feat: add BookOasis document reader and text settings

2026-09-19 10:58:14 +09:00
parent 5848ffb4c4
commit 4900efbca5
20 changed files with 746 additions and 7 deletions
@@ -34,6 +34,27 @@ class EpubReader(private val reader: ArchiveReader) : Closeable by reader {
        return getImagesFromPages(pages, ref)
    }

+    /**
+     * Returns readable text for each spine document in reading order.
+     */
+    fun getTextFromPages(): List<String> {
+        val ref = getPackageHref()
+        val doc = getPackageDocument(ref)
+        val basePath = getParentDirectory(ref)
+        return getPagesFromDocument(doc).mapNotNull { page ->
+            val entryPath = resolveZipPath(basePath, page)
+            getInputStream(entryPath)?.use { input ->
+                val pageDoc = Jsoup.parse(input, null, "")
+                pageDoc.select("br").append("\n")
+                pageDoc.select("p,div,li,h1,h2,h3,h4,h5,h6,blockquote").append("\n")
+                pageDoc.body()?.wholeText()
+                    ?.replace(Regex("[ \\t]+\\n"), "\n")
+                    ?.replace(Regex("\\n{3,}"), "\n\n")
+                    ?.trim()
+            }
+        }
+    }
+
    /**
     * Returns the path to the package document.
     */