cache command updates

src/commands/cache.zig

@@ -1,37 +1,246 @@
 const std = @import("std");
 const zfin = @import("../root.zig");
 const cli = @import("common.zig");
+const srf = @import("srf");
+
+const Store = zfin.cache.Store;
+const DataType = zfin.cache.DataType;
+
+/// Data types to show in the stats table (skip candles_meta and meta — internal bookkeeping).
+const display_types = [_]DataType{
+    .candles_daily,
+    .dividends,
+    .splits,
+    .options,
+    .earnings,
+    .etf_profile,
+};
+
+const display_labels = [_][]const u8{
+    "candles",
+    "dividends",
+    "splits",
+    "options",
+    "earnings",
+    "etf_profile",
+};
 
 pub fn run(allocator: std.mem.Allocator, config: zfin.Config, subcommand: []const u8, out: *std.Io.Writer) !void {
     if (std.mem.eql(u8, subcommand, "stats")) {
-        try out.print("Cache directory: {s}\n", .{config.cache_dir});
+        try out.print("Cache directory: {s}\n\n", .{config.cache_dir});
-        std.fs.cwd().access(config.cache_dir, .{}) catch {
+
-            try out.print("  (empty -- no cached data)\n", .{});
-            return;
-        };
         var dir = std.fs.cwd().openDir(config.cache_dir, .{ .iterate = true }) catch {
             try out.print("  (empty -- no cached data)\n", .{});
             return;
         };
         defer dir.close();
-        var count: usize = 0;
+
+        // Collect and sort symbol names
+        var symbols: std.ArrayList([]const u8) = .empty;
+        defer {
+            for (symbols.items) |s| allocator.free(s);
+            symbols.deinit(allocator);
+        }
+
         var iter = dir.iterate();
         while (iter.next() catch null) |entry| {
             if (entry.kind == .directory) {
-                try out.print("  {s}/\n", .{entry.name});
+                const name = allocator.dupe(u8, entry.name) catch continue;
-                count += 1;
+                symbols.append(allocator, name) catch {
+                    allocator.free(name);
+                    continue;
+                };
             }
         }
-        if (count == 0) {
+
+        if (symbols.items.len == 0) {
             try out.print("  (empty -- no cached data)\n", .{});
-        } else {
+            return;
-            try out.print("\n  {d} symbol(s) cached\n", .{count});
         }
+
+        std.mem.sort([]const u8, symbols.items, {}, struct {
+            fn cmp(_: void, a: []const u8, b: []const u8) bool {
+                return std.mem.order(u8, a, b) == .lt;
+            }
+        }.cmp);
+
+        // Track totals
+        var total_size: u64 = 0;
+        var total_files: usize = 0;
+
+        for (symbols.items) |symbol| {
+            try out.print("{s}\n", .{symbol});
+
+            // Print header
+            try out.print("  {s:<14} {s:>10}  {s}\n", .{ "Type", "Size", "Updated" });
+            try out.print("  {s:->14} {s:->10}  {s:->24}\n", .{ "", "", "" });
+
+            var symbol_size: u64 = 0;
+            var symbol_files: usize = 0;
+
+            for (display_types, display_labels) |dt, label| {
+                const info = getFileInfo(allocator, config.cache_dir, symbol, dt);
+                if (info.exists) {
+                    symbol_files += 1;
+                    symbol_size += info.size;
+
+                    var size_buf: [10]u8 = undefined;
+                    const size_str = formatSize(&size_buf, info.size);
+
+                    if (info.is_negative) {
+                        try out.print("  {s:<14} {s:>10}  (negative cache)\n", .{ label, size_str });
+                    } else if (info.created) |ts| {
+                        var age_buf: [24]u8 = undefined;
+                        const age_str = formatAge(&age_buf, ts);
+                        const thru = info.lastDate() orelse "";
+                        if (info.expired) {
+                            if (thru.len > 0) {
+                                try out.print("  {s:<14} {s:>10}  {s} (stale, thru {s})\n", .{ label, size_str, age_str, thru });
+                            } else {
+                                try out.print("  {s:<14} {s:>10}  {s} (stale)\n", .{ label, size_str, age_str });
+                            }
+                        } else {
+                            if (thru.len > 0) {
+                                try out.print("  {s:<14} {s:>10}  {s} (thru {s})\n", .{ label, size_str, age_str, thru });
+                            } else {
+                                try out.print("  {s:<14} {s:>10}  {s}\n", .{ label, size_str, age_str });
+                            }
+                        }
+                    } else {
+                        try out.print("  {s:<14} {s:>10}  (no timestamp)\n", .{ label, size_str });
+                    }
+                }
+            }
+
+            // Also count candles_meta size (not displayed as its own row but is part of total)
+            const meta_info = getFileInfo(allocator, config.cache_dir, symbol, .candles_meta);
+            if (meta_info.exists) {
+                symbol_size += meta_info.size;
+                symbol_files += 1;
+            }
+
+            if (symbol_files == 0) {
+                try out.print("  (empty)\n", .{});
+            }
+            try out.print("\n", .{});
+            total_size += symbol_size;
+            total_files += symbol_files;
+        }
+
+        // Also count the top-level cusip_tickers.srf if present
+        const cusip_path = std.fs.path.join(allocator, &.{ config.cache_dir, "cusip_tickers.srf" }) catch null;
+        if (cusip_path) |path| {
+            defer allocator.free(path);
+            if (std.fs.cwd().statFile(path)) |stat| {
+                total_size += stat.size;
+                total_files += 1;
+            } else |_| {}
+        }
+
+        var total_buf: [10]u8 = undefined;
+        try out.print("{d} symbol(s), {d} file(s), {s} total\n", .{
+            symbols.items.len,
+            total_files,
+            formatSize(&total_buf, total_size),
+        });
     } else if (std.mem.eql(u8, subcommand, "clear")) {
-        var store = zfin.cache.Store.init(allocator, config.cache_dir);
+        var store = Store.init(allocator, config.cache_dir);
         try store.clearAll();
         try out.writeAll("Cache cleared.\n");
     } else {
         try cli.stderrPrint("Unknown cache subcommand. Use 'stats' or 'clear'.\n");
     }
 }
+
+const FileInfo = struct {
+    exists: bool = false,
+    size: u64 = 0,
+    is_negative: bool = false,
+    created: ?i64 = null,
+    expired: bool = false,
+    last_date_buf: [10]u8 = undefined,
+    last_date_len: u4 = 0,
+
+    fn lastDate(self: *const FileInfo) ?[]const u8 {
+        if (self.last_date_len == 0) return null;
+        return self.last_date_buf[0..self.last_date_len];
+    }
+};
+
+fn getFileInfo(allocator: std.mem.Allocator, cache_dir: []const u8, symbol: []const u8, dt: DataType) FileInfo {
+    var store = Store.init(allocator, cache_dir);
+
+    // Get file size via stat
+    const path = std.fs.path.join(allocator, &.{ cache_dir, symbol, dt.fileName() }) catch return .{};
+    defer allocator.free(path);
+    const stat = std.fs.cwd().statFile(path) catch return .{};
+
+    // Check for negative cache
+    if (store.isNegative(symbol, dt)) {
+        return .{ .exists = true, .size = stat.size, .is_negative = true };
+    }
+
+    // For candles_daily, use candles_meta for freshness and last_date
+    if (dt == .candles_daily) {
+        const meta_result = store.readCandleMeta(symbol) orelse
+            return .{ .exists = true, .size = stat.size };
+
+        var last_date_buf: [10]u8 = undefined;
+        const date_str = meta_result.meta.last_date.format(&last_date_buf);
+
+        return .{
+            .exists = true,
+            .size = stat.size,
+            .created = meta_result.created,
+            .expired = !store.isCandleMetaFresh(symbol),
+            .last_date_buf = last_date_buf,
+            .last_date_len = @intCast(date_str.len),
+        };
+    }
+
+    // For all other types, read the file and use the srf iterator for directives
+    const data = std.fs.cwd().readFileAlloc(allocator, path, 50 * 1024 * 1024) catch
+        return .{ .exists = true, .size = stat.size };
+    defer allocator.free(data);
+
+    var reader = std.Io.Reader.fixed(data);
+    const it = srf.iterator(&reader, allocator, .{ .alloc_strings = false }) catch
+        return .{ .exists = true, .size = stat.size };
+    defer it.deinit();
+
+    return .{
+        .exists = true,
+        .size = stat.size,
+        .created = it.created,
+        .expired = if (it.expires != null) !it.isFresh() else false,
+    };
+}
+
+fn formatSize(buf: *[10]u8, size: u64) []const u8 {
+    if (size < 1024) {
+        return std.fmt.bufPrint(buf, "{d} B", .{size}) catch "?";
+    } else if (size < 1024 * 1024) {
+        const kb: f64 = @as(f64, @floatFromInt(size)) / 1024.0;
+        return std.fmt.bufPrint(buf, "{d:.1} KB", .{kb}) catch "?";
+    } else {
+        const mb: f64 = @as(f64, @floatFromInt(size)) / (1024.0 * 1024.0);
+        return std.fmt.bufPrint(buf, "{d:.1} MB", .{mb}) catch "?";
+    }
+}
+
+fn formatAge(buf: *[24]u8, timestamp: i64) []const u8 {
+    const now = std.time.timestamp();
+    const age = now - timestamp;
+
+    if (age < 0) {
+        return std.fmt.bufPrint(buf, "just now", .{}) catch "?";
+    } else if (age < 60) {
+        return std.fmt.bufPrint(buf, "{d}s ago", .{@as(u64, @intCast(age))}) catch "?";
+    } else if (age < 3600) {
+        return std.fmt.bufPrint(buf, "{d}m ago", .{@as(u64, @intCast(@divTrunc(age, 60)))}) catch "?";
+    } else if (age < 86400) {
+        return std.fmt.bufPrint(buf, "{d}h ago", .{@as(u64, @intCast(@divTrunc(age, 3600)))}) catch "?";
+    } else {
+        return std.fmt.bufPrint(buf, "{d}d ago", .{@as(u64, @intCast(@divTrunc(age, 86400)))}) catch "?";
+    }
+}

src/main.zig

@@ -103,8 +103,14 @@ pub fn main() !u8 {
     var svc = zfin.DataService.init(allocator, config);
     defer svc.deinit();
 
-    // Normalize symbol to uppercase (e.g. "aapl" -> "AAPL")
+    // Normalize symbol to uppercase (e.g. "aapl" -> "AAPL") for commands that take a symbol.
-    if (args.len >= 3) {
+    // Skip normalization for commands where args[2] is a subcommand or file path.
+    if (args.len >= 3 and
+        !std.mem.eql(u8, command, "cache") and
+        !std.mem.eql(u8, command, "enrich") and
+        !std.mem.eql(u8, command, "analysis") and
+        !std.mem.eql(u8, command, "portfolio"))
+    {
         for (args[2]) |*c| c.* = std.ascii.toUpper(c.*);
     }