bytes to str when reading html or convert to text WIP

2ddb82bc · Emmy Vouriot · Jérome Perrin · 8f489804 · 2ddb82bc
Commit 2ddb82bc authored Aug 07, 2023 by Emmy Vouriot Committed by Jérome Perrin May 20, 2024
Hide whitespace changes
Inline Side-by-side

Showing with 5 additions and 5 deletions

bt5/erp5_dms/DocumentTemplateItem/portal_components/document.erp5.PDFDocument.py ...mplateItem/portal_components/document.erp5.PDFDocument.py +5 -5

No files found.
--- a/bt5/erp5_dms/DocumentTemplateItem/portal_components/document.erp5.PDFDocument.py
+++ b/bt5/erp5_dms/DocumentTemplateItem/portal_components/document.erp5.PDFDocument.py
@@ -194,7 +194,7 @@ class PDFDocument(Image):
                                             context=self, filename=filename,
                                             mimetype=self.getContentType())
    if result:
-      return result
+      return bytes2str(result)
    else:
      # Try to use OCR from ghostscript, but tolerate that the command might
      # not be available.
@@ -282,7 +282,7 @@ class PDFDocument(Image):
      command = ['pdftohtml', '-enc', 'UTF-8', '-stdout',
                 '-noframes', '-i', tmp.name]
      try:
-        command_result = Popen(command, stdout=PIPE).communicate()[0]
+        command_result = bytes2str(Popen(command, stdout=PIPE).communicate()[0])
      except OSError as e:
        if e.errno == errno.ENOENT:
          raise ConversionError('pdftohtml was not found')
@@ -291,10 +291,10 @@ class PDFDocument(Image):
    finally:
      tmp.close()
    # Quick hack to remove bg color - XXX
-    h = command_result.replace(b'<BODY bgcolor="#A0A0A0"', b'<BODY ')
+    h = command_result.replace('<BODY bgcolor="#A0A0A0"', '<BODY ')
    # Make links relative
-    h = h.replace(str2bytes('href="%s.html' % tmp.name.split(os.sep)[-1]),
+    h = h.replace('href="%s.html' % tmp.name.split(os.sep)[-1],
-                                                          b'href="asEntireHTML')
+                                                          'href="asEntireHTML')
    return h
  security.declarePrivate('_convertToDJVU')