Use PdfDictionary to try and improve parsing

vbuch · Dec 1, 2023 · 6c4bb13 · 6c4bb13
1 parent cbf7e17
commit 6c4bb13
Show file tree

Hide file tree

Showing 8 changed files with 140 additions and 213 deletions.
diff --git a/packages/placeholder-plain/src/__snapshots__/readRefTable.test.js.snap b/packages/placeholder-plain/src/__snapshots__/readRefTable.test.js.snap
@@ -1,70 +1,6 @@
 // Jest Snapshot v1, https://goo.gl/fbAQLP
 
-exports[`getXref Throws an error when size has unexpected value 1`] = `"Failed to parse size of xref table."`;
-
-exports[`getXref Throws an error when size is not found 1`] = `"Size not found in xref table."`;
-
-exports[`getXref Throws an error when xref is not at its expected position 1`] = `"Cross-Reference Streams not yet implemented."`;
-
-exports[`getXref Throws an error when xref is not found at position 1`] = `"Cross-Reference Streams not yet implemented."`;
-
-exports[`readRefTable Expects to merge correctly the refTable of resources 1`] = `
-Object {
-  "maxIndex": 19,
-  "offsets": Map {
-    1 => 19012,
-    2 => 19,
-    3 => 224,
-    4 => 12330,
-    5 => 244,
-    6 => 11154,
-    7 => 11176,
-    8 => 11368,
-    9 => 11709,
-    10 => 11910,
-    11 => 11943,
-    12 => 12140,
-    13 => 12196,
-    14 => 18928,
-    15 => 12494,
-    17 => 13264,
-    18 => 18742,
-    19 => 18860,
-  },
-  "startingIndex": 0,
-}
-`;
-
-exports[`readRefTable Expects to merge correctly the refTable of resources 2`] = `
-Object {
-  "maxIndex": 21,
-  "offsets": Map {
-    1 => 25091,
-    2 => 19,
-    3 => 224,
-    4 => 12330,
-    5 => 244,
-    6 => 11154,
-    7 => 11176,
-    8 => 11368,
-    9 => 11709,
-    10 => 11910,
-    11 => 11943,
-    12 => 12140,
-    13 => 12196,
-    14 => 18948,
-    15 => 12494,
-    17 => 13264,
-    18 => 18742,
-    19 => 25016,
-    20 => 19431,
-    21 => 24878,
-  },
-  "startingIndex": 0,
-}
-`;
-
-exports[`readRefTable Expects to merge correctly the refTable of resources 3`] = `
+exports[`readRefTable Expects to merge correctly the refTable of contributing.pdf: contributing.pdf 1`] = `
 Object {
   "maxIndex": 24,
   "offsets": Map {
@@ -97,7 +33,7 @@ Object {
 }
 `;
 
-exports[`readRefTable Expects to merge correctly the refTable of resources 4`] = `
+exports[`readRefTable Expects to merge correctly the refTable of formexample.pdf: formexample.pdf 1`] = `
 Object {
   "maxIndex": 62,
   "offsets": Map {
@@ -168,7 +104,7 @@ Object {
 }
 `;
 
-exports[`readRefTable Expects to merge correctly the refTable of resources 5`] = `
+exports[`readRefTable Expects to merge correctly the refTable of incrementally_signed.pdf: incrementally_signed.pdf 1`] = `
 Object {
   "maxIndex": 18,
   "offsets": Map {
@@ -195,7 +131,7 @@ Object {
 }
 `;
 
-exports[`readRefTable Expects to merge correctly the refTable of resources 6`] = `
+exports[`readRefTable Expects to merge correctly the refTable of signed.pdf: signed.pdf 1`] = `
 Object {
   "maxIndex": 13,
   "offsets": Map {
@@ -217,7 +153,63 @@ Object {
 }
 `;
 
-exports[`readRefTable Expects to merge correctly the refTable of resources 7`] = `
+exports[`readRefTable Expects to merge correctly the refTable of signed-once.pdf: signed-once.pdf 1`] = `
+Object {
+  "maxIndex": 19,
+  "offsets": Map {
+    1 => 19012,
+    2 => 19,
+    3 => 224,
+    4 => 12330,
+    5 => 244,
+    6 => 11154,
+    7 => 11176,
+    8 => 11368,
+    9 => 11709,
+    10 => 11910,
+    11 => 11943,
+    12 => 12140,
+    13 => 12196,
+    14 => 18928,
+    15 => 12494,
+    17 => 13264,
+    18 => 18742,
+    19 => 18860,
+  },
+  "startingIndex": 0,
+}
+`;
+
+exports[`readRefTable Expects to merge correctly the refTable of signed-twice.pdf: signed-twice.pdf 1`] = `
+Object {
+  "maxIndex": 21,
+  "offsets": Map {
+    1 => 25091,
+    2 => 19,
+    3 => 224,
+    4 => 12330,
+    5 => 244,
+    6 => 11154,
+    7 => 11176,
+    8 => 11368,
+    9 => 11709,
+    10 => 11910,
+    11 => 11943,
+    12 => 12140,
+    13 => 12196,
+    14 => 18948,
+    15 => 12494,
+    17 => 13264,
+    18 => 18742,
+    19 => 25016,
+    20 => 19431,
+    21 => 24878,
+  },
+  "startingIndex": 0,
+}
+`;
+
+exports[`readRefTable Expects to merge correctly the refTable of w3dummy.pdf: w3dummy.pdf 1`] = `
 Object {
   "maxIndex": 15,
   "offsets": Map {

diff --git a/packages/placeholder-plain/src/findObject.js b/packages/placeholder-plain/src/findObject.js
@@ -1,4 +1,5 @@
 import getIndexFromRef from './getIndexFromRef';
+import PdfDictionary from './PdfDictionary';
 
 /**
  * @typedef {object} FindObjectAtReturnType
@@ -13,20 +14,20 @@ import getIndexFromRef from './getIndexFromRef';
  */
 export const findObjectAt = (pdf, position) => {
     let slice = pdf.subarray(position);
-    slice = slice.subarray(0, slice.indexOf('endobj', 'utf8') - 1);
-    // ^ Buffer from the start position until the first endobj (included).
+    slice = slice.subarray(slice.indexOf('obj') + 3, slice.indexOf('endobj', 'utf8') - 1);
+    // ^ Buffer from the start position until the first endobj.
 
     const dictionary = slice.subarray(
         slice.indexOf('<<', 'utf8') + 2,
-        slice.indexOf('>>', 'utf8') - 1,
+        slice.lastIndexOf('>>', 'utf8'),
     );
     const stream = slice.subarray(
         slice.indexOf('stream', 'utf8') + 6,
         slice.indexOf('endstream', 'utf8') - 1,
     );
 
     return {
-        dictionary,
+        dictionary: new PdfDictionary(dictionary),
         stream,
     };
 };

diff --git a/packages/placeholder-plain/src/getValue.js b/packages/placeholder-plain/src/getValue.js
diff --git a/packages/placeholder-plain/src/getValue.test.js b/packages/placeholder-plain/src/getValue.test.js
diff --git a/packages/placeholder-plain/src/readPdf.js b/packages/placeholder-plain/src/readPdf.js
@@ -1,6 +1,6 @@
-import readRefTable from './readRefTable';
-import findObject from './findObject';
-import {getValue} from './getValue';
+import {SignPdfError} from '@signpdf/utils';
+import readRefTable, {getLastXrefPosition} from './readRefTable';
+import findObject, {findObjectAt} from './findObject';
 
 /**
  * @typedef {object} ReadPdfReturnType
@@ -23,26 +23,33 @@ import {getValue} from './getValue';
  */
 const readPdf = (pdfBuffer) => {
     // Extract the trailer dictionary.
-    const trailerStart = pdfBuffer.lastIndexOf('trailer');
-    // The trailer is followed by xref. Then an EOF. EOF's length is 6 characters.
-    const trailer = pdfBuffer.slice(trailerStart, pdfBuffer.length - 6);
-
-    let xRefPosition = trailer.slice(trailer.lastIndexOf('startxref') + 10).toString();
-
-    xRefPosition = parseInt(xRefPosition);
-    const refTable = readRefTable(pdfBuffer);
-
-    const rootRef = getValue(trailer, '/Root');
+    const xRefPosition = getLastXrefPosition(pdfBuffer);
+
+    let refTable;
+
+    const trailerObject = findObjectAt(pdfBuffer, xRefPosition);
+    if (trailerObject.stream.indexOf('trailer') !== -1) {
+        // assuming trailer
+        refTable = readRefTable(pdfBuffer, xRefPosition);
+    } else {
+        // assuming stream
+        if (!trailerObject.dictionary.has('/Filter')) {
+            throw new Error('Expected /Filter in trailer with streams.');
+        }
+        throw new SignPdfError(
+            '/Filter is not implemented.',
+            SignPdfError.TYPE_PARSE,
+        );
+    }
+    const rootRef = trailerObject.dictionary.get('/Root');
     const root = findObject(pdfBuffer, refTable, rootRef).toString();
-
-    const infoRef = getValue(trailer, '/Info');
+    const infoRef = trailerObject.dictionary.get('/Info');
 
     return {
         xref: refTable,
         rootRef,
         root,
         infoRef,
-        trailerStart,
         xRefPosition,
     };
 };

diff --git a/packages/placeholder-plain/src/readPdf.test.js b/packages/placeholder-plain/src/readPdf.test.js
@@ -2,20 +2,30 @@ import {readTestResource} from '@signpdf/internal-utils';
 import readPdf from './readPdf';
 
 describe(readPdf, () => {
-    it('reads contributing.pdf', () => {
-        const pdfBuffer = readTestResource('contributing.pdf');
+    it.each([
+        {
+            resource: 'signed-once.pdf',
+            xRefPosition: 19174,
+            root: 14,
+            rootByteOffset: 18928,
+            info: 15,
+        },
+        {
+            resource: 'contributing.pdf',
+            xRefPosition: 72203,
+            root: 12,
+            rootByteOffset: 4394,
+            info: 1,
+        },
+    ])('reads $resource', ({
+        resource, root, info, xRefPosition, rootByteOffset,
+    }) => {
+        const pdfBuffer = readTestResource(resource);
         const result = readPdf(pdfBuffer);
 
-        expect(result.xRefPosition).toBe(72203);
-        expect(result.rootRef).toBe('12 0 R');
-        expect(result.infoRef).toBe('1 0 R');
-    });
-    it('reads issue-79-test.pdf', () => {
-        const pdfBuffer = readTestResource('issue-79-test.pdf');
-        const result = readPdf(pdfBuffer);
-
-        expect(result.xRefPosition).toBe(1542);
-        expect(result.rootRef).toBe('2 0 R');
-        expect(result.infoRef).toBe('3 0 R');
+        expect(result.xRefPosition).toBe(xRefPosition);
+        expect(result.rootRef).toBe(`${root} 0 R`);
+        expect(result.infoRef).toBe(`${info} 0 R`);
+        expect(result.xref.offsets.get(root)).toBe(rootByteOffset);
     });
 });