @@ class Submission:
             with metadata_path.open('r') as f:
                 self.metadata = json.load(f)
 
+            # Code dupe
+            self.accession = self.metadata['accession-number']
+            self.filing_date= f"{self.metadata['filing-date'][:4]}-{self.metadata['filing-date'][4:6]}-{self.metadata['filing-date'][6:8]}"
+    
+
 
 
     def document_type(self, document_type):
@@ -65,9 +71,12 @@ class Submission:
                     document_path = self.path / filename
                     extension = document_path.suffix
 
-                    with document_path.open('r') as f:
+                    with document_path.open('rb') as f:
                         content = f.read()
 
+                    if extension in ['.htm','.html','.txt','.xml']:
+                        content = content.decode('utf-8', errors='replace')
+
                     yield Document(type=doc['type'], content=content, extension=extension,filing_date=self.filing_date,accession=self.accession,path=document_path)
                 # if loaded from sgml_content
                 else:
@@ -89,9 +98,12 @@ class Submission:
 
                 # check if the file exists
                 if document_path.exists():
-                    with document_path.open('r') as f:
+                    with document_path.open('rb') as f:
                         content = f.read()
 
+                    if extension in ['.htm','.html','.txt','.xml']:
+                        content = content.decode('utf-8', errors='replace')
+
                     yield Document(type=doc['type'], content=content, extension=extension,filing_date=self.filing_date,accession=self.accession,path=document_path)
                 else:
                     print(f"Warning: File {document_path} does not exist likely due to keep types in downloading.")
@@ -100,28 +112,6 @@ class Submission:
             else:
                 yield self.documents[idx]
 
-    # keep documents by document type
-    def keep(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
-
-        if self.path is not None:
-            for doc in self.metadata['documents']:
-                filename = doc.get('filename')
-                type = doc.get('type')
-                if type not in document_types:
-                    # oh we need handling here for sequences case
-                    if filename is None:
-                        filename = doc.sequence + '.txt'
-
-                    document_path = self.path / filename
-                    # delete the file
-                    document_path.unlink()
-        else:
-            print("Warning: keep() method is only available when loading from path.")
 
 
 

{datamule-1.2.2.dist-info → datamule-1.2.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: datamule
-Version: 1.2.2
+Version: 1.2.3
 Summary: Making it easier to use SEC filings.
 Home-page: https://github.com/john-friedman/datamule-python
 Author: John Friedman

{datamule-1.2.2.dist-info → datamule-1.2.3.dist-info}/RECORD

@@ -5,7 +5,7 @@ datamule/index.py,sha256=0txvbzPcvY1GsdxA-wGdLzAByxSeE_1VyyBp9mZEQRM,2292
 datamule/package_updater.py,sha256=Z9zaa_y0Z5cknpRn8oPea3gg4kquFHfpfhduKKCZ6NU,958
 datamule/portfolio.py,sha256=so6j2KrkcZOToHIqkANAu3CC4QsfgaUN1zk9CrbRe1E,7225
 datamule/sheet.py,sha256=TvFqK9eAYuVoJ2uWdAlx5EN6vS9lke-aZf7FqtUiDBc,22304
-datamule/submission.py,sha256=tc4-8houjT2gfSK0P7ekowPduT31rj5_zt0axwZUacc,8483
+datamule/submission.py,sha256=HXuFL6snLevGk7DGlvPbjcBOJuccAIxEPXnkA1TXX8Y,8121
 datamule/document/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datamule/document/document.py,sha256=BRnHPVt-vIT7EZTF-c-Ulv3N33xX9zE02Q9mKXVDeuY,9474
 datamule/document/processing.py,sha256=fw-1OWfbmZhG1R8XpJx_vcGwz3_djmk0FrblHAMPmwc,27476
@@ -34,7 +34,7 @@ datamule/seclibrary/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSu
 datamule/seclibrary/bq.py,sha256=C8sb_rpXTvchprrFLcbRar4Qi0XWW25tnv1YsHSS5o4,18025
 datamule/seclibrary/downloader.py,sha256=fJztJ_sEfv2oHHbDff07DRlXLmztXnzt3Yvv5YaZgGk,13718
 datamule/seclibrary/query.py,sha256=qGuursTERRbOGfoDcYcpo4oWkW3PCBW6x1Qf1Puiak4,7352
-datamule-1.2.2.dist-info/METADATA,sha256=QpXbg-4cnRknynj-W4Z2Sc1zKlWan62zEG8OrN2_E-A,512
-datamule-1.2.2.dist-info/WHEEL,sha256=oiQVh_5PnQM0E3gPdiz09WCNmwiHDMaGer_elqB3coM,92
-datamule-1.2.2.dist-info/top_level.txt,sha256=iOfgmtSMFVyr7JGl_bYSTDry79JbmsG4p8zKq89ktKk,9
-datamule-1.2.2.dist-info/RECORD,,
+datamule-1.2.3.dist-info/METADATA,sha256=3gODk6YjozgMTYnjvXRX_pox_Otkq7tSDZY2LEl6MiU,512
+datamule-1.2.3.dist-info/WHEEL,sha256=oiQVh_5PnQM0E3gPdiz09WCNmwiHDMaGer_elqB3coM,92
+datamule-1.2.3.dist-info/top_level.txt,sha256=iOfgmtSMFVyr7JGl_bYSTDry79JbmsG4p8zKq89ktKk,9
+datamule-1.2.3.dist-info/RECORD,,