mimetype file content not properly extracted sometimes

[rdeltour]

From [email protected] on December 28, 2007 13:35:44

(using epubcheck-0.9.2.jar)

epub file: http://www.hxa7241.org/articles/content/EpubGuide- hxa7241.epub (including correct mimetype file)
was zipped with: http://www.info-zip.org/Zip.html but produced error:

EpubGuide-hxa7241.epub: mimetype contains wrong type (application/
epub+zip expected)

The problem: In a zip file, there is an extra field between the file name
and the file content (it is zero-length only optionally).

To fix:

Considering the zip file as a byte array, then:

filename starts at [30]
content starts at [30 + filenameLength + extrafieldLength]

filenameLength is ([27] << 8) | [26]
extrafieldLength is ([29] << 8) | [28]
contentLength is ([21] << 24) | ([20] << 16) | ([19] << 8) | [18]

So, summarising comments for possible code would be (separating
deserializing from checking):

// read checkable values (filename, filecontent)
// open file stream

  // read some header values
     // read file content length
        // seek to 18, read 4 (low bytes first)
     // read filename length
        // seek +4, read 2 (low bytes first)
     // read extrafield length
        // read 2 (low bytes first)

  // read checkable values
     // read filename
        // read filenameLength bytes
     // read filecontent (maybe not all)
        // seek +extrafieldLength, read filecontentLength bytes

// check values
// check filename
// make string from filename bytes
// compare with "mimetype"
// check filecontent
// make string from filecontent bytes
// compare with "application/epub+zip"

But, probably it would be better to use java.util.zip.ZipFile...

Original issue: http://code.google.com/p/epubcheck/issues/detail?id=2

[rdeltour]

From [email protected] on December 31, 2007 13:07:45

Not a bug. OCF spec section 4 "ZIP Container" states that actual MIME type (i.e.,
the string "application/XXXX+zip") must begin at position 38. This file does not
satisfy this requirement.

$ od -c EpubGuide-hxa7241.zip | head -5
0000000 P K 003 004 \n \0 \0 \0 \0 \0 \0 ` 234 7 o a
0000020 253 , 024 \0 \0 \0 024 \0 \0 \0 \b \0 021 \0 m i
0000040 m e t y p e U T \r \0 \a @ 345 t G 200
0000060 < t G 0 O q G a p p l i c a t i
0000100 o n / e p u b + z i p P K 003 004 024

Status: Invalid

[rdeltour]

From [email protected] on December 31, 2007 13:07:45

Not a bug. OCF spec section 4 "ZIP Container" states that actual MIME type (i.e.,
the string "application/XXXX+zip") must begin at position 38. This file does not
satisfy this requirement.

$ od -c EpubGuide-hxa7241.zip | head -5
0000000 P K 003 004 \n \0 \0 \0 \0 \0 \0 ` 234 7 o a
0000020 253 , 024 \0 \0 \0 024 \0 \0 \0 \b \0 021 \0 m i
0000040 m e t y p e U T \r \0 \a @ 345 t G 200
0000060 < t G 0 O q G a p p l i c a t i
0000100 o n / e p u b + z i p P K 003 004 024

Status: Invalid