add helper script to decode a PDF string (for debugging) [skip ci]

1 files changed, 29 insertions, 0 deletions

diff --git a/scripts/decode-pdf-string b/scripts/decode-pdf-string
new file mode 100755
index 00000000..f4198ddf
--- /dev/null
+++ b/scripts/decode-pdf-string
@@ -0,0 +1,29 @@
+#!/usr/bin/env ruby
+
+str = ARGV[0]
+
+def string_object_to_utf16 str
+  [str.force_encoding(::Encoding::US_ASCII)].pack('H*').force_encoding ::Encoding::UTF_16BE
+end
+
+def string_object_to_utf8 str
+  [str.force_encoding(::Encoding::US_ASCII)].pack('H*').force_encoding ::Encoding::UTF_8
+end
+
+def utf16_to_utf8 str
+  str.encode ::Encoding::UTF_8
+end
+
+def decode_hexified_utf16_string_object str
+  utf16_to_utf8 string_object_to_utf16 str
+end
+
+def decode_hexified_utf8_string_object str
+  string_object_to_utf8 str
+end
+
+if ARGV[1] == 'utf8'
+  puts decode_hexified_utf8_string_object ARGV[0].dup
+else
+  puts decode_hexified_utf16_string_object ARGV[0].dup
+end