Search results

new_leads uses SELECT DISTINCT ... ORDER BY l.id while l.id isn’t part of the DISTINCT projection. If multiple lead rows share the same (company, location, source, enrichment), SQLite may pick an arbitrary row for the DISTINCT result, making the “stable FIFO order” claim unreliable. Consider selecting the minimum id per lead (e.g., via GROUP BY / MIN(id)) or avoiding DISTINCT and enforcing uniqueness at insert time.

Status: Open.

#33 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

load_history_file claims to ignore lines beginning with #, but the filter checks ln.startswith("#") before stripping. Lines like ' # comment' will be inserted into history; consider checking ln.lstrip().startswith("#") (or using the stripped value) to match the documented behaviour and _load_history’s parsing.

Status: Open.

#32 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

The implementation computes normalised term-frequency weights, but the API/docs call it “TF‑IDF” (_tfidf, module header, _make_vector, search_leads). This is misleading for maintainers/users; either implement an actual IDF component (requires corpus stats) or rename/update docs to reflect TF-only vectors.

Status: Open.

#31 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

sqlite3.connect(..., check_same_thread=False) disables SQLite’s thread-affinity safety checks. This repo doesn’t appear to use threads, so this weakens safety without a clear need; consider leaving the default (True) or documenting/encapsulating the concurrency model (e.g., one connection per thread) if cross-thread access is required.

Status: Open.

#30 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

This test asserts the connection is closed using pytest.raises(Exception), which is very broad and could mask unrelated failures. Since sqlite3 raises a specific exception for operations on a closed connection, prefer asserting that specific exception type to keep the test tight and diagnostic.

Status: Open.

#29 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

run() starts a DB run (start_run) and later calls finish_run, but there’s no try/finally. If an exception occurs mid-run (fetching, extraction, validation, file I/O), the SQLite run record will never get finished_at and the DB connection may remain open. Consider wrapping the body in try/finally to always close the store and to finalize the run (even if only to mark it failed).

Status: Open.

#28 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

Processor.process accepts run_id and the docstring says it’s “forwarded to the store”, but the parameter is currently unused. Either remove it to avoid a misleading API surface, or actually use it (e.g., to scope store queries/updates to a run).

Status: Open.

#27 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

logger.exception(...) is called when html is None, but there is no active exception in that branch (it’s not inside an except). This logs a misleading NoneType: None traceback. Use logger.error(...)/logger.warning(...), or move the logging into _fetch_search_page where the exception is caught so the real error is recorded.

Status: Open.

#25 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

new_tab() catches a blanket Exception and returns None. In asyncio this will also swallow asyncio.CancelledError, preventing task cancellation from propagating correctly. Catch a narrower exception set (e.g., RuntimeError from _get_json) and let cancellations bubble up.

Status: Open.

#24 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

extra_403_retries currently doesn’t increase the total number of attempts beyond max_retries because the loop is bounded by while attempt < max_retries and 403 retries still increment attempt. If the intent is that 403s get additional attempts on top of the base retry budget (as the name/config docs suggest), adjust the accounting (e.g., separate counters or a larger total-attempt cap). If not, consider renaming/clarifying docs to avoid implying extra total retries.

Status: Open.

#23 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

Same issue as above: logger.exception(...) is invoked when html is None without an active exception, producing noisy/misleading tracebacks. Prefer logger.error(...) here, or log the caught exception inside _fetch_search_page.

Status: Open.

#22 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Apr 28, 2026

Lighthouse Audit: Improve index.html Performance, Accessibility, and Best Practices

data-qualityValidation, deduplication, and ensuring output accuracy (e.g., phone/email rules).Validation, deduplication, and ensuring output accuracy (e.g., phone/email rules).

documentationWriting/updating docs, runbooks, manpages, and README sections.Writing/updating docs, runbooks, manpages, and README sections.

opsOperational aspects like scheduling, logging, monitoring, and production readiness.Operational aspects like scheduling, logging, monitoring, and production readiness.

qaQuality assurance, manual testing, and final verification runs.Quality assurance, manual testing, and final verification runs.

reliabilityImproving fetch reliability, retries, timeouts, and error handling.Improving fetch reliability, retries, timeouts, and error handling.

Status: Open.

#19 In 2MuchC0ff33/elvis;

· 2MuchC0ff33 opened on Jan 18, 2026

·  

v1.0.1