mindee 4.7.0.pre.rc1 → 4.7.0.pre.rc2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c88f4c2b47bf3a97cd519d1c658f12a874ed930c8e86d5ff208c0c44d4174cca
-  data.tar.gz: 191850ef611da7ca60fac4827f6c530348adf2d07d4a7790dd46a6abb335e906
+  metadata.gz: 95c963d2c0530278f42fa3543609db9f34156d74e031c08c929be8043794425e
+  data.tar.gz: 92b4cfd60be3f710de9bdab714d4291b152d6d253373dcfae20428a940ad5ec9
 SHA512:
-  metadata.gz: cbf412b01eb4c6a324f6548232ce24e7ed0bd6f20b12d3803a47218e971805b77c32989e644077809aaf3ec404946461deaf4e440385812228ece7d8d839d6b6
-  data.tar.gz: 381aebed32ebee1cbd763f361c46bf76eff8fd6592f1d9652ccb79ca5515a804e20c8a36459c56af189e5af883ec9b02c2f69a921dde46d15ad6bb1b7224267d
+  metadata.gz: 64be40a13e4662730c5749d5348860abf30e0c44002e4501d19a99a8c51c4afb2a83f9d901b1b3b4bbc43d255d8db3a27410b4ff5d274028092b7ec1368d830d
+  data.tar.gz: 5e9201f8c7d9beee93f601e0fe12758c20f55175c7c0476544bada2d1f8a7828962c848c75bb882169dea9c677649f53b3e0eba824cbdc559810f196fb3c0a75

data/CHANGELOG.md

@@ -1,5 +1,13 @@
 # Mindee Ruby API Library Changelog
 
+## v4.7.0-rc2 - 2025-08-20
+### Changes
+* :sparkles: add missing accessors for PDF fixing options in `LocalInputSource`
+### Fixes
+* :recycle: update existing PDF fixing syntax
+* :memo: fix typos & documentation
+
+
 ## v4.7.0-rc1 - 2025-08-13
 ### Changes
 * :sparkles: add support for client V2 & associated features

data/docs/code_samples/default_v2.txt

@@ -8,10 +8,10 @@ model_id = 'MY_MODEL_ID'
 mindee_client = Mindee::ClientV2.new(api_key: api_key)
 
 # Set inference parameters
-params = Mindee::Input::InferenceParameters.new(
+inference_params = Mindee::Input::InferenceParameters.new(
     # ID of the model, required.
     model_id,
-    # If set to `True`, will enable Retrieval-Augmented Generation.
+    # If set to `true`, will enable Retrieval-Augmented Generation.
     rag: false,
 )
 
@@ -21,7 +21,7 @@ input_source = Mindee::Input::Source::PathInputSource.new(input_path)
 # Send for processing
 response = mindee_client.enqueue_and_get_inference(
     input_source,
-    params # Note: this parameter can also be provided as a Hash.
+    inference_params # Note: this parameter can also be provided as a Hash.
 )
 
 # Print a brief summary of the parsed data

data/lib/mindee/errors/mindee_input_error.rb

@@ -16,7 +16,7 @@ module Mindee
       # @param mime_type [String]
       def initialize(mime_type)
         @invalid_mimetype = mime_type
-        super("'#{@invalid_mimetype}' mime type not allowed, must be one of" \
+        super("'#{@invalid_mimetype}' mime type not allowed, must be one of " \
               "#{Mindee::Input::Source::ALLOWED_MIME_TYPES.join(', ')}")
       end
     end

data/lib/mindee/image/image_extractor.rb

@@ -18,12 +18,10 @@ module Mindee
       # @return [Origami::PDF] A PdfDocument handle.
       def self.attach_image_as_new_file(input_buffer, format: 'jpg')
         magick_image = MiniMagick::Image.read(input_buffer)
-        # NOTE: some jpeg images get rendered as three different versions of themselves per output if the format isn't
-        # converted.
+        # NOTE: We force format consolidation to a single format to avoid frames being interpreted as the final output.
         magick_image.format(format)
         original_density = magick_image.resolution
-        scale_factor = original_density[0].to_f / 4.166666 # No clue why the resolution needs to be reduced for
-        # the pdf otherwise the resulting image shrinks.
+        scale_factor = original_density[0].to_f / 4.166666 # Convert from default 300 DPI to 72.
         magick_image.format('pdf', 0, { density: scale_factor.to_s })
         Origami::PDF.read(StringIO.new(magick_image.to_blob))
       end
@@ -32,8 +30,7 @@ module Mindee
       #
       # @param [Input::Source::LocalInputSource] input_source
       # @param [Integer] page_id ID of the Page to extract from.
-      # @param [Array<Array<Geometry::Point>>, Array<Geometry::Quadrilateral>] polygons List of coordinates.
-      # to extract.
+      # @param [Array<Array<Geometry::Point>>, Array<Geometry::Quadrilateral>] polygons List of coordinates to extract.
       # @return [Array<Image::ExtractedImage>] Extracted Images.
       def self.extract_multiple_images_from_source(input_source, page_id, polygons)
         new_stream = load_input_source_pdf_page_as_stringio(input_source, page_id)

data/lib/mindee/input/local_response.rb

@@ -69,10 +69,10 @@ module Mindee
       end
 
       # Deserializes a loaded response
-      # @param response_class [Parsing::V2::JobResponse, Parsing::V2::InferenceResponse] class to return.
+      # @param response_class [Parsing::V2::CommonResponse] class to return.
       # @return [Parsing::V2::JobResponse, Parsing::V2::InferenceResponse]
       def deserialize_response(response_class)
-        response_class.new(as_hash)
+        response_class.new(as_hash) # : Parsing::V2::JobResponse | Parsing::V2::InferenceResponse
       rescue StandardError
         raise Errors::MindeeInputError, 'Invalid response provided.'
       end

data/lib/mindee/input/sources/local_input_source.rb

@@ -47,8 +47,7 @@ module Mindee
           end
 
           if filename.end_with?('.pdf') && repair_pdf
-            rescue_broken_pdf(@io_stream)
-            @file_mimetype = Marcel::MimeType.for @io_stream
+            fix_pdf!
 
             logger.debug("Loaded new input #{@filename} from #{self.class}")
             return if ALLOWED_MIME_TYPES.include? @file_mimetype
@@ -57,27 +56,40 @@ module Mindee
           raise Errors::MindeeMimeTypeError, @file_mimetype.to_s
         end
 
-        # Attempts to fix pdf files if mimetype is rejected.
-        # "Broken PDFs" are often a result of third-party injecting invalid headers.
-        # This attempts to remove them and send the file
-        # @param stream [StringIO, File]
-        def rescue_broken_pdf(stream)
-          stream.gets('%PDF-')
-          raise Errors::MindeePDFError if stream.eof? || stream.pos > 500
-
-          stream.pos = stream.pos - 5
-          data = stream.read
-          @io_stream.close
-
-          @io_stream = StringIO.new
-          @io_stream << data
+        # @deprecated See {#fix_pdf!} or {#self.fix_pdf} instead.
+        def rescue_broken_pdf(_)
+          fix_pdf!
         end
 
-        # Shorthand for pdf mimetype validation.
+        # Shorthand for PDF mimetype validation.
         def pdf?
           @file_mimetype.to_s == 'application/pdf'
         end
 
+        # Attempts to fix the PDF data in the file.
+        # @param maximum_offset [Integer] Maximum offset to look for the PDF header.
+        # @return [void]
+        # @raise [Mindee::Errors::MindeePDFError]
+        def fix_pdf!(maximum_offset: 500)
+          @io_stream = LocalInputSource.fix_pdf(@io_stream, maximum_offset: maximum_offset)
+          @io_stream.rewind
+          @file_mimetype = Marcel::MimeType.for @io_stream
+        end
+
+        # Attempt to fix the PDF data in the given stream.
+        # @param stream [StringIO] The stream to fix.
+        # @param maximum_offset [Integer] Maximum offset to look for the PDF header.
+        # @return [StringIO] The fixed stream.
+        # @raise [Mindee::Errors::MindeePDFError]
+        def self.fix_pdf(stream, maximum_offset: 500)
+          out_stream = StringIO.new
+          stream.gets('%PDF-')
+          raise Errors::MindeePDFError if stream.eof? || stream.pos > maximum_offset
+
+          stream.pos = stream.pos - 5
+          out_stream << stream.read
+        end
+
         # Cuts a PDF file according to provided options.
         # @param options [PageOptions, nil] Page cutting/merge options:
         #
@@ -166,8 +178,8 @@ module Mindee
           @io_stream.rewind
         end
 
-        # Checks whether the file has source text if it is a pdf. False otherwise
-        # @return [bool] True if the file is a PDF and has source text.
+        # Checks whether the file has source text if it is a pdf. `false` otherwise
+        # @return [bool] `true` if the file is a PDF and has source text.
         def source_text?
           Mindee::PDF::PDFTools.source_text?(@io_stream)
         end

data/lib/mindee/parsing/universal/universal_object_field.rb

@@ -78,7 +78,7 @@ module Mindee
         #
         # @param method_name [Symbol] The name of the method being checked.
         # @param include_private [bool] Whether to include private methods in the check.
-        # @return [bool] True if the method can be responded to, false otherwise.
+        # @return [bool] `true` if the method can be responded to, false otherwise.
         def respond_to_missing?(method_name, include_private = false)
           @all_values.key?(method_name.to_s) || super
         end

data/lib/mindee/parsing/v2/field/field_confidence.rb

@@ -40,25 +40,25 @@ module Mindee
           end
 
           # Check if this is a certain confidence level.
-          # @return [Boolean] True if confidence is certain.
+          # @return [Boolean] `true` if confidence is certain.
           def certain?
             @value == CERTAIN
           end
 
           # Check if this is a high confidence level.
-          # @return [Boolean] True if confidence is high.
+          # @return [Boolean] `true` if confidence is high.
           def high?
             @value == HIGH
           end
 
           # Check if this is a medium confidence level.
-          # @return [Boolean] True if confidence is medium.
+          # @return [Boolean] `true` if confidence is medium.
           def medium?
             @value == MEDIUM
           end
 
           # Check if this is a low confidence level.
-          # @return [Boolean] True if confidence is low.
+          # @return [Boolean] `true` if confidence is low.
           def low?
             @value == LOW
           end
@@ -71,7 +71,7 @@ module Mindee
 
           # Compare two FieldConfidence instances.
           # @param other [FieldConfidence] The other confidence to compare.
-          # @return [Boolean] True if they have the same value.
+          # @return [Boolean] `true` if they have the same value.
           def ==(other)
             other.is_a?(FieldConfidence) && @value == other.value
           end

data/lib/mindee/parsing/v2/field/inference_fields.rb

@@ -43,7 +43,7 @@ module Mindee
 
           # Check if method_missing should handle the method.
           # @param method_name [Symbol] The method name.
-          # @return [Boolean] True if the method should be handled.
+          # @return [Boolean] `true` if the method should be handled.
           def respond_to_missing?(method_name, include_private = false)
             key?(method_name.to_s) || super
           end

data/lib/mindee/parsing/v2/field/list_field.rb

@@ -49,7 +49,7 @@ module Mindee
           end
 
           # Check if the list is empty.
-          # @return [Boolean] True if the list has no items.
+          # @return [Boolean] `true` if the list has no items.
           def empty?
             @items.empty?
           end
@@ -74,7 +74,7 @@ module Mindee
           end
 
           # Iterator for Enumerator inheritance.
-          # NOTE: Untyped due to incomplete support in steep.
+          # NOTE: Untyped due to incomplete support in current supported version of RBS.
           def each(&block)
             return to_enum(:each) unless block_given?
 

data/lib/mindee/parsing/v2/field/object_field.rb

@@ -91,7 +91,7 @@ module Mindee
 
           # Check if method_missing should handle the method.
           # @param method_name [Symbol] The method name.
-          # @return [Boolean] True if the method should be handled.
+          # @return [Boolean] `true` if the method should be handled.
           def respond_to_missing?(method_name, include_private = false)
             @fields.respond_to?(method_name) || super
           end

data/lib/mindee/parsing/v2/field/simple_field.rb

@@ -26,7 +26,6 @@ module Mindee
             if @value.is_a?(TrueClass) || @value.is_a?(FalseClass)
               @value ? 'True' : 'False'
             elsif @value.is_a?(Integer) || @value.is_a?(Float)
-              # NOTE: explicitly typing because steep is very, very dumb
               num = @value # @type var num: Integer | Float
               format_numeric_value(num)
             else

data/lib/mindee/pdf/pdf_tools.rb

@@ -44,7 +44,7 @@ module Mindee
       # Checks a PDFs stream content for text operators
       # See https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/PDF32000_2008.pdf page 243-251.
       # @param [StringIO] stream Stream object from a PDFs page.
-      # @return [bool] True if a text operator is found in the stream.
+      # @return [bool] `true` if a text operator is found in the stream.
       def self.stream_has_text?(stream)
         data = stream.data
         return false if data.nil? || data.empty?
@@ -55,7 +55,7 @@ module Mindee
 
       # Checks whether the file has source_text. Sends false if the file isn't a PDF.
       # @param [StringIO] pdf_data Abinary-encoded stream representing the PDF file.
-      # @return [bool] True if the pdf has source text, false otherwise.
+      # @return [bool] `true` if the pdf has source text, false otherwise.
       def self.source_text?(pdf_data)
         begin
           pdf_data.rewind

data/lib/mindee/version.rb

@@ -3,7 +3,7 @@
 # Mindee
 module Mindee
   # Current version.
-  VERSION = '4.7.0-rc1'
+  VERSION = '4.7.0-rc2'
 
   # Finds and return the current platform.
   # @return [Symbol, Hash[String | Symbol, Regexp], Nil?]

data/sig/custom/mini_magick.rbs

@@ -1,5 +1,5 @@
 # Stub for the mini_magick library.
-# Note: though typing annotations for the MiniMagick library now exist, it seems that they aren't strict enough
+# NOTE: though typing annotations for the MiniMagick library now exist, it seems that they aren't strict enough
 # to match the rules we have on the repo, hence the existence of this file and the overrides present below.
 module MiniMagick
   class Image

data/sig/mindee/input/local_response.rbs

@@ -8,7 +8,7 @@ module Mindee
       def self.process_secret_key: (String) -> String
       def get_hmac_signature: (String) -> String
       def valid_hmac_signature?: (String, String) -> bool
-      def deserialize_response: (singleton(Parsing::V2::JobResponse) | singleton(Parsing::V2::InferenceResponse))-> (Parsing::V2::JobResponse | Parsing::V2::InferenceResponse)
+      def deserialize_response: (singleton(Parsing::V2::CommonResponse))-> (Parsing::V2::JobResponse | Parsing::V2::InferenceResponse)
     end
   end
 end

data/sig/mindee/input/sources/local_input_source.rbs

@@ -8,10 +8,14 @@ module Mindee
         attr_reader filename: String
         attr_reader io_stream: StringIO | File
         def initialize: (StringIO | File, String, ?repair_pdf: bool) -> void
+
+        def fix_pdf!: (?maximum_offset: Integer) -> void
+        def self.fix_pdf: (StringIO | File, ?maximum_offset: Integer) -> StringIO
+
         def logger: () -> Logger
 
 
-        def rescue_broken_pdf: (StringIO | File) -> (StringIO | File)
+        def rescue_broken_pdf: (untyped) -> void
         def pdf?: -> bool
         def apply_page_options: (PageOptions) -> StringIO?
         def process_pdf: (PageOptions) -> StringIO?

data/sig/mindee/parsing/v2/field/list_field.rbs

@@ -12,7 +12,6 @@ module Mindee
           def empty?: -> bool
           def size: -> Integer
           def length: -> Integer
-          # NOTE: Steep is incapable of handling typing of `each` when multiple types are used.
           def each: () { (untyped) -> untyped } -> untyped
           def []: (Integer) -> (BaseField)
         end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: mindee
 version: !ruby/object:Gem::Version
-  version: 4.7.0.pre.rc1
+  version: 4.7.0.pre.rc2
 platform: ruby
 authors:
 - Mindee, SA
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-08-13 00:00:00.000000000 Z
+date: 2025-08-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: base64