support for poetry and epigraphs

shared/src/commonMain/kotlin/net/sergeych/toread/fb2/Fb2Book.kt

@@ -15,6 +15,7 @@ data class Fb2Book(
     val documentInfo: Fb2DocumentInfo = Fb2DocumentInfo(),
     val bodyTitle: List<String> = emptyList(),
     val bodyImages: List<Fb2ImageRef> = emptyList(),
+    val bodyEpigraphs: List<Fb2Epigraph> = emptyList(),
     val sections: List<Fb2Section> = emptyList(),
     val binaries: List<Fb2Binary> = emptyList(),
 ) {
@@ -69,10 +70,50 @@ data class Fb2Binary(
 sealed interface Fb2Block {
     data class Paragraph(val content: Fb2Text) : Fb2Block
     data class Subtitle(val content: Fb2Text) : Fb2Block
+    data class Epigraph(val epigraph: Fb2Epigraph) : Fb2Block
+    data class Cite(val cite: Fb2Cite) : Fb2Block
+    data class Poem(val poem: Fb2Poem) : Fb2Block
     data class Image(val image: Fb2ImageRef) : Fb2Block
     data object EmptyLine : Fb2Block
 }
 
+data class Fb2Epigraph(
+    val blocks: List<Fb2EpigraphBlock> = emptyList(),
+    val textAuthors: List<Fb2Text> = emptyList(),
+)
+
+data class Fb2Cite(
+    val blocks: List<Fb2EpigraphBlock> = emptyList(),
+    val textAuthors: List<Fb2Text> = emptyList(),
+)
+
+sealed interface Fb2EpigraphBlock {
+    data class Paragraph(val content: Fb2Text) : Fb2EpigraphBlock
+    data class Subtitle(val content: Fb2Text) : Fb2EpigraphBlock
+    data class Poem(val poem: Fb2Poem) : Fb2EpigraphBlock
+    data class Cite(val cite: Fb2Cite) : Fb2EpigraphBlock
+    data object EmptyLine : Fb2EpigraphBlock
+}
+
+data class Fb2Poem(
+    val title: List<Fb2Text> = emptyList(),
+    val epigraphs: List<Fb2Epigraph> = emptyList(),
+    val blocks: List<Fb2PoemBlock> = emptyList(),
+    val textAuthors: List<Fb2Text> = emptyList(),
+    val date: String? = null,
+)
+
+sealed interface Fb2PoemBlock {
+    data class Subtitle(val content: Fb2Text) : Fb2PoemBlock
+    data class Stanza(val stanza: Fb2Stanza) : Fb2PoemBlock
+}
+
+data class Fb2Stanza(
+    val title: List<Fb2Text> = emptyList(),
+    val subtitle: Fb2Text? = null,
+    val verses: List<Fb2Text> = emptyList(),
+)
+
 data class Fb2Text(
     val spans: List<Fb2TextSpan>,
 ) {

shared/src/commonMain/kotlin/net/sergeych/toread/fb2/Fb2XmlMapper.kt

@@ -32,6 +32,7 @@ internal object Fb2XmlMapper {
             documentInfo = documentInfoFrom(documentInfo),
             bodyTitle = body?.first("title")?.children("p")?.mapNotNull { it.text().ifBlank { null } }.orEmpty(),
             bodyImages = body?.children("image")?.mapNotNull(::imageRefFrom).orEmpty(),
+            bodyEpigraphs = body?.children("epigraph")?.map(::epigraphFrom).orEmpty(),
             sections = body?.children("section")?.map(::sectionFrom).orEmpty(),
             binaries = root.children("binary").mapNotNull(::binaryFrom),
         )
@@ -124,6 +125,9 @@ internal object Fb2XmlMapper {
             when (child.localName) {
                 "p" -> Fb2Block.Paragraph(textFrom(child))
                 "subtitle" -> Fb2Block.Subtitle(textFrom(child))
+                "epigraph" -> Fb2Block.Epigraph(epigraphFrom(child))
+                "cite" -> Fb2Block.Cite(citeFrom(child))
+                "poem" -> Fb2Block.Poem(poemFrom(child))
                 "image" -> imageRefFrom(child)?.let(Fb2Block::Image)
                 "empty-line" -> Fb2Block.EmptyLine
                 else -> null
@@ -140,8 +144,64 @@ internal object Fb2XmlMapper {
         )
     }
 
+    private fun poemFrom(element: XmlElement): Fb2Poem =
+        Fb2Poem(
+            title = titleFrom(element.first("title")),
+            epigraphs = element.children("epigraph").map(::epigraphFrom),
+            blocks = element.nodes.mapNotNull { node ->
+                val child = (node as? XmlNode.ElementNode)?.element ?: return@mapNotNull null
+                when (child.localName) {
+                    "subtitle" -> Fb2PoemBlock.Subtitle(textFrom(child))
+                    "stanza" -> Fb2PoemBlock.Stanza(stanzaFrom(child))
+                    else -> null
+                }
+            },
+            textAuthors = element.children("text-author").map(::textFrom),
+            date = element.first("date")?.text()?.ifBlank { null },
+        )
+
+    private fun epigraphFrom(element: XmlElement): Fb2Epigraph =
+        Fb2Epigraph(
+            blocks = element.nodes.mapNotNull { node ->
+                val child = (node as? XmlNode.ElementNode)?.element ?: return@mapNotNull null
+                when (child.localName) {
+                    "p" -> Fb2EpigraphBlock.Paragraph(textFrom(child))
+                    "poem" -> Fb2EpigraphBlock.Poem(poemFrom(child))
+                    "cite" -> Fb2EpigraphBlock.Cite(citeFrom(child))
+                    "empty-line" -> Fb2EpigraphBlock.EmptyLine
+                    else -> null
+                }
+            },
+            textAuthors = element.children("text-author").map(::textFrom),
+        )
+
+    private fun citeFrom(element: XmlElement): Fb2Cite =
+        Fb2Cite(
+            blocks = element.nodes.mapNotNull { node ->
+                val child = (node as? XmlNode.ElementNode)?.element ?: return@mapNotNull null
+                when (child.localName) {
+                    "p" -> Fb2EpigraphBlock.Paragraph(textFrom(child))
+                    "subtitle" -> Fb2EpigraphBlock.Subtitle(textFrom(child))
+                    "poem" -> Fb2EpigraphBlock.Poem(poemFrom(child))
+                    "empty-line" -> Fb2EpigraphBlock.EmptyLine
+                    else -> null
+                }
+            },
+            textAuthors = element.children("text-author").map(::textFrom),
+        )
+
+    private fun stanzaFrom(element: XmlElement): Fb2Stanza =
+        Fb2Stanza(
+            title = titleFrom(element.first("title")),
+            subtitle = element.first("subtitle")?.let(::textFrom),
+            verses = element.children("v").map(::textFrom),
+        )
+
+    private fun titleFrom(element: XmlElement?): List<Fb2Text> =
+        element?.children("p")?.map(::textFrom).orEmpty()
+
     private fun textFrom(element: XmlElement): Fb2Text =
-        Fb2Text(spansFrom(element.nodes).mergeAdjacent())
+        Fb2Text(spansFrom(element.nodes).mergeAdjacent().trimBoundaryWhitespace())
 
     private fun spansFrom(nodes: List<XmlNode>, styles: Set<Fb2TextStyle> = emptySet()): List<Fb2TextSpan> =
         nodes.flatMap { node ->
@@ -179,6 +239,14 @@ internal object Fb2XmlMapper {
         return merged
     }
 
+    private fun List<Fb2TextSpan>.trimBoundaryWhitespace(): List<Fb2TextSpan> {
+        if (isEmpty()) return emptyList()
+        val trimmed = toMutableList()
+        trimmed[0] = trimmed[0].copy(text = trimmed[0].text.trimStart())
+        trimmed[trimmed.lastIndex] = trimmed.last().copy(text = trimmed.last().text.trimEnd())
+        return trimmed.filter { it.text.isNotEmpty() }
+    }
+
     private fun StringBuilder.appendSection(section: Fb2Section) {
         append("<section>")
         section.title?.takeIf { it.isNotBlank() }?.let {

shared/src/commonTest/kotlin/net/sergeych/toread/fb2/Fb2FormatTest.kt

@@ -81,6 +81,41 @@ class Fb2FormatTest {
         assertEquals("pic.png", (section.blocks[3] as Fb2Block.Image).image.binaryId)
     }
 
+    @Test
+    fun parsesPoemsWithStanzasVersesAndInlineStyles() {
+        val book = Fb2Format.parseXml(poemXml)
+        val poem = (book.sections.single().blocks.single() as Fb2Block.Poem).poem
+
+        assertEquals("Song", poem.title.single().plainText)
+        val stanza = (poem.blocks.single() as Fb2PoemBlock.Stanza).stanza
+        assertEquals("First stanza", stanza.title.single().plainText)
+        assertEquals("Softly", stanza.subtitle?.plainText)
+        assertEquals(listOf("Line one", "Line two."), stanza.verses.map { it.plainText })
+        assertEquals(setOf(Fb2TextStyle.Emphasis), stanza.verses[0].spans.single().styles)
+        assertEquals("Poet", poem.textAuthors.single().plainText)
+        assertEquals("1910", poem.date)
+    }
+
+    @Test
+    fun parsesBodySectionAndPoemEpigraphs() {
+        val book = Fb2Format.parseXml(epigraphXml)
+
+        assertEquals("Body quote", (book.bodyEpigraphs.single().blocks.single() as Fb2EpigraphBlock.Paragraph).content.plainText)
+
+        val sectionBlocks = book.sections.single().blocks
+        val sectionEpigraph = (sectionBlocks[0] as Fb2Block.Epigraph).epigraph
+        assertEquals("Section quote", (sectionEpigraph.blocks[0] as Fb2EpigraphBlock.Paragraph).content.plainText)
+        assertEquals("Author", sectionEpigraph.textAuthors.single().plainText)
+
+        val cite = (sectionEpigraph.blocks[1] as Fb2EpigraphBlock.Cite).cite
+        assertEquals("Cited line", (cite.blocks.single() as Fb2EpigraphBlock.Paragraph).content.plainText)
+        assertEquals("Cited author", cite.textAuthors.single().plainText)
+
+        val poem = (sectionBlocks[1] as Fb2Block.Poem).poem
+        val poemEpigraph = poem.epigraphs.single()
+        assertEquals("Poem quote", (poemEpigraph.blocks.single() as Fb2EpigraphBlock.Paragraph).content.plainText)
+    }
+
     private val sampleXml = """
         <?xml version="1.0" encoding="UTF-8"?>
         <FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:xlink="http://www.w3.org/1999/xlink">
@@ -134,6 +169,78 @@ class Fb2FormatTest {
         </FictionBook>
     """.trimIndent()
 
+    private val poemXml = """
+        <?xml version="1.0" encoding="UTF-8"?>
+        <FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:xlink="http://www.w3.org/1999/xlink">
+          <description>
+            <title-info>
+              <author><nickname>A</nickname></author>
+              <book-title>Poetry</book-title>
+              <lang>en</lang>
+            </title-info>
+            <document-info>
+              <author><nickname>Toread</nickname></author>
+              <date>2026-05-12</date>
+              <id>poetry</id>
+              <version>1.0</version>
+            </document-info>
+          </description>
+          <body>
+            <section>
+              <poem>
+                <title><p>Song</p></title>
+                <stanza>
+                  <title><p>First stanza</p></title>
+                  <subtitle>Softly</subtitle>
+                  <v>
+                    <emphasis>Line one</emphasis>
+                  </v>
+                  <v>Line two.</v>
+                </stanza>
+                <text-author>Poet</text-author>
+                <date>1910</date>
+              </poem>
+            </section>
+          </body>
+        </FictionBook>
+    """.trimIndent()
+
+    private val epigraphXml = """
+        <?xml version="1.0" encoding="UTF-8"?>
+        <FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:xlink="http://www.w3.org/1999/xlink">
+          <description>
+            <title-info>
+              <author><nickname>A</nickname></author>
+              <book-title>Epigraphs</book-title>
+              <lang>en</lang>
+            </title-info>
+            <document-info>
+              <author><nickname>Toread</nickname></author>
+              <date>2026-05-12</date>
+              <id>epigraphs</id>
+              <version>1.0</version>
+            </document-info>
+          </description>
+          <body>
+            <epigraph><p>Body quote</p></epigraph>
+            <section>
+              <epigraph>
+                <p>Section quote</p>
+                <cite>
+                  <p>Cited line</p>
+                  <text-author>Cited author</text-author>
+                </cite>
+                <text-author>Author</text-author>
+              </epigraph>
+              <poem>
+                <epigraph><p>Poem quote</p></epigraph>
+                <stanza><v>Line.</v></stanza>
+              </poem>
+            </section>
+          </body>
+        </FictionBook>
+    """.trimIndent()
+
     private val windows1251Xml = """
         <?xml version="1.0" encoding="windows-1251"?>
         <FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:xlink="http://www.w3.org/1999/xlink">

shared/src/jdbcMain/kotlin/net/sergeych/toread/storage/jdbc/LibraryScanner.kt

@@ -2,7 +2,12 @@ package net.sergeych.toread.storage.jdbc
 
 import net.sergeych.toread.fb2.Fb2Block
 import net.sergeych.toread.fb2.Fb2Book
+import net.sergeych.toread.fb2.Fb2Cite
+import net.sergeych.toread.fb2.Fb2Epigraph
+import net.sergeych.toread.fb2.Fb2EpigraphBlock
 import net.sergeych.toread.fb2.Fb2Format
+import net.sergeych.toread.fb2.Fb2Poem
+import net.sergeych.toread.fb2.Fb2PoemBlock
 import net.sergeych.toread.fb2.Fb2Section
 import net.sergeych.toread.storage.BodyClusterRecord
 import net.sergeych.toread.storage.BookBodyRecord
@@ -246,7 +251,7 @@ private fun String.bookMimeType(): String =
     if (endsWith(".zip", ignoreCase = true)) "application/zip" else "application/x-fictionbook+xml"
 
 private fun Fb2Book.canonicalText(): String =
-    sections.flatMap { it.textBlocks() }
+    (bodyEpigraphs.flatMap { it.textBlocks() } + sections.flatMap { it.textBlocks() })
         .joinToString(separator = "\n") { it.normalizeForBodyHash() }
         .trim()
 
@@ -256,8 +261,11 @@ private fun Fb2Section.textBlocks(): List<String> {
         blocks.forEach { block ->
             when (block) {
                 Fb2Block.EmptyLine -> Unit
+                is Fb2Block.Cite -> addAll(block.cite.textBlocks())
+                is Fb2Block.Epigraph -> addAll(block.epigraph.textBlocks())
                 is Fb2Block.Image -> Unit
                 is Fb2Block.Paragraph -> add(block.content.plainText)
+                is Fb2Block.Poem -> addAll(block.poem.textBlocks())
                 is Fb2Block.Subtitle -> add(block.content.plainText)
             }
         }
@@ -266,6 +274,42 @@ private fun Fb2Section.textBlocks(): List<String> {
     return current + sections.flatMap { it.textBlocks() }
 }
 
+private fun Fb2Poem.textBlocks(): List<String> = buildList {
+    title.forEach { add(it.plainText) }
+    epigraphs.forEach { addAll(it.textBlocks()) }
+    blocks.forEach { block ->
+        when (block) {
+            is Fb2PoemBlock.Stanza -> {
+                block.stanza.title.forEach { add(it.plainText) }
+                block.stanza.subtitle?.let { add(it.plainText) }
+                block.stanza.verses.forEach { add(it.plainText) }
+            }
+            is Fb2PoemBlock.Subtitle -> add(block.content.plainText)
+        }
+    }
+    textAuthors.forEach { add(it.plainText) }
+    date?.let { add(it) }
+}
+
+private fun Fb2Epigraph.textBlocks(): List<String> = buildList {
+    blocks.forEach { addAll(it.textBlocks()) }
+    textAuthors.forEach { add(it.plainText) }
+}
+
+private fun Fb2Cite.textBlocks(): List<String> = buildList {
+    blocks.forEach { addAll(it.textBlocks()) }
+    textAuthors.forEach { add(it.plainText) }
+}
+
+private fun Fb2EpigraphBlock.textBlocks(): List<String> =
+    when (this) {
+        Fb2EpigraphBlock.EmptyLine -> emptyList()
+        is Fb2EpigraphBlock.Cite -> cite.textBlocks()
+        is Fb2EpigraphBlock.Paragraph -> listOf(content.plainText)
+        is Fb2EpigraphBlock.Poem -> poem.textBlocks()
+        is Fb2EpigraphBlock.Subtitle -> listOf(content.plainText)
+    }
+
 private fun String.normalizeForBodyHash(): String =
     lowercase()
         .replace('\u00ad'.toString(), "")