Merge pull request #45 from fleeksoft/develop

enable js resource tests and update with jsoup latest code
fleeksoft · Aug 9, 2024 · 579adab · 579adab
2 parents 0086bd2 + e21c475
commit 579adab
Show file tree

Hide file tree

Showing 67 changed files with 1,991 additions and 1,358 deletions.
diff --git a/.github/workflows/deploy.yml b/.github/workflows/deploy.yml
@@ -5,7 +5,7 @@ on:
     branches: [ "release" ]
     paths-ignore: [ "**.md" ]
   pull_request:
-    branches: [ "release" ]
+    branches: [ "release", "develop" ]
     paths-ignore: [ "**.md" ]
 
 jobs:
@@ -64,6 +64,7 @@ jobs:
 #          ORG_GRADLE_PROJECT_signingInMemoryKeyPassword: ${{ secrets.GPG_KEY_PASSWORD }}
 #
   deploy_docs:
+    if: github.ref == 'refs/heads/release'
     runs-on: macos-latest
     needs:
       - build

diff --git a/ksoup-test/build.gradle.kts b/ksoup-test/build.gradle.kts
@@ -22,7 +22,7 @@ val generateBuildConfigFile: Task by tasks.creating {
     }
 }
 
-tasks.all {
+tasks.configureEach {
     if (name != generateBuildConfigFile.name && !name.contains("publish", ignoreCase = true)) {
         dependsOn(generateBuildConfigFile.name)
     }
@@ -34,4 +34,13 @@ kotlin {
             this.kotlin.srcDir(layout.buildDirectory.file(rootPath))
         }
     }
+    js(IR) {
+        browser {
+            testTask {
+                useMocha {
+                    timeout = "9s"
+                }
+            }
+        }
+    }
 }
diff --git a/ksoup-test/module.yaml b/ksoup-test/module.yaml
@@ -17,6 +17,7 @@ test-dependencies:
     - $libs.codepoints
     - $libs.kotlinx.coroutines.test
     - $libs.kotlinx.datetime
+    - $libs.stately.concurrent
 
 settings:
     kotlin:

diff --git a/ksoup-test/test/com/fleeksoft/ksoup/GzipTest.kt b/ksoup-test/test/com/fleeksoft/ksoup/GzipTest.kt
@@ -10,10 +10,6 @@ import kotlin.test.assertEquals
 class GzipTest {
     @Test
     fun testReadGzipFile() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val gzipFileStr = TestHelper.readGzipResource("htmltests/gzip.html.gz").readAll()
             .toString(charset = Charsets.UTF8)
         val expected = """<title>Gzip test</title>

diff --git a/ksoup-test/test/com/fleeksoft/ksoup/TestHelper.kt b/ksoup-test/test/com/fleeksoft/ksoup/TestHelper.kt
@@ -9,23 +9,24 @@ import korlibs.io.stream.readAll
 object TestHelper {
 
     suspend fun readGzipResource(file: String): SyncStream {
-        return readGzipFile(getResourceAbsolutePath(file))
+        return readGzipFile(getResourceAbsolutePath(file).uniVfs)
     }
 
     fun getResourceAbsolutePath(resourceName: String): String {
-        if (Platform.current == PlatformType.WINDOWS) {
+        if (Platform.isWindows()) {
             return "../../../../testResources/$resourceName"
+        } else if (Platform.isJS()) {
+            return "https://raw.githubusercontent.com/fleeksoft/ksoup/release/ksoup-test/testResources/$resourceName"
         }
         return "${BuildConfig.PROJECT_ROOT}/ksoup-test/testResources/$resourceName"
     }
 
     suspend fun getFileAsString(file: VfsFile): String {
-        val bytes: ByteArray =
-            if (file.fullName.endsWith(".gz")) {
-                readGzipFile(file.absolutePath).readAll()
-            } else {
-                readFile(file.absolutePath).readAll()
-            }
+        val bytes: ByteArray = if (file.fullName.endsWith(".gz")) {
+            readGzipFile(file).readAll()
+        } else {
+            readFile(file).readAll()
+        }
         return bytes.decodeToString()
     }
 
@@ -36,9 +37,9 @@ object TestHelper {
 
     suspend fun pathToStream(file: VfsFile): SyncStream {
         return if (file.fullName.endsWith(".gz")) {
-            readGzipFile(file.absolutePath)
+            readGzipFile(file)
         } else {
-            readFile(file.absolutePath)
+            readFile(file)
         }
     }
 }
diff --git a/ksoup-test/test/com/fleeksoft/ksoup/helper/DataUtilTest.kt b/ksoup-test/test/com/fleeksoft/ksoup/helper/DataUtilTest.kt
@@ -173,10 +173,6 @@ class DataUtilTest {
 
     @Test
     fun supportsBOMinFiles() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         var input = TestHelper.getResourceAbsolutePath("bomtests/bom_utf16be.html")
         var doc: Document =
             Ksoup.parseFile(filePath = input, baseUri = "http://example.com", charsetName = null)
@@ -187,7 +183,7 @@ class DataUtilTest {
         assertTrue(doc.title().contains("UTF-16LE"))
         assertTrue(doc.text().contains("가각갂갃간갅"))
 
-        if (Platform.isJS()) {
+        if (Platform.isJS() || Platform.isWindows()) {
             // FIXME: UTF-32 charset not supported
             return@runTest
         }
@@ -203,32 +199,58 @@ class DataUtilTest {
     }
 
     @Test
-    fun supportsUTF8BOM() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
+    fun streamerSupportsBOMinFiles() = runTest {
+        // test files from http://www.i18nl10n.com/korean/utftest/
+        var file = TestHelper.getResourceAbsolutePath("bomtests/bom_utf16be.html").uniVfs
+        val parser = Parser.htmlParser()
+
+        var doc: Document =
+            DataUtil.streamParser(file = file, baseUri = "http://example.com", charset = null, parser = parser)
+                .complete()
+        assertTrue(doc.title().contains("UTF-16BE"))
+        assertTrue(doc.text().contains("가각갂갃간갅"))
+
+        file = TestHelper.getResourceAbsolutePath("bomtests/bom_utf16le.html").uniVfs
+        doc = DataUtil.streamParser(file = file, baseUri = "http://example.com", charset = null, parser = parser)
+            .complete()
+        assertTrue(doc.title().contains("UTF-16LE"))
+        assertTrue(doc.text().contains("가각갂갃간갅"))
+
+        if (Platform.isJS() || Platform.isWindows()) {
+            // FIXME: UTF-32 charset not supported
             return@runTest
         }
+
+        file = TestHelper.getResourceAbsolutePath("bomtests/bom_utf32be.html").uniVfs
+        doc = DataUtil.streamParser(file = file, baseUri = "http://example.com", charset = null, parser = parser)
+            .complete()
+        assertTrue(doc.title().contains("UTF-32BE"))
+        assertTrue(doc.text().contains("가각갂갃간갅"))
+
+        file = TestHelper.getResourceAbsolutePath("bomtests/bom_utf32le.html").uniVfs
+        doc = DataUtil.streamParser(file = file, baseUri = "http://example.com", charset = null, parser = parser)
+            .complete()
+        assertTrue(doc.title().contains("UTF-32LE"))
+        assertTrue(doc.text().contains("가각갂갃간갅"))
+    }
+
+    @Test
+    fun supportsUTF8BOM() = runTest {
         val input: String = TestHelper.getResourceAbsolutePath("bomtests/bom_utf8.html")
         val doc: Document = Ksoup.parseFile(input, "http://example.com", null)
         assertEquals("OK", doc.head().select("title").text())
     }
 
     @Test
     fun noExtraNULLBytes() {
-        val b =
-            "<html><head><meta charset=\"UTF-8\"></head><body><div><u>ü</u>ü</div></body></html>".toByteArray(
-                Charsets.UTF8,
-            )
+        val b = "<html><head><meta charset=\"UTF-8\"></head><body><div><u>ü</u>ü</div></body></html>"
+            .toByteArray(Charsets.UTF8)
         val doc = Ksoup.parse(b.openSync(), baseUri = "", charsetName = null)
         assertFalse(doc.outerHtml().contains("\u0000"))
     }
 
     @Test
     fun supportsZippedUTF8BOM() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val input: String = TestHelper.getResourceAbsolutePath("bomtests/bom_utf8.html.gz")
         val doc: Document =
             Ksoup.parseFile(
@@ -243,6 +265,22 @@ class DataUtilTest {
         )
     }
 
+    @Test
+    fun streamerSupportsZippedUTF8BOM() = runTest {
+        val file = TestHelper.getResourceAbsolutePath("bomtests/bom_utf8.html.gz").uniVfs
+        val doc = DataUtil.streamParser(
+            file = file,
+            baseUri = "http://example.com",
+            charset = null,
+            parser = Parser.htmlParser()
+        ).complete();
+        assertEquals("OK", doc.head().select("title").text());
+        assertEquals(
+            "There is a UTF8 BOM at the top (before the XML decl). If not read correctly, will look like a non-joining space.",
+            doc.body().text()
+        );
+    }
+
     @Test
     fun supportsXmlCharsetDeclaration() {
         val encoding = "iso-8859-1"
@@ -259,10 +297,6 @@ class DataUtilTest {
 
     @Test
     fun loadsGzipFile() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val input: String = TestHelper.getResourceAbsolutePath("htmltests/gzip.html.gz")
         val doc: Document = Ksoup.parseFile(filePath = input, charsetName = null)
         doc.toString()
@@ -272,10 +306,6 @@ class DataUtilTest {
 
     @Test
     fun loadsZGzipFile() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         // compressed on win, with z suffix
         val input: String = TestHelper.getResourceAbsolutePath("htmltests/gzip.html.z")
         val doc: Document = Ksoup.parseFile(filePath = input, charsetName = null)
@@ -285,10 +315,6 @@ class DataUtilTest {
 
     @Test
     fun handlesFakeGzipFile() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val input: String = TestHelper.getResourceAbsolutePath("htmltests/fake-gzip.html.gz")
         val doc: Document = Ksoup.parseFile(filePath = input, charsetName = null)
         assertEquals("This is not gzipped", doc.title())
@@ -297,10 +323,6 @@ class DataUtilTest {
 
     @Test
     fun handlesChunkedInputStream() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val inputFile: String = TestHelper.getResourceAbsolutePath("htmltests/large.html.gz")
         val input: String = TestHelper.getFileAsString(inputFile.uniVfs)
 //        val stream = VaryingBufferReader(BufferReader(input))
@@ -313,10 +335,6 @@ class DataUtilTest {
 
     @Test
     fun handlesUnlimitedRead() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val inputFile: String = TestHelper.getResourceAbsolutePath("htmltests/large.html.gz")
         val input: String = TestHelper.getFileAsString(inputFile.uniVfs)
         val byteBuffer: ByteArray = DataUtil.readToByteBuffer(input.openSync(), 0)

diff --git a/ksoup-test/test/com/fleeksoft/ksoup/integration/ParseTest.kt b/ksoup-test/test/com/fleeksoft/ksoup/integration/ParseTest.kt
@@ -21,22 +21,17 @@ import kotlin.test.assertTrue
 class ParseTest {
     @Test
     fun testHtml5Charset() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
-        if (Platform.isApple()) {
-//            apple don't support gb2312 or gbk
+        if (Platform.isApple() || Platform.isWindows()) {
+//            don't support gb2312 or gbk
             return@runTest
         }
         // test that <meta charset="gb2312"> works
         var input = TestHelper.getResourceAbsolutePath("htmltests/meta-charset-1.html")
-        var doc: Document =
-            parseFile(
-                filePath = input,
-                baseUri = "http://example.com/",
-                charsetName = null,
-            ) // gb2312, has html5 <meta charset>
+        var doc: Document = parseFile(
+            filePath = input,
+            baseUri = "http://example.com/",
+            charsetName = null,
+        ) // gb2312, has html5 <meta charset>
         if (Platform.isJS()) {
             // FIXME: on js it is returning GBK
             assertEquals("GBK", doc.outputSettings().charset().name.uppercase())
@@ -85,10 +80,6 @@ class ParseTest {
 
     @Test
     fun testLowercaseUtf8Charset() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val input = TestHelper.getResourceAbsolutePath("htmltests/lowercase-charset-test.html")
         val doc: Document = parseFile(filePath = input, charsetName = null)
         val form = doc.select("#form").first()
@@ -98,11 +89,6 @@ class ParseTest {
 
     @Test
     fun testXwiki() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
-        // https://github.com/jhy/jsoup/issues/1324
         // this tests that when in CharacterReader we hit a buffer while marked, we preserve the mark when buffered up and can rewind
         val input = TestHelper.getResourceAbsolutePath("htmltests/xwiki-1324.html.gz")
         val doc: Document =
@@ -122,21 +108,15 @@ class ParseTest {
 
     @Test
     fun testXwikiExpanded() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
-        // https://github.com/jhy/jsoup/issues/1324
         // this tests that if there is a huge illegal character reference, we can get through a buffer and rewind, and still catch that it's an invalid refence,
         // and the parse tree is correct.
         val parser = Parser.htmlParser()
-        val doc =
-            parse(
-                syncStream = TestHelper.resourceFilePathToStream("htmltests/xwiki-edit.html.gz"),
-                baseUri = "https://localhost/",
-                charsetName = "UTF-8",
-                parser = parser.setTrackErrors(100),
-            )
+        val doc = parse(
+            syncStream = TestHelper.resourceFilePathToStream("htmltests/xwiki-edit.html.gz"),
+            baseUri = "https://localhost/",
+            charsetName = "UTF-8",
+            parser = parser.setTrackErrors(100),
+        )
         val errors = parser.getErrors()
         assertEquals("XWiki Jetty HSQLDB 12.1-SNAPSHOT", doc.select("#xwikiplatformversion").text())
         assertEquals(0, errors.size) // not an invalid reference because did not look legit
@@ -150,10 +130,6 @@ class ParseTest {
 
     @Test
     fun testWikiExpandedFromString() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val input = TestHelper.getResourceAbsolutePath("htmltests/xwiki-edit.html.gz")
         val html = TestHelper.getFileAsString(input.uniVfs)
         val doc = parse(html)
@@ -165,10 +141,6 @@ class ParseTest {
 
     @Test
     fun testWikiFromString() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val input = TestHelper.getResourceAbsolutePath("htmltests/xwiki-1324.html.gz")
         val html = TestHelper.getFileAsString(input.uniVfs)
         val doc = parse(html)
@@ -180,10 +152,6 @@ class ParseTest {
 
     @Test
     fun testFileParseNoCharsetMethod() = runTest {
-        if (Platform.isJS()) {
-//            js resource access issue
-            return@runTest
-        }
         val file = TestHelper.getResourceAbsolutePath("htmltests/xwiki-1324.html.gz")
         val doc: Document = parseFile(file)
         assertEquals("XWiki Jetty HSQLDB 12.1-SNAPSHOT", doc.select("#xwikiplatformversion").text())