Fix incorrect category display in report for same merchant names

[terryaney]

Fixes #88 allowing merchants to different categorization rules with same name.

Instead of requiring:

[Costco Grocery]
match: contains("COSTCO") and amount <= 200
category: Food
subcategory: Grocery

[Costco Bulk]
match: contains("COSTCO") and amount > 200
category: Shopping
subcategory: Wholesale

Can instead be:

[Costco]
match: contains("COSTCO") and amount <= 200
category: Food
subcategory: Grocery

[Costco]
match: contains("COSTCO") and amount > 200
category: Shopping
subcategory: Wholesale

Allows easier auditing of spending to a merchant. Of course, if you prefer unique merchant names, that is still possible.

[github-actions]

PR Build Available

Version: 0.0.91-e1adce4

Install from this PR

Linux / macOS:

curl -fsSL https://raw.githubusercontent.com/davidfowl/tally/main/docs/install-pr.sh | bash -s -- 91

Windows PowerShell:

iex "& { $(irm https://raw.githubusercontent.com/davidfowl/tally/main/docs/install-pr.ps1) } 91"

Manual download: View workflow run and download artifacts.

Requirements

• GitHub CLI (gh) must be installed and authenticated
• Run gh auth login if not already authenticated

[terryaney]

The first commit on the PR correctly fixes the Python-side bug — changing by_merchant to use a composite (merchant, category, subcategory) key so same-named merchants with different categories appear as separate rows. This significantly reduces the practical impact of the JavaScript-side bug described here.

However, even after that fix, merchant.tags in the report data is still the union of all transaction tags within that merchant entry. When that union includes tags that not every transaction individually carries, filteredViewTotals misclassifies those transactions. The second commit on my PR addresses this.

Root Cause

In spending_report.js, both affected computed properties hoist the tags lookup outside the transaction loop:

// filteredViewTotals
const tags = merchant.tags || [];          // ← merchant-level aggregate
const txns = merchant.filteredTxns || merchant.transactions || [];
for (const txn of txns) {
    const c = categorizeAmount(txn.amount || 0, tags);  // ← all txns get same tags

// chartAggregations
const tags = merchant.tags || [];          // ← same problem
for (const txn of merchant.filteredTxns || []) {
    const c = categorizeAmount(txn.amount, tags);

Each individual transaction already carries its own tags array in the report JSON (e.g., {"id": "...", "date": "06/22", "amount": 33.51, "tags": ["monthly-bill"]}), so the data is available.

[Copilot]

Pull request overview

This PR updates merchant aggregation/reporting so identical merchant display names can appear as separate rows when categorized differently, while keeping merchant-name filtering logical across those category variants.

Changes:

• Uses composite merchant identity in analysis/report data for same-name merchants with different category/subcategory values.
• Updates HTML filtering/autocomplete behavior to filter by visible merchant name across duplicate category variants.
• Adds regression coverage for duplicate merchant names and updates tests for composite row IDs and “Uncategorized” fallback text.

Show a summary per file

File	Description
src/tally/analyzer.py	Changes merchant aggregation keys to include category/subcategory and propagates merchant names from data.
src/tally/report.py	Builds composite merchant IDs and category-view rows for duplicate merchant names.
src/tally/spending_report.js	Adjusts merchant filtering/autocomplete and transaction-level categorization in filtered totals/charts.
src/tally/commands/explain.py	Updates explain lookup/filter paths for duplicate merchant-name entries.
tests/test_report_html.py	Adds duplicate-name regression tests and updates row locators/fallback expectations.

Copilot's findings

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

• Files reviewed: 5/5 changed files
• Comments generated: 5

[davidfowl]

There are some pr comments from copilot

[terryaney]

There are some pr comments from copilot

Working on them and testing. Curious, I used Copilot to help code up this branch and 4-5 comments above are troubling to me in that I feel Copilot should have found those (all the json format issues). Given that, I'm curious what kind of prompt is running the copilot review? Something custom you would share? Or just a standard review? And my plan/implementation via Copilot missed thinking about these somehow?