| | @@ -0,0 +1,476 @@ |
| 1 | +"""
|
| 2 | +PlanopticonIngester — loads planopticon output into the navegador knowledge graph.
|
| 3 | +
|
| 4 | +Planopticon extracts structured knowledge from videos, meetings, and documents:
|
| 5 | +entities, relationships, key points, action items, and diagrams. This ingester
|
| 6 | +maps that output onto navegador's knowledge layer so agents can query business
|
| 7 | +context alongside code.
|
| 8 | +
|
| 9 | +Supported input:
|
| 10 | + - manifest.json (single video — primary entry point)
|
| 11 | + - interchange.json (canonical planopticon exchange format)
|
| 12 | + - knowledge_graph.json (raw KG export, no manifest required)
|
| 13 | +
|
| 14 | +Mapping:
|
| 15 | + planopticon → navegador
|
| 16 | + ─────────────────────────────────────────
|
| 17 | + node type=person → Person
|
| 18 | + node type=concept/technology → Concept
|
| 19 | + node type=organization → Concept (domain = "organization")
|
| 20 | + PlanningEntity type=decision → Decision
|
| 21 | + PlanningEntity type=requirement/constraint → Rule
|
| 22 | + diagram → WikiPage (content = mermaid/description)
|
| 23 | + key_point → Concept (tagged from source)
|
| 24 | + action_item.assignee → Person + ASSIGNED_TO edge
|
| 25 | + relationship.type → mapped EdgeType (see EDGE_MAP)
|
| 26 | +
|
| 27 | +Usage:
|
| 28 | + from navegador.ingestion.planopticon import PlanopticonIngester
|
| 29 | + store = GraphStore.sqlite(".navegador/graph.db")
|
| 30 | + ing = PlanopticonIngester(store)
|
| 31 | +
|
| 32 | + stats = ing.ingest_manifest("planopticon-output/manifest.json")
|
| 33 | + stats = ing.ingest_kg("planopticon-output/results/knowledge_graph.json")
|
| 34 | + stats = ing.ingest_interchange("planopticon-output/interchange.json")
|
| 35 | +"""
|
| 36 | +
|
| 37 | +import json
|
| 38 | +import logging
|
| 39 | +from pathlib import Path
|
| 40 | +from typing import Any
|
| 41 | +
|
| 42 | +from navegador.graph.schema import EdgeType, NodeLabel
|
| 43 | +from navegador.graph.store import GraphStore
|
| 44 | +
|
| 45 | +logger = logging.getLogger(__name__)
|
| 46 | +
|
| 47 | +# ── Relationship type mapping ─────────────────────────────────────────────────
|
| 48 | +
|
| 49 | +EDGE_MAP: dict[str, EdgeType] = {
|
| 50 | + "related_to": � Rule
|
| 51 | + diagram uses":get("edgDEPENDS_ON,
|
| 52 | + "built_on":depends_on": DEPENDS_ON,
|
| 53 | + "built_on":built_on": DEPENDS_ON,
|
| 54 | + ��─
|
| 55 | + node type=person requires": DEPENDS_ON,
|
| 56 | + "built_on":blocked_by": DEPENDS_ON,
|
| 57 | + "built_on":has_risk": � Rule
|
| 58 | + diagram addresses": � Rule
|
| 59 | + diagram ��───────� ��─
|
| 60 | + node ept/technolontity type=decision → Decowned_by": ntity type=decision → Decowns": tity type=decintity type=decision → Decworks_with": � Rule
|
| 61 | + d ]) -> None:
|
| 62 | + namermaid/description)
|
| 63 | + key_point → Concept (tagged from NodeLabel.Concept,
|
| 64 | + "technology": NodeLabel.Concept,
|
| 65 | + "organizae("planopticon-output/interchandiagram":{
|
| 66 | + planopticon import Pla "content":pticon import Planoptic{
|
| 67 | + ite(".navegador/graph.db")
|
| 68 | + ing = PlanopticonIngester(store)
|
| 69 | +
|
| 70 | + stats = ing.ing output/manifest.json")
|
| 71 | + stats = ing.ingest_kg("planopticon-output/resullts/knowledge_graph.json")
|
| 72 | + trip()
|
| 73 | + Rule,
|
| 74 | + "goal": "content":opticon-output/interchange. NodeLabel.Concept,
|
| 75 | + "feature": NodeLabel.Concept,
|
| 76 | + "milestone": NodeLabel.Concept,
|
| 77 | + "task": "content":Concept,
|
| 78 | + "dependency": NodeLabel.Concept,
|
| 79 | +}
|
| 80 | +
|
| 81 | +
|
| 82 | +classngestion.planopticonraph.store import GraphStore
|
| 83 | +
|
| 84 | +logger = logging.getLogger(__name__)
|
| 85 | +
|
| 86 | +# ── Relationship type mapping ─────────────────────────────────────────────────
|
| 87 | +
|
| 88 | +EDGE_MAP: dict[str, EdgeType] = {
|
| 89 | + "related_to": EdgeType.RELATED_TO,
|
| 90 | + "uses": EdgeType.DEPENDS_ON,
|
| 91 | + "depends_on": EdgeType.DEPENDS_ON,
|
| 92 | + "built_on": EdgeType.DEPENDS_ON,
|
| 93 | + "implements": EdgeType.IMPLEMENTS,
|
| 94 | + "requires": EdgeType.DEPENDS_ON,
|
| 95 | + "blocked_by": EdgeType.DEPENDS_ON,
|
| 96 | + "has_risk": EdgeType.RELATED_TO,
|
| 97 | + "addresses": EdgeType.RELATED_TO,
|
| 98 | + "has_tradeoff": EdgeType.RELATED_TO,
|
| 99 | + "delivers": EdgeType.IMPLEMENTS,
|
| 100 | + "parent_of": EdgeType.CONTAINS,
|
| 101 | + "assigned_to": EdgeType.ASSIGNED_TO,
|
| 102 | + "owned_by": EdgeType.ASSIGNED_TO,
|
| 103 | + "owns": EdgeType.ASSIGNED_TO,
|
| 104 | + "employed_by": EdgeType.ASSIGNED_TO,
|
| 105 | + "works_with": EdgeType.RELATED_TO,
|
| 106 | + "governs": EdgeType.GOVERNS,
|
| 107 | + "documents": EdgeType.DOCUMENTS,
|
| 108 | +}
|
| 109 | +
|
| 110 | +# planopticon node type → navegador NodeLabel
|
| 111 | +NODE_TYPE_MAP: dict[str, NodeLabel] = {
|
| 112 | + "concept": NodeLabel.Concept,
|
| 113 | + "technology": NodeLabel.Concept,
|
| 114 | + "organization": NodeLabel.Concept,
|
| 115 | + "diagram": NodeLabel.WikiPage,
|
| 116 | + "time": NodeLabel.Concept,
|
| 117 | + "person": NodeLabel.Person,
|
| 118 | +}
|
| 119 | +
|
| 120 | +# planning_type → navegador NodeLabel
|
| 121 | +PLANNING_TYPE_MAP: dict[str, NodeLabel] = {
|
| 122 | + "decision": NodeLabel.Decision,
|
| 123 | + "requirement": NodeLabel.Rule,
|
| 124 | + "constraint": NodeLabel.Rule,
|
| 125 | + "risk": NodeLabel.Rule,
|
| 126 | + "goal": NodeLabel.Concept,
|
| 127 | + "assumption": NodeLabel.Concept,
|
| 128 | + "feature": NodeLabel.Concept,
|
| 129 | + "milestone": NodeLabel.Concept,
|
| 130 | + "task": NodeLabel.Concept,
|
| 131 | + "dependency": NodeLabel.Concept,
|
| 132 | +}
|
| 133 | +
|
| 134 | +
|
| 135 | +class PlanopticonIngester:
|
| 136 | + """
|
| 137 | + Reads planopticon output and writes it into a GraphStore.
|
| 138 | +
|
| 139 | + All paths may be relative (resolved against the manifest's parent directory)
|
| 140 | + or absolute.
|
| 141 | + """
|
| 142 | +
|
| 143 | + def __init__(self, store: GraphStore, source_tag: str = "") -> None:
|
| 144 | + self.store = store
|
| 145 | + self.source_tag = source_tag # optional label for provenance
|
| 146 | + self._stats: dic stats.get("nodes", 0), stats.get("edges", 0)ntent[:4000],
|
| 147 | + for rel in data.get("relkg(self, kg_path: str | Path) -> dict[str, int]:
|
| 148 | + """
|
| 149 | + Ingest a knowledge_graph.json (KnowledgeGraphData) directly.
|
| 150 | + Can be used standalone without a manifest.
|
| 151 | + """
|
| 152 | + kg_path = Path(kg_path).resolve()
|
| 153 | + data = self._load_json(kg_path)
|
| 154 | + stats = self._reset_stats()
|
| 155 | +
|
| 156 | + for node in data.get("nodes", []):
|
| 157 | + self._ingest_kg_node(node)
|
| 158 | +
|
| 159 | + for rel in data.get("relationships", []):
|
| 160 | + self._ingest_kg_relationship(rel)
|
| 161 | +
|
| 162 | + # Ingest sources as WikiPage nodes
|
| 163 | + for source in data.get("sources", []):
|
| 164 | + self._ingest_source(source)
|
| 165 | +
|
| 166 | + return stats
|
| 167 | +
|
| 168 | + def ingest_interchange(self, interchange_path: str | Path) -> dict[str, int]:
|
| 169 | + """
|
| 170 | + Ingest a planopticon interchange.json (PlanOpticonExchange format).
|
| 171 | + Includes planning taxonomy, artifacts, and full entity graph.
|
| 172 | + """
|
| 173 | + interchange_path = Path(interchange_path).resolve()
|
| 174 | + data = self._load_json(interchange_path)
|
| 175 | + stats = self._reset_stats()
|
| 176 | +
|
| 177 | + project = data.get("project", {})
|
| 178 | + project_name = project.get("name", interchange_path.parent.name)
|
| 179 | + self.source_tag = self.source_tag or project_name
|
| 180 | +
|
| 181 | + # Domain from project tags
|
| 182 | + for tag in project.get("tags", []):
|
| 183 | + self.store.create_node(NodeLabel.Domain, {"name": tag, "description": ""})
|
| 184 | +
|
| 185 | + # Entities (planning taxonomy takes priority over raw type)
|
| 186 | + for entity in data.get("entities", []):
|
| 187 | + planning_type = entity.get("planning_type")
|
| 188 | + if planning_type and planning_type in PLANNING_TYPE_MAP:
|
| 189 | + self._ingest_planning_entity(entity)
|
| 190 | + else:
|
| 191 | + self._ingest_kg_node(entity)
|
| 192 | +
|
| 193 | + # Relationships
|
| 194 | + for rel in data.get("relationships", []):
|
| 195 | + self._ingest_kg_relationship(rel)
|
| 196 | +
|
| 197 | + # Artifacts → WikiPage nodes
|
| 198 | + for artifact in data.get("artifacts", []):
|
| 199 | + self._ingest_artifact(artifact, project_name)
|
| 200 | +
|
| 201 | + # Sources
|
| 202 | + for source in data.get("sou if no onto navegador's knowl"""
|
| 203 | +P": "",
|
| 204 | + "updated_at": source.get("ingested_at", ""),
|
| 205 | + },
|
| 206 | + )
|
| 207 | + self._stats["nodes"] = self._stats.get("nodes", 0) + 1
|
| 208 | +
|
| 209 | + def _ingest_artifact(self, artifact: dict[str, Any], project_name: str) -> None:
|
| 210 | + name = (artifact.get("name") or "").strip()
|
| 211 | + if not name:
|
| 212 | + return
|
| 213 | + content = artifact.get("content", "")
|
| 214 | + self.store.create_node(
|
| 215 | + NodeLabel.WikiPage,
|
| 216 | + {
|
| 217 | + "name": name,
|
| 218 | + "url": "",
|
| 219 | + "source": project_name,
|
| 220 | + "content": content[:4000],
|
| 221 | + },
|
| 222 | + )
|
| 223 | + self._stats["nodes"] = self._stats.get("nodes", 0) + 1
|
| 224 | +
|
| 225 | + # ── Helpers ───────────────────────────────────────────────────────────────
|
| 226 | +
|
| 227 | + def _ensure_domain(self, name: str) -> None:
|
| 228 | + self.store.create_node(NodeLabel.Domain, {"name": name, "description": ""})
|
| 229 | +
|
| 230 | + def _lazy_wiki_link(self, name: str, label: NodeLabel, source_id: str) -> None:
|
| 231 | + """Create a DOCUMENTS edge from a WikiPage to this node if the page exists."""
|
| 232 | + try:
|
| 233 | + self.store.create_edge(
|
| 234 | + NodeLabel.WikiPage,
|
| 235 | + {"name": source_id},
|
| 236 | + EdgeType.DOCUMENTS,
|
| 237 | + label,
|
| 238 | + {"name": name},
|
| 239 | + )
|
| 240 | + except Exception:
|
| 241 | + logger.debug("Could not link %s to wiki page %s", name, source_id)
|
| 242 | +
|
| 243 | + def _load_json(self, path: Path) -> Any:
|
| 244 | + return json.loads(Path(path).read_text(encoding="utf-8"))
|
| 245 | +
|
| 246 | + def _reset_stats(self) -> dict[str, int]:
|
| 247 | + self._stats = {"nodes": 0, "edges": 0}
|
| 248 | + return self._stats
|
| 249 | +
|
| 250 | + def _merge_stats(self, other: dict[str, int]) -> None:
|
| 251 | + for k, v in other.items():
|
| 252 | + self._stats[k] = self._stats.get(k, 0) + v
|
| 253 | +role": node.get("r, ""),
|
| 254 | + } and documents:
|
| 255 | +entities, relationships, key points, action items, and diagrams"""
|
| 256 | +PlanopticonInge onto navegador's knowledge l"""
|
| 257 | +Planoptic, ""),
|
| 258 | + "content": descr})
|
| 259 | + "n },
|
| 260 | + )
|
| 261 | + else:
|
| 262 | + domain = "organization" if raw_type == "organization" else node.get("domain", "")
|
| 263 | + },
|
| 264 | + )
|
| 265 | +
|
| 266 | + onto navegador's knowl"""
|
| 267 | +PlanopticonI"""
|
| 268 | +PlanopticonIngester — loads sub = self.i})
|
| 269 | + if rn stats
|
| 270 | +
|
| 271 | + # ── Node ingestion ──────────────────────────────� {"name": name},
|
| 272 | + ", 0) + 1
|
| 273 | +
|
| 274 | + def _ingest_kg_relat {"name": domain},
|
| 275 | + )
|
| 276 | +
|
| 277 | + self._stats["nodes"] = self._stats.get("nodes", 0) + 1
|
| 278 | +
|
| 279 | + P.get(raw_type, NodeLabel.Concept)
|
| 280 | +
|
| 281 | + name = (node.get("name") or node.get("id") or "").strip()
|
| 282 | + if not name:
|
| 283 | + return
|
| 284 | +
|
| 285 | + descriptions = node.get("descriptions", [])
|
| 286 | + description = descriptions[0] if descriptions else node.get("description", "")
|
| 287 | +
|
| 288 | + if label == NodeLabel.Person:
|
| 289 | + self.store.create_node(
|
| 290 | + NodeLabel.Person,
|
| 291 | + {
|
| 292 | + "name": name,
|
| 293 | + "email": "",
|
| 294 | + "role": node.get("role", ""),
|
| 295 | + "team": node.get("organization", ""),
|
| 296 | + },
|
| 297 | + )
|
| 298 | + elif label == NodeLabel.WikiPage:
|
| 299 | + self.store.create_node(
|
| 300 | + NodeLabel.WikiPautput/manifest.json onto navegador's knowl"""
|
| 301 | +PlanopticonI"""
|
| 302 | +PlanopticonIngester — loads "status": status else:
|
| 303 | + domain"rationale": entity.get("ra} and documents:
|
| 304 | +entities, relation"""
|
| 305 | +PlanopticoNodeLabel.Rule, onto navegador's knowl"""
|
| 306 | +PlanopticonI"""
|
| 307 | +PlanopticonIngester — loads "severity": "critical" if priority == "high" else "info",
|
| 308 | + "rationale": entity.get("ra})
|
| 309 | + "nels self.store},
|
| 310 | + )
|
| 311 | +
|
| 312 | + onto navegador's knowl"""
|
| 313 | +PlanopticonI"""
|
| 314 | +PlanopticonIngester — loads "status": status,
|
| 315 | + })
|
| 316 | +
|
| 317 | + conceptmerge_stats(sub)
|
| 318 | +
|
| 319 | + ode ingestiondiagrams. Thi"""
|
| 320 | +PlBELONGS_TOr, Any]) -> None:
|
| 321 | +NodeLabel.Domain, {"name": domain},
|
| 322 | + )
|
| 323 | +
|
| 324 | + self._stats["nodes"] = self._stats.get("nodes", 0) + 1
|
| 325 | +
|
| 326 | + def _ingest_kg_relationship(self, rel: dict[str, Any]) -> None:
|
| 327 | + src = (rel.get("source") or "").strip()
|
| 328 | + tgt = (rel.get("target") or "").strip()
|
| 329 | + rel_type = (rel.get("type") or "related_to").lower().replace(" ", "_")
|
| 330 | +
|
| 331 | + if not src or not tgt:
|
| 332 | + return
|
| 333 | +
|
| 334 | + edge_type = EDGE_MAP.get(rel_type, EdgeType.RELATED_TO)
|
| 335 | +
|
| 336 | + # We don't know the exact label of each node — use a label-agnos cypher = (
|
| 337 | + """
|
| 338 | + MATCH (a), (b)
|
| 339 | + WHERE a.name = $src AND b.name = $t + edge_type + """]->(b)
|
| 340 | + """
|
| 341 | + """
|
| 342 | + )
|
| 343 | + try:
|
| 344 | + self.store.query(cypher, {"src": src, "tgt": tgt})
|
| 345 | + self._stats["edges"] = self._stats.get("edges", 0) + 1
|
| 346 | + except Exception:
|
| 347 | + logger.warning("Could not create edge %s -[%s]-> %s", src, edge_type, tgt)
|
| 348 | +
|
| 349 | + def _ingest_key_points(self, key_points: list[dict], source: str) -> None:
|
| 350 | + for kp in key_points:
|
| 351 | + point = (kp.get("point") or "").strip()
|
| 352 | + if not point:
|
| 353 | + continue
|
| 354 | + topic = kp.get("topic") or ""
|
| 355 | + name = point[:120] # use the point text as the concept name
|
| 356 | + },
|
| 357 | + )
|
| 358 | +
|
| 359 | + onto navegador's knowl"""
|
| 360 | +PlanopticonIngester �, ""),
|
| 361 | + "domai
|
| 362 | +PlanopticonIngester — loads plan )
|
| 363 | + if topic:
|
| 364 | + me": name},
|
| 365 | + EdgeType.BELONGS_TO,
|
| 366 | + NodeLabel.Dom {"name": name},
|
| 367 | + ", 0) + 1
|
| 368 | +
|
| 369 | + def _ingest_kg_relatict[str, Any]) -> None:
|
| 370 | + src = (rel.get("sdebu) or "").strip()
|
| 371 | + tgt = (rel.get("target") or "").strip()
|
| 372 | + rel_type = (rel.get("type") or "related_to").lower().replace(" ", "_")
|
| 373 | +
|
| 374 | + if not src or not tgt:
|
| 375 | + return
|
| 376 | +
|
| 377 | + edge_type = EDGE_MAP.get(rel_type, EdgeType.RELATED_TO)
|
| 378 | +
|
| 379 | + # We don't know the exact label of each node — use a label-agnostic match
|
| 380 | + cypher = (
|
| 381 | + """
|
| 382 | + MATCH (a), (b)
|
| 383 | + NodeLabel.Rule, onto navegador's knowledge"""
|
| 384 | +PlanopticonIngester — lo.store.query(cypher, {"src": ": "",
|
| 385 | + "domainms, and diagrams. This ingester"""
|
| 386 | +PlanopticonIngester — er, {"src": src, "tgt": tgt}nodes"] = self._stats.get("aw_type, NodeLabel.Concesource: str) -> None:
|
| 387 | + if no {
|
| 388 | + "name": assignee, "em": "",
|
| 389 | + "updated_at": source.get("ingested_at", ""),
|
| 390 | + },
|
| 391 | + )
|
| 392 | + self._stats["nodes"] = self._stats.get("nodes", 0) + 1
|
| 393 | +
|
| 394 | + def _ingest_artifact(self, artifact: dict[str, Any], project_name: str) -> None:
|
| 395 | + name = (artifact.get("name") or "").strip()
|
| 396 | + if not name:
|
| 397 | + return
|
| 398 | + content = artifact.get("content", "")
|
| 399 | + self.store.create_node(
|
| 400 | + NodeLabel.WikiPage,
|
| 401 | + {
|
| 402 | + "name": name,
|
| 403 | + "url": "",
|
| 404 | + "source": project_name,
|
| 405 | + "content": content[:4000],
|
| 406 | + },
|
| 407 | + )
|
| 408 | + self._stats["nodes"] = self._stats.get("nodes", 0) + 1
|
| 409 | +
|
| 410 | + # ── Helpers ───────────────────────────────────────────────────────────────
|
| 411 | +
|
| 412 | + def _ensure_domain(self, name: str) -> None:
|
| 413 | + self.store.create_node(NodeLabel.Domain, {"name": name, "description": ""})
|
| 414 | +
|
| 415 | + def _lazy_wiki_link(self, name: str, label: NodeLabel, source_id: str) -> None:
|
| 416 | + """Create a DOCUMENTS edge from a WikiPage to this node if the page exists."""
|
| 417 | + try:
|
| 418 | + self.store.create_edge(
|
| 419 | + NodeLabel.WikiPage,
|
| 420 | + {"name": source_id},
|
| 421 | + EdgeType.DOCUMENTS,
|
| 422 | + label,
|
| 423 | + {"name": name},
|
| 424 | + )
|
| 425 | + except Exception:
|
| 426 | + logger.debug("Could not link %s to wiki page %s", name, source_id)
|
| 427 | +
|
| 428 | + def _load_json(self, path: Path) -> Any:
|
| 429 | + return json.loads(Path(path).read_text(encoding="utf-8"))
|
| 430 | +
|
| 431 | + def _reset_stats(self) -> dict[str, int]:
|
| 432 | + self._stats = {"nodes": 0, "edges": 0}
|
| 433 | + return self._stats
|
| 434 | +
|
| 435 | + def _merge_stats(self, other: dict[str, int]) -> None:
|
| 436 | + for k, v in other.items():
|
| 437 | + self._stats[k] = self._stats.get(k, 0) + v
|
| 438 | +}assignee},
|
| 439 | + store.create_edge(
|
| 440 | +
|
| 441 | + + edgeget("details", ""),
|
| 442 | + "domain": topic,
|
| 443 | + {"name": assignee},
|
| 444 | + )
|
| 445 | + self._stats["edges"] = self._stats.get("edges", 0) + 1
|
| 446 | +
|
| 447 | + def _ingest_diagram(self, diagram: dict[str, Any], base_dir: Path, source: str) -> None:
|
| 448 | + dtype = diagram.get("diagram_type", "diagram")
|
| 449 | + desc = diagram.get("description") or diagram.get("text_content") or ""
|
| 450 | + mermaid = diagram.get("mermaid", "")
|
| 451 | + ts = diagram.get("timestamp")
|
| 452 | + name = f"{dtype.capitalize()} @ {ts:.0f}s" if ts is not None else f"{dtype.capitalize()}"
|
| 453 | +
|
| 454 | + content = mermaid or desc
|
| 455 | + self.store.create_node(NodeLabel.WikiPage, {
|
| 456 | +"
|
| 457 | +Planopticngester — loads planoath", ""),
|
| 458 | + "so as concepts
|
| 459 | + for element in}) raw_type = node.get("type", "concept")
|
| 460 | + label = NODE_TYPE_MAP.get(Link diagram elements as concepts
|
| 461 | + for element in diagram.get("elements", []):
|
| 462 | + element = element.strip()
|
| 463 | + if not element:
|
| 464 | + continue
|
| 465 | + self.store.create_node(},
|
| 466 | + )
|
| 467 | +
|
| 468 | + onto navegador's knowledge"""
|
| 469 | +PlanopticonIngester —"""
|
| 470 | +PlanopticonIng "status": "",
|
| 471 | + }) passype", "concept")
|
| 472 | + labtryam elements as concepts
|
| 473 | + except FileNotFoundErrorepts
|
| 474 | + for elemwarning("File not found: %s", path)ador's knowledge la"""
|
| 475 | +PlanopticonInepts
|
| 476 | + |