more permissive fiction link format

Aivean · Aug 21, 2023 · db7f665 · db7f665
1 parent 5ba5cbc
commit db7f665
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 5 deletions.
diff --git a/src/main/scala/com/aivean/royalroad/Args.scala b/src/main/scala/com/aivean/royalroad/Args.scala
@@ -43,5 +43,5 @@ class Args(args: Seq[String]) extends ScallopConf(args) {
 }
 
 object Args {
-  val urlValidationRegex = "https?://(www\\.)?royalroadl?.com/fiction/\\d+(/([^/]+))?/?"
+  val urlValidationRegex = "(https?://(www\\.)?royalroadl?.com/fiction/\\d+)(/.*)?"
 }
diff --git a/src/main/scala/com/aivean/royalroad/Main.scala b/src/main/scala/com/aivean/royalroad/Main.scala
@@ -19,6 +19,13 @@ object Main extends App {
   def handleFromArg[T](chaps: Seq[T], fromChap: Int): Seq[T] =
     if (fromChap > 0) chaps.drop(fromChap - 1) else if (fromChap < 0) chaps.takeRight(-fromChap) else chaps
 
+  def extractFictionLink(link: String): String = {
+    val regex = Args.urlValidationRegex.r
+    link match {
+      case regex(url, _, _) => url
+      case _ => throw new IllegalArgumentException("Invalid fiction link: " + link)
+    }
+  }
 
   val cliArgs = new Args(args)
 
@@ -33,7 +40,7 @@ object Main extends App {
     }
   }
 
-  val doc = browser.get(cliArgs.fictionLink())
+  val doc = browser.get(extractFictionLink(cliArgs.fictionLink()))
   val title = doc >> text("title")
 
   println("Title: " + title)

diff --git a/src/test/scala/com/aivean/royalroad/ArgsTest.scala b/src/test/scala/com/aivean/royalroad/ArgsTest.scala
@@ -1,5 +1,6 @@
 package com.aivean.royalroad
 
+import com.aivean.royalroad.Main.extractFictionLink
 import org.scalatest.FunSuite
 
 /**
@@ -21,12 +22,40 @@ class ArgsTest extends FunSuite {
       assert(s"${http}://${www}royalroad${l}.com/fiction/12345/name".matches(Args.urlValidationRegex))
       assert(s"${http}://${www}royalroad${l}.com/fiction/12345/na-me".matches(Args.urlValidationRegex))
       assert(s"${http}://${www}royalroad${l}.com/fiction/12345/name/".matches(Args.urlValidationRegex))
+      assert(s"${http}://${www}royalroad${l}.com/fiction/12345/name/chapter/123/chapter-name".matches(Args.urlValidationRegex))
+      assert(s"${http}://${www}royalroad${l}.com/fiction/123/name//".matches(Args.urlValidationRegex))
+      assert(s"${http}://${www}royalroad${l}.com/fiction/12345//".matches(Args.urlValidationRegex))
+      assert(s"${http}://${www}royalroad${l}.com/fiction/123//".matches(Args.urlValidationRegex))
 
-      assert(!s"${http}://${www}royalroad${l}.com/fiction/12345//".matches(Args.urlValidationRegex))
       assert(!s"${http}://${www}royalroad${l}.com/fiction/".matches(Args.urlValidationRegex))
       assert(!s"${http}://${www}royalroad${l}.com/fiction//".matches(Args.urlValidationRegex))
-      assert(!s"${http}://${www}royalroad${l}.com/fiction/123//".matches(Args.urlValidationRegex))
-      assert(!s"${http}://${www}royalroad${l}.com/fiction/123/name//".matches(Args.urlValidationRegex))
+    }
+  }
+
+  test("extract fiction link") {
+    for (
+      http <- List("http", "https");
+      www <- List("", "www.");
+      l <- List("", "l")
+    ) {
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/12345") ==
+        s"${http}://${www}royalroad${l}.com/fiction/12345")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/12345/") ==
+        s"${http}://${www}royalroad${l}.com/fiction/12345")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/12345/name") ==
+        s"${http}://${www}royalroad${l}.com/fiction/12345")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/12345/na-me") ==
+        s"${http}://${www}royalroad${l}.com/fiction/12345")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/12345/name/") ==
+        s"${http}://${www}royalroad${l}.com/fiction/12345")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/12345/name/chapter/123/chapter-name") ==
+        s"${http}://${www}royalroad${l}.com/fiction/12345")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/123/name//") ==
+        s"${http}://${www}royalroad${l}.com/fiction/123")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/12345//") ==
+        s"${http}://${www}royalroad${l}.com/fiction/12345")
+      assert(extractFictionLink(s"${http}://${www}royalroad${l}.com/fiction/123//") ==
+        s"${http}://${www}royalroad${l}.com/fiction/123")
     }
   }
 }