pdf to img: convert jpx to jpeg

周伟奇
Showing 1 changed file with 9 additions and 3 deletions
src/common/tools/pdf_to_img.py
--- a/src/common/tools/pdf_to_img.py
View file @e8bbaf4
+++ b/src/common/tools/pdf_to_img.py
View file @e8bbaf4
@@ -84,9 +84,15 @@ class PDFHandler:
    def extract_single_image(self, pdf, xref, smask, colorspace, pno, img_index=0):
        pix = self.recover_pix(pdf, xref, smask, colorspace)
        ext, img_data = self.get_img_data(pix)
-        img_save_path = self.get_img_save_path(pno, img_index=img_index, ext=ext)
+        if ext == 'jpx':
-        with open(img_save_path, "wb") as f:
+            img_save_path = self.get_img_save_path(pno, img_index=img_index, ext='jpeg')
-            f.write(img_data)
+            jpx_pix = fitz.Pixmap(img_data)
+            jpx_pix.writeImage(img_save_path)
+            jpx_pix = None
+        else:
+            img_save_path = self.get_img_save_path(pno, img_index=img_index, ext=ext)
+            with open(img_save_path, "wb") as f:
+                f.write(img_data)
        self.xref_set.add(xref)
        self.img_path_list.append(img_save_path)