toklin 读取word，pdf文件

toklin 读取word，pdf文件

//org.apache.pdfbox:pdfbox:2.0.36

val file = File("input.pdf")   
val pdDocument = PDDocument.load(file)
val pdfTextStripper = PDFTextStripper()
val fullText = pdfTextStripper.getText(pdDocument)
println(fullText)

　　

//org.apache.poi:poi-ooxml:5.5.1

   val textContent = StringBuilder()

    val filePath = File("input.docx")
    try {
        FileInputStream(filePath).let { fis ->
            XWPFDocument(fis).use { document ->
                val paragraphs: MutableList<XWPFParagraph> = document.getParagraphs()
                for (paragraph in paragraphs) {
                    textContent.append(paragraph.getText()).append("\n")
                }
            }
        }
    } catch (e: IOException) {
        e.printStackTrace()
    }
    println("textContent:$textContent")

　　

参考：https://www.oryoy.com/news/android-gao-xiao-jie-xi-word-wen-dang-shi-yong-java-shi-xian-wen-dang-du-qu-yu-nei-rong-ti-qu.html

https://poi.apache.org/index.html