Update app/spec_builder.py

app/spec_builder.py

@@ -1,79 +1,71 @@
 import re
-from typing import List
 from .schemas import ParserSpec, ColumnRole
 
 
 def infer_parser_spec(text: str) -> ParserSpec:
     lower = text.lower()
 
-    # layout detection
-    if re.search(r"^\d{4}-\d{2}-\d{2}t\d{2}:", text, re.MULTILINE):
+    # =========================
+    # EVENT STREAM (PalmPay)
+    # =========================
+    if re.search(r"\d{2}/\d{2}/\d{4}", text) and re.search(r"[+-]\d+\.\d{2}", text):
+        return ParserSpec(
+            document_type="bank_statement",
+            layout_type="event_stream",
+            row_start_pattern="line_contains_datetime",
+            row_continuation_rule="merge_until_next_datetime",
+            amount_style="signed_amount",
+            direction_signal="sign",
+            date_format="dd/mm/yyyy",
+            confidence=0.7,
+            notes=["Detected mobile wallet / PalmPay style"]
+        )
+
+    # =========================
+    # EVENT BLOCK (Moniepoint)
+    # =========================
+    if re.search(r"\d{4}-\d{2}-\d{2}T\d{2}:", text):
         return ParserSpec(
             document_type="bank_statement",
             layout_type="event_block",
             row_start_pattern="line_starts_with_datetime",
-            row_continuation_rule="append_until_next_row_start",
+            row_continuation_rule="merge_until_amount_line",
             amount_style="debit_credit_balance",
             direction_signal="keywords",
-            date_format="yyyy-mm-dd",
-            currency_markers=find_currency_markers(text),
-            columns=[
-                ColumnRole(role="transaction_date", evidence="datetime row start"),
-                ColumnRole(role="description", evidence="middle lines"),
-                ColumnRole(role="debit", evidence="amount triplet"),
-                ColumnRole(role="credit", evidence="amount triplet"),
-                ColumnRole(role="balance", evidence="last amount in block"),
-            ],
-            summary_fields=infer_summary_fields(text),
-            confidence=0.88,
-            notes=["Detected event-block statement structure"],
+            confidence=0.8,
+            notes=["Detected Moniepoint-style event block"]
         )
 
-    if re.search(r"^\d{2}/\d{2}/\d{2}\s+", text, re.MULTILINE):
+    # =========================
+    # TABLE (GT / UBA / Kuda)
+    # =========================
+    if re.search(r"\d{2}-[A-Za-z]{3}-\d{4}", text):
         return ParserSpec(
             document_type="bank_statement",
             layout_type="table",
-            row_start_pattern="line_starts_with_slash_date",
+            row_start_pattern="line_starts_with_two_dates",
             row_continuation_rule="append_until_next_row_start",
-            amount_style="amount_balance",
-            direction_signal="keywords",
-            date_format="dd/mm/yy",
-            currency_markers=find_currency_markers(text),
-            columns=[
-                ColumnRole(role="transaction_date", evidence="dd/mm/yy row start"),
-                ColumnRole(role="direction", evidence="keywords like inward/outward/web/local"),
-                ColumnRole(role="amount", evidence="currency amount inline"),
-                ColumnRole(role="description", evidence="middle text"),
-                ColumnRole(role="balance", evidence="ending currency amount"),
-            ],
-            summary_fields=infer_summary_fields(text),
-            confidence=0.84,
-            notes=["Detected slash-date multiline table structure"],
+            amount_style="debit_credit_balance",
+            direction_signal="separate_columns",
+            confidence=0.8,
+            notes=["Detected table-style statement"]
         )
 
-    if re.search(r"^\d{2}-[A-Za-z]{3}-\d{4}\s+\d{2}-[A-Za-z]{3}-\d{4}", text, re.MULTILINE):
+    if re.search(r"\d{2}/\d{2}/\d{2}", text):
         return ParserSpec(
             document_type="bank_statement",
             layout_type="table",
-            row_start_pattern="line_starts_with_two_dates",
+            row_start_pattern="line_starts_with_slash_date",
             row_continuation_rule="append_until_next_row_start",
-            amount_style="debit_credit_balance",
-            direction_signal="separate_columns",
-            date_format="dd-MMM-yyyy",
-            currency_markers=find_currency_markers(text),
-            columns=[
-                ColumnRole(role="transaction_date", evidence="first date"),
-                ColumnRole(role="value_date", evidence="second date"),
-                ColumnRole(role="description", evidence="wrapped narration"),
-                ColumnRole(role="debit", evidence="numeric amount column"),
-                ColumnRole(role="credit", evidence="numeric amount column"),
-                ColumnRole(role="balance", evidence="last numeric column"),
-            ],
-            summary_fields=infer_summary_fields(text),
-            confidence=0.90,
-            notes=["Detected classic bank statement table"],
+            amount_style="amount_balance",
+            direction_signal="keywords",
+            confidence=0.6,
+            notes=["Detected short date format table"]
         )
 
+    # =========================
+    # FALLBACK
+    # =========================
     return ParserSpec(
         document_type="unknown",
         layout_type="unknown",
@@ -81,52 +73,6 @@ def infer_parser_spec(text: str) -> ParserSpec:
         row_continuation_rule="unknown",
         amount_style="unknown",
         direction_signal="unknown",
-        date_format=None,
-        currency_markers=find_currency_markers(text),
-        columns=[],
-        summary_fields=infer_summary_fields(text),
-        confidence=0.20,
-        notes=["Could not confidently infer parser grammar"],
-    )
-
-
-def find_currency_markers(text: str) -> List[str]:
-    markers = []
-    if "₦" in text:
-        markers.append("₦")
-    if "ngn" in text.lower():
-        markers.append("NGN")
-    if "$" in text:
-        markers.append("$")
-    return markers
-
-
-def infer_summary_fields(text: str) -> dict:
-    lower = text.lower()
-    result = {}
-
-    if "opening balance" in lower:
-        result["opening_balance"] = "Opening Balance"
-    if "closing balance" in lower:
-        result["closing_balance"] = "Closing Balance"
-    if "total debit" in lower or "money out" in lower:
-        result["total_debit"] = "Total Debit/Money out"
-    if "total credit" in lower or "money in" in lower:
-        result["total_credit"] = "Total Credit/Money in"
-    if "account number" in lower or "account no" in lower:
-        result["account_number"] = "Account Number/Account No"
-
-    return result
-
-
-if profile.layout_type == "event_stream":
-    return ParserSpec(
-        document_type="bank_statement",
-        layout_type="event_stream",
-        row_start_pattern="line_contains_datetime",
-        row_continuation_rule="merge_until_next_datetime",
-        amount_style="signed_amount",
-        direction_signal="sign",
-        confidence=0.7,
-        notes=["Detected PalmPay-style stream"]
+        confidence=0.0,
+        notes=["Could not infer structure"]
     )