Back out app auto-seeding; fix report-status + PCTypesStrict bugs (manifest review)

A deep Fable review of the real manifest corpus (READ-ONLY reference) showed the
manifests are an ENFORCEMENT PROGRAM, not an application inventory, and that
auto-seeding the Applications catalog from entry Type + Name was wrong:

- The catalog ALREADY tracks these apps from the classic-shopdb migration, with
  version histories (PC - DMIS, UDC x11 versions, eMX / eDNC, CLM, CSF, Oracle
  Database, FormTracePak). Seeding from manifest labels created DUPLICATES under
  different names (PC-DMIS 2016 vs PC - DMIS; eDNC (bundles NTLARS) vs eMX / eDNC;
  OpenText HostExplorer ShopFloor vs CSF). It also misclassified config drops
  (eMxInfo.txt) as apps and could never match a PC's reported ARP name.
So the seed-applications command + service are removed. Properly linking
manifest entries to the EXISTING catalog is a curated feature, not label-scraping.

Two REAL bugs the review found are fixed and kept:
- Report status (R4): every healthy cycle runs Always/no-detection scripts the
  engine counts as "installed", so keying self-heal off installed>0 marked the
  common scope selfhealed forever and made 'ok' unreachable. Status now derives
  from explicit per-entry self-heal flags only; the stored flag no longer infers
  from action=='installed'; the client kit doc reflects it.
- PCTypesStrict (R5): the runtime engine has no strict handling (preinstall
  runner only). filters.matches_pctype now applies strict only when phase ==
  'preinstall'; simulate + parity thread the scope phase through; the strict test
  uses a preinstall scope.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

plugins/geenforce/api/routes.py

@@ -368,6 +368,7 @@ def simulate_scope(scopeid):
         'hostname': request.args.get('hostname'),
         'machinenumber': request.args.get('machinenumber'),
         'cmmversion': request.args.get('cmmversion'),
+        'phase': scope.phase,
     }
     applied, filtered = [], []
     for entry in scope.entries:
@@ -376,7 +377,9 @@ def simulate_scope(scopeid):
             applied.append(entry.name)
         else:
             reasons = []
-            if not matches_pctype(entry_dict, profile['pctype'], profile['subtype']):
+            strict_allowed = profile.get('phase') == 'preinstall'
+            if not matches_pctype(entry_dict, profile['pctype'],
+                                  profile['subtype'], strict_allowed):
                 reasons.append('PCTypes')
             if not matches_hostname(entry_dict, profile['hostname']):
                 reasons.append('TargetHostnames')

plugins/geenforce/filters.py

@@ -42,21 +42,24 @@ def _alias_sets(name):
     return [g for g in ALIAS_GROUPS if any(n.lower() == lname for n in g)]
 
 
-def matches_pctype(entry, pctype, subtype=None):
+def matches_pctype(entry, pctype, subtype=None, strict_allowed=False):
     """Test-PCTypeMatches: no PCTypes = all; '*' = all; alias-set intersection.
 
     PCTypesStrict=true disables alias expansion (both the PC's identity and the
-    manifest values are compared literally), so e.g. a 'collections'-only strict
+    manifest values are compared literally), so a 'collections'-only strict entry
-    entry does NOT match a 'nocollections' PC via their shared 'Standard' alias
+    does NOT match a 'nocollections' PC via their shared 'Standard' alias group.
-    group. The preinstall runner honors this flag (see preinstall.json UDC entry);
+    IMPORTANT: only the PREINSTALL runner honors this flag - the runtime engine
-    mirroring it keeps the "what would this PC get" simulator correct.
+    (Install-FromManifest.ps1's Test-PCTypeMatches) has no strict handling. So
+    strict is applied only when strict_allowed (i.e. phase == 'preinstall'),
+    else a runtime entry with PCTypesStrict still alias-expands, matching the
+    real engine.
     """
     values = entry.get('PCTypes') or []
     if not values:
         return True
     if not pctype:
         return True
-    strict = bool(entry.get('PCTypesStrict'))
+    strict = strict_allowed and bool(entry.get('PCTypesStrict'))
     # Names the current PC matches: bare type, "type-subtype", and (unless strict)
     # all aliases of either.
     mynames = set()
@@ -125,9 +128,12 @@ def matches_cmmversion(entry, cmmversion):
 def entry_applies(entry, profile):
     """All four filters ANDed, exactly as the engine's main loop applies them.
 
-    `profile` keys: pctype, subtype, hostname, machinenumber, cmmversion.
+    `profile` keys: pctype, subtype, hostname, machinenumber, cmmversion, and
+    optional phase ('preinstall' enables PCTypesStrict, matching the runners).
     """
-    return (matches_pctype(entry, profile.get('pctype'), profile.get('subtype'))
+    strict_allowed = profile.get('phase') == 'preinstall'
+    return (matches_pctype(entry, profile.get('pctype'), profile.get('subtype'),
+                           strict_allowed)
             and matches_hostname(entry, profile.get('hostname'))
             and matches_machinenumber(entry, profile.get('machinenumber'))
             and matches_cmmversion(entry, profile.get('cmmversion')))

plugins/geenforce/parity.py

@@ -59,9 +59,11 @@ def check_scope(scopename, phase, original, fixtures):
                 k for k in set(co) | set(cr) if co.get(k) != cr.get(k))
             firstdiff = f'entry "{name}" differs on {diffkeys}'
 
-    # Check 2: same entries fire in the same order for each profile.
+    # Check 2: same entries fire in the same order for each profile. Inject the
+    # scope phase so PCTypesStrict is honored only for preinstall (see filters).
     profiles_same = 0
-    for profile in fixtures:
+    for base_profile in fixtures:
+        profile = {**base_profile, 'phase': phase}
         if (applicable_entry_names(orig_apps, profile)
                 == applicable_entry_names(rebuilt_apps, profile)):
             profiles_same += 1

plugins/geenforce/plugin.py

@@ -178,27 +178,6 @@ class GeEnforcePlugin(BasePlugin):
                 db.session.commit()
                 click.echo(f"Published {scopename}/{phase} as v{version}.")
 
-        @geenforce_cli.command('seed-applications')
-        @click.option('--shareroot', required=True)
-        @click.option('--preinstall', default=None)
-        def seed_applications_cmd(shareroot, preinstall):
-            """Track the apps the manifests install in the core Applications catalog."""
-            from flask import current_app
-            from .importer import discover_share, load_manifest_file
-            from .service import seed_applications_from_manifests
-
-            with current_app.app_context():
-                sources = list(discover_share(shareroot))
-                if preinstall:
-                    sources.append(('preinstall', 'preinstall',
-                                    load_manifest_file(preinstall)))
-                result = seed_applications_from_manifests(sources)
-                db.session.commit()
-                click.echo(f"Applications: {len(result['created'])} created, "
-                           f"{len(result['existing'])} already tracked.")
-                for name in result['created']:
-                    click.echo(f"  + {name}")
-
         @geenforce_cli.command('export-share')
         @click.argument('scopename')
         @click.option('--shareroot', required=True)

plugins/geenforce/service.py

@@ -94,45 +94,6 @@ def rollback_scope(scopename, phase, versionnumber):
     return versionnumber
 
 
-# Entry types that install actual applications (vs File/Registry/PS1/INF config).
-INSTALLER_TYPES = {'MSI', 'EXE', 'CMD', 'BAT'}
-
-
-def seed_applications_from_manifests(manifests):
-    """Populate the core Applications catalog from what the manifests install.
-
-    For every installer entry (MSI/EXE/CMD/BAT) across the given manifests,
-    create an Application (idempotent, deduped by appname) so shopdb tracks the
-    apps GE-Enforce deploys. `manifests` is a list of (scopename, phase, dict).
-    Returns {'created': [...], 'existing': [...]} (uncommitted).
-    """
-    from shopdb.api import Application
-
-    created, existing, seen = [], [], set()
-    for _scopename, _phase, manifest in manifests:
-        for entry in manifest.get('Applications', []):
-            if entry.get('Type') not in INSTALLER_TYPES:
-                continue
-            name = (entry.get('Name') or '').strip()
-            if not name or name.lower() in seen:
-                continue
-            seen.add(name.lower())
-            if Application.query.filter(Application.appname.ilike(name)).first():
-                existing.append(name)
-                continue
-            description = None
-            comment = entry.get('_comment')
-            if comment:
-                description = comment.split('.')[0].strip()[:255]
-            db.session.add(Application(
-                appname=name[:100],
-                appdescription=description,
-                installpath=(entry.get('Installer') or '')[:255] or None,
-                isinstallable=True))
-            created.append(name)
-    return {'created': created, 'existing': existing}
-
-
 def record_enforcement_report(payload):
     """Record one PC's enforcement cycle (observed state). Upserts the latest
     report per (hostname, scopename, phase) and keeps prior ones as history.
@@ -154,10 +115,14 @@ def record_enforcement_report(payload):
 
     failed = int(counts.get('failed', 0))
     installed = int(counts.get('installed', 0))
-    # Derive status: any failure wins; else drift-corrected installs = selfhealed.
+    # Derive status from EXPLICIT self-heal flags only, never the raw installed
+    # count: every healthy cycle runs Always/no-detection scripts (asset report,
+    # VNC firewall, EventSaver) that the engine counts as "installed", so keying
+    # self-heal off installed>0 would mark the common scope selfhealed forever
+    # and make 'ok' unreachable. A drift correction is one the client flags.
     if failed > 0:
         status = 'failed'
-    elif installed > 0 or any(r.get('selfhealed') for r in results):
+    elif any(r.get('selfhealed') for r in results):
         status = 'selfhealed'
     else:
         status = 'ok'
@@ -190,8 +155,10 @@ def record_enforcement_report(payload):
         report.results.append(ManifestEnforcementResult(
             entryname=item.get('name', ''),
             action=item.get('action', ''),
-            selfhealed=bool(item.get('selfhealed',
+            # Explicit flag only; do NOT infer from action == 'installed'
-                                     item.get('action') == 'installed')),
+            # (Always/no-detection scripts install every cycle without being a
+            # drift correction).
+            selfhealed=bool(item.get('selfhealed', False)),
             exitcode=item.get('exitcode'),
             message=item.get('message')))
     db.session.add(report)

tests/test_plugins/test_geenforce_crud.py

@@ -155,8 +155,13 @@ def test_duplicate_entry_name_is_400_not_500(client, db, auth_headers):
 
 def test_simulate_pctypesstrict_disables_alias(client, db, auth_headers):
     """A collections-only strict entry must NOT match a nocollections PC via the
-    shared Standard alias group (mirrors the preinstall UDC entry)."""
+    shared Standard alias group. PCTypesStrict is preinstall-only, so the scope
-    scopeid = _create_scope(client, auth_headers, 'preinstall')
+    must be the preinstall phase (the runtime engine ignores strict)."""
+    resp = client.post('/api/geenforce/scopes',
+                       json={'scopename': 'preinstall', 'phase': 'preinstall'},
+                       headers=auth_headers)
+    assert resp.status_code == 201, resp.get_json()
+    scopeid = resp.get_json()['data']['scopeid']
     _add_entry(client, auth_headers, scopeid,
                {'Name': 'UDC (strict)', 'Type': 'EXE', 'Installer': 'apps/udc.exe',
                 'PCTypes': ['gea-shopfloor-collections'], 'PCTypesStrict': True})

tests/test_plugins/test_geenforce_seed_apps.py

@@ -1,54 +0,0 @@
-"""GE-Enforce seeds the core Applications catalog from what the manifests install.
-
-Only installer entries (MSI/EXE/CMD/BAT) become Applications; File/Registry/PS1/
-INF config entries do not. Idempotent + deduped by appname.
-"""
-
-from plugins.geenforce import service
-from shopdb.core.models import Application
-
-
-MANIFEST = {
-    'Version': '2.6',
-    'Applications': [
-        {'Name': 'eDNC', 'Type': 'MSI', 'Installer': 'apps/edns.msi',
-         '_comment': 'eDNC (Universal Data Collection). Bundles NTLARS.'},
-        {'Name': 'UDC', 'Type': 'EXE', 'Installer': 'apps/udc.exe'},
-        {'Name': 'Config drop', 'Type': 'File', 'Source': 'configs/x.json',
-         'Destination': 'C:\\x.json'},
-        {'Name': 'Firewall rule', 'Type': 'PS1', 'Script': 'scripts/fw.ps1'},
-        {'Name': 'FMS host pin', 'Type': 'Registry', 'RegPath': 'HKLM:\\x',
-         'RegName': 'y', 'RegValue': 1, 'RegType': 'DWord'},
-    ],
-}
-
-
-def test_seed_only_installer_apps(db):
-    result = service.seed_applications_from_manifests(
-        [('gea-shopfloor-collections', 'runtime', MANIFEST)])
-    service.db.session.commit()
-    assert set(result['created']) == {'eDNC', 'UDC'}
-    names = {a.appname for a in Application.query.all()}
-    assert 'eDNC' in names and 'UDC' in names
-    # config / script / registry entries are not applications
-    assert 'Config drop' not in names
-    assert 'Firewall rule' not in names
-    assert 'FMS host pin' not in names
-    edns = Application.query.filter_by(appname='eDNC').first()
-    assert edns.isinstallable is True
-    assert edns.installpath == 'apps/edns.msi'
-    assert edns.appdescription.startswith('eDNC (Universal Data Collection)')
-
-
-def test_seed_is_idempotent_and_deduped(db):
-    manifests = [('a', 'runtime', MANIFEST), ('b', 'runtime', MANIFEST)]
-    first = service.seed_applications_from_manifests(manifests)
-    service.db.session.commit()
-    # deduped across the two manifests: 2 created, not 4
-    assert len(first['created']) == 2
-    # re-run: nothing new, both already tracked
-    second = service.seed_applications_from_manifests(manifests)
-    service.db.session.commit()
-    assert second['created'] == []
-    assert set(second['existing']) == {'eDNC', 'UDC'}
-    assert Application.query.filter(Application.appname.in_(['eDNC', 'UDC'])).count() == 2