diff --git a/make-pdf/src/smartypants.ts b/make-pdf/src/smartypants.ts index 2dfe097e0..107817e3c 100644 --- a/make-pdf/src/smartypants.ts +++ b/make-pdf/src/smartypants.ts @@ -20,7 +20,12 @@ const CODE_ZONE_RE = /<(pre|code|script|style)\b[^>]*>[\s\S]*?<\/\1>/gi; const TAG_RE = /<[^>]+>/g; -const URL_RE = /\bhttps?:\/\/\S+/g; +// \u0000 is the placeholder sentinel (see PLACEHOLDER below). It must be +// excluded here: URLs are carved AFTER tags, so a URL sitting flush against +// a tag (`https://ex.com`) would otherwise let \S+ swallow +// the placeholder standing in for `` — that placeholder then never +// restores (restore is a single pass) and the tag is lost. +const URL_RE = /\bhttps?:\/\/[^\s\u0000]+/g; /** * Apply smartypants to an HTML string. Zones that should not be touched: diff --git a/make-pdf/test/render.test.ts b/make-pdf/test/render.test.ts index b54085ecb..a5a4d7f24 100644 --- a/make-pdf/test/render.test.ts +++ b/make-pdf/test/render.test.ts @@ -56,6 +56,34 @@ describe("smartypants", () => { expect(out).toContain("\u201cdetails\u201d"); }); + // A URL flush against a following tag (no whitespace between) used to let + // URL_RE's \S+ swallow the placeholder standing in for that tag. The + // placeholder never restored, so the tag vanished and its styling bled + // into the rest of the document. The test above misses this because its + // URL is followed by a space. + test("does not leak placeholders for a bare autolinked URL", () => { + const input = `
see https://ex.com ok
`; + const out = smartypants(input); + expect(out).not.toContain("SMARTPANTS_PRESERVED"); + expect(out).toContain(""); + expect(out).toBe(input); + }); + + test("does not leak placeholders for a bold URL", () => { + const input = `see https://ex.com ok
`; + const out = smartypants(input); + expect(out).not.toContain("SMARTPANTS_PRESERVED"); + expect(out).toContain(""); + expect(out).toBe(input); + }); + + test("does not leak placeholders when a URL is followed by punctuation", () => { + const input = `see https://ex.com, ok
`; + const out = smartypants(input); + expect(out).not.toContain("SMARTPANTS_PRESERVED"); + expect(out).toContain(""); + }); + test("does NOT touch HTML attribute values", () => { const out = smartypants(`link`); expect(out).toContain(`href="it's-a-test.html"`);