JSON Lines (NDJSON) Viewer

JSON Lines, also called NDJSON, is one JSON value per line — the format that streaming logs, machine-learning datasets and tail-friendly pipelines settled on. This viewer parses each line independently, so one malformed record does not poison the file: you get the parsed table and, separately, the bad lines with the exact parser error for each. A field:value filter narrows by content and accepts a nested.path as well as a top-level key, and a column picker keeps only the fields you want in the re-export.

The format exists because a JSON array cannot be appended to. Adding a record to [...] means rewriting the closing bracket, so a crashed writer leaves a file no parser will touch, and a reader has to hold the whole array in memory before it sees the first element. One value per line fixes all three: append by writing a line, recover by discarding the last partial one, and process a hundred-gigabyte file a line at a time.

What makes line-splitting safe rather than merely convenient is a rule in the JSON grammar itself: a literal newline is never allowed inside a string. A line feed in a value has to be written as the two characters \n, so no newline in a well-formed document can be anything other than a record boundary. Splitting on newlines is therefore correct, not a heuristic — which is not true of CSV, where a quoted field may contain a real line break and naive line splitting silently corrupts rows.

Two conventions to respect when you write these files. The encoding is UTF-8 and nothing else, so a file emitted as UTF-16 or Latin-1 will not interoperate however valid the JSON looks locally. And each line should be a complete value on its own — pretty-printing is exactly what the format forbids, since an indented object spans many lines and each of those lines is invalid alone. If a file will not load anywhere, checking that first usually explains it. Everything here runs in the browser; nothing is uploaded.