| 1 | #!/usr/bin/env python3 |
| 2 | """Independent TrueType reader used only to cross-check nox-text. |
| 3 | |
| 4 | Deliberately written from the spec with nothing but `struct`, so that agreement |
| 5 | with the Rust parser is real evidence and not a shared bug. Emits a JSON fixture |
| 6 | consumed by tests/parser.rs. |
| 7 | """ |
| 8 | import json |
| 9 | import struct |
| 10 | import sys |
| 11 | |
| 12 | U16 = lambda b, o: struct.unpack_from(">H", b, o)[0] |
| 13 | S16 = lambda b, o: struct.unpack_from(">h", b, o)[0] |
| 14 | U32 = lambda b, o: struct.unpack_from(">I", b, o)[0] |
| 15 | |
| 16 | PROBE = ("A", "V", "T", "W", "L", "M", "Q", "a", "o", "g", "f", "i", "z", "1", " ", ".", |
| 17 | ",", "@", "~", "\u00e9", "\u00fc", "\u00f1", "\u03a9", "\u2192", "\u5b57") |
| 18 | |
| 19 | |
| 20 | def tables(b): |
| 21 | off = 0 |
| 22 | if b[:4] == b"ttcf": |
| 23 | off = U32(b, 12) |
| 24 | n = U16(b, off + 4) |
| 25 | out = {} |
| 26 | for i in range(n): |
| 27 | e = off + 12 + i * 16 |
| 28 | tag = b[e:e + 4].decode("latin1") |
| 29 | out[tag] = (U32(b, e + 8), U32(b, e + 12)) |
| 30 | return out |
| 31 | |
| 32 | |
| 33 | def cmap_lookup(b, base, ch): |
| 34 | """Walk every subtable, score them, then resolve through the winner.""" |
| 35 | n = U16(b, base + 2) |
| 36 | best, best_score = None, -1 |
| 37 | for i in range(n): |
| 38 | rec = base + 4 + i * 8 |
| 39 | plat, enc, off = U16(b, rec), U16(b, rec + 2), U32(b, rec + 4) |
| 40 | sub = base + off |
| 41 | if sub + 4 > len(b): |
| 42 | continue |
| 43 | fmt = U16(b, sub) |
| 44 | score = {(3, 10, 12): 100, (0, 3, 12): 95, (0, 4, 12): 95, (0, 6, 12): 95, |
| 45 | (3, 1, 4): 90, (0, 3, 4): 85, (0, 4, 4): 85, (0, 6, 4): 85, |
| 46 | (0, 0, 4): 85, (0, 1, 4): 85, (0, 2, 4): 85, |
| 47 | (3, 0, 4): 70, (1, 0, 6): 40, (1, 0, 0): 30}.get((plat, enc, fmt)) |
| 48 | if score is None: |
| 49 | score = 20 if fmt in (0, 4, 6, 12) else None |
| 50 | if score is None: |
| 51 | continue |
| 52 | if score > best_score: |
| 53 | best, best_score = (sub, fmt), score |
| 54 | if best is None: |
| 55 | return None |
| 56 | sub, fmt = best |
| 57 | cp = ord(ch) |
| 58 | if fmt == 0: |
| 59 | return b[sub + 6 + cp] if cp < 256 else 0 |
| 60 | if fmt == 6: |
| 61 | first, count = U16(b, sub + 6), U16(b, sub + 8) |
| 62 | if cp < first or cp >= first + count: |
| 63 | return 0 |
| 64 | return U16(b, sub + 10 + (cp - first) * 2) |
| 65 | if fmt == 4: |
| 66 | if cp > 0xFFFF: |
| 67 | return 0 |
| 68 | seg2 = U16(b, sub + 6) |
| 69 | segs = seg2 // 2 |
| 70 | ends = sub + 14 |
| 71 | starts = ends + seg2 + 2 |
| 72 | deltas = starts + seg2 |
| 73 | ranges = deltas + seg2 |
| 74 | for s in range(segs): |
| 75 | if U16(b, ends + s * 2) >= cp: |
| 76 | start = U16(b, starts + s * 2) |
| 77 | if cp < start: |
| 78 | return 0 |
| 79 | delta = S16(b, deltas + s * 2) |
| 80 | ro = U16(b, ranges + s * 2) |
| 81 | if ro == 0: |
| 82 | return (cp + delta) & 0xFFFF |
| 83 | idx = ranges + s * 2 + ro + (cp - start) * 2 |
| 84 | if idx + 2 > len(b): |
| 85 | return 0 |
| 86 | g = U16(b, idx) |
| 87 | return 0 if g == 0 else (g + delta) & 0xFFFF |
| 88 | return 0 |
| 89 | if fmt == 12: |
| 90 | ngroups = U32(b, sub + 12) |
| 91 | for g in range(ngroups): |
| 92 | e = sub + 16 + g * 12 |
| 93 | lo, hi, sg = U32(b, e), U32(b, e + 4), U32(b, e + 8) |
| 94 | if lo <= cp <= hi: |
| 95 | return min(sg + cp - lo, 0xFFFF) |
| 96 | return 0 |
| 97 | return None |
| 98 | |
| 99 | |
| 100 | def kern_pairs(b, tb): |
| 101 | """Every (left, right) -> value in the font's own kern table.""" |
| 102 | if tb is None: |
| 103 | return {} |
| 104 | base, _ = tb |
| 105 | if U16(b, base) != 0: |
| 106 | return {} |
| 107 | n = U16(b, base + 2) |
| 108 | off = base + 4 |
| 109 | out = {} |
| 110 | for _ in range(n): |
| 111 | length = U16(b, off + 2) |
| 112 | cov = U16(b, off + 4) |
| 113 | if cov & 1 and (cov >> 8) == 0: |
| 114 | npairs = U16(b, off + 6) |
| 115 | for p in range(npairs): |
| 116 | e = off + 14 + p * 6 |
| 117 | if e + 6 > len(b): |
| 118 | break |
| 119 | out.setdefault((U16(b, e), U16(b, e + 2)), S16(b, e + 4)) |
| 120 | if length < 6: |
| 121 | break |
| 122 | off += length |
| 123 | return out |
| 124 | |
| 125 | |
| 126 | def contour_stats(b, tabs, gid, upem, long_loca, num_glyphs, depth=0): |
| 127 | """(contours, points, on_curve_points, simple, bbox) with composites resolved. |
| 128 | |
| 129 | bbox is over decoded points and is only reported for simple glyphs, where it |
| 130 | exercises the flag-repeat and x/y delta decoding end to end.""" |
| 131 | if depth > 6 or gid >= num_glyphs: |
| 132 | return None |
| 133 | lo, ll = tabs["loca"] |
| 134 | go, gl = tabs["glyf"] |
| 135 | if long_loca: |
| 136 | a, z = U32(b, lo + gid * 4), U32(b, lo + gid * 4 + 4) |
| 137 | else: |
| 138 | a, z = U16(b, lo + gid * 2) * 2, U16(b, lo + gid * 2 + 2) * 2 |
| 139 | if z <= a: |
| 140 | return (0, 0, 0, True, None) |
| 141 | p = go + a |
| 142 | nc = S16(b, p) |
| 143 | if nc < 0: |
| 144 | off = p + 10 |
| 145 | tot = [0, 0, 0] |
| 146 | for _ in range(32): |
| 147 | flags, sub = U16(b, off), U16(b, off + 2) |
| 148 | off += 4 |
| 149 | off += 4 if flags & 1 else 2 |
| 150 | if flags & 8: |
| 151 | off += 2 |
| 152 | elif flags & 0x40: |
| 153 | off += 4 |
| 154 | elif flags & 0x80: |
| 155 | off += 8 |
| 156 | r = contour_stats(b, tabs, sub, upem, long_loca, num_glyphs, depth + 1) |
| 157 | if r: |
| 158 | tot = [tot[i] + r[i] for i in range(3)] |
| 159 | if not flags & 0x20: |
| 160 | break |
| 161 | return (tot[0], tot[1], tot[2], False, None) |
| 162 | p += 10 |
| 163 | ends = [U16(b, p + i * 2) for i in range(nc)] |
| 164 | p += nc * 2 |
| 165 | npts = ends[-1] + 1 if ends else 0 |
| 166 | ins = U16(b, p) |
| 167 | p += 2 + ins |
| 168 | flags = [] |
| 169 | while len(flags) < npts: |
| 170 | f = b[p] |
| 171 | p += 1 |
| 172 | flags.append(f) |
| 173 | if f & 8: |
| 174 | rep = b[p] |
| 175 | p += 1 |
| 176 | flags.extend([f] * min(rep, npts - len(flags))) |
| 177 | on = sum(1 for f in flags if f & 1) |
| 178 | xs, v = [], 0 |
| 179 | for f in flags: |
| 180 | if f & 2: |
| 181 | d = b[p]; p += 1 |
| 182 | v += d if f & 16 else -d |
| 183 | elif not f & 16: |
| 184 | v += S16(b, p); p += 2 |
| 185 | xs.append(v) |
| 186 | ys, v = [], 0 |
| 187 | for f in flags: |
| 188 | if f & 4: |
| 189 | d = b[p]; p += 1 |
| 190 | v += d if f & 32 else -d |
| 191 | elif not f & 32: |
| 192 | v += S16(b, p); p += 2 |
| 193 | ys.append(v) |
| 194 | box = [min(xs), min(ys), max(xs), max(ys)] if xs else None |
| 195 | return (nc, npts, on, True, box) |
| 196 | |
| 197 | |
| 198 | def describe(path): |
| 199 | b = open(path, "rb").read() |
| 200 | tabs = tables(b) |
| 201 | for need in ("head", "hhea", "hmtx", "maxp", "loca", "glyf"): |
| 202 | if need not in tabs: |
| 203 | return None |
| 204 | head = tabs["head"][0] |
| 205 | if U32(b, head + 12) != 0x5F0F3CF5: |
| 206 | return None |
| 207 | upem = U16(b, head + 18) |
| 208 | long_loca = S16(b, head + 50) == 1 |
| 209 | num_glyphs = U16(b, tabs["maxp"][0] + 4) |
| 210 | hhea = tabs["hhea"][0] |
| 211 | nhm = U16(b, hhea + 34) |
| 212 | hmtx = tabs["hmtx"][0] |
| 213 | |
| 214 | def adv(g): |
| 215 | return U16(b, hmtx + min(g, nhm - 1) * 4) if nhm else 0 |
| 216 | |
| 217 | cmap = tabs.get("cmap") |
| 218 | glyphs = {} |
| 219 | if cmap: |
| 220 | for ch in PROBE: |
| 221 | g = cmap_lookup(b, cmap[0], ch) |
| 222 | if g is None: |
| 223 | continue |
| 224 | glyphs[ch] = g |
| 225 | out = { |
| 226 | "file": path, |
| 227 | "units_per_em": upem, |
| 228 | "num_glyphs": num_glyphs, |
| 229 | "ascent": S16(b, hhea + 4), |
| 230 | "descent": S16(b, hhea + 6), |
| 231 | "line_gap": S16(b, hhea + 8), |
| 232 | "bbox": [S16(b, head + 36), S16(b, head + 38), S16(b, head + 40), S16(b, head + 42)], |
| 233 | "has_cmap": bool(cmap), |
| 234 | "glyph_ids": glyphs, |
| 235 | "advances": {str(g): adv(g) for g in sorted(set(glyphs.values()))}, |
| 236 | "contours": {}, |
| 237 | "kerning": {}, |
| 238 | } |
| 239 | for ch, g in sorted(glyphs.items()): |
| 240 | st = contour_stats(b, tabs, g, upem, long_loca, num_glyphs) |
| 241 | if st: |
| 242 | out["contours"][str(g)] = st |
| 243 | # Sample real pairs out of the font's own kern table, spread across it so the |
| 244 | # binary search is exercised at both ends, plus pairs known to be absent. |
| 245 | pairs = kern_pairs(b, tabs.get("kern")) |
| 246 | if pairs: |
| 247 | keys = sorted(pairs) |
| 248 | step = max(1, len(keys) // 40) |
| 249 | for k in keys[::step][:40]: |
| 250 | out["kerning"][f"{k[0]},{k[1]}"] = pairs[k] |
| 251 | out["kerning"][f"{keys[0][0]},{keys[0][1]}"] = pairs[keys[0]] |
| 252 | out["kerning"][f"{keys[-1][0]},{keys[-1][1]}"] = pairs[keys[-1]] |
| 253 | for miss in ((0, 0), (65535, 65534), (7, 65533)): |
| 254 | if miss not in pairs: |
| 255 | out["kerning"][f"{miss[0]},{miss[1]}"] = 0 |
| 256 | return out |
| 257 | |
| 258 | |
| 259 | def emit(d): |
| 260 | w = sys.stdout.write |
| 261 | w(f"FONT {d['file']}\n") |
| 262 | w(f"UPEM {d['units_per_em']}\n") |
| 263 | w(f"NGLYPHS {d['num_glyphs']}\n") |
| 264 | w(f"VMETRICS {d['ascent']} {d['descent']} {d['line_gap']}\n") |
| 265 | w("BBOX %d %d %d %d\n" % tuple(d["bbox"])) |
| 266 | w(f"CMAP {int(d['has_cmap'])}\n") |
| 267 | for ch, g in sorted(d["glyph_ids"].items()): |
| 268 | w(f"GID {ord(ch)} {g}\n") |
| 269 | for g, a in sorted(d["advances"].items(), key=lambda kv: int(kv[0])): |
| 270 | w(f"ADV {g} {a}\n") |
| 271 | for g, st in sorted(d["contours"].items(), key=lambda kv: int(kv[0])): |
| 272 | nc, npts, non, simple, box = st |
| 273 | if simple and box: |
| 274 | w(f"GLYPH {g} {nc} {npts} {non} 1 {box[0]} {box[1]} {box[2]} {box[3]}\n") |
| 275 | else: |
| 276 | w(f"GLYPH {g} {nc} {npts} {non} 0 0 0 0 0\n") |
| 277 | for k, v in sorted(d["kerning"].items()): |
| 278 | l, r = k.split(",") |
| 279 | w(f"KERN {l} {r} {v}\n") |
| 280 | w("END\n") |
| 281 | |
| 282 | |
| 283 | def main(): |
| 284 | n = 0 |
| 285 | for path in sys.argv[1:]: |
| 286 | try: |
| 287 | d = describe(path) |
| 288 | except Exception as exc: |
| 289 | print(f"skip {path}: {exc}", file=sys.stderr) |
| 290 | continue |
| 291 | if d: |
| 292 | emit(d) |
| 293 | n += 1 |
| 294 | print(f"{n} fonts described", file=sys.stderr) |
| 295 | |
| 296 | |
| 297 | main() |