WIP

imageroot/actions/seek-snapshot-contents/50seek_snapshot_contents

@@ -0,0 +1,68 @@
+#!/usr/bin/env python3
+
+#
+# Copyright (C) 2024 Nethesis S.r.l.
+# SPDX-License-Identifier: GPL-3.0-or-later
+#
+
+import json
+import sys
+import os
+import subprocess
+import agent
+import hashlib
+
+request = json.load(sys.stdin)
+
+limit_reached = False
+counter_limit = request.get("limit", 1000)
+cache_seed = '-'.join(request['destination'], request['snapshot'], request['share'])
+plocate_cache = hashlib.md5(cache_seed.encode(), usedforsecurity=False).hexdigest()
+
+podman_exec = ["podman", "exec", "samba-dc"]
+pcheck = agent.run_helper(*(podman_exec + ["net", "conf", "showshare", request["share"]]))
+if pcheck.returncode != 0:
+    agent.set_status('validation-failed')
+    json.dump([{'field':'share', 'parameter':'share','value': request['share'], 'error':'share_not_found'}], fp=sys.stdout)
+    sys.exit(2)
+
+cache_exits = False
+if not cache_exists:
+    # Extract the .plocate files from the snapshot and store them under a
+    # temporary cache directory:
+    podman_args = ["--workdir=/srv"] + agent.agent.get_state_volume_args()
+    restic_args = [
+        "restore",
+        f"{request["snapshot"]}:volumes/data/backup",
+        "--include=*.plocate",
+        f"--target=/srv/volumes/data/plocate/{plocate_cache}"
+    ]
+    agent.run_restic(agent.redis_connect(), request["destination"], request["repopath"], podman_args, restic_args).check_returnvalue()
+    # TODO: remove stale cache dirs (age > 8hours?)
+
+# Search share paths matching the query:
+plocate_cmd = ['podman', 'exec', 'samba-dc', 'plocate', '-d', f"/var/lib/samba/plocate/{plocate_cache}/{request['share']}.plocate", request.get('query', "")]
+if request.get("basename"):
+    plocate_cmd += ["-b"] # restrict search to file basename
+with subprocess.Popen(plocate_cmd, stdout=subprocess.PIPE, stderr=sys.stderr, text=True) as vproc:
+    contents = []
+    counter = 0
+    while True:
+        line = vproc.stdout.readline()
+        if not line:
+            break
+        contents.append(line.rstrip())
+        counter += 1
+        if counter >= counter_limit:
+            limit_reached = True
+            break
+if vproc.returncode != 0:
+    print(agent.SD_ERR + f"Plocate command failed with exit code {vproc.returncode}.", file=sys.stderr)
+    sys.exit(1)
+
+contents.sort()
+json.dump({
+    "request": request,
+    "contents": contents,
+    "limit_reached": limit_reached,
+}, fp=sys.stdout)

imageroot/actions/seek-snapshot-contents/validate-input.json

@@ -0,0 +1,46 @@
+{
+    "$schema": "http://json-schema.org/draft-07/schema#",
+    "title": "seek-snapshot-contents input",
+    "$id": "http://schema.nethserver.org/mail/seek-snapshot-contents-input.json",
+    "description": "Extract the list of IMAP folders from a backup snapshot",
+    "examples": [
+
+    ],
+    "type": "object",
+    "required": [
+        "destination",
+        "repopath",
+        "snapshot",
+        "share"
+    ],
+    "properties": {
+        "destination": {
+            "type": "string",
+            "description": "The UUID of the backup destination where the Restic repository resides."
+        },
+        "repopath": {
+            "type": "string",
+            "description": "Restic repository path, under the backup destination"
+        },
+        "snapshot": {
+            "type": "string",
+            "description": "Restic snapshot ID to restore"
+        },
+        "share": {
+            "type": "string",
+            "description": "Seek the paths of this Samba share"
+        },
+        "query": {
+            "type": "string",
+            "description": "Seek matching paths"
+        },
+        "basename": {
+            "type":"boolean",
+            "description": "If true, match the path basename part"
+        },
+        "limit": {
+            "type": "integer",
+            "description": "Limit the number of returned contents"
+        }
+    }
+}

imageroot/actions/seek-snapshot-contents/validate-output.json

@@ -0,0 +1,47 @@
+{
+    "$schema": "http://json-schema.org/draft-07/schema#",
+    "title": "seek-snapshot-content output",
+    "$id": "http://schema.nethserver.org/mail/seek-snapshot-content-output.json",
+    "description": "Extract the list of IMAP folders from a backup snapshot",
+    "examples": [
+        {
+            "request": {
+                "destination": "14030a59-a4e6-54cc-b8ea-cd5f97fe44c8",
+                "repopath": "mail/4372a5d5-0886-45d3-82e7-68d913716a4c",
+                "snapshot": "latest",
+                "share": "myshare",
+                "query": "*.php",
+                "basename": true,
+                "limit": 1000
+            },
+            "contents": [
+                "dir1/file001.php",
+                "dir1/file002.php",
+                "Project/NethServer/Main.php"
+            ],
+            "limit_reached": false
+        }
+    ],
+    "type": "object",
+    "required": [
+        "contents",
+        "limit_reached"
+    ],
+    "properties": {
+        "contents": {
+            "type": "array",
+            "description": "List of absolute share content paths",
+            "items": {
+                "type": "string"
+            }
+        },
+        "limit_reached": {
+            "type":"boolean",
+            "description": "If true, the query matches more contents than the returned items"
+        },
+        "request": {
+            "type": "object",
+            "title": "Original request object"
+        }
+    }
+}