switchroom 0.21.8 → 0.21.10

package/telegram-plugin/tts-normalize.ts

@@ -350,16 +350,54 @@ export function normalizeForTts(text: string): string {
 
   // -- Number + unit suffix glued to the number (500ms, 5km, 10GB). Only a
   //    known unit bounded by a non-letter, so identifiers never match.
+  //
+  //    The leading `(?<![\d.,])` lookbehind is load-bearing: without it
+  //    `\b` is satisfied between a decimal point (or thousands comma) and
+  //    the digits that follow, so "17s" inside "0.17s" — or "500s" inside
+  //    "1,500s" — would match on its own and mangle the number ("0.17s" →
+  //    "0.seventeen seconds"; "1,500s" → "1,five hundred seconds"). Excludes
+  //    both `.` and `,` so a glued decimal/thousands-separated number is
+  //    never re-anchored on. Multi-letter units (ms, GB, …) stay
+  //    case-insensitive since `5MB`/`500ms` must keep working regardless of
+  //    case; single-letter units (s, m, h, d, g) are matched
+  //    case-sensitively lowercase-only in a second pass — the single letter
+  //    "M" is one keystroke from meaning "million" in prose ("$5M",
+  //    "Revenue was 5M") and must never fall through to "minutes" just
+  //    because the whole alternation ran under /i.
+  //
+  //    KNOWN GAP, disclosed deliberately (2026-08, hotfix for #2760-derived
+  //    corpus regression): a decimal- or comma-glued number with a unit
+  //    (27.5s, 1,500s, 89.5g, 0.7-0.8m) is now left completely UNEXPANDED —
+  //    the unit goes unspoken — rather than mangled the way it was before.
+  //    Corpus-measured: samples carrying a digit-adjacent unspoken unit go
+  //    65 → 122 (61 samples change under this fix, every one gaining an
+  //    unspoken unit). Correctly expanding a decimal/comma-glued number is
+  //    explicitly out of scope for this narrow hotfix — see the pinning
+  //    tests below (13.0 GB / 89.5g / 27.5 s) marking the baseline for a
+  //    follow-up decimal-expansion pass.
+  //
+  //    ASYMMETRY WITH pass 1 (voice-normalize-text.ts), pre-existing and
+  //    NOT introduced by this fix: this pass allows an optional space
+  //    (`\s?`) between the number and the unit and maps `m` to METRE; pass
+  //    1 has no `\s?` and maps `m` to MINUTE. That accident is exactly what
+  //    makes "5m" (no space) come out of pass 1 as minutes and "16 m"
+  //    (space) come out of pass 2 as metres in the composed pipeline —
+  //    pinned by the composed-pipeline test in the test suite.
+  const unitReplacer = (m: string, num: string, unitRaw: string): string => {
+    const unit = UNIT_MAP[unitRaw.toLowerCase()]
+    if (!unit) return m
+    const n = Number(num)
+    const w = numberToWords(n)
+    if (!w) return m
+    return `${w} ${n === 1 ? unit.s : unit.p}`
+  }
   s = s.replace(
-    /\b(\d{1,9})\s?(ms|sec|min|km|cm|mm|kg|kb|mb|gb|tb|ghz|mhz|hr|mi|s|m|h|d|g)(?![a-z])/gi,
-    (m, num: string, unitRaw: string) => {
-      const unit = UNIT_MAP[unitRaw.toLowerCase()]
-      if (!unit) return m
-      const n = Number(num)
-      const w = numberToWords(n)
-      if (!w) return m
-      return `${w} ${n === 1 ? unit.s : unit.p}`
-    },
+    /(?<![\d.,])\b(\d{1,9})\s?(ms|sec|min|km|cm|mm|kg|kb|mb|gb|tb|ghz|mhz|hr|mi)(?![a-zA-Z])/gi,
+    unitReplacer,
+  )
+  s = s.replace(
+    /(?<![\d.,])\b(\d{1,9})\s?(s|m|h|d|g)(?![a-zA-Z])/g,
+    unitReplacer,
   )
 
   // -- Symbols in prose.

package/telegram-plugin/uat/scenarios/jtbd-rich-formatting-render-dm.test.ts

@@ -271,13 +271,16 @@ function kindsPresent(msg: ObservedMessage): Set<string> {
 );
 
 // ---------------------------------------------------------------------------
-// Rich-render wiring proof — expandable / collapsible content round-trip.
+// Rich-render wiring proof — legacy `**> ` quote repair round-trip.
 //
 // This is the piece the unit suite CANNOT prove: that a REAL markdown reply
-// carrying the Bot API 10.1 expandable-blockquote marker (`**> `) flows
+// carrying the LEGACY switchroom expandable-quote marker (`**> `) flows
 // through the live send path — parse.ts (marker -> IR `expandable: true`) ->
-// render.ts (IR -> `**> ` markdown) -> renderSafe -> sendRichMessage — and
-// that Telegram actually parses it back to a blockquote entity on the wire.
+// render.ts (IR -> plain `> ` markdown; `**>` is MarkdownV2-only and renders
+// as literal text on the rich path, wire-proved 2026-08-13, so the renderer
+// no longer emits it) -> renderSafe -> sendRichMessage — and that Telegram
+// parses the repaired output back to a blockquote entity with NO literal
+// `**>` glyphs reaching the reader.
 //
 // Runs ONLY when the rich-render flag is on in this process (see
 // RICH_RENDER_ON) AND the driver creds are present; self-skips green
@@ -299,7 +302,7 @@ const EXPANDABLE_SAMPLE = [
   "uat: expandable/collapsible content round-trips through the live renderer",
   () => {
     it(
-      "a `**>` reply parses -> renders -> sends -> decodes as a blockquote entity",
+      "a legacy `**>` reply is repaired to a plain quote and decodes as a blockquote entity",
       async () => {
         const sc = await spinUp({ agent: AGENT });
         try {
@@ -318,9 +321,9 @@ const EXPANDABLE_SAMPLE = [
           ).not.toBe("\x01");
           expect(reply.text).toContain(EXP_MARKER);
 
-          // The collapsible construct must survive as a real blockquote entity
-          // on the wire — proof the parse->render->send path produced valid
-          // Bot API 10.1 markdown, not stray `**>` literal text.
+          // The legacy quote must survive as a real blockquote entity on the
+          // wire — proof the parse->render->send path repaired the `**>` line
+          // into the quote instead of shipping stray literal text.
           const present = kindsPresent(reply);
           expect(
             [...present],
@@ -331,6 +334,12 @@ const EXPANDABLE_SAMPLE = [
           // The quoted body text must round-trip (content never lost).
           expect(reply.text).toContain("first hidden line");
 
+          // And the retired marker must never reach the reader as literal
+          // glyphs — the exact failure mode of the pre-repair renderer
+          // (Telegram rendered the `**> ` opener as a literal `**>` paragraph
+          // while the `> ` continuation lines formed a detached quote).
+          expect(reply.text).not.toContain("**>");
+
           console.info(
             "[uat] expandable round-trip entity structure: " +
               JSON.stringify(

package/telegram-plugin/voice-normalize-text.ts

@@ -561,16 +561,55 @@ export function normalizeForSpeech(input: string): string {
   //     Number + unit suffix → "<number> <unit>" (e.g. 500ms, 2h, 10KB).
   //     Only when the suffix is a known unit glued directly to the number and
   //     bounded by a non-letter (so "my5thing" / "class5" are never touched).
+  //
+  //     The leading `(?<![\d.,])` lookbehind is load-bearing: without it
+  //     `\b` is satisfied between a decimal point (or thousands comma) and
+  //     the digits that follow, so "17s" inside "0.17s" — or "500s" inside
+  //     "1,500s" — would match on its own and mangle the number ("0.17s" →
+  //     "0.seventeen seconds"; "1,500s" → "1,five hundred seconds"). Excludes
+  //     both `.` and `,` so a glued decimal/thousands-separated number is
+  //     never re-anchored on. Multi-letter units (ms, GB, …) stay
+  //     case-insensitive since `5MB`/`500ms` must keep working regardless of
+  //     case; single-letter units (s, m, h, d) are matched case-sensitively
+  //     lowercase-only in a second pass — the single letter "M" is one
+  //     keystroke from meaning "million" in prose ("$5M", "Revenue was 5M")
+  //     and must never fall through to "minutes" just because the whole
+  //     alternation ran under /i.
+  //
+  //     KNOWN GAP, disclosed deliberately (2026-08, hotfix for a corpus
+  //     regression): a decimal- or comma-glued number with a unit (27.5s,
+  //     1,500s, 89.5g, 0.7-0.8m) is now left completely UNEXPANDED — the
+  //     unit goes unspoken — rather than mangled the way it was before.
+  //     Corpus-measured: samples carrying a digit-adjacent unspoken unit go
+  //     65 → 122 (61 samples change under this fix, every one gaining an
+  //     unspoken unit). Correctly expanding a decimal/comma-glued number is
+  //     explicitly out of scope for this narrow hotfix — see the pinning
+  //     tests below (13.0 GB / 89.5g / 27.5 s) marking the baseline for a
+  //     follow-up decimal-expansion pass.
+  //
+  //     ASYMMETRY WITH pass 2 (tts-normalize.ts), pre-existing and NOT
+  //     introduced by this fix: this pass has NO optional space between the
+  //     number and the unit (unlike pass 2's `\s?`) and maps `m` to MINUTE;
+  //     pass 2 maps `m` to METRE. That accident is exactly what makes "5m"
+  //     (no space) come out of THIS pass as minutes and "16 m" (space) fall
+  //     through untouched here but come out of pass 2 as metres in the
+  //     composed pipeline — pinned by the composed-pipeline test in the
+  //     tts-normalize test suite.
+  const unitReplacer = (m: string, num: string, unitRaw: string): string => {
+    const unit = UNIT_MAP[unitRaw.toLowerCase()]
+    if (!unit) return m
+    const n = Number(num)
+    const w = numberToWords(n)
+    if (!w) return m
+    return `${w} ${n === 1 ? unit.s : unit.p}`
+  }
+  s = s.replace(
+    /(?<![\d.,])\b(\d{1,9})(ms|sec|min|kb|mb|gb|tb|hr)(?![a-zA-Z])/gi,
+    unitReplacer,
+  )
   s = s.replace(
-    /\b(\d{1,9})(ms|sec|min|kb|mb|gb|tb|hr|s|m|h|d)(?![a-z])/gi,
-    (m, num, unitRaw) => {
-      const unit = UNIT_MAP[unitRaw.toLowerCase()]
-      if (!unit) return m
-      const n = Number(num)
-      const w = numberToWords(n)
-      if (!w) return m
-      return `${w} ${n === 1 ? unit.s : unit.p}`
-    },
+    /(?<![\d.,])\b(\d{1,9})(s|m|h|d)(?![a-zA-Z])/g,
+    unitReplacer,
   )
   //     Symbols between tokens: standalone & → and, + → plus, = → equals.
   s = s.replace(/(\S)\s*\+\s*(\S)/g, '$1 plus $2')