Sakurai/pdf.js
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Re-factor searching for incomplete objects in XRef.indexObjects (issue 15803)

Browse Source 
When trying to find incomplete objects, i.e. those missing the "endobj"-string at the end, there's unfortunately a number of possible operators that we need to check for. Otherwise we could miss e.g. the "trailer" at the end of a corrupt PDF document, which is why the referenced document didn't work.

Currently we do all searching on the "raw" bytes of the PDF document, for efficiency, however this doesn't really work when we need to check for *multiple* potential command-strings. To keep the complexity manageable we'll instead use regular expressions here, but we can at least avoid creating lots of substrings thanks to the `RegExp.lastIndex` property; which is well supported across browsers according to https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp/lastIndex#browser_compatibility

Note that this repeated regular expression usage could perhaps be slightly less efficient than the old code, however this method is only invoked for corrupt PDF documents.
This commit is contained in:
Jonas Jenwald 2022-12-19 23:01:09 +01:00
parent 6a9a567670
commit 2fcf8bb5be
3 changed files with 39 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

68
src/core/xref.js
View File

@ -431,16 +431,14 @@ class XRef {
}
return skipped;
}
const gEndobjRegExp = /\b(endobj|\d+\s+\d+\s+obj|xref|trailer)\b/g;
const gStartxrefRegExp = /\b(startxref|\d+\s+\d+\s+obj)\b/g;
const objRegExp = /^(\d+)\s+(\d+)\s+obj\b/;
const endobjRegExp = /\bendobj[\b\s]$/;
const nestedObjRegExp = /\s+(\d+\s+\d+\s+obj[\b\s<])$/;
const CHECK_CONTENT_LENGTH = 25;
const trailerBytes = new Uint8Array([116, 114, 97, 105, 108, 101, 114]);
const startxrefBytes = new Uint8Array([
115, 116, 97, 114, 116, 120, 114, 101, 102,
]);
const objBytes = new Uint8Array([111, 98, 106]);
const xrefBytes = new Uint8Array([47, 88, 82, 101, 102]);
// Clear out any existing entries, since they may be bogus.
@ -450,6 +448,7 @@ class XRef {
const stream = this.stream;
stream.pos = 0;
const buffer = stream.getBytes(),
bufferStr = bytesToString(buffer),
length = buffer.length;
let position = stream.start;
const trailers = [],
@ -484,8 +483,8 @@ class XRef {
const num = m[1] | 0,
gen = m[2] | 0;
const startPos = position + token.length;
let contentLength,
startPos = position + token.length,
updateEntries = false;
if (!this.entries[num]) {
updateEntries = true;
@ -519,31 +518,22 @@ class XRef {
// Find the next "obj" string, rather than "endobj", to ensure that
// we won't skip over a new 'obj' operator in corrupt files where
// 'endobj' operators are missing (fixes issue9105_reduced.pdf).
while (startPos < length) {
const endPos = startPos + skipUntil(buffer, startPos, objBytes) + 4;
gEndobjRegExp.lastIndex = startPos;
const match = gEndobjRegExp.exec(bufferStr);
if (match) {
const endPos = gEndobjRegExp.lastIndex + 1;
contentLength = endPos - position;
const checkPos = Math.max(endPos - CHECK_CONTENT_LENGTH, startPos);
const tokenStr = bytesToString(buffer.subarray(checkPos, endPos));
// Check if the current object ends with an 'endobj' operator.
if (endobjRegExp.test(tokenStr)) {
break;
} else {
// Check if an "obj" occurrence is actually a new object,
// i.e. the current object is missing the 'endobj' operator.
const objToken = nestedObjRegExp.exec(tokenStr);
if (objToken && objToken[1]) {
warn(
'indexObjects: Found new "obj" inside of another "obj", ' +
'caused by missing "endobj" -- trying to recover.'
);
contentLength -= objToken[1].length;
break;
}
if (match[1] !== "endobj") {
warn(
`indexObjects: Found "${match[1]}" inside of another "obj", ` +
'caused by missing "endobj" -- trying to recover.'
);
contentLength -= match[1].length + 1;
}
startPos = endPos;
} else {
contentLength = length - position;
}
const content = buffer.subarray(position, position + contentLength);
@ -562,26 +552,26 @@ class XRef {
) {
trailers.push(position);
const contentLength = skipUntil(buffer, position, startxrefBytes);
const startPos = position + token.length;
let contentLength;
// Attempt to handle (some) corrupt documents, where no 'startxref'
// operators are present (fixes issue15590.pdf).
if (position + contentLength >= length) {
const endPos = position + skipUntil(buffer, position, objBytes) + 4;
gStartxrefRegExp.lastIndex = startPos;
const match = gStartxrefRegExp.exec(bufferStr);
const checkPos = Math.max(endPos - CHECK_CONTENT_LENGTH, position);
const tokenStr = bytesToString(buffer.subarray(checkPos, endPos));
if (match) {
const endPos = gStartxrefRegExp.lastIndex + 1;
contentLength = endPos - position;
// Find the first "obj" occurrence after the 'trailer' operator.
const objToken = nestedObjRegExp.exec(tokenStr);
if (objToken && objToken[1]) {
if (match[1] !== "startxref") {
warn(
'indexObjects: Found first "obj" after "trailer", ' +
`indexObjects: Found "${match[1]}" after "trailer", ` +
'caused by missing "startxref" -- trying to recover.'
);
position = endPos - objToken[1].length;
continue;
contentLength -= match[1].length + 1;
}
} else {
contentLength = length - position;
}
position += contentLength;
} else {

1
test/pdfs/issue15803.pdf.link Normal file
View File

@ -0,0 +1 @@
https://github.com/mozilla/pdf.js/files/10200431/ocg.pdf

9
test/test_manifest.json
View File

@ -1761,6 +1761,15 @@
"link": false,
"type": "eq"
},
{ "id": "issue15803",
"file": "pdfs/issue15803.pdf",
"md5": "e501a4418d4ece5be7ce4e8acf029100",
"rounds": 1,
"link": true,
"lastPage": 1,
"type": "eq",
"annotations": true
},
{ "id": "issue9105_other",
"file": "pdfs/issue9105_other.pdf",
"md5": "4c8b9c2cceb9c5d621e1d50b3dc38efc",