fix: Facebook sanitizer for mobile domain and story URLs (#246)

svenjacobs · Sep 14, 2023 · af6e296 · af6e296
1 parent b1e0fe4
commit af6e296
Show file tree

Hide file tree

Showing 15 changed files with 62 additions and 47 deletions.
diff --git a/core-common/src/main/kotlin/com/svenjacobs/app/leon/core/common/domain/DomainExtensions.kt b/core-common/src/main/kotlin/com/svenjacobs/app/leon/core/common/domain/DomainExtensions.kt
@@ -18,7 +18,7 @@
 
 package com.svenjacobs.app.leon.core.common.domain
 
-fun String.matchesDomain(domain: String, isRegex: Boolean = false): Boolean {
-	val regexDomain = if (!isRegex) domain.replace(".", "\\.") else domain
-	return Regex("^(?:https?://)?(?:www\\.)?$regexDomain.*").matches(this)
-}
+fun String.matchesDomainRegex(domain: String): Boolean =
+	Regex("^(?:https?://)?(?:www\\.)?$domain.*").matches(this)
+
+fun String.matchesDomain(domain: String): Boolean = matchesDomainRegex(domain.replace(".", "\\."))
diff --git a/core-common/src/main/kotlin/com/svenjacobs/app/leon/core/common/regex/RegexFactory.kt b/core-common/src/main/kotlin/com/svenjacobs/app/leon/core/common/regex/RegexFactory.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
@@ -26,7 +26,7 @@ object RegexFactory {
 	val AllParameters = Regex("\\?.*")
 
 	/**
-	 * Returns a regex string which matches a certain parameter.
+	 * Returns a Regex which matches a certain parameter.
 	 *
 	 * For example `ofParameter("abc")` returns a regex string which matches `?abc=` or `&abc=`.
 	 *
@@ -36,7 +36,7 @@ object RegexFactory {
 	fun ofParameter(parameter: String): Regex = Regex("[?&](?:$parameter)=([^&#]*)")
 
 	/**
-	 * Returns a regex string which matches a certain parameter prefix.
+	 * Returns a Regex which matches a certain parameter prefix.
 	 *
 	 * For example `ofWildcardParameter("abc_")` returns a regex string which matches `?abc_x=`,
 	 * `&abc_y=`, `&abc_zzz=` et cetera.
@@ -45,4 +45,14 @@ object RegexFactory {
 	 */
 	@Suppress("RegExpUnnecessaryNonCapturingGroup")
 	fun ofWildcardParameter(parameter: String): Regex = Regex("[?&](?:$parameter)[^=]*=([^&#]*)")
+
+	/**
+	 * Returns a Regex which removes all parameters except the specified one.
+	 *
+	 * Use regular expression group and OR (`|`) to include several parameters, for example
+	 * `(a|b)`.
+	 *
+	 * @param parameter Parameter to exclude
+	 */
+	fun exceptParameter(parameter: String): Regex = Regex("[?&](?!$parameter=)[^&]+")
 }
diff --git a/...common/src/test/kotlin/com/svenjacobs/app/leon/core/common/domain/DomainExtensionsTest.kt b/...common/src/test/kotlin/com/svenjacobs/app/leon/core/common/domain/DomainExtensionsTest.kt
@@ -44,9 +44,8 @@ class DomainExtensionsTest : WordSpec(
 			}
 
 			"match domain with regular expression values" {
-				"https://aliexpress.com/item/32948511896".matchesDomain(
+				"https://aliexpress.com/item/32948511896".matchesDomainRegex(
 					domain = "aliexpress\\..+/item/",
-					isRegex = true,
 				) shouldBe true
 			}
 

diff --git a/...in/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/aliexpress/AliexpressSanitizer.kt b/...in/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/aliexpress/AliexpressSanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.aliexpress
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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
@@ -36,6 +36,5 @@ class AliexpressSanitizer : RegexSanitizer(
 		name = context.getString(R.string.sanitizer_aliexpress_name),
 	)
 
-	override fun matchesDomain(input: String) =
-		input.matchesDomain("aliexpress\\..+/item/", isRegex = true)
+	override fun matchesDomain(input: String) = input.matchesDomainRegex("aliexpress\\..+/item/")
 }
diff --git a/...n/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/amazon/AmazonSanitizer.kt b/...n/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/amazon/AmazonSanitizer.kt
@@ -19,7 +19,7 @@
 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.domain.matchesDomainRegex
 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
@@ -40,5 +40,5 @@ class AmazonSanitizer : RegexSanitizer(
 		name = context.getString(R.string.sanitizer_amazon_name),
 	)
 
-	override fun matchesDomain(input: String) = input.matchesDomain("amazon\\..+/", isRegex = true)
+	override fun matchesDomain(input: String) = input.matchesDomainRegex("amazon\\..+/")
 }
diff --git a/...omain/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/ebay/EbaySanitizer.kt b/...omain/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/ebay/EbaySanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.ebay
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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
@@ -36,5 +36,5 @@ class EbaySanitizer : RegexSanitizer(
 		name = context.getString(R.string.sanitizer_ebay_name),
 	)
 
-	override fun matchesDomain(input: String) = input.matchesDomain("ebay\\..+/itm/", isRegex = true)
+	override fun matchesDomain(input: String) = input.matchesDomainRegex("ebay\\..+/itm/")
 }
diff --git a/...c/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/facebook/FacebookSanitizer.kt b/...c/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/facebook/FacebookSanitizer.kt
@@ -19,15 +19,15 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.facebook
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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 FacebookSanitizer : RegexSanitizer(
-	regex = RegexFactory.AllParameters,
+	regex = RegexFactory.exceptParameter("(id|story_fbid)"),
 ) {
 
 	override val id = SanitizerId("facebook_com")
@@ -36,5 +36,7 @@ class FacebookSanitizer : RegexSanitizer(
 		name = context.getString(R.string.sanitizer_facebook_name),
 	)
 
-	override fun matchesDomain(input: String) = input.matchesDomain("facebook.com")
+	override fun matchesDomain(input: String) = input.matchesDomainRegex(
+		domain = "(m\\.)?facebook.com",
+	)
 }
diff --git a/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/georiot/GeoRiotSanitizer.kt b/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/georiot/GeoRiotSanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.georiot
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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.Sanitizer
@@ -37,5 +37,5 @@ class GeoRiotSanitizer : SearchResultSanitizer(
 	)
 
 	override fun matchesDomain(input: String) =
-		input.matchesDomain("target.georiot\\.[^/]+/Proxy.ashx", isRegex = true)
+		input.matchesDomainRegex("target.georiot\\.[^/]+/Proxy.ashx")
 }
diff --git a/...main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/google/GoogleSearchSanitizer.kt b/...main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/google/GoogleSearchSanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.google
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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.Sanitizer
@@ -36,6 +36,5 @@ class GoogleSearchSanitizer : SearchResultSanitizer(
 		name = context.getString(R.string.sanitizer_google_search_name),
 	)
 
-	override fun matchesDomain(input: String) =
-		input.matchesDomain("google\\.[^/]+/url", isRegex = true)
+	override fun matchesDomain(input: String) = input.matchesDomainRegex("google\\.[^/]+/url")
 }
diff --git a/.../kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/linksynergy/LinkSynergySanitizer.kt b/.../kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/linksynergy/LinkSynergySanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.linksynergy
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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.Sanitizer
@@ -36,6 +36,5 @@ class LinkSynergySanitizer : SearchResultSanitizer(
 		name = context.getString(R.string.sanitizer_linksynergy_name),
 	)
 
-	override fun matchesDomain(input: String) =
-		input.matchesDomain("linksynergy\\.[^/]+/link", isRegex = true)
+	override fun matchesDomain(input: String) = input.matchesDomainRegex("linksynergy\\.[^/]+/link")
 }
diff --git a/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/netflix/NetflixSanitizer.kt b/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/netflix/NetflixSanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.netflix
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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
@@ -36,8 +36,7 @@ class NetflixSanitizer : RegexSanitizer(
 		name = context.getString(R.string.sanitizer_netflix_name),
 	)
 
-	override fun matchesDomain(input: String) = input.matchesDomain(
+	override fun matchesDomain(input: String) = input.matchesDomainRegex(
 		domain = "(help\\.)?netflix.com",
-		isRegex = true,
 	)
 }
diff --git a/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/spotify/SpotifySanitizer.kt b/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/spotify/SpotifySanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.spotify
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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
@@ -36,6 +36,5 @@ class SpotifySanitizer : RegexSanitizer(
 		name = context.getString(R.string.sanitizer_spotify_name),
 	)
 
-	override fun matchesDomain(input: String) =
-		input.matchesDomain("(open\\.)?spotify\\.com", isRegex = true)
+	override fun matchesDomain(input: String) = input.matchesDomainRegex("(open\\.)?spotify\\.com")
 }
diff --git a/core-domain/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/x/XSanitizer.kt b/core-domain/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/x/XSanitizer.kt
@@ -19,7 +19,7 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.x
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
 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
@@ -36,8 +36,7 @@ class XSanitizer : RegexSanitizer(
 		name = context.getString(R.string.sanitizer_x_name),
 	)
 
-	override fun matchesDomain(input: String) = input.matchesDomain(
+	override fun matchesDomain(input: String) = input.matchesDomainRegex(
 		domain = "(twitter|x)\\.com",
-		isRegex = true,
 	)
 }
diff --git a/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/youtube/YoutubeSanitizer.kt b/...src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/youtube/YoutubeSanitizer.kt
@@ -19,25 +19,22 @@
 package com.svenjacobs.app.leon.core.domain.sanitizer.youtube
 
 import android.content.Context
-import com.svenjacobs.app.leon.core.common.domain.matchesDomain
+import com.svenjacobs.app.leon.core.common.domain.matchesDomainRegex
+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 YoutubeSanitizer : Sanitizer {
+class YoutubeSanitizer : RegexSanitizer(
+	RegexFactory.exceptParameter("(v|search_query)"),
+) {
 
 	override val id = SanitizerId("youtube")
 
 	override fun getMetadata(context: Context) = Sanitizer.Metadata(
 		name = context.getString(R.string.sanitizer_youtube_name),
 	)
 
-	override fun matchesDomain(input: String) =
-		input.matchesDomain("(m\\.)?youtube\\.com", isRegex = true)
-
-	override fun invoke(input: String) = PARAMS_REGEX.replace(input, "")
-
-	private companion object {
-		private val PARAMS_REGEX = Regex("[?&](?!(v|search_query)=)[^&]+")
-	}
+	override fun matchesDomain(input: String) = input.matchesDomainRegex("(m\\.)?youtube\\.com")
 }
diff --git a/...st/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/facebook/FacebookSanitizerTest.kt b/...st/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/facebook/FacebookSanitizerTest.kt
@@ -27,17 +27,30 @@ class FacebookSanitizerTest : WordSpec(
 
 		"invoke" should {
 
-			"clean facebook.com URLs" {
+			"clean facebook.com reel URLs" {
 				sanitizer("https://www.facebook.com/reel/1242384407160280?sfnsn=scwspmo") shouldBe
 					"https://www.facebook.com/reel/1242384407160280"
 			}
+
+			"clean m.facebook.com story URLs" {
+				sanitizer(
+					"https://m.facebook.com/story.php?story_fbid=pfbid0HqS6zLZvNrQt6ACvjv3h" +
+						"Kq6khpVse437nWSq2jBifKRD5sVH2XRLC3zz8aA7TKkWl&id=4&sfnsn=wiwspmo&mibext" +
+						"id=XzsMCV",
+				) shouldBe "https://m.facebook.com/story.php?story_fbid=pfbid0HqS6zLZvNrQt6ACvjv" +
+					"3hKq6khpVse437nWSq2jBifKRD5sVH2XRLC3zz8aA7TKkWl&id=4"
+			}
 		}
 
 		"matchesDomain" should {
 
 			"match facebook.com" {
 				sanitizer.matchesDomain("https://facebook.com") shouldBe true
 			}
+
+			"match m.facebook.com" {
+				sanitizer.matchesDomain("https://m.facebook.com") shouldBe true
+			}
 		}
 	},
 )