Cleanup findings from #100 review (reuse, simplification, efficiency)

src/main/kotlin/io/heapy/kotbusta/ktor/routes/books/DownloadBookRoute.kt

@@ -9,7 +9,7 @@ import io.heapy.kotbusta.ktor.routes.requireApprovedUser
 import io.heapy.kotbusta.ktor.routes.requiredParameter
 import io.heapy.kotbusta.service.BookFileException
 import io.heapy.kotbusta.service.ConversionFormat
-import io.heapy.kotbusta.util.asciiFallbackFileName
+import io.heapy.kotbusta.util.attachmentContentDisposition
 import io.ktor.http.*
 import io.ktor.server.response.*
 import io.ktor.server.routing.*
@@ -62,19 +62,5 @@ fun Route.downloadBookRoute() {
     }
 }
 
-internal fun downloadContentDisposition(fileName: String): ContentDisposition {
-    val fallback = asciiFallbackFileName(fileName)
-    val disposition = ContentDisposition.Attachment.withParameter(
-        ContentDisposition.Parameters.FileName,
-        fallback,
-    )
-
-    return if (fallback == fileName) {
-        disposition
-    } else {
-        disposition.withParameter(
-            ContentDisposition.Parameters.FileNameAsterisk,
-            fileName,
-        )
-    }
-}
+internal fun downloadContentDisposition(fileName: String): ContentDisposition =
+    attachmentContentDisposition(fileName, includeAsciiFallback = true)

src/main/kotlin/io/heapy/kotbusta/model/ImportJob.kt

@@ -19,7 +19,6 @@ data class ImportJob(
     val booksAdded: Int,
     val bookErrors: Int,
     val sequentialBookErrors: Int,
-    val maxSequentialBookErrors: Int,
     val bookDeleted: Int,
     val startedAt: Instant,
     val completedAt: Instant? = null,
@@ -98,7 +97,6 @@ fun ImportStats.toImportJob() = ImportJob(
     booksAdded = booksAdded.load(),
     bookErrors = bookErrors.load(),
     sequentialBookErrors = sequentialBookErrors.load(),
-    maxSequentialBookErrors = ImportStats.MAX_SEQUENTIAL_BOOK_ERRORS,
     bookDeleted = booksDeleted.load(),
     startedAt = startedAt.load(),
     completedAt = completedAt.load(),

src/main/kotlin/io/heapy/kotbusta/service/BookFileService.kt

@@ -2,6 +2,9 @@ package io.heapy.kotbusta.service
 
 import io.heapy.komok.tech.logging.Logger
 import io.heapy.kotbusta.model.Book
+import io.heapy.kotbusta.util.UNDERSCORE_RUN
+import io.heapy.kotbusta.util.WHITESPACE_RUN
+import io.heapy.kotbusta.util.takeCodePointSafe
 import java.io.File
 import java.nio.file.Files
 import java.nio.file.Path
@@ -92,27 +95,15 @@ class ZipBookFileService(
             .trim()
             .ifBlank { "book_${book.id}" }
             .replace(UNSAFE_FILENAME_CHARS, "_")
-            .replace(WHITESPACE, "_")
-            .replace(UNDERSCORES, "_")
+            .replace(WHITESPACE_RUN, "_")
+            .replace(UNDERSCORE_RUN, "_")
             .trim('.', '_')
             .ifBlank { "book_${book.id}" }
-            .truncateForFilename()
-
-    private fun String.truncateForFilename(): String {
-        val truncated = take(100)
-        val withoutTrailingHighSurrogate = if (truncated.lastOrNull()?.isHighSurrogate() == true) {
-            truncated.dropLast(1)
-        } else {
-            truncated
-        }
-
-        return withoutTrailingHighSurrogate.trimEnd('.', '_')
-    }
+            .takeCodePointSafe(100)
+            .trimEnd('.', '_')
 
     private companion object : Logger() {
         private val UNSAFE_FILENAME_CHARS = Regex("""[\p{Cntrl}/\\:*?"<>|]+""")
-        private val WHITESPACE = Regex("""\s+""")
-        private val UNDERSCORES = Regex("""_+""")
     }
 }
 

src/main/kotlin/io/heapy/kotbusta/service/EmailService.kt

@@ -4,7 +4,11 @@ import aws.sdk.kotlin.services.ses.SesClient
 import aws.sdk.kotlin.services.ses.model.RawMessage
 import aws.sdk.kotlin.services.ses.model.SendRawEmailRequest
 import io.heapy.komok.tech.logging.Logger
+import io.heapy.kotbusta.util.WHITESPACE_RUN
 import io.heapy.kotbusta.util.asciiFallbackFileName
+import io.heapy.kotbusta.util.attachmentContentDisposition
+import io.heapy.kotbusta.util.takeCodePointSafe
+import java.io.ByteArrayOutputStream
 import java.io.File
 import java.util.*
 
@@ -19,6 +23,7 @@ interface EmailService {
         recipientEmail: String,
         bookFile: File,
         bookTitle: String,
+        attachmentFileName: String,
         format: String,
     ): EmailResult
 }
@@ -31,6 +36,7 @@ class SesEmailService(
         recipientEmail: String,
         bookFile: File,
         bookTitle: String,
+        attachmentFileName: String,
         format: String,
     ): EmailResult {
         return try {
@@ -46,7 +52,7 @@ class SesEmailService(
                 subject = "Your book: $title",
                 body = "Please find your requested book attached.",
                 attachmentBytes = bookFile.readBytes(),
-                attachmentName = "$title.${format.lowercase()}",
+                attachmentName = attachmentFileName,
                 mimeType = mimeType,
             )
 
@@ -98,22 +104,15 @@ class SesEmailService(
 }
 
 private val CONTROL_OR_QUOTE = Regex("""[\p{Cntrl}"]+""")
-private val WHITESPACE = Regex("""\s+""")
 private const val MAX_SANITIZED_BOOK_TITLE_LENGTH = 100
 
 internal fun sanitizeBookTitle(title: String): String =
     title
         .replace(CONTROL_OR_QUOTE, " ")
-        .replace(WHITESPACE, " ")
+        .replace(WHITESPACE_RUN, " ")
+        .trim()
+        .takeCodePointSafe(MAX_SANITIZED_BOOK_TITLE_LENGTH)
         .trim()
-        .let { sanitizedTitle ->
-            val truncated = sanitizedTitle.take(MAX_SANITIZED_BOOK_TITLE_LENGTH)
-            if (truncated.lastOrNull()?.let { Character.isHighSurrogate(it) } == true) {
-                truncated.dropLast(1)
-            } else {
-                truncated
-            }.trim()
-        }
         .ifBlank { "book" }
 
 internal fun buildRawEmail(
@@ -130,21 +129,14 @@ internal fun buildRawEmail(
 
     val boundary = "----=_Part_${System.currentTimeMillis()}"
     val fallbackName = asciiFallbackFileName(attachmentName)
-    // For non-ASCII names send ONLY the RFC 2231 filename*. Common parsers
-    // (Python's email, and likely Amazon's ingestion) prefer a plain
-    // filename= when both are present, which would replace the real title
-    // with the ASCII fallback; extended-only is what calibre sends to
-    // Kindle and is known to decode correctly there.
-    val disposition = if (attachmentName.all { it.code in 0x20..0x7E }) {
-        """attachment; filename="$attachmentName""""
-    } else {
-        "attachment; filename*=${rfc2231FileName(attachmentName)}"
-    }
+    val disposition = attachmentContentDisposition(attachmentName, includeAsciiFallback = false)
 
-    val emailBuilder = StringBuilder()
+    val out = ByteArrayOutputStream()
+    fun write(text: String) = out.write(text.toByteArray())
     fun line(text: String = "") {
         // RFC 5322 requires CRLF line endings in raw messages.
-        emailBuilder.append(text).append("\r\n")
+        write(text)
+        write("\r\n")
     }
 
     // Headers
@@ -169,13 +161,18 @@ internal fun buildRawEmail(
     line("Content-Transfer-Encoding: base64")
     line("Content-Disposition: $disposition")
     line()
-    line(Base64.getMimeEncoder().encodeToString(attachmentBytes))
-    line()
+    // Stream-encode the attachment straight into `out` to avoid holding a
+    // separate full-size base64 String. wrap(...).close() flushes the final
+    // base64 group; ByteArrayOutputStream.close() is a no-op, so `out` stays
+    // usable afterwards.
+    Base64.getMimeEncoder().wrap(out).use { it.write(attachmentBytes) }
+    line()  // terminate the final base64 line
+    line()  // blank line before the closing boundary
 
     // End boundary
     line("--$boundary--")
 
-    return emailBuilder.toString().toByteArray()
+    return out.toByteArray()
 }
 
 /**
@@ -213,17 +210,3 @@ internal fun encodeMimeHeaderValue(text: String): String {
         "=?UTF-8?B?${Base64.getEncoder().encodeToString(it.toByteArray())}?="
     }
 }
-
-/**
- * RFC 2231/5987 extended parameter value (`UTF-8''%D0%9C...`) carrying the
- * real Unicode file name alongside the ASCII fallback `filename`.
- */
-internal fun rfc2231FileName(fileName: String): String =
-    fileName.toByteArray().joinToString(separator = "", prefix = "UTF-8''") { byte ->
-        val char = byte.toInt().toChar()
-        if (char in 'a'..'z' || char in 'A'..'Z' || char in '0'..'9' || char in "!#$&+-.^_`|~") {
-            char.toString()
-        } else {
-            "%%%02X".format(byte.toInt() and 0xFF)
-        }
-    }

src/main/kotlin/io/heapy/kotbusta/service/EmbeddingService.kt

@@ -78,19 +78,22 @@ class DjlEmbeddingService(
 }
 
 internal class AutoTokenTypeTextEmbeddingTranslator(
-    tokenizer: HuggingFaceTokenizer,
+    private val tokenizer: HuggingFaceTokenizer,
 ) : Translator<String, FloatArray> {
-    private val withoutTokenTypes = textEmbeddingTranslator(tokenizer, includeTokenTypes = false)
-    private val withTokenTypes = textEmbeddingTranslator(tokenizer, includeTokenTypes = true)
+    @Volatile
     private var delegate: TextEmbeddingTranslator? = null
 
     override fun getBatchifier(): Batchifier =
         Batchifier.STACK
 
     override fun prepare(ctx: TranslatorContext) {
-        val includeTokenTypes = requiresTokenTypes(ctx.model.describeInput().keys())
-        delegate = if (includeTokenTypes) withTokenTypes else withoutTokenTypes
-        selectedDelegate().prepare(ctx)
+        val translator = delegate ?: synchronized(this) {
+            delegate ?: textEmbeddingTranslator(
+                tokenizer,
+                requiresTokenTypes(ctx.model.describeInput().keys()),
+            ).also { delegate = it }
+        }
+        translator.prepare(ctx)
     }
 
     override fun processInput(ctx: TranslatorContext, input: String): NDList =

src/main/kotlin/io/heapy/kotbusta/util/AsciiFileName.kt

@@ -22,7 +22,7 @@ internal fun asciiFallbackFileName(fileName: String): String {
             }
         }
         .joinToString("")
-        .replace(Regex("_+"), "_")
+        .replace(UNDERSCORE_RUN, "_")
         .trim('.', '-', '_')
         .ifBlank { "book" }
         .take(100)

src/main/kotlin/io/heapy/kotbusta/util/AttachmentDisposition.kt

@@ -0,0 +1,37 @@
+package io.heapy.kotbusta.util
+
+import io.ktor.http.ContentDisposition
+
+/**
+ * Builds an attachment Content-Disposition for [fileName].
+ *
+ * Pure-ASCII names get a plain `filename=`. A non-ASCII name always carries the
+ * real name in an RFC 5987 `filename*`; [includeAsciiFallback] controls whether
+ * a plain ASCII `filename=` fallback is sent alongside it. Kindle wants the
+ * extended parameter ONLY (false) so mail parsers don't prefer the ASCII
+ * fallback and show a transliterated title; HTTP downloads send both (true) so
+ * clients without RFC 5987 support still get a usable name.
+ */
+internal fun attachmentContentDisposition(
+    fileName: String,
+    includeAsciiFallback: Boolean,
+): ContentDisposition {
+    val isAscii = fileName.all { it.code in 0x20..0x7E }
+    if (isAscii) {
+        return ContentDisposition.Attachment.withParameter(
+            ContentDisposition.Parameters.FileName,
+            fileName,
+        )
+    }
+    val extendedOnly = ContentDisposition.Attachment.withParameter(
+        ContentDisposition.Parameters.FileNameAsterisk,
+        fileName,
+    )
+    return if (includeAsciiFallback) {
+        ContentDisposition.Attachment
+            .withParameter(ContentDisposition.Parameters.FileName, asciiFallbackFileName(fileName))
+            .withParameter(ContentDisposition.Parameters.FileNameAsterisk, fileName)
+    } else {
+        extendedOnly
+    }
+}

src/main/kotlin/io/heapy/kotbusta/util/TextSanitization.kt

@@ -0,0 +1,20 @@
+package io.heapy.kotbusta.util
+
+/** Matches a run of whitespace characters. */
+internal val WHITESPACE_RUN = Regex("""\s+""")
+
+/** Matches a run of underscores. */
+internal val UNDERSCORE_RUN = Regex("""_+""")
+
+/**
+ * Takes at most [maxUtf16Units] UTF-16 code units, then drops a trailing
+ * unpaired high surrogate so truncation never splits a surrogate pair.
+ */
+internal fun String.takeCodePointSafe(maxUtf16Units: Int): String {
+    val truncated = take(maxUtf16Units)
+    return if (truncated.lastOrNull()?.isHighSurrogate() == true) {
+        truncated.dropLast(1)
+    } else {
+        truncated
+    }
+}

src/main/kotlin/io/heapy/kotbusta/worker/KindleSendWorker.kt

@@ -187,7 +187,8 @@ class KindleSendWorker(
                 recipientEmail = device.email,
                 bookFile = materialized.file,
                 bookTitle = book.title,
-                format = item.format,
+                attachmentFileName = materialized.fileName,
+                format = materialized.format,
             )
         } finally {
             materialized.cleanup()

src/main/resources/static/js/components/AdminPanel.js

@@ -98,7 +98,6 @@ export function AdminPanel() {
     : `${processedFiles} files`;
   const currentFile = jobStatus?.currentInpFile;
   const sequentialErrors = jobStatus?.sequentialBookErrors || 0;
-  const maxSequentialErrors = jobStatus?.maxSequentialBookErrors || 100;
   const messages = Object.entries(jobStatus?.messages || {})
     .sort((a, b) => new Date(a[0]) - new Date(b[0]));
 
@@ -186,7 +185,7 @@ export function AdminPanel() {
             h('div', { className: 'metric-label' }, 'Error Streak'),
             h('div', {
               className: `metric-value ${sequentialErrors > 0 ? 'danger-text' : ''}`
-            }, `${sequentialErrors} / ${maxSequentialErrors}`)
+            }, `${sequentialErrors}`)
           )
         ),