addresses #120 (URL percent decode / sanitize)

readium · Nov 10, 2014 · bc86c60 · bc86c60
1 parent cb174c1
commit bc86c60
Show file tree

Hide file tree

Showing 5 changed files with 55 additions and 22 deletions.
diff --git a/ePub3/ePub/package.cpp b/ePub3/ePub/package.cpp
@@ -38,6 +38,10 @@
 #include <ePub3/xml/document.h>
 #include <ePub3/xml/element.h>
 
+//#include "iri.h"
+#include <google-url/url_canon.h>
+#include <google-url/url_util.h>
+
 EPUB3_BEGIN_NAMESPACE
 
 #define _XML_OVERRIDE_SWITCHES (EPUB_USE(LIBXML2) && PROMISCUOUS_LIBXML_OVERRIDES == 0)
@@ -174,6 +178,30 @@ ConstManifestItemPtr PackageBase::ManifestItemAtRelativePath(const string& path)
 		if (item.second->AbsolutePath() == absPath)
 			return item.second;
 	}
+
+    // Edge case...
+    // before giving up, let's check for lower/upper-case percent encoding mismatch (e.g. %2B vs. %2b)
+
+    //if ( path.find("%") != std::string::npos ) SOMETIMES OPF MANIFEST ITEM HREF IS PERCENT-ESCAPED, BUT NOT HTML SRC !!
+
+    url_canon::RawCanonOutputW<256> output;
+    url_util::DecodeURLEscapeSequences(path.c_str(), static_cast<int>(path.size()), &output);
+    string path_(output.data(), output.length());
+
+    string absPath_ = _pathBase + (path_[0] == '/' ? path_.substr(1) : path_);
+    for (auto& item : _manifest)
+    {
+        string absolute = item.second->AbsolutePath();
+
+        url_canon::RawCanonOutputW<256> output_;
+        url_util::DecodeURLEscapeSequences(absolute.c_str(), static_cast<int>(absolute.size()), &output_);
+        string absolute_(output_.data(), output_.length());
+
+        if (absolute_ == absPath_)
+            return item.second;
+    }
+
+    printf("MISSING ManifestItemAtRelativePath %s (%s)\n", path.c_str(), absPath.c_str());
 	return nullptr;
 }
 shared_ptr<NavigationTable> PackageBase::NavigationTable(const string &title) const

diff --git a/ePub3/ePub/zip_archive.cpp b/ePub3/ePub/zip_archive.cpp
@@ -276,12 +276,6 @@ ArchiveItemInfo ZipArchive::InfoAtPath(const string & path) const
         throw std::runtime_error(std::string("zip_stat("+path.stl_str()+") - " + zip_strerror(_zip)));
     return ZipItemInfo(sbuf);
 }
-string ZipArchive::Sanitized(const string& path) const
-{
-    if ( path.find('/') == 0 )
-        return path.substr(1);
-    return path;
-}
 
 void ZipWriter::DataBlob::Append(const void *data, size_t len)
 {

diff --git a/ePub3/ePub/zip_archive.h b/ePub3/ePub/zip_archive.h
@@ -89,10 +89,7 @@ class ZipArchive : public Archive
 
     typedef std::list<zip_source*>  ZipSourceList;
     ZipSourceList   _liveSources;   ///< A list of live zip sources, which must be cleaned up upon closing.
-
-    ///
-    /// Sanitizes a path string, since `libzip` can be finnicky about them.
-    string Sanitized(const string& path) const;
+
 };
 
 EPUB3_END_NAMESPACE

diff --git a/ePub3/utilities/byte_stream.cpp b/ePub3/utilities/byte_stream.cpp
@@ -763,12 +763,7 @@ bool ZipFileByteStream::IsOpen() const _NOEXCEPT
 {
     return _file != nullptr;
 }
-string ZipFileByteStream::Sanitized(const string& path) const
-{
-    if ( path.find('/') == 0 )
-        return path.substr(1);
-    return path;
-}
+
 bool ZipFileByteStream::Open(struct zip *archive, const string &path, int flags)
 {
     if ( _file != nullptr )

diff --git a/ePub3/utilities/byte_stream.h b/ePub3/utilities/byte_stream.h
@@ -32,9 +32,31 @@
 struct zip;
 struct zip_file;
 
+//#include "iri.h"
+#include <google-url/url_canon.h>
+#include <google-url/url_util.h>
+
 EPUB3_BEGIN_NAMESPACE
 
-/**
+static string Sanitized(const string& path)
+{
+    if ( path.find("%") != std::string::npos )
+    {
+        url_canon::RawCanonOutputW<256> output;
+        url_util::DecodeURLEscapeSequences(path.c_str(), static_cast<int>(path.size()), &output);
+        string path_(output.data(), output.length());
+
+        if ( path_.find('/') == 0 )
+            return path_.substr(1);
+        return path_;
+    }
+
+    if ( path.find('/') == 0 )
+        return path.substr(1);
+    return path;
+}
+
+    /**
  The abstract base class for all stream and pipe objects used by the Readium SDK.
  
  This class declares the standard interface for a stream-- that is, an object to
@@ -57,7 +79,7 @@ class ByteStream
     ///
     /// A value to be returned when a real count is not possible.
     static const size_type          UnknownSize = 0;
-    
+
 public:
                              ByteStream() : _eof(false), _err(0)    {}
     virtual                 ~ByteStream()                           {}
@@ -606,10 +628,7 @@ class ZipFileByteStream : public SeekableByteStream
 protected:
     struct zip_file*        _file;      ///< The underlying Zip file stream.
 	std::ios::openmode		_mode;		///< The mode used to open the file (used by Clone()).
-
-    ///
-    /// B
-    string Sanitized(const string& path) const;
+
 };
 
 #ifdef SUPPORT_ASYNC