polytypo 1.1.0 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3d928b45252e84a40709cad7cf3f07865ba9bb0cd2836b7d50a926be8c808f78
-  data.tar.gz: a7f438eba68f12e06f3f428be9d253ba5558d8e12ebf17d4e1998a81972c1cd8
+  metadata.gz: 83d0c7e30bb5525df1fcde26cfc166534b11b92e83a68490b2fdb468af059a93
+  data.tar.gz: fba93f881b2fffc39b572cabf93dd5d0a0af8073ffe3179c9594b76c17e6d052
 SHA512:
-  metadata.gz: 3844b1bf0a840dd0321160d4f96f04057fbb9a09448242874785f8f030486f3c86fde4859812e25494ee6e4454cebc21944ace0fe7bfaa55802ad1bd46bce0a7
-  data.tar.gz: a59e7a77a2cd1deef32f5a57d44ad111e56d5b7120c66c2d8c0c69836b08fc167bf33f3fd31b9a0936f4875a09408524747695e989320d8c333163c0fcb40076
+  metadata.gz: 736115dda633084592307b6402ce41c40183c2b71d04b62ad95512f7ad8d62c5f43a29ea4380e718f8f1ee6d7b83da5af527f24a049e947b389b4b0e56b83980
+  data.tar.gz: dd20929d2101f1151bfd004bb9cdcfb89e3a7793442864fb35e897bf9295a21fbca7224f55efc3077ade7ed9d2079487493c8f26c051b70f6b0f39cfe701709c

data/lib/polytypo/data/VERSION

@@ -1 +1 @@
-1.1.0
+1.2.0

data/lib/polytypo/data/fixtures/de-CH.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "de-CH",
   "cases": [
     {
@@ -496,6 +496,14 @@
       "in": "rock 'n' roll",
       "out": "rock ’n’ roll",
       "note": "spec 1.1.0, quotes.md §3.2's universal medial-n elision veto: by operator decision the idiom is international, so `quotes` declines the pairing in every locale rather than only where a cited quotes.elisionIdioms entry exists, and `apostrophe` (order 50) converts each mark by its own case ladder — leading elision (apostrophe.md §3.3 case 4), then trailing elision (case 3). Same output and same two case decisions as en-us-quotes-rock-n-roll, which reaches them through the cited idiom instead. History: spec 0.5.0's general ambiguous-medial-span veto preserved both marks here as literal U+0027; before 0.5.0 `quotes` paired them as an ordinary depth-1 quotation in this locale's primary glyphs."
+    },
+    {
+      "id": "de-ch-apostrophe-elision-before-guillemet",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "die Zeitung l'\"Express\"",
+      "out": "die Zeitung l’«Express»",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0): `quotes` gives de-CH's « », and the elision before « converts."
     }
   ]
 }

data/lib/polytypo/data/fixtures/de-DE.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "de-DE",
   "cases": [
     {
@@ -542,6 +542,38 @@
       "in": "rock 'n' roll",
       "out": "rock ’n’ roll",
       "note": "spec 1.1.0, quotes.md §3.2's universal medial-n elision veto: by operator decision the idiom is international, so `quotes` declines the pairing in every locale rather than only where a cited quotes.elisionIdioms entry exists, and `apostrophe` (order 50) converts each mark by its own case ladder — leading elision (apostrophe.md §3.3 case 4), then trailing elision (case 3). Same output and same two case decisions as en-us-quotes-rock-n-roll, which reaches them through the cited idiom instead. History: spec 0.5.0's general ambiguous-medial-span veto preserved both marks here as literal U+0027; before 0.5.0 `quotes` paired them as an ordinary depth-1 quotation in this locale's primary glyphs."
+    },
+    {
+      "id": "de-de-spaces-dot-word-start-dotfile",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "Die Datei .gitignore ist leer .",
+      "out": "Die Datei .gitignore ist leer.",
+      "note": "spaces.md §3.4 word-start clause (spec 1.2.0): the space before `.gitignore` survives because a letter follows the dot; the space before the final dot is deleted because nothing follows it."
+    },
+    {
+      "id": "de-de-apostrophe-possessive-before-closing-quote",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "Er sagte „Hans'“ und ging.",
+      "out": "Er sagte „Hans’“ und ging.",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0), §6 row 17: U+201C closes a de-DE quotation but is not in CLOSEISH, so case 3 missed the possessive; case 3a reaches the same U+2019 without reading locale data."
+    },
+    {
+      "id": "de-de-apostrophe-closing-single-before-closing-double-not-elision",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "\"Er sagte 'Hans'\" und ging.",
+      "out": "„Er sagte ‚Hans‘“ und ging.",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0), §7 item 7: in de-DE the closing double quote U+201C is in OPENQUOTE, so a closing single quote directly before it has exactly case 3a's shape. `quotes` (order 40) pairs both levels first, so no U+0027 reaches this rule and the nested quotation keeps its de-DE glyphs."
+    },
+    {
+      "id": "de-de-apostrophe-mixed-curly-and-straight-closing-single",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "„Er sagte ‚Hans'“ und ging.",
+      "out": "„Er sagte ‚Hans‘“ und ging.",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0): the same shape with the outer pair and the opening single quote already curly. `quotes` still pairs the straight closing single quote with ‚, so case 3a never sees it."
     }
   ]
 }

data/lib/polytypo/data/fixtures/el.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "el",
   "cases": [
     {
@@ -234,6 +234,14 @@
         "ranges": true
       },
       "note": "spec 0.5.0: el's dash.range is \"none\" (no verified convention), so `ranges` substitutes nothing even when explicitly enabled (ranges.md §2, §3.3) — the same invariant the pre-0.5.0 `dashes`-tagged no-op fixtures for this locale already proved for the combined rule."
+    },
+    {
+      "id": "el-apostrophe-elision-before-guillemet",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "η εφημερίδα l'\"Humanité\"",
+      "out": "η εφημερίδα l’«Humanité»",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0): `quotes` gives el's « », and the elision before « converts."
     }
   ]
 }

data/lib/polytypo/data/fixtures/en-GB.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "en-GB",
   "cases": [
     {
@@ -1269,6 +1269,22 @@
       "in": "'90s were fun,' he said",
       "out": "‘90s were fun,’ he said",
       "note": "quotes.md §6 row E2s: the straight-input control, confirming the corruption in E2 predates mandate 1."
+    },
+    {
+      "id": "en-gb-spaces-dot-word-start-file-extensions",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "CAD files (.DWG, .STEP) on request",
+      "out": "CAD files (.DWG, .STEP) on request",
+      "note": "spaces.md §3.4 word-start clause (spec 1.2.0), §6 row 10g: `.STEP` starts a token, so the space after the comma survives. Spec 1.1.0 produced `(.DWG,.STEP)`."
+    },
+    {
+      "id": "en-gb-apostrophe-elision-before-curly-quote",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "dell'“arte” moderna",
+      "out": "dell’‘arte’ moderna",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0), §6 row 16: `quotes` re-typesets the curly pair as en-GB's primary single quotes, and the elision before the opening glyph converts to U+2019."
     }
   ]
 }

data/lib/polytypo/data/fixtures/en-US.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "en-US",
   "cases": [
     {
@@ -1802,6 +1802,70 @@
         "ranges": true
       },
       "note": "ranges.md §3.2, G3 — not part of a decimal or a path. `before` is U+002E (.) for the first candidate and U+002F (/) for the second; `after` is U+002F (/) for the second. This is currently the only canonical fixture exercising G3 — previously untested by any canonical case even before the spec 0.5.0 ownership correction (dashes.test.ts's engine-level 'must not touch' list covered it, but no fixture did)."
+    },
+    {
+      "id": "en-us-spaces-dot-word-start-dotnet",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "Use .NET, .NET Core or .NET Framework",
+      "out": "Use .NET, .NET Core or .NET Framework",
+      "note": "spaces.md §3.4 word-start clause (spec 1.2.0), §6 row 10f: each dot is followed by a letter, so it starts a token and the space before it is not deleted. Spec 1.1.0 produced `Use.NET,.NET Core or.NET Framework`."
+    },
+    {
+      "id": "en-us-spaces-dot-word-start-decimal",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "the values .5 and .9 apply",
+      "out": "the values .5 and .9 apply",
+      "note": "spaces.md §3.4 word-start clause (spec 1.2.0), §6 row 10h: an ASCII digit after the dot counts the same as a letter."
+    },
+    {
+      "id": "en-us-spaces-dot-word-start-misplaced-full-stop",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "the end .Next sentence",
+      "out": "the end .Next sentence",
+      "note": "spaces.md §3.4, §6 row 10i: the accepted cost of the word-start clause. A misplaced full stop followed directly by a letter keeps its stray space, because deleting it would glue two words that a later pass cannot separate again."
+    },
+    {
+      "id": "en-us-spaces-dot-terminal-still-strips",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "See p. 12 . Next",
+      "out": "See p. 12. Next",
+      "note": "spaces.md §3.4, §6 row 10j: a dot followed by a space is terminal punctuation, so the space before it is still deleted under spec 1.2.0."
+    },
+    {
+      "id": "en-us-nbsp-span-boundary-no-punctuation-data",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "<strong>Note :</strong> read this",
+      "out": "<strong>Note:</strong> read this",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0): accepting the marker changes nothing in a locale whose punctuation lists are empty. `spaces` deletes the space and nothing is put back, the same result as in text mode."
+    },
+    {
+      "id": "en-us-apostrophe-elision-before-quote",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "the magazine l'\"Uomo\" today",
+      "out": "the magazine l’“Uomo” today",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0): `quotes` converts the straight double quotes to U+201C/U+201D, and the elision before U+201C converts to U+2019."
+    },
+    {
+      "id": "en-us-apostrophe-prime-before-bracket-untouched",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "f'(x) = 2x",
+      "out": "f'(x) = 2x",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0), §6 row 18: `(` is in OPENISH but not in OPENQUOTE, so a prime on a function name is left alone."
+    },
+    {
+      "id": "en-us-spaces-dot-before-span-boundary-is-terminal",
+      "rule": "spaces",
+      "mode": "html",
+      "in": "Use .<b>NET</b> today",
+      "out": "Use.<b>NET</b> today",
+      "note": "spaces.md §3.4 word-start clause (spec 1.2.0): the code point after the dot is the −1 span boundary marker, which is neither a LETTER nor an ASCII digit (modes.md §3.3), so the dot counts as terminal and the space before it is deleted."
     }
   ]
 }

data/lib/polytypo/data/fixtures/fi.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "fi",
   "cases": [
     {
@@ -1309,6 +1309,14 @@
       "in": "'90s were fun,' he said",
       "out": "”90s were fun,” he said",
       "note": "quotes.md §6 row E3s: the straight-input control for E3."
+    },
+    {
+      "id": "fi-apostrophe-elision-before-shared-quote-glyph",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "lehti l'\"Humanité\"",
+      "out": "lehti l’”Humanité”",
+      "note": "apostrophe.md §3.3 cases 3 and 3a (spec 1.2.0): fi opens and closes with U+201D, which is in CLOSEISH, so case 3 already converted this before 1.2.0. Pinned so that case 3a is not read as changing a shared-glyph locale."
     }
   ]
 }

data/lib/polytypo/data/fixtures/fr-CA.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "fr-CA",
   "cases": [
     {
@@ -263,6 +263,31 @@
         "ranges": true
       },
       "note": "spec 0.5.0: fr-CA's dash.range is \"none\" (no verified convention), so `ranges` substitutes nothing even when explicitly enabled (ranges.md §2, §3.3) — the same invariant the pre-0.5.0 `dashes`-tagged no-op fixtures for this locale already proved for the combined rule."
+    },
+    {
+      "id": "fr-ca-nbsp-span-boundary-strong-colon",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "<strong>Résistant au gel :</strong> il reste souple",
+      "out": "<strong>Résistant au gel :</strong> il reste souple",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0): same as the `fr` case; `fr-CA` lists `:` in its own punctuation data."
+    },
+    {
+      "id": "fr-ca-nbsp-span-boundary-markdown-strong-colon",
+      "rule": "nbsp",
+      "mode": "markdown",
+      "dialect": "commonmark",
+      "in": "**Note :** voir plus bas\n",
+      "out": "**Note :** voir plus bas\n",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0): the colon ends the strong span; the −1 marker after it is in CLOSEISH, so N1 restores the U+00A0 that `spaces` deleted. (fr-CA lists no punctuation for N2, so `?` would take nothing back in any mode.)"
+    },
+    {
+      "id": "fr-ca-apostrophe-elision-before-straight-quote",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "l'\"idée\" est simple",
+      "out": "l’« idée » est simple",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0): same as `fr`."
     }
   ]
 }

data/lib/polytypo/data/fixtures/fr.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "fr",
   "cases": [
     {
@@ -598,6 +598,112 @@
         "ranges": true
       },
       "note": "spec 0.5.0: fr's dash.range is \"none\" (no verified convention), so `ranges` substitutes nothing even when explicitly enabled (ranges.md §2, §3.3) — the same invariant the pre-0.5.0 `dashes`-tagged no-op fixtures for this locale already proved for the combined rule."
+    },
+    {
+      "id": "fr-spaces-dot-word-start-dotfile",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "Voir le fichier .htaccess .",
+      "out": "Voir le fichier .htaccess.",
+      "note": "spaces.md §3.4 word-start clause (spec 1.2.0): the space before `.htaccess` survives, the space before the terminal dot is deleted. `.` is in no `nbsp` list for `fr`, so nothing is put back."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-strong-colon",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "<strong>Résistant au gel :</strong> il reste souple",
+      "out": "<strong>Résistant au gel :</strong> il reste souple",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0), §6 row 6c; modes.md §3.3: `spaces` deletes the typed space before `:`, and the −1 span boundary marker after the mark is in `nbsp`'s CLOSEISH, so N1 inserts U+00A0 at the mark's index, inside the span. Spec 1.1.0 runtimes produced `gel:</strong>`."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-br-question",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "Et la réglementation ?<br>Oui",
+      "out": "Et la réglementation ?<br>Oui",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0), §6 row 6d: `<br>` leaves no line terminator in the gap, so the marker is −1; it is in CLOSEISH and N2 restores U+202F."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-link-question",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "<a href=\"/faq\">Comment ?</a> Oui",
+      "out": "<a href=\"/faq\">Comment ?</a> Oui",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0): the mark ends the link text, the marker follows it, and N2 restores U+202F instead of leaving `Comment?`."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-markdown-strong-colon",
+      "rule": "nbsp",
+      "mode": "markdown",
+      "dialect": "commonmark",
+      "in": "**Résistant au gel :** il reste souple\n",
+      "out": "**Résistant au gel :** il reste souple\n",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0): the Markdown form of the French definition pattern; the closing `**` is a skipped region with no line terminator, so the marker is −1 and N1 restores U+00A0."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-not-openish-em",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "Il dit <em>non</em> ! Oui",
+      "out": "Il dit <em>non</em> ! Oui",
+      "note": "nbsp.md §3.3 step 3 (spec 1.2.0), §6 row 6e; modes.md §3.3: the marker is not in `nbsp`'s OPENISH, so the quote-glyph guard does not decline and N2 converts the space. A port that applied the quote-glyph guard to the marker would leave a plain U+0020 here."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-not-openish-link",
+      "rule": "nbsp",
+      "mode": "markdown",
+      "dialect": "commonmark",
+      "in": "Voir [ceci](http://example.org) : oui\n",
+      "out": "Voir [ceci](http://example.org) : oui\n",
+      "note": "nbsp.md §3.3 step 3 (spec 1.2.0): the second witness that the quote-glyph guard does not decline at a span boundary. The link text ends two places left of `:`, and N1 still converts the space. The marker's absence from the rest of `nbsp`'s OPENISH is pinned by ru-nbsp-span-boundary-not-openish-short-word."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-mark-alone-in-span",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "mot<em>!</em>",
+      "out": "mot<em>!</em>",
+      "note": "nbsp.md §3.3 step 2; modes.md §3.4 edge-growth rule: the mark is alone in its span, so the insertion point is the span edge and the insertion is discarded, as it was before spec 1.2.0."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-colon-before-digit-span",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "Il est 12:<b>30</b>",
+      "out": "Il est 12 :<b>30</b>",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0), §6 row 6f: the stated cost of accepting the marker. Step 2 cannot see past the span boundary, so a colon that ends one span while a digit starts the next is treated as sentence punctuation."
+    },
+    {
+      "id": "fr-apostrophe-elision-before-guillemet",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "un mélange d'« urine diluée »",
+      "out": "un mélange d’« urine diluée »",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0), §6 row 15: letter left, opening quotation glyph right. Spec 1.1.0 left the apostrophe straight, because no case matched a letter followed by an OPENISH-only glyph."
+    },
+    {
+      "id": "fr-apostrophe-elision-before-straight-quote",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "l'\"idée\" est simple",
+      "out": "l’« idée » est simple",
+      "note": "apostrophe.md §3.2, §3.3 case 3a (spec 1.2.0): `quotes` (order 40) turns the straight `\"` into `«` first, so the mark reaches this rule next to an opening glyph and case 3a converts it."
+    },
+    {
+      "id": "fr-apostrophe-elision-qu-before-guillemet",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "Il faut qu'« il vienne »",
+      "out": "Il faut qu’« il vienne »",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0): `qu'` before a quotation."
+    },
+    {
+      "id": "fr-nbsp-span-boundary-colon-before-url-span",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "Voir http:<a href=\"//x.org\">//x.org</a>",
+      "out": "Voir http :<a href=\"//x.org\">//x.org</a>",
+      "note": "nbsp.md §3.3 step 2 (spec 1.2.0): the second stated cost of accepting the marker. Step 2 cannot see the `/` in the next span, so the colon is treated as sentence punctuation and N1 inserts U+00A0."
     }
   ]
 }

data/lib/polytypo/data/fixtures/locale-resolution.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "cases": [
     {
       "id": "exact-en-us",

data/lib/polytypo/data/fixtures/ru.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "ru",
   "cases": [
     {
@@ -691,6 +691,30 @@
       "in": "rock ’n’ roll",
       "out": "rock ’n’ roll",
       "note": "quotes.md §3.2, the regression witness for spec 1.1.0's widening of the veto from straight ASCII to the whole NARROW class. This is the output of ru-quotes-rock-n-roll fed back in, so it is that case's idempotency obligation made explicit — and it is the case a naive implementation gets wrong. A port that matches U+0027 only (as spec 0.5.0's withdrawn predicate did, correctly for its own preserve-then-stop outcome) does not recognise the U+2019 marks it just produced: the pair falls through to ordinary NARROW quotation and this input becomes `rock «n» roll`, ru's primary pair at depth 1. Measured, not hypothesised. `apostrophe` emits U+2019 only in place of U+0027 (apostrophe.md §1), so once the veto matches, nothing edits these marks and the form is stable."
+    },
+    {
+      "id": "ru-spaces-dot-word-start-dotfile",
+      "rule": "spaces",
+      "mode": "text",
+      "in": "Проверьте файл .env сейчас",
+      "out": "Проверьте файл .env сейчас",
+      "note": "spaces.md §3.4 word-start clause (spec 1.2.0): a dot followed by a letter starts a token in every locale; `spaces` reads no locale data (§2)."
+    },
+    {
+      "id": "ru-apostrophe-elision-before-guillemet",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "газета l'\"Humanité\"",
+      "out": "газета l’«Humanité»",
+      "note": "apostrophe.md §3.3 case 3a (spec 1.2.0): a French title in Russian text; `quotes` gives « », and the elision before « converts."
+    },
+    {
+      "id": "ru-nbsp-span-boundary-not-openish-short-word",
+      "rule": "nbsp",
+      "mode": "html",
+      "in": "Он живёт в <em>Москве</em>",
+      "out": "Он живёт в <em>Москве</em>",
+      "note": "nbsp.md §3.1, §7 item 12 (spec 1.2.0); modes.md §3.3: the −1 span boundary marker is not in `nbsp`'s OPENISH, so N3's following-token guard does not accept it and the space after `в` stays U+0020. This pins the membership split itself: a port that also put the marker in OPENISH would bind here."
     }
   ]
 }

data/lib/polytypo/data/fixtures/sv.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "locale": "sv",
   "cases": [
     {
@@ -1285,6 +1285,14 @@
       "in": "'90s were fun,' he said",
       "out": "”90s were fun,” he said",
       "note": "quotes.md §6 row E4s: the straight-input control for E4."
+    },
+    {
+      "id": "sv-apostrophe-elision-before-shared-quote-glyph",
+      "rule": "apostrophe",
+      "mode": "text",
+      "in": "tidningen l'\"Humanité\"",
+      "out": "tidningen l’”Humanité”",
+      "note": "apostrophe.md §3.3 cases 3 and 3a (spec 1.2.0): same as fi; U+201D is in CLOSEISH and the output is unchanged from spec 1.1.0."
     }
   ]
 }

data/lib/polytypo/data/locales/registry.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "$comment": "Locale resolution input. Algorithm is specified in spec/rules/locale-resolution.md and is identical in every runtime — never delegate it to a platform locale-negotiation library. \"spec\" here must track spec/VERSION exactly — it is not itself the global version source; scripts/validate-spec.mjs enforces the match.",
   "locales": ["en-US", "en-GB", "de-DE", "de-CH", "fr", "fr-CA", "ru", "fi", "sv", "el"],
   "aliases": {

data/lib/polytypo/data/rules/order.json

@@ -1,5 +1,5 @@
 {
-  "spec": "1.1.0",
+  "spec": "1.2.0",
   "$comment": "Single source of truth for pipeline order. Rules run in ascending `order`. Disabling a rule removes it from the sequence and never reorders the rest. Rule ids are public API (see docs/ARCHITECTURE.md section 5). \"spec\" here must track spec/VERSION exactly — it is not itself the global version source; scripts/validate-spec.mjs enforces the match.",
   "rules": [
     {

data/lib/polytypo/engine/rules/apostrophe.rb

@@ -9,7 +9,7 @@ require_relative "../registry"
 module Polytypo
   module Engine
     module Rules
-      # spec/rules/apostrophe.md (spec 0.5.0), order 50.
+      # spec/rules/apostrophe.md (spec 1.2.0), order 50.
       #
       # Converts a straight U+0027 to U+2019 where it is genuinely an apostrophe: a contraction,
       # an elision, a possessive, or a decade elision. Runs immediately after `quotes` (order 40)
@@ -47,6 +47,10 @@ module Polytypo
           0x2D, 0x2011, 0x2013, 0x2014
         ].freeze
 
+        # OPENQUOTE is apostrophe.md 3.1 OPENQUOTE (spec 1.2.0): the quotation glyphs of OPENISH,
+        # without its brackets, its dashes and Engine::MARKER. Read only by case 3a.
+        OPENQUOTE = [0xAB, 0x2018, 0x201A, 0x201B, 0x201C, 0x201E, 0x201F, 0x2039].freeze
+
         BREAK = [0x0A, 0x0D, 0x0B, 0x0C, 0x85, 0x2028, 0x2029, Engine::LINE_MARKER].freeze
 
         # SPACELIKE, including Engine::LINE_MARKER as a member of BREAK for every rule everywhere
@@ -81,6 +85,11 @@ module Polytypo
             return true
           end
 
+          # Case 3a (spec 1.2.0) -- elision before a quotation: `d'« urine »`, `l'“idea”`. Case 3
+          # misses it because an opening quotation glyph is not in CLOSEISH; `f'(x)` stays straight
+          # because a bracket is not in OPENQUOTE.
+          return true if UnicodeUtil.letter?(left) && OPENQUOTE.include?(right)
+
           # Case 4 -- leading elision: `'90s`, `'tis`, `'em`, `'n'` (the leading mark). The
           # replacement is U+2019, never U+2018 -- a leading elision is a raised comma, not an
           # opening quotation mark, and `quotes` has already had its chance to claim the mark as a

data/lib/polytypo/engine/rules/nbsp.rb

@@ -9,7 +9,7 @@ require_relative "../../errors"
 module Polytypo
   module Engine
     module Rules
-      # `nbsp` -- spec/rules/nbsp.md (spec 0.6.0), order 70 (last).
+      # `nbsp` -- spec/rules/nbsp.md (spec 1.2.0), order 70 (last).
       #
       # Ten sub-rules, N1 through N10, evaluated in that fixed order (3.2); each produces
       # candidate edits keyed by the index of the space (or insertion point) it claims, and the
@@ -54,16 +54,7 @@ module Polytypo
         end
 
         # BREAK (nbsp.md 3.1), including LINE_MARKER -- a member of BREAK for every rule,
-        # everywhere (modes.md 3.2).
-        #
-        # MARKER is deliberately NOT a member here, and openish?/closeish? below do not add it
-        # either. modes.md 3.3's table says nbsp's OPENISH/CLOSEISH include the span-boundary
-        # MARKER, the same way quotes' and apostrophe's do -- but the JS reference
-        # implementation's own isBreak/isOpenish/isCloseish never test for MARKER, only
-        # LINE_MARKER via isBreak. That is a documented, preserved discrepancy (tracked, not
-        # resolved, in the roadmap), carried forward here unchanged rather than "corrected"
-        # against the table, for cross-runtime consistency with the already-shipped JS, Python
-        # and Go ports.
+        # everywhere (modes.md 3.2). MARKER is not a member of BREAK; see closeish? below.
         def self.break?(cp)
           cp == 0x0A || cp == 0x0D || cp == 0x0B || cp == 0x0C || cp == 0x85 ||
             cp == 0x2028 || cp == 0x2029 || cp == LINE_MARKER
@@ -197,13 +188,19 @@ module Polytypo
         end
 
         # OPENISH / CLOSEISH (nbsp.md 3.1): the ASCII brackets plus every locale quote glyph.
-        # See break?'s comment for why MARKER is not a member here.
+        #
+        # Since spec 1.2.0 the span boundary MARKER is a member of CLOSEISH and not of OPENISH
+        # (nbsp.md 3.1 and 7 item 12, modes.md 3.3). CLOSEISH is read only by N1/N2's right-context
+        # guard, where membership lets French `<strong>gel :</strong>` keep its no-break space
+        # after `spaces` deletes the U+0020. OPENISH stays marker-free: with the marker in it,
+        # N1/N2's quote-glyph guard would decline `<em>non</em> ! Oui`, and the left-boundary
+        # tests of N3, N7, N9, N10 would widen at span edges.
         def self.openish?(prep, cp)
           prep.opens.include?(cp)
         end
 
         def self.closeish?(prep, cp)
-          prep.closes.include?(cp)
+          cp == MARKER || prep.closes.include?(cp)
         end
 
         def self.mark?(prep, cp)

data/lib/polytypo/engine/rules/spaces.rb

@@ -107,10 +107,17 @@ module Polytypo
         # after any edit had been applied would break the Chicago spaced ellipsis
         # "Hello . . .", where every dot is a lone dot at decision time and all three spaces
         # must still strip.
+        #
+        # Spec 1.2.0's word-start clause: a single dot directly followed by a letter or an ASCII
+        # digit starts a token (".NET", ".env", ".5"), so "Use .NET" keeps its space. A span
+        # boundary marker after the dot is neither, so "a .</em>" still strips.
         def self.lone_dot?(cp, e)
           return true if at(cp, e) != FULL_STOP
 
-          !dotlike?(at(cp, e + 1))
+          after = at(cp, e + 1)
+          return false if UnicodeUtil.letter?(after) || digit_ascii?(after)
+
+          !dotlike?(after)
         end
 
         def self.digit_ascii?(value)

data/lib/polytypo/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Polytypo
-  VERSION = "1.1.0"
+  VERSION = "1.2.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: polytypo
 version: !ruby/object:Gem::Version
-  version: 1.1.0
+  version: 1.2.0
 platform: ruby
 authors:
 - Iurii Rogulia