feat(web/todolist): Implement a "todo-list" page generator r/1404

This invokes ripgrep & jq to construct a list of TODOs from known
users across depot sources, and dumps it into a static page that we
can serve.

The structure is relatively simple, but it might be useful. See here
for an example of what this looks like:

https: //tazj.in/blobs/todos.png
Change-Id: I1edef56606273584ab886b9e762c8ed4d210919d
Reviewed-on: https://cl.tvl.fyi/c/depot/+/1296
Tested-by: BuildkiteCI
Reviewed-by: Alyssa Ross <hi@alyssa.is>

1 files changed, 25 insertions, 0 deletions

diff --git a/web/todolist/extract-todos.jq b/web/todolist/extract-todos.jq
new file mode 100644
index 0000000000..cda717fceb
--- /dev/null
+++ b/web/todolist/extract-todos.jq
@@ -0,0 +1,25 @@
+# Simple jq script to extract all TODO comments in the code base from
+# ripgrep's JSON output.
+#
+# This assumes that the filter used is something like 'TODO\(\w+\):'
+
+# Construct a structure with only the fields we need to populate the
+# page.
+def simplify_match:
+  .data.submatches[0].match.text as $todo
+  | {
+     file: (.data.path.text | sub("/nix/store/.+-depot/"; "")),
+     line: .data.line_number,
+     todo: $todo,
+     user: ($todo | capture("TODO\\((?<user>\\w+)\\)") | .user),
+     };
+
+# Group all matches first by the user and return them in sorted order
+# by the file in which they appear. This matches the presentation
+# order on the website.
+def group_by_user: .
+    | group_by(.user)
+    | map(sort_by(.file));
+
+# main:
+map(select(.type == "match") | simplify_match) | group_by_user