Source code for egaia.egaia_parsefn

import os
import mimetypes
import magic
import re

[docs]def parseFilename(path):
    """Parse a tagged file and return a triple: (basename, uuid, extension)."""
    
    filepath = os.path.basename(path)
    
    if not '.' in filepath:
        # Add an extension to extensionless files and directories
        ext = None
        if os.path.isdir(filepath):
            # Use "dir" extension for directories, so we can identify them 
            # from tree listings alone
            ext = '.dir'
        elif os.path.exists(filepath):
            # If we are processing a real file (not just a filename), 
            # get the filetype from the file signature
            mtype = magic.from_file(filepath, mime=True)
            ext = mimetypes.guess_extension(mtype, strict=True)
        if not ext:
            # Unknown mimetype
            ext = '.null'
        
        filepath = filepath + ext
    
    return separate(filepath)

[docs]def separate(filepath):
    """Search for the UUID in a string representing a filepath, and separate"""
    
    # Caution: re.search/match fails if we have more than one UUID in the path!
    re_uuid = re.compile("[0-F]{8}-[0-F]{4}-[0-F]{4}-[0-F]{4}-[0-F]{12}", re.I)
    try:
        uuid = re.findall(re_uuid, filepath)[-1]
        fileparts = filepath.split(uuid)
        return (fileparts[0].strip('.'), uuid, fileparts[1].strip('.'))
    except:
        (base, ext) = os.path.splitext(filepath)
        return (base, None, ext.strip('.'))

[docs]def getUuid(filepath):
    """Return the uuid for a tagged filename, if available, or None."""
    parts = parseFilename(filepath)
    return parts[1]

[docs]def getBasename(filepath):
    """Return the base part of a tagged filename."""
    parts = parseFilename(filepath)
    return parts[0]

[docs]def getExtension(filepath):
    """Return the extension for a filename."""
    parts = parseFilename(filepath)
    return parts[2]

[docs]def getFormat(filepath):
    """Return the format of a derivative file."""

    m = re.search("df-(.*?)\.", filepath)
    try:
        fmt = m.group(1)
    except:
        return None
    
    return fmt

def _cli(args):
    """egaia parsefn
    
    Parse a given filename and return the basename, uuid, and extension.
    This is intended primarily as a utility function for other tools. If only
    the filename is given, a triple is returned containing the base, uuid,
    and extension of the filename.
    
    Usage:
        egaia parsefn --help
        egaia parsefn [ --base | --uuid | --ext ] FILENAME
    
    """
    
    if args['--base']:
        print getBasename(args['FILENAME'])
    elif args['--uuid']:
        print getUuid(args['FILENAME'])
    elif args['--ext']:
        print getExtension(args['FILENAME'])
    else:
        print parseFilename(args['FILENAME'])