From 3c0b8073be4880d50cca05bdd4776157083246ac Mon Sep 17 00:00:00 2001
From: Rob Wu <gwnRob@gmail.com>
Date: Wed, 22 Jan 2014 12:27:44 +0100
Subject: [PATCH 1/2] Try to recover from bad URI value
See https://github.com/mozilla/pdf.js/issues/4159
BAD (http://cms.di.unipi.it/files/bbec7791fac20e98127c77531e4031912392156c/testo.pdf)
<< /S /URI /URI /v#2findex.php#2fFile:Logo.png >>
GOOD (http://www.ioi2012.org/wp-content/uploads/2011/12/practice.pdf):
<< /S /URI /URI (http://127.0.0.1/v/index.php/File:Logo.png >>
The URL should be wrapped in parentheses, but sometimes it isn't.
Consequently, the value is interpreted as a Name (because of the leading "/"),
and the resulting object is `{name: "v/index.php/File:Logo.png" }`. Obviously,
this is not a string, so `url.indexOf` throws an error here.
---
src/shared/annotation.js | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/src/shared/annotation.js b/src/shared/annotation.js
index ea504e8aa..f54732ea0 100644
--- a/src/shared/annotation.js
+++ b/src/shared/annotation.js
@@ -640,7 +640,13 @@ var LinkAnnotation = (function LinkAnnotationClosure() {
if (action) {
var linkType = action.get('S').name;
if (linkType === 'URI') {
- var url = addDefaultProtocolToUrl(action.get('URI'));
+ var url = action.get('URI');
+ if (isName(url)) {
+ // Some bad PDFs do not put parentheses around relative URLs.
+ url = '/' + url.name;
+ } else {
+ url = addDefaultProtocolToUrl(url);
+ }
// TODO: pdf spec mentions urls can be relative to a Base
// entry in the dictionary.
if (!isValidUrl(url, false)) {
From b35ced8c9e8ca8c0004ab8187e0c1ab771b08be7 Mon Sep 17 00:00:00 2001
From: Rob Wu <gwnRob@gmail.com>
Date: Wed, 22 Jan 2014 12:39:42 +0100
Subject: [PATCH 2/2] RFC 3986-compliant isValidUrl (protocol parsing)
To avoid misinterpreting URLs like "/File:wikipedia.png" as a
non-relative URLs.
---
src/shared/util.js | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/src/shared/util.js b/src/shared/util.js
index 51fd42bee..312445565 100644
--- a/src/shared/util.js
+++ b/src/shared/util.js
@@ -223,7 +223,7 @@ var UnsupportedManager = PDFJS.UnsupportedManager =
function combineUrl(baseUrl, url) {
if (!url)
return baseUrl;
- if (url.indexOf(':') >= 0)
+ if (/^[a-z][a-z0-9+\-.]*:/i.test(url))
return url;
if (url.charAt(0) == '/') {
// absolute path
@@ -247,11 +247,13 @@ function isValidUrl(url, allowRelative) {
if (!url) {
return false;
}
- var colon = url.indexOf(':');
- if (colon < 0) {
+ // RFC 3986 (http://tools.ietf.org/html/rfc3986#section-3.1)
+ // scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
+ var protocol = /^[a-z][a-z0-9+\-.]*(?=:)/i.exec(url);
+ if (!protocol) {
return allowRelative;
}
- var protocol = url.substr(0, colon);
+ protocol = protocol[0].toLowerCase();
switch (protocol) {
case 'http':
case 'https':