Sort ZIP Entries again

[dlitz]

This ensures that the entries in the zip file's central directory are grouped
together by folder, with each folder's directory entry coming before the
folder's contents, rather than being in the random order dictated by Ruby's
Hash implementation.

Before:

  $ unzip -l before.zip
  Archive:  before.zip
    Length      Date    Time    Name
  ---------  ---------- -----   ----
          7  03-11-2011 14:14   fruit/apple
          8  03-11-2011 14:14   fruit/orange
          6  03-11-2011 14:14   fruit/kiwi
          0  03-11-2011 14:14   uuid/5D8976B4-C3B5-4FC6-A9BA-E307173F0064
          0  03-11-2011 14:14   uuid/
          0  03-11-2011 14:14   uuid/B1D55A18-860D-4BE1-BCB3-7DFBCD7C8BC4
          0  03-11-2011 14:14   uuid/E5E1F8F6-60A3-4733-9648-86711E7CD02A
          0  03-11-2011 14:14   fruit/
          8  03-11-2011 14:14   fruit/mango
          0  03-11-2011 14:14   Vegetable/
          0  03-11-2011 14:14   uuid/47FD714B-1A38-46AB-B9EF-0E8B2098D3A5
          0  03-11-2011 14:14   uuid/8B6FD77A-9A4C-4D74-B1E0-5B0EBBF0C3A9
          7  03-11-2011 14:14   Vegetable/celery
          8  03-11-2011 14:14   Vegetable/carrot
          6  03-11-2011 14:14   Vegetable/bean
  ---------                     -------
         50                     15 files

After:

  $ unzip -l after.zip
  Archive:  after.zip
    Length      Date    Time    Name
  ---------  ---------- -----   ----
          0  03-11-2011 14:15   Vegetable/
          6  03-11-2011 14:15   Vegetable/bean
          8  03-11-2011 14:15   Vegetable/carrot
          7  03-11-2011 14:15   Vegetable/celery
          0  03-11-2011 14:15   fruit/
          7  03-11-2011 14:15   fruit/apple
          6  03-11-2011 14:15   fruit/kiwi
          8  03-11-2011 14:15   fruit/mango
          8  03-11-2011 14:15   fruit/orange
          0  03-11-2011 14:15   uuid/
          0  03-11-2011 14:15   uuid/47FD714B-1A38-46AB-B9EF-0E8B2098D3A5
          0  03-11-2011 14:15   uuid/5D8976B4-C3B5-4FC6-A9BA-E307173F0064
          0  03-11-2011 14:15   uuid/8B6FD77A-9A4C-4D74-B1E0-5B0EBBF0C3A9
          0  03-11-2011 14:15   uuid/B1D55A18-860D-4BE1-BCB3-7DFBCD7C8BC4
          0  03-11-2011 14:15   uuid/E5E1F8F6-60A3-4733-9648-86711E7CD02A
  ---------                     -------
         50                     15 files

Similar functionality has been committed before (see commit 74e4512),
but was reverted in commit 963f23c because it
breaks things (like EPUB) that require entries to appear in a particular order.

This re-enables sorting by default, but still allows custom ZipEntrySet
implementations to be used when a custom sort order is required.

[simonoff]

Sorry but I think better way when you will sort entries before include them into zip archive.
In this case EPUB files will be work and unzip will work too.
ZipEntrySet is not a Hash, it class with Enumeration mixin.

[dlitz]

Alex, please take a closer look at the code:

https://github.com/aussiegeek/rubyzip/blob/156c0c04b54857ab07df845c6722148e1b821683/lib/zip/zip_entry_set.rb:

    def initialize(anEnumerable = [])
      super()
      @entrySet = {}
      anEnumerable.each { |o| push(o) }
    end

    # ...

    def each(&aProc)
      @entrySet.values.each(&aProc)
    end

It's definitely returning entries in arbitrary Hash order.

[simonoff]

Hm.. ok. In any case I will look deep on it tomorrow for better understanding how to fix EPUP files and sorting.

[dlitz]

Ping?

[marcuserronius]

I second not sorting by filename, for epub related reasons. Under Ruby 1.9, ordered hashes currently help with this, but it's just a side effect; and it would be nice to have the option of defining the exact order of the files in the archive. I'm looking into how much work would be required for this.

[marcuserronius]

Just issued a pull request for changes to ZipEntrySet that preserve the original ordering, and insertion ordering. If desired, one could sort the entries and reinsert them to get whatever ordering one wants :)

[simonoff]

@dlitz sorry for long delay. Can we continue conversation to choose proper way of fix this bug?

[dlitz]

@simonoff Sorry I didn't respond (I no longer do much Ruby work). I would have made :sort_entries=>true the default (most users don't need a specific ordering for .zip files, so it makes sense to just do the right thing for them), but otherwise it looks good to me.

Thanks!

[simonoff]

@dlitz I think i will release this feature and will look what others thinks. May be you will be right and I will set sort_entries to true by default.

PS. To where you are moving? Node? Clojure? Erlang?

[dlitz]

I went back to C and Python; I'm working at Dropbox now.

Cheers!
Dwayne

Alexander Simonov notifications@github.com wrote:

@dlitz I think i will release this feature and will look what others
thinks. May be you will be right and I will set sort_entries to true by
default.

PS. To where you are moving? Node? Clojure? Erlang?

---

Reply to this email directly or view it on GitHub:
#28 (comment)

Sent from my Android device with K-9 Mail. Please excuse my brevity.