Convert all locale files to MiniMessage by tastybento · Pull Request #555 · BentoBoxWorld/AOneBlock

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento · 2026-08-14T15:30:49Z

BentoBox text is MiniMessage now. The old &-codes still parse, so nothing is broken today — but leaving the files in the legacy format is what makes formatting bugs possible: any code that builds a fragment and drops it into a translation has to guess which format the target line is written in, and guessing wrong renders colour codes as literal text. (That is exactly the bug this started as, in ChunkBlock's chunk map.)

All 18 locale files are now MiniMessage.

How

Converted with BentoBox's own Util.legacyToMiniMessage, so each value now literally says what the runtime was already rendering — nothing hand-translated. Then tidied: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code (every line of a multi-line flag description, every run split by a placeholder), so those </aqua><aqua> seams are closed up.

# beforedescription: |- &b Rank that can break the magic &b block if they can break blocks.# afterdescription: |- <aqua>Rank that can break the magic block if they can break blocks.</aqua>

Verification

Every one of the 1506 values rendered before and after and compared per character, with the style each character ends up with. Nothing moved. (Comparing component trees was too strict — the tidy pass can nest <bold> inside <green> rather than the reverse, which is the same thing to a player.)
Placeholder checker reports the same 47 pre-existing issues across the translations as before, none new.
Comments, quoting style and numeric-looking values (wrap-at: '50') survive the round trip.
Full test suite green (570 tests).

Left alone deliberately

Ten values in vi, tr, it, es, fr, uk and ja contain a broken colour code — & c, & l, &un bloc — that never rendered as colour. They still render as literal text, exactly as today. Fixing them changes what players see, so that is a translator's call, not this PR's.

BentoBox text is MiniMessage now; the &-codes still parse, but leaving the files in the old format is what makes formatting bugs possible — anything that builds a fragment and puts it into a translation has to guess which format the target line is in. Converted with BentoBox's own Util.legacyToMiniMessage, so each value now says exactly what the runtime was already rendering, then tidied up: a straight conversion closes and immediately reopens a tag wherever the old text repeated a colour code, which is every line of a multi-line description. Verified by rendering all 1506 values before and after and comparing the style of every character: no value moved. Placeholder checks report the same 47 pre-existing issues as before, none new. Tests green. Ten values across vi, tr, it, es, fr, uk and ja were left alone: they contain a broken colour code ("& c", "&un", "& l") that never rendered as colour and still doesn't. Worth a translator's eye, separately. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017EZEwab2kL4i1FNnBYvSmp

sonarqubecloud · 2026-08-14T15:32:24Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

tastybento merged commit d0017fc into developAug 14, 2026
3 checks passed

tastybento deleted the chore/minimessage-locales branch August 14, 2026 22:43