fix investor parsing logic for new cas formats

codereverser · codereverser · commit 7e7ee3f228b9 · 2023-09-11T09:01:56.000+05:30
diff --git a/casparser/parsers/mupdf.py b/casparser/parsers/mupdf.py
@@ -129,7 +129,7 @@ def parse_investor_info(page_dict, page_rect: fitz.Rect) -> InvestorInfo:
                 else:
                     if (
                         re.search(
-                            r"Date\s+Transaction|Folio\s+No|^Date\s*$",
+                            r"Mutual\s+Fund|Date\s+Transaction|Folio\s+No|^Date\s*$",
                             txt,
                             re.I | re.MULTILINE,
                         )
diff --git a/casparser/parsers/pdfminer.py b/casparser/parsers/pdfminer.py
@@ -44,7 +44,11 @@ def parse_investor_info(layout, width, height) -> InvestorInfo:
             name = txt
         else:
             if (
-                re.search(r"Date\s+Transaction|Folio\s+No|^Date\s*$", txt, re.I | re.MULTILINE)
+                re.search(
+                    r"Portfolio\s+Summary|Mutual\s+Fund|Date\s+Transaction|Folio\s+No|^Date\s*$",
+                    txt,
+                    re.I | re.MULTILINE,
+                )
                 or mobile is not None
             ):
                 return InvestorInfo(

Original file line number	Diff line number	Diff line change
`@@ -129,7 +129,7 @@ def parse_investor_info(page_dict, page_rect: fitz.Rect) -> InvestorInfo:`
`129`	`129`	`else:`
`130`	`130`	`if (`
`131`	`131`	`re.search(`
`132`		`- r"Date\s+Transaction\|Folio\s+No\|^Date\s*$",`
	`132`	`+ r"Mutual\s+Fund\|Date\s+Transaction\|Folio\s+No\|^Date\s*$",`
`133`	`133`	`txt,`
`134`	`134`	`re.I \| re.MULTILINE,`
`135`	`135`	`)`