TheAnachronism/docspell
1
0
Fork 0
mirror of https://github.com/TheAnachronism/docspell.git synced 2025-06-22 02:18:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update pdfbox to 3.0.0

Browse Source
This commit is contained in:
eikek
2023-11-05 23:34:51 +01:00
parent 84612bc7e7
commit fe4a300b0e
6 changed files with 18 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
modules/extract/src/main/scala/docspell/extract/pdfbox/PdfLoader.scala
View File

@ -11,11 +11,12 @@ import cats.implicits._
import fs2.Stream
import org.apache.pdfbox.pdmodel.PDDocument
import org.apache.pdfbox.{Loader => PdfboxLoader}
object PdfLoader {
private def readBytes1[F[_]: Sync](bytes: Array[Byte]): F[PDDocument] =
Sync[F].delay(PDDocument.load(bytes))
Sync[F].delay(PdfboxLoader.loadPDF(bytes))
private def closePDDocument[F[_]: Sync](pd: PDDocument): F[Unit] =
Sync[F].delay(pd.close())

15
modules/extract/src/main/scala/docspell/extract/pdfbox/PdfboxExtract.scala
View File

@ -6,8 +6,6 @@
package docspell.extract.pdfbox
import java.io.InputStream
import scala.util.{Try, Using}
import cats.effect.Sync
@ -20,6 +18,7 @@ import docspell.extract.internal.Text
import org.apache.pdfbox.pdmodel.PDDocument
import org.apache.pdfbox.text.PDFTextStripper
import org.apache.pdfbox.{Loader => PdfboxLoader}
object PdfboxExtract {
@ -44,11 +43,8 @@ object PdfboxExtract {
.attempt
.map(_.flatten)
def getText(is: InputStream): Either[Throwable, Text] =
Using(PDDocument.load(is))(readText).toEither.flatten
def getText(inFile: Path): Either[Throwable, Text] =
Using(PDDocument.load(inFile.toNioPath.toFile))(readText).toEither.flatten
Using(PdfboxLoader.loadPDF(inFile.toNioPath.toFile))(readText).toEither.flatten
private def readText(doc: PDDocument): Either[Throwable, Text] =
Try {
@ -64,11 +60,8 @@ object PdfboxExtract {
.attempt
.map(_.flatten)
def getMetaData(is: InputStream): Either[Throwable, PdfMetaData] =
Using(PDDocument.load(is))(readMetaData).toEither.flatten
def getMetaData(inFile: Path): Either[Throwable, PdfMetaData] =
Using(PDDocument.load(inFile.toNioPath.toFile))(readMetaData).toEither.flatten
Using(PdfboxLoader.loadPDF(inFile.toNioPath.toFile))(readMetaData).toEither.flatten
private def readMetaData(doc: PDDocument): Either[Throwable, PdfMetaData] =
Try {
@ -83,7 +76,7 @@ object PdfboxExtract {
mkValue(info.getKeywords),
mkValue(info.getCreator),
Option(info.getCreationDate).map(c => Timestamp(c.toInstant)),
doc.getNumberOfPages()
doc.getNumberOfPages
)
}.toEither
}

20
modules/extract/src/test/scala/docspell/extract/pdfbox/PdfboxExtractTest.scala
View File

@ -21,16 +21,6 @@ class PdfboxExtractTest extends FunSuite with TestLoggingConfig {
ExampleFiles.letter_en_pdf -> TestFiles.letterENText
)
test("extract text from text PDFs by inputstream") {
textPDFs.foreach { case (file, txt) =>
val url = file.toJavaUrl.fold(sys.error, identity)
val str = PdfboxExtract.getText(url.openStream()).fold(throw _, identity)
val received = removeFormatting(str.value)
val expect = removeFormatting(txt)
assertEquals(received, expect)
}
}
test("extract text from text PDFs via Stream") {
textPDFs.foreach { case (file, txt) =>
val data = file.readURL[IO](8192)
@ -42,18 +32,18 @@ class PdfboxExtractTest extends FunSuite with TestLoggingConfig {
}
test("extract text from image PDFs") {
val url = ExampleFiles.scanner_pdf13_pdf.toJavaUrl.fold(sys.error, identity)
val pdfData = ExampleFiles.scanner_pdf13_pdf.readURL[IO](8192)
val str = PdfboxExtract.getText(url.openStream()).fold(throw _, identity)
val str = PdfboxExtract.getText(pdfData).unsafeRunSync().fold(throw _, identity)
assertEquals(str.value, "")
}
test("extract metadata from pdf") {
val url = ExampleFiles.keywords_pdf.toJavaUrl.fold(sys.error, identity)
val str = PdfboxExtract.getText(url.openStream()).fold(throw _, identity)
val pdfData = ExampleFiles.keywords_pdf.readURL[IO](8192)
val str = PdfboxExtract.getText(pdfData).unsafeRunSync().fold(throw _, identity)
assert(str.value.startsWith("Keywords in PDF"))
val md = PdfboxExtract.getMetaData(url.openStream()).fold(throw _, identity)
val md = PdfboxExtract.getMetaData(pdfData).unsafeRunSync().fold(throw _, identity)
assertEquals(md.author, Some("E.K."))
assertEquals(md.title, Some("Keywords in PDF"))
assertEquals(md.subject, Some("This is a subject"))