From 0783060b91e23940bb70ef1602ef6b775737909c Mon Sep 17 00:00:00 2001
From: sabine <6594573+sabine@users.noreply.github.com>
Date: Thu, 6 Nov 2025 16:44:31 +0100
Subject: [PATCH 1/2] increase yaml library buffer to fix video scraper

---
 tool/ood-gen/lib/youtube.ml | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

diff --git a/tool/ood-gen/lib/youtube.ml b/tool/ood-gen/lib/youtube.ml
index 19bf3a5899..48664a88df 100644
--- a/tool/ood-gen/lib/youtube.ml
+++ b/tool/ood-gen/lib/youtube.ml
@@ -187,17 +187,28 @@ let scrape yaml_file =
   in
   match fetched with
   | Ok fetched ->
-      let yaml =
+      let all_videos =
         VideoSet.union fetched scraped
         |> VideoSet.to_seq |> List.of_seq
         |> List.sort (fun a b -> compare b.Vid.published a.Vid.published)
-        |> Vid.video_list_to_yaml
       in
-      let output =
-        Yaml.pp Format.str_formatter yaml;
-        Format.flush_str_formatter ()
+      let yaml = Vid.video_list_to_yaml all_videos in
+      (* The yaml library uses a fixed-size output buffer. The default is 262140 bytes,
+         which was exceeded when we had 203 videos (~262KB output). This caused the
+         document_end operation to fail with "doc_end failed" error.
+         
+         Current stats: 203 videos ≈ 260KB, average ~1.3KB per video.
+         We use a 2MB buffer to accommodate growth to ~1500 videos before hitting limits.
+         If the list grows beyond that, this will fail with a clear error message. *)
+      let buffer_size = 2 * 1024 * 1024 in (* 2MB *)
+      let output = match Yaml.to_string ~len:buffer_size yaml with
+        | Ok s -> s
+        | Error (`Msg err) -> 
+            failwith (Printf.sprintf 
+              "YAML serialization failed (tried %d videos, buffer size %d bytes): %s" 
+              (List.length all_videos) buffer_size err)
       in
       let oc = open_out yaml_file in
-      Printf.fprintf oc "%s" output;
+      output_string oc output;
       close_out oc
   | Error (`Msg msg) -> failwith msg

From 662a13aa53fc519327da56a205d1a474c6c3cb38 Mon Sep 17 00:00:00 2001
From: sabine <6594573+sabine@users.noreply.github.com>
Date: Thu, 6 Nov 2025 18:07:38 +0100
Subject: [PATCH 2/2] fmt

---
 tool/ood-gen/lib/youtube.ml | 31 ++++++++++++++++++-------------
 1 file changed, 18 insertions(+), 13 deletions(-)

diff --git a/tool/ood-gen/lib/youtube.ml b/tool/ood-gen/lib/youtube.ml
index 48664a88df..0506e509b0 100644
--- a/tool/ood-gen/lib/youtube.ml
+++ b/tool/ood-gen/lib/youtube.ml
@@ -193,20 +193,25 @@ let scrape yaml_file =
         |> List.sort (fun a b -> compare b.Vid.published a.Vid.published)
       in
       let yaml = Vid.video_list_to_yaml all_videos in
-      (* The yaml library uses a fixed-size output buffer. The default is 262140 bytes,
-         which was exceeded when we had 203 videos (~262KB output). This caused the
-         document_end operation to fail with "doc_end failed" error.
-         
-         Current stats: 203 videos ≈ 260KB, average ~1.3KB per video.
-         We use a 2MB buffer to accommodate growth to ~1500 videos before hitting limits.
-         If the list grows beyond that, this will fail with a clear error message. *)
-      let buffer_size = 2 * 1024 * 1024 in (* 2MB *)
-      let output = match Yaml.to_string ~len:buffer_size yaml with
+      (* The yaml library uses a fixed-size output buffer. The default is 262140
+         bytes, which was exceeded when we had 203 videos (~262KB output). This
+         caused the document_end operation to fail with "doc_end failed" error.
+
+         Current stats: 203 videos ≈ 260KB, average ~1.3KB per video. We use a
+         2MB buffer to accommodate growth to ~1500 videos before hitting limits.
+         If the list grows beyond that, this will fail with a clear error
+         message. *)
+      let buffer_size = 2 * 1024 * 1024 in
+      (* 2MB *)
+      let output =
+        match Yaml.to_string ~len:buffer_size yaml with
         | Ok s -> s
-        | Error (`Msg err) -> 
-            failwith (Printf.sprintf 
-              "YAML serialization failed (tried %d videos, buffer size %d bytes): %s" 
-              (List.length all_videos) buffer_size err)
+        | Error (`Msg err) ->
+            failwith
+              (Printf.sprintf
+                 "YAML serialization failed (tried %d videos, buffer size %d \
+                  bytes): %s"
+                 (List.length all_videos) buffer_size err)
       in
       let oc = open_out yaml_file in
       output_string oc output;