Rules and CSV

Moderation rules and CSV import/export

Word/phrase rules are project-wide; rejection takes precedence over review when several rules match. The rule list supports phrase search, action/enabled filters and 25 records per page, retaining filters across pages. Adding and editing use separate forms; all rule endpoints require the project-wide rules capability.

Import/export buttons support bulk rule management. Export downloads all rules as UTF-8 CSV with a BOM, independent of active list filters. The columns are phrase, action (review / reject) and enabled (1 / 0; import also accepts true / false). Import accepts comma or semicolon separators and preserves quoted Unicode phrases, commas and line breaks:

phrase,action,enabled
spam,review,1
unwanted phrase,reject,1

Import adds new phrases and updates action/enabled on exact phrase matches; unmentioned rules remain unchanged. Reimporting the same file is idempotent. Invalid rows, duplicate phrases in the file or ambiguous existing matches abort the entire import and show a row error. Spreadsheet-sensitive phrase prefixes are escaped with an apostrophe on export and restored on import; literal leading apostrophes are doubled. Use the exported format when transferring such phrases. Default limits are 5 MiB and 20,000 rules per import, overridable only in COMMENTAIL["default"] as RULES_CSV_MAX_BYTES and RULES_CSV_MAX_ROWS.

How text matches

Rules run only in Apply rules mode. Matching uses complete word-token sequences, case-insensitively; punctuation is ignored, whitespace/line breaks are normalized, and Russian ё equals е. A word does not match inside another word. Displayed text is unchanged. Rejection wins over review when multiple rules match.

Bulk workflow

1. Export the existing list as a backup. Export includes all rules, regardless of filters.
2. Keep the column names and UTF-8 encoding when editing in a spreadsheet.
3. Import a small sample and inspect the result before a large batch.
4. Reimporting identical phrases updates without creating duplicates.

Import is an upsert, not a replacement/deletion tool. Literal phrase equality is the CSV key; different phrases that normalize identically are not automatically merged. A single invalid row aborts the entire import.

---

← Moderation · Documentation home · Guest comments →