From 0287b68de0cccc4da06fca48aaeafeea8151d891 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20G=C3=B6pfert?=
 <94385965+jangoepfert@users.noreply.github.com>
Date: Fri, 8 Dec 2023 11:47:52 +0100
Subject: [PATCH] Added option to directly pass the PDF as bytes to process_pdf
 instead of loading the PDF from disk

---
 grobid_client/grobid_client.py | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/grobid_client/grobid_client.py b/grobid_client/grobid_client.py
index b63cf20..61fe167 100644
--- a/grobid_client/grobid_client.py
+++ b/grobid_client/grobid_client.py
@@ -243,9 +243,18 @@ def process_pdf(
         include_raw_citations,
         include_raw_affiliations,
         tei_coordinates,
-        segment_sentences
+        segment_sentences,
+        from_memory=False
     ):
-        pdf_handle = open(pdf_file, "rb")
+        if from_memory:
+            # PDF already loaded into memory
+            # expects pdf_file to be of type 'bytes'
+            pdf_handle = io.BytesIO(pdf_file)            
+            pdf_file = ""
+        else:
+            # expects pdf_file to be path to PDF file
+            pdf_handle = open(pdf_file, "rb")
+            
         files = {
             "input": (
                 pdf_file,