added dump_data and update_info

README.rst

@@ -110,6 +110,25 @@ Read PDF and output form field statistics.
 
 -  ``pdf_path`` : input PDF file
 
+``dump_data``
+~~~~~~~~~~~~~~~~~~~~
+
+Read PDF and output metadata, bookmarks, page metrics.
+
+-  ``pdf_path`` : input PDF file
+-  ``out_file`` (default=None) : text file containing dumped data
+
+
+``update_info``
+~~~~~~~~~~~~~~~~~~~~
+
+Read PDF ``pdf_path``, update metadata according to ``info_data`` and
+write the finale PDF to ``out_file``.
+
+-  ``pdf_path`` : input PDF file
+-  ``info_data`` : text file containing new info data
+-  ``out_file`` : output PDF file
+
 
 Example
 -------

pypdftk.py

@@ -305,3 +305,23 @@ def uncompress(pdf_path, out_file=None, flatten=True):
     '''
 
     return pdftk_cmd_util(pdf_path, "uncompress", out_file, flatten)
+
+def dump_data(pdf_path, out_file=None):
+    '''
+    Reads a single input PDF file and returns its metadata, bookmarks
+    (a/k/a outlines), page metrics (media, rotation and labels) and
+    other data
+
+    :param pdf_path: input PDF file
+    :param out_file: (default=None) : data will also be written to this file if provided
+    :return: dumped_data
+    '''
+    info_data = run_command([PDFTK_PATH, pdf_path, 'dump_data'])
+    if out_file:
+        with open(out_file, 'w') as f:
+            f.writelines(info_data)
+
+    return "\n".join(info_data)
+
+def update_info(pdf_path, info_data, out_file):
+    run_command([PDFTK_PATH, pdf_path, 'update_info', info_data, 'output', out_file])

test.py

@@ -145,6 +145,64 @@ def test_replace_page_at_end(self):
         pypdftk.replace_page(TEST_PDF_PATH, last_page, pdf_to_insert)
         self.assertEqual(total_pages, pypdftk.get_num_pages(TEST_PDF_PATH))
 
+    def test_dump_data(self):
+        form_info_data = """InfoBegin
+InfoKey: Keywords
+InfoValue: PDF Form
+InfoBegin
+InfoKey: Creator
+InfoValue: Writer
+InfoBegin
+InfoKey: CreationDate
+InfoValue: D:20130629204853+02'00'
+InfoBegin
+InfoKey: Producer
+InfoValue: OpenOffice.org 3.4
+InfoBegin
+InfoKey: Title
+InfoValue: PDF Form Example
+PdfID0: 5e0a553555622a0516e9877ca55217a6
+PdfID1: 5e0a553555622a0516e9877ca55217a6
+NumberOfPages: 1
+PageMediaBegin
+PageMediaNumber: 1
+PageMediaRotation: 0
+PageMediaRect: 0 0 595 842
+PageMediaDimensions: 595 842"""
+        dumped_data = pypdftk.dump_data('test_files/form.pdf')
+        self.assertEqual(dumped_data, form_info_data)
+
+    def test_update_info(self):
+        form_info_data = """InfoBegin
+InfoKey: Keywords
+InfoValue: My fancy form
+InfoBegin
+InfoKey: Creator
+InfoValue: Ghostwriter
+InfoBegin
+InfoKey: CreationDate
+InfoValue: D:20210101204853+02'00'
+InfoBegin
+InfoKey: Producer
+InfoValue: PDFTK
+InfoBegin
+InfoKey: Title
+InfoValue: Form with updated metadata
+PdfID0: 5e0a553555622a0516e9877ca55217a6
+PdfID1: 5e0a553555622a0516e9877ca55217a6
+NumberOfPages: 1
+PageMediaBegin
+PageMediaNumber: 1
+PageMediaRotation: 0
+PageMediaRect: 0 0 595 842
+PageMediaDimensions: 595 842"""
+
+        with open("test_files/form_info_data.txt", 'w') as f:
+            f.write(form_info_data)
+        pypdftk.update_info('test_files/form.pdf', 'test_files/form_info_data.txt', 'test_files/form_updated.pdf')
+        dumped_data = pypdftk.dump_data('test_files/form_updated.pdf')
+        self.assertEqual(dumped_data, form_info_data)
+
     @unittest.skip('Not implemented yet')
     def test_stamp(self):
         pass