From 7242159aaa89d0ef2f035423cd2dab22e1e3c0fa Mon Sep 17 00:00:00 2001
From: Alireza Adli <alireza.adli@mail.concordia.ca>
Date: Fri, 23 Aug 2024 16:41:08 -0400
Subject: [PATCH] Add centroid fields & Convert to a DF

---
 .../remove_nrcan_duplicates.py                | 61 ++++++++++++++++---
 1 file changed, 54 insertions(+), 7 deletions(-)

diff --git a/varennes_single_processes/remove_nrcan_duplicates.py b/varennes_single_processes/remove_nrcan_duplicates.py
index a3bf0bb..64c08b3 100644
--- a/varennes_single_processes/remove_nrcan_duplicates.py
+++ b/varennes_single_processes/remove_nrcan_duplicates.py
@@ -7,24 +7,71 @@ Project Developer: Alireza Adli alireza.adli@concordia.ca
 # You need to clone mtl_gis_oo project and
 # add it as a dependency of this new project
 from scrub_layer_class import *
+import pandas as pd
 
 # Change the paths by the location of your QGIS installation and datalayers
 qgis_path = 'C:/Program Files/QGIS 3.34.1/apps/qgis'
 
 varennes_nrcan_extra_polygons = \
   'C:/Users/a_adli/PycharmProjects/varennes_gis_oo/' \
-  'data/initial_data/endeavor/nrcan_centroid_main/nrcan_centroid_main.shp'
+  'data/initial_data/endeavor/nrcan_without_centroids/auto_building_2.shp'
 
-# first we duplicate the layer to preserve the main data layer.
+# First we duplicate the layer to preserve the main data layer.
 duplicated = \
   'C:/Users/a_adli/PycharmProjects/varennes_gis_oo/' \
-  'data/initial_data/endeavor/nrcan_centroids_1/auto_building_2.shp'
-
-# First, the layer will be duplicated as some records are going to be removed
+  'data/initial_data/endeavor/nrcan_centroids/nrcan_centroids.shp'
 
 varennes_nrcan = ScrubLayer(
   qgis_path, varennes_nrcan_extra_polygons, 'NRCan Varennes')
+varennes_nrcan.duplicate_layer(duplicated)
+
+varennes_nrcan_centroids = ScrubLayer(
+  qgis_path, duplicated, 'NRCan Varennes with Coordinates')
+
+# Then we add coordinates to each polygon so we can remove the
+# very similar polygons based on coordinates.
+varennes_nrcan_centroids.layer.startEditing()
+
+# Add new fields for the centroid coordinates
+varennes_nrcan_centroids.layer.dataProvider().\
+  addAttributes([QgsField("centroid_x", QVariant.Double),
+                 QgsField("centroid_y", QVariant.Double)])
+varennes_nrcan_centroids.layer.updateFields()
+
+
+centroid_x_index = varennes_nrcan_centroids.\
+  layer.fields().indexFromName("centroid_x")
+centroid_y_index = varennes_nrcan_centroids.\
+  layer.fields().indexFromName("centroid_y")
+
+for feature in varennes_nrcan_centroids.layer.getFeatures():
+    centroid = feature.geometry().centroid().asPoint()
+    feature.setAttribute(centroid_x_index, centroid.x())
+    feature.setAttribute(centroid_y_index, centroid.y())
+    varennes_nrcan_centroids.layer.updateFeature(feature)
+
+# Commit the changes for adding centroids fields
+varennes_nrcan_centroids.layer.commitChanges()
+
+
+# Pandas is a better option to compare polygons and remove the duplicates
+# so we make a dataframe. We just transfer the necessary fields
+# to the dataframe.
+
+field_names = ['feature_id', 'centroid_x', 'centroid_y']
+
+# Get the indices of the specified fields
+field_indices = [varennes_nrcan_centroids.layer.fields().indexOf(field)
+                 for field in field_names]
+
+# Extract the attribute values and store them in a list of dictionaries
+data = []
+for feature in varennes_nrcan_centroids.layer.getFeatures():
+    attributes = [feature.attributes()[index] for index in field_indices]
+    data.append(dict(zip(field_names, attributes)))
+
+# Create a DataFrame from the list of dictionaries
+varennes_nrcan_centroids_df = pd.DataFrame(data)
+varennes_nrcan_centroids_df['ID'] = range(len(varennes_nrcan_centroids_df))
 
-varennes_nrcan_duplicate = varennes_nrcan.duplicate_layer(
-  duplicated, 'NRCan duplicated')