HAR Request Extractor

About this tool

HAR Request Extractor pulls the request list out of a HAR (HTTP Archive) capture — the JSON file browser DevTools export from the Network tab via Save all as HAR. Instead of scrolling a 40 MB JSON blob, you get one line per request:

• Method + URL — what was requested.
• Status — the HTTP status code; failed requests that never got a response are listed with status 0.
• Type — the response mimeType, with ; charset=… parameters stripped.
• Size — bytes on the wire: response.bodySize when the exporter recorded it, else Chrome's _transferSize, else the decoded content.size. Unknown sizes show as -.
• Time — the entry's total time in milliseconds.

The # column is each request's position in the original capture, so a row stays findable in DevTools even after filtering or sorting.

Filters and sorting

• Status filter: 2xx, 3xx, 4xx, 5xx, or errors (4xx + 5xx + failed status-0 requests).
• Method filter: case-insensitive exact match (GET, POST, …).
• URL contains: case-insensitive substring, e.g. /api/, .png, or a hostname.
• Sort: capture order, slowest first, or largest first.

Output formats

• Table — aligned text columns plus a summary line (n of m requests · X transferred; the byte count sums the listed rows).
• CSV — header + one row per request (index,method,url,status,status_text,mime_type,size_bytes,time_ms,started), ready for a spreadsheet.
• JSON — an array of request objects with the same fields (unknown values are null).
• URL list — one URL per line, in row order, ready to pipe.

Worked example

A three-request capture (an HTML page, a slow API call, a missing image):

{"log":{"entries":[
  {"startedDateTime":"2024-01-01T00:00:00.000Z","time":102.5,
   "request":{"method":"GET","url":"https://example.com/"},
   "response":{"status":200,"statusText":"OK","content":{"mimeType":"text/html","size":5120},"bodySize":2048}},
  {"startedDateTime":"2024-01-01T00:00:01.000Z","time":812,
   "request":{"method":"POST","url":"https://example.com/api/search"},
   "response":{"status":200,"statusText":"OK","content":{"mimeType":"application/json","size":20480},"bodySize":-1,"_transferSize":10240}},
  {"startedDateTime":"2024-01-01T00:00:02.000Z","time":54,
   "request":{"method":"GET","url":"https://cdn.example.com/logo.png"},
   "response":{"status":404,"statusText":"Not Found","content":{"mimeType":"image/png","size":0},"bodySize":512}}
]}}

extracts to this table:

3 of 3 requests · 12.5 KB transferred

#  METHOD  STATUS  TYPE              SIZE     TIME    URL
1  GET     200     text/html         2.0 KB   103 ms  https://example.com/
2  POST    200     application/json  10.0 KB  812 ms  https://example.com/api/search
3  GET     404     image/png         512 B    54 ms   https://cdn.example.com/logo.png

With the status filter on Errors, only row 3 remains; with Slowest first, the 812 ms API call moves to the top (and keeps its #2).

Limits and edge cases

• Only the request LIST is extracted — request/response headers, cookies, and bodies are never read into the output, so the result is safe to share even when the capture itself isn't.
• The extractor is deliberately forgiving: entries missing fields still list, with - (table/CSV) or null (JSON) for the unknowns. Only non-JSON input or JSON without a log.entries array is an error.
• Times are rounded to whole milliseconds in the table; CSV/JSON keep the exact values. Sizes in the table are 1024-based (KB/MB).
• Everything runs in one pass in your browser's WebAssembly — captures of tens of thousands of entries are fine; a multi-hundred-MB capture is limited by your browser's memory, not by an upload cap.

Handy for

• Turning a bug-report HAR into a shareable request summary without leaking cookies or auth headers.
• Finding the slowest or heaviest requests behind a slow page load.
• Pulling every API call (/api/, graphql, …) out of a session as CSV for a spreadsheet or JSON for a script.

FAQ