Merge pull request #20038 from calixteman/bug1974112

Don't remove a dash at the end of a line when guessing urls (bug 1974112)

test/unit/autolinker_spec.js

@@ -195,4 +195,13 @@ describe("autolinker", function () {
       ["httptest@email.com", "mailto:httptest@email.com"],
     ]);
   });
+
+  it("shouldn't remove the dash when it's an the end of a line (bug 1974112)", function () {
+    testLinks([
+      [
+        "https://github.com/pypi/linehaul-cloud-\nfunction",
+        "https://github.com/pypi/linehaul-cloud-function",
+      ],
+    ]);
+  });
 });

web/autolinker.js

@@ -138,7 +138,7 @@ class Autolinker {
     this.#regex ??=
       /\b(?:https?:\/\/|mailto:|www\.)(?:[\S--[\p{P}<>]]|\/|[\S--[\[\]]]+[\S--[\p{P}<>]])+|\b[\S--[@\p{Ps}\p{Pe}<>]]+@([\S--[\p{P}<>]]+(?:\.[\S--[\p{P}<>]]+)+)/gmv;
 
-    const [normalizedText, diffs] = normalize(text);
+    const [normalizedText, diffs] = normalize(text, { ignoreDashEOL: true });
     const matches = normalizedText.matchAll(this.#regex);
     const links = [];
     for (const match of matches) {

web/pdf_find_controller.js

@@ -97,7 +97,7 @@ const NFKC_CHARS_TO_NORMALIZE = new Map();
 let noSyllablesRegExp = null;
 let withSyllablesRegExp = null;
 
-function normalize(text) {
+function normalize(text, options = {}) {
   // The diacritics in the text or in the query can be composed or not.
   // So we use a decomposed text using NFD (and the same for the query)
   // in order to be sure that diacritics are in the same order.
@@ -118,6 +118,7 @@ function normalize(text) {
   }
 
   const hasSyllables = syllablePositions.length > 0;
+  const ignoreDashEOL = options.ignoreDashEOL ?? false;
 
   let normalizationRegex;
   if (!hasSyllables && noSyllablesRegExp) {
@@ -294,6 +295,12 @@ function normalize(text) {
       }
 
       if (p5) {
+        if (ignoreDashEOL) {
+          // Keep the - but remove the EOL.
+          shiftOrigin += 1;
+          eol += 1;
+          return p5.slice(0, -1);
+        }
         // In "X-\ny", "-\n" is removed because an hyphen at the end of a line
         // between two letters is likely here to mark a break in a word.
         // If X is encoded with UTF-32 then it can have a length greater than 1.