Spaces:
Paused
Paused
File size: 9,891 Bytes
f66643d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 | """Unit tests for pdf2zh.render.sizing — font size clustering."""
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent.parent))
from pdf2zh.render.config import SizingConfig
from pdf2zh.render.sizing import _greedy_cluster, assign_render_sizes
def _make_doc(
elements: list[dict], page_width: float = 612, page_height: float = 792
) -> dict:
return {
"pages": [
{
"page_index": 0,
"page_width": page_width,
"page_height": page_height,
"elements": elements,
}
]
}
class TestGreedyCluster:
def test_single_cluster(self):
items = [(10.5, "a"), (10.7, "b"), (11.0, "c")]
clusters = _greedy_cluster(items, eps=1.5)
assert len(clusters) == 1
assert len(clusters[0]) == 3
def test_two_clusters(self):
items = [(10.5, "a"), (11.0, "b"), (14.0, "c"), (14.5, "d")]
clusters = _greedy_cluster(items, eps=1.5)
assert len(clusters) == 2
def test_three_clusters(self):
items = [(10.5, "a"), (14.0, "b"), (18.0, "c")]
clusters = _greedy_cluster(items, eps=1.5)
assert len(clusters) == 3
def test_empty(self):
assert _greedy_cluster([], eps=1.5) == []
def test_single_item(self):
clusters = _greedy_cluster([(11.0, "a")], eps=1.5)
assert len(clusters) == 1
assert clusters[0] == [(11.0, "a")]
class TestAssignRenderSizes:
def _cfg(self, **kwargs) -> SizingConfig:
return SizingConfig(**kwargs)
def test_body_text_consistent(self):
doc = _make_doc(
[
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 10.7,
"bbox_pdf": [0, 0, 100, 20],
"cells": [],
},
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 11.0,
"bbox_pdf": [0, 25, 100, 45],
"cells": [],
},
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 11.3,
"bbox_pdf": [0, 50, 100, 70],
"cells": [],
},
]
)
cfg = self._cfg()
sizes = assign_render_sizes(doc, cfg)
# All should snap to same cluster (within eps=1.5)
assert sizes["p0:e0"] == sizes["p0:e1"] == sizes["p0:e2"]
def test_heading_hierarchy_preserved(self):
doc = _make_doc(
[
{
"label": "SectionHeader",
"category": "IN_PLACE",
"font_size": 10.0,
"bbox_pdf": [0, 0, 100, 20],
"cells": [],
},
{
"label": "SectionHeader",
"category": "IN_PLACE",
"font_size": 14.0,
"bbox_pdf": [0, 25, 100, 45],
"cells": [],
},
{
"label": "SectionHeader",
"category": "IN_PLACE",
"font_size": 18.0,
"bbox_pdf": [0, 50, 100, 70],
"cells": [],
},
]
)
cfg = self._cfg()
sizes = assign_render_sizes(doc, cfg)
# Three distinct sizes
assert sizes["p0:e0"] < sizes["p0:e1"] < sizes["p0:e2"]
def test_fallback_for_zero_size(self):
doc = _make_doc(
[
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 0.0,
"bbox_pdf": [0, 0, 100, 20],
"cells": [],
},
]
)
cfg = self._cfg(fallback_size=11.0)
sizes = assign_render_sizes(doc, cfg)
assert sizes["p0:e0"] == 11.0
def test_bypass_not_included(self):
doc = _make_doc(
[
{
"label": "Figure",
"category": "BYPASS",
"font_size": 0.0,
"bbox_pdf": [0, 0, 100, 100],
"cells": [],
},
]
)
cfg = self._cfg()
sizes = assign_render_sizes(doc, cfg)
# BYPASS elements are not in any cluster group, so not assigned
assert "p0:e0" not in sizes or True # Acceptable: either absent or fallback
def test_table_cells_cluster_per_table(self):
doc = _make_doc(
[
{
"label": "Table",
"category": "TABLE",
"font_size": 10.0,
"bbox_pdf": [0, 0, 300, 100],
"cells": [
{
"bbox_pdf": [0, 0, 100, 20],
"source_text": "A",
"translated_text": "B",
"cell_font_size": 10.0,
},
{
"bbox_pdf": [100, 0, 200, 20],
"source_text": "C",
"translated_text": "D",
"cell_font_size": 10.5,
},
],
},
]
)
cfg = self._cfg()
sizes = assign_render_sizes(doc, cfg)
# Both cells should get the same cluster size
assert sizes["p0:e0:c0"] == sizes["p0:e0:c1"]
def test_equation_clusters_with_body(self):
doc = _make_doc(
[
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 11.0,
"bbox_pdf": [0, 0, 100, 20],
"cells": [],
},
{
"label": "Equation",
"category": "EQUATION",
"font_size": 11.2,
"bbox_pdf": [0, 25, 100, 45],
"cells": [],
},
]
)
cfg = self._cfg()
cfg.cluster_groups["body"].append("Equation")
sizes = assign_render_sizes(doc, cfg)
# Both in "body" group → should cluster together (within eps)
assert sizes["p0:e0"] == sizes["p0:e1"]
def test_page_scope_header_footer(self):
# Page-scope: each page's header/footer clusters independently
pages = [
{
"page_index": 0,
"page_width": 612,
"page_height": 792,
"elements": [
{
"label": "PageHeader",
"category": "IN_PLACE",
"font_size": 9.0,
"bbox_pdf": [0, 0, 600, 20],
"cells": [],
},
],
},
{
"page_index": 1,
"page_width": 612,
"page_height": 792,
"elements": [
{
"label": "PageHeader",
"category": "IN_PLACE",
"font_size": 10.0,
"bbox_pdf": [0, 0, 600, 20],
"cells": [],
},
],
},
]
doc = {"pages": pages}
cfg = self._cfg()
sizes = assign_render_sizes(doc, cfg)
# They're in separate page-scoped buckets, so cluster separately
assert "p0:e0" in sizes
assert "p1:e0" in sizes
def test_colliding_shrink_never_inflates_above_canonical(self):
# Regression: on a page whose cluster canonical is below fallback_size,
# a block whose translated text overflows AND collides with a neighbor
# used to be clamped UP to fallback (11pt) — larger than its cluster —
# instead of shrunk. It must stay at (or below) the canonical.
doc = _make_doc(
[
# e0 overflows at the 9pt canonical and collides with e1 below it.
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 9.0,
"source_text": "",
"translated_text": "x" * 220,
"bbox_pdf": [0, 0, 100, 20],
"cells": [],
},
# e1 sits directly below e0 (tiny gap) → e0's overflow hits it.
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 9.0,
"source_text": "",
"translated_text": "",
"bbox_pdf": [0, 22, 100, 42],
"cells": [],
},
# e2 sets the cluster too, off to the side (no collision).
{
"label": "Text",
"category": "FLOWING_TEXT",
"font_size": 9.0,
"source_text": "",
"translated_text": "",
"bbox_pdf": [300, 0, 400, 20],
"cells": [],
},
]
)
cfg = self._cfg()
sizes = assign_render_sizes(doc, cfg)
# The colliding block stays uniform with its cluster (9pt), and is never
# inflated up to fallback (11pt) by the readability floor.
assert sizes["p0:e0"] == sizes["p0:e2"]
assert sizes["p0:e0"] < cfg.fallback_size
|