Migrate file formats from Hachior project (in python)

[tonyqus]

I found that there is a long file format parser list from Hachior project. I have thought of migrating the whole Hachoir to .NET with the project name NHachoir. But it looks like your project works well. I'm considering directly migrating some file format parsers to your project.

Let me know your thoughts on this.

[tonyqus]

• implement 7zip file format parser #76
• Implement GZip file parser #77
• Implement Flac file format parser #78
• Implement RAR file format parser #79
• Implement Tar file format parser #80
• Implement Midi file format parser #81
• Implement Wmf file format parser #82
• implement PSD file format parser #83
• Implement Flv file format parser #84

[neilharvey]

Sure, feel free to send some PRs for formats if you like. It's probably only worth focusing on ones which are likely to be commonly used though. The original intent with this library was that I wanted the list of formats to be easily extensible and customisable to fit the use case so that people can plug in the formats they want to use rather than it being a comprehensive 'catch all' solution.

[tonyqus]

Ok. I chose a few commonly used file formats from the Hachior file format list.

TODO List

• rar: Roshal archive (RAR)
• tar: TAR archive
• swf: Macromedia Flash data
• flv: Macromedia Flash video
• wmf: Microsoft Windows Metafile (WMF)
• targa: Truevision Targa Graphic (TGA)
• midi: MIDI audio
• mod: Uncompressed amiga module
• flac: FLAC audio
• psd: Photoshop (PSD) picture

[neilharvey]

Sure, that looks like a reasonable list - thanks.

Although you don't need to worry about Quicktime and Ogg as we already have those covered. I think the rest are all new :)

[tonyqus]

Can you merge #79 first? Since I need to use the newly added FileFormat constructor.

[neilharvey]

Sorry, but could you write the signature headers in hex instead please? It is much easier to debug issues because we are just comparing like-for-like rather than converting between text and hex.

[tonyqus]

I don't think string signature may hurt your debugging capacity. It still converts string to hex byte array. There is nothing different.

And Hachior also uses string signature
https://github.com/vstinner/hachoir/blob/e4b13af02e19cae27231b8466a1949125786c88b/hachoir/parser/audio/flac.py#L150

[neilharvey]

I disagree, sorry. I prefer to compare hex to hex than have to translate from the string representation.

I also have a concern that Convert.ToByte to parse the string could have some unexpected runtime behaviour as it is based on the consumer's current culture.

[tonyqus]

Since it's your project, I'd like to follow your rule. The code in #79 is revoked. No string signature anymore.

[neilharvey]

Thanks, I've merged in the latest round of PRs. I've also implemented swf from the todo list.