From 3618c2617006f1c6fc08f88f04b122fe34406185 Mon Sep 17 00:00:00 2001
From: Daniel Bosk <dbosk@kth.se>
Date: Tue, 17 Dec 2024 13:21:27 +0100
Subject: [PATCH] Converts PDFs to text using pdf2txt

---
 src/canvaslms/cli/submissions.nw | 22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/src/canvaslms/cli/submissions.nw b/src/canvaslms/cli/submissions.nw
index a20be10..6e0eae6 100644
--- a/src/canvaslms/cli/submissions.nw
+++ b/src/canvaslms/cli/submissions.nw
@@ -905,6 +905,7 @@ def convert_to_md(attachment: canvasapi.file.File,
   <<download [[attachment]] to [[outfile]] in [[tmpdir]]>>
   content_type = getattr(attachment, "content-type")
   <<if [[content_type]] is text, just use [[outfile]] contents>>
+  <<else if [[content_type]] is PDF, convert [[outfile]] using [[pdf2txt]]>>
   <<else convert [[outfile]] using [[pypandoc]]>>
 <<let [[contents]] be the converted [[attachment]]>>=
 contents = convert_to_md(attachment, tmpdir)
@@ -951,6 +952,8 @@ def text_to_md(content_type):
 This leaves us with the following.
 The advantage of reading the content from the file is that Python will solve 
 the encoding for us.
+Instead of using an [[if]] statement, we'll go all Python and use a 
+[[try-except]] block.
 <<if [[content_type]] is text, just use [[outfile]] contents>>=
 try:
   md_type = text_to_md(content_type)
@@ -961,7 +964,22 @@ except ValueError:
   pass
 @
 
-If the content type is not text, we use [[pypandoc]] to convert it to Markdown.
+Now we'll do the same for PDF files.
+We'll use [[pdf2txt]] to convert the PDF to text.
+However, here we'll use an if statement.
+We'll check for the content type to end with [[pdf]], that will capture also 
+[[x-pdf]].
+<<else if [[content_type]] is PDF, convert [[outfile]] using [[pdf2txt]]>>=
+if content_type.endswith("pdf"):
+  try:
+    return subprocess.check_output(["pdf2txt", str(outfile)],
+                                    text=True)
+  except subprocess.CalledProcessError:
+    pass
+@
+
+Finally, as a last attempt, we use [[pypandoc]] to try to convert it to 
+Markdown.
 Here we'll use Pandoc's ability to infer the file type on its own.
 This means we'll have to download the attachment as a file in a temporary 
 location and let Pandoc convert the file to Markdown.
@@ -969,7 +987,7 @@ location and let Pandoc convert the file to Markdown.
 try:
   return pypandoc.convert_file(outfile, "markdown")
 except Exception as err:
-  return f"Pandoc cannot convert this file. " \
+  return f"Cannot convert this file. " \
          f"The file is located at\n\n  {outfile}\n\n"
 @