moved files

This commit is contained in:
Jun-te Kim 2025-03-08 06:29:07 +00:00
parent 3257852003
commit baf93e220d
12 changed files with 57 additions and 10 deletions

View file

@ -1 +1,27 @@
#!/bin/bash
cd etl && poetry install;
# Get the Poetry virtual environment path
VENV_PATH=$(poetry env info --path 2>/dev/null)
if [ -z "$VENV_PATH" ]; then
echo "No Poetry environment found. Did you run 'poetry install'?"
exit 1
fi
# Ensure VS Code settings directory exists
SETTINGS_DIR="/home/vscode/.vscode-server/data/Machine"
SETTINGS_FILE="$SETTINGS_DIR/settings.json"
mkdir -p "$SETTINGS_DIR"
# If settings.json doesn't exist, create a default one
if [ ! -f "$SETTINGS_FILE" ]; then
echo "{}" > "$SETTINGS_FILE"
fi
# Update VS Code settings to use the Poetry virtual environment
jq --arg venv "$VENV_PATH/bin/python" '.["python.defaultInterpreterPath"] = $venv' \
"$SETTINGS_FILE" > "$SETTINGS_FILE.tmp" && mv "$SETTINGS_FILE.tmp" "$SETTINGS_FILE"
echo "✅ Updated VS Code to use Poetry environment: $VENV_PATH"

View file

@ -1,10 +1,10 @@
import os
from pdfReader.pdfReaderToText import pdfReaderToText
from scraper.scraper import SharePointScraper, SharePointInstaller
from etl.scraper.scraper import SharePointScraper, SharePointInstaller
from pprint import pprint, pformat
import logging
import tempfile
from utils.logger import Logger
from etl.utils.logger import Logger
from etl.validator.validator import DomnaSharePointValidator
@ -29,9 +29,16 @@ def main():
# logger.info(pformat(list_of_house_ass_names))
# POC of work completed
south_coast_scraper = SharePointScraper(SharePointInstaller.SOUTH_COAST_INSULATION_SERVICE)
surverys_completed = south_coast_scraper.get_number_of_surverys_completed()
logger.info(pformat(surverys_completed))
# south_coast_scraper = SharePointScraper(SharePointInstaller.SOUTH_COAST_INSULATION_SERVICE)
# surverys_completed = south_coast_scraper.get_number_of_surverys_completed()
# logger.info(pformat(surverys_completed))
print("hello world")
print("1")
a = 1
import sys
print(sys.executable)
print("Current working directory:", os.getcwd())
# def create_temp_file(dict_content):
# with tempfile.NamedTemporaryFile(mode='w+', delete=False, suffix='.txt') as temp_file:
@ -46,12 +53,21 @@ if __name__ == "__main__":
main()
# Make a validaation checker for housing assosciation
# Get and read files
# CSV file of the survey number
# Work out productivity metirc (number of address in submission folder, with at least one file included)
# Khalim would like these metrics from the pdf
# address, uprn, assessor's name validation, current sap rating, current annual emissions. DImension
# address, uprn, assessor's name validation, current sap rating, current annual emissions. DImension
# next week mvp
# Gather file data form sharepoint
# Start making a scheme or csv with the data we 'care about'
# {id: uuid, 'surveryor_name': 'khalim', 'address': '123 fake street', 'prorperty_type': 'house', sharepoint_loc}
# data base of historical data. SQL data

View file

@ -1,10 +1,10 @@
from pprint import pformat
from enum import Enum
import os
from utils.logger import Logger
from etl.utils.logger import Logger
import logging
# Awesome Khalim's Sharepoint code that I'll probably will need to learn how it work in the future
from utils.sharepoint.sharepoint import SharePointClient
from etl.utils.sharepoint.sharepoint import SharePointClient
from functools import wraps
import re
from etl.validator.validator import DomnaSharePointValidator
@ -42,6 +42,7 @@ class SharePointScraper():
self.surveyor_work_completed = {}
def ensure_surveyor_names_loaded(func):
"""Decorator to ensure surveyor names are loaded before calling the function."""
@wraps(func)
@ -174,3 +175,7 @@ class SharePointScraper():
self.surveyor_work_completed.update({name: len(address_folders['value'])})
return self.surveyor_work_completed
@ensure_housing_assosiation_is_loaded
def download_file_for_each_address(self):
pass