Add local receipt OCR + grouping (OCR/)
Two local, GPU-backed tools that talk to a local ollama server (no cloud, no API key): - receipt_ocr.py: receipt image -> ordered list of rows of text, via Qwen2.5-VL 7B. Zero deps (stdlib + ollama HTTP API). - receipt_group.py: OCR rows -> structured receipt (store, line items with multi-row details merged, total, metadata), via Mistral Small 24B. Detail rows are attached by arithmetic reconciliation, not position; a prompt completeness clause guards against dropped rows; a deterministic --json reconciliation pass audits item math, subtotal/total, and unit count. README documents setup, an ollama operating runbook, and design rationale. Includes the ALDI example image + its OCR rows. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Jean-Michel Tremblay
parent
e94a17160b
commit
344663d87d
5 changed files with 725 additions and 0 deletions
263
OCR/README.md
Normal file
263
OCR/README.md
Normal file
|
|
@ -0,0 +1,263 @@
|
||||||
|
I want to create an OCR feature (ollama, whatever) that's
|
||||||
|
|
||||||
|
1) local (assume I have a GPU with 24GB VRAM)
|
||||||
|
2) works on receipts (scanned or digital)
|
||||||
|
3) given a receipt, return an ordered list of rows of text in that receipt
|
||||||
|
4) example /home/jm/programming/HSAmanager/OCR/example/data/aldi.jpg
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Getting started (working prototype)
|
||||||
|
|
||||||
|
Local, GPU-backed, no cloud / no API key. Engine: **Qwen2.5-VL 7B** served by
|
||||||
|
**ollama**. The model transcribes the receipt top-to-bottom and returns one
|
||||||
|
string per physical row, columns joined left-to-right.
|
||||||
|
|
||||||
|
### One-time setup (already done on this machine)
|
||||||
|
|
||||||
|
- ollama binary installed (no root) at `~/.local/bin/ollama`.
|
||||||
|
- Model pulled: `ollama pull qwen2.5vl:7b` (~6 GB).
|
||||||
|
|
||||||
|
### Run
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# 1. start the server (leave running; ~6 GB VRAM when the model loads)
|
||||||
|
~/.local/bin/ollama serve &
|
||||||
|
|
||||||
|
# 2. transcribe a receipt -> ordered, numbered rows
|
||||||
|
python3 OCR/receipt_ocr.py OCR/example/data/aldi.jpg
|
||||||
|
|
||||||
|
# raw JSON {"rows": [...]} instead of numbered text
|
||||||
|
python3 OCR/receipt_ocr.py OCR/example/data/aldi.jpg --json
|
||||||
|
```
|
||||||
|
|
||||||
|
No third-party Python deps — `receipt_ocr.py` uses the stdlib and the ollama
|
||||||
|
HTTP API. Override with `OLLAMA_MODEL` / `OLLAMA_HOST` env vars.
|
||||||
|
|
||||||
|
### Notes / next steps
|
||||||
|
|
||||||
|
- Works on the ALDI example with near-perfect line ordering and column joining.
|
||||||
|
- Caveat: the model sometimes *expands* thermal-print abbreviations
|
||||||
|
(`Fire Roasted Tom` → `Fire Roasted Tomatoes`). Helpful for readability but
|
||||||
|
not byte-for-byte faithful; tighten the prompt if you need verbatim text.
|
||||||
|
- Accuracy upgrade: `qwen2.5vl:32b` (q4 ≈ 20 GB, fits the 3090 but tighter/slower).
|
||||||
|
- This is local-only and complementary to `../receiptscan/`, which uses the
|
||||||
|
Claude cloud API for line-item categorization + annotated rendering.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Operating ollama (runbook)
|
||||||
|
|
||||||
|
ollama is the local model server both scripts talk to over HTTP
|
||||||
|
(`http://127.0.0.1:11434`). It must be running before you call either script.
|
||||||
|
Installed **without root** at `~/.local/bin/ollama` (the binary was extracted
|
||||||
|
from the `ollama-linux-amd64.tar.zst` GitHub release; `~/.local/bin` is on PATH,
|
||||||
|
so plain `ollama` works too).
|
||||||
|
|
||||||
|
### Start / check / stop
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# Start the server (does NOT survive a reboot or terminal close — restart it):
|
||||||
|
ollama serve & # or: nohup ollama serve > /tmp/ollama.log 2>&1 &
|
||||||
|
|
||||||
|
# Is it up? (prints a JSON version string if so)
|
||||||
|
curl -fsS http://127.0.0.1:11434/api/version
|
||||||
|
|
||||||
|
# Is the process alive?
|
||||||
|
pgrep -af "ollama serve"
|
||||||
|
|
||||||
|
# Stop it:
|
||||||
|
pkill -f "ollama serve"
|
||||||
|
```
|
||||||
|
|
||||||
|
After a reboot the server is **not** running — just `ollama serve &` again. The
|
||||||
|
pulled models persist on disk (under `~/.ollama`), so you don't re-download them.
|
||||||
|
|
||||||
|
### Models
|
||||||
|
|
||||||
|
```bash
|
||||||
|
ollama list # models on disk
|
||||||
|
ollama ps # models currently loaded in VRAM (and how long until idle-unload)
|
||||||
|
ollama pull <model> # download a model, e.g. ollama pull qwen2.5:14b
|
||||||
|
ollama rm <model> # delete a model to reclaim disk
|
||||||
|
```
|
||||||
|
|
||||||
|
Currently pulled: `qwen2.5vl:7b` (OCR), `mistral-small3.2:24b` (grouping, default),
|
||||||
|
plus `qwen2.5:7b` / `qwen2.5:14b` (earlier grouping experiments — `ollama rm`
|
||||||
|
them if you want the disk back).
|
||||||
|
|
||||||
|
### VRAM / GPU notes
|
||||||
|
|
||||||
|
- A model loads into VRAM on first request and **auto-unloads after ~5 min idle**
|
||||||
|
(so `ollama ps` is often empty — that's normal; the next call reloads it in a
|
||||||
|
few seconds).
|
||||||
|
- Only run one large model at a time on the 24 GB 3090. The OCR model (~6 GB) and
|
||||||
|
the grouping model (~15 GB) can coexist, but bumping grouping to
|
||||||
|
`qwen2.5:32b`/`qwen2.5vl:32b` (~20 GB) leaves little room — expect a reload
|
||||||
|
when switching between OCR and grouping.
|
||||||
|
- Force-unload now (free VRAM without stopping the server):
|
||||||
|
`ollama stop <model>`.
|
||||||
|
- Keep a model resident longer / shorter via the request `keep_alive` field, or
|
||||||
|
globally with `OLLAMA_KEEP_ALIVE` (e.g. `OLLAMA_KEEP_ALIVE=30m ollama serve`).
|
||||||
|
|
||||||
|
### Pointing the scripts elsewhere
|
||||||
|
|
||||||
|
Both scripts honor `OLLAMA_HOST` and `OLLAMA_MODEL`:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
OLLAMA_MODEL=qwen2.5:14b python3 OCR/receipt_group.py # try another model
|
||||||
|
OLLAMA_HOST=http://other-box:11434 python3 OCR/receipt_ocr.py img.jpg # remote server
|
||||||
|
```
|
||||||
|
|
||||||
|
### Logs / troubleshooting
|
||||||
|
|
||||||
|
- If a script hangs or errors connecting, the server is probably down — check
|
||||||
|
with the `curl` above and restart.
|
||||||
|
- Startup logs (GPU discovery, VRAM, errors) go wherever you redirected `serve`
|
||||||
|
(e.g. `/tmp/ollama.log`); tail that if a model fails to load.
|
||||||
|
- "model not found" → you haven't `ollama pull`ed it (or typo'd the tag); see
|
||||||
|
`ollama list`.
|
||||||
|
|
||||||
|
|
||||||
|
# grouping the OCR results
|
||||||
|
next task is ... given an OCR output, like:
|
||||||
|
```text
|
||||||
|
1 ALDI
|
||||||
|
2 Store #145
|
||||||
|
3 1501 Rockville Pike
|
||||||
|
4 Rockville, MD
|
||||||
|
5 https://help.aldi.us
|
||||||
|
6 Your cashier today was Wendy
|
||||||
|
7 382175 Unsalted Peanuts 2.29 FA
|
||||||
|
8 382175 Unsalted Peanuts 2.29 FA
|
||||||
|
9 384773 4 lb. Sugar 2.79 FA
|
||||||
|
10 384773 4 lb. Sugar 2.79 FA
|
||||||
|
11 382437 Fire Roasted Tomatoes 2.30 FA
|
||||||
|
12 2 @ 1.15
|
||||||
|
13 371607 Canned Cat Food 2.28 NB
|
||||||
|
14 4 @ 0.57
|
||||||
|
15 371607 Canned Cat Food 0.57 NB
|
||||||
|
16 416787 1% Milk, Gallon 3.38 FA
|
||||||
|
17 343557 Protein Powder 18.49 FA
|
||||||
|
18 416645 Chocolate Milk 1.91 FA
|
||||||
|
19 382260 Plain NF Greek Yogurt 2.79 FA
|
||||||
|
20 382260 Plain NF Greek Yogurt 2.79 FA
|
||||||
|
21 297956 PureandSimple Bars 3.99 FA
|
||||||
|
22 356508 Broccoli Crowns 2.21 FA
|
||||||
|
23 1.17 lb x 1.89/lb
|
||||||
|
24 469529 Corn Tortillas 1.95 FA
|
||||||
|
25 609748 Assorted Cashews 6.49 FA
|
||||||
|
26 634568 Chicken Skewers 7.49 FA
|
||||||
|
27 382473 Shredded Mozzarella 3.29 FA
|
||||||
|
28 341878 Yellow Onions 1.85 FA
|
||||||
|
29 535290 ABF B/S Thighs 6.61 FA
|
||||||
|
30 272135 Salmon Portions 9.35 FA
|
||||||
|
31 272135 Salmon Portions 8.54 FA
|
||||||
|
32 382310 Feta Crumbles 1.29 FA
|
||||||
|
33 356508 Broccoli Crowns 2.53 FA
|
||||||
|
34 1.34 lb x 1.89/lb
|
||||||
|
35 262137 Stuffed Olives 2.89 FA
|
||||||
|
36 356490 Bagged Avocados 2.99 FA
|
||||||
|
37 356607 Red Delic Apples 2.75 FA
|
||||||
|
38 356527 Celery 1.89 FA
|
||||||
|
39 382653 Indian Sauces 3.69 FA
|
||||||
|
40 356628 Seedless Cucumber 0.89 FA
|
||||||
|
41 356684 WildTwist Apf LRW 5.44 FA
|
||||||
|
42 (G) 3.10 lb - (T) 0.06 lb
|
||||||
|
43 (N) 3.04 lb x 1.79/lb
|
||||||
|
44 282119 Pineapples 1.89 FA
|
||||||
|
45 356419 Mangoes 3.80 FA
|
||||||
|
46 4 @ 0.95
|
||||||
|
47 356691 Zucchini 2.01 FA
|
||||||
|
48 1.69 lb x 1.19/lb
|
||||||
|
49 356522 Cantaloupe 1.89 FA
|
||||||
|
50 356504 Blueberries 9.95 FA
|
||||||
|
51 5 @ 1.99
|
||||||
|
52 388137 Large Eggs 2.92 FA
|
||||||
|
53 2 @ 1.46
|
||||||
|
54 262747 Bananas LRW 1.45 FA
|
||||||
|
55 (G) 2.97 lb - (T) 0.01 lb
|
||||||
|
56 (N) 2.96 lb x 0.49/lb
|
||||||
|
57 262747 Bananas LRW 1.13 FA
|
||||||
|
58 (G) 2.32 lb - (T) 0.01 lb
|
||||||
|
59 (N) 2.31 lb x 0.49/lb
|
||||||
|
60 356427 Multi-Peppers 3pk. 2.69 FA
|
||||||
|
61 341876 Red Grapes LRW 5.48 FA
|
||||||
|
62 (G) 3.96 lb - (T) 0.02 lb
|
||||||
|
63 (N) 3.94 lb x 1.39/lb
|
||||||
|
64 479744 Campari Tomatoes 2.99 FA
|
||||||
|
65 479744 Campari Tomatoes 2.99 FA
|
||||||
|
66 569269 Protein Bread 3.99 FA
|
||||||
|
67 343989 Family AsstCookie 1.99 FA
|
||||||
|
68 356646 Strawberries 2.09 FA
|
||||||
|
69 356646 Strawberries 2.09 FA
|
||||||
|
70 356646 Strawberries 2.09 FA
|
||||||
|
71 VISA 172.42
|
||||||
|
72 ************0294 ONLINE
|
||||||
|
73 06/21/26 12:43 Ref/Seq # 0
|
||||||
|
74 Auth# 00401D
|
||||||
|
75 AID A000000031010
|
||||||
|
76 TVR 0000000000
|
||||||
|
77 IAD 06021203A00000
|
||||||
|
78 TSI 0000 ARC 00 EntryMode 07
|
||||||
|
79 ++APPROVED++
|
||||||
|
80 SUBTOTAL 172.24
|
||||||
|
81 B-Taxable @6.00% 0.18
|
||||||
|
82 A-Taxable @0.00% 0.00
|
||||||
|
83 AMOUNT DUE 172.42
|
||||||
|
84 TOTAL $172.42
|
||||||
|
85 60 ITEMS
|
||||||
|
86 Credit Card $ 172.42
|
||||||
|
87 *7367 L411/005/064 06/21/26 12:43PM
|
||||||
|
88 ************
|
||||||
|
89 Sign up for ALDI emails
|
||||||
|
90 for a sneak peek on the weekly ad!
|
||||||
|
91 www.aldi.us/signup
|
||||||
|
```
|
||||||
|
|
||||||
|
write a LLM call, still using my local ollama (you can switch the model). I want the output to be an object roughly like that:
|
||||||
|
|
||||||
|
store name:
|
||||||
|
store branch:
|
||||||
|
store address:
|
||||||
|
list of rows that belong together as one item purchased
|
||||||
|
total in dollars.
|
||||||
|
(anything else is relevant here)?
|
||||||
|
|
||||||
|
you can hardcode the lines for this test and let's call the LLM to see what it does
|
||||||
|
|
||||||
|
the output needs to be a new script (receipt_group.py) that (for now) read this examle as is from an example file that you will create, and with the correct prompt organizes the information correctly.
|
||||||
|
|
||||||
|
## Grouping — implemented
|
||||||
|
|
||||||
|
`receipt_group.py` reads OCR rows (default
|
||||||
|
`example/data/aldi_ocr.txt`) and asks a local ollama text model to reconstruct
|
||||||
|
the receipt: store name/branch/address, line items (merging multi-row items
|
||||||
|
such as weighed produce and `qty @ price` lines), total, and metadata
|
||||||
|
(cashier, datetime, payment, subtotal, tax, item count).
|
||||||
|
|
||||||
|
```bash
|
||||||
|
python3 OCR/receipt_group.py # groups only: each item + its rows
|
||||||
|
python3 OCR/receipt_group.py OCR/example/data/aldi_ocr.txt --json # full object + reconciliation
|
||||||
|
```
|
||||||
|
|
||||||
|
- Default model: `mistral-small3.2:24b` (~15 GB). Override with `OLLAMA_MODEL`.
|
||||||
|
Tried qwen2.5:7b (many merge errors) and qwen2.5:14b (~90%); Mistral Small got
|
||||||
|
the grouping right. Swapping models is just the env var — no code change.
|
||||||
|
- **Why an LLM groups, not a regex**: detecting item vs detail rows is trivial,
|
||||||
|
but deciding *which* item a detail row attaches to is not — wrapped
|
||||||
|
descriptions, discount lines, and qty/weight lines can sit above or below the
|
||||||
|
item, so a positional rule breaks. The prompt tells the model to attach each
|
||||||
|
detail row to the item its **math** reconciles with (count × unit, weight ×
|
||||||
|
rate = the item's charged price), never by a fixed above/below position.
|
||||||
|
- **Completeness clause** in the prompt: every input row must land in exactly one
|
||||||
|
item's `rows`, copied verbatim, and the model must walk the input top-to-bottom
|
||||||
|
to confirm nothing was dropped. This fixed the model silently swallowing Large
|
||||||
|
Eggs' `2 @ 1.46` line. Prompting lowers the odds of a drop but cannot guarantee
|
||||||
|
it — hence the backstop below.
|
||||||
|
- **Deterministic reconciliation pass** (the part you *can* trust), shown in
|
||||||
|
`--json` under `validation`: Python recomputes each item from its detail math,
|
||||||
|
sums items vs the printed subtotal, checks total = subtotal + tax, and sums
|
||||||
|
unit quantities vs the printed `N ITEMS`. When the model slips (e.g. dropping a
|
||||||
|
`qty @` line), the unit-count check flags it (59 vs 60) even though dollars
|
||||||
|
still balance.
|
||||||
BIN
OCR/example/data/aldi.jpg
Normal file
BIN
OCR/example/data/aldi.jpg
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 438 KiB |
91
OCR/example/data/aldi_ocr.txt
Normal file
91
OCR/example/data/aldi_ocr.txt
Normal file
|
|
@ -0,0 +1,91 @@
|
||||||
|
1 ALDI
|
||||||
|
2 Store #145
|
||||||
|
3 1501 Rockville Pike
|
||||||
|
4 Rockville, MD
|
||||||
|
5 https://help.aldi.us
|
||||||
|
6 Your cashier today was Wendy
|
||||||
|
7 382175 Unsalted Peanuts 2.29 FA
|
||||||
|
8 382175 Unsalted Peanuts 2.29 FA
|
||||||
|
9 384773 4 lb. Sugar 2.79 FA
|
||||||
|
10 384773 4 lb. Sugar 2.79 FA
|
||||||
|
11 382437 Fire Roasted Tomatoes 2.30 FA
|
||||||
|
12 2 @ 1.15
|
||||||
|
13 371607 Canned Cat Food 2.28 NB
|
||||||
|
14 4 @ 0.57
|
||||||
|
15 371607 Canned Cat Food 0.57 NB
|
||||||
|
16 416787 1% Milk, Gallon 3.38 FA
|
||||||
|
17 343557 Protein Powder 18.49 FA
|
||||||
|
18 416645 Chocolate Milk 1.91 FA
|
||||||
|
19 382260 Plain NF Greek Yogurt 2.79 FA
|
||||||
|
20 382260 Plain NF Greek Yogurt 2.79 FA
|
||||||
|
21 297956 PureandSimple Bars 3.99 FA
|
||||||
|
22 356508 Broccoli Crowns 2.21 FA
|
||||||
|
23 1.17 lb x 1.89/lb
|
||||||
|
24 469529 Corn Tortillas 1.95 FA
|
||||||
|
25 609748 Assorted Cashews 6.49 FA
|
||||||
|
26 634568 Chicken Skewers 7.49 FA
|
||||||
|
27 382473 Shredded Mozzarella 3.29 FA
|
||||||
|
28 341878 Yellow Onions 1.85 FA
|
||||||
|
29 535290 ABF B/S Thighs 6.61 FA
|
||||||
|
30 272135 Salmon Portions 9.35 FA
|
||||||
|
31 272135 Salmon Portions 8.54 FA
|
||||||
|
32 382310 Feta Crumbles 1.29 FA
|
||||||
|
33 356508 Broccoli Crowns 2.53 FA
|
||||||
|
34 1.34 lb x 1.89/lb
|
||||||
|
35 262137 Stuffed Olives 2.89 FA
|
||||||
|
36 356490 Bagged Avocados 2.99 FA
|
||||||
|
37 356607 Red Delic Apples 2.75 FA
|
||||||
|
38 356527 Celery 1.89 FA
|
||||||
|
39 382653 Indian Sauces 3.69 FA
|
||||||
|
40 356628 Seedless Cucumber 0.89 FA
|
||||||
|
41 356684 WildTwist Apf LRW 5.44 FA
|
||||||
|
42 (G) 3.10 lb - (T) 0.06 lb
|
||||||
|
43 (N) 3.04 lb x 1.79/lb
|
||||||
|
44 282119 Pineapples 1.89 FA
|
||||||
|
45 356419 Mangoes 3.80 FA
|
||||||
|
46 4 @ 0.95
|
||||||
|
47 356691 Zucchini 2.01 FA
|
||||||
|
48 1.69 lb x 1.19/lb
|
||||||
|
49 356522 Cantaloupe 1.89 FA
|
||||||
|
50 356504 Blueberries 9.95 FA
|
||||||
|
51 5 @ 1.99
|
||||||
|
52 388137 Large Eggs 2.92 FA
|
||||||
|
53 2 @ 1.46
|
||||||
|
54 262747 Bananas LRW 1.45 FA
|
||||||
|
55 (G) 2.97 lb - (T) 0.01 lb
|
||||||
|
56 (N) 2.96 lb x 0.49/lb
|
||||||
|
57 262747 Bananas LRW 1.13 FA
|
||||||
|
58 (G) 2.32 lb - (T) 0.01 lb
|
||||||
|
59 (N) 2.31 lb x 0.49/lb
|
||||||
|
60 356427 Multi-Peppers 3pk. 2.69 FA
|
||||||
|
61 341876 Red Grapes LRW 5.48 FA
|
||||||
|
62 (G) 3.96 lb - (T) 0.02 lb
|
||||||
|
63 (N) 3.94 lb x 1.39/lb
|
||||||
|
64 479744 Campari Tomatoes 2.99 FA
|
||||||
|
65 479744 Campari Tomatoes 2.99 FA
|
||||||
|
66 569269 Protein Bread 3.99 FA
|
||||||
|
67 343989 Family AsstCookie 1.99 FA
|
||||||
|
68 356646 Strawberries 2.09 FA
|
||||||
|
69 356646 Strawberries 2.09 FA
|
||||||
|
70 356646 Strawberries 2.09 FA
|
||||||
|
71 VISA 172.42
|
||||||
|
72 ************0294 ONLINE
|
||||||
|
73 06/21/26 12:43 Ref/Seq # 0
|
||||||
|
74 Auth# 00401D
|
||||||
|
75 AID A000000031010
|
||||||
|
76 TVR 0000000000
|
||||||
|
77 IAD 06021203A00000
|
||||||
|
78 TSI 0000 ARC 00 EntryMode 07
|
||||||
|
79 ++APPROVED++
|
||||||
|
80 SUBTOTAL 172.24
|
||||||
|
81 B-Taxable @6.00% 0.18
|
||||||
|
82 A-Taxable @0.00% 0.00
|
||||||
|
83 AMOUNT DUE 172.42
|
||||||
|
84 TOTAL $172.42
|
||||||
|
85 60 ITEMS
|
||||||
|
86 Credit Card $ 172.42
|
||||||
|
87 *7367 L411/005/064 06/21/26 12:43PM
|
||||||
|
88 ************
|
||||||
|
89 Sign up for ALDI emails
|
||||||
|
90 for a sneak peek on the weekly ad!
|
||||||
|
91 www.aldi.us/signup
|
||||||
284
OCR/receipt_group.py
Executable file
284
OCR/receipt_group.py
Executable file
|
|
@ -0,0 +1,284 @@
|
||||||
|
#!/usr/bin/env python3
|
||||||
|
"""Group raw receipt-OCR rows into a structured object with a local LLM.
|
||||||
|
|
||||||
|
Takes the text output of receipt_ocr.py (one row per line, optionally prefixed
|
||||||
|
with a line number) and asks a local ollama model to reconstruct the receipt:
|
||||||
|
store identity, the line items (merging multi-row items such as weighed produce
|
||||||
|
and "qty @ price" lines), the total, and any other useful metadata.
|
||||||
|
|
||||||
|
Local-only: no cloud, no API key. Uses a text model (default qwen2.5:7b).
|
||||||
|
|
||||||
|
Usage:
|
||||||
|
python3 receipt_group.py [ocr_rows.txt] # pretty summary
|
||||||
|
python3 receipt_group.py ocr_rows.txt --json # raw structured JSON
|
||||||
|
|
||||||
|
Env:
|
||||||
|
OLLAMA_HOST default http://127.0.0.1:11434
|
||||||
|
OLLAMA_MODEL default qwen2.5:7b
|
||||||
|
"""
|
||||||
|
|
||||||
|
import json
|
||||||
|
import os
|
||||||
|
import re
|
||||||
|
import sys
|
||||||
|
import urllib.request
|
||||||
|
|
||||||
|
DEFAULT_INPUT = os.path.join(os.path.dirname(__file__), "example", "data", "aldi_ocr.txt")
|
||||||
|
HOST = os.environ.get("OLLAMA_HOST", "http://127.0.0.1:11434").rstrip("/")
|
||||||
|
MODEL = os.environ.get("OLLAMA_MODEL", "mistral-small3.2:24b")
|
||||||
|
|
||||||
|
PROMPT = """You are given the OCR transcription of a single retail receipt, one \
|
||||||
|
physical row per line. Each line may start with a row number (e.g. "23 ...") — \
|
||||||
|
ignore the row number itself, it is not part of the receipt text.
|
||||||
|
|
||||||
|
Reconstruct the receipt as structured data. Rules:
|
||||||
|
|
||||||
|
- A purchased ITEM begins on a row that has a product code (a run of digits) and \
|
||||||
|
a price. EVERY row that begins with a product code starts a NEW item — never \
|
||||||
|
attach a code-bearing row to the item above it, even when the name and price are \
|
||||||
|
identical to the row above. Two identical consecutive code rows are TWO separate \
|
||||||
|
purchases, each its own entry. Do not merge or de-duplicate them.
|
||||||
|
- A DETAIL row is any row that does NOT begin with a product code, e.g.:
|
||||||
|
* quantity lines like "2 @ 1.15" (count @ unit price)
|
||||||
|
* weight lines like "1.17 lb x 1.89/lb"
|
||||||
|
* gross/tare/net lines like "(G) 3.10 lb - (T) 0.06 lb" and "(N) 3.04 lb x 1.79/lb"
|
||||||
|
Attach each detail row to the item it actually describes. Decide that by the \
|
||||||
|
MATH, not by position: count x unit price, or weight x rate, equals that item's \
|
||||||
|
charged price. (A gross/tare line that has no price of its own stays with the \
|
||||||
|
net line it accompanies.) The describing item is usually adjacent, but verify by \
|
||||||
|
reconciliation rather than assuming it is the one above or below.
|
||||||
|
- The item's "price" is the dollar amount printed on the item's own (first) row — \
|
||||||
|
the amount actually charged. Keep two decimals.
|
||||||
|
- Put the exact original row strings that belong to each item in its "rows" array, \
|
||||||
|
in order, WITHOUT the leading row number.
|
||||||
|
- Header rows (store name, branch, address, website, cashier) are NOT items.
|
||||||
|
- Payment, auth, subtotal, tax, total, item-count and marketing/footer rows are \
|
||||||
|
NOT items — pull the useful values into the metadata fields instead.
|
||||||
|
- Do not invent values. If something is not present, omit it.
|
||||||
|
|
||||||
|
COMPLETENESS — this is critical:
|
||||||
|
- Every item row AND every detail row from the input must appear in exactly ONE \
|
||||||
|
item's "rows" array, copied VERBATIM (same digits, same spacing, same price). \
|
||||||
|
You may not drop, skip, summarize, or rewrite any row.
|
||||||
|
- NEVER discard a quantity line ("2 @ 1.46"), a weight line ("1.17 lb x 1.89/lb"), \
|
||||||
|
or a gross/tare/net line. Attach each to the item its math reconciles with.
|
||||||
|
- Before you answer, walk the input top to bottom and confirm that every row is \
|
||||||
|
either placed in an item's "rows" or is a header/payment/footer row that you \
|
||||||
|
deliberately excluded. No item or detail row may be missing.
|
||||||
|
|
||||||
|
Return STRICT JSON only, matching the requested schema."""
|
||||||
|
|
||||||
|
SCHEMA = {
|
||||||
|
"type": "object",
|
||||||
|
"properties": {
|
||||||
|
"store_name": {"type": "string"},
|
||||||
|
"store_branch": {"type": "string"},
|
||||||
|
"store_address": {"type": "string"},
|
||||||
|
"items": {
|
||||||
|
"type": "array",
|
||||||
|
"items": {
|
||||||
|
"type": "object",
|
||||||
|
"properties": {
|
||||||
|
"name": {"type": "string"},
|
||||||
|
"code": {"type": "string"},
|
||||||
|
"price": {"type": "number"},
|
||||||
|
"rows": {"type": "array", "items": {"type": "string"}},
|
||||||
|
},
|
||||||
|
"required": ["name", "price", "rows"],
|
||||||
|
},
|
||||||
|
},
|
||||||
|
"total": {"type": "number"},
|
||||||
|
"metadata": {
|
||||||
|
"type": "object",
|
||||||
|
"properties": {
|
||||||
|
"cashier": {"type": "string"},
|
||||||
|
"purchase_datetime": {"type": "string"},
|
||||||
|
"payment_method": {"type": "string"},
|
||||||
|
"subtotal": {"type": "number"},
|
||||||
|
"tax": {"type": "number"},
|
||||||
|
"item_count": {"type": "integer"},
|
||||||
|
},
|
||||||
|
},
|
||||||
|
},
|
||||||
|
"required": ["store_name", "items", "total"],
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
def group(rows_text):
|
||||||
|
payload = {
|
||||||
|
"model": MODEL,
|
||||||
|
"messages": [
|
||||||
|
{"role": "user", "content": PROMPT + "\n\n--- RECEIPT OCR ---\n" + rows_text},
|
||||||
|
],
|
||||||
|
"stream": False,
|
||||||
|
"format": SCHEMA,
|
||||||
|
"options": {"temperature": 0, "num_ctx": 8192},
|
||||||
|
}
|
||||||
|
req = urllib.request.Request(
|
||||||
|
f"{HOST}/api/chat",
|
||||||
|
data=json.dumps(payload).encode("utf-8"),
|
||||||
|
headers={"Content-Type": "application/json"},
|
||||||
|
)
|
||||||
|
with urllib.request.urlopen(req, timeout=600) as resp:
|
||||||
|
body = json.load(resp)
|
||||||
|
return json.loads(body["message"]["content"])
|
||||||
|
|
||||||
|
|
||||||
|
# --- Deterministic reconciliation -------------------------------------------
|
||||||
|
# The LLM decides the grouping; arithmetic decides whether to trust it. Detail
|
||||||
|
# rows carry their own math, so we recompute each item's price from its details
|
||||||
|
# and balance the whole receipt against the printed subtotal / total / count.
|
||||||
|
|
||||||
|
QTY_RE = re.compile(r"(\d+)\s*@\s*\$?(\d+(?:\.\d+)?)") # "4 @ 0.57"
|
||||||
|
WEIGHT_RE = re.compile(r"(\d+(?:\.\d+)?)\s*lb\s*[xX]\s*\$?(\d+(?:\.\d+)?)\s*/\s*lb") # "1.17 lb x 1.89/lb"
|
||||||
|
TOL = 0.02 # cents of slack for rounding on weighed items
|
||||||
|
|
||||||
|
|
||||||
|
def check_item(it):
|
||||||
|
"""Reconcile one item from its detail rows.
|
||||||
|
|
||||||
|
Returns (status, qty, note) where status is 'ok' | 'warn' | 'none'.
|
||||||
|
'none' = no math-bearing detail row, so there's nothing to verify.
|
||||||
|
"""
|
||||||
|
price = float(it.get("price", 0))
|
||||||
|
qty = 1
|
||||||
|
expected = None
|
||||||
|
note = ""
|
||||||
|
for row in it.get("rows", [])[1:]: # skip the anchor row; details only
|
||||||
|
m = QTY_RE.search(row)
|
||||||
|
if m:
|
||||||
|
qty = int(m.group(1))
|
||||||
|
expected = round(qty * float(m.group(2)), 2)
|
||||||
|
note = f"{m.group(1)} @ {m.group(2)} = {expected:.2f}"
|
||||||
|
continue
|
||||||
|
w = WEIGHT_RE.search(row)
|
||||||
|
if w:
|
||||||
|
expected = round(float(w.group(1)) * float(w.group(2)), 2)
|
||||||
|
note = f"{w.group(1)} lb x {w.group(2)}/lb = {expected:.2f}"
|
||||||
|
if expected is None:
|
||||||
|
return ("none", qty, "")
|
||||||
|
if abs(expected - price) <= TOL:
|
||||||
|
return ("ok", qty, note)
|
||||||
|
return ("warn", qty, f"{note} != charged {price:.2f}")
|
||||||
|
|
||||||
|
|
||||||
|
def validate(obj):
|
||||||
|
items = obj.get("items", [])
|
||||||
|
meta = obj.get("metadata", {}) or {}
|
||||||
|
|
||||||
|
per_item = [check_item(it) for it in items]
|
||||||
|
items_sum = round(sum(float(it.get("price", 0)) for it in items), 2)
|
||||||
|
unit_count = sum(q for _, q, _ in per_item)
|
||||||
|
|
||||||
|
subtotal = meta.get("subtotal")
|
||||||
|
tax = meta.get("tax") or 0
|
||||||
|
total_printed = obj.get("total")
|
||||||
|
count_printed = meta.get("item_count")
|
||||||
|
|
||||||
|
report = {
|
||||||
|
"per_item": [{"status": s, "qty": q, "note": n} for s, q, n in per_item],
|
||||||
|
"items_sum": items_sum,
|
||||||
|
"subtotal_printed": subtotal,
|
||||||
|
"subtotal_ok": subtotal is not None and abs(items_sum - subtotal) <= TOL,
|
||||||
|
"total_computed": round(items_sum + tax, 2),
|
||||||
|
"total_printed": total_printed,
|
||||||
|
"total_ok": total_printed is not None and abs(items_sum + tax - total_printed) <= TOL,
|
||||||
|
"unit_count": unit_count,
|
||||||
|
"unit_count_printed": count_printed,
|
||||||
|
"unit_count_ok": count_printed is not None and unit_count == count_printed,
|
||||||
|
}
|
||||||
|
return report
|
||||||
|
|
||||||
|
|
||||||
|
# --- Rendering ----------------------------------------------------------------
|
||||||
|
|
||||||
|
MARK = {"ok": "✓", "warn": "⚠", "none": " "}
|
||||||
|
|
||||||
|
|
||||||
|
def yn(ok):
|
||||||
|
return "✓" if ok else "⚠"
|
||||||
|
|
||||||
|
|
||||||
|
def pretty(obj, report):
|
||||||
|
out = []
|
||||||
|
out.append(f"Store: {obj.get('store_name', '')}")
|
||||||
|
if obj.get("store_branch"):
|
||||||
|
out.append(f"Branch: {obj['store_branch']}")
|
||||||
|
if obj.get("store_address"):
|
||||||
|
out.append(f"Address: {obj['store_address']}")
|
||||||
|
out.append("")
|
||||||
|
|
||||||
|
items = obj.get("items", [])
|
||||||
|
checks = report["per_item"]
|
||||||
|
out.append(f"Items ({len(items)}):")
|
||||||
|
for it, chk in zip(items, checks):
|
||||||
|
code = it.get("code", "")
|
||||||
|
head = f" {MARK[chk['status']]} {it.get('name', '')} ${it.get('price', 0):.2f}"
|
||||||
|
if code:
|
||||||
|
head += f" [{code}]"
|
||||||
|
out.append(head)
|
||||||
|
for extra in it.get("rows", [])[1:]: # detail rows only
|
||||||
|
out.append(f" · {extra}")
|
||||||
|
if chk["status"] == "warn":
|
||||||
|
out.append(f" ⚠ {chk['note']}")
|
||||||
|
out.append("")
|
||||||
|
|
||||||
|
# Receipt-level balance
|
||||||
|
out.append("Reconciliation:")
|
||||||
|
out.append(
|
||||||
|
f" Items sum: ${report['items_sum']:.2f}"
|
||||||
|
+ (
|
||||||
|
f" (printed subtotal ${report['subtotal_printed']:.2f}) {yn(report['subtotal_ok'])}"
|
||||||
|
if report["subtotal_printed"] is not None
|
||||||
|
else ""
|
||||||
|
)
|
||||||
|
)
|
||||||
|
if report["total_printed"] is not None:
|
||||||
|
out.append(
|
||||||
|
f" + tax/total: ${report['total_computed']:.2f}"
|
||||||
|
f" (printed TOTAL ${report['total_printed']:.2f}) {yn(report['total_ok'])}"
|
||||||
|
)
|
||||||
|
if report["unit_count_printed"] is not None:
|
||||||
|
out.append(
|
||||||
|
f" Unit count: {report['unit_count']}"
|
||||||
|
f" (printed {report['unit_count_printed']} ITEMS) {yn(report['unit_count_ok'])}"
|
||||||
|
)
|
||||||
|
|
||||||
|
meta = obj.get("metadata", {})
|
||||||
|
if meta:
|
||||||
|
out.append("")
|
||||||
|
out.append("Metadata:")
|
||||||
|
for k, v in meta.items():
|
||||||
|
out.append(f" {k}: {v}")
|
||||||
|
return "\n".join(out)
|
||||||
|
|
||||||
|
|
||||||
|
def main():
|
||||||
|
args = [a for a in sys.argv[1:] if not a.startswith("--")]
|
||||||
|
as_json = "--json" in sys.argv
|
||||||
|
path = args[0] if args else DEFAULT_INPUT
|
||||||
|
if not os.path.exists(path):
|
||||||
|
sys.exit(f"input not found: {path}")
|
||||||
|
|
||||||
|
with open(path, encoding="utf-8") as f:
|
||||||
|
rows_text = f.read()
|
||||||
|
|
||||||
|
obj = group(rows_text)
|
||||||
|
|
||||||
|
if as_json:
|
||||||
|
obj["validation"] = validate(obj)
|
||||||
|
print(json.dumps(obj, ensure_ascii=False, indent=2))
|
||||||
|
else:
|
||||||
|
# Just the groups: each purchased item, with the rows that belong together.
|
||||||
|
items = obj.get("items", [])
|
||||||
|
width = len(str(len(items)))
|
||||||
|
for i, it in enumerate(items, 1):
|
||||||
|
rows = it.get("rows", [])
|
||||||
|
head = rows[0] if rows else f"{it.get('name', '')} {it.get('price', 0):.2f}"
|
||||||
|
print(f"{str(i).rjust(width)} {head}")
|
||||||
|
for extra in rows[1:]:
|
||||||
|
print(f"{' ' * width} {extra}")
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == "__main__":
|
||||||
|
main()
|
||||||
87
OCR/receipt_ocr.py
Executable file
87
OCR/receipt_ocr.py
Executable file
|
|
@ -0,0 +1,87 @@
|
||||||
|
#!/usr/bin/env python3
|
||||||
|
"""Local receipt OCR: given a receipt image, print an ordered list of rows of text.
|
||||||
|
|
||||||
|
Uses a local Qwen2.5-VL model served by ollama. No cloud calls, no API key.
|
||||||
|
Zero third-party dependencies — talks to the ollama HTTP API with the stdlib.
|
||||||
|
|
||||||
|
Usage:
|
||||||
|
python3 receipt_ocr.py [image.jpg] # pretty, numbered rows
|
||||||
|
python3 receipt_ocr.py image.jpg --json # raw JSON {"rows": [...]}
|
||||||
|
|
||||||
|
Env:
|
||||||
|
OLLAMA_HOST default http://127.0.0.1:11434
|
||||||
|
OLLAMA_MODEL default qwen2.5vl:7b
|
||||||
|
"""
|
||||||
|
|
||||||
|
import base64
|
||||||
|
import json
|
||||||
|
import os
|
||||||
|
import sys
|
||||||
|
import urllib.request
|
||||||
|
|
||||||
|
DEFAULT_IMAGE = os.path.join(os.path.dirname(__file__), "example", "data", "aldi.jpg")
|
||||||
|
HOST = os.environ.get("OLLAMA_HOST", "http://127.0.0.1:11434").rstrip("/")
|
||||||
|
MODEL = os.environ.get("OLLAMA_MODEL", "qwen2.5vl:7b")
|
||||||
|
|
||||||
|
PROMPT = (
|
||||||
|
"You are an OCR engine for retail receipts. Transcribe EVERY line of text on "
|
||||||
|
"this receipt, from top to bottom, exactly as printed. Each physical row of the "
|
||||||
|
"receipt becomes one string in the output, with its columns joined left-to-right "
|
||||||
|
"and separated by single spaces (e.g. an item code, its name, and its price all "
|
||||||
|
"go in the same row). Do not merge or split rows, do not reorder, do not add "
|
||||||
|
"commentary, do not correct spelling, do not invent text. Preserve numbers and "
|
||||||
|
"punctuation verbatim. Return strict JSON only."
|
||||||
|
)
|
||||||
|
|
||||||
|
# Structured-output schema: forces the model to return {"rows": ["...", ...]}.
|
||||||
|
SCHEMA = {
|
||||||
|
"type": "object",
|
||||||
|
"properties": {"rows": {"type": "array", "items": {"type": "string"}}},
|
||||||
|
"required": ["rows"],
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
def ocr(image_path):
|
||||||
|
with open(image_path, "rb") as f:
|
||||||
|
b64 = base64.b64encode(f.read()).decode("ascii")
|
||||||
|
|
||||||
|
payload = {
|
||||||
|
"model": MODEL,
|
||||||
|
"messages": [{"role": "user", "content": PROMPT, "images": [b64]}],
|
||||||
|
"stream": False,
|
||||||
|
"format": SCHEMA,
|
||||||
|
"options": {"temperature": 0, "num_ctx": 8192},
|
||||||
|
}
|
||||||
|
|
||||||
|
req = urllib.request.Request(
|
||||||
|
f"{HOST}/api/chat",
|
||||||
|
data=json.dumps(payload).encode("utf-8"),
|
||||||
|
headers={"Content-Type": "application/json"},
|
||||||
|
)
|
||||||
|
with urllib.request.urlopen(req, timeout=600) as resp:
|
||||||
|
body = json.load(resp)
|
||||||
|
|
||||||
|
content = body["message"]["content"]
|
||||||
|
return json.loads(content)["rows"]
|
||||||
|
|
||||||
|
|
||||||
|
def main():
|
||||||
|
args = [a for a in sys.argv[1:] if not a.startswith("--")]
|
||||||
|
as_json = "--json" in sys.argv
|
||||||
|
image_path = args[0] if args else DEFAULT_IMAGE
|
||||||
|
|
||||||
|
if not os.path.exists(image_path):
|
||||||
|
sys.exit(f"image not found: {image_path}")
|
||||||
|
|
||||||
|
rows = ocr(image_path)
|
||||||
|
|
||||||
|
if as_json:
|
||||||
|
print(json.dumps({"rows": rows}, ensure_ascii=False, indent=2))
|
||||||
|
else:
|
||||||
|
width = len(str(len(rows)))
|
||||||
|
for i, row in enumerate(rows, 1):
|
||||||
|
print(f"{str(i).rjust(width)} {row}")
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == "__main__":
|
||||||
|
main()
|
||||||
Loading…
Reference in a new issue