feat: HTMLを整形し、不正なHTMLを含まないように

build.gradle.kts

@@ -215,6 +215,7 @@ dependencies {
     implementation("org.flywaydb:flyway-core")
 
     implementation("dev.usbharu:emoji-kt:2.0.0")
+    implementation("org.jsoup:jsoup:1.17.2")
 
     implementation("io.ktor:ktor-client-logging-jvm:$ktor_version")
 

src/main/kotlin/dev/usbharu/hideout/core/service/post/DefaultPostContentFormatter.kt

@@ -0,0 +1,107 @@
+package dev.usbharu.hideout.core.service.post
+
+import org.jsoup.Jsoup
+import org.jsoup.nodes.Attributes
+import org.jsoup.nodes.Element
+import org.jsoup.nodes.TextNode
+import org.jsoup.parser.Tag
+import org.jsoup.select.Elements
+
+class DefaultPostContentFormatter() : PostContentFormatter {
+    override suspend fun format(content: String): FormattedPostContent {
+        val document =
+            Jsoup.parseBodyFragment(content).getElementsByTag("body").first() ?: return FormattedPostContent("", "")
+
+        val flattenHtml = document.childNodes().mapNotNull {
+            if (it is Element) {
+                if (it.tagName() == "p") {
+                    p(it)
+                } else {
+                    p(Element("p").appendChildren(document.childNodes()))
+                }
+            } else if (it is TextNode) {
+                Element("p").appendText(it.text())
+            } else {
+                null
+            }
+        }.filter { it.text().isNotBlank() }
+
+        val formattedHtml = mutableListOf<Element>()
+
+        for (element in flattenHtml) {
+            var brCount = 0
+            var prevIndex = 0
+            val childNodes = element.childNodes()
+            for ((index, childNode) in childNodes.withIndex()) {
+                if (childNode is Element && childNode.tagName() == "br") {
+                    brCount++
+                } else if (brCount >= 2) {
+                    formattedHtml.add(Element("p").appendChildren(childNodes.subList(prevIndex, index - brCount)))
+                    prevIndex = index
+                }
+            }
+            formattedHtml.add(Element("p").appendChildren(childNodes.subList(prevIndex, childNodes.size)))
+        }
+
+
+        val elements = Elements(formattedHtml)
+
+        return FormattedPostContent(elements.outerHtml().replace("\n", ""), printHtml(elements))
+    }
+
+    private fun p(element: Element): Element {
+        val childNodes = element.childNodes()
+
+        if (childNodes.size == 1 && childNodes.first() is TextNode) {
+            val pTag = Element("p")
+
+            pTag.appendText(element.text())
+            return pTag
+        }
+
+        val map = childNodes.mapNotNull {
+            if (it is Element) {
+                if (it.tagName() == "a") {
+                    a(it)
+                } else if (it.tagName() == "br") {
+                    Element("br")
+                } else {
+                    TextNode(it.text())
+                }
+            } else if (it is TextNode) {
+                it
+            } else {
+                null
+            }
+        }
+
+        val pTag = Element("p")
+
+        pTag.appendChildren(map)
+
+        return pTag
+    }
+
+    private fun a(element: Element): Element {
+        val attributes = Attributes()
+
+        attributes.put("href", element.attribute("href").value)
+        return Element(Tag.valueOf("a"), "", attributes).appendText(element.text())
+    }
+
+    private fun printHtml(element: Elements): String {
+        return element.joinToString("\n\n") {
+            it.childNodes().joinToString("") {
+                if (it is Element && it.tagName() == "br") {
+                    "\n"
+                } else if (it is Element) {
+                    it.text()
+                } else if (it is TextNode) {
+                    it.text()
+                } else {
+                    ""
+                }
+            }
+        }
+    }
+}

src/main/kotlin/dev/usbharu/hideout/core/service/post/FormattedPostContent.kt

@@ -0,0 +1,6 @@
+package dev.usbharu.hideout.core.service.post
+
+data class FormattedPostContent(
+    val html: String,
+    val content: String
+)

src/main/kotlin/dev/usbharu/hideout/core/service/post/PostContentFormatter.kt

@@ -0,0 +1,5 @@
+package dev.usbharu.hideout.core.service.post
+
+interface PostContentFormatter {
+    suspend fun format(content: String): FormattedPostContent
+}

src/test/kotlin/dev/usbharu/hideout/core/service/post/DefaultPostContentFormatterTest.kt

@@ -0,0 +1,117 @@
+package dev.usbharu.hideout.core.service.post
+
+import kotlinx.coroutines.test.runTest
+import org.assertj.core.api.Assertions.assertThat
+import org.junit.jupiter.api.Test
+
+class DefaultPostContentFormatterTest {
+    @Test
+    fun pタグがpタグになる() = runTest {
+        //language=HTML
+        val html = """<p>hoge</p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p>", "hoge"))
+    }
+
+    @Test
+    fun hタグがpタグになる() = runTest {
+        //language=HTML
+        val html = """<h1>hoge</h1>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p>", "hoge"))
+    }
+
+    @Test
+    fun pタグのネストは破棄される() = runTest {
+        //language=HTML
+        val html = """<p>hoge<p>fuga</p>piyo</p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p><p>fuga</p><p>piyo</p>", "hoge\n\nfuga\n\npiyo"))
+    }
+
+    @Test
+    fun spanタグは無視される() = runTest {
+        //language=HTML
+        val html = """<p><span>hoge</span></p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p>", "hoge"))
+    }
+
+    @Test
+    fun `2連続改行は段落に変換される`() = runTest {
+        //language=HTML
+        val html = """<p>hoge<br><br>fuga</p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p><p>fuga</p>", "hoge\n\nfuga"))
+    }
+
+    @Test
+    fun iタグは無視される() = runTest {
+        //language=HTML
+        val html = """<p><i>hoge</i></p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p>", "hoge"))
+    }
+
+    @Test
+    fun aタグはhrefの中身のみ引き継がれる() = runTest {
+        //language=HTML
+        val html = """<p><a href='https://example.com' class='u-url' target='_blank'>hoge</a></p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p><a href=\"https://example.com\">hoge</a></p>", "hoge"))
+    }
+
+    @Test
+    fun aタグの中のspanは無視される() = runTest {
+        //language=HTML
+        val html = """<p><a href='https://example.com'><span>hoge</span></a></p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p><a href=\"https://example.com\">hoge</a></p>", "hoge"))
+    }
+
+    @Test
+    fun brタグのコンテンツは改行になる() = runTest {
+        //language=HTML
+        val html = """<p>hoge<br>fuga</p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge<br> fuga</p>", "hoge\nfuga"))
+    }
+
+    @Test
+    fun いきなりテキストが来たらpタグで囲む() = runTest {
+        //language=HTML
+        val html = """hoge"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p>", "hoge"))
+    }
+
+    @Test
+    fun bodyタグが含まれていた場合消す() = runTest {
+        //language=HTML
+        val html = """</body><p>hoge</p>"""
+
+        val actual = DefaultPostContentFormatter().format(html)
+
+        assertThat(actual).isEqualTo(FormattedPostContent("<p>hoge</p>", "hoge"))
+    }
+}