acrofill 0.1.2 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8192b1e65d4be56876296608eecae9371866700b8a2016d6bbedad79062fca33
-  data.tar.gz: 750c2b6ccd33d9e335933c6f5e5f906f0ca8adb0f5aa07e98a4eb6019104c0f9
+  metadata.gz: d7b9246e478c6ab2f8206ba44e766a2b4e9e9a946a48245d171d7e212bac54d2
+  data.tar.gz: e6f70298a5008e1419089328788cb1d66908cd3e2edfa2e3bd724e0a220ff176
 SHA512:
-  metadata.gz: c081d29edbd239103797267d4f47cf2ed18ef3fa46026acae8ffc03b8d27e1b13e9477bba89c7646a554774059905bd833cd7d5df41717d95eb90ab72286d0f6
-  data.tar.gz: a0b575761f075a4fc8f7690b85ec7addee099fcf5ef29ff5f1201e72ee47fd873398645c2dd51af07476073f3cb85f8cbc7570b46af158900372201c60841098
+  metadata.gz: 4f858c6d8e34cf948d8ef554d31d9e0a7187408f256fadff63c15bd4af3ad6ad6e61329770e9db506482f179635b9305393d432a9176742893aa45643810f22c
+  data.tar.gz: 75b8174ebb610ec9be61a7b214a69a47071d5aa797bb13378917eb3d41b4279c9ee81e1348641703dd29e5436ca1dcda30148b3e50c70759b2d47fad8c137cb1

data/CHANGELOG.md

@@ -7,6 +7,94 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.3.0] - 2026-07-27
+
+### Fixed
+
+- Single-line baselines now match pdftk when the text is taller than the
+  field, which real forms hit routinely (a 12pt `/DA` in a 10.8pt-high box
+  is common). Centering is bounded on both sides: the ascender is kept
+  inside the box and the baseline never drops below the box floor. This was
+  the last geometry difference on a 551-field sample of real templates.
+- Text is now measured with the metrics of the font it is actually drawn
+  with. A template that embeds its own face declares `/Widths` and a
+  `/FontDescriptor`, and those were ignored in favour of standard-14
+  tables — which put centered and right-aligned values as much as tens of
+  points away from where pdftk puts them (43pt on a 300pt-wide field in
+  one measured case). `/Widths` now drives glyph widths, `/FontDescriptor`
+  `/Ascent` the baseline and its `/FontBBox` the multiline row spacing,
+  falling back to the standard-14 tables only when the dictionary is
+  silent. This is the layout most real-world forms hit, since almost all
+  of them embed a subset face.
+
+### Changed
+
+- Vertical geometry is now font-aware and matches pdftk-java 3.3.3 exactly.
+  Baselines are placed from the font's own AFM ascender instead of a fixed
+  Helvetica value (Times sat 0.18pt low, Courier 0.45pt), and multiline rows
+  are spaced by the font's `FontBBox` extent with pdftk's 1pt top offset
+  instead of a flat `1.15 * size` leading.
+- `Acrofill::Metrics` exposes `.font_for`, returning widths plus ascender,
+  descender and `FontBBox` for one of the twelve standard-14 text cuts.
+  Vertical metrics are stored per cut, since Courier-Bold and Times-Italic
+  differ there even where their widths do not.
+- Font resolution moved out of `Appearance` into `Acrofill::Fonts`, which
+  owns the `/DR /Font` dictionary: metrics for a `/DA` resource name and
+  the reference a generated appearance points at.
+
+### Added
+
+- `benchmark/geometry_diff.rb` compares acrofill's appearance streams with
+  pdftk's field by field on your own templates.
+- `spec/pdftk_parity_spec.rb` pins alignment, baseline and multiline row
+  geometry — for standard-14 faces, template-supplied metrics, and text
+  taller than its field — to numbers measured from pdftk's own output.
+
+### Known differences from pdftk
+
+- Auto-sized fields (`0 Tf`): pdftk picks a font-dependent size that fills
+  the box (16.33pt in a 20pt box for Helvetica, and 20.73pt — taller than
+  the box — for Courier), with a hard 4pt floor. Acrofill keeps its own
+  `min(height * 0.66, 12pt)` and is not going to reproduce that.
+- Values too wide for the field: acrofill shrinks the font to fit, pdftk
+  keeps the size and clips.
+- Non-ASCII values: pdftk writes UTF-8 bytes into a `/WinAnsiEncoding`
+  font, which renders as mojibake; acrofill writes Windows-1252, so the
+  text is correct and the measured width differs accordingly.
+
+## [0.2.0] - 2026-07-27
+
+### Fixed
+
+- Pages whose `/Type` entry is missing are no longer skipped: flattening
+  such a document silently produced an unflattened file.
+- A widget whose appearance cannot be regenerated (degenerate `/Rect`)
+  now has its `/AP` dropped instead of keeping the appearance of the
+  *previous* value while `/V` already holds the new one.
+- A page `/Resources` or `/Resources /XObject` that is not a dictionary
+  no longer raises `TypeError` while flattening, and a `/Root` that is
+  missing or not a dictionary raises `Acrofill::Error` instead of
+  `NoMethodError`.
+- Filling a field whose fully-qualified name is shared by several field
+  dictionaries no longer depends on the return value of the first fill.
+
+### Changed
+
+- Text metrics now cover all standard-14 text cuts (Times bold/italic,
+  the Courier family, the oblique Helvetica cuts) over the full
+  WinAnsiEncoding range instead of ASCII-only Helvetica/Courier/
+  Times-Roman, so accented characters and bold or serif faces are
+  measured rather than approximated. Widths are WinAnsi-correct: `'`
+  measured 222 (StandardEncoding `quoteright`) where appearances
+  actually emit `quotesingle`.
+- Unknown `/BaseFont` names (`ArialMT`, `TimesNewRomanPS-BoldMT`, subset
+  faces) are classified by family and weight instead of all falling back
+  to Helvetica.
+- One fallback Helvetica font object is now shared by every generated
+  appearance; a 50-field form previously wrote 50 identical font
+  dictionaries.
+- Flattening moved out of `Form` into its own `Acrofill::Flattener`.
+
 ## [0.1.2] - 2026-07-24
 
 ### Added
@@ -54,7 +142,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - `PdfForms`-compatible entry points (`Acrofill.new`, `fill_form`, `fields`,
   `field_names`).
 
-[Unreleased]: https://github.com/stiig/acrofill/compare/v0.1.2...HEAD
+[Unreleased]: https://github.com/stiig/acrofill/compare/v0.3.0...HEAD
+[0.3.0]: https://github.com/stiig/acrofill/compare/v0.2.0...v0.3.0
+[0.2.0]: https://github.com/stiig/acrofill/compare/v0.1.2...v0.2.0
 [0.1.2]: https://github.com/stiig/acrofill/compare/v0.1.1...v0.1.2
 [0.1.1]: https://github.com/stiig/acrofill/compare/v0.1.0...v0.1.1
 [0.1.0]: https://github.com/stiig/acrofill/releases/tag/v0.1.0

data/README.md

@@ -75,6 +75,28 @@ filler.fill_form(tpl, out, data, flatten: true)
 
 Unknown field names are silently ignored, matching pdftk.
 
+### Geometry parity
+
+Filled text lands where pdftk puts it: across four real-world
+government claim forms — 551 filled widgets — every appearance agrees with
+pdftk-java 3.3.3 to the two decimals it prints, except where acrofill
+deliberately differs (below). `spec/pdftk_parity_spec.rb` pins that
+placement against numbers read out of pdftk's own appearance streams —
+alignment, baselines for Helvetica/Times/Courier across box heights, text
+taller than its field, multiline row spacing, and the widths, ascent and
+`FontBBox` a template's own font dictionary supplies. Check it against
+your own templates with:
+
+```bash
+ruby benchmark/geometry_diff.rb path/to/form.pdf
+```
+
+Three differences are deliberate: acrofill shrinks an overlong value to fit
+where pdftk clips it, it writes Windows-1252 for non-ASCII values where
+pdftk emits UTF-8 bytes into a WinAnsi font (which renders as mojibake), and
+it caps auto-sized (`0 Tf`) text at 12pt where pdftk scales it to fill the
+box. See the changelog for the measured numbers.
+
 ## Performance
 
 Because Acrofill runs in-process, it avoids the JVM (or C++ process)
@@ -147,9 +169,10 @@ corruption or code execution. Specifically it defends against:
   name and only well-formed colour operators (`g`/`rg`/`k`) are copied
   into generated appearances. Field **values** are always written as
   escaped literals / hex strings, never as operators.
-- **Malformed scalars** — non-finite reals, deeply nested arrays, and
-  mistyped dictionaries are clamped or rejected rather than crashing the
-  writer.
+- **Malformed scalars and mistyped structure** — non-finite reals, deeply
+  nested arrays, a `/Resources` that is not a dictionary, a `/Root` that
+  points nowhere: each is clamped, replaced or rejected rather than
+  raising a `TypeError`/`NoMethodError` out of the middle of a fill.
 - **Encrypted / unparseable input** — rejected up front; every failure
   at the parse boundary surfaces as `Acrofill::Error` (including lazy
   per-object parse errors and parser stack overflow on pathologically
@@ -162,8 +185,13 @@ Supported:
 - Text fields (`/Tx`) — hierarchical names (`parent.kid`), inherited
   `/DA`, alignment via `/Q` (left/center/right), auto font size (`0 Tf`),
   shrink-to-fit for overflowing values, multiline fields (`/Ff` bit 13)
-  with word wrapping, standard-14 width metrics
-  (Helvetica/Courier/Times).
+  with word wrapping, and text measured with the metrics of the font it is
+  drawn with: a template's own `/Widths` and `/FontDescriptor` when the
+  face carries them (as embedded subsets do), otherwise WinAnsi tables for
+  all standard-14 cuts — widths, ascender and `FontBBox`, so baselines and
+  row spacing follow the actual face. A `/BaseFont` naming no standard cut
+  (`ArialMT`, `TimesNewRomanPS-BoldMT`) is classified by family and weight
+  rather than all measured as Helvetica.
 - Checkboxes and radio groups (`/Btn`) — state selection via `/V`+`/AS`
   using the template's own appearance states.
 - Choice fields (`/Ch`) — value set and rendered like text.

data/lib/acrofill/appearance.rb

@@ -5,14 +5,15 @@ module Acrofill
   # text-field widget, honouring the field's /DA string and /Q alignment.
   class Appearance
     PADDING = 2.0
-    ASCENT = 0.718 # Helvetica cap-height-ish ascent, em fractions
-    DESCENT = 0.207
+    # pdftk offsets the first multiline row by 1pt from the box top.
+    TOP_OFFSET = 1.0
     # Colour-setting operators allowed in a /DA string, and their operand counts.
     COLOR_OP_ARITY = { 'g' => 1, 'rg' => 3, 'k' => 4 }.freeze
 
     def initialize(doc, acroform)
       @doc = doc
       @acroform = acroform
+      @fonts = Fonts.new(doc, acroform)
     end
 
     # Returns a Reference to the new appearance XObject, or nil when the
@@ -26,22 +27,19 @@ module Acrofill
       return nil if width <= 0 || height <= 0
 
       font_name, size, color_ops = parse_da(field_node)
-      base_font = base_font_for(font_name)
+      font = @fonts.metrics(font_name)
       align = alignment(field_node)
 
       body =
         if multiline
           size = 12.0 if size <= 0
           size = size.clamp(2.0, 144.0)
-          multiline_body(value, base_font, size, width, height, align)
+          multiline_body(value, font, size, width, height, align)
         else
           text = printable_text(value)
           size = [height * 0.66, 12.0].min if size.zero?
-          size = shrink_to_fit(text, base_font, size, width)
-          # Vertically center the ascent box, matching pdftk's baseline
-          # placement exactly: ty = (h - ascent*size) / 2.
-          ty = [(height - (size * ASCENT)) / 2.0, size * DESCENT].max
-          "#{fmt(line_x(text, base_font, size, width, align))} #{fmt(ty)} Td\n" \
+          size = shrink_to_fit(text, font, size, width)
+          "#{fmt(line_x(text, font, size, width, align))} #{fmt(baseline(height, font, size))} Td\n" \
             "(#{escape_literal(text)}) Tj\n"
         end
 
@@ -56,7 +54,7 @@ module Acrofill
         Subtype: :Form,
         FormType: 1,
         BBox: [0, 0, width, height],
-        Resources: { Font: { font_name.to_sym => font_ref(font_name) } }
+        Resources: { Font: { font_name.to_sym => @fonts.reference(font_name) } }
       }
       @doc.add(StreamObject.new(dict, content.b))
     end
@@ -67,14 +65,27 @@ module Acrofill
       Serializer.format_number(num.to_f)
     end
 
+    # Where pdftk puts a single line's baseline: the ascent box centered in
+    # the field, but never so low that the descender leaves the box, never so
+    # high that the ascender does, and never below the box floor. The last
+    # two only bind when the text is taller than the field it sits in, which
+    # real forms do hit — a 12pt /DA in a 10.8pt-high field is common.
+    def baseline(height, font, size)
+      ascent = font.ascent(size)
+      highest = height - ascent # baseline putting the ascender at the box top
+      return 0.0 if highest <= 0 # the glyphs are taller than the box
+
+      [highest / 2.0, font.descent(size)].max.clamp(0.0, highest)
+    end
+
     # /Q (0 left, 1 center, 2 right), inheritable and possibly indirect.
     def alignment(field_node)
       align = @doc.deref(@doc.inherited_value(field_node, :Q) || @acroform[:Q])
       align.is_a?(Integer) ? align : 0
     end
 
-    def line_x(text, base_font, size, width, align)
-      text_width = Metrics.string_width(text, base_font, size)
+    def line_x(text, font, size, width, align)
+      text_width = Metrics.string_width(text, font.widths, size)
       case align
       when 1 then [(width - text_width) / 2.0, PADDING].max
       when 2 then [width - PADDING - text_width, PADDING].max
@@ -84,18 +95,20 @@ module Acrofill
 
     # Greedy word wrap, top-down, honouring explicit line breaks. Lines
     # that would fall below the box are clipped by the BBox.
-    def multiline_body(value, base_font, size, width, height, align)
+    def multiline_body(value, font, size, width, height, align)
       max_width = width - (2 * PADDING)
       lines = value.to_s.split(/\r\n|[\r\n]/).flat_map do |paragraph|
-        wrap_line(printable_text(paragraph), base_font, size, max_width)
+        wrap_line(printable_text(paragraph), font, size, max_width)
       end
 
-      leading = size * 1.15
-      first_y = height - PADDING - (size * ASCENT)
+      # pdftk spaces rows by the font's FontBBox extent and drops the first
+      # baseline by that extent from the top of the box.
+      leading = font.line_height(size)
+      first_y = height - font.top(size) + TOP_OFFSET
       body = "#{fmt(leading)} TL\n"
       previous_x = 0.0
       lines.each_with_index do |line, index|
-        x = line_x(line, base_font, size, width, align)
+        x = line_x(line, font, size, width, align)
         body << "#{fmt(x - previous_x)} #{index.zero? ? fmt(first_y) : '0'} Td\n"
         body << "(#{escape_literal(line)}) Tj\nT*\n"
         previous_x = x
@@ -105,21 +118,21 @@ module Acrofill
 
     # Greedy wrap. Line and space widths are accumulated incrementally so
     # the cost is O(total characters), not O(words * line-length).
-    def wrap_line(text, base_font, size, max_width)
-      space = Metrics.string_width(' ', base_font, size)
+    def wrap_line(text, font, size, max_width)
+      space = Metrics.string_width(' ', font.widths, size)
       lines = ['']
-      widths = [0.0]
+      so_far = [0.0]
       text.split.each do |word|
-        word_width = Metrics.string_width(word, base_font, size)
+        word_width = Metrics.string_width(word, font.widths, size)
         if lines.last.empty?
           lines[-1] = word
-          widths[-1] = word_width
-        elsif widths.last + space + word_width <= max_width
+          so_far[-1] = word_width
+        elsif so_far.last + space + word_width <= max_width
           lines[-1] = "#{lines.last} #{word}"
-          widths[-1] += space + word_width
+          so_far[-1] += space + word_width
         else
           lines << word
-          widths << word_width
+          so_far << word_width
         end
       end
       lines
@@ -182,37 +195,13 @@ module Acrofill
 
     # Fixed sizes that overflow the box are scaled down so the whole value
     # stays visible (Acrobat-style best-fit; pdftk would clip instead).
-    def shrink_to_fit(text, base_font, size, width)
+    def shrink_to_fit(text, font, size, width)
       max_width = width - (2 * PADDING)
-      text_width = Metrics.string_width(text, base_font, size)
+      text_width = Metrics.string_width(text, font.widths, size)
       size *= max_width / text_width if text_width > max_width && text_width.positive?
       size.clamp(2.0, 144.0)
     end
 
-    # The font resource dictionary from /AcroForm /DR /Font, or {} when the
-    # template supplies a malformed (non-dictionary) /DR or /Font.
-    def dr_fonts
-      dr = @doc.deref(@acroform[:DR])
-      return {} unless dr.is_a?(Hash)
-
-      fonts = @doc.deref(dr[:Font])
-      fonts.is_a?(Hash) ? fonts : {}
-    end
-
-    def base_font_for(resource_name)
-      font = @doc.deref(dr_fonts[resource_name.to_sym])
-      base = font.is_a?(Hash) ? font[:BaseFont].to_s : ''
-      base.sub(/\A[A-Z]{6}\+/, '')
-    end
-
-    def font_ref(resource_name)
-      entry = dr_fonts[resource_name.to_sym]
-      return @doc.ref_for(entry) if entry
-
-      # Font not present in /DR: register a plain Helvetica.
-      @doc.add(Type: :Font, Subtype: :Type1, BaseFont: :Helvetica, Encoding: :WinAnsiEncoding)
-    end
-
     def printable_text(value)
       value.to_s.gsub(/[[:space:]]+/, ' ').strip
            .encode('Windows-1252', invalid: :replace, undef: :replace, replace: '?')

data/lib/acrofill/document.rb

@@ -11,6 +11,8 @@ module Acrofill
   # materialized once via pdf-reader (which transparently handles xref
   # streams and object streams), then mutated in place before writing.
   class Document
+    PAGE_NODE_TYPES = %i[Pages Page].freeze
+
     attr_reader :objects, :trailer
 
     def initialize(path)
@@ -96,8 +98,14 @@ module Acrofill
       obj.is_a?(PDF::Reader::Reference) ? obj : add(obj)
     end
 
+    # The document catalog. The trailer may point at a missing or mistyped
+    # object, so this is a parse-boundary check too: callers get an
+    # Acrofill::Error rather than a NoMethodError on nil.
     def root
-      deref(@trailer[:Root])
+      node = deref(@trailer[:Root])
+      raise Error, 'PDF has no document catalog' unless node.is_a?(Hash)
+
+      node
     end
 
     # Depth-first, document-order walk over the /Pages tree. Iterative with
@@ -118,7 +126,7 @@ module Acrofill
           seen[node.id] = true
         end
 
-        case dict[:Type]
+        case node_type(dict)
         when :Pages
           kids = deref(dict[:Kids])
           stack.concat(kids.reverse) if kids.is_a?(Array)
@@ -140,5 +148,19 @@ module Acrofill
       end
       nil
     end
+
+    private
+
+    # /Type is required on page-tree nodes but plenty of real generators omit
+    # it, and a page skipped here is a page silently not flattened. Infer it
+    # from the shape instead: a node with /Kids is an internal node, anything
+    # else reached from the page tree is a leaf.
+    def node_type(dict)
+      type = deref(dict[:Type])
+      return type if PAGE_NODE_TYPES.include?(type)
+      return nil if type.is_a?(Symbol) # something else entirely, e.g. /Font
+
+      dict.key?(:Kids) ? :Pages : :Page
+    end
   end
 end

data/lib/acrofill/flattener.rb

@@ -0,0 +1,159 @@
+# frozen_string_literal: true
+
+module Acrofill
+  # Burns widget appearances into page content and removes the interactive
+  # layer, like pdftk's `output ... flatten`. Every geometry value it reads
+  # comes from the template, so each one is validated before use.
+  class Flattener
+    HIDDEN_FLAG = 2
+
+    def initialize(doc)
+      @doc = doc
+      @stamp_counter = 0
+    end
+
+    def flatten!
+      @doc.each_page { |page| flatten_page(page) }
+      @doc.root.delete(:AcroForm)
+    end
+
+    private
+
+    # Annotation /F flags. Must be dereferenced before to_i: on a
+    # PDF::Reader::Reference, to_i returns the object *number*.
+    def annotation_flags(widget)
+      flags = @doc.deref(widget[:F])
+      flags.is_a?(Integer) ? flags : 0
+    end
+
+    def flatten_page(page)
+      annot_refs = @doc.deref(page[:Annots])
+      annot_refs = [] unless annot_refs.is_a?(Array)
+      annots = annot_refs.map { |a| [a, @doc.deref(a)] }
+      widgets, others = annots.partition { |_ref, dict| dict.is_a?(Hash) && dict[:Subtype] == :Widget }
+      return if widgets.empty?
+
+      stamps = []
+      widgets.each do |_ref, widget|
+        stamp = stamp_operations(page, widget)
+        stamps << stamp if stamp
+      end
+
+      unless stamps.empty?
+        wrap = ->(bytes) { @doc.add(StreamObject.new({}, bytes.b)) }
+        derefed = @doc.deref(page[:Contents])
+        contents = (derefed.is_a?(Array) ? derefed : [page[:Contents]]).compact
+        contents = contents.map { |stream| @doc.ref_for(stream) }
+        page[:Contents] = [wrap.call("q\n"), *contents, wrap.call("\nQ\n#{stamps.join("\n")}\n")]
+      end
+
+      remaining = others.map(&:first)
+      if remaining.empty?
+        page.delete(:Annots)
+      else
+        page[:Annots] = remaining
+      end
+    end
+
+    # Returns content-stream operations placing the widget's normal
+    # appearance onto the page, or nil when there is nothing to draw.
+    # Implements the appearance-box algorithm of PDF 32000 §12.5.5: the
+    # form's /Matrix is applied to its BBox, and the resulting extent is
+    # mapped onto the annotation rectangle.
+    def stamp_operations(page, widget)
+      return nil if annotation_flags(widget).anybits?(HIDDEN_FLAG)
+
+      ap_ref = normal_appearance(widget)
+      xobject = @doc.deref(ap_ref)
+      dict = xobject.is_a?(StreamObject) ? xobject.dict : xobject&.hash
+      return nil unless dict.is_a?(Hash)
+
+      bbox = normalize_box(@doc.deref(dict[:BBox]))
+      rect = normalize_box(@doc.deref(widget[:Rect]))
+      return nil unless bbox && rect
+
+      # Appearance streams are form XObjects, but /Type and /Subtype are
+      # sometimes omitted; /Do requires them.
+      dict[:Type] ||= :XObject
+      dict[:Subtype] ||= :Form
+
+      llx, lly, urx, ury = rect
+      bx0, by0, bx1, by1 = transformed_bbox(bbox, @doc.deref(dict[:Matrix]))
+      bw = bx1 - bx0
+      bh = by1 - by0
+      return nil if bw <= 0 || bh <= 0
+
+      sx = (urx - llx) / bw
+      sy = (ury - lly) / bh
+      name = register_xobject(page, ap_ref)
+      matrix = [sx, 0, 0, sy, llx - (bx0 * sx), lly - (by0 * sy)]
+      ops = matrix.map { |n| Serializer.format_number(n.to_f) }
+      "q #{ops.join(' ')} cm /#{name} Do Q"
+    end
+
+    # Bounding box of the (already normalized) BBox corners after the
+    # form's /Matrix (identity when absent or malformed).
+    def transformed_bbox(bbox, matrix)
+      x0, y0, x1, y1 = bbox
+      matrix = matrix.map { |m| @doc.deref(m) } if matrix.is_a?(Array)
+      return [x0, y0, x1, y1] unless matrix.is_a?(Array) && matrix.size == 6 &&
+                                     matrix.all?(Numeric)
+
+      a, b, c, d, e, f = matrix.map(&:to_f)
+      xs = []
+      ys = []
+      [[x0, y0], [x1, y0], [x0, y1], [x1, y1]].each do |x, y|
+        xs << ((a * x) + (c * y) + e)
+        ys << ((b * x) + (d * y) + f)
+      end
+      [xs.min, ys.min, xs.max, ys.max]
+    end
+
+    def normal_appearance(widget)
+      ap = @doc.deref(widget[:AP])
+      return nil unless ap.is_a?(Hash)
+
+      normal = ap[:N]
+      states = @doc.deref(normal)
+      if states.is_a?(Hash) && !states.is_a?(PDF::Reader::Stream)
+        # Pick the widget's current state; without /AS default to /Off
+        # (never an arbitrary "on" appearance for an unset checkbox).
+        state = @doc.deref(widget[:AS])
+        state = :Off unless state.is_a?(Symbol) && states.key?(state)
+        normal = states[state]
+      end
+      normal
+    end
+
+    # Derefs each element (array entries may legally be indirect objects)
+    # and returns [llx, lly, urx, ury], or nil when the box is not four
+    # numbers.
+    def normalize_box(box)
+      return nil unless box.is_a?(Array) && box.size == 4
+
+      nums = box.map { |n| @doc.deref(n) }
+      return nil unless nums.all?(Numeric)
+
+      xs = [nums[0].to_f, nums[2].to_f].sort
+      ys = [nums[1].to_f, nums[3].to_f].sort
+      [xs[0], ys[0], xs[1], ys[1]]
+    end
+
+    # A page's /Resources (and its /XObject subdictionary) are template
+    # data: anything that is not a dictionary is replaced rather than
+    # indexed, which would raise TypeError on an Array or a stream.
+    def register_xobject(page, ap_ref)
+      resources = @doc.deref(page[:Resources]) || @doc.deref(@doc.inherited_value(page, :Resources))
+      resources = resources.is_a?(Hash) ? resources.dup : {}
+      xobjects = @doc.deref(resources[:XObject])
+      xobjects = xobjects.is_a?(Hash) ? xobjects.dup : {}
+
+      @stamp_counter += 1
+      name = :"AcrofillAP#{@stamp_counter}"
+      xobjects[name] = @doc.ref_for(ap_ref)
+      resources[:XObject] = xobjects
+      page[:Resources] = resources
+      name
+    end
+  end
+end

data/lib/acrofill/fonts.rb

@@ -0,0 +1,144 @@
+# frozen_string_literal: true
+
+module Acrofill
+  # The form's /DR /Font dictionary: the metrics a widget's /DA font resource
+  # actually implies, and the reference to put in a generated appearance's
+  # /Resources.
+  #
+  # Real templates embed their own faces, and those font dictionaries carry
+  # their own /Widths and /FontDescriptor. The appearance stream is drawn
+  # with that very font, so it has to be measured with it too — laying an
+  # embedded face out against standard-14 tables misplaces every centered or
+  # right-aligned value. pdftk reads the dictionary; these are the same rules,
+  # measured against pdftk-java 3.3.3 (see benchmark/geometry_diff.rb):
+  #
+  #   widths     /Widths + /FirstChar, else the standard-14 table
+  #   ascender   /FontDescriptor /Ascent, else standard-14 AFM, else 800
+  #   FontBBox   /FontDescriptor /FontBBox, else standard-14 AFM, else 900/-200
+  #
+  # The standard-14 fallbacks apply only to a BaseFont that literally names
+  # one of the fourteen; a face merely *resembling* one (ArialMT and friends)
+  # gets pdftk's generic defaults vertically, while its widths still fall back
+  # to the closest standard table, there being nothing better to measure with.
+  class Fonts
+    DEFAULT_ASCENDER = 800
+    DEFAULT_DESCENDER = -200
+    DEFAULT_BBOX_TOP = 900
+    DEFAULT_BBOX_BOTTOM = -200
+    # A code outside /FirstChar../LastChar draws as zero-width: pdftk does
+    # not consult /MissingWidth (verified — setting it changes nothing).
+    OUT_OF_RANGE_WIDTH = 0
+    SUBSET_PREFIX = /\A[A-Z]{6}\+/
+
+    def initialize(doc, acroform)
+      @doc = doc
+      @acroform = acroform
+      @metrics = {}
+      @references = {}
+    end
+
+    # Metrics::Font for the font a /DA string names.
+    def metrics(resource_name)
+      key = resource_name.to_sym
+      @metrics[key] ||= build(@doc.deref(entry(key)))
+    end
+
+    # Indirect reference to that font, for the appearance /Resources. Fonts
+    # absent from /DR share one registered Helvetica; a /DR font stored as a
+    # direct dictionary is promoted to an indirect object once, not once per
+    # widget.
+    def reference(resource_name)
+      key = resource_name.to_sym
+      @references[key] ||=
+        begin
+          found = entry(key)
+          found ? @doc.ref_for(found) : fallback
+        end
+    end
+
+    private
+
+    def dr_fonts
+      @dr_fonts ||=
+        begin
+          dr = @doc.deref(@acroform[:DR])
+          fonts = dr.is_a?(Hash) ? @doc.deref(dr[:Font]) : nil
+          fonts.is_a?(Hash) ? fonts : {}
+        end
+    end
+
+    def entry(key)
+      dr_fonts[key]
+    end
+
+    def fallback
+      @fallback ||=
+        @doc.add(Type: :Font, Subtype: :Type1, BaseFont: :Helvetica, Encoding: :WinAnsiEncoding)
+    end
+
+    def build(dict)
+      return Metrics.font_for('') unless dict.is_a?(Hash)
+
+      name = base_font_name(dict)
+      standard = Metrics.standard_font(name)
+      ascender, descender, top, bottom = vertical(dict, standard)
+      Metrics::Font.new(widths(dict) || Metrics.widths_for(name), ascender, descender, top, bottom)
+                   .freeze
+    end
+
+    def base_font_name(dict)
+      @doc.deref(dict[:BaseFont]).to_s.sub(SUBSET_PREFIX, '')
+    end
+
+    # The font's own glyph widths, laid out the way Metrics tables are
+    # (index = code - 32), or nil when the dictionary does not supply usable
+    # ones. Only the codes acrofill can emit are read, so a /Widths array of
+    # any declared length costs the same.
+    def widths(dict)
+      first = @doc.deref(dict[:FirstChar])
+      declared = @doc.deref(dict[:Widths])
+      return nil unless first.is_a?(Integer) && declared.is_a?(Array)
+
+      low = [Metrics::FIRST_CODE - first, 0].max
+      high = [Metrics::LAST_CODE - first, declared.size - 1].min
+      return nil if high < low
+
+      table = Array.new(Metrics::LAST_CODE - Metrics::FIRST_CODE + 1, OUT_OF_RANGE_WIDTH)
+      (low..high).each do |index|
+        width = number(declared[index])
+        table[first + index - Metrics::FIRST_CODE] = width if width
+      end
+      table
+    end
+
+    def vertical(dict, standard)
+      descriptor = @doc.deref(dict[:FontDescriptor])
+      descriptor = nil unless descriptor.is_a?(Hash)
+      bottom, top = font_bbox(descriptor)
+      [
+        (descriptor && number(descriptor[:Ascent])) || standard&.ascender || DEFAULT_ASCENDER,
+        (descriptor && number(descriptor[:Descent])) || standard&.descender || DEFAULT_DESCENDER,
+        top || standard&.bbox_top || DEFAULT_BBOX_TOP,
+        bottom || standard&.bbox_bottom || DEFAULT_BBOX_BOTTOM
+      ]
+    end
+
+    # [lower y, upper y] of /FontBBox, or [nil, nil] when it is unusable.
+    def font_bbox(descriptor)
+      box = descriptor && @doc.deref(descriptor[:FontBBox])
+      return [nil, nil] unless box.is_a?(Array) && box.size == 4
+
+      [number(box[1]), number(box[3])]
+    end
+
+    # Font dictionaries are template data: every scalar may be indirect, and
+    # a non-finite one would poison the geometry it feeds.
+    def number(raw)
+      value = @doc.deref(raw)
+      return nil unless value.is_a?(Numeric)
+
+      value = value.to_f
+      value.finite? ? value : nil
+    end
+  end
+end

data/lib/acrofill/form.rb

@@ -4,7 +4,6 @@ module Acrofill
   # The interactive form of a document: field lookup by fully-qualified
   # name, value filling with appearance regeneration, and flattening.
   class Form
-    HIDDEN_FLAG = 2
     MULTILINE_FLAG = 1 << 12
     PUSHBUTTON_FLAG = 1 << 16
 
@@ -46,7 +45,7 @@ module Acrofill
 
       case field_type(groups.first[:node])
       when :Btn then fill_button(groups, value)
-      when :Tx, :Ch, nil then groups.each { |group| fill_text(group, value) }.any?
+      when :Tx, :Ch, nil then fill_text_groups(groups, value)
       else false # signatures and unknown types are left untouched
       end
     end
@@ -54,8 +53,7 @@ module Acrofill
     # Stamps every visible widget appearance into its page's content and
     # removes the interactive layer, like pdftk's `output ... flatten`.
     def flatten!
-      @doc.each_page { |page| flatten_page(page) }
-      @doc.root.delete(:AcroForm)
+      Flattener.new(@doc).flatten!
     end
 
     private
@@ -76,11 +74,11 @@ module Acrofill
       flags.is_a?(Integer) ? flags : 0
     end
 
-    # Annotation /F flags. Must be dereferenced before to_i: on a
-    # PDF::Reader::Reference, to_i returns the object *number*.
-    def annotation_flags(widget)
-      flags = @doc.deref(widget[:F])
-      flags.is_a?(Integer) ? flags : 0
+    # Every group sharing this name is filled, so the walk must not
+    # short-circuit on the first one.
+    def fill_text_groups(groups, value)
+      groups.each { |group| fill_text(group, value) }
+      true
     end
 
     def fill_text(group, value)
@@ -91,11 +89,14 @@ module Acrofill
       multiline = field_flags(node).anybits?(MULTILINE_FLAG)
       group[:widgets].each do |widget|
         widget.delete(:AS)
-        if value.empty?
-          widget.delete(:AP)
+        ap_ref = @appearance.build(node, widget, value, multiline: multiline) unless value.empty?
+        # An unusable geometry yields no appearance; dropping /AP is still
+        # required, or the widget would keep rendering the *previous* value
+        # while /V already holds the new one.
+        if ap_ref
+          widget[:AP] = { N: ap_ref }
         else
-          ap_ref = @appearance.build(node, widget, value, multiline: multiline)
-          widget[:AP] = { N: ap_ref } if ap_ref
+          widget.delete(:AP)
         end
       end
       true
@@ -222,132 +223,5 @@ module Acrofill
 
       "\xFE\xFF".b + value.encode('UTF-16BE').b
     end
-
-    def flatten_page(page)
-      annot_refs = @doc.deref(page[:Annots])
-      annot_refs = [] unless annot_refs.is_a?(Array)
-      annots = annot_refs.map { |a| [a, @doc.deref(a)] }
-      widgets, others = annots.partition { |_ref, dict| dict.is_a?(Hash) && dict[:Subtype] == :Widget }
-      return if widgets.empty?
-
-      stamps = []
-      widgets.each do |_ref, widget|
-        stamp = stamp_operations(page, widget)
-        stamps << stamp if stamp
-      end
-
-      unless stamps.empty?
-        wrap = ->(bytes) { @doc.add(StreamObject.new({}, bytes.b)) }
-        derefed = @doc.deref(page[:Contents])
-        contents = (derefed.is_a?(Array) ? derefed : [page[:Contents]]).compact
-        contents = contents.map { |stream| @doc.ref_for(stream) }
-        page[:Contents] = [wrap.call("q\n"), *contents, wrap.call("\nQ\n#{stamps.join("\n")}\n")]
-      end
-
-      remaining = others.map(&:first)
-      if remaining.empty?
-        page.delete(:Annots)
-      else
-        page[:Annots] = remaining
-      end
-    end
-
-    # Returns content-stream operations placing the widget's normal
-    # appearance onto the page, or nil when there is nothing to draw.
-    # Implements the appearance-box algorithm of PDF 32000 §12.5.5: the
-    # form's /Matrix is applied to its BBox, and the resulting extent is
-    # mapped onto the annotation rectangle.
-    def stamp_operations(page, widget)
-      return nil if annotation_flags(widget).anybits?(HIDDEN_FLAG)
-
-      ap_ref = normal_appearance(widget)
-      xobject = @doc.deref(ap_ref)
-      dict = xobject.is_a?(StreamObject) ? xobject.dict : xobject&.hash
-      return nil unless dict.is_a?(Hash)
-
-      bbox = normalize_box(@doc.deref(dict[:BBox]))
-      rect = normalize_box(@doc.deref(widget[:Rect]))
-      return nil unless bbox && rect
-
-      # Appearance streams are form XObjects, but /Type and /Subtype are
-      # sometimes omitted; /Do requires them.
-      dict[:Type] ||= :XObject
-      dict[:Subtype] ||= :Form
-
-      llx, lly, urx, ury = rect
-      bx0, by0, bx1, by1 = transformed_bbox(bbox, @doc.deref(dict[:Matrix]))
-      bw = bx1 - bx0
-      bh = by1 - by0
-      return nil if bw <= 0 || bh <= 0
-
-      sx = (urx - llx) / bw
-      sy = (ury - lly) / bh
-      name = register_xobject(page, ap_ref)
-      matrix = [sx, 0, 0, sy, llx - (bx0 * sx), lly - (by0 * sy)]
-      ops = matrix.map { |n| Serializer.format_number(n.to_f) }
-      "q #{ops.join(' ')} cm /#{name} Do Q"
-    end
-
-    # Bounding box of the (already normalized) BBox corners after the
-    # form's /Matrix (identity when absent or malformed).
-    def transformed_bbox(bbox, matrix)
-      x0, y0, x1, y1 = bbox
-      matrix = matrix.map { |m| @doc.deref(m) } if matrix.is_a?(Array)
-      return [x0, y0, x1, y1] unless matrix.is_a?(Array) && matrix.size == 6 &&
-                                     matrix.all?(Numeric)
-
-      a, b, c, d, e, f = matrix.map(&:to_f)
-      xs = []
-      ys = []
-      [[x0, y0], [x1, y0], [x0, y1], [x1, y1]].each do |x, y|
-        xs << ((a * x) + (c * y) + e)
-        ys << ((b * x) + (d * y) + f)
-      end
-      [xs.min, ys.min, xs.max, ys.max]
-    end
-
-    def normal_appearance(widget)
-      ap = @doc.deref(widget[:AP])
-      return nil unless ap.is_a?(Hash)
-
-      normal = ap[:N]
-      states = @doc.deref(normal)
-      if states.is_a?(Hash) && !states.is_a?(PDF::Reader::Stream)
-        # Pick the widget's current state; without /AS default to /Off
-        # (never an arbitrary "on" appearance for an unset checkbox).
-        state = @doc.deref(widget[:AS])
-        state = :Off unless state.is_a?(Symbol) && states.key?(state)
-        normal = states[state]
-      end
-      normal
-    end
-
-    # Derefs each element (array entries may legally be indirect objects)
-    # and returns [llx, lly, urx, ury], or nil when the box is not four
-    # numbers.
-    def normalize_box(box)
-      return nil unless box.is_a?(Array) && box.size == 4
-
-      nums = box.map { |n| @doc.deref(n) }
-      return nil unless nums.all?(Numeric)
-
-      xs = [nums[0].to_f, nums[2].to_f].sort
-      ys = [nums[1].to_f, nums[3].to_f].sort
-      [xs[0], ys[0], xs[1], ys[1]]
-    end
-
-    def register_xobject(page, ap_ref)
-      resources = @doc.deref(page[:Resources]) || @doc.inherited_value(page, :Resources)
-      resources = @doc.deref(resources) || {}
-      resources = resources.dup
-      xobjects = (@doc.deref(resources[:XObject]) || {}).dup
-
-      @stamp_counter = (@stamp_counter || 0) + 1
-      name = :"AcrofillAP#{@stamp_counter}"
-      xobjects[name] = @doc.ref_for(ap_ref)
-      resources[:XObject] = xobjects
-      page[:Resources] = resources
-      name
-    end
   end
 end

data/lib/acrofill/metrics.rb

@@ -1,29 +1,238 @@
 # frozen_string_literal: true
 
-# Glyph widths (1/1000 em) for standard-14 fonts, ASCII 32..126,
-# extracted from Adobe AFM metrics. Index = char code - 32.
 module Acrofill
+  # Glyph widths (1/1000 em) for the standard-14 text fonts, indexed by
+  # WinAnsiEncoding code minus 32 (appearance streams are written as
+  # Windows-1252 bytes against a /WinAnsiEncoding font, so the byte emitted
+  # is the byte measured). Extracted from Adobe AFM metrics.
+  #
+  # Only six width tables are stored: the oblique cuts of Helvetica have the
+  # same widths as the upright face, and the whole Courier family is
+  # monospaced. Vertical metrics (ascender, descender, FontBBox extent) are
+  # per cut, since they differ where widths do not.
   module Metrics
-    WIDTHS = {
-      'Helvetica' => [278, 278, 355, 556, 556, 889, 667, 222, 333, 333, 389, 584, 278, 333, 278, 278, 556,
-                      556, 556, 556, 556, 556, 556, 556, 556, 556, 278, 278, 584, 584, 584, 556, 1015, 667, 667, 722, 722, 667, 611, 778, 722, 278, 500, 667, 556, 833, 722, 778, 667, 778, 722, 667, 611, 722, 667, 944, 667, 667, 611, 278, 278, 278, 469, 556, 222, 556, 556, 500, 556, 556, 278, 556, 556, 222, 222, 500, 222, 833, 556, 556, 556, 556, 333, 500, 278, 556, 500, 722, 500, 500, 500, 334, 260, 334, 584],
-      'Helvetica-Bold' => [278, 333, 474, 556, 556, 889, 722, 278, 333, 333, 389, 584, 278, 333, 278, 278,
-                           556, 556, 556, 556, 556, 556, 556, 556, 556, 556, 333, 333, 584, 584, 584, 611, 975, 722, 722, 722, 722, 667, 611, 778, 722, 278, 556, 722, 611, 833, 722, 778, 667, 778, 722, 667, 611, 722, 667, 944, 667, 667, 611, 333, 278, 333, 584, 556, 278, 556, 611, 556, 611, 556, 333, 611, 611, 278, 278, 556, 278, 889, 611, 611, 611, 611, 389, 556, 333, 611, 556, 778, 556, 556, 500, 389, 280, 389, 584],
-      'Courier' => [600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600,
-                    600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600, 600],
-      'Times-Roman' => [250, 333, 408, 500, 500, 833, 778, 333, 333, 333, 500, 564, 250, 333, 250, 278, 500,
-                        500, 500, 500, 500, 500, 500, 500, 500, 500, 278, 278, 564, 564, 564, 444, 921, 722, 667, 667, 722, 611, 556, 722, 722, 333, 389, 722, 611, 889, 722, 722, 556, 722, 667, 556, 611, 722, 722, 944, 722, 722, 611, 333, 278, 333, 469, 500, 333, 444, 500, 444, 500, 444, 333, 500, 500, 278, 278, 500, 278, 778, 500, 500, 500, 500, 333, 389, 278, 500, 500, 722, 500, 500, 444, 480, 200, 480, 541]
+    DEFAULT_WIDTH = 556
+    COURIER_WIDTH = 600
+    FIRST_CODE = 32
+    LAST_CODE = 255
+    WINDOWS_1252 = Encoding::Windows_1252
+
+    # The standard-14 text cuts, indexed by bold + 2 * italic per family.
+    CUTS = {
+      helvetica: %w[Helvetica Helvetica-Bold Helvetica-Oblique Helvetica-BoldOblique].freeze,
+      times: %w[Times-Roman Times-Bold Times-Italic Times-BoldItalic].freeze,
+      courier: %w[Courier Courier-Bold Courier-Oblique Courier-BoldOblique].freeze
     }.freeze
+    # Family and weight hints for BaseFont names outside the standard 14.
+    SERIF = /times|roman|serif|georgia|garamond/
+    SANS = /sans/
+    FIXED = /courier|mono/
+    BOLD = /bold|black|heavy/
+    ITALIC = /italic|oblique/
 
-    DEFAULT_WIDTH = 556
+    TABLES = {
+      'Helvetica' => [
+        278, 278, 355, 556, 556, 889, 667, 191, 333, 333, 389, 584, 278, 333, 278, 278, 556,
+        556, 556, 556, 556, 556, 556, 556, 556, 556, 278, 278, 584, 584, 584, 556, 1015, 667,
+        667, 722, 722, 667, 611, 778, 722, 278, 500, 667, 556, 833, 722, 778, 667, 778, 722,
+        667, 611, 722, 667, 944, 667, 667, 611, 278, 278, 278, 469, 556, 333, 556, 556, 500,
+        556, 556, 278, 556, 556, 222, 222, 500, 222, 833, 556, 556, 556, 556, 333, 500, 278,
+        556, 500, 722, 500, 500, 500, 334, 260, 334, 584, nil, 556, nil, 222, 556, 333, 1000,
+        556, 556, 333, 1000, 667, 333, 1000, nil, 611, nil, nil, 222, 222, 333, 333, 350, 556,
+        1000, 333, 1000, 500, 333, 944, nil, 500, 667, 278, 333, 556, 556, 556, 556, 260, 556,
+        333, 737, 370, 556, 584, 333, 737, 333, 400, 584, 333, 333, 333, 556, 537, 278, 333,
+        333, 365, 556, 834, 834, 834, 611, 667, 667, 667, 667, 667, 667, 1000, 722, 667, 667,
+        667, 667, 278, 278, 278, 278, 722, 722, 778, 778, 778, 778, 778, 584, 778, 722, 722,
+        722, 722, 667, 667, 611, 556, 556, 556, 556, 556, 556, 889, 500, 556, 556, 556, 556,
+        278, 278, 278, 278, 556, 556, 556, 556, 556, 556, 556, 584, 611, 556, 556, 556, 556,
+        500, 556, 500
+      ].freeze,
+      'Helvetica-Bold' => [
+        278, 333, 474, 556, 556, 889, 722, 238, 333, 333, 389, 584, 278, 333, 278, 278, 556,
+        556, 556, 556, 556, 556, 556, 556, 556, 556, 333, 333, 584, 584, 584, 611, 975, 722,
+        722, 722, 722, 667, 611, 778, 722, 278, 556, 722, 611, 833, 722, 778, 667, 778, 722,
+        667, 611, 722, 667, 944, 667, 667, 611, 333, 278, 333, 584, 556, 333, 556, 611, 556,
+        611, 556, 333, 611, 611, 278, 278, 556, 278, 889, 611, 611, 611, 611, 389, 556, 333,
+        611, 556, 778, 556, 556, 500, 389, 280, 389, 584, nil, 556, nil, 278, 556, 500, 1000,
+        556, 556, 333, 1000, 667, 333, 1000, nil, 611, nil, nil, 278, 278, 500, 500, 350, 556,
+        1000, 333, 1000, 556, 333, 944, nil, 500, 667, 278, 333, 556, 556, 556, 556, 280, 556,
+        333, 737, 370, 556, 584, 333, 737, 333, 400, 584, 333, 333, 333, 611, 556, 278, 333,
+        333, 365, 556, 834, 834, 834, 611, 722, 722, 722, 722, 722, 722, 1000, 722, 667, 667,
+        667, 667, 278, 278, 278, 278, 722, 722, 778, 778, 778, 778, 778, 584, 778, 722, 722,
+        722, 722, 667, 667, 611, 556, 556, 556, 556, 556, 556, 889, 556, 556, 556, 556, 556,
+        278, 278, 278, 278, 611, 611, 611, 611, 611, 611, 611, 584, 611, 611, 611, 611, 611,
+        556, 611, 556
+      ].freeze,
+      'Times-Roman' => [
+        250, 333, 408, 500, 500, 833, 778, 180, 333, 333, 500, 564, 250, 333, 250, 278, 500,
+        500, 500, 500, 500, 500, 500, 500, 500, 500, 278, 278, 564, 564, 564, 444, 921, 722,
+        667, 667, 722, 611, 556, 722, 722, 333, 389, 722, 611, 889, 722, 722, 556, 722, 667,
+        556, 611, 722, 722, 944, 722, 722, 611, 333, 278, 333, 469, 500, 333, 444, 500, 444,
+        500, 444, 333, 500, 500, 278, 278, 500, 278, 778, 500, 500, 500, 500, 333, 389, 278,
+        500, 500, 722, 500, 500, 444, 480, 200, 480, 541, nil, 500, nil, 333, 500, 444, 1000,
+        500, 500, 333, 1000, 556, 333, 889, nil, 611, nil, nil, 333, 333, 444, 444, 350, 500,
+        1000, 333, 980, 389, 333, 722, nil, 444, 722, 250, 333, 500, 500, 500, 500, 200, 500,
+        333, 760, 276, 500, 564, 333, 760, 333, 400, 564, 300, 300, 333, 500, 453, 250, 333,
+        300, 310, 500, 750, 750, 750, 444, 722, 722, 722, 722, 722, 722, 889, 667, 611, 611,
+        611, 611, 333, 333, 333, 333, 722, 722, 722, 722, 722, 722, 722, 564, 722, 722, 722,
+        722, 722, 722, 556, 500, 444, 444, 444, 444, 444, 444, 667, 444, 444, 444, 444, 444,
+        278, 278, 278, 278, 500, 500, 500, 500, 500, 500, 500, 564, 500, 500, 500, 500, 500,
+        500, 500, 500
+      ].freeze,
+      'Times-Bold' => [
+        250, 333, 555, 500, 500, 1000, 833, 278, 333, 333, 500, 570, 250, 333, 250, 278, 500,
+        500, 500, 500, 500, 500, 500, 500, 500, 500, 333, 333, 570, 570, 570, 500, 930, 722,
+        667, 722, 722, 667, 611, 778, 778, 389, 500, 778, 667, 944, 722, 778, 611, 778, 722,
+        556, 667, 722, 722, 1000, 722, 722, 667, 333, 278, 333, 581, 500, 333, 500, 556, 444,
+        556, 444, 333, 500, 556, 278, 333, 556, 278, 833, 556, 500, 556, 556, 444, 389, 333,
+        556, 500, 722, 500, 500, 444, 394, 220, 394, 520, nil, 500, nil, 333, 500, 500, 1000,
+        500, 500, 333, 1000, 556, 333, 1000, nil, 667, nil, nil, 333, 333, 500, 500, 350, 500,
+        1000, 333, 1000, 389, 333, 722, nil, 444, 722, 250, 333, 500, 500, 500, 500, 220, 500,
+        333, 747, 300, 500, 570, 333, 747, 333, 400, 570, 300, 300, 333, 556, 540, 250, 333,
+        300, 330, 500, 750, 750, 750, 500, 722, 722, 722, 722, 722, 722, 1000, 722, 667, 667,
+        667, 667, 389, 389, 389, 389, 722, 722, 778, 778, 778, 778, 778, 570, 778, 722, 722,
+        722, 722, 722, 611, 556, 500, 500, 500, 500, 500, 500, 722, 444, 444, 444, 444, 444,
+        278, 278, 278, 278, 500, 556, 500, 500, 500, 500, 500, 570, 500, 556, 556, 556, 556,
+        500, 556, 500
+      ].freeze,
+      'Times-Italic' => [
+        250, 333, 420, 500, 500, 833, 778, 214, 333, 333, 500, 675, 250, 333, 250, 278, 500,
+        500, 500, 500, 500, 500, 500, 500, 500, 500, 333, 333, 675, 675, 675, 500, 920, 611,
+        611, 667, 722, 611, 611, 722, 722, 333, 444, 667, 556, 833, 667, 722, 611, 722, 611,
+        500, 556, 722, 611, 833, 611, 556, 556, 389, 278, 389, 422, 500, 333, 500, 500, 444,
+        500, 444, 278, 500, 500, 278, 278, 444, 278, 722, 500, 500, 500, 500, 389, 389, 278,
+        500, 444, 667, 444, 444, 389, 400, 275, 400, 541, nil, 500, nil, 333, 500, 556, 889,
+        500, 500, 333, 1000, 500, 333, 944, nil, 556, nil, nil, 333, 333, 556, 556, 350, 500,
+        889, 333, 980, 389, 333, 667, nil, 389, 556, 250, 389, 500, 500, 500, 500, 275, 500,
+        333, 760, 276, 500, 675, 333, 760, 333, 400, 675, 300, 300, 333, 500, 523, 250, 333,
+        300, 310, 500, 750, 750, 750, 500, 611, 611, 611, 611, 611, 611, 889, 667, 611, 611,
+        611, 611, 333, 333, 333, 333, 722, 667, 722, 722, 722, 722, 722, 675, 722, 722, 722,
+        722, 722, 556, 611, 500, 500, 500, 500, 500, 500, 500, 667, 444, 444, 444, 444, 444,
+        278, 278, 278, 278, 500, 500, 500, 500, 500, 500, 500, 675, 500, 500, 500, 500, 500,
+        444, 500, 444
+      ].freeze,
+      'Times-BoldItalic' => [
+        250, 389, 555, 500, 500, 833, 778, 278, 333, 333, 500, 570, 250, 333, 250, 278, 500,
+        500, 500, 500, 500, 500, 500, 500, 500, 500, 333, 333, 570, 570, 570, 500, 832, 667,
+        667, 667, 722, 667, 667, 722, 778, 389, 500, 667, 611, 889, 722, 722, 611, 722, 667,
+        556, 611, 722, 667, 889, 667, 611, 611, 333, 278, 333, 570, 500, 333, 500, 500, 444,
+        500, 444, 333, 500, 556, 278, 278, 500, 278, 778, 556, 500, 500, 500, 389, 389, 278,
+        556, 444, 667, 500, 444, 389, 348, 220, 348, 570, nil, 500, nil, 333, 500, 500, 1000,
+        500, 500, 333, 1000, 556, 333, 944, nil, 611, nil, nil, 333, 333, 500, 500, 350, 500,
+        1000, 333, 1000, 389, 333, 722, nil, 389, 611, 250, 389, 500, 500, 500, 500, 220, 500,
+        333, 747, 266, 500, 606, 333, 747, 333, 400, 570, 300, 300, 333, 576, 500, 250, 333,
+        300, 300, 500, 750, 750, 750, 500, 667, 667, 667, 667, 667, 667, 944, 667, 667, 667,
+        667, 667, 389, 389, 389, 389, 722, 722, 722, 722, 722, 722, 722, 570, 722, 722, 722,
+        722, 722, 611, 611, 500, 500, 500, 500, 500, 500, 500, 722, 444, 444, 444, 444, 444,
+        278, 278, 278, 278, 500, 556, 500, 500, 500, 500, 500, 570, 500, 556, 556, 556, 556,
+        444, 500, 444
+      ].freeze,
+      'Courier' => Array.new(LAST_CODE - FIRST_CODE + 1, COURIER_WIDTH).freeze
+    }.freeze
+
+    # Cuts that share a width table with another cut.
+    WIDTHS = TABLES.merge(
+      'Helvetica-Oblique' => TABLES['Helvetica'],
+      'Helvetica-BoldOblique' => TABLES['Helvetica-Bold'],
+      'Courier-Bold' => TABLES['Courier'],
+      'Courier-Oblique' => TABLES['Courier'],
+      'Courier-BoldOblique' => TABLES['Courier']
+    ).freeze
+
+    # [ascender, descender, FontBBox top, FontBBox bottom] in 1/1000 em,
+    # from the same Adobe AFM data as the widths. pdftk places baselines
+    # from the ascender and spaces multiline rows by the FontBBox extent,
+    # so these drive vertical geometry.
+    VERTICAL = {
+      'Helvetica' => [718, -207, 931, -225].freeze,
+      'Helvetica-Bold' => [718, -207, 962, -228].freeze,
+      'Helvetica-Oblique' => [718, -207, 931, -225].freeze,
+      'Helvetica-BoldOblique' => [718, -207, 962, -228].freeze,
+      'Times-Roman' => [683, -217, 898, -218].freeze,
+      'Times-Bold' => [683, -217, 935, -218].freeze,
+      'Times-Italic' => [683, -217, 883, -217].freeze,
+      'Times-BoldItalic' => [683, -217, 921, -218].freeze,
+      'Courier' => [629, -157, 805, -250].freeze,
+      'Courier-Bold' => [629, -157, 801, -250].freeze,
+      'Courier-Oblique' => [629, -157, 805, -250].freeze,
+      'Courier-BoldOblique' => [629, -157, 801, -250].freeze
+    }.freeze
+
+    # Everything the appearance code needs about one resolved face.
+    Font = Struct.new(:widths, :ascender, :descender, :bbox_top, :bbox_bottom) do
+      def ascent(size) = ascender * size / 1000.0
 
-    def self.string_width(str, base_font, size)
-      table = WIDTHS[base_font] || WIDTHS['Helvetica']
-      units = str.each_char.sum do |ch|
-        code = ch.ord
-        code.between?(32, 126) ? table[code - 32] : DEFAULT_WIDTH
+      def descent(size) = -descender * size / 1000.0
+
+      # Distance from the box top to the first baseline, and between rows.
+      def top(size) = bbox_top * size / 1000.0
+
+      def line_height(size) = (bbox_top - bbox_bottom) * size / 1000.0
+    end
+
+    FONTS = VERTICAL.to_h do |name, (asc, desc, top, bottom)|
+      [name, Font.new(WIDTHS[name], asc, desc, top, bottom).freeze]
+    end.freeze
+
+    # Width of +str+ at +size+ points. +widths+ is a table from #widths_for,
+    # or a BaseFont name (resolved here, at the cost of the lookup).
+    def self.string_width(str, widths, size)
+      widths = widths_for(widths) unless widths.is_a?(Array)
+      str = to_win_ansi(str)
+      units = str.each_byte.sum do |code|
+        (code >= FIRST_CODE && widths[code - FIRST_CODE]) || DEFAULT_WIDTH
       end
       units * size / 1000.0
     end
+
+    # Metrics for a BaseFont name. Standard-14 names resolve directly;
+    # anything else (ArialMT, TimesNewRomanPS-BoldMT, a template's embedded
+    # face) is classified by family and weight, which is far closer than
+    # treating everything as Helvetica.
+    def self.font_for(base_font)
+      FONTS.fetch(canonical_name(base_font))
+    end
+
+    def self.widths_for(base_font)
+      font_for(base_font).widths
+    end
+
+    # Metrics for a name that *is* one of the standard 14, or nil. Unlike
+    # #font_for this does not classify: a template's own face only inherits
+    # standard-14 vertical metrics when it actually names one.
+    def self.standard_font(base_font)
+      FONTS[base_font.to_s]
+    end
+
+    # Maps any BaseFont name onto one of the twelve standard-14 text cuts.
+    def self.canonical_name(base_font)
+      name = base_font.to_s
+      return name if FONTS.key?(name)
+
+      lower = name.downcase
+      family = if FIXED.match?(lower) then :courier
+               elsif serif?(lower) then :times
+               else :helvetica
+               end
+      CUTS[family][weight_index(lower)]
+    end
+
+    def self.serif?(lower)
+      SERIF.match?(lower) && !SANS.match?(lower)
+    end
+
+    def self.weight_index(lower)
+      (BOLD.match?(lower) ? 1 : 0) + (ITALIC.match?(lower) ? 2 : 0)
+    end
+
+    # Appearance text is already Windows-1252; anything else is converted so
+    # that measuring and rendering agree byte for byte.
+    def self.to_win_ansi(str)
+      return str if str.encoding == WINDOWS_1252
+
+      str.encode(WINDOWS_1252, invalid: :replace, undef: :replace, replace: '?')
+    end
+
+    private_class_method :serif?, :weight_index, :to_win_ansi
   end
 end

data/lib/acrofill/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Acrofill
-  VERSION = '0.1.2'
+  VERSION = '0.3.0'
 end

data/lib/acrofill.rb

@@ -5,7 +5,9 @@ require_relative 'acrofill/document'
 require_relative 'acrofill/metrics'
 require_relative 'acrofill/serializer'
 require_relative 'acrofill/writer'
+require_relative 'acrofill/fonts'
 require_relative 'acrofill/appearance'
+require_relative 'acrofill/flattener'
 require_relative 'acrofill/form'
 require_relative 'acrofill/filler'
 require_relative 'acrofill/template'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: acrofill
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.3.0
 platform: ruby
 authors:
 - stiig
@@ -37,6 +37,8 @@ files:
 - lib/acrofill/appearance.rb
 - lib/acrofill/document.rb
 - lib/acrofill/filler.rb
+- lib/acrofill/flattener.rb
+- lib/acrofill/fonts.rb
 - lib/acrofill/form.rb
 - lib/acrofill/metrics.rb
 - lib/acrofill/serializer.rb