Added all of chat-topics, persona;-history, projects, receipts and vehicle. This is the knowledge base section after all....

receipts/README.md

@@ -0,0 +1,55 @@
+# Receipts Archive
+
+Local receipt/warranty archive for Michael.
+
+## Structure
+
+- `receipts.sqlite` — searchable SQLite database.
+- `images/YY-MM-images/` — original receipt/warranty images, grouped by month.
+
+## Current database tables
+
+- `receipts` — one row per receipt/document image, including receipt-level `notes`.
+- `receipt_items` — line items linked to `receipts.id`.
+
+## Suggested conventions
+
+- `category` is broad and filterable, e.g. `Home purchases`, `Business expenses`, `Vehicle`, `Tools`, `Medical`, `Travel`.
+- `doc_type` distinguishes `receipt`, `invoice`, `warranty_certificate`, `quote`, etc.
+- Keep images immutable; fix OCR/database fields rather than replacing the source image.
+- Use `image_sha256` to avoid duplicate imports.
+
+## Local import/OCR wrapper
+
+```bash
+# Dry-run OCR/parse without touching the DB
+knowledge/receipts/scripts/receipt_import.py /path/to/receipt.jpg --dry-run --json
+
+# Import a new receipt image/PDF into the archive and DB
+knowledge/receipts/scripts/receipt_import.py /path/to/receipt.jpg --category "Home purchases" --notes "Warranty for office"
+
+# Re-OCR/update an existing DB row
+knowledge/receipts/scripts/receipt_import.py knowledge/receipts/images/26-05-images/example.jpg --receipt-id 2 --category "Personal care"
+```
+
+The wrapper keeps originals immutable, writes OCR derivatives under `knowledge/receipts/ocr/<sha-prefix>/`, uses local Tesseract/OCRmyPDF only, and stores parsed fields plus raw OCR text in `receipts.sqlite`.
+
+## Useful queries
+
+```sql
+-- Recent records
+SELECT id, category, doc_type, merchant_name, receipt_date, total, image_path
+FROM receipts
+ORDER BY COALESCE(receipt_date, captured_at) DESC;
+
+-- Filter by category
+SELECT id, merchant_name, receipt_date, total, image_path
+FROM receipts
+WHERE category = 'Home purchases';
+
+-- Search items/text
+SELECT r.id, r.category, r.merchant_name, i.description, i.sku, r.image_path
+FROM receipts r
+JOIN receipt_items i ON i.receipt_id = r.id
+WHERE i.description LIKE '%powerbank%' OR r.ocr_raw_text LIKE '%powerbank%';
+```

receipts/ocr/4eba4bf8f3eaf083/ocr_raw.txt

@@ -0,0 +1,123 @@
+--- OCR original.png psm=6 ---
+é AliExpress Qgare#
+Orders ve
+Viewall Topay(0) Toship(2) Shipped(0) Process
+To ship May 23, 2026 >
+vChoice| SPEEDRACING Official Store >
+Ag PC USB Sequential Shifter Gear Shifter For Logitech ...
+Clamp
+| Bingen
+AUS1 coupon if delayed
+citty
+Free returns
+Total:AU$ 25.80
+PC USB Sequential Shifter Gear Shifter For Logitech ...
+black
+Hit AUS 83.69 xi
++2 AUS1 coupon if delayed
+“ Free returns
+Total:AU$ 92.06
+--- OCR original.png psm=11 ---
+AliExpress
+
+Orders
+
+Qa
+
+o°
+
+View all
+
+To pay (0)
+
+To ship (2)
+
+Shipped (0)
+
+Process
+
+—=
+
+May 23, 2026 >
+
+To ship
+
+e
+
+AUS 23.32 x1
+
+aaae
+
+Total:AUS$ 25.80
+
+AUS 83.69 x1
+
+<
+
+Total:AUS 92.06
+--- OCR gray_contrast_2x.png psm=6 ---
+< AliExpress Q Q oy)
+Orders oe
+Viewall Topay(0) Toship(2) Shipped(0O) Process
+V¥Choice| SPEEDRACING Official Store >
+oss,
+sore AUS 83.69 x1
+--- OCR gray_contrast_2x.png psm=11 ---
+
+Q 8
+
+°°
+
+=
+
+Total:AUS 25.80
+
+——~
+
+boee
+
+--- OCR threshold_2x.png psm=6 ---
+¢ AliExpress agar
+Orders oo
+VChoice | SPEEDRACING Official Store >
+8068,
+ors AUS 83.69 x1
+Total:AUS$ 92.06
+--- OCR threshold_2x.png psm=11 ---
+
+Q &
+
+R
+
+qn»
+
+vVChoice| SPEEDRACING Official Store >
+
+6000
+
+ttt
+
+--- OCR crop_1_2x.png psm=6 ---
+< Orders Q ran R
+--- OCR crop_1_2x.png psm=11 ---
+
+RF
+
+[ss
+
+--- OCR crop_2_2x.png psm=6 ---
+————- ll i
+--- OCR crop_2_2x.png psm=11 ---
+
+--- OCR crop_3_2x.png psm=6 ---
+| PC USB Sequential Shifter Gear Shifter For Logitech ...
+--- OCR crop_3_2x.png psm=11 ---
+
+|
+
+--- OCR crop_4_2x.png psm=6 ---
+erat AUS 83.69 x1
+Edit address
+--- OCR crop_4_2x.png psm=11 ---
+
+seoe

receipts/ocr/4eba4bf8f3eaf083/ocr_runs.json

@@ -0,0 +1,86 @@
+[
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/original.png",
+    "psm": 6,
+    "score": 715,
+    "chars": 465
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/original.png",
+    "psm": 11,
+    "score": 694,
+    "chars": 494
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/gray_contrast_2x.png",
+    "psm": 6,
+    "score": 665,
+    "chars": 465
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/gray_contrast_2x.png",
+    "psm": 11,
+    "score": 649,
+    "chars": 499
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/threshold_2x.png",
+    "psm": 6,
+    "score": 712,
+    "chars": 462
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/threshold_2x.png",
+    "psm": 11,
+    "score": 703,
+    "chars": 503
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_1_2x.png",
+    "psm": 6,
+    "score": 184,
+    "chars": 134
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_1_2x.png",
+    "psm": 11,
+    "score": 203,
+    "chars": 153
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_2_2x.png",
+    "psm": 6,
+    "score": 242,
+    "chars": 142
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_2_2x.png",
+    "psm": 11,
+    "score": 187,
+    "chars": 137
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_3_2x.png",
+    "psm": 6,
+    "score": 124,
+    "chars": 124
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_3_2x.png",
+    "psm": 11,
+    "score": 129,
+    "chars": 129
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_4_2x.png",
+    "psm": 6,
+    "score": 132,
+    "chars": 82
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/4eba4bf8f3eaf083/crop_4_2x.png",
+    "psm": 11,
+    "score": 126,
+    "chars": 76
+  }
+]

receipts/ocr/92677db8f57247de/ocr_raw.txt

@@ -0,0 +1,96 @@
+--- OCR original.png psm=6 ---
+Merchant AIAU MARKETS PTY LTD
+Transaction date 23 May 2026
+Order ID 202605238503130000436825
+8230069
+Deliver to Michael Manning
+Forsters Bay Road
+7
+Narooma NSW 2546
+Australia
+Track Package
+PC USB Sequential Sh... $25.80
+Qty: 1
+Product ID: 8211773665403987
+PC USB Sequential Sh... $92.06
+Product ID: 8211773665383987
+Subtotal $117.86
+Total $117.86 AUD
+Paid AIAU MARKETS PTY LTD with
+fase svIsA $117.86 AUD
+Ss Credit --4942
+This charge will appear on your credit card
+statement as "PAYPAL *AIAUMARKETS".
+Transaction ID: 4EA50033E33881234
+--- OCR original.png psm=11 ---
+Merchant
+
+AIAU MARKETS PTY LTD
+
+Transaction date
+
+23 May 2026
+
+Order ID
+
+202605238503130000436825
+
+Deliver to
+
+Michael Manning
+
+$25.80
+
+PC USB Sequential Sh...
+
+$92.06
+
+Subtotal
+
+$117.86
+
+Total
+
+$117.86 AUD
+
+VISA
+
+Credit +-4942
+
+--- OCR gray_contrast_2x.png psm=6 ---
+
+VISA $117.86 AUD
+
+--- OCR gray_contrast_2x.png psm=11 ---
+
+=
+
+Credit +4942
+
+--- OCR threshold_2x.png psm=6 ---
+4 VISA $117.86 AUD
+--- OCR threshold_2x.png psm=11 ---
+
+Credit --4942
+
+--- OCR crop_1_2x.png psm=6 ---
+--- OCR crop_1_2x.png psm=11 ---
+
+--- OCR crop_2_2x.png psm=6 ---
+Dradiict IN- 2911772RASANZOR7
+--- OCR crop_2_2x.png psm=11 ---
+
+Drodiict IN: 28911772ARA5SANZORT
+--- OCR crop_3_2x.png psm=6 ---
+
+--- OCR crop_3_2x.png psm=11 ---
+ee er el
+
+—
+
+eet ete ee Se ee es Oa
+
+--- OCR crop_4_2x.png psm=6 ---
+--- OCR crop_4_2x.png psm=11 ---
+
+Credit +*-4942

receipts/ocr/92677db8f57247de/ocr_runs.json

@@ -0,0 +1,86 @@
+[
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/original.png",
+    "psm": 6,
+    "score": 992,
+    "chars": 542
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/original.png",
+    "psm": 11,
+    "score": 1015,
+    "chars": 565
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/gray_contrast_2x.png",
+    "psm": 6,
+    "score": 990,
+    "chars": 540
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/gray_contrast_2x.png",
+    "psm": 11,
+    "score": 1016,
+    "chars": 566
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/threshold_2x.png",
+    "psm": 6,
+    "score": 984,
+    "chars": 534
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/threshold_2x.png",
+    "psm": 11,
+    "score": 1014,
+    "chars": 564
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_1_2x.png",
+    "psm": 6,
+    "score": 214,
+    "chars": 164
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_1_2x.png",
+    "psm": 11,
+    "score": 224,
+    "chars": 174
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_2_2x.png",
+    "psm": 6,
+    "score": 141,
+    "chars": 91
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_2_2x.png",
+    "psm": 11,
+    "score": 149,
+    "chars": 99
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_3_2x.png",
+    "psm": 6,
+    "score": 403,
+    "chars": 103
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_3_2x.png",
+    "psm": 11,
+    "score": 445,
+    "chars": 145
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_4_2x.png",
+    "psm": 6,
+    "score": 225,
+    "chars": 175
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/92677db8f57247de/crop_4_2x.png",
+    "psm": 11,
+    "score": 235,
+    "chars": 185
+  }
+]

receipts/ocr/ae6c6d9927838a7c/ocr_raw.txt

@@ -0,0 +1,194 @@
+--- OCR original.png psm=6 ---
+1:00 O©4ZQ4H° WY 8: 529 He at 66%
+3 CE |
+i vy Receipt for Your Payment to GoDaddy 0...
+"PayPal" <service@paypal.com.au> 7 - =
+GoDaddy Online
+°
+Servi...
+Merchant GoDaddy Online Servi...
++61 0280428920
+Transaction date 14 May 2026
+Order ID 4071702979_AUD_2415
+View Payment Details
+
+{ts ]
+
+-AU (.NET.AU) Domain... $21.95
+
+Qty: 1
+
+Subtotal $21.95
+
+Tax $2.20
+
+Total $24.15 AUD
+
+See
+
+~ 00OOGOE®
+© https://sm.mmanningau.net/# g A (&) :
+--- OCR original.png psm=11 ---
+5.00
+
+1:00 ©FQASK
+
+B-
+
+8
+
+vas
+
+66%
+
+1 + Receipt for Your Payment to ame fo)
+
+"PayPal" <service@paypal.com.au> 7
+
+Merchant
+
+GoDaddy Online Servi
+
+Transaction date
+
+14 May 2026
+
+Order ID
+
+4071702979_AUD_2415
+
+lets
+
+$21.95
+
+AU (.NET.AU) Domain.
+
+Subtotal
+
+Tax
+
+$2.20
+
+Total
+
+$24.15 AUD
+
+26
+
+© https://sm.mmanningau.net/# g A (&)
+--- OCR gray_contrast_2x.png psm=6 ---
+1:00 O4ZQ4BH * AY B= 3:99 SZ te at! 66%
+3 CE [a+ |
+1 YY Receipt for Your Payment to GoDaddy 0... x
+"PayPal" < service@paypal.com.au> # -« =
+e
+
+| [ents |
+
+aA se oe Ge -©
+@ hitps://sm.mmanningau.net/# 7 A (23) :
+--- OCR gray_contrast_2x.png psm=11 ---
+1:00 © 4Q4B °
+
+WY B= $99 ZS te atl! 66%
+
+1 Yy Receipt for Your Payment to soe 0
+
+x
+
+—
+
+"PayPal" < service@paypal.com.au> #7
+
+<5 A
+
+(tests
+
+AU (.NET.AU) Domain
+
+aN
+
+“@
+
+@ https://sm.mmanningau.net/# 7 A (23)
+--- OCR threshold_2x.png psm=6 ---
+1:00 ©4Q4B « A B= 5:29 S: te al 66%
+3 CSE Ex
+1 v Receipt for Your Payment to GoDaddy 0... x
+“PayPal” < service@paypal.com.au> 7 eo =
+
+Llets
+
+~ se oe Ye -&
+@ fites://sm.mmanningau.net/# g A ®& :
+--- OCR threshold_2x.png psm=11 ---
+1:00 ©4@Q4BH °
+
+& %:
+
+5.00 =
+
+tt ail 66%
+
+Kas oe
+
+1 Yy Receipt for Your Payment to Goaddy 0 Oo
+
+“PayPal” < service@paypal.com.au> 7
+
+“py.
+
+Lite
+
+a
+
+¢
+
+© bttps://sm.mmanningau.net/# g A &
+--- OCR crop_1_2x.png psm=6 ---
+1:00 O©4Q4B » RY B= $99 ZR Me at 66%
+3 CE
+i YY Receipt for Your Payment to GoDaddy 0... x
+"PayPal" < service@paypal.com.au> ¥ * =
+--- OCR crop_1_2x.png psm=11 ---
+1:00 © 4Q4BH
+
+WY 8: 5:99 ZS te at 66%
+
+xo} otat<|>
+
+1 YY Receipt for Your Payment to omees* O...
+
+"PayPal" < service@paypal.com.au> %
+
+--- OCR crop_2_2x.png psm=6 ---
+--- OCR crop_2_2x.png psm=11 ---
+
+GoDaddy Online Servi...
+
+--- OCR crop_3_2x.png psm=6 ---
+--- OCR crop_3_2x.png psm=11 ---
+pis
+
+-AU (.NET.AU) Domain...
+
+§21.95
+
+=
+
+es
+
+--- OCR crop_4_2x.png psm=6 ---
+1ax 92.20
+~ 0 oe Ge -@
+@ hitps://sm.mmanningau.net/# g A (23) :
+--- OCR crop_4_2x.png psm=11 ---
+1ax
+
+92.20
+
+NN
+
+- «© 0 » Ge -@
+
+@ https://sm.mmanningau.net/# g A &)

receipts/ocr/ae6c6d9927838a7c/ocr_runs.json

@@ -0,0 +1,86 @@
+[
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/original.png",
+    "psm": 6,
+    "score": 978,
+    "chars": 428
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/original.png",
+    "psm": 11,
+    "score": 886,
+    "chars": 386
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/gray_contrast_2x.png",
+    "psm": 6,
+    "score": 995,
+    "chars": 445
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/gray_contrast_2x.png",
+    "psm": 11,
+    "score": 1037,
+    "chars": 437
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/threshold_2x.png",
+    "psm": 6,
+    "score": 970,
+    "chars": 420
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/threshold_2x.png",
+    "psm": 11,
+    "score": 927,
+    "chars": 427
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_1_2x.png",
+    "psm": 6,
+    "score": 303,
+    "chars": 153
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_1_2x.png",
+    "psm": 11,
+    "score": 259,
+    "chars": 159
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_2_2x.png",
+    "psm": 6,
+    "score": 226,
+    "chars": 126
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_2_2x.png",
+    "psm": 11,
+    "score": 160,
+    "chars": 110
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_3_2x.png",
+    "psm": 6,
+    "score": 253,
+    "chars": 53
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_3_2x.png",
+    "psm": 11,
+    "score": 222,
+    "chars": 72
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_4_2x.png",
+    "psm": 6,
+    "score": 180,
+    "chars": 80
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/ae6c6d9927838a7c/crop_4_2x.png",
+    "psm": 11,
+    "score": 187,
+    "chars": 87
+  }
+]

receipts/ocr/b0c012b61bd94d08/ocr_raw.txt

@@ -0,0 +1,270 @@
+--- OCR original.png psm=6 ---
+ai7am 39 9s ee @
+thes | fmf trist: | @ Aust | © Code | MWe | Ri tran: | GD (ec x) @NoP | @nop | +
+@ < > GB nopong.comau/check-out/orderreceivea/siesize%7 QO A WH }
+= NO PONG oe
+CHECKOUT
+© THANK YOU! YOUR ORDER HAS BEEN RECEIVED.
+ORDER NUMBER: DATE:
+3465186 May 14, 2026
+EMAIL: TOTAL:
+michael@mmanningau.net $99.98
+PAYMENT METHOD:
+PayPal
+ORDER DETAILS
+PRODUCT TOTAL
+LUSCIOUS LIME BICARB FREE 85G x 2 $29.90
+BICARB FREE, LOW FRAGRANCE 85 x 4 $59.80
+Subtotal: $89.70
+Points Redeemed: ~$3.14. 314 points
+Shipping: $13.42 via Parcel Post
+Payment method: PayPal
+Total: $99.98 (includes $9.09 GST)
+Points: You've earned 433 points
+a) ka)
+BILLING ADDRESS SHIPPING ADDRESS
+Michael Manning Michael Manning
+7 Forsters Bay Road 7 Forsters Bay Rood
+Narooma New South Wales 2546 Narooma New South Wales 2546
+& +erareat2207
+BH weno @+1iry @
+--- OCR original.png psm=11 ---
+si7am 39 9 a:
+
+ee @ @
+
+= the
+
+fi Tris:
+
+@ Aus:
+
+| © coos
+
+| wet
+
+| Biter
+
+@NoF
+
+op |
+
+co
+
+@
+
+<
+
+>
+
+e@
+
+oS
+
+nopong.com.au/check-out/orderreceived/saesie6/7 1] Q@ A &
+
+—
+
+cd
+
+NO PONG
+
+@ THANK YOU! YOUR ORDER HAS BEEN RECEIVED.
+
+ORDER NUMBER:
+
+DATE:
+
+3465186
+
+May 14, 2026
+
+EMAIL:
+
+TOTAL:
+
+michael(@mmanningau.net
+
+$99.98
+
+PRODUCT
+
+TOTAL
+
+LUSCIOUS LIME BICARB FREE 85G x 2
+
+$29.90
+
+BICARB FREE, LOW FRAGRANCE 856 x 4
+
+$59.80
+
+Subtotal:
+
+$89.70
+
+Points Redeemed:
+
+~$3.4 314 points
+
+Shipping:
+
+$13.42 via Parcel Post
+
+$99.98 (includes $9.09 GsT)
+
+Points:
+
+You've earned 433 points
+
+i)
+
+BILLING ADDRESS
+
+SHIPPING ADDRESS
+
+Michael Manning
+
+7 Forsters Bay Road
+
+Narooma New South Wales 2546
+
+& +61418412227
+
+anor
+
+® +
+
+--- OCR gray_contrast_2x.png psm=6 ---
+817am 3B GF 9 Ge On| = OD Gwe)
+Wi thes | fm|trist | @ aust | © Code | Bwelc | KB tran EDlec x| @Nor | Q@Nop | +
+@ << > © _ & nopong.com.au/check-out/order-received/3465186/% [] g A ® :
+iv} THANK YOU! YOUR ORDER HAS BEEN RECEIVED.
+BICARB FREE, LOW FRAGRANCE 85G x 4 $59.80
+Points Redeemed: -$3.14 314 points
+ey w)
+7 Forsters Bay Road 7 Forsters Bay Road
+Q, +67418412227
+i Wwnes @+)y @
+--- OCR gray_contrast_2x.png psm=11 ---
+8:17am 3B GO 9 Gs
+
+{= © te
+
+®ci x
+
+@NoP | QNoP |
+
++
+
+Mi The | fl tis | @ Aust’ | © Code | MWele | Ki tren: | QD
+
+@ <
+
+©
+
+o
+
+=o
+
+nopong.com.au/check-out/order-received/3465186/% g A &
+
+NO PONG’
+
+eo
+
+i ceckour
+
+michael@mmanningau.net
+
+BICARB FREE, LOW FRAGRANCE 85G x 4
+
+-$3.14 314 points
+
+2
+
+4)
+
+ber
+
+wwe @ +) 5
+
+--- OCR threshold_2x.png psm=6 ---
+si7am 3B) 9 9 Bs on = @ oD
+Mi the: | fr trist: | @ aust | © code | wel | BB tran: | Gp | Qc x| @Nop | QNoP | +
+@ < > © & nopong.com.au/check-out/order-received/3465186/% [J g A ® :
+= @ NO PONG o
+iv) THANK YOU! YOUR ORDER HAS BEEN RECEIVED.
+PROOUCT TOTAL
+a) ~)
+ii wnuwosop® @+iy @
+--- OCR threshold_2x.png psm=11 ---
+8:17am 3B F 9 Es
+
+ee © aD
+
+Mi The | fmf trist: | @ Aust | © Code | Ml wei | BB Tran: | ap @c x) Gnor | QNor | +
+
+Cc
+
+~o
+
+nopong.com.au/check-out/order-received/3465186/% [J g A &
+
+@® NO PONG
+
+~$3.14 314 points
+
+a
+
+.
+
+weno @+¢! .
+
+]
+--- OCR crop_1_2x.png psm=6 ---
+817am 3H OF 9 ae on = OO iw)
+Wi The* | fj Trist | @Aust) | © Code | MP Wele | FATran: | 3 obs Me x] @wor | @nop | +
+@ < > © & nopong.com.au/check-out/order-received/3465186/% [) g A ® :
+= NO PONG’ 9
+--- OCR crop_1_2x.png psm=11 ---
+8:17am 3B 9 9 Gs
+
+on = OO Ge
+
+Wi Thes | fl tists | @ aus | © Code | HWele | Ki tren | ED
+
+@®ci x
+
+oO
+
+nopong.com.au/check-out/order-received/3465186/% [] @ A
+
+--- OCR crop_2_2x.png psm=6 ---
+--- OCR crop_2_2x.png psm=11 ---
+
+--- OCR crop_3_2x.png psm=6 ---
+Points Redeemed: -$3.14 374 points
+--- OCR crop_3_2x.png psm=11 ---
+
+Points Redeemed
+
+-$3.14 374 points
+
+Payment method
+
+--- OCR crop_4_2x.png psm=6 ---
+2) w)
+
+Hi Wwne™ @ +)» g
+--- OCR crop_4_2x.png psm=11 ---
+
+Ka)
+
+:
+
+@o2
+
+Ww “M fF
+
+Q® +

receipts/ocr/b0c012b61bd94d08/ocr_runs.json

@@ -0,0 +1,86 @@
+[
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/original.png",
+    "psm": 6,
+    "score": 1825,
+    "chars": 825
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/original.png",
+    "psm": 11,
+    "score": 1853,
+    "chars": 853
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/gray_contrast_2x.png",
+    "psm": 6,
+    "score": 1824,
+    "chars": 824
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/gray_contrast_2x.png",
+    "psm": 11,
+    "score": 1897,
+    "chars": 897
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/threshold_2x.png",
+    "psm": 6,
+    "score": 1848,
+    "chars": 848
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/threshold_2x.png",
+    "psm": 11,
+    "score": 1902,
+    "chars": 902
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_1_2x.png",
+    "psm": 6,
+    "score": 343,
+    "chars": 243
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_1_2x.png",
+    "psm": 11,
+    "score": 360,
+    "chars": 260
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_2_2x.png",
+    "psm": 6,
+    "score": 371,
+    "chars": 121
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_2_2x.png",
+    "psm": 11,
+    "score": 381,
+    "chars": 131
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_3_2x.png",
+    "psm": 6,
+    "score": 889,
+    "chars": 239
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_3_2x.png",
+    "psm": 11,
+    "score": 902,
+    "chars": 252
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_4_2x.png",
+    "psm": 6,
+    "score": 236,
+    "chars": 236
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/b0c012b61bd94d08/crop_4_2x.png",
+    "psm": 11,
+    "score": 259,
+    "chars": 259
+  }
+]

receipts/ocr/f1ef054a94e5632c/ocr_raw.txt

@@ -0,0 +1,423 @@
+--- OCR original.png psm=6 ---
+| 2UNIN INS ;
+a . a
+(Beg MANSe myiK
+‘ ABN 26 008 672 179 ae
+; Ph: (02) 4412-9600 =
+. Fri BS 02:33:44 PH ;
+| RETURNS/TNFO DESK RO? oe
+Sale eS a a
+ae TAX INVOTCE 4 5
+9311644072980 LIGHT DESK LAMP VERVE we
+sree SRO ic == UI
+9311644106489 PONE Be ANDAR CLICK mz.00 =
+B OUTLET WHT CLRPBD™ ae o
+3 @ SubTotal: $88.90 BX
+of Fe! IN THE TOTAL $56 a ee ay
+cap i 405221-942 ies :
+me) cheb
+-. | @hanse , sa.og fag
+: itr it Sey f
+a $7349 ROT P956 Cy 4221 #007-B6289-7 4-2 6-05-22 3
+TNT TAT TT
+i. Ue aaa ee |
+--- OCR original.png psm=11 ---
+Kvarchguse
+
+BATEMANSBAY
+
+asi Hage, ar
+
+Ph (ba) Hit 800
+
+7
+
+Fri 2
+
+2/0
+
+6/2026 02:33:44 PA
+
+RETURK
+
+:
+
+/IN DESK i)
+
+eat eee
+
+aR TAX” TNVOICE. RE
+
+9311644072980
+
+4
+
+H
+
+TGH
+
+50
+
+td
+
+ESK LAMP U
+
+LED AREN
+
+RUE
+
+$71.40
+
+9311644139799 4 OUT!
+
+WR BRD AR'
+
+tec
+
+$12.00
+
+9311644106463 POWE
+
+6 OUTLET Wi
+
+i
+
+TC
+
+BOA
+
+KPB
+
+DS
+
+TeAtaHD CLICK
+
+3d Subtotal:
+
+$88.90
+
+CLUDED IN THE TOTAL
+
+GST i
+
+405221 -942
+
+$88 *
+
+00
+
+Hg ea nee
+
+$0.00
+
+*" ig
+
+Flybuy
+
+4a)
+
+956 C|
+
+22
+
+TTT
+
+Hii
+
+"
+--- OCR gray_contrast_2x.png psm=6 ---
+a: 2 warehouse (a
+all aIBATEMANSH AY Wie :
+
+‘ ABN 26 0B 672 173 |
+
+b, Ph: (02) 4412 3500 |
+aoe Fri fo 1 ie
+“4 ~—RETURNS/INEU DESK RO?
+~~ 9311644072980 LIGHT DESK LAMP UERUE =
+2 1 a ee nh o
+“| —-9311644106463 rE | i ANDARD CLICK a
+
+| B OUTLET GHY CL RPBG |
+
+| 3 SubTotal: $88.90 |
+4 7 oY Fea aw rHe room $a “$8.08
+
+% ms 405221-942 is '
+--- OCR gray_contrast_2x.png psm=11 ---
+(5
+
+warehou € |
+
+ats Lani
+
+Ph! (02) Mas 9400
+
+Fri 22/0
+
+0/2026 02:33:44 Ph
+
+mt M
+
+|
+
+yt Ina iT
+
+=e
+
+EE TAX, "ft NVOICE
+
+931164407
+
+SCH oh B01
+
+ES
+
+L
+
+LA
+
+A
+
+Rev'f
+
+C
+
+RVE
+
+9311644139
+
+99 4 OUT
+
+R BRD AR
+
+9311644106
+
+6 OUTLET
+
+463 POW
+
+ae
+
+oe
+
+pppanoaio CLICK
+
+3 @ Subsotal:
+
+$5
+
+Af i fa, IN THE TOTAL
+
+CRED
+
+<
+
+405221-942
+
+ye
+
+oii
+
+“Ta
+--- OCR threshold_2x.png psm=6 ---
+‘is ion os ey
+4 Zi warehouse =~"
+So TBATEMARS RAY Mi
+a7 ABN 26 00B 672 179
+a Ph: (02) 4412 3600
+See of . ’ '
+4 Fri Ce |
+et] RETR TIL-OESK ROT
+ee ne TAX ENVOT CE}
+7 9311644072980 LTOHT DESK LAMP UERUE
+PR Me ki tulades
+| " 4p 93tteaat des ie i “non CLICK
+wae | f .3@ SubTotal: i
+é «| “ie. IN THE TOYA sq |
+
+‘3 KARR £ 405221-942 Save
+
+Bai
+AA
+--- OCR threshold_2x.png psm=11 ---
+%
+
+ze
+
+yt Fi
+
+L warehouse
+
+saat wane Aicie
+
+2 179
+
+Say
+
+Pr ‘t2) Mae 3500
+
+be
+
+Sos at
+
+a?
+
+Fri 72/05/2006 02:33:44
+
+AS? 7
+
+FET al DEK WO
+
+><
+
+_ EE Tax” ; NGOTCE
+
+a reef em
+
+fin’ ob Bop
+
+ESK LA
+
+Rev fe
+
+°
+
+93116441
+
+9799 4 OUT
+
+!
+
+ALEC
+
+6 0
+
+LE
+
+vi
+
+cw
+
+a CLICK
+
+3 @ SubTotal:
+
+;
+
+SE
+
+UDED IN THE rot
+
+cH
+
+405221-942_
+
+eyany
+
+vy
+
+Change SS
+
+Fini
+
+"Til iin
+--- OCR crop_1_2x.png psm=6 ---
+oat.
+Bs ‘y
+os i ~~ ‘
+BURNING
+- BATEMANSBAYWH
+--- OCR crop_1_2x.png psm=11 ---
+
+ee
+
+2Un NINGS
+
+/‘BATEMANSBAYWH
+
+BUNNINGS GROUP LTMITED
+--- OCR crop_2_2x.png psm=6 ---
+=) rr NBG O84 POSCS:«*d
+d HANES a
+we TaOPNVoTCE see
+--- OCR crop_2_2x.png psm=11 ---
+Ph: (02) 4412 3500
+
+ABN 26 U0G 672 179
+
+Fri 22/04
+
+0/2026 02:33:44.
+
+RET if Me UE A
+
+Ke TAX ? NVOTCE
+--- OCR crop_3_2x.png psm=6 ---
+~§ JITTO9HS0 72300 LIGHT UEOR LAMP VERVE in
+>. an RL ata Zl
+at | 931164410 is Pui i a a
+98 Subota: ; se
+r 7 od IN THE TOTAL $& ,
+| canna: 405221-942 $08.90
+--- OCR crop_3_2x.png psm=11 ---
+JTTO44U7
+
+Ah 9 5001
+
+AREN TC
+
+AVE
+
+931164413
+
+REC
+
+9311644106463 POW
+
+'
+
+STaNOARD CLICK
+
+3 @ SubTotal
+
+@ }
+
+GST
+
+LUDED IN THE TOTAL
+
+CARD
+
+--- OCR crop_4_2x.png psm=6 ---
+Be PN ty,
+-. | Change $0.00" |
+ME, A wou +.” il my Ste . le) - . A i
+= . FA yous s card: oH Set ‘ she
+57348 07555 oat anions 1549-26-05 -22 oe
+TIE TMT
+--- OCR crop_4_2x.png psm=11 ---
+
+Ro
+
+1ng
+
+ange
+
+$0.
+
+ec,
+
+Bra
+
+*" Tndicé
+
+bh’
+
+Flybuys
+
+an
+
+A@)
+
+» Tiiii

receipts/ocr/f1ef054a94e5632c/ocr_runs.json

@@ -0,0 +1,86 @@
+[
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/original.png",
+    "psm": 6,
+    "score": 793,
+    "chars": 493
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/original.png",
+    "psm": 11,
+    "score": 928,
+    "chars": 478
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/gray_contrast_2x.png",
+    "psm": 6,
+    "score": 584,
+    "chars": 334
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/gray_contrast_2x.png",
+    "psm": 11,
+    "score": 453,
+    "chars": 353
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/threshold_2x.png",
+    "psm": 6,
+    "score": 438,
+    "chars": 338
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/threshold_2x.png",
+    "psm": 11,
+    "score": 564,
+    "chars": 414
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_1_2x.png",
+    "psm": 6,
+    "score": 47,
+    "chars": 47
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_1_2x.png",
+    "psm": 11,
+    "score": 57,
+    "chars": 57
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_2_2x.png",
+    "psm": 6,
+    "score": 51,
+    "chars": 51
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_2_2x.png",
+    "psm": 11,
+    "score": 100,
+    "chars": 100
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_3_2x.png",
+    "psm": 6,
+    "score": 305,
+    "chars": 155
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_3_2x.png",
+    "psm": 11,
+    "score": 395,
+    "chars": 195
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_4_2x.png",
+    "psm": 6,
+    "score": 200,
+    "chars": 150
+  },
+  {
+    "file": "/home/openclaw/.openclaw/workspace/knowledge/receipts/ocr/f1ef054a94e5632c/crop_4_2x.png",
+    "psm": 11,
+    "score": 131,
+    "chars": 81
+  }
+]

receipts/scripts/receipt_import.py

@@ -0,0 +1,491 @@
+#!/usr/bin/env python3
+"""Local-first receipt/document importer for knowledge/receipts.
+
+Pipeline:
+1. Keep the original image/PDF immutable in knowledge/receipts/images/YY-MM-images/.
+2. Generate local OCR derivatives in knowledge/receipts/ocr/<sha-prefix>/.
+3. OCR locally with Tesseract/OCRmyPDF/Poppler.
+4. Parse likely receipt fields with conservative heuristics.
+5. Insert a searchable row into receipts.sqlite, or update an existing receipt id.
+
+No cloud APIs. No deletion. Designed as a wrapper Soren can call quickly from chat.
+"""
+
+from __future__ import annotations
+
+import argparse
+import datetime as dt
+import difflib
+import hashlib
+import json
+import os
+import re
+import shutil
+import sqlite3
+import subprocess
+import sys
+import tempfile
+from pathlib import Path
+from typing import Any
+
+try:
+    from PIL import Image, ImageEnhance, ImageOps
+except Exception as e:  # pragma: no cover
+    print(f"Pillow/PIL required: {e}", file=sys.stderr)
+    raise SystemExit(2)
+
+ROOT = Path(__file__).resolve().parents[1]
+DB = ROOT / "receipts.sqlite"
+IMAGES = ROOT / "images"
+OCR_ROOT = ROOT / "ocr"
+TEXT_EXTS = {".txt"}
+IMAGE_EXTS = {".jpg", ".jpeg", ".png", ".webp", ".tif", ".tiff", ".bmp"}
+PDF_EXTS = {".pdf"}
+
+
+def run(cmd: list[str], *, capture: bool = False, check: bool = True, timeout: int = 120) -> subprocess.CompletedProcess[str]:
+    try:
+        return subprocess.run(cmd, text=True, capture_output=capture, check=check, timeout=timeout)
+    except subprocess.CalledProcessError as e:
+        if e.stdout:
+            print(e.stdout, end="")
+        if e.stderr:
+            print(e.stderr, end="", file=sys.stderr)
+        raise SystemExit(e.returncode)
+
+
+def have(name: str) -> bool:
+    return shutil.which(name) is not None
+
+
+def sha256_file(path: Path) -> str:
+    h = hashlib.sha256()
+    with path.open("rb") as f:
+        for chunk in iter(lambda: f.read(1024 * 1024), b""):
+            h.update(chunk)
+    return h.hexdigest()
+
+
+def now_local() -> dt.datetime:
+    return dt.datetime.now().astimezone()
+
+
+def month_dir(ts: dt.datetime) -> Path:
+    return IMAGES / f"{ts:%y-%m}-images"
+
+
+def slug(s: str) -> str:
+    s = re.sub(r"[^A-Za-z0-9._-]+", "-", s.strip()).strip("-_.")
+    return s[:80] or "receipt"
+
+
+def copy_original(src: Path, *, category: str, captured: dt.datetime, sha: str) -> Path:
+    ext = src.suffix.lower() or ".jpg"
+    dest_dir = month_dir(captured)
+    dest_dir.mkdir(parents=True, exist_ok=True)
+    cat = slug(category.lower())
+    dest = dest_dir / f"{captured:%Y-%m-%d_%H%M}_{cat}_{sha[:8]}{ext}"
+    if not dest.exists():
+        shutil.copy2(src, dest)
+    return dest
+
+
+def image_derivatives(src: Path, ocr_dir: Path) -> list[Path]:
+    ocr_dir.mkdir(parents=True, exist_ok=True)
+    im = Image.open(src).convert("RGB")
+    outputs: list[Path] = []
+
+    # 1. Original as PNG, useful for consistent tesseract input.
+    original = ocr_dir / "original.png"
+    im.save(original)
+    outputs.append(original)
+
+    # 2. Grayscale + contrast + upscale. Good for screenshots and receipts.
+    g = ImageOps.grayscale(im)
+    g = ImageEnhance.Contrast(g).enhance(2.2)
+    up = g.resize((g.width * 2, g.height * 2))
+    gray = ocr_dir / "gray_contrast_2x.png"
+    up.save(gray)
+    outputs.append(gray)
+
+    # 3. Binary-ish threshold, helps faint receipt paper.
+    bw = up.point(lambda p: 255 if p > 175 else 0)
+    bw_path = ocr_dir / "threshold_2x.png"
+    bw.save(bw_path)
+    outputs.append(bw_path)
+
+    # 4. Vertical crops, useful for long phone screenshots/receipts.
+    if im.height > im.width * 1.2:
+        for i in range(4):
+            y0 = i * im.height // 4
+            y1 = (i + 1) * im.height // 4
+            crop = im.crop((0, y0, im.width, y1))
+            crop = ImageEnhance.Contrast(ImageOps.grayscale(crop)).enhance(2.0)
+            crop = crop.resize((crop.width * 2, crop.height * 2))
+            p = ocr_dir / f"crop_{i+1}_2x.png"
+            crop.save(p)
+            outputs.append(p)
+
+    return outputs
+
+
+def tesseract_text(image: Path, *, psm: int) -> str:
+    if not have("tesseract"):
+        raise RuntimeError("tesseract not installed")
+    cp = run(["tesseract", str(image), "stdout", "-l", "eng", "--psm", str(psm)], capture=True, check=False)
+    return cp.stdout or ""
+
+
+def ocr_image(src: Path, ocr_dir: Path) -> dict[str, Any]:
+    variants = image_derivatives(src, ocr_dir)
+    runs = []
+    combined_parts: list[str] = []
+    for img in variants:
+        for psm in (6, 11):
+            text = tesseract_text(img, psm=psm)
+            cleaned = clean_text(text)
+            score = score_text(cleaned)
+            runs.append({"file": str(img), "psm": psm, "score": score, "chars": len(cleaned)})
+            if cleaned:
+                combined_parts.append(f"--- OCR {img.name} psm={psm} ---\n{cleaned}")
+    combined = dedupe_lines("\n".join(combined_parts))
+    (ocr_dir / "ocr_raw.txt").write_text(combined)
+    (ocr_dir / "ocr_runs.json").write_text(json.dumps(runs, indent=2))
+    return {"text": combined, "runs": runs, "dir": str(ocr_dir)}
+
+
+def ocr_pdf(src: Path, ocr_dir: Path) -> dict[str, Any]:
+    ocr_dir.mkdir(parents=True, exist_ok=True)
+    raw_text = ""
+    if have("pdftotext"):
+        cp = run(["pdftotext", "-layout", str(src), "-"], capture=True, check=False)
+        raw_text = clean_text(cp.stdout or "")
+    ocr_pdf_path = None
+    if len(raw_text.strip()) < 40 and have("ocrmypdf"):
+        ocr_pdf_path = ocr_dir / "ocr.pdf"
+        run(["ocrmypdf", "--skip-text", "--deskew", "--rotate-pages", str(src), str(ocr_pdf_path)], check=False, timeout=300)
+        if ocr_pdf_path.exists() and have("pdftotext"):
+            cp = run(["pdftotext", "-layout", str(ocr_pdf_path), "-"], capture=True, check=False)
+            raw_text = clean_text(cp.stdout or raw_text)
+    (ocr_dir / "ocr_raw.txt").write_text(raw_text)
+    return {"text": raw_text, "runs": [{"pdf_ocr": bool(ocr_pdf_path and ocr_pdf_path.exists())}], "dir": str(ocr_dir)}
+
+
+def clean_text(text: str) -> str:
+    text = text.replace("\r", "\n")
+    text = re.sub(r"[ \t]+", " ", text)
+    text = re.sub(r"\n{3,}", "\n\n", text)
+    return text.strip()
+
+
+def dedupe_lines(text: str) -> str:
+    seen = set()
+    out = []
+    for line in text.splitlines():
+        key = re.sub(r"\s+", " ", line.strip()).lower()
+        if not key:
+            if out and out[-1] != "":
+                out.append("")
+            continue
+        # Preserve OCR section headers even when similar.
+        if key.startswith("--- ocr") or key not in seen:
+            out.append(line.rstrip())
+            seen.add(key)
+    return "\n".join(out).strip()
+
+
+def score_text(text: str) -> int:
+    score = len(text)
+    for word in ("total", "subtotal", "gst", "receipt", "invoice", "order", "payment", "$"):
+        score += text.lower().count(word) * 50
+    return score
+
+
+def money_to_float(s: str | None, *, require_dollars: bool = False, last: bool = False) -> float | None:
+    if not s:
+        return None
+    s = s.replace(",", "")
+    pat = r"-?\$\s*([0-9]+(?:\.[0-9]{2})?)" if require_dollars else r"-?\$?\s*([0-9]+(?:\.[0-9]{2})?)"
+    matches = list(re.finditer(pat, s))
+    if not matches:
+        return None
+    m = matches[-1] if last else matches[0]
+    return float(m.group(1))
+
+
+def parse_date(text: str) -> str | None:
+    patterns = [
+        (r"\b(20\d{2})[-/](\d{1,2})[-/](\d{1,2})\b", "%Y-%m-%d"),
+        (r"\b(\d{1,2})[-/](\d{1,2})[-/](20\d{2})\b", "%d-%m-%Y"),
+        (r"\b([A-Z][a-z]+)\s+(\d{1,2}),\s*(20\d{2})\b", "%B %d %Y"),
+        (r"\b(\d{1,2})\s+([A-Z][a-z]+)\s+(20\d{2})\b", "%d %B %Y"),
+    ]
+    for pat, fmt in patterns:
+        for m in re.finditer(pat, text):
+            val = " ".join(m.groups()) if "%B" in fmt else "-".join(m.groups())
+            try:
+                return dt.datetime.strptime(val, fmt).date().isoformat()
+            except ValueError:
+                continue
+    return None
+
+
+def likely_merchant(lines: list[str], text: str = "") -> str | None:
+    # Specific useful hint: many web receipts contain the domain even when the logo OCR is noisy.
+    domain = re.search(r"\b(?:www\.)?([a-z0-9-]+)\.com\.au\b", text, re.I)
+    if domain:
+        known = {"nopong": "No Pong"}
+        key = domain.group(1).lower().replace("-", "")
+        if key in known:
+            return known[key]
+        return domain.group(1).replace("-", " ").title()
+
+    bad = re.compile(r"^(--- ocr|receipt|tax invoice|invoice|order|checkout|date|total|subtotal|payment|product|qty|amount|thank you)", re.I)
+    for line in lines[:30]:
+        l = line.strip(" :-|•")
+        if len(l) < 3 or bad.search(l) or re.search(r"@|https?://|www\.|\$|\d{4,}", l):
+            continue
+        # Prefer line with letters, not mostly symbols.
+        if sum(ch.isalpha() for ch in l) >= 3:
+            return l[:120]
+    return None
+
+
+def parse_fields(text: str) -> dict[str, Any]:
+    lines = [l.strip() for l in text.splitlines() if l.strip()]
+    lower = text.lower()
+    fields: dict[str, Any] = {
+        "merchant_name": likely_merchant(lines, text),
+        "receipt_date": parse_date(text),
+        "currency": "AUD" if "$" in text else None,
+    }
+
+    # Order / invoice / transaction refs. Prefer same-line matches, then label-next-line matches.
+    ref_patterns = {
+        "docket_number": r"(?:order\s*(?:number|no\.?|#)|receipt\s*(?:number|no\.?|#)|docket\s*(?:number|no\.?|#))\s*[:#]?\s*([A-Z0-9-]{4,})",
+        "invoice_number": r"invoice\s*(?:number|no\.?|#)?\s*[:#]?\s*([A-Z0-9-]{4,})",
+        "transaction_reference": r"(?:transaction|txn|reference|ref)\s*(?:number|no\.?|#)?\s*[:#]?\s*([A-Z0-9-]{4,})",
+    }
+    for key, pat in ref_patterns.items():
+        m = re.search(pat, text, re.I)
+        if m and m.group(1).lower() not in {"date", "total", "email", "payment"}:
+            fields[key] = m.group(1)
+    for i, line in enumerate(lines[:-1]):
+        if "order number" in line.lower() and not fields.get("docket_number"):
+            # Common receipt layout: ORDER NUMBER: DATE: / 3465186 May 14, 2026
+            nxt = lines[i + 1]
+            m = re.search(r"\b([0-9]{4,}|[A-Z0-9-]{6,})\b", nxt)
+            if m:
+                fields["docket_number"] = m.group(1)
+
+    pm = re.search(r"payment\s*method\s*:?\s*([A-Za-z0-9 /_-]{3,40})", text, re.I)
+    if pm:
+        fields["payment_method"] = pm.group(1).strip()
+    else:
+        for method in ("PayPal", "Visa", "Mastercard", "EFTPOS", "Cash", "Amex", "Afterpay", "Apple Pay", "Google Pay"):
+            if method.lower() in lower:
+                fields["payment_method"] = method
+                break
+
+    # Amounts. Line-based parsing avoids confusing PRODUCT TOTAL headers and 85G sizes.
+    for line in lines:
+        low = line.lower()
+        amount_last = money_to_float(line, require_dollars=True, last=True)
+        amount_first = money_to_float(line, require_dollars=True, last=False)
+        if amount_last is None:
+            continue
+        if "subtotal" in low:
+            fields["subtotal"] = amount_last
+        if re.search(r"\bgst\b|\btax\b", low):
+            fields["gst"] = amount_last
+        if re.search(r"^(grand\s+total|amount\s+paid|total)\b", low):
+            # For "Total: $99.98 (includes $9.09 GST)", the total is the first amount.
+            fields["total"] = amount_first
+    # Embedded form: includes $9.09 GST
+    gst_m = re.search(r"includes\s+\$\s*([0-9,]+\.[0-9]{2})\s+GST", text, re.I)
+    if gst_m:
+        fields["gst"] = float(gst_m.group(1).replace(",", ""))
+
+    # Fallback total = largest money amount, conservative.
+    if not fields.get("total"):
+        monies = [money_to_float(m.group(0), require_dollars=True) for m in re.finditer(r"\$\s*[0-9,]+\.[0-9]{2}", text)]
+        monies = [m for m in monies if m is not None]
+        if monies:
+            fields["total"] = max(monies)
+
+    # Simple item extraction: lines with money, excluding summary/payment lines.
+    item_lines = []
+    seen_items = set()
+    skip = re.compile(r"subtotal|total|gst|tax|shipping|payment|points|redeemed|balance|change|cash|visa|mastercard|email|@", re.I)
+    for line in lines:
+        if "$" not in line or skip.search(line):
+            continue
+        amount = money_to_float(line, require_dollars=True, last=True)
+        desc = re.sub(r"\s*-?\$\s*[0-9,]+\.[0-9]{2}.*$", "", line).strip(" :-|•")
+        if amount is not None and len(desc) >= 3:
+            qty = None
+            qm = re.search(r"(?:x|×)\s*(\d+(?:\.\d+)?)\b", desc, re.I)
+            if qm:
+                qty = float(qm.group(1))
+                desc = re.sub(r"\s*(?:x|×)\s*\d+(?:\.\d+)?\b", "", desc, flags=re.I).strip()
+            norm_desc = re.sub(r"[^a-z0-9]+", " ", desc.lower()).strip()
+            key = (norm_desc, qty, amount)
+            if key in seen_items:
+                continue
+            # OCR variants can repeat the same item from different preprocessing runs.
+            duplicate_idx = None
+            for idx, existing in enumerate(item_lines):
+                if existing.get("quantity") == qty and existing.get("line_total") == amount:
+                    ex_norm = re.sub(r"[^a-z0-9]+", " ", existing["description"].lower()).strip()
+                    if difflib.SequenceMatcher(None, norm_desc, ex_norm).ratio() >= 0.92:
+                        duplicate_idx = idx
+                        break
+            if duplicate_idx is not None:
+                if len(desc) > len(item_lines[duplicate_idx]["description"]):
+                    item_lines[duplicate_idx]["description"] = desc[:200]
+                continue
+            seen_items.add(key)
+            item_lines.append({"description": desc[:200], "quantity": qty, "line_total": amount})
+    fields["items"] = item_lines[:50]
+    return fields
+
+
+def connect() -> sqlite3.Connection:
+    con = sqlite3.connect(DB)
+    con.row_factory = sqlite3.Row
+    return con
+
+
+def existing_by_sha(con: sqlite3.Connection, sha: str) -> sqlite3.Row | None:
+    return con.execute("SELECT * FROM receipts WHERE image_sha256=?", (sha,)).fetchone()
+
+
+def insert_or_update(con: sqlite3.Connection, *, source: Path, stored: Path, sha: str, parsed: dict[str, Any], text: str, ocr_info: dict[str, Any], args: argparse.Namespace) -> int:
+    captured = args.captured_at or now_local().isoformat(timespec="seconds")
+    confidence = "Local OCR via Tesseract. Parsed fields are heuristic; verify important warranty/tax data against original image."
+    ocr_json = json.dumps({"parsed": parsed, "ocr": ocr_info}, ensure_ascii=False)
+
+    values = {
+        "category": args.category,
+        "doc_type": args.doc_type,
+        "merchant_name": parsed.get("merchant_name"),
+        "receipt_date": parsed.get("receipt_date"),
+        "docket_number": parsed.get("docket_number"),
+        "invoice_number": parsed.get("invoice_number"),
+        "transaction_reference": parsed.get("transaction_reference"),
+        "payment_method": parsed.get("payment_method"),
+        "subtotal": parsed.get("subtotal"),
+        "gst": parsed.get("gst"),
+        "total": parsed.get("total"),
+        "currency": parsed.get("currency") or "AUD",
+        "image_path": str(stored),
+        "source_media_path": str(source),
+        "image_sha256": sha,
+        "ocr_raw_text": text,
+        "ocr_json": ocr_json,
+        "confidence_notes": confidence,
+        "captured_at": captured,
+        "notes": args.notes,
+    }
+
+    if args.receipt_id:
+        rid = args.receipt_id
+        assignments = ", ".join(f"{k}=?" for k in values.keys()) + ", updated_at=CURRENT_TIMESTAMP"
+        con.execute(f"UPDATE receipts SET {assignments} WHERE id=?", [*values.values(), rid])
+    else:
+        row = existing_by_sha(con, sha)
+        if row:
+            rid = int(row["id"])
+            if not args.update_existing:
+                return rid
+            assignments = ", ".join(f"{k}=?" for k in values.keys()) + ", updated_at=CURRENT_TIMESTAMP"
+            con.execute(f"UPDATE receipts SET {assignments} WHERE id=?", [*values.values(), rid])
+        else:
+            cols = ", ".join(values.keys())
+            qs = ", ".join("?" for _ in values)
+            cur = con.execute(f"INSERT INTO receipts ({cols}) VALUES ({qs})", list(values.values()))
+            rid = int(cur.lastrowid)
+
+    if args.replace_items:
+        con.execute("DELETE FROM receipt_items WHERE receipt_id=?", (rid,))
+        for item in parsed.get("items", []):
+            con.execute(
+                "INSERT INTO receipt_items (receipt_id, description, quantity, unit_price, line_total, notes) VALUES (?, ?, ?, ?, ?, ?)",
+                (rid, item.get("description"), item.get("quantity"), item.get("unit_price"), item.get("line_total"), "auto-parsed from OCR"),
+            )
+    con.commit()
+    return rid
+
+
+def main() -> None:
+    ap = argparse.ArgumentParser(description="Import/OCR a receipt image/PDF into knowledge/receipts")
+    ap.add_argument("source", help="Path to receipt image/PDF")
+    ap.add_argument("--category", default="Uncategorised")
+    ap.add_argument("--doc-type", default="receipt")
+    ap.add_argument("--captured-at", help="ISO timestamp; default now")
+    ap.add_argument("--receipt-id", type=int, help="Update an existing receipts.id instead of inserting")
+    ap.add_argument("--notes", help="Receipt-level note from Michael/Soren, separate from OCR confidence notes")
+    ap.add_argument("--update-existing", action="store_true", help="If SHA already exists, update that row")
+    ap.add_argument("--replace-items", action=argparse.BooleanOptionalAction, default=True)
+    ap.add_argument("--dry-run", action="store_true", help="OCR/parse only; no DB write/copy")
+    ap.add_argument("--json", action="store_true", help="Print machine-readable result")
+    args = ap.parse_args()
+
+    src = Path(args.source).expanduser().resolve()
+    if not src.exists():
+        raise SystemExit(f"source not found: {src}")
+    if not DB.exists():
+        raise SystemExit(f"database not found: {DB}")
+
+    ext = src.suffix.lower()
+    sha = sha256_file(src)
+    captured = dt.datetime.fromisoformat(args.captured_at) if args.captured_at else now_local()
+    ocr_dir = OCR_ROOT / sha[:16]
+
+    if ext in IMAGE_EXTS:
+        ocr_info = ocr_image(src, ocr_dir)
+    elif ext in PDF_EXTS:
+        ocr_info = ocr_pdf(src, ocr_dir)
+    elif ext in TEXT_EXTS:
+        ocr_dir.mkdir(parents=True, exist_ok=True)
+        ocr_info = {"text": clean_text(src.read_text(errors="replace")), "runs": [], "dir": str(ocr_dir)}
+        (ocr_dir / "ocr_raw.txt").write_text(ocr_info["text"])
+    else:
+        raise SystemExit(f"unsupported file type: {ext}")
+
+    text = ocr_info["text"]
+    parsed = parse_fields(text)
+    result: dict[str, Any] = {"source": str(src), "sha256": sha, "ocr_dir": str(ocr_dir), "parsed": parsed}
+
+    if args.dry_run:
+        result["dry_run"] = True
+    else:
+        try:
+            # If the source is already inside the receipts image archive, don't duplicate it.
+            src.relative_to(IMAGES.resolve())
+            stored = src.relative_to(Path.cwd()) if src.is_relative_to(Path.cwd()) else src
+        except Exception:
+            stored = copy_original(src, category=args.category, captured=captured, sha=sha)
+        with connect() as con:
+            rid = insert_or_update(con, source=src, stored=stored, sha=sha, parsed=parsed, text=text, ocr_info=ocr_info, args=args)
+        result.update({"receipt_id": rid, "image_path": str(stored)})
+
+    if args.json:
+        print(json.dumps(result, indent=2, ensure_ascii=False))
+    else:
+        print(f"OCR dir: {result['ocr_dir']}")
+        if "receipt_id" in result:
+            print(f"Receipt ID: {result['receipt_id']}")
+            print(f"Image: {result['image_path']}")
+        print("Parsed:")
+        for k, v in parsed.items():
+            if k == "items":
+                print(f"  items: {len(v)}")
+                for item in v[:10]:
+                    print(f"    - {item}")
+            else:
+                print(f"  {k}: {v}")
+
+
+if __name__ == "__main__":
+    main()