|
4be3595…
|
lmata
|
1 |
""" |
|
4be3595…
|
lmata
|
2 |
KnowledgeIngester — manual curation of business concepts, rules, decisions, |
|
4be3595…
|
lmata
|
3 |
people, and domains into the navegador graph. |
|
4be3595…
|
lmata
|
4 |
|
|
4be3595…
|
lmata
|
5 |
These are the things that don't live in code but belong in the knowledge graph: |
|
4be3595…
|
lmata
|
6 |
business rules, architectural decisions, domain groupings, ownership, wiki refs. |
|
4be3595…
|
lmata
|
7 |
""" |
|
4be3595…
|
lmata
|
8 |
|
|
4be3595…
|
lmata
|
9 |
import logging |
|
4be3595…
|
lmata
|
10 |
from typing import Any |
|
4be3595…
|
lmata
|
11 |
|
|
4be3595…
|
lmata
|
12 |
from navegador.graph.schema import EdgeType, NodeLabel |
|
4be3595…
|
lmata
|
13 |
from navegador.graph.store import GraphStore |
|
4be3595…
|
lmata
|
14 |
|
|
4be3595…
|
lmata
|
15 |
logger = logging.getLogger(__name__) |
|
4be3595…
|
lmata
|
16 |
|
|
4be3595…
|
lmata
|
17 |
|
|
4be3595…
|
lmata
|
18 |
class KnowledgeIngester: |
|
4be3595…
|
lmata
|
19 |
""" |
|
4be3595…
|
lmata
|
20 |
Writes business knowledge nodes and relationships into the graph. |
|
4be3595…
|
lmata
|
21 |
|
|
4be3595…
|
lmata
|
22 |
Usage: |
|
4be3595…
|
lmata
|
23 |
store = GraphStore.sqlite(".navegador/graph.db") |
|
4be3595…
|
lmata
|
24 |
k = KnowledgeIngester(store) |
|
4be3595…
|
lmata
|
25 |
|
|
4be3595…
|
lmata
|
26 |
k.add_domain("auth", description="Authentication and authorisation") |
|
4be3595…
|
lmata
|
27 |
k.add_concept("JWT", description="Stateless token auth", domain="auth") |
|
4be3595…
|
lmata
|
28 |
k.add_rule("tokens must expire", domain="auth", severity="critical", |
|
4be3595…
|
lmata
|
29 |
rationale="Security requirement per OWASP") |
|
4be3595…
|
lmata
|
30 |
k.annotate_code("validate_token", "Function", concept="JWT") |
|
4be3595…
|
lmata
|
31 |
k.wiki_page("JWT Auth", url="...", content="...") |
|
4be3595…
|
lmata
|
32 |
""" |
|
4be3595…
|
lmata
|
33 |
|
|
4be3595…
|
lmata
|
34 |
def __init__(self, store: GraphStore) -> None: |
|
4be3595…
|
lmata
|
35 |
self.store = store |
|
4be3595…
|
lmata
|
36 |
|
|
4be3595…
|
lmata
|
37 |
# ── Domains ─────────────────────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
38 |
|
|
4be3595…
|
lmata
|
39 |
def add_domain(self, name: str, description: str = "") -> None: |
|
758c6b7…
|
lmata
|
40 |
self.store.create_node( |
|
758c6b7…
|
lmata
|
41 |
NodeLabel.Domain, |
|
758c6b7…
|
lmata
|
42 |
{ |
|
758c6b7…
|
lmata
|
43 |
"name": name, |
|
758c6b7…
|
lmata
|
44 |
"description": description, |
|
758c6b7…
|
lmata
|
45 |
}, |
|
758c6b7…
|
lmata
|
46 |
) |
|
4be3595…
|
lmata
|
47 |
logger.info("Domain: %s", name) |
|
4be3595…
|
lmata
|
48 |
|
|
4be3595…
|
lmata
|
49 |
# ── Concepts ────────────────────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
50 |
|
|
4be3595…
|
lmata
|
51 |
def add_concept( |
|
4be3595…
|
lmata
|
52 |
self, |
|
4be3595…
|
lmata
|
53 |
name: str, |
|
4be3595…
|
lmata
|
54 |
description: str = "", |
|
4be3595…
|
lmata
|
55 |
domain: str = "", |
|
4be3595…
|
lmata
|
56 |
status: str = "", |
|
4be3595…
|
lmata
|
57 |
rules: str = "", |
|
4be3595…
|
lmata
|
58 |
examples: str = "", |
|
4be3595…
|
lmata
|
59 |
wiki_refs: str = "", |
|
4be3595…
|
lmata
|
60 |
) -> None: |
|
758c6b7…
|
lmata
|
61 |
self.store.create_node( |
|
758c6b7…
|
lmata
|
62 |
NodeLabel.Concept, |
|
758c6b7…
|
lmata
|
63 |
{ |
|
758c6b7…
|
lmata
|
64 |
"name": name, |
|
758c6b7…
|
lmata
|
65 |
"description": description, |
|
758c6b7…
|
lmata
|
66 |
"domain": domain, |
|
758c6b7…
|
lmata
|
67 |
"status": status, |
|
758c6b7…
|
lmata
|
68 |
"rules": rules, |
|
758c6b7…
|
lmata
|
69 |
"examples": examples, |
|
758c6b7…
|
lmata
|
70 |
"wiki_refs": wiki_refs, |
|
758c6b7…
|
lmata
|
71 |
}, |
|
758c6b7…
|
lmata
|
72 |
) |
|
4be3595…
|
lmata
|
73 |
if domain: |
|
4be3595…
|
lmata
|
74 |
self._link_to_domain(name, NodeLabel.Concept, domain) |
|
4be3595…
|
lmata
|
75 |
logger.info("Concept: %s", name) |
|
4be3595…
|
lmata
|
76 |
|
|
4be3595…
|
lmata
|
77 |
def relate_concepts(self, a: str, b: str) -> None: |
|
4be3595…
|
lmata
|
78 |
"""Mark two concepts as related (bidirectional intent).""" |
|
4be3595…
|
lmata
|
79 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
80 |
NodeLabel.Concept, |
|
758c6b7…
|
lmata
|
81 |
{"name": a}, |
|
4be3595…
|
lmata
|
82 |
EdgeType.RELATED_TO, |
|
758c6b7…
|
lmata
|
83 |
NodeLabel.Concept, |
|
758c6b7…
|
lmata
|
84 |
{"name": b}, |
|
4be3595…
|
lmata
|
85 |
) |
|
4be3595…
|
lmata
|
86 |
|
|
4be3595…
|
lmata
|
87 |
# ── Rules ───────────────────────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
88 |
|
|
4be3595…
|
lmata
|
89 |
def add_rule( |
|
4be3595…
|
lmata
|
90 |
self, |
|
4be3595…
|
lmata
|
91 |
name: str, |
|
4be3595…
|
lmata
|
92 |
description: str = "", |
|
4be3595…
|
lmata
|
93 |
domain: str = "", |
|
4be3595…
|
lmata
|
94 |
severity: str = "info", |
|
4be3595…
|
lmata
|
95 |
rationale: str = "", |
|
4be3595…
|
lmata
|
96 |
examples: str = "", |
|
4be3595…
|
lmata
|
97 |
) -> None: |
|
758c6b7…
|
lmata
|
98 |
self.store.create_node( |
|
758c6b7…
|
lmata
|
99 |
NodeLabel.Rule, |
|
758c6b7…
|
lmata
|
100 |
{ |
|
758c6b7…
|
lmata
|
101 |
"name": name, |
|
758c6b7…
|
lmata
|
102 |
"description": description, |
|
758c6b7…
|
lmata
|
103 |
"domain": domain, |
|
758c6b7…
|
lmata
|
104 |
"severity": severity, |
|
758c6b7…
|
lmata
|
105 |
"rationale": rationale, |
|
758c6b7…
|
lmata
|
106 |
"examples": examples, |
|
758c6b7…
|
lmata
|
107 |
}, |
|
758c6b7…
|
lmata
|
108 |
) |
|
4be3595…
|
lmata
|
109 |
if domain: |
|
4be3595…
|
lmata
|
110 |
self._link_to_domain(name, NodeLabel.Rule, domain) |
|
4be3595…
|
lmata
|
111 |
logger.info("Rule: %s", name) |
|
4be3595…
|
lmata
|
112 |
|
|
4be3595…
|
lmata
|
113 |
def rule_governs(self, rule_name: str, target_name: str, target_label: NodeLabel) -> None: |
|
4be3595…
|
lmata
|
114 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
115 |
NodeLabel.Rule, |
|
758c6b7…
|
lmata
|
116 |
{"name": rule_name}, |
|
4be3595…
|
lmata
|
117 |
EdgeType.GOVERNS, |
|
758c6b7…
|
lmata
|
118 |
target_label, |
|
758c6b7…
|
lmata
|
119 |
{"name": target_name}, |
|
4be3595…
|
lmata
|
120 |
) |
|
4be3595…
|
lmata
|
121 |
|
|
4be3595…
|
lmata
|
122 |
# ── Decisions ───────────────────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
123 |
|
|
4be3595…
|
lmata
|
124 |
def add_decision( |
|
4be3595…
|
lmata
|
125 |
self, |
|
4be3595…
|
lmata
|
126 |
name: str, |
|
4be3595…
|
lmata
|
127 |
description: str = "", |
|
4be3595…
|
lmata
|
128 |
domain: str = "", |
|
4be3595…
|
lmata
|
129 |
status: str = "accepted", |
|
4be3595…
|
lmata
|
130 |
rationale: str = "", |
|
4be3595…
|
lmata
|
131 |
alternatives: str = "", |
|
4be3595…
|
lmata
|
132 |
date: str = "", |
|
4be3595…
|
lmata
|
133 |
) -> None: |
|
758c6b7…
|
lmata
|
134 |
self.store.create_node( |
|
758c6b7…
|
lmata
|
135 |
NodeLabel.Decision, |
|
758c6b7…
|
lmata
|
136 |
{ |
|
758c6b7…
|
lmata
|
137 |
"name": name, |
|
758c6b7…
|
lmata
|
138 |
"description": description, |
|
758c6b7…
|
lmata
|
139 |
"domain": domain, |
|
758c6b7…
|
lmata
|
140 |
"status": status, |
|
758c6b7…
|
lmata
|
141 |
"rationale": rationale, |
|
758c6b7…
|
lmata
|
142 |
"alternatives": alternatives, |
|
758c6b7…
|
lmata
|
143 |
"date": date, |
|
758c6b7…
|
lmata
|
144 |
}, |
|
758c6b7…
|
lmata
|
145 |
) |
|
4be3595…
|
lmata
|
146 |
if domain: |
|
4be3595…
|
lmata
|
147 |
self._link_to_domain(name, NodeLabel.Decision, domain) |
|
4be3595…
|
lmata
|
148 |
logger.info("Decision: %s", name) |
|
4be3595…
|
lmata
|
149 |
|
|
4be3595…
|
lmata
|
150 |
# ── People ──────────────────────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
151 |
|
|
4be3595…
|
lmata
|
152 |
def add_person( |
|
4be3595…
|
lmata
|
153 |
self, |
|
4be3595…
|
lmata
|
154 |
name: str, |
|
4be3595…
|
lmata
|
155 |
email: str = "", |
|
4be3595…
|
lmata
|
156 |
role: str = "", |
|
4be3595…
|
lmata
|
157 |
team: str = "", |
|
4be3595…
|
lmata
|
158 |
) -> None: |
|
758c6b7…
|
lmata
|
159 |
self.store.create_node( |
|
758c6b7…
|
lmata
|
160 |
NodeLabel.Person, |
|
758c6b7…
|
lmata
|
161 |
{ |
|
758c6b7…
|
lmata
|
162 |
"name": name, |
|
758c6b7…
|
lmata
|
163 |
"email": email, |
|
758c6b7…
|
lmata
|
164 |
"role": role, |
|
758c6b7…
|
lmata
|
165 |
"team": team, |
|
758c6b7…
|
lmata
|
166 |
}, |
|
758c6b7…
|
lmata
|
167 |
) |
|
4be3595…
|
lmata
|
168 |
logger.info("Person: %s", name) |
|
4be3595…
|
lmata
|
169 |
|
|
4be3595…
|
lmata
|
170 |
def assign(self, target_name: str, target_label: NodeLabel, person_name: str) -> None: |
|
4be3595…
|
lmata
|
171 |
"""Assign a person as owner of any node.""" |
|
4be3595…
|
lmata
|
172 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
173 |
target_label, |
|
758c6b7…
|
lmata
|
174 |
{"name": target_name}, |
|
4be3595…
|
lmata
|
175 |
EdgeType.ASSIGNED_TO, |
|
758c6b7…
|
lmata
|
176 |
NodeLabel.Person, |
|
758c6b7…
|
lmata
|
177 |
{"name": person_name}, |
|
4be3595…
|
lmata
|
178 |
) |
|
4be3595…
|
lmata
|
179 |
|
|
4be3595…
|
lmata
|
180 |
# ── Wiki pages ──────────────────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
181 |
|
|
4be3595…
|
lmata
|
182 |
def wiki_page( |
|
4be3595…
|
lmata
|
183 |
self, |
|
4be3595…
|
lmata
|
184 |
name: str, |
|
4be3595…
|
lmata
|
185 |
url: str = "", |
|
4be3595…
|
lmata
|
186 |
source: str = "github", |
|
4be3595…
|
lmata
|
187 |
content: str = "", |
|
4be3595…
|
lmata
|
188 |
updated_at: str = "", |
|
4be3595…
|
lmata
|
189 |
) -> None: |
|
758c6b7…
|
lmata
|
190 |
self.store.create_node( |
|
758c6b7…
|
lmata
|
191 |
NodeLabel.WikiPage, |
|
758c6b7…
|
lmata
|
192 |
{ |
|
758c6b7…
|
lmata
|
193 |
"name": name, |
|
758c6b7…
|
lmata
|
194 |
"url": url, |
|
758c6b7…
|
lmata
|
195 |
"source": source, |
|
758c6b7…
|
lmata
|
196 |
"content": content, |
|
758c6b7…
|
lmata
|
197 |
"updated_at": updated_at, |
|
758c6b7…
|
lmata
|
198 |
}, |
|
758c6b7…
|
lmata
|
199 |
) |
|
4be3595…
|
lmata
|
200 |
logger.info("WikiPage: %s", name) |
|
4be3595…
|
lmata
|
201 |
|
|
758c6b7…
|
lmata
|
202 |
def wiki_documents( |
|
758c6b7…
|
lmata
|
203 |
self, |
|
758c6b7…
|
lmata
|
204 |
wiki_page_name: str, |
|
758c6b7…
|
lmata
|
205 |
target_name: str, |
|
758c6b7…
|
lmata
|
206 |
target_props: dict[str, Any], |
|
758c6b7…
|
lmata
|
207 |
target_label: NodeLabel, |
|
758c6b7…
|
lmata
|
208 |
) -> None: |
|
4be3595…
|
lmata
|
209 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
210 |
NodeLabel.WikiPage, |
|
758c6b7…
|
lmata
|
211 |
{"name": wiki_page_name}, |
|
4be3595…
|
lmata
|
212 |
EdgeType.DOCUMENTS, |
|
758c6b7…
|
lmata
|
213 |
target_label, |
|
758c6b7…
|
lmata
|
214 |
target_props, |
|
4be3595…
|
lmata
|
215 |
) |
|
4be3595…
|
lmata
|
216 |
|
|
4be3595…
|
lmata
|
217 |
# ── Code ↔ Knowledge bridges ────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
218 |
|
|
4be3595…
|
lmata
|
219 |
def annotate_code( |
|
4be3595…
|
lmata
|
220 |
self, |
|
4be3595…
|
lmata
|
221 |
code_name: str, |
|
4be3595…
|
lmata
|
222 |
code_label: str, |
|
4be3595…
|
lmata
|
223 |
concept: str | None = None, |
|
4be3595…
|
lmata
|
224 |
rule: str | None = None, |
|
4be3595…
|
lmata
|
225 |
) -> None: |
|
4be3595…
|
lmata
|
226 |
""" |
|
4be3595…
|
lmata
|
227 |
Link a code node to a concept or rule via ANNOTATES. |
|
4be3595…
|
lmata
|
228 |
code_label should be a string matching a NodeLabel value. |
|
4be3595…
|
lmata
|
229 |
""" |
|
4be3595…
|
lmata
|
230 |
label = NodeLabel(code_label) |
|
4be3595…
|
lmata
|
231 |
if concept: |
|
4be3595…
|
lmata
|
232 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
233 |
NodeLabel.Concept, |
|
758c6b7…
|
lmata
|
234 |
{"name": concept}, |
|
4be3595…
|
lmata
|
235 |
EdgeType.ANNOTATES, |
|
758c6b7…
|
lmata
|
236 |
label, |
|
758c6b7…
|
lmata
|
237 |
{"name": code_name}, |
|
4be3595…
|
lmata
|
238 |
) |
|
4be3595…
|
lmata
|
239 |
if rule: |
|
4be3595…
|
lmata
|
240 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
241 |
NodeLabel.Rule, |
|
758c6b7…
|
lmata
|
242 |
{"name": rule}, |
|
4be3595…
|
lmata
|
243 |
EdgeType.ANNOTATES, |
|
758c6b7…
|
lmata
|
244 |
label, |
|
758c6b7…
|
lmata
|
245 |
{"name": code_name}, |
|
4be3595…
|
lmata
|
246 |
) |
|
4be3595…
|
lmata
|
247 |
|
|
4be3595…
|
lmata
|
248 |
def code_implements(self, code_name: str, code_label: str, concept_name: str) -> None: |
|
4be3595…
|
lmata
|
249 |
"""Mark a function/class as implementing a concept.""" |
|
4be3595…
|
lmata
|
250 |
label = NodeLabel(code_label) |
|
4be3595…
|
lmata
|
251 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
252 |
label, |
|
758c6b7…
|
lmata
|
253 |
{"name": code_name}, |
|
4be3595…
|
lmata
|
254 |
EdgeType.IMPLEMENTS, |
|
758c6b7…
|
lmata
|
255 |
NodeLabel.Concept, |
|
758c6b7…
|
lmata
|
256 |
{"name": concept_name}, |
|
4be3595…
|
lmata
|
257 |
) |
|
4be3595…
|
lmata
|
258 |
|
|
4be3595…
|
lmata
|
259 |
# ── Helpers ─────────────────────────────────────────────────────────────── |
|
4be3595…
|
lmata
|
260 |
|
|
4be3595…
|
lmata
|
261 |
def _link_to_domain(self, name: str, label: NodeLabel, domain: str) -> None: |
|
4be3595…
|
lmata
|
262 |
# Ensure domain node exists |
|
4be3595…
|
lmata
|
263 |
self.store.create_node(NodeLabel.Domain, {"name": domain, "description": ""}) |
|
4be3595…
|
lmata
|
264 |
self.store.create_edge( |
|
758c6b7…
|
lmata
|
265 |
label, |
|
758c6b7…
|
lmata
|
266 |
{"name": name}, |
|
4be3595…
|
lmata
|
267 |
EdgeType.BELONGS_TO, |
|
758c6b7…
|
lmata
|
268 |
NodeLabel.Domain, |
|
758c6b7…
|
lmata
|
269 |
{"name": domain}, |
|
4be3595…
|
lmata
|
270 |
) |