Class: Fontisan::Hints::TrueTypeInstructionAnalyzer

Inherits:
Object
  • Object
show all
Defined in:
lib/fontisan/hints/truetype_instruction_analyzer.rb

Overview

Analyzes TrueType bytecode instructions to extract hint parameters

This analyzer parses fpgm (Font Program) and prep (Control Value Program) bytecode to extract semantic hint information that can be converted to PostScript Private dict parameters.

Key Extracted Parameters:

  • Blue zones (alignment zones for baseline, x-height, cap-height)
  • Stem widths (from CVT setup in prep)
  • Delta base and shift values
  • Twilight zone setup

Examples:

Analyze prep program

analyzer = TrueTypeInstructionAnalyzer.new
params = analyzer.analyze_prep(prep_bytecode, cvt_values)

Constant Summary collapse

NPUSHB =

TrueType instruction opcodes relevant for hint extraction

0x40
NPUSHW =

Push N bytes

0x41
PUSHB =

Push N words

(0xB0..0xB7).to_a
PUSHW =

Push 1-8 words

(0xB8..0xBF).to_a
SVTCA_Y =

Set freedom and projection vectors to Y-axis

0x00
SVTCA_X =

Set freedom and projection vectors to X-axis

0x01
RCVT =

Read CVT

0x45
WCVTP =

Write CVT (in Pixels)

0x44
WCVTF =

Write CVT (in FUnits)

0x70
MDAP =

Move Direct Absolute Point

[0x2E, 0x2F].freeze
SCVTCI =

Set Control Value Table Cut In

0x1D
SSWCI =

Set Single Width Cut In

0x1E
SSW =

Set Single Width

0x1F

Instance Method Summary collapse

Instance Method Details

#analyze_fpgm(fpgm) ⇒ Hash

Analyze Font Program (fpgm) for complexity indicators

The fpgm contains font-level function definitions. While we don't fully decompile it, we can extract useful metadata about hint complexity.

Parameters:

  • fpgm (String) —

    Binary fpgm data

Returns:

  • (Hash) —

    Analysis results with complexity indicators



155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
 
# File 'lib/fontisan/hints/truetype_instruction_analyzer.rb', line 155

def analyze_fpgm(fpgm)
  return {} if fpgm.nil? || fpgm.empty?

  size = fpgm.bytesize

  # Estimate complexity based on size
  complexity = if size < 100
                 :simple
               elsif size < 200
                 :moderate
               else
                 :complex
               end

  {
    fpgm_size: size,
    has_functions: size.positive?,
    complexity: complexity,
  }
rescue StandardError
  # Return empty hash on any error
  {}
end

#analyze_prep(prep, cvt = []) ⇒ Hash

Analyze prep program to extract hint parameters

Parameters:

  • prep (String) —

    Control value program bytecode

  • cvt (Array<Integer>) (defaults to: []) —

    Control values

Returns:

  • (Hash) —

    Extracted hint parameters



42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
 
# File 'lib/fontisan/hints/truetype_instruction_analyzer.rb', line 42

def analyze_prep(prep, cvt = [])
  return {} if prep.nil? && (cvt.nil? || cvt.empty?)

  params = {}

  # Parse prep bytecode if present
  if prep && !prep.empty?
    bytes = prep.bytes
    stack = []
    i = 0

    while i < bytes.length
      opcode = bytes[i]

      case opcode
      when NPUSHB
        # Push N bytes
        i += 1
        count = bytes[i]
        i += 1
        count.times do
          stack.push(bytes[i])
          i += 1
        end
        next

      when NPUSHW
        # Push N words (16-bit values)
        i += 1
        count = bytes[i]
        i += 1
        count.times do
          value = (bytes[i] << 8) | bytes[i + 1]
          # Convert to signed
          value = value - 65536 if value > 32767
          stack.push(value)
          i += 2
        end
        next

      when *PUSHB
        # Push 1-8 bytes
        count = opcode - 0xB0 + 1
        i += 1
        count.times do
          stack.push(bytes[i])
          i += 1
        end
        next

      when *PUSHW
        # Push 1-8 words
        count = opcode - 0xB8 + 1
        i += 1
        count.times do
          value = (bytes[i] << 8) | bytes[i + 1]
          value = value - 65536 if value > 32767
          stack.push(value)
          i += 2
        end
        next

      when WCVTP, WCVTF
        # Write to CVT - this shows which CVT indices are being set up
        # Pattern: value cvt_index WCVTP (stack top to bottom)
        if stack.length >= 2
          value = stack.pop
          stack.pop
          # Track CVT modifications (useful for understanding setup)
        end

      when SSW
        # Set Single Width - used for stem width control
        if stack.length >= 1
          width = stack.pop
          params[:single_width] = width unless params[:single_width]
        end

      when SSWCI
        # Set Single Width Cut In
        if stack.length >= 1
          params[:single_width_cut_in] = stack.pop
        end

      when SCVTCI
        # Set CVT Cut In
        if stack.length >= 1
          params[:cvt_cut_in] = stack.pop
        end
      end

      i += 1
    end
  end

  # Extract blue zones from CVT analysis (always do this if CVT is present)
  if cvt && !cvt.empty?
    params.merge!(extract_blue_zones_from_cvt(cvt))
  end

  params
rescue StandardError => e
  warn "Error analyzing prep program: #{e.message}"
  {}
end

#extract_blue_zones_from_cvt(cvt) ⇒ Hash

Extract blue zones from CVT values using heuristics

Blue zones in PostScript define alignment constraints for baseline, x-height, cap-height, ascender, and descender. TrueType doesn't have explicit blue zones, but we can derive them from CVT values using common patterns.

Heuristics:

  • Negative values near -250 to -200: Descender zones
  • Values near 0: Baseline zones
  • Values near 500-550: X-height zones
  • Values near 700-750: Cap-height zones
  • For large UPM (>2000): Scale thresholds proportionally

Parameters:

  • cvt (Array<Integer>) —

    Control Value Table entries

Returns:

  • (Hash) —

    Extracted blue zone parameters



195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
 
# File 'lib/fontisan/hints/truetype_instruction_analyzer.rb', line 195

def extract_blue_zones_from_cvt(cvt)
  return {} if cvt.nil? || cvt.empty?

  zones = {}

  # Detect scale from maximum absolute value
  max_value = cvt.map(&:abs).max
  scale_factor = max_value > 1000 ? (max_value / 1000.0) : 1.0

  # Scaled thresholds
  descender_min = (-300 * scale_factor).to_i
  descender_max = (-150 * scale_factor).to_i
  baseline_range = (50 * scale_factor).to_i
  xheight_min = (450 * scale_factor).to_i
  xheight_max = (600 * scale_factor).to_i
  capheight_min = (650 * scale_factor).to_i
  capheight_max = (1500 * scale_factor).to_i # Wider range for large UPM

  # Group CVT values by typical alignment zones
  descender_values = cvt.select do |v|
    v < descender_max && v > descender_min
  end
  baseline_values = cvt.select do |v|
    v >= -baseline_range && v <= baseline_range
  end
  cvt.select { |v| v >= xheight_min && v <= xheight_max }
  capheight_values = cvt.select do |v|
    v >= capheight_min && v <= capheight_max
  end

  # Build blue_values (baseline and top zones)
  blue_values = []

  # Add baseline zone if detected
  if baseline_values.any?
    min_baseline = baseline_values.min
    max_baseline = baseline_values.max
    blue_values << min_baseline << max_baseline
  end

  # Add cap-height zone if detected (or any top zone for large UPM)
  if capheight_values.any?
    min_cap = capheight_values.min
    max_cap = capheight_values.max
    blue_values << min_cap << max_cap
  end

  zones[:blue_values] = blue_values unless blue_values.empty?

  # Build other_blues (descender zones)
  if descender_values.any?
    min_desc = descender_values.min
    max_desc = descender_values.max
    zones[:other_blues] = [min_desc, max_desc]
  end

  zones
end