modified mapping dict for 8k

datamule/datamule/mapping_dicts/txt_mapping_dicts.py

@@ -30,9 +30,11 @@
 }
 
 item_pattern_mapping = r"^\n\n\s*(ITEM|Item)\s+(\d+[a-zA-Z]?|ONE|TWO|THREE|FOUR|FIVE|SIX|SEVEN|EIGHT|NINE|TEN|ELEVEN|TWELVE|THIRTEEN|FOURTEEN|FIFTEEN|SIXTEEN|[0-9]+[a-zA-Z]?)\.?"
+item_pattern_mapping_8k = r"^\n\n\s*(ITEM|Item)\s+(\d+(?:\.\d+)?[a-zA-Z]?|ONE|TWO|THREE|FOUR|FIVE|SIX|SEVEN|EIGHT|NINE|TEN|ELEVEN|TWELVE|THIRTEEN|FOURTEEN|FIFTEEN|SIXTEEN|[0-9]+[a-zA-Z]?)\.?"
 part_pattern_mapping = r"^\n\n\s*(PART|Part)\s+(?:I{1,3}|IV)\.?"
 
 item_pattern_standardization = r"^\s*(?:ITEM|Item)\s+(\d+[a-zA-Z]?|ONE|TWO|THREE|FOUR|FIVE|SIX|SEVEN|EIGHT|NINE|TEN|ELEVEN|TWELVE|THIRTEEN|FOURTEEN|FIFTEEN|SIXTEEN|[0-9]+[a-zA-Z]?)\.?"
+item_pattern_standardization_8k = r"^\s*(?:ITEM|Item)\s+(\d+(?:\.\d+)?[a-zA-Z]?|ONE|TWO|THREE|FOUR|FIVE|SIX|SEVEN|EIGHT|NINE|TEN|ELEVEN|TWELVE|THIRTEEN|FOURTEEN|FIFTEEN|SIXTEEN)\.?"
 part_pattern_standardization =  r"^\s*(?:PART|Part)\s+([IVX]+)"
 
 
@@ -194,7 +196,7 @@
             {
                 "type": "hierarchy",
                 "name": "item",
-                "pattern": item_pattern_mapping,
+                "pattern": item_pattern_mapping_8k,
                 "hierarchy": 0
             },
             ])
@@ -204,7 +206,7 @@
         "type": "standardize", 
         "match": {
             "type": "item",
-            "text_pattern": item_pattern_standardization
+            "text_pattern": item_pattern_standardization_8k
         },
         "output": {
             "format": "item{}",

datamule/docs/source/changelog.rst

@@ -1,7 +1,8 @@
 Changelog
 =========
-v01.0.3 2025-02-10
+v01.0.3
 - modified item mapping dict regex to be more robust.
+- modified Monitor().monitor_submissions() to have a start_date parameter.
 
 v01.0.2 2025-02-06
 - modified .parse() to return dictionary in previous package format.

examples/earnings_report.py

@@ -0,0 +1,32 @@
+# Download earnings announcements for a given date range and save them to a new folder.
+
+from datamule import Portfolio
+from pathlib import Path
+import shutil
+
+portfolio = Portfolio('8K')
+
+portfolio.download_submissions(submission_type='8-K',filing_date=('2020-01-01','2020-01-31'))
+
+def process_submission(submission):
+    try:
+        for document in submission.document_type(['8-K']):
+            document.parse()
+
+            if 'item2.02' in document.data['document'].keys():
+                for document in submission.document_type(['EX-99.1']):
+                    return document.path
+    except:
+        return None
+
+
+paths = portfolio.process_submissions(process_submission)
+paths = [p for p in paths if p is not None]
+
+new_folder = Path('earnings_announcements')
+new_folder.mkdir(exist_ok=True)
+
+for path in paths:
+    new_path = new_folder / path.name
+    shutil.copy(path, new_path)  # Copy the file
+