@dev-pi2pie/word-counter 0.1.3-canary.2 → 0.1.4-canary.1

package/README.md

@@ -139,6 +139,29 @@ word-counter --path ./examples/test-case-multi-files-support --keep-progress
 
 Progress is transient by default, auto-disabled for single-input runs, and suppressed in `--format raw` and `--format json`.
 
+### Batch Concurrency (`--jobs`)
+
+Use `--jobs` to control batch concurrency:
+
+```bash
+word-counter --path ./examples/test-case-multi-files-support --jobs 1
+word-counter --path ./examples/test-case-multi-files-support --jobs 4
+```
+
+Quick policy:
+
+- no `--jobs` and `--jobs 1` are equivalent baseline behavior.
+- `--jobs > 1` enables concurrent `load+count`.
+- if requested `--jobs` exceeds host `suggestedMaxJobs` (from `--print-jobs-limit`), the CLI warns and runs with the suggested limit as a safety cap.
+
+Inspect host jobs diagnostics:
+
+```bash
+word-counter --print-jobs-limit
+```
+
+For full policy details, JSON parity expectations (`--misc`, `--total-of whitespace,words`), and benchmark standards, see [`docs/batch-jobs-usage-guide.md`](docs/batch-jobs-usage-guide.md).
+
 ### Stable Path Resolution Contract
 
 - Repeated `--path` values are accepted as mixed inputs (file + directory).
@@ -591,27 +614,10 @@ Example JSON (trimmed):
 
 ## Locale Tag Detection Notes
 
-- Detection is regex/script based (Unicode script checks), not a statistical language-ID model.
-- Ambiguous Latin text uses `und-Latn` unless a Latin hint is provided.
-- Han-script fallback uses `und-Hani` by default because regex script checks cannot natively distinguish `zh-Hans` vs `zh-Hant`.
-- Current built-in Latin diacritic heuristics include:
-  - `de`: `äöüÄÖÜß`
-  - `es`: `ñÑ¿¡`
-  - `pt`: `ãõÃÕ`
-  - `fr`: `œŒæÆ`
-  - `pl`: `ąćęłńśźżĄĆĘŁŃŚŹŻ`
-  - `tr`: `ıİğĞşŞ`
-  - `ro`: `ăĂâÂîÎșȘțȚ`
-  - `hu`: `őŐűŰ`
-  - `is`: `ðÐþÞ`
-- Latin text with other European diacritics may still remain in `und-Latn` unless a hint is provided.
-- Use `--mode chunk`/`--mode segments` or `--format json` to see the exact locale tag assigned to each chunk.
-- Regex/script-only detection cannot reliably identify English vs. other Latin-script languages; 100% certainty requires explicit metadata (document language tags, user-provided locale, headers) or a language-ID model.
-- Use `--latin-language <tag>` or `--latin-tag <tag>` for ambiguous Latin text.
-- Use `--latin-hint <tag>=<pattern>` (repeatable) and `--latin-hints-file <path>` to add custom Latin rules.
-- Use `--no-default-latin-hints` to disable built-in Latin diacritic rules.
-- Use `--han-language <tag>` or `--han-tag <tag>` for Han-script fallback.
-- `--latin-locale` remains supported as a legacy alias for now and is planned for future deprecation.
+- Detection is regex/script based, not statistical language-ID.
+- Ambiguous Latin defaults to `und-Latn`; Han fallback defaults to `und-Hani`.
+- Use explicit tag and hint flags when you need deterministic tagging.
+- Full notes (built-in heuristics, limitations, and override guidance) are tracked in `docs/locale-tag-detection-notes.md`.
 
 ## Breaking Changes Notes