--vision: warn at startup when the chat template has no image path, and say so per request

zig/src/server/images.zig

@@ -240,6 +240,8 @@ pub fn expand(vision: ?api.Vision, fetch: ?Fetch, text: model_text.Text, cx: *Cx
         marks += @intFromBool(t == v.image_token);
         video_marks += @intFromBool(video_token != null and t == video_token.?);
     }
+    // none at all: the template has no image path (it swaps media parts for text), a configuration problem
+    if (marks == 0 and n_images > 0) return cx.fail(.server, "the chat template wrote no image placeholder for {d} image part{s}; a template that turns media parts into text has no image path: serve with --chat-template naming one that writes the image token", .{ n_images, if (n_images == 1) "" else "s" });
     if (marks != n_images) return cx.fail(.server, "the chat template wrote {d} image placeholders for {d} image parts", .{ marks, n_images });
     if (video_marks != n_videos) return cx.fail(.server, "the chat template wrote {d} video placeholders for {d} video parts", .{ video_marks, n_videos });
     if (ps.len == 0) return null;

zig/src/server/server.zig

@@ -90,6 +90,8 @@ pub const Server = struct {
     keys: ?*auth.Store,
     late_system: []const u8 = "system",
     needs_user_after_tool: bool = false,
+    /// With image input: whether the chat template writes an image placeholder for an image part (templateImages).
+    template_images: bool = true,
     markers: reply_text.Markers = reply_text.think_markers,
     chunks: chunk_plan.Plan = .{},
     effort_levels: []const []const u8 = &.{},
@@ -148,6 +150,10 @@ pub const Server = struct {
             std.mem.indexOf(u8, source, "thinking_mode") != null;
         srv.late_system = try lateSystem(a, text);
         srv.needs_user_after_tool = needsUserAfterTool(source);
+        if (srv.info.vision != null) {
+            srv.template_images = templateImages(srv);
+            if (!srv.template_images) log.line("warning: --vision: the chat template writes no image placeholder for an image part, so image requests will fail; serve with --chat-template naming a template that has an image path", .{});
+        }
         if (text.tokenId("</think>")) |end| {
             const lead = try text.encode(a, "\n", false);
             const trail = try text.encode(a, "\n\n", false);
@@ -389,6 +395,19 @@ fn lateSystem(a: Allocator, text: model_text.Text) ![]const u8 {
     return if (std.mem.indexOf(u8, rendered, probe_text) != null) "system" else "user";
 }
 
+/// With image input: whether a user turn holding an image part renders an image placeholder (else a startup warning).
+fn templateImages(srv: *Server) bool {
+    const v = srv.info.vision orelse return true;
+    var arena: std.heap.ArenaAllocator = .init(srv.gpa);
+    defer arena.deinit();
+    const a = arena.allocator();
+    var cx: errors.Cx = .{ .a = a };
+    const probe = (json.parse(a, "[{\"role\": \"user\", \"content\": [{\"type\": \"image_url\", \"image_url\": {\"url\": \"data:image/png;base64,AA==\"}}, {\"type\": \"text\", \"text\": \"x\"}]}]") catch return true).ok;
+    const ids = @import("prompt.zig").renderIds(srv, &cx, probe, &.{}, false, null, true) catch return true; // a render failure is the request's to report
+    for (ids) |t| if (t == v.image_token) return true;
+    return false;
+}
+
 /// Templates requiring a user query gain a user turn after trailing tool results replace their last user turn.
 fn needsUserAfterTool(source: []const u8) bool {
     return std.mem.indexOf(u8, source, "No user query found") != null;

zig/src/server/videos.zig

@@ -237,3 +237,24 @@ test "a request's videos are limited by the engine's offer, not by a constant" {
     try std.testing.expectError(error.Refused, images.expand(v, null, images.MarkText.text(), &cx, two, &[_]u32{ 8, 8 }, 0));
     try std.testing.expectEqualStrings("a request supports at most 1 videos", cx.message);
 }
+
+test "a template that writes no image placeholder is named as the cause, a miscount keeps its counts" {
+    var arena = std.heap.ArenaAllocator.init(std.testing.allocator);
+    defer arena.deinit();
+    const a = arena.allocator();
+    var dummy: u8 = 0;
+    const v: api.Vision = .{ .ctx = &dummy, .prepare = noPrepare, .image_token = 9, .image_tokens = 64, .max_images = 4 };
+    const two = switch (try json.parseText(a,
+        \\[{"role": "user", "content": [{"type": "image_url", "image_url": {"url": "data:image/png;base64,YQ=="}},
+        \\  {"type": "image_url", "image_url": {"url": "data:image/png;base64,YQ=="}}, {"type": "text", "text": "x"}]}]
+    )) {
+        .ok => |m| m,
+        .err => return error.TestUnexpectedResult,
+    };
+    var cx: Cx = .{ .a = a };
+    try std.testing.expectError(error.Refused, images.expand(v, null, images.MarkText.text(), &cx, two, &[_]u32{ 1, 2 }, 0));
+    try std.testing.expectEqualStrings("the chat template wrote no image placeholder for 2 image parts; a template that turns media parts into text has no image path: serve with --chat-template naming one that writes the image token", cx.message);
+    cx = .{ .a = a };
+    try std.testing.expectError(error.Refused, images.expand(v, null, images.MarkText.text(), &cx, two, &[_]u32{ 1, 9 }, 0));
+    try std.testing.expectEqualStrings("the chat template wrote 1 image placeholders for 2 image parts", cx.message);
+}