About

App classifying parcels by date and departments from the following inputs:

• https://files.data.gouv.fr/cadastre/etalab-cadastre/
• https://cadastre.data.gouv.fr/data/etalab-cadastre/

Generates the following output as parcels-index.csv:

...
97611000BN0032,2024-01-01,976
97617000AH0219,2024-01-01,976
97617000AH0218,2023-10-01,976
97617000AH0217,2024-01-01,976
97617000AH0216,2024-01-01,976

The output is useful for finding the date and department for a given parcel ID to identify which file must be looked into for the parcel GeoJSON.

The app can be run a second time to look search for given parcels.

Pre-requisites

Install OpenJDK 21 and maven.

Build the app

mvn clean install

Usage

Run the app to generate parcels-index.csv

java -Xmx7G -jar target/parcels-index-0.1.jar

At the end, parcels-index.csv will be created.

Find parcel IDs in parcels-index.csv

Prepare a file with parce IDs to look for - for example file missing_parcel_ids_2023.txt:

01033458ZB0427
01053000AN0130
01071000AD0200
...

If you have used dvf, you can do this to find parcels with missing latitude and longitude:

grep -E ',,$' dist/2023/full.csv | awk -F, '{ print $16 }' > missing_parcel_ids_2023.txt

Split the parcels-index.csv in smaller files:

split -l 20000000 parcels-index.csv parcels-index-part-

This will generate several files:

parcels-index-part-aa
parcels-index-part-ab
parcels-index-part-ac
parcels-index-part-ad
parcels-index-part-ae

Find parcels from the missing_parcel_ids_2023.txt with a small script:

#!/usr/bin/env bash

echo -n "" > parcels-matches.csv

for part in parcels-index-part-*;do
  java -Xmx7G -jar target/parcels-index-0.1.jar missing_parcel_ids_2023.txt $part
done

At the end, parcels-matches.csv will be created.

Update an existing parcels-index.csv

If you already have a parcels-index.csv, you can update it with 2024-04-01 new data like this:

java -Xmx13G -jar target/parcels-index-0.1.jar update 2024-04-01

You need at least 16GB of memory if having all parcels from 2017. 8GB of memory + 12GB of swap will also work.