Fix: sc13g period of report label lookup

edgar/attachments.py

@@ -1025,20 +1025,41 @@ def get_filing_dates(self)-> Optional[Tuple[str,str, Optional[str]]]:
         grouping_divs = self._soup.find_all("div", class_="formGrouping")
         if len(grouping_divs) == 0:
             return None
-        date_grouping_div = grouping_divs[0]
-        info_divs = date_grouping_div.find_all("div", class_="info")
-        filing_date = info_divs[0].text.strip()
-        accepted_date = info_divs[1].text.strip()
-
-        if len(grouping_divs) > 1:
-            period_grouping_div = grouping_divs[1]
-            first_info_div = period_grouping_div.find("div", class_="info")
-            if first_info_div:
-                period = first_info_div.text.strip()
-                result = filing_date, accepted_date, period
-                self._cached_filing_dates = result
-                return result
-        result = filing_date, accepted_date, None
+
+        # Build a label -> value map by pairing each <div class="infoHead">
+        # with the next sibling <div class="info"> under the same grouping.
+        # This is label-based rather than positional: on multi-filer Schedule
+        # 13D/G filings the Filer(s) block can appear at index 1, and a
+        # positional lookup would return its concatenated filer names as the
+        # period of report, which then crashes downstream isoformat parsers.
+        label_to_value: Dict[str, str] = {}
+        for grouping in grouping_divs:
+            children = [c for c in grouping.find_all("div", recursive=False)
+                        if c.get("class")]
+            current_label: Optional[str] = None
+            for child in children:
+                classes = child.get("class") or []
+                if "infoHead" in classes:
+                    current_label = child.text.strip().lower()
+                elif "info" in classes and current_label is not None:
+                    # Only keep the first value for each label.
+                    label_to_value.setdefault(current_label, child.text.strip())
+                    current_label = None
+
+        filing_date = label_to_value.get("filing date")
+        accepted_date = label_to_value.get("accepted")
+        period = label_to_value.get("period of report")
+
+        # Fall back to the legacy positional layout if the label-based lookup
+        # missed either of the always-present date fields.
+        if filing_date is None or accepted_date is None:
+            info_divs = grouping_divs[0].find_all("div", class_="info")
+            if filing_date is None and len(info_divs) >= 1:
+                filing_date = info_divs[0].text.strip()
+            if accepted_date is None and len(info_divs) >= 2:
+                accepted_date = info_divs[1].text.strip()
+
+        result = filing_date, accepted_date, period
         self._cached_filing_dates = result
         return result
 

tests/issues/regression/test_issue_sc13g_period_of_report.py

@@ -0,0 +1,115 @@
+"""
+Regression test for Schedule 13D/G multi-filer crash.
+
+FilingHomepage.get_filing_dates used to pick the "Period of Report" block
+by position (grouping_divs[1]). On multi-filer Schedule 13D/G filings, the
+Filer(s) block can appear at index 1 and its .text.strip() concatenates
+each filer's name with no delimiter. Downstream callers (e.g.
+sec-edgar-mcp) then blow up with "Invalid isoformat string: ...". The fix
+is to pick the block by matching its <div class="infoHead"> label instead
+of by position.
+"""
+
+import pytest
+from bs4 import BeautifulSoup
+
+from edgar.attachments import Attachments, FilingHomepage
+
+pytestmark = pytest.mark.fast
+
+
+def _homepage_from_html(html: str) -> FilingHomepage:
+    soup = BeautifulSoup(html, "html.parser")
+    # Attachments is irrelevant for these tests; pass an empty instance.
+    attachments = Attachments(document_files=[], data_files=[], primary_documents=[])
+    return FilingHomepage(url="http://example/index.html", soup=soup, attachments=attachments)
+
+
+_FILING_AND_ACCEPTED = """
+<div class="formGrouping">
+    <div class="infoHead">Filing Date</div>
+    <div class="info">2024-05-01</div>
+    <div class="infoHead">Accepted</div>
+    <div class="info">2024-05-01 16:30:00</div>
+</div>
+"""
+
+
+def test_multi_filer_does_not_return_concatenated_filer_names():
+    """When the Filer(s) block sits at grouping_divs[1], period should be None."""
+    html = f"""
+    <html><body>
+    {_FILING_AND_ACCEPTED}
+    <div class="formGrouping">
+        <div class="infoHead">Filer(s)</div>
+        <div class="info">FILER ONE</div>
+        <div class="info">FILER TWO, LLC</div>
+        <div class="info">FILER THREE</div>
+        <div class="info">FILER FOUR, INC.</div>
+    </div>
+    </body></html>
+    """
+    homepage = _homepage_from_html(html)
+    filing_date, accepted_date, period = homepage.get_filing_dates()
+
+    assert filing_date == "2024-05-01"
+    assert accepted_date == "2024-05-01 16:30:00"
+    assert period is None, f"Expected None, got {period!r}"
+
+
+def test_period_block_found_by_label_when_at_index_1():
+    """Traditional layout: grouping_divs[1] is Period of Report."""
+    html = f"""
+    <html><body>
+    {_FILING_AND_ACCEPTED}
+    <div class="formGrouping">
+        <div class="infoHead">Period of Report</div>
+        <div class="info">2024-03-31</div>
+    </div>
+    </body></html>
+    """
+    homepage = _homepage_from_html(html)
+    filing_date, accepted_date, period = homepage.get_filing_dates()
+
+    assert filing_date == "2024-05-01"
+    assert accepted_date == "2024-05-01 16:30:00"
+    assert period == "2024-03-31"
+
+
+def test_period_block_found_even_when_not_at_index_1():
+    """Label-based lookup should work regardless of position."""
+    html = f"""
+    <html><body>
+    {_FILING_AND_ACCEPTED}
+    <div class="formGrouping">
+        <div class="infoHead">Filer(s)</div>
+        <div class="info">FILER ONE</div>
+        <div class="info">FILER TWO, LLC</div>
+    </div>
+    <div class="formGrouping">
+        <div class="infoHead">Period of Report</div>
+        <div class="info">2024-03-31</div>
+    </div>
+    </body></html>
+    """
+    homepage = _homepage_from_html(html)
+    filing_date, accepted_date, period = homepage.get_filing_dates()
+
+    assert filing_date == "2024-05-01"
+    assert accepted_date == "2024-05-01 16:30:00"
+    assert period == "2024-03-31"
+
+
+def test_no_period_block_returns_none_for_period():
+    """No Period of Report block at all -> period is None."""
+    html = f"""
+    <html><body>
+    {_FILING_AND_ACCEPTED}
+    </body></html>
+    """
+    homepage = _homepage_from_html(html)
+    filing_date, accepted_date, period = homepage.get_filing_dates()
+
+    assert filing_date == "2024-05-01"
+    assert accepted_date == "2024-05-01 16:30:00"
+    assert period is None