From 548500ffd09fd694f8f18c8559e6ffca521f2c92 Mon Sep 17 00:00:00 2001
From: Dawid Makar <dawid.makar@tgs.com>
Date: Tue, 7 Jan 2025 13:21:34 +0100
Subject: [PATCH] PENGSOL-483: Format float numbers.

---
 .../script_components/tab_processors/basic.py | 60 ++++++++++++-------
 .../tab_processors/specified.py               |  1 -
 2 files changed, 40 insertions(+), 21 deletions(-)

diff --git a/scripts/excel_dwh_filler/script_components/tab_processors/basic.py b/scripts/excel_dwh_filler/script_components/tab_processors/basic.py
index 40243ae..47ca340 100644
--- a/scripts/excel_dwh_filler/script_components/tab_processors/basic.py
+++ b/scripts/excel_dwh_filler/script_components/tab_processors/basic.py
@@ -4,6 +4,7 @@
 import logging
 from typing import Any
 
+from openpyxl.styles import numbers
 from openpyxl.worksheet import worksheet
 
 from script_components import helpers
@@ -13,6 +14,12 @@ class BaseExcelTabProcessor(abc.ABC):
     DATE_FORMAT = "%Y-%m-%d"
     DATE_TIME_FORMAT = "%Y-%m-%dT%H:%M:%S"
 
+    PYTHON_TYPE_TO_EXCEL_TYPE = {
+        "str": numbers.FORMAT_TEXT,
+        "int": numbers.FORMAT_NUMBER,
+        "float": numbers.FORMAT_NUMBER_COMMA_SEPARATED1,
+    }
+
     def __init__(self, data: list[Any], tab: worksheet.Worksheet):
         self.data = data
         self.tab = tab
@@ -25,25 +32,48 @@ def run_tab_processing(self) -> None:
     def clear_data_from_tab(self) -> None:
         raise NotImplementedError
 
-    def _format_not_convertible_to_string_values(self, value):
+    def _determine_data_type_and_value(self, value) -> dict:
+        data_type = None
+
         match value:
             case bool():
                 value = int(value)
+                data_type = "int"
             case float():
                 if value.is_integer():
                     value = int(value)
+                    data_type = "int"
                 else:
                     value = helpers.as_text(value)
-                    value = self._change_separator_from_dot_to_comma(value)
+                    data_type = "float"
             case datetime.datetime():
                 value = value.strftime(self.DATE_TIME_FORMAT)
+                data_type = "str"
             case datetime.date():
                 value = value.strftime(self.DATE_FORMAT)
-        return value
+                data_type = "str"
+
+        if data_type is None:
+            data_type = "str"
+
+        return {"data_type": data_type, "value": value}
 
-    def _change_separator_from_dot_to_comma(self, float_as_str: str) -> str:
-        float_with_comma_separator = helpers.as_text(float_as_str).replace(".", ",")
-        return float_with_comma_separator
+    def _place_value_at_location(self, cell_location: str, row_data: dict) -> None:
+        """
+        Places the value in the Excel tab at the specified cell location.
+        Example:
+            cell_location = "A1"
+            row_data = {"data_type": "int", "value": "123"}
+        """
+        formatted_row_value = helpers.as_text(row_data["value"])
+        cell = self.tab[cell_location]
+        cell.value = formatted_row_value
+
+        if row_data["data_type"] in ["int", "float"]:
+            number_data_format = "n"
+            cell.data_type = number_data_format
+
+        cell.number_format = self.PYTHON_TYPE_TO_EXCEL_TYPE[row_data["data_type"]]
 
 
 class SingleColumnExcelTabProcessor(BaseExcelTabProcessor):
@@ -64,19 +94,12 @@ def _fill_excel_column_with_values(self) -> None:
         excel_row_index = copy.copy(self.FIRST_ROW_WITH_VALUE_TO_FILL)
 
         for row_datum in self.data_tuple:
-            transformed_row_datum = self._format_not_convertible_to_string_values(
-                row_datum
-            )
-            formatted_row_datum = helpers.as_text(transformed_row_datum)
-
+            transformed_row_data = self._determine_data_type_and_value(row_datum)
             cell_location = f"{self.VALUE_COLUMN}{excel_row_index}"
-            self.__place_value_at_location(cell_location, formatted_row_datum)
+            self._place_value_at_location(cell_location, transformed_row_data)
 
             excel_row_index += 1
 
-    def __place_value_at_location(self, cell_location: str, formatted_row_datum: str):
-        self.tab[cell_location] = helpers.as_text(formatted_row_datum)
-
     def clear_data_from_tab(self) -> None:
         logging.warning(
             "Clearing data for the single-column values tabs not implemented. "
@@ -99,11 +122,8 @@ def _fill_excel_rows_with_values(self) -> None:
                 column_index = next(excel_column_names_iterator)
                 cell_location = f"{column_index}{excel_row_index}"
 
-                transformed_cell_datum = self._format_not_convertible_to_string_values(
-                    cell_datum
-                )
-                formatted_cell_datum = helpers.as_text(transformed_cell_datum)
-                self.tab[cell_location] = helpers.as_text(formatted_cell_datum)
+                transformed_cell_datum = self._determine_data_type_and_value(cell_datum)
+                self._place_value_at_location(cell_location, transformed_cell_datum)
             excel_row_index += 1
 
     def clear_data_from_tab(self) -> None:
diff --git a/scripts/excel_dwh_filler/script_components/tab_processors/specified.py b/scripts/excel_dwh_filler/script_components/tab_processors/specified.py
index a30ea5f..05ffd75 100644
--- a/scripts/excel_dwh_filler/script_components/tab_processors/specified.py
+++ b/scripts/excel_dwh_filler/script_components/tab_processors/specified.py
@@ -34,7 +34,6 @@ def _fill_excel_column_with_values(self) -> None:
                     row_datum = int(row_datum)
                 else:
                     row_datum = helpers.as_text(row_datum)
-                    row_datum = self._change_separator_from_dot_to_comma(row_datum)
             elif isinstance(row_datum, datetime.date):
                 row_datum = row_datum.strftime(self.DATE_FORMAT)
             cell_location = f"{self.VALUE_COLUMN}{excel_row_index}"