From 765224fe7e938a9c5ab8fc626386703341ec3911 Mon Sep 17 00:00:00 2001
From: s <102841251+duckduckgrayduck@users.noreply.github.com>
Date: Tue, 19 Sep 2023 15:24:40 -0500
Subject: [PATCH] Barebones of main.py

---
 main.py | 38 ++++++++++++--------------------------
 1 file changed, 12 insertions(+), 26 deletions(-)

diff --git a/main.py b/main.py
index fc5f449..d39e46a 100644
--- a/main.py
+++ b/main.py
@@ -1,38 +1,24 @@
 """
-This is a hello world add-on for DocumentCloud.
-
-It demonstrates how to write a add-on which can be activated from the
-DocumentCloud add-on system and run using Github Actions.  It receives data
-from DocumentCloud via the request dispatch and writes data back to
-DocumentCloud using the standard API
+This is an Add-On that uses docTR https://github.com/mindee/doctr to OCR documents for DocumentCloud
 """
 
 from documentcloud.addon import AddOn
 
 
-class HelloWorld(AddOn):
-    """An example Add-On for DocumentCloud."""
+class docTR(AddOn):
+    """Class definition"""
 
     def main(self):
         """The main add-on functionality goes here."""
-        # fetch your add-on specific data
-        name = self.data.get("name", "world")
-
-        self.set_message("Hello World start!")
-
-        # add a hello note to the first page of each selected document
+        model = ocr_predictor(pretrained=True)
         for document in self.get_documents():
-            # get_documents will iterate through all documents efficiently,
-            # either selected or by query, dependeing on which is passed in
-            document.annotations.create(f"Hello {name}!", 0)
-
-        with open("hello.txt", "w+") as file_:
-            file_.write("Hello world!")
-            self.upload_file(file_)
-
-        self.set_message("Hello World end!")
-        self.send_mail("Hello World!", "We finished!")
-
+            pdf_name = f"'{document.title}.pdf'"
+            with archive.open(f"{document.slug} - {document.id}.pdf", "w") as pdf:
+                pdf.write(document.pdf)
+                doc = DocumentFile.from_pdf(pdf_name)
+                result = model(doc)
+                json_export = result.export()
+                print(json_export)
 
 if __name__ == "__main__":
-    HelloWorld().main()
+    docTR().main()