From 0923337fffb72f17736d37d973725a66974d4af3 Mon Sep 17 00:00:00 2001
From: Carlos Escalante <cescalante2988@gmail.com>
Date: Wed, 1 Apr 2026 10:09:51 -0600
Subject: [PATCH] Fix pension paste parser for split-line format from bank
 website

The bank website puts field labels and amounts on separate lines.
Parser now handles both inline and split formats.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 frontend/src/lib/parsePensionPaste.ts | 90 ++++++++++++++++++++-------
 1 file changed, 68 insertions(+), 22 deletions(-)

diff --git a/frontend/src/lib/parsePensionPaste.ts b/frontend/src/lib/parsePensionPaste.ts
index a890ddc..6882ff6 100644
--- a/frontend/src/lib/parsePensionPaste.ts
+++ b/frontend/src/lib/parsePensionPaste.ts
@@ -14,14 +14,12 @@ export interface PensionParsedEntry {
 }
 
 function parseAmount(raw: string): number {
-  // "¢ 18,684,764.98" or "¢ -552,213.24" or just "18,684,764.98"
   const cleaned = raw.replace(/[¢\s]/g, '').replace(/,/g, '');
   const num = parseFloat(cleaned);
   return isNaN(num) ? 0 : num;
 }
 
 function parseDateDMY(raw: string): string {
-  // "01/03/2026" → "2026-03-01"
   const m = raw.match(/(\d{2})\/(\d{2})\/(\d{4})/);
   if (!m) return '';
   return `${m[3]}-${m[2]}-${m[1]}`;
@@ -33,6 +31,17 @@ function extractAmounts(line: string): number[] {
   return matches.map(parseAmount);
 }
 
+// Field labels in the order they appear in the bank statement
+const FIELD_LABELS: [RegExp, string][] = [
+  [/saldo\s*anterior/i, 'saldo_anterior'],
+  [/aportes/i, 'aportes'],
+  [/rendimientos/i, 'rendimientos'],
+  [/retiros/i, 'retiros'],
+  [/traslados/i, 'traslados'],
+  [/comisi[oó]n/i, 'comision'],
+  [/bonificaci[oó]n/i, 'bonificacion'],
+];
+
 interface BlockResult {
   funds: string[];
   fields: Record<string, number[]>;
@@ -60,32 +69,69 @@ function parseBlock(lines: string[]): BlockResult | null {
     return null;
   }
 
-  const fieldMap: [RegExp, string][] = [
-    [/saldo\s*anterior/i, 'saldo_anterior'],
-    [/aportes/i, 'aportes'],
-    [/rendimientos/i, 'rendimientos'],
-    [/retiros/i, 'retiros'],
-    [/traslados/i, 'traslados'],
-    [/comisi[oó]n/i, 'comision'],
-    [/bonificaci[oó]n/i, 'bonificacion'],
-    [/saldo\s*actual/i, 'saldo_final'],
-  ];
+  // Strategy 1: Try same-line parsing (label + amounts on same line)
+  // Strategy 2: Collect standalone amount lines for split-format parsing
+  const detectedFieldOrder: string[] = [];
+  const standaloneAmounts: number[] = [];
 
   for (const line of lines) {
-    for (const [regex, key] of fieldMap) {
-      if (regex.test(line)) {
-        const amounts = extractAmounts(line);
-        if (amounts.length > 0) {
-          result.fields[key] = amounts;
-        }
-      }
-    }
-
-    // Period
+    // Check for period
     const periodMatch = line.match(/del\s+(\d{2}\/\d{2}\/\d{4})\s+al\s+(\d{2}\/\d{2}\/\d{4})/i);
     if (periodMatch) {
       result.period_start = parseDateDMY(periodMatch[1]);
       result.period_end = parseDateDMY(periodMatch[2]);
+      continue;
+    }
+
+    // Check for "Saldo Actual" line (always has amounts inline)
+    if (/saldo\s*actual/i.test(line)) {
+      const amounts = extractAmounts(line);
+      if (amounts.length > 0) {
+        result.fields['saldo_final'] = amounts;
+      }
+      continue;
+    }
+
+    // Check if this line matches a field label
+    let matchedLabel = false;
+    for (const [regex, key] of FIELD_LABELS) {
+      if (regex.test(line)) {
+        matchedLabel = true;
+        const amounts = extractAmounts(line);
+        if (amounts.length > 0) {
+          // Strategy 1: amounts on same line as label
+          result.fields[key] = amounts;
+        } else {
+          // Strategy 2: label-only line, record the order
+          detectedFieldOrder.push(key);
+        }
+        break;
+      }
+    }
+
+    // If not a label line, check if it's a standalone amount line
+    if (!matchedLabel) {
+      const amounts = extractAmounts(line);
+      if (amounts.length === 1) {
+        standaloneAmounts.push(amounts[0]);
+      }
+    }
+  }
+
+  // If we have standalone amounts and field labels, map them
+  // Format: N labels, then N amounts for fund1, then N amounts for fund2, ...
+  if (detectedFieldOrder.length > 0 && standaloneAmounts.length > 0) {
+    const numFields = detectedFieldOrder.length;
+    const numFunds = result.funds.length;
+
+    if (standaloneAmounts.length >= numFields * numFunds) {
+      for (let f = 0; f < numFunds; f++) {
+        for (let i = 0; i < numFields; i++) {
+          const key = detectedFieldOrder[i];
+          if (!result.fields[key]) result.fields[key] = [];
+          result.fields[key].push(standaloneAmounts[f * numFields + i]);
+        }
+      }
     }
   }