Directory mapper

gistfile1.py

import os
from collections import defaultdict


class DB(object):

    def __init__(self, location, columns):
        self.location = location  # path to the location
        self.columns = columns  # names of the table columns
        self.data = []

    def _group(self, group_by, results):
        """ Imitates the GROUP BY SQL operator.
        """
        grouped = defaultdict(list)
        for row in results:
            grouped[row[group_by]].append(row)
        return grouped

    def _folders_from_path(self, path):
        """ Helper functions that extracts a list of folders from a given
        path. A naive solution would be to split the string with delimiter '/',
        but that approach fails to deal with backspaces and double crosses.
        """
        folders = []
        while True:
            if path == '/':
                break
            path, folder = os.path.split(path)
            folders.append(folder)
        return folders[::-1]  # reversed

    def sync(self):
        """ A function that generates the db table out of the specified location.
        """
        for path, dirs, files in os.walk(self.location):
            for f in files:
                # Folders + the filename split by `.`
                all_columns = self._folders_from_path(path) + f.split('.')

                # The location of the first useful (real) column
                start = len(all_columns) - len(self.columns)

                # Get the last N columns
                values = all_columns[start:]

                # Generate the rows
                row = {}
                for key, val in zip(self.columns, values):
                    row[key] = val

                # Add the full path to the row
                row['full_path'] = os.path.join(path, f)

                self.data.append(row)

    def get(self, group_by=None, **conditions):
        """ Returns a list od rows that satisfy the conditions.
        Optionally, you can group the results.

        Usage:
        >>> db.get()  # get all the rows
        >>> db.get(project='my_project', version='v1')
        >>> db.get(image_id='img1', group_by='size')
        """
        self.sync()
        if conditions:
            results = []
            for row in self.data:
                if all(row[col] == conditions[col] for col in conditions):
                    results.append(row)
        else:
            results = self.data
        if group_by:
            results = self._group(group_by, results)
        return results

db = DB(location="/home/whoeverest/code/swissnet/previewer/projects",
        columns=[
            'project',
            'image_id',
            'size',
            'version',
            'extension',
        ])

print db.get(image_id='img1')
print db.get(group_by='project', version='v1')
print db.get(group_by='size')