reverse-engineering-anything/src/process/ProcessOwnershipEvidence.ts

import { processIdentityHasExited } from "./ProcessLifecycle.js";
import type {
  ProcessIdentityObservation,
  ProcessNativeRelationObservation,
  ProcessOwnershipFact,
  ProcessRunTokenObservation,
  ProcessTableEntry,
} from "./ProcessOwnership.js";

export interface RetainedProcessOwnershipCandidate {
  readonly pid: number;
  /** Undefined means the generation has never been established, not PID ownership. */
  readonly identity: string | undefined;
  readonly process: ProcessTableEntry;
  readonly applicability:
    | "token-verified"
    | "related-ownership-unresolved"
    | "identity-unresolved";
  /** Why this generation remains in the ledger; capture anchors are only scope uncertainty. */
  readonly scopeEvidence:
    | "run-token"
    | "current-token-owned-parent"
    | "capture-leader-pid"
    | "captured-process-group-id"
    | "native-identity-unresolved"
    | "previously-related-generation";
  readonly token: ProcessRunTokenObservation;
  readonly relation:
    | Extract<ProcessNativeRelationObservation, { readonly state: "observed" }>
    | undefined;
  readonly diagnostic?: string;
}

export interface ProcessOwnershipEvidenceCoverage {
  readonly snapshotCount: number;
  /** Rows supplied by the caller for the latest snapshot, not a completeness claim. */
  readonly listedRows: number;
  /** Distinct PIDs given before/token/after reads, including retained-only PIDs. */
  readonly examinedPids: number;
  readonly retainedOnlyPids: number;
  readonly retainedOwned: number;
  readonly retainedUnresolved: number;
  readonly uncertainPids: readonly number[];
  readonly processTableCoverage: "unreported-by-host";
}

/** Per-run facts retained across failed cleanup attempts; identity is required for authority. */
export class ProcessOwnershipEvidenceOwner {
  readonly #candidates = new Map<number, RetainedProcessOwnershipCandidate>();
  #snapshotCount = 0;
  #coverage: ProcessOwnershipEvidenceCoverage = {
    snapshotCount: 0,
    listedRows: 0,
    examinedPids: 0,
    retainedOnlyPids: 0,
    retainedOwned: 0,
    retainedUnresolved: 0,
    uncertainPids: [],
    processTableCoverage: "unreported-by-host",
  };

  get candidates(): readonly RetainedProcessOwnershipCandidate[] {
    return [...this.#candidates.values()];
  }

  get coverage(): ProcessOwnershipEvidenceCoverage {
    return this.#coverage;
  }

  /**
   * Reconcile facts from one explicit snapshot. Unrelated transient rows are
   * never retained; unknown related candidates remain evidence, not authority.
   */
  reconcile(
    facts: readonly ProcessOwnershipFact[],
    options: {
      readonly runId: string;
      readonly listedRows: number;
      readonly listedPids: ReadonlySet<number>;
      readonly leaderPid: number;
      readonly processGroupIds: ReadonlySet<number>;
      /** First identity/token/identity result before bounded token-only retries. */
      readonly scopeFacts?: readonly ProcessOwnershipFact[];
    },
  ): readonly RetainedProcessOwnershipCandidate[] {
    this.#snapshotCount += 1;
    const factsByPid = new Map(facts.map((fact) => [fact.process.pid, fact]));
    const sameGeneration = new Map<
      number,
      Extract<
        ProcessOwnershipFact["identityAfter"],
        { readonly state: "readable" }
      >
    >();
    const sameGenerationBefore = new Map<
      number,
      Extract<ProcessIdentityObservation, { readonly state: "readable" }>
    >();
    const tokenMatchedPids = new Set<number>();
    const uncertainPids = new Set<number>();
    const retiredPids = new Set<number>();

    for (const fact of facts) {
      if (fact.state !== "same-generation") continue;
      if (fact.identityAfter.state !== "readable") continue;
      sameGeneration.set(fact.process.pid, fact.identityAfter);
      if (
        fact.identityBefore.state === "readable" &&
        fact.identityBefore.identity === fact.identityAfter.identity
      )
        sameGenerationBefore.set(fact.process.pid, fact.identityBefore);
      if (fact.token.state === "readable" && fact.token.runId === options.runId)
        tokenMatchedPids.add(fact.process.pid);
    }

    const findScopeEvidence = (
      pid: number,
      generations: ReadonlyMap<
        number,
        Extract<ProcessIdentityObservation, { readonly state: "readable" }>
      >,
      tokenOwners: ReadonlySet<number>,
    ):
      | Exclude<
          RetainedProcessOwnershipCandidate["scopeEvidence"],
          "run-token" | "previously-related-generation"
        >
      | undefined => {
      const identity = generations.get(pid);
      if (identity === undefined) return undefined;
      const relation =
        identity.relation?.state === "observed" ? identity.relation : undefined;
      if (relation === undefined) return undefined;
      if (options.processGroupIds.has(relation.processGroupId))
        return "captured-process-group-id";
      const visited = new Set<number>([pid]);
      let parentPid = relation.parentPid;
      while (!visited.has(parentPid)) {
        const parent = generations.get(parentPid);
        if (parent === undefined) {
          return parentPid === options.leaderPid
            ? "capture-leader-pid"
            : undefined;
        }
        const previousParent = this.#candidates.get(parentPid);
        if (
          tokenOwners.has(parentPid) ||
          (previousParent?.applicability === "token-verified" &&
            parent.identity === previousParent.identity)
        )
          return "current-token-owned-parent";
        if (parentPid === options.leaderPid) return "capture-leader-pid";
        visited.add(parentPid);
        if (parent.relation?.state !== "observed") return undefined;
        parentPid = parent.relation.parentPid;
      }
      return undefined;
    };

    const scopeGenerations = new Map<
      number,
      Extract<ProcessIdentityObservation, { readonly state: "readable" }>
    >();
    const scopeBeforeGenerations = new Map<
      number,
      Extract<ProcessIdentityObservation, { readonly state: "readable" }>
    >();
    const scopeOwnedPids = new Set<number>();
    // Keep the original batch together: retries can reparent a child after its
    // verified parent exits. Historical parent proof is scope evidence only;
    // it neither establishes current liveness nor authorizes a signal.
    for (const observed of options.scopeFacts ?? facts) {
      const fact = observed.beforeExitSettlement ?? observed;
      if (
        fact.state !== "same-generation" ||
        fact.identityAfter.state !== "readable"
      )
        continue;
      scopeGenerations.set(fact.process.pid, fact.identityAfter);
      if (
        fact.identityBefore.state === "readable" &&
        fact.identityAfter.identity === fact.identityBefore.identity
      )
        scopeBeforeGenerations.set(fact.process.pid, fact.identityBefore);
      // Capture earlier proof before applying terminal observations below.
      // Its generation must match this historical native observation exactly.
      const previous = this.#candidates.get(fact.process.pid);
      if (
        (fact.token.state === "readable" &&
          fact.token.runId === options.runId) ||
        (previous?.applicability === "token-verified" &&
          previous.identity === fact.identityAfter.identity)
      )
        scopeOwnedPids.add(fact.process.pid);
    }
    const related = (pid: number) => {
      const current =
        findScopeEvidence(pid, sameGeneration, tokenMatchedPids) ??
        findScopeEvidence(pid, sameGenerationBefore, tokenMatchedPids);
      if (current !== undefined) return current;
      // The child must still be the generation observed in that original
      // batch. A reused PID cannot inherit its predecessor's historical scope.
      const original = scopeGenerations.get(pid);
      if (
        original === undefined ||
        sameGeneration.get(pid)?.identity !== original.identity
      )
        return undefined;
      const initial =
        findScopeEvidence(pid, scopeGenerations, scopeOwnedPids) ??
        findScopeEvidence(pid, scopeBeforeGenerations, scopeOwnedPids);
      return initial === undefined
        ? undefined
        : "previously-related-generation";
    };

    for (const [pid, previous] of this.#candidates) {
      const fact = factsByPid.get(pid);
      if (fact === undefined) continue;
      // Retire only this generation's current ledger entry. Darwin's
      // `not-in-snapshot` and Linux's corroborated ENOENT/ESRCH facts are
      // bounded observations, not claims about later PID reuse or global
      // liveness.
      if (
        fact.state === "absent" ||
        (previous.identity !== undefined &&
          fact.identityAfter.state === "readable" &&
          fact.identityAfter.identity !== previous.identity) ||
        (fact.state === "same-generation" &&
          processIdentityHasExited(fact.identityAfter))
      ) {
        this.#candidates.delete(pid);
        retiredPids.add(pid);
        continue;
      }
      if (fact.state !== "same-generation") {
        uncertainPids.add(pid);
        continue;
      }
      const identity = fact.identityAfter;
      if (identity.state !== "readable") {
        uncertainPids.add(pid);
        continue;
      }
      const matched =
        fact.token.state === "readable" && fact.token.runId === options.runId;
      if (previous.identity === undefined && !matched) {
        uncertainPids.add(pid);
        this.#candidates.set(pid, {
          ...previous,
          process: fact.process,
          token: fact.token,
          ...(fact.token.state === "unavailable"
            ? { diagnostic: fact.token.reason }
            : {}),
        });
        continue;
      }
      const relation =
        identity.relation?.state === "observed"
          ? {
              state: "observed" as const,
              parentPid: identity.relation.parentPid,
              processGroupId: identity.relation.processGroupId,
            }
          : undefined;
      this.#candidates.set(pid, {
        pid,
        identity: identity.identity,
        process: fact.process,
        applicability:
          matched || previous.applicability === "token-verified"
            ? "token-verified"
            : "related-ownership-unresolved",
        scopeEvidence:
          matched || previous.applicability === "token-verified"
            ? "run-token"
            : (related(pid) ?? "previously-related-generation"),
        token: fact.token,
        relation,
        ...(fact.token.state === "unavailable"
          ? { diagnostic: fact.token.reason }
          : {}),
      });
    }

    for (const fact of facts) {
      const pid = fact.process.pid;
      if (
        this.#candidates.has(pid) ||
        (retiredPids.has(pid) && fact.state !== "same-generation") ||
        fact.state === "absent" ||
        (fact.state === "same-generation" &&
          processIdentityHasExited(fact.identityAfter))
      )
        continue;
      const identity = fact.identityAfter;
      if (fact.state !== "same-generation" || identity.state !== "readable") {
        this.#candidates.set(pid, {
          pid,
          identity: undefined,
          process: fact.process,
          applicability: "identity-unresolved",
          scopeEvidence: "native-identity-unresolved",
          token: fact.token,
          relation: undefined,
          ...(fact.token.state === "unavailable"
            ? { diagnostic: fact.token.reason }
            : {}),
        });
        uncertainPids.add(pid);
        continue;
      }
      const matched =
        fact.token.state === "readable" && fact.token.runId === options.runId;
      const unresolvedScope = matched ? undefined : related(pid);
      if (!matched && unresolvedScope === undefined) continue;
      const relation =
        identity.relation?.state === "observed"
          ? {
              state: "observed" as const,
              parentPid: identity.relation.parentPid,
              processGroupId: identity.relation.processGroupId,
            }
          : undefined;
      this.#candidates.set(pid, {
        pid,
        identity: identity.identity,
        process: fact.process,
        applicability: matched
          ? "token-verified"
          : "related-ownership-unresolved",
        scopeEvidence: matched
          ? "run-token"
          : (unresolvedScope ?? "previously-related-generation"),
        token: fact.token,
        relation,
        ...(fact.token.state === "unavailable"
          ? { diagnostic: fact.token.reason }
          : {}),
      });
    }

    const retainedOnlyPids = [...this.#candidates.keys()].filter(
      (pid) => !options.listedPids.has(pid),
    );
    this.#coverage = {
      snapshotCount: this.#snapshotCount,
      listedRows: options.listedRows,
      examinedPids: facts.length,
      retainedOnlyPids: retainedOnlyPids.length,
      retainedOwned: [...this.#candidates.values()].filter(
        ({ applicability }) => applicability === "token-verified",
      ).length,
      retainedUnresolved: [...this.#candidates.values()].filter(
        ({ applicability }) => applicability !== "token-verified",
      ).length,
      uncertainPids: [...uncertainPids].sort((left, right) => left - right),
      processTableCoverage: "unreported-by-host",
    };
    return this.candidates;
  }

  /** Merge current discovery rows with retained candidates without PID-only proof. */
  rowsForObservation(
    processes: readonly ProcessTableEntry[],
  ): readonly ProcessTableEntry[] {
    const byPid = new Map(processes.map((process) => [process.pid, process]));
    for (const candidate of this.#candidates.values())
      if (!byPid.has(candidate.pid))
        byPid.set(candidate.pid, candidate.process);
    return [...byPid.values()];
  }
}