bug: handle invalid URLs in site CSV uploads

app/models/site_upload.rb

@@ -16,6 +16,7 @@ class SiteUpload
   MAX_FILE_SIZE = 5.megabytes
   REQUIRED_HEADERS = ["url"].freeze
   SUPPORTED_SEPARATORS = [",", ";"].freeze
+  FIRST_DATA_ROW_NUMBER = 2 # Row 1 contains CSV headers
   BOM = /^\xEF\xBB\xBF/
 
   attr_accessor :file, :team, :tag_ids, :tags, :new_sites, :existing_sites
@@ -67,13 +68,30 @@ def count
   def parse_sites
     require "csv"
 
-    CSV.foreach(file.path, headers: true, encoding: "bom|utf-8", col_sep:) do |row|
+    CSV.foreach(file.path, headers: true, encoding: "bom|utf-8", col_sep:).with_index(FIRST_DATA_ROW_NUMBER) do |row, line_number|
       row = row.to_h.transform_keys { |header| header.to_s.downcase } # Case-insensitive headers
 
       raw_url = row["url"].to_s.strip
       next if raw_url.empty?
 
-      url = Link.normalize(raw_url)
+      begin
+        parsed_url = Link.parse(raw_url)
+        raise Link::InvalidUriError.new(raw_url) if parsed_url.relative?
+
+        url = Link.normalize(parsed_url)
+      rescue Link::InvalidUriError => error
+        Rails.logger.warn(
+          "site_upload_invalid_url " \
+          "team_id=#{team&.id} " \
+          "filename=#{file&.original_filename} " \
+          "line_number=#{line_number} " \
+          "raw_url=#{raw_url} " \
+          "error_class=#{error.class.name} " \
+          "error_message=#{error.message}"
+        )
+        errors.add(:file, :invalid_row_url, line_number:, url: raw_url)
+        next
+      end
       name = row["nom"] || row["name"]
       tag_names = row["tags"].present? ? row["tags"].split(",").map(&:strip).compact_blank.uniq : []
 
@@ -92,10 +110,10 @@ def parse_sites
   rescue CSV::MalformedCSVError => error
     Rails.logger.warn(
       "site_upload_malformed_csv " \
-      "team_id=#{team&.id.inspect} " \
-      "filename=#{file&.original_filename.inspect} " \
-      "error_class=#{error.class.name.inspect} " \
-      "error_message=#{error.message.inspect}"
+      "team_id=#{team&.id} " \
+      "filename=#{file&.original_filename} " \
+      "error_class=#{error.class.name} " \
+      "error_message=#{error.message}"
     )
     errors.add(:file, :malformed_csv)
   end

config/locales/fr.yml

@@ -10,6 +10,7 @@ fr:
             file:
               invalid_format: Le fichier n'est pas reconnu comme un fichier CSV.
               invalid_headers: Le fichier n'a pas les bons en-têtes de colonne.
+              invalid_row_url: "Le fichier contient une URL invalide à la ligne %{line_number} : %{url}"
               invalid_size: Le fichier est trop lourd. Merci de le découper en plusieurs fichiers.
               malformed_csv: Le fichier CSV est mal formé. Veuillez ré-exporter votre fichier depuis votre tableur et vérifier sa structure.
   activerecord:

spec/models/site_upload_spec.rb

@@ -163,6 +163,36 @@
       expect(site_upload.new_sites.keys).to contain_exactly("https://example.com/", "https://test.com/")
     end
 
+    it "logs an invalid URL row" do
+      csv.write("url,name\nhttps://example.com/,Example Site\nhttp://,Broken")
+      csv.rewind
+      allow(Rails.logger).to receive(:warn)
+
+      site_upload.parse_sites
+
+      expect(site_upload.errors.added?(:file, :invalid_row_url, line_number: 3, url: "http://")).to be(true)
+      expect(Rails.logger).to have_received(:warn).with(
+        include("site_upload_invalid_url", "line_number=3", "raw_url=http://", "filename=sites.csv")
+      )
+    end
+
+    it "continues parsing and collects multiple invalid URL rows" do
+      csv.write("url,name\nhttps://example.com/,Example Site\nhttp://,Broken\n/contact,Relative\nhttps://test.com/,Test Site")
+      csv.rewind
+      allow(Rails.logger).to receive(:warn)
+
+      site_upload.parse_sites
+
+      expect(site_upload.new_sites.keys).to contain_exactly("https://example.com/", "https://test.com/")
+      expect(site_upload.errors.details[:file]).to include(
+        error: :invalid_row_url, line_number: 3, url: "http://"
+      )
+      expect(site_upload.errors.details[:file]).to include(
+        error: :invalid_row_url, line_number: 4, url: "/contact"
+      )
+      expect(Rails.logger).to have_received(:warn).twice
+    end
+
     it "ignores duplicate existing sites" do
       existing_site = build(:site, url: "https://example.com/")
       allow(team.sites).to receive(:find_by_url) { |args| existing_site if args[:url] == "https://example.com/" }
@@ -354,7 +384,7 @@
         expect(site_upload.save).to be false
         expect(site_upload.errors.added?(:file, :malformed_csv)).to be(true)
         expect(Rails.logger).to have_received(:warn).with(
-          include("site_upload_malformed_csv", "filename=\"sites.csv\"")
+          include("site_upload_malformed_csv", "filename=sites.csv")
         )
       end
     end