attribution_report/attribution.py

# -*- coding: utf-8 -*-
import os
import datetime
import logging
import subprocess

import pandas as pd
from dateutil.relativedelta import relativedelta

L = logging.getLogger(__name__)


class AttributionReport(object):
    def __init__(self, months=6, footer_length=None):
        self.months = months
        self.footer_length = footer_length

        self.SF_DATE_COLUMN = "Date"
        self.DP_DATE_COLUMN = "Date Received"
        self.PI_COLUMN = "PI_Name"
        self.ORG_COLUMN = "Org Name"
        # Output the XLSX in this order

        self.OUTPUT_COLUMN_ORDER = ["Addgene Assigned", "Plasmid ID", "Deposit ID", "Institute", "PI Name",
                                    "Date Received", "Original Date", "Original ORG", "Original PI"]

        self.ACCEPTABLE_EXTENSIONS = ["*.csv", "*.xls", "*.xlsx"]

        # columns that need to be in the files
        self.REQUIRED_SF_COLUMNS = ["First Name", "Last Name", "Account Name", "Date", "Assigned"]
        self.REQUIRED_DP_COLUMNS = ["Org Name", "Deposit ID", "Plasmid ID", "PI_Name", "Date Received"]

        # After load and merging, delete these columns
        self.SF_TRIM_COLUMNS = ["Subject", "Created Date", "LIMS Organization ID",
                                "Account Description"]
        self.DP_TRIM_COLUMNS = ["Org ID", "Deposit Status", "PI_ID", "Date Available", "# Orders",
                                "# Plasmids in the Deposit", "Addgene Contact", "Country"]

        self.salesforce_df = None
        self.deposit_df = None
        self.output_dir = None
        self.frames = None

    def _get_dataframe_by_extension(self, path, date_cols):
        """
            Gets a dataframe either by .csv, or .xls(x),
            or erroring and exiting.
        """
        _, ext = os.path.splitext(path)

        if ext == ".csv":
            df = pd.read_csv(path, parse_dates=date_cols, encoding='utf-8')
        elif ext in [".xlsx", ".xls"]:
            df = pd.read_excel(path, parse_dates=date_cols, encoding='utf-8')
        else:
            raise Exception("File was not of type {0}.\nQuitting".format(
                " ".join(self.ACCEPTABLE_EXTENSIONS)))
        return df

    def set_dataframe_sf(self, fname):
        self.salesforce_df = None
        try:
            salesforce_df = self._get_dataframe_by_extension(fname, date_cols=[self.SF_DATE_COLUMN, ])
        except IndexError:
            return False
        except ValueError:
            return False
        except:
            raise

        if set(self.REQUIRED_SF_COLUMNS) < set(salesforce_df.columns):
            self.salesforce_df = salesforce_df
            return True
        L.info("Wrong columns")
        return False

    def set_dataframe_deposit(self, fname):
        self.deposit_df = None
        try:
            deposit_df = self._get_dataframe_by_extension(fname, date_cols=[self.DP_DATE_COLUMN, ])
        except IndexError:
            return False
        except:
            raise
        if set(self.REQUIRED_DP_COLUMNS) < set(deposit_df.columns):
            self.deposit_df = deposit_df
            return True
        L.info("Wrong columns")
        return False

    def set_output_dir(self, dir):
        self.output_dir = dir

    def get_dataframes(self):
        salesforce_df, deposit_df = self.clean_dataframes()
        return salesforce_df, deposit_df

    def clean_dataframes(self):
        # Get rid of the footer that Salesforce adds.
        if self.footer_length:
            length_with_footer = len(self.salesforce_df.index)
            self.salesforce_df = self.salesforce_df.head(length_with_footer - self.footer_length)

        # Clean up Salesforce
        self.salesforce_df.sort(self.SF_DATE_COLUMN, ascending=1)

        # Cleanup Deposit Data
        self.deposit_df['Org Name'].fillna('', inplace=True)
        self.deposit_df.sort(self.DP_DATE_COLUMN, ascending=1)
        self.deposit_df['PI_Name'].astype(unicode)

        # Cleanup not needed columns
        for col in self.SF_TRIM_COLUMNS:
            del self.salesforce_df[col]
        for col in self.DP_TRIM_COLUMNS:
            del self.deposit_df[col]

    def get_filtered(self, filtered_df, sf_row, pi_name, pi_org, org=False):
        """
            Assume kind is PI by default.
            Filter where either the PI and PI match, or the Org and Org match
            If both match, add it to the the double list
            if only one matches, add it to the single list.
        """
        filter_column = self.PI_COLUMN
        filter_value = pi_name
        single, double = [], []

        if org:
            filter_column = self.ORG_COLUMN
            filter_value = pi_org

        name_match = filtered_df[filtered_df[filter_column] == filter_value]

        if not name_match.empty:
            for _, row in name_match.iterrows():
                data = {
                    "Addgene Assigned": sf_row['Assigned'],
                    "Plasmid ID": row['Plasmid ID'],
                    "Deposit ID": row['Deposit ID'],
                    "Institute": row['Org Name'],
                    "PI Name": row['PI_Name'],
                    "Date Received": row[self.DP_DATE_COLUMN],
                    "Original Date": sf_row[self.SF_DATE_COLUMN],
                    "Original ORG": pi_org,
                    "Original PI": pi_name,
                }
                if (data['Institute'] == data['Original ORG']) and \
                        (data['PI Name'] == data['Original PI']):
                    double.append(data)
                else:
                    single.append(data)
        return single, double

    def get_attribution_dataframes(self):
        self.clean_dataframes()

        name_matches = []
        org_matches = []
        double_matches = []
        mismatches = []

        # Iterate through the Salesforce report as the master document
        for index, sf_row in self.salesforce_df.iterrows():
            # Get a start date and an end date for filtering.
            start_date = sf_row[self.SF_DATE_COLUMN]
            end_date = start_date + relativedelta(months=self.months)

            start = self.deposit_df[self.DP_DATE_COLUMN].searchsorted(start_date)[0]
            end = self.deposit_df[self.DP_DATE_COLUMN].searchsorted(end_date)[0]

            # Filter the deposit data to grab only things within that timeframe.
            filtered_df = self.deposit_df.ix[start:end]

            # Variables for short names, and not having to type index a lot.
            pi_name = unicode(sf_row['First Name']) + " " + unicode(sf_row['Last Name'])
            pi_org = sf_row['Account Name']

            # Get matches by the PI's name
            by_name, pi_by_both = self.get_filtered(filtered_df, sf_row, pi_name, pi_org)
            name_matches.extend(by_name)
            mismatches.extend(by_name)
            double_matches.extend(pi_by_both)

            # Get matches by the organization name
            by_org, org_by_both = self.get_filtered(filtered_df, sf_row, pi_name, pi_org, org=True)
            org_matches.extend(by_org)
            mismatches.extend(by_org)
            double_matches.extend(org_by_both)

        return (
            ("PI", pd.DataFrame(name_matches, columns=self.OUTPUT_COLUMN_ORDER)),
            ("Institute", pd.DataFrame(org_matches, columns=self.OUTPUT_COLUMN_ORDER)),
            ("Double", pd.DataFrame(double_matches, columns=self.OUTPUT_COLUMN_ORDER)),
            ("Single", pd.DataFrame(mismatches, columns=self.OUTPUT_COLUMN_ORDER))
        )

    def run(self):
        self.frames = None
        self.frames = self.get_attribution_dataframes()

    def save(self):
        for key, df in self.frames:
            fname = '{0}_Attribution_Report_{1}_Match.xlsx'.format(datetime.date.today(), key)

            output_path = os.path.join(self.output_dir, fname)

            deduped_df = df.drop_duplicates()

            with pd.ExcelWriter(output_path, engine='xlsxwriter') as writer:
                deduped_df.to_excel(writer, sheet_name='Sheet1', index=False)

        # Open the window where the files are
        if os.name == "nt":
            subprocess.call(["explorer", self.output_dir], shell=True)
        else:
            subprocess.call(["open", self.output_dir])