feat: Improve Amazon product sanitizer (#180)

svenjacobs · May 22, 2023 · f9fe760 · f9fe760
1 parent 701d285
commit f9fe760
Show file tree

Hide file tree

Showing 2 changed files with 53 additions and 13 deletions.
diff --git a/...ain/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/amazon/AmazonProductSanitizer.kt b/...ain/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/amazon/AmazonProductSanitizer.kt
@@ -19,22 +19,33 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.amazon
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.regex.RegexFactory
 import com.svenjacobs.app.leon.core.domain.R
-import com.svenjacobs.app.leon.core.domain.sanitizer.RegexSanitizer
 import com.svenjacobs.app.leon.core.domain.sanitizer.Sanitizer
 import com.svenjacobs.app.leon.core.domain.sanitizer.SanitizerId
 
-class AmazonProductSanitizer : RegexSanitizer(
-	regex = Regex("ref=[^?&]+|[?&][^=]+=.[^&]*"),
-) {
+class AmazonProductSanitizer : Sanitizer {
 
 	override val id = SanitizerId("amazon")
 
 	override fun getMetadata(context: Context) = Sanitizer.Metadata(
 		name = context.getString(R.string.sanitizer_amazon_product_name),
 	)
 
-	override fun matchesDomain(input: String) =
-		input.matchesDomain("amazon\\..+/dp/[0-9A-Z]+", isRegex = true)
+	override fun matchesDomain(input: String) = REGEX.containsMatchIn(input)
+
+	override fun invoke(input: String): String {
+		val result = REGEX.find(input)
+		// First group contains everything between top level domain and /dp/ argument
+		val group = result?.groups?.get(1) ?: return input
+
+		return RegexFactory.AllParameters.replace(
+			input = input.removeRange(group.range),
+			replacement = "",
+		)
+	}
+
+	private companion object {
+		private val REGEX = Regex("amazon\\..+?(/.*)?/dp/[^/]+")
+	}
 }
diff --git a/...kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/amazon/AmazonProductSanitizerTest.kt b/...kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/amazon/AmazonProductSanitizerTest.kt
@@ -1,6 +1,6 @@
 /*
  * Léon - The URL Cleaner
- * Copyright (C) 2022 Sven Jacobs
+ * Copyright (C) 2023 Sven Jacobs
  *
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -23,21 +23,50 @@ import io.kotest.matchers.shouldBe
 
 class AmazonProductSanitizerTest : WordSpec(
 	{
+		val sanitizer = AmazonProductSanitizer()
 
 		"invoke" should {
 
-			"remove various Amazon parameters" {
-				val sanitizer = AmazonProductSanitizer()
-				val result = sanitizer(
+			"remove everything except /dp/* path argument" {
+				var result = sanitizer(
 					"https://www.amazon.de/Xiaomi-Aktivit%C3%A4tstracker-Trainings-Puls%C3%" +
 						"BCberwachung-Akkulaufzeit/dp/B091G3FLL7/?_encoding=UTF8&pd_rd_w=xDcJP&pf" +
 						"_rd_p=bf172aca-3277-41f6-babb-6ce7fc34cf7f&pf_rd_r=ZC6FZ5G6W9K8DEZTPBYW&" +
 						"pd_rd_r=11b3ec4e-047c-4f37-8302-62dedb8f502b&pd_rd_wg=Ozi90&ref_=pd_gw_c" +
 						"i_mcx_mr_hp_atf_m",
 				)
 
-				result shouldBe "https://www.amazon.de/Xiaomi-Aktivit%C3%A4tstracker-Trainings-Pu" +
-					"ls%C3%BCberwachung-Akkulaufzeit/dp/B091G3FLL7/"
+				result shouldBe "https://www.amazon.de/dp/B091G3FLL7/"
+
+				result = sanitizer(
+					"https://www.amazon.co.uk/Xiaomi-Aktivit%C3%A4tstracker-Trainings-Puls%C3%" +
+						"BCberwachung-Akkulaufzeit/dp/B091G3FLL7/?_encoding=UTF8&pd_rd_w=xDcJP&pf" +
+						"_rd_p=bf172aca-3277-41f6-babb-6ce7fc34cf7f&pf_rd_r=ZC6FZ5G6W9K8DEZTPBYW&" +
+						"pd_rd_r=11b3ec4e-047c-4f37-8302-62dedb8f502b&pd_rd_wg=Ozi90&ref_=pd_gw_c" +
+						"i_mcx_mr_hp_atf_m",
+				)
+
+				result shouldBe "https://www.amazon.co.uk/dp/B091G3FLL7/"
+			}
+
+			"keep already cleaned Amazon URL" {
+				sanitizer("https://www.amazon.com/dp/B091G3FLL7/") shouldBe
+					"https://www.amazon.com/dp/B091G3FLL7/"
+			}
+		}
+
+		"matchesDomain" should {
+
+			"match for Amazon product domain" {
+				sanitizer.matchesDomain(
+					"https://www.amazon.de/Xiaomi-Aktivit%C3%A4tstracke" +
+						"r-Trainings-Puls%C3%BCberwachung-Akkulaufzeit/dp/B091G3FLL7/",
+				) shouldBe true
+
+				sanitizer.matchesDomain(
+					"https://www.amazon.co.uk/Xiaomi-Aktivit%C3%A4tstracke" +
+						"r-Trainings-Puls%C3%BCberwachung-Akkulaufzeit/dp/B091G3FLL7/",
+				) shouldBe true
 			}
 		}
 	},