# Buffer for body body = bytearray()
# Prepare header magic = b'FGOA' version = data['version'] rules = data['optional_features'] rule_count = len(rules)
def parse_header(self): self.magic = self.data[:4] self.version = int.from_bytes(self.data[4:8], 'little') self.rule_count = int.from_bytes(self.data[8:12], 'little') # ... parse rules When shaping Arabic text, after mandatory shaping, apply optional rules: fg-optional-arabic.bin
# Write each rule as: input_len (1B), inputs (4B each), output_len (1B), outputs (4B each), priority (1B) for rule in rules: inputs = rule['input'] outputs = rule['output'] if isinstance(rule['output'], list) else [rule['output']]
print(f"Generated output_bin_path with rule_count rules") If this file is intended for a specific engine, use its tooling: # Buffer for body body = bytearray() #
body.append(rule.get('priority', 50))
import struct with open("fg-optional-arabic.bin", "rb") as f: magic = f.read(4) version = struct.unpack("<I", f.read(4))[0] rule_count = struct.unpack("<I", f.read(4))[0] print(f"Magic: magic, Version: version, Rules: rule_count") If you need to create such a file (e.g., for a custom Arabic shaper), follow this pipeline: 3.1 Source Rule Definition (e.g., rules.json ) "version": 2, "language": "ara", "optional_features": [ "name": "lam_alef_ligature", "input": [0x0644, 0x0627], "output": 0xFEF5, "condition": "diacritic_free", "priority": 90 , "name": "kashida_after_alef", "input": [0x0627], "output": [0x0627, 0x0640], "condition": "justify_mode", "priority": 30 ] 'little') self.rule_count = int.from_bytes(self.data[8:12]
// ... load rule table from body_offset fclose(f); return hdr;
import mmap class OptionalArabicRules: def init (self, bin_path): with open(bin_path, 'rb') as f: self.data = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ) self.parse_header()
body.append(len(inputs)) for ch in inputs: body.extend(struct.pack('<I', ch)) # Unicode codepoint