Fix, Convert & View 20+ Data Formats — JSON, CSV, Parquet, Excel & Mock Webhook APIs · yellorn

Fix, convert, and view 20+ data formats in one editor. 29 auto-fix passes repair broken JSON, XML, YAML, CSV, and SQL from LLMs, log dumps, and Excel exports. Open binary files too — Parquet, Excel, Arrow, MessagePack, CBOR, Protobuf, BSON, gzip — plus AI model files (Safetensors, GGUF, ComfyUI). Convert across formats, publish payloads as mock REST APIs, and dispatch HTTP requests with saved templates.

Universal data-file viewer — 20 formats, from JSON to Parquet to Safetensors

• Text & structured formats (edit, fix, and convert): JSON, JSON Lines / NDJSON, XML, YAML, TOML, CSV, SQL, Python, and Raw Text — each with its own best-effort recovery ladder and Smart Format ladder.
• Binary data files (open and inspect, decoded to a JSON preview in a worker): Apache Parquet, Apache Arrow / Feather, Excel (.xlsx / .xlsm / .xls), MessagePack, CBOR, Protobuf (schema-less decode_raw), BSON (MongoDB), and gzip. Parquet and single-sheet Excel open straight into the sortable table view.
• AI / ML artifacts: Safetensors (header-only — tensor names, dtypes, shapes, parameter count, never the multi-GB weights), GGUF (llama.cpp model metadata + tensor map), and ComfyUI workflows extracted from a generated PNG (the API prompt + UI graph + a recipe summary: checkpoint, LoRAs, sampler, image size, resolved prompts).
• No upload required — binary files are decoded locally in your browser; large files are previewed with a byte cap so the tab stays responsive.

Data Doctor — repair broken structured data instantly

• 29 documented auto-fix passes across four formats: 15 for JSON (markdown fences, comments, Python literals, single quotes, unquoted keys, trailing commas, missing commas, auto-close brackets, Python timestamps / datetime / Decimal / tuples / sets), 5 for XML, 6 for YAML, 3 for CSV.
• LLM output cleanup — paste JSON from ChatGPT, Claude, or any LLM. Markdown fences and explanatory comments are stripped automatically; Python-style None / True / False are normalised; smart quotes from rich-text pastes are replaced.
• Server-log recovery — recover concatenated JSON documents glued by bad \X escapes and truncated "response": "{... log lines via the Smart Format catch path.
• Excel-on-Windows BOM removal — every UTF-8-BOM CSV export from Excel is repaired so the first column header doesn't end up as \uFEFFColumnName.
• Auditable fixes — every applied repair is labelled in the status bar ("Auto-fixed: trailing-commas, single-quotes, unquoted-keys"), so the user reviews and consents before publishing.

SQL auto-format — MySQL, Postgres, Athena, BigQuery, Snowflake

• Multi-dialect formatter — Smart Format (Cmd+Shift+F) auto-indents and uppercases keywords across MySQL, PostgreSQL, MS SQL Server, Amazon Athena (Trino/Presto), Google BigQuery, Snowflake, MariaDB, IBM DB2, and SQLite via sql-formatter.
• Best-effort SQL repair — pasted queries are cleaned of markdown ```sql fences, UTF-8 BOMs, smart quotes, and zero-width characters before formatting; every fix is labelled in the status bar.
• Simple / Safe / Compress modes — preserve keyword case (Simple), layout-only whitespace normalisation (Safe), or minify to a single line for embedding in code (Compress).
• Python escape hatch — Convert to Python emits a ready-to-run cursor.execute(dedent(...)) snippet so you can drop the query straight into your codebase.

Cross-format converter — JSON ↔ XML ↔ YAML ↔ CSV

• Convert JSON to XML, YAML, or CSV with a single click; CSV → JSON emits row-of-objects; XML → JSON preserves @_attr / #text / #cdata conventions.
• Python-dict clipboard export from every format (datetime roundtrips through datetime.fromisoformat, Decimal, tuples, sets).
• Lossy-conversion warnings (e.g. CSV ← deeply-nested JSON) surface a confirmation toast before re-shaping.
• Atomic undo — one Cmd+Z reverts both the text and the active format together.

Mock webhook REST API — publish payloads as live endpoints

• Click Publish, get a Cloudflare-edge HTTPS URL. All HTTP methods accepted at the slug; configurable response status code (200 / 401 / 429 / 500) to simulate failure paths.
• Format-aware Content-Type — publish XML and the URL serves application/xml; publish CSV and it serves text/csv.
• Owner-only request log portal at /webhook/<slug>: see method, headers, body, source IP, and timestamp for every incoming request.
• Free tier ships 5 active URLs with 24h TTL; Pro subscribers get 1,000 new URLs/month with persistent TTL while the subscription is active; Team gets 10,000/month at 300K calls/day.

Multi-view visualizer + JMESPath search

• Tree view — virtualized via react-window, supports inline edit, hover-menu actions (copy JMESPath path, copy value, set null, delete), and search highlighting.
• Graph view — node-link diagram via React Flow + Dagre layout in a worker; collapse / expand subgraphs.
• Table view — sortable header (worker-backed sort for large datasets), cell edit, copy column / row.
• JMESPath playground — type users[?age > `30`].name in the search bar to filter the tree live; match counts and Prev/Next navigation.

Request Sender — compose and dispatch HTTP requests

• Send GET, POST, PUT, PATCH, DELETE, HEAD, and OPTIONS requests directly from the browser with full header, auth, and body configuration.
• Save request templates for reuse — name, method, URL, headers, auth, and body are all persisted per account.
• Execution history tracks every dispatch with status, timing, response headers, and truncated response body (up to 1 MB).
• Built-in SSRF protection and loop guard prevent abuse; the sender adds an X-Yellorn-Internal-Dispatch header to block recursive calls to your own webhooks.

Cloud Backup — tabs, palette, and layout across devices

• Use Cloud Backup in the Account panel to back up your open tabs, pane split ratio, and active palette to every signed-in browser.
• Per-tab sync status badges in the tab bar show which tabs are synced, local-only, or too large to sync.
• Sync is quota-aware — usage bar and limits are visible in the Account panel.

Compare mode + smart format ladder

• Compare two tabs in Monaco's read-only diff editor (JSON sorted before diffing for stable output).
• Smart Format (Cmd+Shift+F) — full recovery + unwrap + type-convert.
• Simple Format (Alt+Shift+F) — semantic preservation; pretty-print without changing types.
• Safe Format — layout-only indent / line breaks; never re-parses.
• Compress (Alt+Shift+C) — minify whitespace.