File size: 8,630 Bytes
f9bc0c1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from __future__ import annotations

import importlib.util
import json
import sys
from argparse import Namespace
from pathlib import Path


def _load_tool():
    path = Path(__file__).resolve().parents[1] / "tools" / "apply_page_reading_order_geometry_profile.py"
    spec = importlib.util.spec_from_file_location("apply_page_reading_order_geometry_profile", path)
    assert spec is not None
    assert spec.loader is not None
    module = importlib.util.module_from_spec(spec)
    sys.modules[spec.name] = module
    spec.loader.exec_module(module)
    return module


def _write_document(root: Path) -> None:
    doc_dir = root / "sample-1"
    doc_dir.mkdir(parents=True)
    lines = [
        ("right-low", "right low", [500, 100, 40, 10]),
        ("left-top", "left top", [100, 10, 40, 10]),
        ("right-top", "right top", [500, 10, 40, 10]),
    ]
    document_lines = [
        {"line_id": line_id, "page_index": 0, "text": text, "bbox": bbox, "confidence": 0.9, "metadata": {}}
        for line_id, text, bbox in lines
    ]
    blocks = [
        {
            "block_id": f"block-{index}",
            "block_type": "text",
            "page_index": 0,
            "bbox": bbox,
            "order": index,
            "text": text,
            "confidence": 0.9,
            "line_ids": [line_id],
            "table_id": None,
            "figure_id": None,
            "metadata": {},
        }
        for index, (line_id, text, bbox) in enumerate(lines, start=1)
    ]
    payload = {
        "schema_version": "1.0",
        "source_path": "page.png",
        "tables": [],
        "figures": [],
        "metadata": {},
        "pages": [{"page_index": 0, "width": 800, "height": 200, "metadata": {}, "text_lines": document_lines, "blocks": blocks}],
    }
    (doc_dir / "document.json").write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")


def _write_document_with_sample(root: Path, sample_id: str) -> None:
    doc_dir = root / sample_id
    doc_dir.mkdir(parents=True)
    lines = [
        ("a", f"{sample_id} a", [100, 10, 40, 10]),
        ("b", f"{sample_id} b", [500, 10, 40, 10]),
    ]
    payload = {
        "schema_version": "1.0",
        "source_path": f"{sample_id}.png",
        "tables": [],
        "figures": [],
        "metadata": {},
        "pages": [
            {
                "page_index": 0,
                "width": 800,
                "height": 200,
                "metadata": {},
                "text_lines": [
                    {"line_id": line_id, "page_index": 0, "text": text, "bbox": bbox, "confidence": 0.9, "metadata": {}}
                    for line_id, text, bbox in lines
                ],
                "blocks": [
                    {
                        "block_id": f"block-{index}",
                        "block_type": "text",
                        "page_index": 0,
                        "bbox": bbox,
                        "order": index,
                        "text": text,
                        "confidence": 0.9,
                        "line_ids": [line_id],
                        "table_id": None,
                        "figure_id": None,
                        "metadata": {},
                    }
                    for index, (line_id, text, bbox) in enumerate(lines, start=1)
                ],
            }
        ],
    }
    (doc_dir / "document.json").write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")


def test_apply_page_reading_order_geometry_profile_orders_by_chunks(tmp_path: Path) -> None:
    tool = _load_tool()
    source = tmp_path / "docs"
    _write_document(source)
    profile = tmp_path / "profile.json"
    profile.write_text(
        json.dumps(
            {
                "chunks": [
                    {"chunk_index": 1, "cx_median": 520, "y_min": 0, "y_max": 120, "width_median": 40},
                    {"chunk_index": 2, "cx_median": 120, "y_min": 0, "y_max": 120, "width_median": 40},
                ]
            },
            ensure_ascii=False,
        )
        + "\n",
        encoding="utf-8",
    )

    report = tool.apply_profile(
        Namespace(
            source_documents_dir=source,
            profile=profile,
            out=tmp_path / "out",
            x_scale=1.0,
            y_scale=80.0,
            outside_weight=2.0,
        )
    )

    assert report["assignment_count"] == 3
    document = json.loads((tmp_path / "out" / "sample-1" / "document.json").read_text(encoding="utf-8"))
    page = document["pages"][0]
    assert [line["text"] for line in page["text_lines"]] == ["right top", "right low", "left top"]
    assert page["text_lines"][0]["metadata"]["reading_order_geometry_profile_assignment"]["chunk_index"] == 1
    assert (tmp_path / "out" / "reading-order-geometry-profile-assignments.jsonl").is_file()


def test_apply_page_reading_order_geometry_profile_requires_claim_safe_when_requested(tmp_path: Path) -> None:
    tool = _load_tool()
    source = tmp_path / "docs"
    _write_document(source)
    profile = tmp_path / "profile.json"
    profile.write_text(
        json.dumps(
            {
                "profile_claim_safe": False,
                "chunks": [
                    {"chunk_index": 1, "cx_median": 520, "y_min": 0, "y_max": 120, "width_median": 40},
                ],
            },
            ensure_ascii=False,
        )
        + "\n",
        encoding="utf-8",
    )

    try:
        tool.apply_profile(
            Namespace(
                source_documents_dir=source,
                profile=profile,
                out=tmp_path / "out",
                x_scale=1.0,
                y_scale=80.0,
                outside_weight=2.0,
                require_profile_claim_safe=True,
            )
        )
    except ValueError as exc:
        assert "profile_claim_safe is not true" in str(exc)
    else:
        raise AssertionError("unsafe profile should be rejected when claim-safe profile is required")


def test_apply_page_reading_order_geometry_profile_uses_profile_map_per_sample(tmp_path: Path) -> None:
    tool = _load_tool()
    source = tmp_path / "docs"
    _write_document_with_sample(source, "left-first")
    _write_document_with_sample(source, "right-first")
    left_profile = tmp_path / "left-profile.json"
    right_profile = tmp_path / "right-profile.json"
    left_profile.write_text(
        json.dumps(
            {
                "profile_claim_safe": True,
                "chunks": [
                    {"chunk_index": 1, "cx_median": 120, "y_min": 0, "y_max": 100, "width_median": 40},
                    {"chunk_index": 2, "cx_median": 520, "y_min": 0, "y_max": 100, "width_median": 40},
                ],
            },
            ensure_ascii=False,
        )
        + "\n",
        encoding="utf-8",
    )
    right_profile.write_text(
        json.dumps(
            {
                "profile_claim_safe": True,
                "chunks": [
                    {"chunk_index": 1, "cx_median": 520, "y_min": 0, "y_max": 100, "width_median": 40},
                    {"chunk_index": 2, "cx_median": 120, "y_min": 0, "y_max": 100, "width_median": 40},
                ],
            },
            ensure_ascii=False,
        )
        + "\n",
        encoding="utf-8",
    )
    profile_map = tmp_path / "profile-map.json"
    profile_map.write_text(
        json.dumps(
            {
                "profiles": [
                    {"sample_id": "left-first", "profile": str(left_profile)},
                    {"sample_id": "right-first", "profile": str(right_profile)},
                ]
            },
            ensure_ascii=False,
            indent=2,
        )
        + "\n",
        encoding="utf-8",
    )

    report = tool.apply_profile(
        Namespace(
            source_documents_dir=source,
            profile=None,
            profile_map=profile_map,
            out=tmp_path / "out",
            x_scale=1.0,
            y_scale=80.0,
            outside_weight=2.0,
            require_profile_claim_safe=True,
        )
    )

    assert report["profile_claim_safe"] is True
    assert len(report["profile_usage"]) == 2
    left_doc = json.loads((tmp_path / "out" / "left-first" / "document.json").read_text(encoding="utf-8"))
    right_doc = json.loads((tmp_path / "out" / "right-first" / "document.json").read_text(encoding="utf-8"))
    assert [line["text"] for line in left_doc["pages"][0]["text_lines"]] == ["left-first a", "left-first b"]
    assert [line["text"] for line in right_doc["pages"][0]["text_lines"]] == ["right-first b", "right-first a"]