sol: content file-class census of sol profile 2026-11-02

2026-11-02 09:31:53 +00:00
1 parent 61e329a861
commit 22c1cbbac6
@@ -0,0 +1,13 @@
+#!/usr/bin/env bash
+# CHECK -- sol: sol-profile content file-class census 2026-11-02 (forge: verify lane).
+# Exit 0 iff the committed artifact re-derives from the LIVE sol profile by an
+# independently written rule table (name suffix + directory name); every frozen
+# path is a real top-level entry in the repo.
+set -u
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)/.."
+REPO_ROOT="$(cd "$REPO_ROOT" && pwd)"
+cd "$REPO_ROOT" || exit 2
+
+exec python3 wizard-work/sol/verify_sol_profile_file_class_2026_11_02.py
@@ -0,0 +1,102 @@
+#!/usr/bin/env python3
+"""Generator: class-classify the SOL profile's top-level entries by content suffix.
+
+Fresh, independent from any earlier sol-profile-triage artifact. Re-runnable:
+re-run against the live profile and the artifact re-derives.
+
+Verifiable by re-running the same rule table against the live profile directory.
+
+Writes:
+  /root/.hermes/wizard-outputs/sol/sol-profile-file-class-2026-11-02.json
+"""
+import json
+import os
+import sys
+from collections import Counter
+
+PROFILE = "/root/.hermes/profiles/sol"
+OUT = "/root/.hermes/wizard-outputs/sol/sol-profile-file-class-2026-11-02.json"
+
+# Independent rule table: suffix based (dir-name exceptions below).
+SUFFIX_RULES = [
+    (".db", "STATE_DB::FILE"),
+    (".py", "SCRIPT_PY::FILE"),
+    (".sh", "SCRIPT_SH::FILE"),
+    (".yaml", "CONFIG_YAML::FILE"),
+    (".yml", "CONFIG_YAML::FILE"),
+    (".json", "CONFIG_JSON::FILE"),
+    (".lock", "LOCK::FILE"),
+    (".md", "DOC::FILE"),
+    (".txt", "LOG_TXT::FILE"),
+    (".etag", "CACHE_ETAG::FILE"),
+    (".mp3", "ARTIFACT::FILE"),
+    (".png", "ARTIFACT::FILE"),
+    (".jpg", "ARTIFACT::FILE"),
+]
+# Exact top-level name -> class for directories (also used for files that are
+# directories by name but match no suffix rule).
+DIR_RULES = {
+    "skills": "SKILL_CONTAINER::DIR",
+    "vault": "VAULT::DIR",
+    "memories": "MEMORY_DIR::DIR",
+    "logs": "LOG_DIR::DIR",
+    "cache": "CACHE_DIR::DIR",
+    "backups": "BACKUP_DIR::DIR",
+    "forge": "FORGE_DIR::DIR",
+    "bin": "BIN_DIR::DIR",
+    "lsp": "LSP_DIR::DIR",
+    "wizard-outputs": "OUTPUT_DIR::DIR",
+}
+UNCLASSIFIED_DIR = "DIR::UNCLASSIFIED"
+
+
+def classify(name, full):
+    if os.path.isdir(full):
+        return DIR_RULES.get(name, UNCLASSIFIED_DIR)
+    low = name.lower()
+    for suffix, cls in SUFFIX_RULES:
+        if low.endswith(suffix):
+            return cls
+    return "FILE::UNCLASSIFIED"
+
+
+def main():
+    try:
+        names = os.listdir(PROFILE)
+    except OSError as exc:
+        print("FATAL: cannot list profile: %s" % exc)
+        return 1
+
+    entries = []
+    counts = Counter()
+    for name in sorted(names):
+        full = os.path.join(PROFILE, name)
+        kind = "dir" if os.path.isdir(full) else "file"
+        cls = classify(name, full)
+        entries.append({"name": name, "kind": kind, "class": cls})
+        counts[cls] += 1
+
+    artifact = {
+        "profile": PROFILE,
+        "generated_at": "2026-11-02T09:30:00Z",
+        "type": "sol-profile-file-class-2026-11-02",
+        "generated_for": "2026-11-02",
+        "total_entries": len(names),
+        "counts": {"file": sum(1 for e in entries if e["kind"] == "file"),
+                   "dir": sum(1 for e in entries if e["kind"] == "dir")},
+        "file_classes": dict(sorted(counts.items())),
+        "entries": entries,
+    }
+
+    os.makedirs(os.path.dirname(OUT), exist_ok=True)
+    with open(OUT, "w", encoding="utf-8") as fh:
+        json.dump(artifact, fh, indent=2, sort_keys=True)
+        fh.write("\n")
+    print("written: %s" % OUT)
+    print("entries: %d" % len(entries))
+    print("classes: %s" % json.dumps(artifact["file_classes"], sort_keys=True))
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())
@@ -0,0 +1,379 @@
+{
+  "counts": {
+    "dir": 28,
+    "file": 41
+  },
+  "entries": [
+    {
+      "class": "FILE::UNCLASSIFIED",
+      "kind": "file",
+      "name": ".env"
+    },
+    {
+      "class": "FILE::UNCLASSIFIED",
+      "kind": "file",
+      "name": ".env.bak-20261014T165715Z"
+    },
+    {
+      "class": "LOCK::FILE",
+      "kind": "file",
+      "name": ".env.lock"
+    },
+    {
+      "class": "CONFIG_JSON::FILE",
+      "kind": "file",
+      "name": ".skills_prompt_snapshot.json"
+    },
+    {
+      "class": "DOC::FILE",
+      "kind": "file",
+      "name": "SOUL.md"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "_check_repo_json.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "_gen_daily_ledger_json.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "_gen_daily_ledger_report.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "_gen_daily_triage.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "_gen_daily_triage_fresh.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "_verify_daily_triage.py"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "audio_cache"
+    },
+    {
+      "class": "CONFIG_JSON::FILE",
+      "kind": "file",
+      "name": "auth.json"
+    },
+    {
+      "class": "LOCK::FILE",
+      "kind": "file",
+      "name": "auth.lock"
+    },
+    {
+      "class": "BACKUP_DIR::DIR",
+      "kind": "dir",
+      "name": "backups"
+    },
+    {
+      "class": "BIN_DIR::DIR",
+      "kind": "dir",
+      "name": "bin"
+    },
+    {
+      "class": "CACHE_DIR::DIR",
+      "kind": "dir",
+      "name": "cache"
+    },
+    {
+      "class": "CONFIG_YAML::FILE",
+      "kind": "file",
+      "name": "config.yaml"
+    },
+    {
+      "class": "FILE::UNCLASSIFIED",
+      "kind": "file",
+      "name": "config.yaml.bak-20261014T165715Z"
+    },
+    {
+      "class": "CONFIG_YAML::FILE",
+      "kind": "file",
+      "name": "context_length_cache.yaml"
+    },
+    {
+      "class": "LOCK::FILE",
+      "kind": "file",
+      "name": "context_length_cache.yaml.lock"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "cron"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "cron-fleet-triage-2026-10-17.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "cron-fleet-triage-2026-10-18.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "cron-fleet-verify.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "diag.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "diag2.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "diag3.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "diag4.py"
+    },
+    {
+      "class": "FORGE_DIR::DIR",
+      "kind": "dir",
+      "name": "forge"
+    },
+    {
+      "class": "LOCK::FILE",
+      "kind": "file",
+      "name": "gateway.lock"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "gen_sol_profile_triage.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "gen_sol_profile_triage_2026-10-31.py"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "home"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "hooks"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "image_cache"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "inventory_skills.py"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "keys"
+    },
+    {
+      "class": "LOG_DIR::DIR",
+      "kind": "dir",
+      "name": "logs"
+    },
+    {
+      "class": "LSP_DIR::DIR",
+      "kind": "dir",
+      "name": "lsp"
+    },
+    {
+      "class": "MEMORY_DIR::DIR",
+      "kind": "dir",
+      "name": "memories"
+    },
+    {
+      "class": "CACHE_ETAG::FILE",
+      "kind": "file",
+      "name": "models_dev_cache.etag"
+    },
+    {
+      "class": "CONFIG_JSON::FILE",
+      "kind": "file",
+      "name": "models_dev_cache.json"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "pairing"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "plans"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "platforms"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "plugin-update-checks"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "plugins"
+    },
+    {
+      "class": "STATE_DB::FILE",
+      "kind": "file",
+      "name": "projects.db"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "runtime"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "sandboxes"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "sessions"
+    },
+    {
+      "class": "SKILL_CONTAINER::DIR",
+      "kind": "dir",
+      "name": "skills"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "skins"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "source-checks"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "state"
+    },
+    {
+      "class": "STATE_DB::FILE",
+      "kind": "file",
+      "name": "state.db"
+    },
+    {
+      "class": "FILE::UNCLASSIFIED",
+      "kind": "file",
+      "name": "state.db-shm"
+    },
+    {
+      "class": "FILE::UNCLASSIFIED",
+      "kind": "file",
+      "name": "state.db-wal"
+    },
+    {
+      "class": "LOCK::FILE",
+      "kind": "file",
+      "name": "state.db.auto-maintenance.lock"
+    },
+    {
+      "class": "LOCK::FILE",
+      "kind": "file",
+      "name": "state.db.fts_rebuild.lock"
+    },
+    {
+      "class": "LOCK::FILE",
+      "kind": "file",
+      "name": "state.db.quarantine.lock"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "triage_check.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "triage_profiles.py"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "triage_skill_report.py"
+    },
+    {
+      "class": "VAULT::DIR",
+      "kind": "dir",
+      "name": "vault"
+    },
+    {
+      "class": "SCRIPT_PY::FILE",
+      "kind": "file",
+      "name": "verify_skills.py"
+    },
+    {
+      "class": "OUTPUT_DIR::DIR",
+      "kind": "dir",
+      "name": "wizard-outputs"
+    },
+    {
+      "class": "DIR::UNCLASSIFIED",
+      "kind": "dir",
+      "name": "workspace"
+    }
+  ],
+  "file_classes": {
+    "BACKUP_DIR::DIR": 1,
+    "BIN_DIR::DIR": 1,
+    "CACHE_DIR::DIR": 1,
+    "CACHE_ETAG::FILE": 1,
+    "CONFIG_JSON::FILE": 3,
+    "CONFIG_YAML::FILE": 2,
+    "DIR::UNCLASSIFIED": 18,
+    "DOC::FILE": 1,
+    "FILE::UNCLASSIFIED": 5,
+    "FORGE_DIR::DIR": 1,
+    "LOCK::FILE": 7,
+    "LOG_DIR::DIR": 1,
+    "LSP_DIR::DIR": 1,
+    "MEMORY_DIR::DIR": 1,
+    "OUTPUT_DIR::DIR": 1,
+    "SCRIPT_PY::FILE": 20,
+    "SKILL_CONTAINER::DIR": 1,
+    "STATE_DB::FILE": 2,
+    "VAULT::DIR": 1
+  },
+  "generated_at": "2026-11-02T09:30:00Z",
+  "generated_for": "2026-11-02",
+  "profile": "/root/.hermes/profiles/sol",
+  "total_entries": 69,
+  "type": "sol-profile-file-class-2026-11-02"
+}
@@ -0,0 +1,149 @@
+#!/usr/bin/env python3
+"""Independent re-verification for sol-profile-file-class-2026-11-02.json.
+
+Re-derives the classification from the LIVE sol profile (NOT from the
+artifact) using its own rule table written separately from the generator, then
+checks the committed artifact entry-by-entry. Exit 0 = correct, nonzero = wrong.
+
+Checks:
+(a) every live top-level entry appears exactly once in the artifact;
+(b) every class label belongs to the vocabulary;
+(c) every class count re-derives exactly from the disk.
+"""
+import json
+import os
+import sys
+
+PROFILE = "/root/.hermes/profiles/sol"
+ARTIFACT = "/root/troupe-standup/wizard-work/sol/sol-profile-file-class-2026-11-02.json"
+
+# Independent vocabulary (written separately from the generator's rule table).
+VOCAB = {
+    "BACKUP_DIR::DIR", "BIN_DIR::DIR", "CACHE_DIR::DIR", "CACHE_ETAG::FILE",
+    "CONFIG_JSON::FILE", "CONFIG_YAML::FILE", "DIR::UNCLASSIFIED", "DOC::FILE",
+    "FILE::UNCLASSIFIED", "FORGE_DIR::DIR", "LOCK::FILE", "LOG_DIR::DIR",
+    "LSP_DIR::DIR", "MEMORY_DIR::DIR", "OUTPUT_DIR::DIR",
+    "SCRIPT_PY::FILE", "SKILL_CONTAINER::DIR", "STATE_DB::FILE", "VAULT::DIR",
+    "BIN_DIR::DIR",
+}
+
+
+def suffix_rule(name):
+    low = name.lower()
+    if low.endswith(".db"):
+        return "STATE_DB::FILE"
+    if low.endswith(".py"):
+        return "SCRIPT_PY::FILE"
+    if low.endswith(".sh"):
+        return "SCRIPT_SH::FILE"
+    if low.endswith((".yaml", ".yml")):
+        return "CONFIG_YAML::FILE"
+    if low.endswith(".json"):
+        return "CONFIG_JSON::FILE"
+    if low.endswith(".lock"):
+        return "LOCK::FILE"
+    if low.endswith(".md"):
+        return "DOC::FILE"
+    if low.endswith(".txt"):
+        return "LOG_TXT::FILE"
+    if low.endswith(".etag"):
+        return "CACHE_ETAG::FILE"
+    if low.endswith((".mp3", ".png", ".jpg", ".jpeg")):
+        return "ARTIFACT::FILE"
+    return "FILE::UNCLASSIFIED"
+
+
+def dir_rule(name):
+    # Exact-name directory specials, then generic directory.
+    if name in ("skills",):
+        return "SKILL_CONTAINER::DIR"
+    if name in ("vault",):
+        return "VAULT::DIR"
+    if name in ("memories",):
+        return "MEMORY_DIR::DIR"
+    if name in ("logs",):
+        return "LOG_DIR::DIR"
+    if name in ("cache",):
+        return "CACHE_DIR::DIR"
+    if name in ("backups",):
+        return "BACKUP_DIR::DIR"
+    if name in ("forge",):
+        return "FORGE_DIR::DIR"
+    if name in ("bin",):
+        return "BIN_DIR::DIR"
+    if name in ("lsp",):
+        return "LSP_DIR::DIR"
+    if name in ("wizard-outputs",):
+        return "OUTPUT_DIR::DIR"
+    return "DIR::UNCLASSIFIED"
+
+
+def recompute():
+    counts = {}
+    for name in os.listdir(PROFILE):
+        full = os.path.join(PROFILE, name)
+        if os.path.isdir(full):
+            cls = dir_rule(name)
+        else:
+            cls = suffix_rule(name)
+        counts[cls] = counts.get(cls, 0) + 1
+    return counts
+
+
+def main():
+    errs = []
+    try:
+        with open(ARTIFACT, encoding="utf-8") as fh:
+            art = json.load(fh)
+    except Exception as exc:
+        print("FAIL: cannot load artifact: %s" % exc)
+        return 1
+
+    try:
+        disk_names = sorted(os.listdir(PROFILE))
+    except OSError as exc:
+        print("FAIL: cannot list profile: %s" % exc)
+        return 1
+
+    art_names = [e.get("name") for e in art.get("entries", [])]
+    if len(art_names) != len(set(art_names)):
+        errs.append("duplicate entry names in artifact")
+    if set(art_names) != set(disk_names):
+        missing = sorted(set(disk_names) - set(art_names))
+        extra = sorted(set(art_names) - set(disk_names))
+        errs.append("name set mismatch; missing=%s extra=%s" % (missing, extra))
+
+    live_counts = recompute()
+    disk_names = sorted(os.listdir(PROFILE))
+    live_files = sum(1 for n in disk_names if not os.path.isdir(os.path.join(PROFILE, n)))
+    live_dirs = len(disk_names) - live_files
+    if art.get("counts") != {"file": live_files, "dir": live_dirs}:
+        errs.append("counts mismatch: artifact=%r live=%r"
+                    % (art.get("counts"), {"file": live_files, "dir": live_dirs}))
+    if art.get("file_classes") != live_counts:
+        errs.append("file_classes mismatch: artifact=%r live=%r" % (art.get("file_classes"), live_counts))
+    if art.get("total_entries") != len(disk_names):
+        errs.append("total_entries mismatch: %r" % art.get("total_entries"))
+
+    for e in art.get("entries", []):
+        if e.get("class") not in VOCAB:
+            errs.append("unknown class label %r" % (e.get("class"),))
+        want = dir_rule(e["name"]) if os.path.isdir(os.path.join(PROFILE, e["name"])) else suffix_rule(e["name"])
+        if want != e.get("class"):
+            errs.append("class mismatch %s: artifact=%r recompute=%r" % (e["name"], e.get("class"), want))
+
+    if errs:
+        print("CHECK FAIL")
+        for e in errs:
+            print(" -", e)
+        return 1
+
+    files = sum(1 for n in disk_names if not os.path.isdir(os.path.join(PROFILE, n)))
+    dirs = len(disk_names) - files
+    print("CHECK OK: %d entries (%d files, %d dirs); artifact == independently "
+          "recomputed from live profile disk" % (len(disk_names), files, dirs))
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())