punctilio 3.7.3 → 3.7.5

package/src/symbols.ts

@@ -190,24 +190,28 @@ function contextAwareLegalReplace(
   })
 }
 
+const LEGAL_COPYRIGHT_RE = /\(c\)/gi
+const LEGAL_REGISTERED_RE = /\(r\)/gi
+const LEGAL_TRADEMARK_RE = /\(tm\)/gi
+
 /** Convert (c), (r), (tm) to ©, ®, ™. */
 export function legalSymbols(text: string, options: SymbolOptions = {}): string {
   const separator = options.separator ?? DEFAULT_SEPARATOR
   // (c) → © only with positive copyright evidence (year or "copyright" keyword)
   // and not in a path context (e.g., example.com/path(c))
-  text = contextAwareLegalReplace(text, /\(c\)/gi, COPYRIGHT, (before, after) =>
+  text = contextAwareLegalReplace(text, LEGAL_COPYRIGHT_RE, COPYRIGHT, (before, after) =>
     !isPathContext(before) && (/^\s*(?:19|20)\d{2}\b/.test(after) || /\bcopyright\s*$/i.test(before)),
     separator
   )
 
   // (r) → ® unless in enumeration "(q), (r)", legal citation "(r)(1)", or path context
-  text = contextAwareLegalReplace(text, /\(r\)/gi, REGISTERED, (before, after) =>
+  text = contextAwareLegalReplace(text, LEGAL_REGISTERED_RE, REGISTERED, (before, after) =>
     !/\([a-z]\)[,;]\s*$/i.test(before) && !/^\(\d/.test(after) && !isPathContext(before),
     separator
   )
 
   // (tm) → ™ unless in a path context
-  text = contextAwareLegalReplace(text, /\(tm\)/gi, TRADEMARK, (before) =>
+  text = contextAwareLegalReplace(text, LEGAL_TRADEMARK_RE, TRADEMARK, (before) =>
     !isPathContext(before),
     separator
   )
@@ -452,10 +456,12 @@ export function superscriptOrdinal(text: string, options: SymbolOptions = {}): s
   })
 }
 
-/** Collapse multiple spaces (including tabs) to single space. Prefers nbsp if any nbsp is present. */
+/** Collapse multiple spaces (including tabs) to single space. Preserves the highest-priority space type present in the run: NBSP > NNBSP > regular. Leading whitespace at the start of a line (after `\n` or start-of-string) is preserved so indented blocks (e.g. HN-style code) survive. */
 export function collapseSpaces(text: string): string {
-  return text.replace(cachedRegExp(`[${SPACE_CHARS}]{2,}`, "g"), (match) => {
-    return match.includes(NBSP) ? NBSP : " "
+  return text.replace(cachedRegExp(`(?<=[^\\n${SPACE_CHARS}])[${SPACE_CHARS}]{2,}`, "g"), (match) => {
+    if (match.includes(NBSP)) return NBSP
+    if (match.includes(UNICODE_SYMBOLS.NNBSP)) return UNICODE_SYMBOLS.NNBSP
+    return " "
   })
 }
 

package/src/utils.ts

@@ -60,6 +60,7 @@ export function assertSeparatorAbsent(textValues: string[], separator: string):
 }
 
 export function countSeparators(text: string, separator: string = DEFAULT_SEPARATOR): number {
+  if (separator.length === 0) return 0
   let count = 0
   let index = 0
   while ((index = text.indexOf(separator, index)) !== -1) {

package/tsconfig.json

@@ -16,5 +16,5 @@
     "isolatedModules": true
   },
   "include": ["src/**/*"],
-  "exclude": ["node_modules", "dist", "**/*.test.ts"]
+  "exclude": ["node_modules", "dist", "**/*.test.ts", "src/tests"]
 }

package/dist/tests/test-helpers.d.ts

@@ -1,33 +0,0 @@
-/**
- * Shared test utilities for punctilio test suites.
- */
-/**
- * Builds a deterministic pseudo-random string of approximately `charCount`
- * characters by randomly sampling from FRAGMENTS using a seeded PRNG.
- *
- * @param charCount - Approximate desired character count
- * @param seed - PRNG seed (default: "42")
- */
-export declare function buildMixedContent(charCount: number, seed?: string): string;
-/**
- * Asserts that a function scales linearly (not quadratically) with input size.
- *
- * Compares ms/char at the two largest of 4 input sizes (4x vs 8x of startingN)
- * after a JIT warmup pass. For a linear algorithm, doubling input keeps ms/char
- * constant (ratio ~1). For a quadratic one, doubling input doubles ms/char (~2).
- *
- * Default startingN of 5000 produces inputs from ~5K to ~40K chars, which is
- * past V8's string allocation overhead inflection point where ms/char stabilizes.
- *
- * Each trial measures `fn` and a known-linear baseline back-to-back at both
- * sizes, then divides `fn`'s 8x/4x ratio by the baseline's. Shared platform
- * noise (CPU contention, scheduler jitter) inflates both numerator and
- * denominator and cancels out; only `fn`-specific super-linear growth pushes
- * the normalized ratio past 1. Min across trials picks the cleanest sample.
- *
- * @param fn - The function to benchmark
- * @param buildInput - Builds an input string of approximately `n` units
- * @param startingN - Size parameter for the smallest input (default: 5000)
- */
-export declare function assertLinearScaling(fn: (input: string) => unknown, buildInput: (n: number) => string, startingN?: number): void;
-//# sourceMappingURL=test-helpers.d.ts.map

package/dist/tests/test-helpers.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"test-helpers.d.ts","sourceRoot":"","sources":["../../src/tests/test-helpers.ts"],"names":[],"mappings":"AAAA;;GAEG;AAgCH;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,SAAS,EAAE,MAAM,EAAE,IAAI,GAAE,MAAa,GAAG,MAAM,CAQhF;AAUD;;;;;;;;;;;;;;;;;;;GAmBG;AACH,wBAAgB,mBAAmB,CACjC,EAAE,EAAE,CAAC,KAAK,EAAE,MAAM,KAAK,OAAO,EAC9B,UAAU,EAAE,CAAC,CAAC,EAAE,MAAM,KAAK,MAAM,EACjC,SAAS,GAAE,MAAa,GACvB,IAAI,CAwCN"}

package/dist/tests/test-helpers.js

@@ -1,109 +0,0 @@
-/**
- * Shared test utilities for punctilio test suites.
- */
-import seedrandom from "seedrandom";
-/**
- * Fragments covering every transform type: quotes, dashes, apostrophes,
- * ellipses, multiplication, math symbols, legal symbols, arrows, primes,
- * fractions, degrees, ordinals, nbsp-triggering short words, and plain text.
- */
-const FRAGMENTS = [
-    `"Hello," she said. `,
-    `It's a beautiful day -- isn't it? `,
-    `Wait... `,
-    `The temperature was 20 C. `,
-    `5x5 != 25. `,
-    `Add 1/2 cup. `,
-    `He's 5'10" tall. `,
-    `(c) 2024 Acme Corp. `,
-    `See p. 42 for details. `,
-    `Dr. Smith arrived. `,
-    `The 1st place winner. `,
-    `Go -> click -> done. `,
-    `Really?! No way!! `,
-    `A B C D E F G H. `,
-    `Pages 10-20 contain important info. `,
-    `January-March was cold. `,
-    `The value is -5. `,
-    `"Don't stop," he said, "keep going." `,
-    `J. K. Rowling wrote it. `,
-    `100 km in 2 hrs. `,
-];
-/**
- * Builds a deterministic pseudo-random string of approximately `charCount`
- * characters by randomly sampling from FRAGMENTS using a seeded PRNG.
- *
- * @param charCount - Approximate desired character count
- * @param seed - PRNG seed (default: "42")
- */
-export function buildMixedContent(charCount, seed = "42") {
-    const random = seedrandom(seed);
-    let result = "";
-    while (result.length < charCount) {
-        const index = Math.floor(random() * FRAGMENTS.length);
-        result += FRAGMENTS[index];
-    }
-    return result;
-}
-/**
- * Provably linear baseline: a single regex pass that touches every char and
- * exercises the same V8 String.replace path the production transforms use.
- * Its ms/char ratio is the platform's "what linear looks like right now"
- * fingerprint — we divide fn's ratio by it to cancel shared CPU/GC noise.
- */
-const linearBaseline = (s) => s.replace(/[a-z]/gi, "x");
-/**
- * Asserts that a function scales linearly (not quadratically) with input size.
- *
- * Compares ms/char at the two largest of 4 input sizes (4x vs 8x of startingN)
- * after a JIT warmup pass. For a linear algorithm, doubling input keeps ms/char
- * constant (ratio ~1). For a quadratic one, doubling input doubles ms/char (~2).
- *
- * Default startingN of 5000 produces inputs from ~5K to ~40K chars, which is
- * past V8's string allocation overhead inflection point where ms/char stabilizes.
- *
- * Each trial measures `fn` and a known-linear baseline back-to-back at both
- * sizes, then divides `fn`'s 8x/4x ratio by the baseline's. Shared platform
- * noise (CPU contention, scheduler jitter) inflates both numerator and
- * denominator and cancels out; only `fn`-specific super-linear growth pushes
- * the normalized ratio past 1. Min across trials picks the cleanest sample.
- *
- * @param fn - The function to benchmark
- * @param buildInput - Builds an input string of approximately `n` units
- * @param startingN - Size parameter for the smallest input (default: 5000)
- */
-export function assertLinearScaling(fn, buildInput, startingN = 5000) {
-    const input4x = buildInput(startingN * 4);
-    const input8x = buildInput(startingN * 8);
-    // Warmup both with the largest input so JIT compiles all code paths.
-    fn(input8x);
-    linearBaseline(input8x);
-    const minTrials = 8;
-    const minElapsedMs = 200;
-    let bestRatio = Infinity;
-    let totalElapsed = 0;
-    let trials = 0;
-    while (trials < minTrials || totalElapsed < minElapsedMs) {
-        const startFn4 = performance.now();
-        fn(input4x);
-        const fnTime4 = performance.now() - startFn4;
-        const startBase4 = performance.now();
-        linearBaseline(input4x);
-        const baseTime4 = performance.now() - startBase4;
-        const startFn8 = performance.now();
-        fn(input8x);
-        const fnTime8 = performance.now() - startFn8;
-        const startBase8 = performance.now();
-        linearBaseline(input8x);
-        const baseTime8 = performance.now() - startBase8;
-        const fnRatio = (fnTime8 / input8x.length) / (fnTime4 / input4x.length);
-        const baseRatio = (baseTime8 / input8x.length) / (baseTime4 / input4x.length);
-        const ratio = fnRatio / baseRatio;
-        bestRatio = Math.min(bestRatio, ratio);
-        totalElapsed += fnTime4 + fnTime8 + baseTime4 + baseTime8;
-        trials++;
-    }
-    // After baseline-normalization: linear ≈ 1, quadratic ≈ 2.
-    expect(bestRatio).toBeLessThan(1.5);
-}
-//# sourceMappingURL=test-helpers.js.map

package/dist/tests/test-helpers.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"test-helpers.js","sourceRoot":"","sources":["../../src/tests/test-helpers.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,UAAU,MAAM,YAAY,CAAA;AAEnC;;;;GAIG;AACH,MAAM,SAAS,GAAG;IAChB,qBAAqB;IACrB,oCAAoC;IACpC,UAAU;IACV,4BAA4B;IAC5B,aAAa;IACb,eAAe;IACf,mBAAmB;IACnB,sBAAsB;IACtB,yBAAyB;IACzB,qBAAqB;IACrB,wBAAwB;IACxB,uBAAuB;IACvB,oBAAoB;IACpB,mBAAmB;IACnB,sCAAsC;IACtC,0BAA0B;IAC1B,mBAAmB;IACnB,uCAAuC;IACvC,0BAA0B;IAC1B,mBAAmB;CACpB,CAAA;AAED;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAAC,SAAiB,EAAE,OAAe,IAAI;IACtE,MAAM,MAAM,GAAG,UAAU,CAAC,IAAI,CAAC,CAAA;IAC/B,IAAI,MAAM,GAAG,EAAE,CAAA;IACf,OAAO,MAAM,CAAC,MAAM,GAAG,SAAS,EAAE,CAAC;QACjC,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,SAAS,CAAC,MAAM,CAAC,CAAA;QACrD,MAAM,IAAI,SAAS,CAAC,KAAK,CAAC,CAAA;IAC5B,CAAC;IACD,OAAO,MAAM,CAAA;AACf,CAAC;AAED;;;;;GAKG;AACH,MAAM,cAAc,GAAG,CAAC,CAAS,EAAU,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC,CAAA;AAEvE;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,UAAU,mBAAmB,CACjC,EAA8B,EAC9B,UAAiC,EACjC,YAAoB,IAAI;IAExB,MAAM,OAAO,GAAG,UAAU,CAAC,SAAS,GAAG,CAAC,CAAC,CAAA;IACzC,MAAM,OAAO,GAAG,UAAU,CAAC,SAAS,GAAG,CAAC,CAAC,CAAA;IAEzC,qEAAqE;IACrE,EAAE,CAAC,OAAO,CAAC,CAAA;IACX,cAAc,CAAC,OAAO,CAAC,CAAA;IAEvB,MAAM,SAAS,GAAG,CAAC,CAAA;IACnB,MAAM,YAAY,GAAG,GAAG,CAAA;IACxB,IAAI,SAAS,GAAG,QAAQ,CAAA;IACxB,IAAI,YAAY,GAAG,CAAC,CAAA;IACpB,IAAI,MAAM,GAAG,CAAC,CAAA;IACd,OAAO,MAAM,GAAG,SAAS,IAAI,YAAY,GAAG,YAAY,EAAE,CAAC;QACzD,MAAM,QAAQ,GAAG,WAAW,CAAC,GAAG,EAAE,CAAA;QAClC,EAAE,CAAC,OAAO,CAAC,CAAA;QACX,MAAM,OAAO,GAAG,WAAW,CAAC,GAAG,EAAE,GAAG,QAAQ,CAAA;QAE5C,MAAM,UAAU,GAAG,WAAW,CAAC,GAAG,EAAE,CAAA;QACpC,cAAc,CAAC,OAAO,CAAC,CAAA;QACvB,MAAM,SAAS,GAAG,WAAW,CAAC,GAAG,EAAE,GAAG,UAAU,CAAA;QAEhD,MAAM,QAAQ,GAAG,WAAW,CAAC,GAAG,EAAE,CAAA;QAClC,EAAE,CAAC,OAAO,CAAC,CAAA;QACX,MAAM,OAAO,GAAG,WAAW,CAAC,GAAG,EAAE,GAAG,QAAQ,CAAA;QAE5C,MAAM,UAAU,GAAG,WAAW,CAAC,GAAG,EAAE,CAAA;QACpC,cAAc,CAAC,OAAO,CAAC,CAAA;QACvB,MAAM,SAAS,GAAG,WAAW,CAAC,GAAG,EAAE,GAAG,UAAU,CAAA;QAEhD,MAAM,OAAO,GAAG,CAAC,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,CAAA;QACvE,MAAM,SAAS,GAAG,CAAC,SAAS,GAAG,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,SAAS,GAAG,OAAO,CAAC,MAAM,CAAC,CAAA;QAC7E,MAAM,KAAK,GAAG,OAAO,GAAG,SAAS,CAAA;QACjC,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,SAAS,EAAE,KAAK,CAAC,CAAA;QACtC,YAAY,IAAI,OAAO,GAAG,OAAO,GAAG,SAAS,GAAG,SAAS,CAAA;QACzD,MAAM,EAAE,CAAA;IACV,CAAC;IAED,2DAA2D;IAC3D,MAAM,CAAC,SAAS,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC,CAAA;AACrC,CAAC"}