Add unoconv-based script to normalize office docs

normalization/preservation-office-pdfa.py

@@ -0,0 +1,41 @@
+import subprocess
+import sys
+
+# Python 2/3 compatibility
+try:
+    range = xrange
+except NameError:
+    pass
+
+MAX_TRIES = 16
+
+
+def convert(document, target, max_tries=MAX_TRIES):
+    """
+    Attempt to convert a document into a PDF at the target location.
+
+    This continues up to max_tries times; it's possible for initial calls
+    to fail due to delays in spinning up a LibreOffice server, where
+    retries may succeed.
+    """
+    for _ in range(0, max_tries):
+        try:
+            # SelectPdfVersion=1 converts to PDF/A-1a, instead of
+            # the default PDF 1.4
+            subprocess.check_call(['unoconv', '-eSelectPdfVersion=1',
+                                   '--output', target, document])
+            return 0
+        except subprocess.CalledProcessError:
+            continue
+
+    return 1
+
+if __name__ == '__main__':
+    import argparse
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--file-full-name", dest="input")
+    parser.add_argument("--output-file-path", dest="output")
+    args, _ = parser.parse_known_args()
+
+    sys.exit(convert(args.input, args.output + '.pdf'))