纯文本PDF压缩方案求助：现有代码仅支持含图片PDF压缩

问题描述 #

我需要针对仅含文本、无图片的PDF文件的压缩方案。目前使用基于iText的Android代码进行PDF压缩，但该代码仅针对包含图片的PDF有效，对纯文本PDF几乎无压缩效果（仅能压缩100-200KB）。然而应用商店中的同类APP均能实现纯文本PDF的压缩，特此求助优化方案。

当前代码 #

(context as AppCompatActivity).lifecycleScope.launch(Dispatchers.IO) {
    try {
        val reader = PdfReader(inputPath, password.toByteArray())
        //pdfOptimize(reader)
        compressReader(reader)
        saveReader(reader)
        reader.close()

        onPDFCompletion(outputPath)
    } catch (e: IOException) {
        Log.d("PDFCompressionActivityTEST", "execute: ${e.message}")
        onPDFFailed(e.message)
    } catch (e: DocumentException) {
        onPDFFailed(e.message)
    } catch (e: Exception) {
        onPDFFailed(e.message)
    }

}


@Throws(IOException::class)
private fun compressReader(reader: PdfReader) {
    val n = reader.xrefSize
    var `object`: PdfObject?
    var stream: PRStream
    for (i in 0 until n) {
        `object` = reader.getPdfObject(i)
        if (`object` == null || !`object`.isStream) continue
        stream = `object` as PRStream
        compressStream(stream)
    }
    reader.removeUnusedObjects()
}

@Throws(IOException::class)
private fun compressStream(stream: PRStream) {
    val pdfSubType = stream[PdfName.SUBTYPE]
    println(stream.type())
    if (pdfSubType != null && pdfSubType.toString() == PdfName.IMAGE.toString()) {
        val image = PdfImageObject(stream)
        val imageBytes = image.imageAsBytes
        val bmp: Bitmap =
            BitmapFactory.decodeByteArray(imageBytes, 0, imageBytes.size) ?: return
        val width = bmp.width
        val height = bmp.height
        val outBitmap = Bitmap.createBitmap(width, height, Bitmap.Config.ARGB_8888)
        val outCanvas = Canvas(outBitmap)
        outCanvas.drawBitmap(bmp, 0f, 0f, null)
        val imgBytes = ByteArrayOutputStream()
        outBitmap.compress(Bitmap.CompressFormat.JPEG, quality, imgBytes)
        stream.clear()
        stream.setData(imgBytes.toByteArray(), false, PRStream.BEST_COMPRESSION)
        stream.put(PdfName.TYPE, PdfName.XOBJECT)
        stream.put(PdfName.SUBTYPE, PdfName.IMAGE)
        stream.put(PdfName.FILTER, PdfName.DCTDECODE)
        stream.put(PdfName.WIDTH, PdfNumber(width))
        stream.put(PdfName.HEIGHT, PdfNumber(height))
        stream.put(PdfName.BITSPERCOMPONENT, PdfNumber(8))
        stream.put(PdfName.COLORSPACE, PdfName.DEVICERGB)
    }
}


@Throws(DocumentException::class, IOException::class)
private fun saveReader(reader: PdfReader) {
    val stamper = PdfStamper(reader, FileOutputStream(outputPath))
    stamper.setFullCompression()
    stamper.close()
}

优化方案 #

1. 优化内容流压缩 #

当前代码仅处理图片流，纯文本PDF的核心体积来自内容流，需新增文本流的压缩逻辑：
修改compressStream方法，对非图片流使用最高压缩级别重新处理：

@Throws(IOException::class)
private fun compressStream(stream: PRStream) {
    val pdfSubType = stream[PdfName.SUBTYPE]
    if (pdfSubType != null && pdfSubType.toString() == PdfName.IMAGE.toString()) {
        // 保留原图片压缩逻辑
        val image = PdfImageObject(stream)
        val imageBytes = image.imageAsBytes
        val bmp: Bitmap =
            BitmapFactory.decodeByteArray(imageBytes, 0, imageBytes.size) ?: return
        val width = bmp.width
        val height = bmp.height
        val outBitmap = Bitmap.createBitmap(width, height, Bitmap.Config.ARGB_8888)
        val outCanvas = Canvas(outBitmap)
        outCanvas.drawBitmap(bmp, 0f, 0f, null)
        val imgBytes = ByteArrayOutputStream()
        outBitmap.compress(Bitmap.CompressFormat.JPEG, quality, imgBytes)
        stream.clear()
        stream.setData(imgBytes.toByteArray(), false, PRStream.BEST_COMPRESSION)
        stream.put(PdfName.TYPE, PdfName.XOBJECT)
        stream.put(PdfName.SUBTYPE, PdfName.IMAGE)
        stream.put(PdfName.FILTER, PdfName.DCTDECODE)
        stream.put(PdfName.WIDTH, PdfNumber(width))
        stream.put(PdfName.HEIGHT, PdfNumber(height))
        stream.put(PdfName.BITSPERCOMPONENT, PdfNumber(8))
        stream.put(PdfName.COLORSPACE, PdfName.DEVICERGB)
    } else {
        // 对文本内容流重新压缩
        val streamData = PdfReader.getStreamBytes(stream)
        stream.setData(streamData, true, PRStream.BEST_COMPRESSION)
    }
}

2. 字体优化（纯文本PDF压缩核心） #

纯文本PDF体积大的主要原因是嵌入字体，可通过以下方式优化：

• 移除未使用的字体子集：遍历文档资源中的字体，删除未被页面引用的字体对象
• 替换嵌入字体为系统字体（需注意版权和显示兼容性）：

private fun optimizeFonts(reader: PdfReader) {
    val catalog = reader.catalog
    val resources = catalog.getAsDict(PdfName.RESOURCES) ?: return
    val fonts = resources.getAsDict(PdfName.FONT) ?: return
    val fontKeys = fonts.keys.iterator()
    while (fontKeys.hasNext()) {
        val key = fontKeys.next()
        val fontDict = fonts.getAsDict(key) ?: continue
        val baseFont = fontDict.getAsString(PdfName.BASEFONT) ?: continue
        
        // 针对常见嵌入字体（如Arial），移除嵌入文件，改为引用系统字体
        if (baseFont.startsWith("/Arial") || baseFont.startsWith("/TimesNewRoman")) {
            fontDict.remove(PdfName.FONTFILE)
            fontDict.remove(PdfName.FONTFILE2)
            fontDict.remove(PdfName.FONTFILE3)
            fontDict.put(PdfName.EMBEDDED, PdfBoolean.FALSE)
        }
    }
}

调用时在compressReader后添加optimizeFonts(reader)即可。

3. 强化对象清理与合并 #

在compressReader方法中补充更多清理操作：

@Throws(IOException::class)
private fun compressReader(reader: PdfReader) {
    val n = reader.xrefSize
    var `object`: PdfObject?
    var stream: PRStream
    for (i in 0 until n) {
        `object` = reader.getPdfObject(i)
        if (`object` == null || !`object`.isStream) continue
        stream = `object` as PRStream
        compressStream(stream)
    }
    reader.removeUnusedObjects()
    // 合并重复命名目标
    reader.consolidateNamedDestinations()
    // 清理冗余的页面资源
    reader.removeRedundantObjects()
}

4. 禁用不必要的PDF元素 #

如果不需要保留PDF的交互功能（如表单、注释），可以在保存时移除：

@Throws(DocumentException::class, IOException::class)
private fun saveReader(reader: PdfReader) {
    val stamper = PdfStamper(reader, FileOutputStream(outputPath))
    // 移除注释
    stamper.removeAnnotations()
    // 移除表单域（若无需交互）
    stamper.getAcroFields().removeFields()
    stamper.setFullCompression()
    stamper.close()
}

内容的提问来源于stack exchange，提问作者Kaunain