candle-annotator/openspec/specs/data-ingestion/spec.md at 4121a87875278d0b4e93a8e0c4ef785caa31b809

Marko Djordjevic 4121a87875 sync: apply multi-chart-management delta specs to main specs

- Created new chart-management capability spec
- Updated data-ingestion: chart-scoped candles, duplicate filename handling
- Updated backend-api: all endpoints gain chartId parameter, chart CRUD
- Updated chart-canvas: chart switching, scoped data fetching
- Updated label-management: annotations scoped to active chart
- Updated ui-shell: upload creates/selects chart, theme-aware link styling

2026-02-13 09:06:37 +01:00

3.3 KiB

Raw Blame History

ADDED Requirements

Requirement: CSV file upload

The system SHALL provide a file upload component that accepts CSV files containing OHLC candle data. The CSV format MUST have columns: time, open, high, low, close. The time column SHALL accept both YYYY-MM-DD date strings and Unix timestamps (integer seconds). Uploading a CSV SHALL create a new chart (named from the filename without extension) and insert all candle rows associated with that chart, rather than replacing existing data.

Scenario: Valid CSV upload

WHEN user uploads a CSV file with valid headers (time, open, high, low, close) and valid data rows
THEN system creates a new chart named from the filename (without .csv extension), parses all rows, and stores them in the candles table with the new chart's chart_id

Scenario: CSV with Unix timestamps

WHEN user uploads a CSV where the time column contains Unix timestamps (e.g., 1700000000)
THEN system stores the timestamps as integers in the database with the correct chart_id and renders candles correctly on the chart

Scenario: CSV with date strings

WHEN user uploads a CSV where the time column contains date strings (e.g., "2024-01-15")
THEN system converts dates to Unix timestamps and stores them in the database with the correct chart_id

Scenario: Invalid CSV format

WHEN user uploads a CSV missing required headers or containing malformed data
THEN system displays an error message describing the issue, does not create a chart, and does not store any partial data

Scenario: Duplicate filename upload

WHEN user uploads a CSV whose filename (without extension) matches an existing chart name
THEN system appends a numeric suffix to the chart name (e.g., "btc-daily-2") and creates a new chart with the suffixed name

Requirement: CSV parsing with papaparse

The system SHALL use the papaparse library for CSV parsing. Parsing SHALL handle large files by using streaming mode for files exceeding 10,000 rows. Parsed records SHALL be inserted into SQLite within a single database transaction for atomicity.

Scenario: Large file parsing

WHEN user uploads a CSV with more than 10,000 rows
THEN system uses streaming parse and batch inserts within a transaction, completing without memory issues

Scenario: Transaction atomicity

WHEN a parse error occurs midway through a CSV file
THEN system rolls back the entire transaction and no partial data is stored

Requirement: Candles database table

The system SHALL store candle data in a candles table with columns: id (integer primary key, auto-increment), chart_id (integer, foreign key to charts.id, NOT NULL), time (integer, Unix timestamp), open (real), high (real), low (real), close (real). The table MUST have a composite unique constraint on (chart_id, time).

Scenario: Schema structure

WHEN the database is initialized
THEN the candles table exists with all required columns including chart_id and the composite unique constraint on (chart_id, time)

Scenario: Same timestamp across different charts

WHEN two different charts have candles with the same Unix timestamp
THEN both records are stored successfully because the unique constraint is per-chart

3.3 KiB Raw Blame History