From 514e521abe14f61847d68f28c8dd0b4ef6879210 Mon Sep 17 00:00:00 2001
From: Markus Maiwald <markus@maiwald.work>
Date: Mon, 2 Feb 2026 12:19:40 +0100
Subject: [PATCH] feat(ncp): add L1 index layer with path-based addressing

Implement B-Tree style index:
- Path to CID mapping (exact lookup)
- List by prefix (directory listing)
- Glob pattern matching (* and ? wildcards)
- Insert, remove, paths operations

Supports hierarchical addressing: /agents/frankie/tasks

Refs RFC-NCP-001 L1 Index
Developed with Frankie assistance
---
 ncp-prototype/src/l1_index.nim | 114 +++++++++++++++++++++++++++++++++
 1 file changed, 114 insertions(+)
 create mode 100644 ncp-prototype/src/l1_index.nim

diff --git a/ncp-prototype/src/l1_index.nim b/ncp-prototype/src/l1_index.nim
new file mode 100644
index 0000000..7180f04
--- /dev/null
+++ b/ncp-prototype/src/l1_index.nim
@@ -0,0 +1,114 @@
+## l1_index.nim: L1 Index Layer for NCP
+## B-Tree index for path-based addressing
+## RFC-NCP-001 Implementation
+
+import std/[tables, sequtils, algorithm, strutils]
+import types
+
+## Index Entry: Maps path to CID
+type IndexEntry* = object
+  path*: string      ## Hierarchical path (e.g., "/agents/frankie/tasks")
+  cid*: CID          ## Content identifier
+  timestamp*: int64  ## When indexed
+
+## B-Tree Node (simplified for prototype)
+type BTreeNode* = object
+  isLeaf*: bool
+  keys*: seq[string]      ## Paths
+  values*: seq[CID]       ## CIDs
+  children*: seq[int]     ## Child node indices (for internal nodes)
+
+## L1 Index Handle
+type L1Index* = object
+  entries*: Table[string, IndexEntry]  ## Path -> Entry (simplified B-Tree)
+  root*: string
+
+proc initL1Index*(): L1Index =
+  ## Initialize empty L1 Index
+  result.entries = initTable[string, IndexEntry]()
+  result.root = "/"
+
+## Insert or update path -> CID mapping
+proc insert*(index: var L1Index, path: string, cid: CID, timestamp: int64 = 0) =
+  ## Index a path to CID mapping
+  index.entries[path] = IndexEntry(
+    path: path,
+    cid: cid,
+    timestamp: if timestamp == 0: getTime().toUnix() else: timestamp
+  )
+
+## Lookup CID by exact path
+proc lookup*(index: L1Index, path: string): Option[CID] =
+  ## Find CID by exact path
+  if index.entries.hasKey(path):
+    return some(index.entries[path].cid)
+  return none(CID)
+
+## List all paths under a prefix (directory listing)
+proc list*(index: L1Index, prefix: string): seq[string] =
+  ## List all paths starting with prefix
+  result = @[]
+  for path in index.entries.keys:
+    if path.startsWith(prefix):
+      result.add(path)
+  result.sort()
+
+## Find paths matching glob pattern (simplified)
+proc glob*(index: L1Index, pattern: string): seq[string] =
+  ## Find paths matching pattern
+  ## Supports: * (any chars), ? (single char)
+  result = @[]
+  for path in index.entries.keys:
+    # Simple glob matching (can be improved)
+    if matchGlob(path, pattern):
+      result.add(path)
+  result.sort()
+
+## Simple glob matcher
+proc matchGlob*(s, pattern: string): bool =
+  ## Match string against glob pattern
+  var sIdx = 0
+  var pIdx = 0
+  
+  while pIdx < pattern.len:
+    if pattern[pIdx] == '*':
+      # Match any sequence
+      if pIdx == pattern.len - 1:
+        return true  # * at end matches everything
+      # Find next char after *
+      let nextChar = pattern[pIdx + 1]
+      while sIdx < s.len and s[sIdx] != nextChar:
+        sIdx.inc
+      pIdx += 2
+    elif pattern[pIdx] == '?':
+      # Match single char
+      if sIdx >= s.len:
+        return false
+      sIdx.inc
+      pIdx.inc
+    else:
+      # Match literal
+      if sIdx >= s.len or s[sIdx] != pattern[pIdx]:
+        return false
+      sIdx.inc
+      pIdx.inc
+  
+  return sIdx == s.len
+
+## Delete path from index
+proc remove*(index: var L1Index, path: string): bool =
+  ## Remove path from index
+  if index.entries.hasKey(path):
+    index.entries.del(path)
+    return true
+  return false
+
+## Get all indexed paths
+proc paths*(index: L1Index): seq[string] =
+  ## Return all indexed paths (sorted)
+  result = toSeq(index.entries.keys)
+  result.sort()
+
+## Export
+export L1Index, IndexEntry
+export initL1Index, insert, lookup, list, glob, remove, paths, matchGlob