fix: Yahoo sanitizer for search query links (#289)

svenjacobs · Oct 4, 2023 · ce0776b · ce0776b
1 parent f4db4f3
commit ce0776b
Show file tree

Hide file tree

Showing 3 changed files with 38 additions and 16 deletions.
diff --git a/...in/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/SearchResultSanitizer.kt b/...in/src/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/SearchResultSanitizer.kt
@@ -29,9 +29,11 @@ abstract class SearchResultSanitizer(
 	private val regex: Regex,
 ) : Sanitizer {
 
-	override fun invoke(input: String): String {
-		val result = regex.find(input) ?: return input
-		val group = result.groups[1] ?: return input
-		return decodeUrl(group.value)
-	}
+	override fun invoke(input: String): String = extractSearchResultValue(regex, input)
+}
+
+internal fun extractSearchResultValue(regex: Regex, input: String): String {
+	val result = regex.find(input) ?: return input
+	val group = result.groups[1] ?: return input
+	return decodeUrl(group.value)
 }
diff --git a/...c/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/yahoo/YahooSearchSanitizer.kt b/...c/main/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/yahoo/YahooSearchSanitizer.kt
@@ -20,14 +20,13 @@ package com.svenjacobs.app.leon.core.domain.sanitizer.yahoo
 
 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.Sanitizer
 import com.svenjacobs.app.leon.core.domain.sanitizer.SanitizerId
-import com.svenjacobs.app.leon.core.domain.sanitizer.SearchResultSanitizer
+import com.svenjacobs.app.leon.core.domain.sanitizer.extractSearchResultValue
 
-class YahooSearchSanitizer : SearchResultSanitizer(
-	Regex("RU=([^/]+)"),
-) {
+class YahooSearchSanitizer : Sanitizer {
 
 	override val id = SanitizerId("yahoo_search")
 
@@ -36,4 +35,14 @@ class YahooSearchSanitizer : SearchResultSanitizer(
 	)
 
 	override fun matchesDomain(input: String) = input.matchesDomain("search.yahoo.com")
+
+	override fun invoke(input: String): String = if (input.contains("/search")) {
+		RegexFactory.exceptParameter("p").replace(input, "")
+	} else {
+		extractSearchResultValue(REGEX_SEARCH_RESULT, input)
+	}
+
+	companion object {
+		private val REGEX_SEARCH_RESULT = Regex("RU=([^/]+)")
+	}
 }
diff --git a/...st/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/yahoo/YahooSearchSanitizerTest.kt b/...st/kotlin/com/svenjacobs/app/leon/core/domain/sanitizer/yahoo/YahooSearchSanitizerTest.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
@@ -23,19 +23,30 @@ import io.kotest.matchers.shouldBe
 
 class YahooSearchSanitizerTest : WordSpec(
 	{
+		val sanitizer = YahooSearchSanitizer()
 
 		"invoke" should {
 
-			"extract URL from Yahoo search link" {
-				val sanitizer = YahooSearchSanitizer()
-
-				val result = sanitizer(
+			"extract URL from Yahoo search result link" {
+				sanitizer(
 					"https://r.search.yahoo.com/_ylt=A0geKLovoVtisIEAUapx.9w4;_ylu=Y29sbwNi" +
 						"ZjEEcG9zAzQEdnRpZAMEc2VjA3Ny/RV=2/RE=1650201007/RO=10/RU=https%3a%2f%2fg" +
 						"ithub.com%2fsvenjacobs%2fleon/RK=2/RS=rHoItccMzwyZAXsJuDMkBaKUMx0-",
-				)
+				) shouldBe "https://github.com/svenjacobs/leon"
+			}
+
+			"clean Yahoo search query link" {
+				sanitizer(
+					"https://search.yahoo.com/search?p=hi&fr=yfp-hrmob&fr2=p%3Afp%2Cm%3Asb&" +
+						".tsrc=yfp-hrmob&ei=UTF-8&fp=1&toggle=1&cop=mss",
+				) shouldBe "https://search.yahoo.com/search?p=hi"
+			}
+		}
+
+		"matchesDomain" should {
 
-				result shouldBe "https://github.com/svenjacobs/leon"
+			"match search.yahoo.com" {
+				sanitizer.matchesDomain("https://search.yahoo.com") shouldBe true
 			}
 		}
 	},