From 7f5397c6df3c9dad670a097cf6f58a15f24072ea Mon Sep 17 00:00:00 2001
From: Antoine Goutenoir <antoine.goutenoir@irap.omp.eu>
Date: Wed, 24 Nov 2021 11:58:51 +0100
Subject: [PATCH] fix: don't crash when encoding is not supported

---
 flaskr/controllers/main_controller.py | 22 +++++++++++++++++++++-
 requirements.txt                      |  1 +
 2 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/flaskr/controllers/main_controller.py b/flaskr/controllers/main_controller.py
index 9362cd6..d7bb72d 100644
--- a/flaskr/controllers/main_controller.py
+++ b/flaskr/controllers/main_controller.py
@@ -6,6 +6,7 @@ from io import StringIO
 from os import unlink, getenv
 from os.path import join
 
+import chardet
 import geopy
 import pandas
 import sqlalchemy
@@ -83,7 +84,14 @@ def gather_addresses(from_list, from_file):
     addresses = []
     if from_file:
         file_mimetype = from_file.mimetype
-        file_contents = from_file.read().decode()
+        file_contents_raw = from_file.read()
+        detected = chardet.detect(file_contents_raw)
+        if detected['encoding']:
+            file_contents = file_contents_raw.decode(
+                encoding=detected['encoding']
+            )
+        else:
+            file_contents = file_contents_raw.decode()
 
         rows_dicts = None
 
@@ -198,6 +206,12 @@ def estimate():  # register new estimation request, more accurately
         except validators.ValidationError as e:
             form.origin_addresses_file.errors.append(str(e))
             return show_form()
+        except UnicodeDecodeError as e:
+            form.origin_addresses_file.errors.append(
+                "We only accept UTF-8 and UTF-16 encoded files, \n" +
+                "or files we can detect encoding from."
+            )
+            return show_form()
 
         try:
             estimation.destination_addresses = gather_addresses(
@@ -207,6 +221,12 @@ def estimate():  # register new estimation request, more accurately
         except validators.ValidationError as e:
             form.destination_addresses_file.errors.append(str(e))
             return show_form()
+        except UnicodeDecodeError as e:
+            form.origin_addresses_file.errors.append(
+                "We only accept UTF-8 and UTF-16 encoded files, \n" +
+                "or files we can detect encoding from."
+            )
+            return show_form()
 
         estimation.use_train_below_km = form.use_train_below_km.data
 
diff --git a/requirements.txt b/requirements.txt
index fd0a7ae..11cf054 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -26,6 +26,7 @@ enum34==1.1.6
 geopy==1.23.0
 certifi==2019.11.28
 python-dotenv==0.10.3
+chardet==4.0.0
 
 # Force stable werkzeug
 # see https://gitlab.irap.omp.eu/carbon/travel-carbon-footprint.irap.omp.eu/issues/38
--
libgit2 0.21.2