From 15d9af835a6377d190e750f4f760da5a3f4920d2 Mon Sep 17 00:00:00 2001
From: Alexis de Lattre <alexis.delattre@akretion.com>
Date: Tue, 11 Jun 2019 17:28:14 +0200
Subject: [PATCH] Add getAttachFiles() on PdfFileReader

---
 pypdf/pdf.py | 99 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 99 insertions(+)

diff --git a/pypdf/pdf.py b/pypdf/pdf.py
index 82f625671..af2c10f05 100644
--- a/pypdf/pdf.py
+++ b/pypdf/pdf.py
@@ -2605,6 +2605,105 @@ def _authenticateUserPassword(self, password):
     def isEncrypted(self):
         return "/Encrypt" in self._trailer
 
+    def _get_dict_entry(self, node, entry):
+        if not isinstance(node, dict):
+            raise ValueError('The node must be a dict')
+        dict_entry = node.get(entry)
+        if isinstance(dict_entry, dict):
+            return dict_entry
+        elif isinstance(dict_entry, IndirectObject):
+            res_dict_entry = dict_entry.getObject()
+            if isinstance(res_dict_entry, dict):
+                return res_dict_entry
+            else:
+                return False
+        else:
+            return False
+
+    def _parse_embeddedfiles_kids_node(self, kids_node, level, res):
+        if level not in [1, 2]:
+            raise ValueError('Level argument should be 1 or 2')
+        # The /Kids entry of the EmbeddedFiles name tree must be an array
+        if not isinstance(kids_node, list):
+            return False
+        for kid_entry in kids_node:
+            # The /Kids entry of the EmbeddedFiles name tree must be a
+            # list of IndirectObjects
+            if not isinstance(kid_entry, IndirectObject):
+                return False
+            kids_node = kid_entry.getObject()
+            # The /Kids entry of the EmbeddedFiles name tree
+            # must be a list of IndirectObjects that point to dict objects
+            if not isinstance(kids_node, dict):
+                return False
+            if '/Names' in kids_node:
+                # The /Names entry in EmbeddedFiles must be an array
+                if not isinstance(kids_node['/Names'], list):
+                    return False
+                res += kids_node['/Names']
+            elif '/Kids' in kids_node and level == 1:
+                kids_node_l2 = kids_node['/Kids']
+                self._parse_embeddedfiles_kids_node(kids_node_l2, 2, res)
+            else:
+                # /Kids node should have a /Names or /Kids entry
+                return False
+        return True
+
+    def _get_embeddedfiles(self, embeddedfiles_node):
+        if not isinstance(embeddedfiles_node, dict):
+            raise ValueError('The EmbeddedFiles node must be a dict')
+        res = []
+        if '/Names' in embeddedfiles_node:
+            # The /Names entry of the EmbeddedFiles name tree must be an array
+            if not isinstance(embeddedfiles_node['/Names'], list):
+                return False
+            res = embeddedfiles_node['/Names']
+        elif '/Kids' in embeddedfiles_node:
+            kids_node = embeddedfiles_node['/Kids']
+            parse_result = self._parse_embeddedfiles_kids_node(
+                kids_node, 1, res)
+            if parse_result is False:
+                return False
+        else:
+            # The EmbeddedFiles name tree should have either a /Names or a
+            # /Kids entry
+            return False
+        # The EmbeddedFiles name tree should point to an even number
+        # of elements
+        if len(res) % 2 != 0:
+            return False
+        return res
+
+    def getAttachFiles(self):
+        """
+        Retrieves all the attachments.
+
+        :return: a list of tuples (filename, file)
+        :rtype: list
+        """
+        res = []
+        catalog = self.trailer['/Root']
+        # get the name tree
+        catalog_name = self._get_dict_entry(catalog, '/Names')
+        if not catalog_name:
+            return res
+        embeddedfiles_node = self._get_dict_entry(
+            catalog_name, '/EmbeddedFiles')
+        if not embeddedfiles_node:
+            return res
+        embeddedfiles = self._get_embeddedfiles(embeddedfiles_node)
+        if not embeddedfiles:
+            return res
+        embeddedfiles_by_two = list(zip(embeddedfiles, embeddedfiles[1:]))[::2]
+        for (filename, file_obj) in embeddedfiles_by_two:
+            file_dict = file_obj.getObject()
+            try:
+                file_content = file_dict['/EF']['/F'].getData()
+                res.append((filename, file_content))
+            except KeyError:
+                pass
+        return res
+
 
 def _convertToInt(d, size):
     if size > 8: