pdfrb 0.8.8 → 0.8.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: af5b6aea283bcc7325c584008268aaedc228f78882cb5408b615c55576cb8154
-  data.tar.gz: 29c3a794f7829144e4fa2275ba4a7b4124087abd7279f076689b06d14d5ac4c3
+  metadata.gz: c9ccd2f148c786f735078767b672eacf4e8861100894f68b248905074c73e7eb
+  data.tar.gz: eaceeb41569d8fab78bcd46891608f6760d4c2a3d47ef45da896c500e32dd736
 SHA512:
-  metadata.gz: 6491c0a75e0ef6eb6fa849fdf25de0d546c2055b099e3f533595a073e370a30c37693604b31620e5d3707237e1960f18f6ba7ee05555698c21b081e6300cf493
-  data.tar.gz: 569d900685cc748edba60dfcdc426987343025c4ccae008a715b188ce41ce9d56abf29901debb39e1292d856ba34951394b85a57efb297ac9b5af0b67f706eb6
+  metadata.gz: 909be54b72b05e78ec5bda296a335cfb2b1e78cf2ade15695d45b59e29bb9c73ef95cbfa5f1c4bb4c61a8a191055aaff2dfb3e36e0efb4970b7c0bf71d19cee8
+  data.tar.gz: 609afef48ccdeed968a0f1791dd1b7f5e295ce1169e6ab4c4ace378fdd6bd6f5eab116b67c15b4feb40eab0f590a6a4dc29cb0e30910c393624dd8a61f72d112

data/lib/pdfrb/document/form.rb

@@ -86,7 +86,9 @@ module Pdfrb
 
       # Find a field by fully qualified name.
       def find(name)
-        each_field.find { |f| f.value[:T] == name }
+        each_field.find do |f|
+          Pdfrb::Model::Cos::TextString.decode(f.value[:T]) == name
+        end
       end
 
       # Enumerate all top-level fields.
@@ -105,7 +107,9 @@ module Pdfrb
 
       # List of fully qualified field names (T values).
       def field_names
-        each_field.filter_map { |f| f.value[:T] }
+        each_field.filter_map do |f|
+          Pdfrb::Model::Cos::TextString.decode(f.value[:T])
+        end
       end
 
       # Set a field value by name. Auto-regenerates the appearance stream
@@ -129,7 +133,7 @@ module Pdfrb
         field = find(name)
         return nil unless field
 
-        field.value[:V]
+        Pdfrb::Model::Cos::TextString.decode(field.value[:V])
       end
 
       # Flatten all form fields: stamp their appearance streams into page

data/lib/pdfrb/document/metadata.rb

@@ -45,7 +45,10 @@ module Pdfrb
         raw = info.value[field]
         return nil if raw.nil?
 
-        raw.is_a?(::String) ? raw.dup.force_encoding("UTF-8") : raw
+        # Info values are text strings (s7.9.2.2): they may arrive as
+        # UTF-16BE+BOM or PDFDocEncoding bytes (decryption hands back
+        # wire bytes). Decode rather than relabel the encoding.
+        Pdfrb::Model::Cos::TextString.decode(raw)
       end
 
       def write_field(field, value)

data/lib/pdfrb/model/cos/text_string.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Cos
+      # Semantic accessors for spec-defined text strings (s7.9.2.2).
+      # On the wire these are UTF-16BE+BOM or PDFDocEncoding bytes —
+      # and read-side decryption hands them back as raw bytes — so
+      # every accessor that means "this field is text" decodes via
+      # this seam instead of relabelling the encoding. The Model
+      # itself stays lossless: only semantic accessors decode.
+      module TextString
+        module_function
+
+        # @return [String] the decoded UTF-8 text, or +raw+ unchanged
+        #   when it is not a String (names, arrays, dates pass
+        #   through untouched).
+        def decode(raw)
+          return raw unless raw.is_a?(::String)
+
+          StringEncoding.decode_text(raw)
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/cos.rb

@@ -16,6 +16,7 @@ module Pdfrb
       autoload :Dictionary, "pdfrb/model/cos/dictionary"
       autoload :Stream, "pdfrb/model/cos/stream"
       autoload :ArlingtonBacked, "pdfrb/model/cos/arlington_backed"
+      autoload :TextString, "pdfrb/model/cos/text_string"
     end
   end
 end

data/lib/pdfrb/model/type/annotation.rb

@@ -26,8 +26,15 @@ module Pdfrb
         # ---- Accessors ----
         def subtype; self[:Subtype]; end
         def rect; self[:Rect]; end
-        def contents; self[:Contents]; end
-        def name; self[:NM]; end
+
+        def contents
+          Pdfrb::Model::Cos::TextString.decode(self[:Contents])
+        end
+
+        def name
+          Pdfrb::Model::Cos::TextString.decode(self[:NM])
+        end
+
         def modified_date; self[:M]; end
         def flags; self[:F] || 0; end
         def color; self[:C]; end

data/lib/pdfrb/model/type/form_field.rb

@@ -8,11 +8,27 @@ module Pdfrb
         register_type :Field
 
         def field_type; self[:FT]; end
-        def field_name; self[:T]; end
-        def alternate_field_name; self[:TU]; end
-        def mapping_field_name; self[:TM]; end
-        def field_value; self[:V]; end
-        def default_value; self[:DV]; end
+
+        def field_name
+          Pdfrb::Model::Cos::TextString.decode(self[:T])
+        end
+
+        def alternate_field_name
+          Pdfrb::Model::Cos::TextString.decode(self[:TU])
+        end
+
+        def mapping_field_name
+          Pdfrb::Model::Cos::TextString.decode(self[:TM])
+        end
+
+        def field_value
+          Pdfrb::Model::Cos::TextString.decode(self[:V])
+        end
+
+        def default_value
+          Pdfrb::Model::Cos::TextString.decode(self[:DV])
+        end
+
         def kids; self[:Kids]; end
         def parent; self[:Parent]; end
         def flags; self[:Ff] || 0; end

data/lib/pdfrb/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Pdfrb
-  VERSION = "0.8.8"
+  VERSION = "0.8.10"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: pdfrb
 version: !ruby/object:Gem::Version
-  version: 0.8.8
+  version: 0.8.10
 platform: ruby
 authors:
 - Ribose Inc.
@@ -999,6 +999,7 @@ files:
 - lib/pdfrb/model/cos/name_encoding.rb
 - lib/pdfrb/model/cos/stream.rb
 - lib/pdfrb/model/cos/string_encoding.rb
+- lib/pdfrb/model/cos/text_string.rb
 - lib/pdfrb/model/date.rb
 - lib/pdfrb/model/matrix.rb
 - lib/pdfrb/model/object.rb