this repo has no description
Track regenerated typeset splits with dark mode
+66
-34
+2
-2
data/typeset_mixed_test.dvc
+2
-2
data/typeset_mixed_test.dvc
+2
-2
data/typeset_mixed_train.dvc
+2
-2
data/typeset_mixed_train.dvc
+2
-2
data/typeset_mixed_val.dvc
+2
-2
data/typeset_mixed_val.dvc
+2
-2
data/typeset_prose_test.dvc
+2
-2
data/typeset_prose_test.dvc
+3
-3
data/typeset_prose_train.dvc
+3
-3
data/typeset_prose_train.dvc
+2
-2
data/typeset_prose_val.dvc
+2
-2
data/typeset_prose_val.dvc
+2
-2
data/typeset_uniform_test.dvc
+2
-2
data/typeset_uniform_test.dvc
+2
-2
data/typeset_uniform_train.dvc
+2
-2
data/typeset_uniform_train.dvc
+2
-2
data/typeset_uniform_val.dvc
+2
-2
data/typeset_uniform_val.dvc
+1
-1
src/augment_vis.py
+1
-1
src/augment_vis.py
···
31
31
description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter
32
32
)
33
33
parser.add_argument("--split", default="typeset_mixed_train")
34
-
parser.add_argument("--n", type=int, default=12)
34
+
parser.add_argument("--n", type=int, default=20)
35
35
parser.add_argument("--seed", type=int, default=42)
36
36
parser.add_argument("--out", default="/tmp/augment_vis")
37
37
parser.add_argument("--images", nargs="+", metavar="IMG",
+46
-14
src/generate_typeset.py
+46
-14
src/generate_typeset.py
···
61
61
# "blackboard-style"; smaller sizes give tighter notes appearance.
62
62
_FONT_SIZES = [10, 10, 11, 11, 11, 12, 12, 13, 14]
63
63
64
-
# Ink colours: mostly black, occasional dark blue / dark grey
64
+
# Light-mode ink: mostly black, occasional coloured ink
65
65
_INK_COLOURS = [
66
-
"#000000", "#000000", "#000000", "#000000", # 4/7 pure black
66
+
"#000000", "#000000", "#000000", "#000000", # 4/10 pure black
67
67
"#1a1a1a", # near-black
68
68
"#0d0d4d", # dark navy (pen ink)
69
69
"#2b2b2b", # dark grey
70
+
"#1a4d1a", # dark green
71
+
"#5c0000", # dark red / maroon
72
+
"#2d0060", # dark purple
70
73
]
71
74
75
+
# Light-mode page fills: mostly white, occasional warm/cool tints
76
+
_PAGE_FILLS_LIGHT = [
77
+
"#ffffff", "#ffffff", "#ffffff", "#ffffff", # 4/7 pure white
78
+
"#fffef0", # warm cream
79
+
"#fdf6e3", # solarized-light cream
80
+
"#f0f4ff", # very light blue
81
+
]
82
+
83
+
# Dark-mode variants: light ink on dark background
84
+
_INK_COLOURS_DARK = [
85
+
"#ffffff", "#ffffff", "#ffffff", # pure white
86
+
"#f0f0f0", # near-white
87
+
"#e8e8ff", # blue-white (chalk feel)
88
+
"#d0ffd0", # light green (green chalkboard)
89
+
]
90
+
91
+
_PAGE_FILLS_DARK = [
92
+
"#1a1a1a", "#1a1a1a", "#1a1a1a", # near-black
93
+
"#1e1e1e", # editor dark
94
+
"#0d1117", # deep black
95
+
"#2d2d2d", # dark grey
96
+
]
97
+
98
+
DARK_MODE_PROB = 0.50
99
+
72
100
73
101
def _available_fonts(font_dir: Path) -> list[str]:
74
102
"""Return Typst family names whose TTF files exist in font_dir."""
···
90
118
# Math equations continue using Typst's built-in math fonts (realistic).
91
119
_TEMPLATE_HW = (
92
120
"#set page(width: {width}, height: auto, "
93
-
"margin: (x: 12pt, y: 12pt), fill: white)\n"
121
+
"margin: (x: 12pt, y: 12pt), fill: rgb(\"{fill}\"))\n"
94
122
'#set text(font: ("{font}", "New Computer Modern"), '
95
123
"size: {size}pt, fill: rgb(\"{ink}\"), fallback: true)\n"
96
124
"#set list(spacing: 1.2em)\n"
···
102
130
# wrapped in #text(font: ...) -- handled in _apply_mixed_fonts().
103
131
_TEMPLATE_MIX = (
104
132
"#set page(width: {width}, height: auto, "
105
-
"margin: (x: 12pt, y: 12pt), fill: white)\n"
133
+
"margin: (x: 12pt, y: 12pt), fill: rgb(\"{fill}\"))\n"
134
+
"#set text(fill: rgb(\"{ink}\"))\n"
106
135
"#set list(spacing: 1.0em)\n"
107
136
"#set enum(spacing: 1.0em)\n"
108
137
"{body}\n"
···
153
182
font_dir: Path,
154
183
size: int,
155
184
ink: str,
185
+
fill: str,
156
186
rng: random.Random,
157
187
) -> tuple[bool, str]:
158
188
"""
···
163
193
"""
164
194
if mode == "uniform" and font is not None:
165
195
src = _TEMPLATE_HW.format(
166
-
width=page_width, font=font, size=size, ink=ink, body=body
196
+
width=page_width, font=font, size=size, ink=ink, fill=fill, body=body
167
197
)
168
198
else:
169
199
# font=None (default) or mix mode: use plain template
170
-
src = _TEMPLATE_MIX.format(width=page_width, body=body)
200
+
src = _TEMPLATE_MIX.format(width=page_width, ink=ink, fill=fill, body=body)
171
201
172
202
with tempfile.NamedTemporaryFile(suffix=".typ", mode="w", delete=False) as f:
173
203
f.write(src)
···
257
287
seen.add(body)
258
288
font = _pick_font(rng, available)
259
289
size = rng.choice(_FONT_SIZES)
260
-
ink = rng.choice(_INK_COLOURS)
261
-
candidates.append((body, page_width, font, size, ink))
290
+
dark = rng.random() < DARK_MODE_PROB
291
+
ink = rng.choice(_INK_COLOURS_DARK if dark else _INK_COLOURS)
292
+
fill = rng.choice(_PAGE_FILLS_DARK if dark else _PAGE_FILLS_LIGHT)
293
+
candidates.append((body, page_width, font, size, ink, fill))
262
294
pbar.update(1)
263
295
264
296
print(f" {attempts:,} attempts ({attempts / len(candidates):.1f}x overhead)")
···
271
303
shown_failures = 0
272
304
273
305
def _task(
274
-
clean_body: str, page_width: str, font: str, size: int, ink: str
306
+
clean_body: str, page_width: str, font: str, size: int, ink: str, fill: str
275
307
) -> tuple[str, str, bool, str]:
276
308
# prose renders as a uniform-font document (one font, whole page)
277
309
if args.mode in ("uniform", "prose"):
···
281
313
clean_body, rng, font, hw_prob=args.hw_prob, size=size, ink=ink
282
314
)
283
315
284
-
# Hash over render body + font + size + ink + mode for uniqueness
285
-
h_key = f"{args.mode}:{font}:{size}:{ink}:{page_width}:{render_body}"
316
+
# Hash over render body + font + size + ink + fill + mode for uniqueness
317
+
h_key = f"{args.mode}:{font}:{size}:{ink}:{fill}:{page_width}:{render_body}"
286
318
h = hashlib.sha1(h_key.encode()).hexdigest()[:16]
287
319
out_path = img_dir / f"{h}.png"
288
320
···
290
322
ok, err = _render(
291
323
render_body, out_path, page_width,
292
324
mode=render_mode, font=font, font_dir=font_dir, size=size, ink=ink,
293
-
rng=rng,
325
+
fill=fill, rng=rng,
294
326
)
295
327
# Manifest always records the CLEAN body (no font directives).
296
328
return clean_body, f"images/{h}.png", ok, err, font
297
329
298
330
with ThreadPoolExecutor(max_workers=args.jobs) as pool:
299
331
futs = {
300
-
pool.submit(_task, body, pw, font, size, ink): body
301
-
for body, pw, font, size, ink in candidates
332
+
pool.submit(_task, body, pw, font, size, ink, fill): body
333
+
for body, pw, font, size, ink, fill in candidates
302
334
}
303
335
with tqdm(total=len(candidates), unit="img") as pbar:
304
336
for fut in as_completed(futs):