AltF Flow recipe

Filter a log file down to its unique errors

To see which distinct errors a log contains, keep only ERROR and FATAL lines, remove the timestamp at the start of each line, replace numbers such as IDs and durations with a placeholder, and remove duplicates. This recipe does that in four steps, turning hundreds of repeated lines into a short list of unique error messages in first-seen order.

  • Free, no sign-up
  • Runs in your browser
  • Nothing uploaded
  • Share recipes as a link

4 steps · runs in your browser · nothing uploaded

Pipeline

  1. Keep, Contains: \\b(ERROR|FATAL)\\b, Regular expression

    9 → 6 lines (-3)

  2. Find: ^\\S+[ T]\\S+\\s+, Regular expression

    6 → 6 lines

  3. Find: \\b\\d+\\b, Replace with: <n>, Regular expression

    6 → 6 lines

  4. Keep the first copy of each line and drop repeats.

    6 → 3 lines (-3)

9lines79words573chars

3lines21words149chars-6 lines vs input

The 4 steps in this recipe

  1. 1

    Filter lines - Keep, Contains: \\b(ERROR|FATAL)\\b, Regular expression

    Keep or remove lines that contain a word or match a pattern.

  2. 2

    Find and replace - Find: ^\\S+[ T]\\S+\\s+, Regular expression

    Replace text everywhere. Supports regular expressions with $1 back-references.

  3. 3

    Find and replace - Find: \\b\\d+\\b, Replace with: <n>, Regular expression

    Replace text everywhere. Supports regular expressions with $1 back-references.

  4. 4

    Remove duplicate lines

    Keep the first copy of each line and drop repeats.

Change any option in the workspace above, switch a step off to compare, or add your own steps. Save it to reuse it, or share the recipe as a link.

Worked example

Before
2026-10-31 09:14:02 INFO  Server started on port 8080
2026-10-31 09:14:05 ERROR Payment failed for order 1042: card declined
2026-10-31 09:14:06 WARN  Slow query took 812 ms
2026-10-31 09:15:11 ERROR Payment failed for order 1077: card declined
2026-10-31 09:15:40 INFO  User 55 signed in
2026-10-31 09:16:02 ERROR Timeout calling https://api.example.com after 30 s
2026-10-31 09:16:30 ERROR Payment failed for order 1093: card declined
2026-10-31 09:17:45 FATAL Database connection lost (pool 2)
2026-10-31 09:18:02 ERROR Timeout calling https://api.example.com after 30 s
After
ERROR Payment failed for order <n>: card declined
ERROR Timeout calling https://api.example.com after <n> s
FATAL Database connection lost (pool <n>)

Why each step is there

  • Filter lines keeps lines with the whole word ERROR or FATAL. Ignore case is off so a message that merely mentions "error" in lowercase is not picked up.
  • The first replace removes the leading date and time, which make every line unique.
  • The second replace turns standalone numbers into <n>, so "order 1042" and "order 1077" become the same message.
  • Remove duplicate lines then leaves one line per distinct error.

Adapting it to your log format

If your timestamps look like [31/Oct/2026:09:14:02 +0000], change the first replace to ^\[[^\]]*\]\s*. To count how often each error happens rather than just listing it, run the recipe without the last step and paste the output into a word frequency counter. Add WARN to the filter pattern to include warnings.

Filter errors from a log: questions and answers

Why mask numbers instead of keeping them?

IDs and durations make otherwise identical errors look different. Delete step 3 if you need the exact values.

Can it handle multi-line stack traces?

Filter lines works line by line, so only the line with the level is kept. Stack frame lines below it are dropped.

How large a log can I paste?

Up to 2 MB per run. For bigger files, cut out the time window you care about first.

More recipes like this

Developers and data

List to SQL IN clauseTurn a pasted column of IDs or codes into a SQL IN (...) list: trimmed, deduped, single-quoted with apostrophes escaped and joined with commas.
  1. Trim every line
  2. Remove empty lines
  3. Remove duplicate lines
  4. Quote each line
  5. +2 more
CSV to JSON with cleanupTrim stray spaces around CSV values and drop blank rows before converting to a JSON array of objects keyed by the header row. Runs locally.
  1. Trim whitespace
  2. Remove empty lines
  3. Trim every line
  4. Find and replace
  5. +1 more
JSON to CSV, sorted columnsConvert a JSON array of objects to CSV with a header row, sorting keys first so columns come out in a predictable A-Z order. Free, no upload.
  1. Format JSON
  2. JSON to CSV
Base64 to pretty JSONPaste a Base64 or Base64url string, including wrapped lines, and get readable, indented JSON. Useful for webhook payloads and encoded configs.
  1. Encode or decode
  2. Format JSON
Wrap text at 72 charactersTrim a commit message or plain-text email, squeeze extra blank lines to one and hard wrap every paragraph at 72 characters on word boundaries.
  1. Trim whitespace
  2. Remove empty lines
  3. Wrap at a width
See all 22 recipes

Every step you can add (41)

Clean up

  • Trim whitespace
  • Trim every line
  • Remove empty lines
  • Collapse repeated spaces
  • Remove line breaks
  • Strip HTML tags
  • Straighten smart quotes
  • Remove accents
  • Remove punctuation
  • Remove emoji
  • Remove non-ASCII characters

Change case

  • Change case
  • Make URL slugs

Lines

  • Sort lines
  • Remove duplicate lines
  • Keep only duplicated lines
  • Compare with another list
  • Count repeated lines
  • Reverse line order
  • Shuffle lines
  • Keep a range of lines
  • Truncate long lines
  • Wrap at a width
  • Split into lines
  • Join lines

Find and filter

  • Find and replace
  • Filter lines
  • Remove words or characters

Extract

  • Extract

Data and formats

  • Format JSON
  • Minify JSON
  • CSV to JSON
  • JSON to CSV
  • Lines to JSON array
  • JSON array to lines

Encode and decode

  • Encode or decode

Build output

  • Add prefix / suffix
  • Quote each line
  • Number lines
  • Apply a line template
  • Add timestamp
What each step does, with its options

AltF Flow runs entirely in your browser. Your text is never sent to AltFTool; saved recipes stay in this browser and shared links carry the steps only.