[IMP] pdf_helper: add pdf_embed_xml

pdf_helper/models/helper.py

@@ -3,9 +3,11 @@
 # Copyright 2023 Jacques-Etienne Baudoux (BCIM) <[email protected]>
 # License LGPL-3.0 or later (http://www.gnu.org/licenses/lgpl).
 
+import io
 import logging
 
 from odoo import api, models
+from odoo.tools.pdf import OdooPdfFileReader, OdooPdfFileWriter
 
 from ..utils import PDFParser
 
@@ -29,5 +31,19 @@ def pdf_get_xml_files(self, pdf_file):
         try:
             return parser.get_xml_files()
         except parser.get_xml_files_swallable_exceptions() as err:
+            # TODO: can't we catch specific exceptions?
+            # This try/except block was added to reflect what done
+            # in base_business_document_import till now.
             _logger.error("PDF file parsing failed: %s", str(err))
             return {}
+
+    @api.model
+    def pdf_embed_xml(self, pdf_content, xml_filename, xml_string):
+        """Add an XML attachment in a pdf"""
+        with io.BytesIO(pdf_content) as reader_buffer, io.BytesIO() as new_pdf_stream:
+            reader = OdooPdfFileReader(reader_buffer, strict=False)
+            writer = OdooPdfFileWriter()
+            writer.cloneReaderDocumentRoot(reader)
+            writer.addAttachment(xml_filename, xml_string, subtype="text/xml")
+            writer.write(new_pdf_stream)
+            return new_pdf_stream.getvalue()

pdf_helper/readme/USAGE.rst

@@ -2,6 +2,8 @@ Inside Odoo env::
 
     res = env["pdf.helper"].pdf_get_xml_files(pdf_filecontent)
 
+    new_pdf_filecontent = env["pdf.helper"].pdf_embed_xml(pdf_filecontent, filename, xml)
+
 Outside Odoo env::
 
     from odoo.addons.pdf_helper.utils import PDFParser

pdf_helper/tests/test_helper.py

@@ -43,3 +43,15 @@ def test_get_xml_fail(self):
                 "PDF file parsing failed: Cannot read an empty file",
                 log_catcher.output[0],
             )
+
+    def test_embed_xml(self):
+        pdf_content = read_test_file("pdf_with_xml_test.pdf", mode="rb")
+        filename = "test"
+        xml = b"<root>test</root>"
+        newpdf_content = self.env["pdf.helper"].pdf_embed_xml(
+            pdf_content, filename, xml
+        )
+        attachments = self.env["pdf.helper"].pdf_get_xml_files(newpdf_content)
+        self.assertTrue(filename in attachments)
+        etree_content = attachments[filename]
+        self.assertEqual(xml, etree.tostring(etree_content))