Log Error Parser

Implement top_errors(log_text, limit=3) for application logs.

Requirements

• Input is one string containing newline-separated log lines.
• Count only lines that contain " ERROR ".
• The error key is the entire suffix after the first " ERROR " marker, including any later occurrences.
• Strip leading and trailing whitespace from the extracted message.
• Exclude empty messages, including whitespace-only messages after stripping.
• Ignore indented continuation lines for counting.
• Return a list of (message, count) pairs.
• Sort by count descending, then message ascending.
• Return at most limit rows.
• If limit <= 0, return an empty list.

Example

The repeated timeout must rank first. The one-count message follows it.

log_text = '''
2026-01-01 INFO start
2026-01-01 ERROR timeout
2026-01-01 ERROR bad gateway
2026-01-01 ERROR timeout
'''

assert top_errors(log_text, limit=2) == [("timeout", 2), ("bad gateway", 1)]

Constraints

• Use standard-library Python only.
• Prefer splitlines, Counter, or an explicit state machine.
• limit is an integer. Messages are case-sensitive; preserve internal whitespace. Only trim the ends.
• A line is indented when its first character is whitespace. Ignore it even if it contains the marker. Accept LF and CRLF line endings.

The marker uses literal ASCII spaces and uppercase ERROR: "t\tERROR\tfailed" does not count. This exercise matches a delimiter anywhere on an unindented line; it does not validate timestamps or parse a structured severity field.