m4mstat.py

import requests
import sys
import xml
import hashlib
import getpass
import datetime

from collections import OrderedDict

from django.conf import settings
from django.contrib.auth.models import User

from reporter.models import Rendering

LOCAL_USER_FIELDS = (
    'first_name',
    'last_name',
    'username',
    'date_joined',
    'last_login',
    'email'
)
KC_PROFILE_FIELDS = ('organization',)

RENDERING_FIELDS = ('created', 'submission_count()', 'template__name')

KC_PROFILE_ENDPOINT = '{}/api/v1/profiles'.format(settings.KC_URL)
KC_FORM_ENDPOINT = '{}/api/v1/forms'.format(settings.KC_URL)

def print_tabular(list_of_dicts):
    # TODO: Handle data that includes tab characters
    if not len(list_of_dicts):
        return
    print '\t'.join(list_of_dicts[0].keys())
    for r in list_of_dicts:
        print '\t'.join(r.values())

def render_field(obj, field_name):
    if field_name.endswith('()'):
        is_method = True
        field_name = field_name[:-2]
    else:
        is_method = False
    field = get_related_field(obj, field_name)
    if is_method:
        field = field()
    if type(field) is datetime.datetime:
        # Excel compatibility
        return field.strftime('%Y-%m-%d %H:%M:%S')
    else:
        return str(field)

def get_related_field(obj, field_name):
    # Is there really not a Django utility function for this?
    path = field_name.split('__')
    field = obj
    for p in path:
       field = getattr(field, p)
    return field

def user_report():
    sys.stderr.write('Loading profiles from KC...')
    kc_response = requests.get(KC_PROFILE_ENDPOINT)
    kc_profiles = {p['username']: p for p in kc_response.json()}
    sys.stderr.write(' Done!\n')

    user_report = []
    for user in User.objects.all():
        row = OrderedDict()
        try:
            profile = kc_profiles[user.username]
        except KeyError:
            sys.stdout.write(u'No profile for {}!\n'.format(user.username))
            profile = None
        for f in LOCAL_USER_FIELDS:
            row[f] = render_field(user, f)
        for f in KC_PROFILE_FIELDS:
            if profile is not None:
                row[f] = profile[f]
            else:
                row[f] = str(None)
        user_report.append(row)
    print_tabular(user_report)

def project_report():
    print 'Please log in with a KC superuser.'
    kc_super_username = raw_input('Username: ')
    kc_super_password = getpass.getpass()
    project_report = []
    sys.stderr.write('Processing projects')
    for rendering in Rendering.objects.all():
        # Get a MD5 sum of the XML body of the KC form, hopefully to ease
        # identification. Once `Rendering` is changed to include which `Form`
        # was used, this won't be necessary
        parsed_url = urlparse(rendering.url)
        numeric_id = int(parsed_url.path.split('/')[-1])
        xml_url = '{}/{}/form.xml'.format(KC_FORM_ENDPOINT, numeric_id)
        kc_response = requests.get(
            xml_url, auth=(kc_super_username, kc_super_password))
        if kc_response.status_code != requests.codes.ok:
            sys.stderr.write(
                '\nHTTP {}: Failed to get XML for form {}!\n'.format(
                    kc_response.status_code, numeric_id)
            )
            form_md5 = str(None)
            form_title = str(None)
        else:
            root_el = xml.etree.ElementTree.fromstring(kc_response.content)
            # Yo dawg, I heard you liked XML...
            root_el = xml.etree.ElementTree.fromstring(
                root_el.text.encode('utf-8'))
            body_el = root_el.find('{http://www.w3.org/1999/xhtml}body')
            form_md5 = hashlib.md5(
                ''.join(body_el.itertext()).encode('utf-8')
            ).hexdigest()
            # There are better ways to get the title, but since we already have
            # the XML...
            title_el = root_el.find('.//{http://www.w3.org/1999/xhtml}title')
            form_title = title_el.text
            sys.stderr.write('.')
        #######################################################################

        row = OrderedDict()
        row['md5'] = form_md5
        row['title'] = form_title
        for f in RENDERING_FIELDS:
            row[f] = render_field(rendering, f)
        project_report.append(row)
    sys.stderr.write('\n')
    print_tabular(project_report)