|
1
|
"""Generate PDF reports from knowledge graph data. |
|
2
|
|
|
3
|
Uses reportlab for PDF generation. Falls back gracefully if not installed. |
|
4
|
No LLM required — pure template-based generation from KG data. |
|
5
|
""" |
|
6
|
|
|
7
|
import logging |
|
8
|
from datetime import datetime |
|
9
|
from pathlib import Path |
|
10
|
from typing import Any, Dict, List, Optional |
|
11
|
|
|
12
|
logger = logging.getLogger(__name__) |
|
13
|
|
|
14
|
|
|
15
|
def _get_styles(): |
|
16
|
"""Import and configure reportlab styles.""" |
|
17
|
from reportlab.lib import colors |
|
18
|
from reportlab.lib.pagesizes import letter |
|
19
|
from reportlab.lib.styles import ParagraphStyle, getSampleStyleSheet |
|
20
|
from reportlab.lib.units import inch |
|
21
|
|
|
22
|
styles = getSampleStyleSheet() |
|
23
|
|
|
24
|
styles.add( |
|
25
|
ParagraphStyle( |
|
26
|
"KGTitle", |
|
27
|
parent=styles["Title"], |
|
28
|
fontSize=24, |
|
29
|
spaceAfter=20, |
|
30
|
textColor=colors.HexColor("#1a1a2e"), |
|
31
|
) |
|
32
|
) |
|
33
|
styles.add( |
|
34
|
ParagraphStyle( |
|
35
|
"KGHeading2", |
|
36
|
parent=styles["Heading2"], |
|
37
|
fontSize=16, |
|
38
|
spaceBefore=16, |
|
39
|
spaceAfter=8, |
|
40
|
textColor=colors.HexColor("#16213e"), |
|
41
|
) |
|
42
|
) |
|
43
|
styles.add( |
|
44
|
ParagraphStyle( |
|
45
|
"KGBody", |
|
46
|
parent=styles["Normal"], |
|
47
|
fontSize=10, |
|
48
|
leading=14, |
|
49
|
spaceBefore=4, |
|
50
|
spaceAfter=4, |
|
51
|
) |
|
52
|
) |
|
53
|
styles.add( |
|
54
|
ParagraphStyle( |
|
55
|
"KGBullet", |
|
56
|
parent=styles["Normal"], |
|
57
|
fontSize=10, |
|
58
|
leading=14, |
|
59
|
leftIndent=20, |
|
60
|
bulletIndent=10, |
|
61
|
spaceBefore=2, |
|
62
|
spaceAfter=2, |
|
63
|
) |
|
64
|
) |
|
65
|
|
|
66
|
return styles, letter, inch, colors |
|
67
|
|
|
68
|
|
|
69
|
def _build_entity_table(nodes: List[dict], colors) -> Any: |
|
70
|
"""Build a table of entities grouped by type.""" |
|
71
|
from reportlab.lib.units import inch |
|
72
|
from reportlab.platypus import Table, TableStyle |
|
73
|
|
|
74
|
by_type: Dict[str, list] = {} |
|
75
|
for n in nodes: |
|
76
|
t = n.get("type", "concept") |
|
77
|
by_type.setdefault(t, []).append(n) |
|
78
|
|
|
79
|
data = [["Type", "Count", "Examples"]] |
|
80
|
for etype, elist in sorted(by_type.items(), key=lambda x: -len(x[1])): |
|
81
|
examples = ", ".join(e.get("name", "") for e in elist[:3]) |
|
82
|
if len(elist) > 3: |
|
83
|
examples += f" (+{len(elist) - 3} more)" |
|
84
|
data.append([etype.title(), str(len(elist)), examples]) |
|
85
|
|
|
86
|
table = Table(data, colWidths=[1.2 * inch, 0.8 * inch, 4.0 * inch]) |
|
87
|
table.setStyle( |
|
88
|
TableStyle( |
|
89
|
[ |
|
90
|
("BACKGROUND", (0, 0), (-1, 0), colors.HexColor("#e8eaf6")), |
|
91
|
("FONTNAME", (0, 0), (-1, 0), "Helvetica-Bold"), |
|
92
|
("FONTSIZE", (0, 0), (-1, -1), 9), |
|
93
|
("GRID", (0, 0), (-1, -1), 0.5, colors.grey), |
|
94
|
("VALIGN", (0, 0), (-1, -1), "TOP"), |
|
95
|
("TOPPADDING", (0, 0), (-1, -1), 4), |
|
96
|
("BOTTOMPADDING", (0, 0), (-1, -1), 4), |
|
97
|
] |
|
98
|
) |
|
99
|
) |
|
100
|
return table |
|
101
|
|
|
102
|
|
|
103
|
def _build_relationship_table(rels: List[dict], colors, max_rows: int = 30) -> Any: |
|
104
|
"""Build a table of relationships.""" |
|
105
|
from reportlab.lib.units import inch |
|
106
|
from reportlab.platypus import Table, TableStyle |
|
107
|
|
|
108
|
data = [["Source", "Relationship", "Target"]] |
|
109
|
for r in rels[:max_rows]: |
|
110
|
data.append([r.get("source", ""), r.get("type", ""), r.get("target", "")]) |
|
111
|
if len(rels) > max_rows: |
|
112
|
data.append(["...", f"({len(rels) - max_rows} more)", "..."]) |
|
113
|
|
|
114
|
table = Table(data, colWidths=[2.0 * inch, 2.0 * inch, 2.0 * inch]) |
|
115
|
table.setStyle( |
|
116
|
TableStyle( |
|
117
|
[ |
|
118
|
("BACKGROUND", (0, 0), (-1, 0), colors.HexColor("#e8eaf6")), |
|
119
|
("FONTNAME", (0, 0), (-1, 0), "Helvetica-Bold"), |
|
120
|
("FONTSIZE", (0, 0), (-1, -1), 9), |
|
121
|
("GRID", (0, 0), (-1, -1), 0.5, colors.grey), |
|
122
|
("VALIGN", (0, 0), (-1, -1), "TOP"), |
|
123
|
("TOPPADDING", (0, 0), (-1, -1), 4), |
|
124
|
("BOTTOMPADDING", (0, 0), (-1, -1), 4), |
|
125
|
] |
|
126
|
) |
|
127
|
) |
|
128
|
return table |
|
129
|
|
|
130
|
|
|
131
|
def _build_key_entities_table(rels: List[dict], colors) -> Any: |
|
132
|
"""Build a table of top entities by connection count.""" |
|
133
|
from reportlab.lib.units import inch |
|
134
|
from reportlab.platypus import Table, TableStyle |
|
135
|
|
|
136
|
degree: Dict[str, int] = {} |
|
137
|
for r in rels: |
|
138
|
degree[r.get("source", "")] = degree.get(r.get("source", ""), 0) + 1 |
|
139
|
degree[r.get("target", "")] = degree.get(r.get("target", ""), 0) + 1 |
|
140
|
|
|
141
|
top = sorted(degree.items(), key=lambda x: -x[1])[:10] |
|
142
|
if not top: |
|
143
|
return None |
|
144
|
|
|
145
|
data = [["Entity", "Connections"]] |
|
146
|
for name, deg in top: |
|
147
|
data.append([name, str(deg)]) |
|
148
|
|
|
149
|
table = Table(data, colWidths=[4.0 * inch, 1.5 * inch]) |
|
150
|
table.setStyle( |
|
151
|
TableStyle( |
|
152
|
[ |
|
153
|
("BACKGROUND", (0, 0), (-1, 0), colors.HexColor("#e8eaf6")), |
|
154
|
("FONTNAME", (0, 0), (-1, 0), "Helvetica-Bold"), |
|
155
|
("FONTSIZE", (0, 0), (-1, -1), 9), |
|
156
|
("GRID", (0, 0), (-1, -1), 0.5, colors.grey), |
|
157
|
("TOPPADDING", (0, 0), (-1, -1), 4), |
|
158
|
("BOTTOMPADDING", (0, 0), (-1, -1), 4), |
|
159
|
] |
|
160
|
) |
|
161
|
) |
|
162
|
return table |
|
163
|
|
|
164
|
|
|
165
|
def generate_pdf( |
|
166
|
kg_data: dict, |
|
167
|
output_path: Path, |
|
168
|
title: Optional[str] = None, |
|
169
|
diagrams_dir: Optional[Path] = None, |
|
170
|
) -> Path: |
|
171
|
"""Generate a PDF report from knowledge graph data. |
|
172
|
|
|
173
|
Args: |
|
174
|
kg_data: Knowledge graph dict with 'nodes' and 'relationships'. |
|
175
|
output_path: Path to write the PDF file. |
|
176
|
title: Optional report title. |
|
177
|
diagrams_dir: Optional directory containing diagram images to embed. |
|
178
|
|
|
179
|
Returns: |
|
180
|
Path to the generated PDF. |
|
181
|
|
|
182
|
Raises: |
|
183
|
ImportError: If reportlab is not installed. |
|
184
|
""" |
|
185
|
from reportlab.platypus import ( |
|
186
|
Paragraph, |
|
187
|
SimpleDocTemplate, |
|
188
|
Spacer, |
|
189
|
) |
|
190
|
|
|
191
|
styles, letter, inch, colors = _get_styles() |
|
192
|
|
|
193
|
output_path = Path(output_path) |
|
194
|
output_path.parent.mkdir(parents=True, exist_ok=True) |
|
195
|
|
|
196
|
doc = SimpleDocTemplate( |
|
197
|
str(output_path), |
|
198
|
pagesize=letter, |
|
199
|
topMargin=0.75 * inch, |
|
200
|
bottomMargin=0.75 * inch, |
|
201
|
leftMargin=0.75 * inch, |
|
202
|
rightMargin=0.75 * inch, |
|
203
|
) |
|
204
|
|
|
205
|
story = [] |
|
206
|
nodes = kg_data.get("nodes", []) |
|
207
|
rels = kg_data.get("relationships", []) |
|
208
|
|
|
209
|
# Title |
|
210
|
report_title = title or "Knowledge Graph Report" |
|
211
|
story.append(Paragraph(report_title, styles["KGTitle"])) |
|
212
|
story.append( |
|
213
|
Paragraph( |
|
214
|
f"Generated {datetime.now().strftime('%Y-%m-%d %H:%M')} • " |
|
215
|
f"{len(nodes)} entities • {len(rels)} relationships", |
|
216
|
styles["KGBody"], |
|
217
|
) |
|
218
|
) |
|
219
|
story.append(Spacer(1, 20)) |
|
220
|
|
|
221
|
# Entity breakdown |
|
222
|
if nodes: |
|
223
|
story.append(Paragraph("Entity Breakdown", styles["KGHeading2"])) |
|
224
|
story.append(_build_entity_table(nodes, colors)) |
|
225
|
story.append(Spacer(1, 12)) |
|
226
|
|
|
227
|
# Key entities |
|
228
|
if rels: |
|
229
|
key_table = _build_key_entities_table(rels, colors) |
|
230
|
if key_table: |
|
231
|
story.append(Paragraph("Key Entities (by connections)", styles["KGHeading2"])) |
|
232
|
story.append(key_table) |
|
233
|
story.append(Spacer(1, 12)) |
|
234
|
|
|
235
|
# Embed diagram images |
|
236
|
if diagrams_dir and diagrams_dir.exists(): |
|
237
|
_embed_diagrams(story, styles, diagrams_dir, inch) |
|
238
|
|
|
239
|
# Relationship table |
|
240
|
if rels: |
|
241
|
story.append(Paragraph("Relationships", styles["KGHeading2"])) |
|
242
|
story.append(_build_relationship_table(rels, colors)) |
|
243
|
story.append(Spacer(1, 12)) |
|
244
|
|
|
245
|
# Entity details |
|
246
|
if nodes: |
|
247
|
story.append(Paragraph("Entity Details", styles["KGHeading2"])) |
|
248
|
for node in sorted(nodes, key=lambda n: n.get("name", "")): |
|
249
|
name = node.get("name", "") |
|
250
|
etype = node.get("type", "concept") |
|
251
|
descs = node.get("descriptions", []) |
|
252
|
desc = descs[0] if descs else "No description." |
|
253
|
story.append(Paragraph(f"<b>{name}</b> <i>({etype})</i>: {desc}", styles["KGBullet"])) |
|
254
|
|
|
255
|
doc.build(story) |
|
256
|
logger.info(f"Generated PDF report: {output_path}") |
|
257
|
return output_path |
|
258
|
|
|
259
|
|
|
260
|
def _embed_diagrams(story, styles, diagrams_dir: Path, inch): |
|
261
|
"""Embed diagram PNG images from a directory.""" |
|
262
|
from reportlab.platypus import Image, Paragraph, Spacer |
|
263
|
|
|
264
|
pngs = sorted(diagrams_dir.glob("*.png")) |
|
265
|
if not pngs: |
|
266
|
return |
|
267
|
|
|
268
|
story.append(Paragraph("Diagrams", styles["KGHeading2"])) |
|
269
|
|
|
270
|
for png in pngs: |
|
271
|
try: |
|
272
|
img = Image(str(png), width=5 * inch, height=3.5 * inch) |
|
273
|
img.hAlign = "CENTER" |
|
274
|
story.append(img) |
|
275
|
story.append(Spacer(1, 8)) |
|
276
|
except Exception as e: |
|
277
|
logger.warning(f"Could not embed diagram {png.name}: {e}") |
|
278
|
|