From dbe487e0f4dc7ee8d635f42f41aca026d3f66e50 Mon Sep 17 00:00:00 2001
From: Emil Lerch <emil@lerch.org>
Date: Mon, 1 Jun 2026 08:29:56 -0700
Subject: [PATCH] use arena for refresh loop

---
 src/main.zig | 51 ++++++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 40 insertions(+), 11 deletions(-)

diff --git a/src/main.zig b/src/main.zig
index 51bc063..c6c4e26 100644
--- a/src/main.zig
+++ b/src/main.zig
@@ -508,27 +508,31 @@ fn writeNewPortfolio(io: std.Io, allocator: std.mem.Allocator, path: []const u8,
 
 // ── Refresh command ──────────────────────────────────────────
 
-fn refresh(io: std.Io, allocator: std.mem.Allocator, environ: *const std.process.Environ.Map) !void {
-    var config = zfin.Config.fromEnv(io, allocator, environ);
+fn refresh(
+    io: std.Io,
+    gpa: std.mem.Allocator,
+    environ: *const std.process.Environ.Map,
+) !void {
+    var config = zfin.Config.fromEnv(io, gpa, environ);
     defer config.deinit();
-    var svc = zfin.DataService.init(io, allocator, config);
+    var svc = zfin.DataService.init(io, gpa, config);
     defer svc.deinit();
 
     const portfolio_path = environ.get("ZFIN_PORTFOLIO") orelse "portfolio.srf";
 
-    const data = std.Io.Dir.cwd().readFileAlloc(io, portfolio_path, allocator, .limited(10 * 1024 * 1024)) catch {
+    const data = std.Io.Dir.cwd().readFileAlloc(io, portfolio_path, gpa, .limited(10 * 1024 * 1024)) catch {
         log.err("failed to read portfolio: {s}", .{portfolio_path});
         return error.ReadFailed;
     };
-    defer allocator.free(data);
+    defer gpa.free(data);
 
-    var portfolio = zfin.cache.deserializePortfolio(allocator, data) catch {
+    var portfolio = zfin.cache.deserializePortfolio(gpa, data) catch {
         log.err("failed to parse portfolio", .{});
         return error.ParseFailed;
     };
     defer portfolio.deinit();
 
-    var symbols = std.StringHashMap(void).init(allocator);
+    var symbols = std.StringHashMap(void).init(gpa);
     defer symbols.deinit();
     for (portfolio.lots) |lot| {
         if (lot.security_type != .stock and lot.security_type != .watch) continue;
@@ -577,6 +581,25 @@ fn refresh(io: std.Io, allocator: std.mem.Allocator, environ: *const std.process
         try stdout.flush();
     }
 
+    // Per-iteration scratch arena. Wraps the gpa so iteration-scoped
+    // allocations get freed in bulk via `reset(.retain_capacity)` at
+    // the bottom of each loop iteration — no per-allocation defers,
+    // no manual `free` mistakes. Capacity is retained so we're not
+    // re-acquiring pages every symbol.
+    //
+    // What goes through this arena: any per-iteration scratch we
+    // allocate ourselves (CIK dupes, future symbol-derived strings).
+    // NOT the `FetchResult` payloads — those are owned by the
+    // service's gpa and freed via `result.deinit()` at their own
+    // call sites.
+    //
+    // We don't use juicy-main's `init.arena` because that's
+    // process-lifetime; resetting it mid-run would clobber other
+    // startup allocations.
+    var iter_arena_state = std.heap.ArenaAllocator.init(gpa);
+    defer iter_arena_state.deinit();
+    const iter_arena = iter_arena_state.allocator();
+
     var it = symbols.iterator();
     while (it.next()) |entry| {
         const sym = entry.key_ptr.*;
@@ -637,14 +660,16 @@ fn refresh(io: std.Io, allocator: std.mem.Allocator, environ: *const std.process
         // had it — used to chain into entity_facts below.
         // NotFound is logged as `n/a` (symbol genuinely has no
         // Wikidata entry) and doesn't flip sym_ok.
-        var cik_buf: ?[]u8 = null;
-        defer if (cik_buf) |b| allocator.free(b);
+        //
+        // The CIK dupe goes through `iter_arena`. No defer-free
+        // needed — the arena is reset at end of iteration.
+        var cik_str: ?[]const u8 = null;
         try printRateLimitWait(&svc, stdout);
         if (svc.getClassification(sym, .{})) |result| {
             defer result.deinit();
             if (result.data.len > 0) {
                 if (result.data[0].cik) |cik| {
-                    cik_buf = allocator.dupe(u8, cik) catch null;
+                    cik_str = iter_arena.dupe(u8, cik) catch null;
                 }
             }
             try stdout.print(", classification ok", .{});
@@ -678,7 +703,7 @@ fn refresh(io: std.Io, allocator: std.mem.Allocator, environ: *const std.process
         // have an EDGAR CIK from the ticker map (production
         // zfin chains entity_facts off Wikidata's CIK, so the
         // server warms the cache the same way).
-        if (cik_buf) |cik| {
+        if (cik_str) |cik| {
             try printRateLimitWait(&svc, stdout);
             if (svc.getEntityFacts(cik, .{})) |result| {
                 result.deinit();
@@ -696,6 +721,10 @@ fn refresh(io: std.Io, allocator: std.mem.Allocator, environ: *const std.process
         try stdout.flush();
 
         if (sym_ok) success_count += 1 else fail_count += 1;
+
+        // Reset per-iteration scratch. Retains capacity so the
+        // next iteration's allocations reuse the same pages.
+        _ = iter_arena_state.reset(.retain_capacity);
     }
 
     try stdout.print("\nRefresh complete: {d} ok, {d} failed\n", .{ success_count, fail_count });