[omatheusmesmo]

Describe Changes

The completion display assumes a candidate always extends the typed word by a shared prefix: a single candidate is appended after the prefix and multiple candidates insert their common prefix. That corrupts the buffer for completion styles where the typed word is not a prefix of the candidates, such as the fuzzy matching planned for aeshell/aesh#510: a typo like comit with candidate commit appends the full name and produces comitcommit.

This adds an opt-in replacingCurrentWord flag to CompleteOperation. When set, a single candidate removes the whole typed word, dashes included, before the full candidate is written, and common-prefix insertion is skipped in favour of listing. The removal uses unix-word-rubout because backward-kill-word treats - as a word delimiter and leaves typed dashes behind. The flag defaults to false, so existing prefix completion behavior is unchanged.

It also moves the org.aesh.readline.fuzzy package from the readline module to readline-api, unchanged in name, so that the aesh completion fallback loads for consumers that exclude the full readline module: jbang, for example, depends on aesh with readline excluded plus readline-api. Nothing outside the package location changes; the fuzzy history search in readline resolves the classes through its readline-api dependency and all its tests keep passing.

---

• Supports Fuzzy/camel-case completion matching aesh#510

[stalep]

Thanks for this, and sorry for the slow turnaround. Two requests before merge — one naming, one build hygiene. I'm fine with the fuzzy package move to readline-api as designed.

1. Please rename isReplaceCurrentWord(), it doesn't read naturally.
   The interface convention is isX/setX with an adjective-like phrase (isIgnoreOffset, isAppendSeparator, isIgnoreStartsWith), but isReplaceCurrentWord pairs is with an imperative, so the call site reads ungrammatically:

if (!possibleCompletions.get(0).isIgnoreStartsWith()
        && !possibleCompletions.get(0).isReplaceCurrentWord())  // "is replace…?"

My preference is the present participle, which keeps the is/set symmetry and reads as plain English:
- isReplacingCurrentWord() / setReplacingCurrentWord(boolean)
Alternative, if you prefer a noun phrase in the style of isAppendSeparator: isWordReplacement() / setWordReplacement(boolean). Either works for me, just not the imperative under is.
2. Please don't hardcode the JUnit version in readline-api/pom.xml.
The parent pom already defines <junit.version>4.13.2</junit.version> (pom.xml:68) , the new test dependency should use ${junit.version}. (Yes, readline/pom.xml:91 hardcodes the same version; that's pre-existing, but if you're touching this anyway, aligning both lines would be a welcome cleanup, otherwise I'll file a follow-up.)

[stalep]

@omatheusmesmo let me know if you have any questions or disagree with my review

[omatheusmesmo]

Sorry @stalep , it's been a busy week, but I will get it done by the end of the weekend.

[stalep]

Sorry @stalep , it's been a busy week, but I will get it done by the end of the weekend.

Sure, no rush!

[omatheusmesmo]

Thanks @stalep, both points addressed and pushed:

• Renamed to isReplacingCurrentWord() / setReplacingCurrentWord(boolean)
• readline-api/pom.xml now uses ${junit.version}. I aligned readline/pom.xml too, so no follow-up is needed

I'll update aeshell/aesh#695 to the new method name next.

[stalep]

Thanks a lot, merged!