You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
OnHoldReconciler/src/reports.py

344 lines
12 KiB

from pandas import DataFrame, merge, to_datetime, NaT, concat, ExcelWriter
from openpyxl import Workbook, load_workbook
from abc import ABC
from logging import getLogger
import re
import datetime
from copy import deepcopy
from dataclasses import dataclass
from helpers import CN_REGEX, drop_unnamed
from memory import get_prev_reconciled, hash_cols, col_hash, create_identifier
from pathlib import Path
logger = getLogger(__name__)
@dataclass
class ReconciledReports:
no_match: DataFrame
amt_mismatch: DataFrame
prev_rec: DataFrame
gp_filtered: DataFrame
ob_overdue: DataFrame
def save_reports(self, output_path: Path):
with ExcelWriter(output_path, mode='w') as writer:
self.no_match.drop_duplicates(inplace=True)
self.no_match.to_excel(writer, sheet_name="No Match",
index=False, freeze_panes=(1,3)
)
self.amt_mismatch.drop_duplicates(inplace=True)
self.amt_mismatch.to_excel(writer, sheet_name="Amount Mismatch",
index=False, freeze_panes=(1,3)
)
self.ob_overdue.to_excel(writer, sheet_name="Overdue",
index=False
)
self.prev_rec.to_excel(writer, sheet_name="Previously Reconciled",
index=False, freeze_panes=(1,3)
)
self.gp_filtered.to_excel(writer, sheet_name="Filtered from GP",
index=False, freeze_panes=(1,0)
)
wb: Workbook = load_workbook(output_path)
for sheet in ["No Match", "Amount Mismatch"]:
ws = wb[sheet]
ws.column_dimensions['A'].hidden = True
ws.column_dimensions['B'].hidden = True
for sheet in ["Filtered from GP", "Previously Reconciled"]:
wb[sheet].sheet_state = "hidden"
wb.save(output_path)
wb.close()
class HoldReport(ABC):
source = ""
def __init__(self, dataframe: DataFrame, reports_config: dict) -> None:
self.config = reports_config
drop_unnamed(dataframe)
self.df = dataframe
self.df = self._add_work_columns(self.df)
self._normalize()
def _normalize(self):
# Rename the columns to standardize the column names
self.df.rename( columns= { unique_cols[self.source] : common_col
for common_col, unique_cols in self.config["shared_columns"].items()
}, inplace=True)
# Convert the on-hold amount column to float format and round to two decimal places
self.df["onhold_amount"] = self.df["onhold_amount"].astype(float).round(2)
# Use regex to extract the contract number from the column values and create a new column with the standardized format
self.df["contract_number"] = self.df["contract_number"].apply(
lambda cn: str(cn) if not re.search(CN_REGEX, str(cn))
else re.search(CN_REGEX, str(cn)).group(0)
)
# Create a new column with a unique transaction ID
self.df["ID"] = self.df["contract_number"] +'_'+\
self.df["onhold_amount"].astype(str)
# Create a new column with the data source
self.df["Source"] = self.source
@staticmethod
def _remove_prev_recs(contract_match, no_match) -> \
tuple[DataFrame, DataFrame, DataFrame]:
"""
"""
idents: list[col_hash] = create_identifier(contract_match)["Indentifier"].to_list()
idents.extend(create_identifier(no_match)["Indentifier"].to_list())
logger.debug(f"{idents=}")
# Get previsouly reced
prev_recs: DataFrame|None = get_prev_reconciled(idents)
if prev_recs is None:
logger.info("No previously reconciled!")
return DataFrame(), contract_match, no_match
dfs = []
for df in [contract_match, no_match]:
start_size = df.shape[0]
logger.debug(f"Report DF: \n{df}")
logger.debug(f"prev_rec: \n{prev_recs}")
df = merge(
df,
prev_recs,
how="left",
on= "Indentifier",
suffixes=("_cur", "_prev")
)
df = HoldReport._created_combined_col("HideNextMonth", df, ["prev", "cur"])
df = HoldReport._created_combined_col("Resolution", df, ["prev", "cur"])
df["ID_OB"] = df["ID_OB_cur"]
df["ID_GP"] = df["ID_GP_cur"]
# Drop anything that should be ignored
df = df[df["HideNextMonth"] != True]
logger.info(f"Prev res added:\n{df}")
col_to_drop = []
for c in df.keys().to_list():
if "_prev" in c in c or "_cur" in c:
col_to_drop.append(c)
logger.debug(f"{col_to_drop=}")
df.drop(
columns= col_to_drop,
inplace=True
)
# Restandardize
end_size = df.shape[0]
logger.info(f"Reduced df by {start_size-end_size}")
dfs.append(df)
return prev_recs, dfs[0], dfs[1]
def _remove_full_matches(self, other: 'HoldReport'):
"""
Removes any contracts that match both contract number and hold amount.
These do not need to be reconciled.
This id done 'in place' to both dataframes
"""
filter_id_match: DataFrame = self.df[~(self.df["ID"].isin(other.df["ID"]))]
other.df: DataFrame = other.df[~(other.df["ID"].isin(self.df["ID"]))]
self.df = filter_id_match
self.combined_missing: DataFrame = concat([self.df, other.df], ignore_index=True)
#self.combined_missing.to_excel("ALL MISSING.xlsx")
logger.debug(f"Combined Missing:\n{self.combined_missing}")
logger.info(f"Payments with errors: {self.combined_missing.shape[0]}")
@staticmethod
def _created_combined_col(column: str, target_df: DataFrame, sources: tuple[str, str]) -> DataFrame :
"""
Creates a new column by filling empty columns of this source, with the matching column from another source
"""
this, that = sources
target_df[column] = target_df[f"{column}_{this}"].fillna(
target_df[f"{column}_{that}"]
)
return target_df
def _requires_rec(self, other: 'HoldReport') -> tuple[DataFrame, DataFrame]:
"""
To be run after full matches have been re
"""
# Merge the two filtered DataFrames on the contract number
contract_match = merge(
self.df, other.df,
how="inner",
on=["contract_number"],
suffixes=('_'+self.source, '_'+other.source)
)
contract_match = create_identifier(contract_match)
#contract_match.to_excel("CONTRACT_MATCH.xlsx")
for col in ["vendor_name", "HideNextMonth", "Resolution"]:
self._created_combined_col(col, contract_match, (self.source, other.source))
logger.debug(f"_requires_rec | contract_match:\n{contract_match.columns} ({contract_match.shape})")
no_match: DataFrame = self.combined_missing[~(
self.combined_missing["contract_number"].isin(
contract_match["contract_number"]
))
]
no_match[f"ID_{self.source}"] = no_match.apply(lambda row:
row["ID"] if row["Source"] == self.source else None
, axis=1)
no_match[f"ID_{other.source}"] = no_match.apply(lambda row:
row["ID"] if row["Source"] == other.source else None
, axis=1)
no_match = create_identifier(no_match)
logger.debug(f"_requires_rec | no_match:\n{no_match.columns} ({no_match.shape})")
self.prev_recs, contract_match, no_match = self._remove_prev_recs(contract_match, no_match)
return contract_match, no_match
@staticmethod
def _add_work_columns(df: DataFrame) -> DataFrame:
"""
Add empty columns to the dataframe to faciliate working through the report.
"""
logger.debug("Adding work columns!")
df_cols: list[str] = df.columns.to_list()
WORK_COLS = ["HideNextMonth","Resolution"]
for col in WORK_COLS:
if col not in df_cols:
df[col] = ''
return df
def reconcile(self, other: 'HoldReport') -> ReconciledReports:
"""
"""
assert self.source != other.source, f"Reports to reconcile must be from different sources.\
({self.source} , {other.source})."
self._remove_full_matches(other)
if self.source == "OB":
over_due: DataFrame = self.overdue
filtered_gp: DataFrame = other.filtered
elif self.source == "GP":
over_due: DataFrame = other.overdue
filtered_gp: DataFrame = self.filtered
logger.debug(f"Removed matches:\n{self.df}")
amount_mismatch, no_match = self._requires_rec(other)
logger.debug(f"reconcile | no_match unaltered\n{no_match.columns} ({no_match.shape})")
logger.debug(f"reconcile | am_mm unaltered:\n{amount_mismatch.columns} ({amount_mismatch.shape})")
# Formatting
columns: list[str] = ["ID_GP", "ID_OB"]
columns.extend(self.config["output_columns"])
nm_cols:list[str] = deepcopy(columns)
nm_cols.insert(3,"onhold_amount")
nm_cols.insert(4,"Source")
columns.insert(3,"onhold_amount_GP")
columns.insert(4, "onhold_amount_OB")
# Select and reorder columns
no_match = no_match[
nm_cols
]
amount_mismatch = amount_mismatch[
columns
]
logger.info(f"no_match: {no_match.shape[0]}")
logger.info(f"am_mm: {amount_mismatch.shape[0]}")
reconciled: ReconciledReports = ReconciledReports(
no_match=no_match,
amt_mismatch=amount_mismatch,
prev_rec=self.prev_recs,
gp_filtered=filtered_gp,
ob_overdue = over_due
)
return reconciled
class OnBaseReport(HoldReport):
source = "OB"
def __init__(self, dataframe: DataFrame, reports_config: dict) -> None:
self.overdue = self._get_overdue(dataframe)
super().__init__(dataframe, reports_config)
@staticmethod
def _get_overdue(dataframe: DataFrame) -> DataFrame:
"""
"""
dataframe["InstallDate"] = to_datetime(dataframe["InstallDate"])
dataframe["InstallDate"].fillna(NaT, inplace=True)
overdue: DataFrame = dataframe[dataframe["InstallDate"].dt.date\
< datetime.date.today()]
return overdue
class GreatPlainsReport(HoldReport):
source = "GP"
def __init__(self, dataframe: DataFrame, report_config: dict) -> None:
self.filtered: DataFrame = self._filter(
gp_report_df= dataframe,
doc_num_filters= report_config["gp_filters"]["doc_num_filters"],
good_po_num_regex= report_config["gp_filters"]["po_filter"]
)
super().__init__(dataframe, report_config)
@staticmethod
def _filter(gp_report_df: DataFrame,
doc_num_filters: list[str], good_po_num_regex: str
) -> DataFrame:
GOOD_PO_NUM = re.compile(good_po_num_regex, re.IGNORECASE)
bad_doc_num = ''
rx : str
for rx in doc_num_filters:
bad_doc_num += f"({rx})|"
bad_doc_num = re.compile(bad_doc_num[:-1], re.IGNORECASE)
# Create a mask/filter that will keep rows that match these
# requirments
keep_mask = (
(gp_report_df["Document Type"] == "Invoice") &
(gp_report_df["Purchase Order Number"].str.contains(GOOD_PO_NUM))
)
# Get the rows that DO NOT fit the keep_mask
dropped_posotives: DataFrame = gp_report_df[~keep_mask]
# Drop the rows to filter
gp_report_df.drop(dropped_posotives.index, inplace=True)
# Create a filter to remove rows that meet this requirment
# Making this a negative in the keep mask is more trouble than
# it's worth
remove_mask = gp_report_df["Document Number"].str.contains(bad_doc_num)
dropped_negatives: DataFrame = gp_report_df[remove_mask]
gp_report_df.drop(dropped_negatives.index, inplace=True)
return concat([dropped_posotives,dropped_negatives], ignore_index=False)