fontisan 0.2.22 → 0.2.23

data/lib/fontisan/models/audit/library_summary.rb

@@ -1,40 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # Aggregate view over a directory (tree) of audited fonts.
-      #
-      # Built by {Audit::LibraryAuditor}. Combines a flat list of
-      # per-face {AuditReport}s with derived cross-face rollups:
-      # script coverage matrix, duplicate detection (by source_sha256),
-      # and license distribution. Lets a librarian inventory a font
-      # collection in one pass.
-      class LibrarySummary < Lutaml::Model::Serializable
-        attribute :root_path, :string
-        attribute :total_files, :integer
-        attribute :total_faces, :integer
-        attribute :scanned_extensions, :string, collection: true
-        attribute :aggregate_metrics, :hash
-        attribute :script_coverage, ScriptCoverageRow, collection: true
-        attribute :duplicate_groups, DuplicateGroup, collection: true
-        attribute :license_distribution, :hash
-        attribute :per_face_reports, AuditReport, collection: true
-
-        key_value do
-          map "root_path", to: :root_path
-          map "total_files", to: :total_files
-          map "total_faces", to: :total_faces
-          map "scanned_extensions", to: :scanned_extensions
-          map "aggregate_metrics", to: :aggregate_metrics
-          map "script_coverage", to: :script_coverage
-          map "duplicate_groups", to: :duplicate_groups
-          map "license_distribution", to: :license_distribution
-          map "per_face_reports", to: :per_face_reports
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit/licensing.rb

@@ -1,48 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # Licensing + embedding + vendor provenance fields for a face.
-      #
-      # Combines the human-readable legal/identity fields from the name
-      # table with the machine-readable embedding permissions from OS/2.
-      # Type 1 fonts have no OS/2 — callers must tolerate a nil
-      # embedding_type / fs_selection_flags / vendor_id.
-      class Licensing < Lutaml::Model::Serializable
-        # Name-table fields (English name IDs)
-        attribute :copyright,          :string
-        attribute :trademark,          :string
-        attribute :manufacturer,       :string
-        attribute :designer,           :string
-        attribute :description,        :string
-        attribute :vendor_url,         :string
-        attribute :designer_url,       :string
-        attribute :license_description, :string
-        attribute :license_url, :string
-
-        # OS/2 fields
-        attribute :vendor_id,           :string
-        attribute :embedding_type,      :string
-        attribute :fs_selection_flags,  :string, collection: true
-
-        key_value do
-          map "copyright",           to: :copyright
-          map "trademark",           to: :trademark
-          map "manufacturer",        to: :manufacturer
-          map "designer",            to: :designer
-          map "description",         to: :description
-          map "vendor_url",          to: :vendor_url
-          map "designer_url",        to: :designer_url
-          map "license_description", to: :license_description
-          map "license_url",         to: :license_url
-          map "vendor_id",           to: :vendor_id
-          map "embedding_type",      to: :embedding_type
-          map "fs_selection_flags",  to: :fs_selection_flags
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit/metrics.rb

@@ -1,111 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # Layout-critical metrics for a face, consolidated from head, hhea,
-      # OS/2, and post tables. Designers and engineers can read all
-      # spacing-relevant numbers in one place instead of cross-referencing
-      # raw table dumps.
-      #
-      # All fields are nil-safe — Type 1 fonts and stripped WOFF builds
-      # may not carry every table. Derived booleans (e.g. metrics_consistent?)
-      # tolerate nil inputs and return false rather than raising.
-      class Metrics < Lutaml::Model::Serializable
-        # head
-        attribute :units_per_em, :integer
-        attribute :bbox_x_min,   :integer
-        attribute :bbox_y_min,   :integer
-        attribute :bbox_x_max,   :integer
-        attribute :bbox_y_max,   :integer
-
-        # hhea (horizontal)
-        attribute :hhea_ascent,  :integer
-        attribute :hhea_descent, :integer
-        attribute :hhea_line_gap, :integer
-
-        # OS/2 typo
-        attribute :typo_ascender, :integer
-        attribute :typo_descender, :integer
-        attribute :typo_line_gap, :integer
-
-        # OS/2 win
-        attribute :win_ascent,  :integer
-        attribute :win_descent, :integer
-
-        # OS/2 v2+ (optional)
-        attribute :x_height,   :integer
-        attribute :cap_height, :integer
-
-        # OS/2 subscript/superscript
-        attribute :subscript_x_size,    :integer
-        attribute :subscript_y_size,    :integer
-        attribute :subscript_x_offset,  :integer
-        attribute :subscript_y_offset,  :integer
-        attribute :superscript_x_size,    :integer
-        attribute :superscript_y_size,    :integer
-        attribute :superscript_x_offset,  :integer
-        attribute :superscript_y_offset,  :integer
-
-        # OS/2 strikeout
-        attribute :strikeout_size,     :integer
-        attribute :strikeout_position, :integer
-
-        # post underline
-        attribute :underline_position,  :float
-        attribute :underline_thickness, :float
-
-        key_value do
-          map "units_per_em",     to: :units_per_em
-          map "bbox_x_min",       to: :bbox_x_min
-          map "bbox_y_min",       to: :bbox_y_min
-          map "bbox_x_max",       to: :bbox_x_max
-          map "bbox_y_max",       to: :bbox_y_max
-
-          map "hhea_ascent",   to: :hhea_ascent
-          map "hhea_descent",  to: :hhea_descent
-          map "hhea_line_gap", to: :hhea_line_gap
-
-          map "typo_ascender",  to: :typo_ascender
-          map "typo_descender", to: :typo_descender
-          map "typo_line_gap",  to: :typo_line_gap
-
-          map "win_ascent",  to: :win_ascent
-          map "win_descent", to: :win_descent
-
-          map "x_height",   to: :x_height
-          map "cap_height", to: :cap_height
-
-          map "subscript_x_size",   to: :subscript_x_size
-          map "subscript_y_size",   to: :subscript_y_size
-          map "subscript_x_offset", to: :subscript_x_offset
-          map "subscript_y_offset", to: :subscript_y_offset
-          map "superscript_x_size",   to: :superscript_x_size
-          map "superscript_y_size",   to: :superscript_y_size
-          map "superscript_x_offset", to: :superscript_x_offset
-          map "superscript_y_offset", to: :superscript_y_offset
-
-          map "strikeout_size",     to: :strikeout_size
-          map "strikeout_position", to: :strikeout_position
-
-          map "underline_position",  to: :underline_position
-          map "underline_thickness", to: :underline_thickness
-        end
-
-        # True when hhea ascent/descent match OS/2 typo ascent/descent.
-        # Mismatch is a common font bug that causes inconsistent line
-        # height across platforms.
-        #
-        # @return [Boolean]
-        def metrics_consistent?
-          return false if hhea_ascent.nil? || typo_ascender.nil?
-          return false if hhea_descent.nil? || typo_descender.nil?
-
-          hhea_ascent == typo_ascender && hhea_descent == typo_descender
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit/named_instance.rb

@@ -1,41 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # One fvar named instance (e.g. "Bold", "SemiCondensed").
-      #
-      # `coordinates` is serialized as a compact "tag=value,tag=value" string
-      # (e.g. "wght=700,wdth=100") for human readability. The AuditReport is
-      # primarily a human-facing artifact; downstream tooling that needs
-      # structured coordinates can re-derive them from fvar.
-      class NamedInstance < Lutaml::Model::Serializable
-        attribute :subfamily_name,  :string
-        attribute :postscript_name, :string
-        attribute :coordinates,     :string
-
-        key_value do
-          map "subfamily_name",  to: :subfamily_name
-          map "postscript_name", to: :postscript_name
-          map "coordinates",     to: :coordinates
-        end
-
-        # Build the coordinates string from a parallel array of axis tags
-        # and fvar coordinate values. Returns nil if either side is empty.
-        #
-        # @param axis_tags [Array<String>] ordered axis tags (e.g. ["wght", "wdth"])
-        # @param values [Array<Numeric>] ordered coordinate values
-        # @return [String, nil]
-        def self.format_coordinates(axis_tags, values)
-          return nil if axis_tags.nil? || values.nil?
-          return nil if axis_tags.empty? || values.empty?
-
-          pairs = axis_tags.zip(values).map { |tag, val| "#{tag}=#{val}" }
-          pairs.join(",")
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit/opentype_layout.rb

@@ -1,40 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # Structured OpenType layout summary for one face.
-      #
-      # Replaces the previous flat `opentype_scripts` + `features` pair
-      # on AuditReport for MECE cleanliness. Carries:
-      #
-      # - `scripts`: union of GSUB + GPOS script tags (sorted, unique).
-      # - `features`: union of GSUB + GPOS feature tags across every
-      #   script (sorted, unique).
-      # - `by_script`: per-script breakdown preserving the
-      #   "feature X is for script Y" relationship that the flat arrays
-      #   discarded.
-      # - `has_gsub` / `has_gpos`: presence flags so consumers can tell
-      #   "font has no layout" from "font has GSUB but no GPOS".
-      #
-      # nil for Type 1 fonts (no SFNT table structure).
-      class OpenTypeLayout < Lutaml::Model::Serializable
-        attribute :scripts,   :string,         collection: true
-        attribute :features,  :string,         collection: true
-        attribute :by_script, ScriptFeatures,  collection: true
-        attribute :has_gsub,  Lutaml::Model::Type::Boolean
-        attribute :has_gpos,  Lutaml::Model::Type::Boolean
-
-        key_value do
-          map "scripts",   to: :scripts
-          map "features",  to: :features
-          map "by_script", to: :by_script
-          map "has_gsub",  to: :has_gsub
-          map "has_gpos",  to: :has_gpos
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit/script_coverage_row.rb

@@ -1,26 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # One row in a LibrarySummary's script-coverage matrix.
-      #
-      # Lists every face (by postscript_name) whose cmap covers at least
-      # one codepoint assigned to a Unicode script. Lets a librarian
-      # answer "which fonts cover Cyrillic?" without re-auditing.
-      class ScriptCoverageRow < Lutaml::Model::Serializable
-        attribute :script, :string
-        attribute :face_count, :integer
-        attribute :faces, :string, collection: true
-
-        key_value do
-          map "script", to: :script
-          map "face_count", to: :face_count
-          map "faces", to: :faces
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit/script_features.rb

@@ -1,28 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # Per-script breakdown of OpenType features.
-      #
-      # Pairs a script tag (e.g. "latn", "kana ") with the GSUB features
-      # and GPOS features that apply to it. The two collections are
-      # kept separate because substitution and positioning have different
-      # semantics — consumers answering "does this font support kerning
-      # for Latin?" want to look at GPOS only.
-      class ScriptFeatures < Lutaml::Model::Serializable
-        attribute :script,         :string
-        attribute :gsub_features,  :string, collection: true
-        attribute :gpos_features,  :string, collection: true
-
-        key_value do
-          map "script",        to: :script
-          map "gsub_features", to: :gsub_features
-          map "gpos_features", to: :gpos_features
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit/variation_detail.rb

@@ -1,44 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Audit
-      # Variable-font detail for one face.
-      #
-      # Bundles everything fvar-derived (axes + named instances) with the
-      # presence flags for every variation side-table (avar/cvar/HVAR/VVAR/
-      # MVAR/gvar). Replaces the previous flat `axes` + `is_variable` pair
-      # on AuditReport for MECE cleanliness — a face is variable iff this
-      # object is non-nil.
-      #
-      # `axes` reuses the existing AuditAxis shape; `named_instances` is a
-      # parallel NamedInstance collection. The has_* booleans are presence
-      # checks only — they don't validate the table contents.
-      class VariationDetail < Lutaml::Model::Serializable
-        attribute :axes,            AuditAxis,      collection: true
-        attribute :named_instances, NamedInstance,  collection: true
-
-        # Variation side-table presence flags.
-        attribute :has_avar, Lutaml::Model::Type::Boolean # axis variation
-        attribute :has_cvar, Lutaml::Model::Type::Boolean # CVT variation
-        attribute :has_hvar, Lutaml::Model::Type::Boolean # horizontal metrics
-        attribute :has_vvar, Lutaml::Model::Type::Boolean # vertical metrics
-        attribute :has_mvar, Lutaml::Model::Type::Boolean # metrics variation
-        attribute :has_gvar, Lutaml::Model::Type::Boolean # glyph variation (TT)
-
-        key_value do
-          map "axes",            to: :axes
-          map "named_instances", to: :named_instances
-          map "has_avar", to: :has_avar
-          map "has_cvar", to: :has_cvar
-          map "has_hvar", to: :has_hvar
-          map "has_vvar", to: :has_vvar
-          map "has_mvar", to: :has_mvar
-          map "has_gvar", to: :has_gvar
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/audit.rb

@@ -1,33 +0,0 @@
-# frozen_string_literal: true
-
-# Namespace hub for audit-related models.
-#
-# All Models::Audit::* constants are autoloaded from here.
-
-module Fontisan
-  module Models
-    module Audit
-      autoload :AuditBlock,  "fontisan/models/audit/audit_block"
-      autoload :AuditAxis,   "fontisan/models/audit/audit_axis"
-      autoload :AuditDiff,   "fontisan/models/audit/audit_diff"
-      autoload :AuditReport, "fontisan/models/audit/audit_report"
-      autoload :CodepointRange, "fontisan/models/audit/codepoint_range"
-      autoload :CodepointSetDiff, "fontisan/models/audit/codepoint_set_diff"
-      autoload :ColorCapabilities, "fontisan/models/audit/color_capabilities"
-      autoload :DuplicateGroup, "fontisan/models/audit/duplicate_group"
-      autoload :EmbeddingType, "fontisan/models/audit/embedding_type"
-      autoload :FieldChange, "fontisan/models/audit/field_change"
-      autoload :FsSelectionFlags, "fontisan/models/audit/fs_selection_flags"
-      autoload :GaspRange, "fontisan/models/audit/gasp_range"
-      autoload :Hinting, "fontisan/models/audit/hinting"
-      autoload :LibrarySummary, "fontisan/models/audit/library_summary"
-      autoload :Licensing, "fontisan/models/audit/licensing"
-      autoload :Metrics, "fontisan/models/audit/metrics"
-      autoload :NamedInstance, "fontisan/models/audit/named_instance"
-      autoload :OpenTypeLayout, "fontisan/models/audit/opentype_layout"
-      autoload :ScriptCoverageRow, "fontisan/models/audit/script_coverage_row"
-      autoload :ScriptFeatures, "fontisan/models/audit/script_features"
-      autoload :VariationDetail, "fontisan/models/audit/variation_detail"
-    end
-  end
-end

data/lib/fontisan/models/ucd/ucd.rb

@@ -1,38 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Ucd
-      # Root <ucd> element of the UCDXML flat file.
-      #
-      # The flat UCDXML file has the structure:
-      #
-      #   <ucd>
-      #     <description>...</description>
-      #     <last_revision date="2025-..." version="17.0.0" />
-      #     <char cp="0000" .../>
-      #     <char cp="0001" .../>
-      #     ...
-      #     <char first-cp="3400" last-cp="4DBF" .../>
-      #     ...
-      #   </ucd>
-      #
-      # The flat variant merges all per-category UCD files (Blocks.txt,
-      # Scripts.txt, UnicodeData.txt, etc.) into one stream of <char>
-      # elements. Roughly 340,000 entries for Unicode 17.0.0.
-      class Ucd < Lutaml::Model::Serializable
-        attribute :last_revision, :string
-        attribute :chars, UcdChar, collection: true
-
-        xml do
-          element "ucd"
-
-          map_element "last_revision", to: :last_revision
-          map_element "char", to: :chars
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/ucd/ucd_char.rb

@@ -1,67 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    module Ucd
-      # Single <char> element from the UCDXML flat file.
-      #
-      # UCDXML uses two forms:
-      #   <char cp="0041" name="..." script="Latin" block="Basic Latin" .../>
-      #   <char first-cp="3400" last-cp="4DBF" name="..." script="Han" .../>
-      #
-      # The first form describes one codepoint. The second form describes a
-      # closed range of codepoints that share the same properties (used for
-      # CJK ideograph ranges where each codepoint would otherwise need its
-      # own <char> entry).
-      #
-      # Both forms can appear in the same document; `cp` is mutually
-      # exclusive with `first-cp`/`last-cp`.
-      class UcdChar < Lutaml::Model::Serializable
-        attribute :cp, :string
-        attribute :first_cp, :string
-        attribute :last_cp, :string
-        attribute :name, :string
-        attribute :general_category, :string
-        attribute :script, :string
-        attribute :block, :string
-        attribute :age, :string
-
-        xml do
-          element "char"
-
-          map_attribute "cp",               to: :cp
-          map_attribute "first-cp",         to: :first_cp
-          map_attribute "last-cp",          to: :last_cp
-          map_attribute "name",             to: :name
-          map_attribute "general-category", to: :general_category
-          map_attribute "script",           to: :script
-          map_attribute "block",            to: :block
-          map_attribute "age",              to: :age
-        end
-
-        # True if this entry describes a codepoint range rather than a
-        # single codepoint.
-        def range?
-          !first_cp.nil? && !last_cp.nil?
-        end
-
-        # The codepoints covered by this entry, as Integers.
-        # For a single-codepoint entry, returns a one-element array.
-        # For a range entry, returns the inclusive range as an array
-        # (caller should treat this lazily if the range is huge — CJK
-        # ranges can have tens of thousands of codepoints).
-        def codepoints
-          if range?
-            (first_cp.to_i(16)..last_cp.to_i(16)).to_a
-          elsif cp
-            [cp.to_i(16)]
-          else
-            []
-          end
-        end
-      end
-    end
-  end
-end

data/lib/fontisan/models/ucd.rb

@@ -1,19 +0,0 @@
-# frozen_string_literal: true
-
-require "lutaml/model"
-
-module Fontisan
-  module Models
-    # Namespace for UCDXML deserialization models.
-    #
-    # These classes deserialize the upstream UCDXML flat file
-    # (https://www.unicode.org/Public/<version>/ucdxml/ucd.all.flat.zip)
-    # into Ruby objects. They are used by Fontisan::Ucd::IndexBuilder to
-    # derive compact run-length-encoded indices for Unicode block and
-    # script lookup.
-    module Ucd
-      autoload :UcdChar, "fontisan/models/ucd/ucd_char"
-      autoload :Ucd,     "fontisan/models/ucd/ucd"
-    end
-  end
-end

data/lib/fontisan/ucd/aggregator.rb

@@ -1,73 +0,0 @@
-# frozen_string_literal: true
-
-module Fontisan
-  module Ucd
-    # Produces audit-ready aggregations from a codepoint list + UCD indices.
-    #
-    # Pure: no I/O, no side effects. Caller passes the codepoints and the
-    # blocks/scripts indices; Aggregator returns the aggregated summaries.
-    module Aggregator
-      module_function
-
-      # Aggregate codepoints per Unicode block.
-      #
-      # Returns one hash per overlapping block, sorted by first_cp:
-      #
-      #   { name:, first_cp:, last_cp:, total:, covered:, fill_ratio:, complete: }
-      #
-      # @param codepoints [Array<Integer>] sorted not required
-      # @param blocks_index [Index]
-      # @return [Array<Hash>]
-      def aggregate_blocks(codepoints, blocks_index)
-        sorted = codepoints.sort
-        return [] if sorted.empty?
-
-        coverage = Hash.new { |h, k| h[k] = 0 }
-        coverage.compare_by_identity
-        first_cp = sorted.first
-        last_cp = sorted.last
-
-        overlapping = blocks_index.each_overlapping(first_cp, last_cp).to_a
-        overlapping.each do |entry|
-          coverage[entry] = count_in_range(sorted, [entry.first_cp, entry.last_cp])
-        end
-
-        overlapping.map do |entry|
-          covered = coverage[entry]
-          total = entry.size
-          {
-            name: entry.name,
-            first_cp: entry.first_cp,
-            last_cp: entry.last_cp,
-            total: total,
-            covered: covered,
-            fill_ratio: covered.fdiv(total).round(4),
-            complete: covered == total,
-          }
-        end
-      end
-
-      # Aggregate unique script names from codepoints.
-      #
-      # @param codepoints [Array<Integer>]
-      # @param scripts_index [Index]
-      # @return [Array<String>] sorted unique script names
-      def aggregate_scripts(codepoints, scripts_index)
-        scripts = codepoints.filter_map { |cp| scripts_index.lookup(cp) }
-        scripts.uniq.sort
-      end
-
-      # Count codepoints in `sorted` that fall within [first, last].
-      # `sorted` must be sorted ascending.
-      def count_in_range(sorted, range)
-        first, last = range
-        left = sorted.bsearch_index { |cp| cp >= first } || sorted.size
-        return 0 if left == sorted.size
-
-        right = sorted.bsearch_index { |cp| cp > last } || sorted.size
-        right - left
-      end
-      private_class_method :count_in_range
-    end
-  end
-end