getPDFSignatureCount: fix "too much recursion" error with some PDFs

The regex used to detect signatures uses "(.|\\s)" to emulate dotall, since
old browsers don't support it.

For some reason that leads to internal errors in Firefox with larger files in some cases.

Replacing it with a simpler "[\\s\\S]" class makes things work.

Also adds some tests with real PDF files, "QPDF-367-0.pdf" was triggering this issue.

Fixes #46

karma.conf.js

@@ -9,7 +9,8 @@ module.exports = function (config) {
         },
         files: [
             {pattern: './*.js', included: true, watched: true, served: true, type: 'module'},
-            {pattern: './**/*', included: false, watched: true, served: true},
+            // XXX: nocache is required or karma serves garbage binary data for some reason
+            {pattern: './**/*', included: false, watched: true, served: true, nocache: true},
         ],
         autoWatch: true,
         browsers: ['ChromiumHeadlessNoSandbox', 'FirefoxHeadless'],

rollup.config.js

@@ -183,6 +183,14 @@ Dependencies:
                 emitFiles: true,
                 fileName: 'shared/[name].[hash][extname]',
             }),
+            appEnv == 'test' && copy({
+                targets: [
+                    {
+                        src: 'test/data/*',
+                        dest: 'dist/test',
+                    },
+                ]
+            }),
             copy({
                 targets: [
                     {

src/utils.js

@@ -132,7 +132,7 @@ export const readArrayBufferFileContent = async (file) => {
  */
 export const getPDFSignatureCount = async (file) => {
     const sigRegex = new RegExp(
-        '/Type\\s*/Sig(.|\\s)*?/SubFilter\\s*(/ETSI\\.CAdES\\.detached|/adbe\\.pkcs7\\.detached)',
+        '/Type\\s*/Sig[\\s\\S]*?/SubFilter\\s*(/ETSI\\.CAdES\\.detached|/adbe\\.pkcs7\\.detached)',
         'g'
     );
     const content = await readBinaryFileContent(file);

test/data/README.md

+QPDF-367-0.pdf:
+    From https://corpora.tika.apache.org/base/packaged/pdfs/archive/pdfs_202002/qpdf.zip
\ No newline at end of file

test/unit.js

@@ -66,4 +66,18 @@ suite('pdf signature detection', () => {
         assert((await getPDFSignatureCount(getPDFFile('foobar'))) === 0);
         assert((await getPDFSignatureCount(getPDFFile(''))) === 0);
     });
+
+    test('getPDFSignatureCount real files', async () => {
+
+        async function getRealPDFFile(name) {
+            let url = new URL('test/' + name, import.meta.url).href;
+            let resp = await fetch(url);
+            assert(resp.ok);
+            return getPDFFile(await resp.arrayBuffer());
+        }
+
+        assert.equal((await getPDFSignatureCount(await getRealPDFFile('QPDF-367-0.pdf'))), 0);
+        assert.equal((await getPDFSignatureCount(await getRealPDFFile('qual-sig-simple.pdf'))), 1);
+        assert.equal((await getPDFSignatureCount(await getRealPDFFile('qual-sig-tugraz-multiple.pdf'))), 2);
+    });
 });