From d41622674b2998e178a53fed02351ba2ae1379b8 Mon Sep 17 00:00:00 2001
From: Jonas Smedegaard <dr@jones.dk>
Date: Wed, 21 Feb 2024 18:55:52 +0100
Subject: add helper script xmp2rdfxml, and make targets turtle-from-*
 dotgraph-from-*

---
 bin/xmp2rdfxml | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)
 create mode 100755 bin/xmp2rdfxml

(limited to 'bin/xmp2rdfxml')

diff --git a/bin/xmp2rdfxml b/bin/xmp2rdfxml
new file mode 100755
index 0000000..81011b9
--- /dev/null
+++ b/bin/xmp2rdfxml
@@ -0,0 +1,25 @@
+#!/usr/bin/perl
+
+# extract XMP data from PDF file and unwrap as generic RDF/XML
+
+use v5.36;
+use strict;
+
+use PDF::API2;
+
+my $xml = PDF::API2->open( $ARGV[0] )->xml_metadata();
+
+# strip noise
+$xml =~ s,^PDF version of requested feature [^<]*,,;
+
+# replace whitespace and surrounding XMP boilerplate with XML boilerplate
+$xml =~ s,\s*<\?xpacket [^>]+>\s*,,;
+$xml =~ s,<x:xmpmeta [^>]+>,<?xml version="1.0" encoding="utf-8"?>,;
+
+# strip trailing boilerplate
+$xml =~ s,\s*</x:xmpmeta>,,;
+$xml =~ s,\s*<\?xpacket [^>]+>,,;
+
+say $xml;
+
+1;
-- 
cgit v1.2.3