Add support for Guardian puzzles.

See #46
jpd236 · Oct 29, 2024 · 4b4caa4 · 4b4caa4
1 parent 2ecb0ef
commit 4b4caa4
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 1 deletion.
diff --git a/build.gradle.kts b/build.gradle.kts
@@ -32,7 +32,7 @@ kotlin {
                 implementation("org.jetbrains.kotlinx:kotlinx-serialization-json-js:1.6.2")
                 implementation("com.github.ajalt.colormath:colormath-js:3.3.3")
 
-                implementation("com.jeffpdavidson.kotwords:kotwords-js:1.4.0")
+                implementation("com.jeffpdavidson.kotwords:kotwords-js:1.4.2")
 
                 // TODO: Migrate to kotlinx-datetime if parsing/formatting support is added.
                 implementation("com.soywiz.korlibs.klock:klock-js:4.0.10")

diff --git a/src/jsMain/kotlin/com/jeffpdavidson/crosswordscraper/CrosswordScraper.kt b/src/jsMain/kotlin/com/jeffpdavidson/crosswordscraper/CrosswordScraper.kt
@@ -7,6 +7,7 @@ import com.jeffpdavidson.crosswordscraper.sources.BostonGlobeSource
 import com.jeffpdavidson.crosswordscraper.sources.CrosshareSource
 import com.jeffpdavidson.crosswordscraper.sources.CrosswordCompilerSource
 import com.jeffpdavidson.crosswordscraper.sources.CrosswordNexusSource
+import com.jeffpdavidson.crosswordscraper.sources.GuardianSource
 import com.jeffpdavidson.crosswordscraper.sources.NewYorkTimesSource
 import com.jeffpdavidson.crosswordscraper.sources.PuzzleLinkSource
 import com.jeffpdavidson.crosswordscraper.sources.PuzzleSocietySource
@@ -69,6 +70,7 @@ object CrosswordScraper {
         CrosshareSource,
         CrosswordCompilerSource,
         CrosswordNexusSource,
+        GuardianSource,
         NewYorkTimesSource,
         PuzzleSocietySource,
         PzzlSource,

diff --git a/src/jsMain/kotlin/com/jeffpdavidson/crosswordscraper/sources/GuardianSource.kt b/src/jsMain/kotlin/com/jeffpdavidson/crosswordscraper/sources/GuardianSource.kt
@@ -0,0 +1,30 @@
+package com.jeffpdavidson.crosswordscraper.sources
+
+import com.jeffpdavidson.crosswordscraper.Scraping
+import com.jeffpdavidson.crosswordscraper.sources.Source.Companion.hostIsDomainOrSubdomainOf
+import com.jeffpdavidson.kotwords.formats.Guardian
+import org.w3c.dom.url.URL
+
+object GuardianSource : FixedHostSource() {
+
+    override val sourceName: String = "The Guardian"
+    override fun neededHostPermissions(url: URL) = listOf("https://*.theguardian.com/*")
+
+    override fun matchesUrl(url: URL): Boolean = url.hostIsDomainOrSubdomainOf("theguardian.com")
+    override suspend fun scrapePuzzlesWithPermissionGranted(url: URL, tabId: Int, frameId: Int): ScrapeResult {
+        val scrapeFn = js(
+            """function() {
+                var crosswordElems = document.getElementsByClassName("js-crossword");
+                if (crosswordElems.length == 0 || !crosswordElems[0].dataset["crosswordData"]) {
+                  return '';
+                }
+                return crosswordElems[0].dataset["crosswordData"];
+            }"""
+        )
+        val puzzleJson = Scraping.executeFunctionForString(tabId, frameId, scrapeFn)
+        if (puzzleJson.isNotEmpty()) {
+            return ScrapeResult.Success(listOf(Guardian(puzzleJson)))
+        }
+        return ScrapeResult.Success(listOf())
+    }
+}