Merge pull request #19992 from calixteman/hashtag_filename

Correctly handle files with a hash sign in their names (bug 1894166)

Author: calixteman

src/display/display_utils.js

@@ -338,31 +338,82 @@ function getPdfFilenameFromUrl(url, defaultFilename = "document.pdf") {
     warn('getPdfFilenameFromUrl: ignore "data:"-URL for performance reasons.');
     return defaultFilename;
   }
-  const reURI = /^(?:(?:[^:]+:)?\/\/[^/]+)?([^?#]*)(\?[^#]*)?(#.*)?$/;
-  //              SCHEME        HOST        1.PATH  2.QUERY   3.REF
-  // Pattern to get last matching NAME.pdf
-  const reFilename = /[^/?#=]+\.pdf\b(?!.*\.pdf\b)/i;
-  const splitURI = reURI.exec(url);
-  let suggestedFilename =
-    reFilename.exec(splitURI[1]) ||
-    reFilename.exec(splitURI[2]) ||
-    reFilename.exec(splitURI[3]);
-  if (suggestedFilename) {
-    suggestedFilename = suggestedFilename[0];
-    if (suggestedFilename.includes("%")) {
-      // URL-encoded %2Fpath%2Fto%2Ffile.pdf should be file.pdf
+
+  const getURL = urlString => {
+    try {
+      return new URL(urlString);
+    } catch {
       try {
-        suggestedFilename = reFilename.exec(
-          decodeURIComponent(suggestedFilename)
-        )[0];
+        return new URL(decodeURIComponent(urlString));
       } catch {
-        // Possible (extremely rare) errors:
-        // URIError "Malformed URI", e.g. for "%AA.pdf"
-        // TypeError "null has no properties", e.g. for "%2F.pdf"
+        try {
+          // Attempt to parse the URL using the document's base URI.
+          return new URL(urlString, "https://foo.bar");
+        } catch {
+          try {
+            return new URL(decodeURIComponent(urlString), "https://foo.bar");
+          } catch {
+            return null;
+          }
+        }
+      }
+    }
+  };
+
+  const newURL = getURL(url);
+  if (!newURL) {
+    // If the URL is invalid, return the default filename.
+    return defaultFilename;
+  }
+
+  const decode = name => {
+    try {
+      let decoded = decodeURIComponent(name);
+      if (decoded.includes("/")) {
+        decoded = decoded.split("/").at(-1);
+        if (decoded.test(/^\.pdf$/i)) {
+          return decoded;
+        }
+        return name;
+      }
+      return decoded;
+    } catch {
+      return name;
+    }
+  };
+
+  const pdfRegex = /\.pdf$/i;
+  const filename = newURL.pathname.split("/").at(-1);
+  if (pdfRegex.test(filename)) {
+    return decode(filename);
+  }
+
+  if (newURL.searchParams.size > 0) {
+    const values = Array.from(newURL.searchParams.values()).reverse();
+    for (const value of values) {
+      if (pdfRegex.test(value)) {
+        // If any of the search parameters ends with ".pdf", return it.
+        return decode(value);
+      }
+    }
+    const keys = Array.from(newURL.searchParams.keys()).reverse();
+    for (const key of keys) {
+      if (pdfRegex.test(key)) {
+        // If any of the search parameter keys ends with ".pdf", return it.
+        return decode(key);
       }
     }
   }
-  return suggestedFilename || defaultFilename;
+
+  if (newURL.hash) {
+    const reFilename = /[^/?#=]+\.pdf\b(?!.*\.pdf\b)/i;
+    const hashFilename = reFilename.exec(newURL.hash);
+    if (hashFilename) {
+      return decode(hashFilename[0]);
+    }
+  }
+
+  return defaultFilename;
 }
 
 class StatTimer {

test/integration/viewer_spec.mjs

@@ -1220,4 +1220,28 @@ describe("PDF viewer", () => {
       );
     });
   });
+
+  describe("Filename with a hash sign", () => {
+    let pages;
+
+    beforeEach(async () => {
+      pages = await loadAndWait("empty%23hash.pdf", ".textLayer .endOfContent");
+    });
+
+    afterEach(async () => {
+      await closePages(pages);
+    });
+
+    it("must extract the filename correctly", async () => {
+      await Promise.all(
+        pages.map(async ([browserName, page]) => {
+          const filename = await page.evaluate(() => document.title);
+
+          expect(filename)
+            .withContext(`In ${browserName}`)
+            .toBe("empty#hash.pdf");
+        })
+      );
+    });
+  });
 });

test/pdfs/.gitignore

@@ -726,3 +726,4 @@
 !chrome-text-selection-markedContent.pdf
 !bug1963407.pdf
 !issue19517.pdf
+!empty#hash.pdf

test/pdfs/empty#hash.pdf

@@ -193,6 +193,20 @@ describe("display_utils", function () {
         "document.pdf"
       );
     });
+
+    it("gets PDF filename with a hash sign", function () {
+      expect(getPdfFilenameFromUrl("/foo.html?file=foo%23.pdf")).toEqual(
+        "foo#.pdf"
+      );
+
+      expect(getPdfFilenameFromUrl("/foo.html?file=%23.pdf")).toEqual("#.pdf");
+
+      expect(getPdfFilenameFromUrl("/foo.html?foo%23.pdf")).toEqual("foo#.pdf");
+
+      expect(getPdfFilenameFromUrl("/foo%23.pdf?a=b#c")).toEqual("foo#.pdf");
+
+      expect(getPdfFilenameFromUrl("foo.html#%23.pdf")).toEqual("#.pdf");
+    });
   });
 
   describe("isValidFetchUrl", function () {

test/unit/display_utils_spec.js

@@ -727,6 +727,7 @@ const PDFViewerApplication = {
       const queryString = document.location.search.substring(1);
       const params = parseQueryString(queryString);
       file = params.get("file") ?? AppOptions.get("defaultUrl");
+      file = encodeURIComponent(file).replaceAll("%2F", "/");
       validateFileURL(file);
     } else if (PDFJSDev.test("MOZCENTRAL")) {
       file = window.location.href;
@@ -2336,7 +2337,7 @@ if (typeof PDFJSDev === "undefined" || PDFJSDev.test("GENERIC")) {
 
     this.open({
       url: URL.createObjectURL(file),
-      originalUrl: file.name,
+      originalUrl: encodeURIComponent(file.name),
     });
   };