From 651efc1e6bd470b4f9803c7aa4766d798cc63eb7 Mon Sep 17 00:00:00 2001
From: Emery Hemingway <ehmry@posteo.net>
Date: Tue, 9 Jan 2024 20:49:54 +0200
Subject: [PATCH] syndesizer: add file-system-usage

---
 README.md                            | 17 +++++++++++++++++
 assertions.prs                       |  3 +++
 config.prs                           |  4 ++++
 src/schema/config.nim                |  8 ++++++++
 src/syndesizer.nim                   |  2 ++
 src/syndesizer/file_system_usage.nim | 27 +++++++++++++++++++++++++++
 6 files changed, 61 insertions(+)
 create mode 100644 assertions.prs
 create mode 100644 src/syndesizer/file_system_usage.nim

diff --git a/README.md b/README.md
index f1a5729..ef0837a 100644
--- a/README.md
+++ b/README.md
@@ -28,6 +28,23 @@ Example configuration:
     $cap <cache { dataspace: $nixspace lifetime: 3600.0 }> ]
 ]
 ```
+### File System Usage
+
+Summarize the size of file-system directory. Equivalent to `du -s -b`.
+Query the size of a directory in bytes by observing `<file-system-usage "/SOME/PATH" ?size>`.
+
+```
+# Configuration example
+
+? <exposed-dataspace ?ds> [
+
+  <require-service <daemon syndesizer>>
+  ? <service-object <daemon syndesizer> ?cap> [
+    $cap <file-system-usage { dataspace: $ds }>
+  ]
+
+]
+```
 
 ### JSON Socket Translator
 
diff --git a/assertions.prs b/assertions.prs
new file mode 100644
index 0000000..a556e81
--- /dev/null
+++ b/assertions.prs
@@ -0,0 +1,3 @@
+version 1.
+
+FileSystemUsage = <file-system-usage @path string @size int>.
diff --git a/config.prs b/config.prs
index 02094ce..3793d4b 100644
--- a/config.prs
+++ b/config.prs
@@ -6,6 +6,10 @@ CacheArguments = <cache {
   lifetime: float
 }>.
 
+FileSystemUsageArguments = <file-system-usage {
+  dataspace: #!any
+}>.
+
 JsonTranslatorArguments = <json-stdio-translator {
   argv: [string ...]
   dataspace: #!any
diff --git a/src/schema/config.nim b/src/schema/config.nim
index a22bf44..7f1ac5e 100644
--- a/src/schema/config.nim
+++ b/src/schema/config.nim
@@ -34,6 +34,12 @@ type
   WebhooksArguments* {.preservesRecord: "webhooks".} = object
     `field0`*: WebhooksArgumentsField0
 
+  FileSystemUsageArgumentsField0* {.preservesDictionary.} = object
+    `dataspace`* {.preservesEmbedded.}: EmbeddedRef
+
+  FileSystemUsageArguments* {.preservesRecord: "file-system-usage".} = object
+    `field0`*: FileSystemUsageArgumentsField0
+
   SqliteArgumentsField0* {.preservesDictionary.} = object
     `database`*: string
     `dataspace`* {.preservesEmbedded.}: EmbeddedRef
@@ -56,6 +62,7 @@ proc `$`*(x: WebsocketArguments | JsonTranslatorArguments |
     JsonTranslatorConnected |
     JsonSocketTranslatorArguments |
     WebhooksArguments |
+    FileSystemUsageArguments |
     SqliteArguments |
     CacheArguments |
     Tcp): string =
@@ -65,6 +72,7 @@ proc encode*(x: WebsocketArguments | JsonTranslatorArguments |
     JsonTranslatorConnected |
     JsonSocketTranslatorArguments |
     WebhooksArguments |
+    FileSystemUsageArguments |
     SqliteArguments |
     CacheArguments |
     Tcp): seq[byte] =
diff --git a/src/syndesizer.nim b/src/syndesizer.nim
index e77d93f..2c83606 100644
--- a/src/syndesizer.nim
+++ b/src/syndesizer.nim
@@ -7,6 +7,7 @@ import syndicate, syndicate/relays, syndicate/actors/timers
 
 import ./syndesizer/[
   cache_actor,
+  file_system_usage,
   json_socket_translator,
   json_translator,
   sqlite_actor,
@@ -17,6 +18,7 @@ runActor("syndesizer") do (turn: var Turn; root: Cap):
   connectStdio(turn, root)
   discard spawnTimers(turn, root)
   discard spawnCacheActor(turn, root)
+  discard spawnFileSystemUsageActor(turn, root)
   discard spawnJsonSocketTranslator(turn, root)
   discard spawnJsonStdioTranslator(turn, root)
   discard spawnWebhookActor(turn, root)
diff --git a/src/syndesizer/file_system_usage.nim b/src/syndesizer/file_system_usage.nim
new file mode 100644
index 0000000..1c6f10c
--- /dev/null
+++ b/src/syndesizer/file_system_usage.nim
@@ -0,0 +1,27 @@
+# SPDX-FileCopyrightText: ☭ Emery Hemingway
+# SPDX-License-Identifier: Unlicense
+
+import std/[dirs, os, paths]
+import preserves
+import syndicate, syndicate/relays
+
+import ../schema/[assertions, config]
+
+proc spawnFileSystemUsageActor*(turn: var Turn; root: Cap): Actor {.discardable.} =
+  spawn("file-system-usage", turn) do (turn: var Turn):
+    during(turn, root, ?:FileSystemUsageArguments) do (ds: Cap):
+      var pat = ?Observe(pattern: !FileSystemUsage) ?? { 0: grab() }
+      during(turn, ds, pat) do (lit: Literal[string]):
+        var ass = FileSystemUsage(path: lit.value)
+        if fileExists(ass.path): ass.size = getFileSize(ass.path)
+        else:
+          for fp in walkDirRec(paths.Path(lit.value), yieldFilter={pcFile}):
+            var fs = getFileSize(string fp)
+            inc(ass.size, fs)
+        discard publish(turn, ds, ass)
+          # TODO: updates?
+
+when isMainModule:
+  runActor("main") do (turn: var Turn; root: Cap):
+    connectStdio(turn, root)
+    discard spawnFileSystemUsageActor(turn, root)