Creates a .yaml file 'psuedo-schema' from a folder directory (including ESRI gdb's)

gdb_schema_yaml.py

# -*- coding: utf-8 -*-
"""distribution_yaml.py: Converts Paths and GBDs to dicts, then to a yaml file.

This script 'walks' through a folder structure (including ESRI Geodatabases)
and outputs a .yaml pseudo-schema of the directory-tree and geodatabase(s).

MIT License (MIT)
Copyright (C) 2015 Garin Wally / Windfall Spatial

This code was modified from code by Blake Miller, found here:
http://btmiller.com/2015/03/17/represent-file-structure-as-yaml-with-python.html
"""

import os
import sys

import arcpy
import yaml


def gdb2dict(path):
    """Creates a dictionary from a Geodatabase schema.
    """
    gdb_dir = {}
    
    for gdbname, dsnames, fcs in arcpy.da.Walk(path):
        gdb = os.path.basename(gdbname)
        gdb_dir[gdb] = []
        
        if dsnames:
            for ds in dsnames:
                gdb_dir[gdb].append(gdb2dict(os.path.join(path, ds)))
        
            for fc in fcs:
                gdb_dir[gdb].append(fc)
                
        else:
            gdb_dir[gdb] = fcs
            
        return gdb_dir


def dir2dict(path):
    """Creates a dictionary from a directory tree.
    Credit: Blake Miller
    """
    directory = {}

    for dirname, dirnames, filenames in os.walk(path):
        dn = os.path.basename(dirname)
        directory[dn] = []

        if dirnames:
            for d in dirnames:
                if not d.endswith(".gdb"):
                    directory[dn].append(dir2dict(path=os.path.join(path, d)))
                    
                elif d.endswith(".gdb"):
                    gdb = os.path.join(path, d)
                    directory[dn].append(gdb2dict(gdb))

            for f in filenames:
                directory[dn].append(f)
                
        else:
            directory[dn] = filenames

        return directory


def path2yaml(input_path, output_yaml=None):
    """Converts an input_dict to an output_yaml file."""
    if output_yaml is None or output_yaml == "":
        output_yaml = "{}.yaml".format(os.path.basename(input_path))
        
    assert output_yaml.endswith(".yaml"), \
        "Output file must have extension '.yaml'"
    
    if input_path.endswith(".gdb"):
        out_file = os.path.join(os.path.dirname(input_path), output_yaml)
    else:
        out_file = os.path.join(input_path, output_yaml)
    
    with open(out_file, 'w') as f:
        if not input_path.endswith(".gdb"):
            yaml.safe_dump(dir2dict(input_path), f,
                           default_flow_style=False, allow_unicode=True)
        else:
            yaml.safe_dump(gdb2dict(input_path), f,
                           default_flow_style=False, allow_unicode=True)
        
    print("Dict written to {}".format(out_file))
    return


# Allow script to be run from either commandline or with raw_input
if __name__ == '__main__':
    try:
        if len(sys.argv) == 1:
            path2yaml(os.path.abspath(sys.argv[1]))
        elif len(sys.argv) == 2:
            path2yaml(os.path.abspath(sys.argv[1]), sys.argv[2])
    except IndexError:
        path2yaml(raw_input("Enter directory or path: "),
                  raw_input("Enter an output .yaml name (Optional) "))