From a2d36d51922252ad40d9f6e70a494af1d4bf8bb1 Mon Sep 17 00:00:00 2001
From: thatben <thatbenbierens@gmail.com>
Date: Fri, 7 Feb 2025 15:35:40 +0100
Subject: [PATCH] basic crawling

---
 codexcrawler/crawler.nim | 19 +++++++++++++++++--
 codexcrawler/list.nim    | 24 +++++++++++++++++++-----
 2 files changed, 36 insertions(+), 7 deletions(-)

diff --git a/codexcrawler/crawler.nim b/codexcrawler/crawler.nim
index 680bf6e..c3b8124 100644
--- a/codexcrawler/crawler.nim
+++ b/codexcrawler/crawler.nim
@@ -7,6 +7,8 @@ import ./dht
 import ./list
 import ./nodeentry
 
+import std/sequtils
+
 logScope:
   topics = "crawler"
 
@@ -16,6 +18,11 @@ type Crawler* = ref object
   okNodes: List
   nokNodes: List
 
+# This is not going to stay this way.
+proc isNew(c: Crawler, node: Node): bool =
+  not c.todoNodes.contains(node.id) and not c.okNodes.contains(node.id) and
+    not c.nokNodes.contains(node.id)
+
 proc handleNodeNotOk(c: Crawler, target: NodeEntry) {.async.} =
   if err =? (await c.nokNodes.add(target)).errorOption:
     error "Failed to add not-OK-node to list", err = err.msg
@@ -34,16 +41,24 @@ proc addNewTodoNodes(c: Crawler, newNodes: seq[Node]) {.async.} =
       error "Failed to add todo-node to list", err = err.msg
 
 proc step(c: Crawler) {.async.} =
+  logScope:
+    todo = $c.todoNodes.len
+    ok = $c.okNodes.len
+    nok = $c.nokNodes.len
+
   without target =? (await c.todoNodes.pop()), err:
     error "Failed to get todo node", err = err.msg
 
   # todo: update target timestamp
 
-  without newNodes =? (await c.dht.getNeighbors(target.id)), err:
-    trace "getNeighbors call failed", node = $target.id, err = err.msg
+  without receivedNodes =? (await c.dht.getNeighbors(target.id)), err:
+    trace "Call failed", node = $target.id, err = err.msg
     await c.handleNodeNotOk(target)
     return
 
+  let newNodes = receivedNodes.filterIt(isNew(c, it))
+
+  trace "Received nodes", receivedNodes = receivedNodes.len, newNodes = newNodes.len
   await c.handleNodeOk(target)
   await c.addNewTodoNodes(newNodes)
 
diff --git a/codexcrawler/list.nim b/codexcrawler/list.nim
index 82e4719..e083e1a 100644
--- a/codexcrawler/list.nim
+++ b/codexcrawler/list.nim
@@ -12,6 +12,7 @@ import pkg/codexdht
 
 import std/sets
 import std/strutils
+import std/sequtils
 import std/os
 
 import ./nodeentry
@@ -25,7 +26,7 @@ type
   List* = ref object
     name: string
     store: TypedDatastore
-    items: HashSet[NodeEntry]
+    items: seq[NodeEntry]
     onMetric: OnUpdateMetric
 
 proc encode(s: NodeEntry): seq[byte] =
@@ -40,7 +41,8 @@ proc decode(T: type NodeEntry, bytes: seq[byte]): ?!T =
   if tokens.len != 2:
     return failure("expected 2 tokens")
 
-  success(NodeEntry(id: NodeId(tokens[0].u256), value: tokens[1]))
+  let id = UInt256.fromHex(tokens[0])
+  success(NodeEntry(id: id, value: tokens[1]))
 
 proc saveItem(this: List, item: NodeEntry): Future[?!void] {.async.} =
   without itemKey =? Key.init(this.name / $item.id), err:
@@ -66,7 +68,7 @@ proc load*(this: List): Future[?!void] {.async.} =
     without value =? item.value, err:
       return failure(err)
     if value.value.len > 0:
-      this.items.incl(value)
+      this.items.add(value)
 
   this.onMetric(this.items.len.int64)
   info "Loaded list", name = this.name, items = this.items.len
@@ -77,8 +79,17 @@ proc new*(
 ): List =
   List(name: name, store: store, onMetric: onMetric)
 
+proc contains*(this: List, nodeId: NodeId): bool =
+  this.items.anyIt(it.id == nodeId)
+
+proc contains*(this: List, item: NodeEntry): bool =
+  this.contains(item.id)
+
 proc add*(this: List, item: NodeEntry): Future[?!void] {.async.} =
-  this.items.incl(item)
+  if this.contains(item):
+    return success()
+
+  this.items.add(item)
   this.onMetric(this.items.len.int64)
 
   if err =? (await this.saveItem(item)).errorOption:
@@ -89,7 +100,10 @@ proc pop*(this: List): Future[?!NodeEntry] {.async.} =
   if this.items.len < 1:
     return failure(this.name & "List is empty.")
 
-  let item = this.items.pop()
+  let item = this.items[0]
+  this.items.delete(0)
+  this.onMetric(this.items.len.int64)
+
   if err =? (await this.removeItem(item)).errorOption:
     return failure(err)
   return success(item)