fix: metadata fix for content_hash diffs

This commit is contained in:
2026-04-27 11:08:21 +02:00
parent 56c5ec1861
commit 53dd9deeab
6 changed files with 137 additions and 4 deletions

View File

@@ -119,7 +119,7 @@ defmodule BDS.Desktop.ShellCommands do
defp dispatch("rebuild_posts_from_files", project, _params) do
queue_task(project, "rebuild_posts_from_files", "Rebuild Posts From Files", "Maintenance", fn report ->
{:ok, posts} = Maintenance.rebuild_from_filesystem(project.id, "post", on_progress: report, rebuild_embeddings: false)
{:ok, posts} = Maintenance.rebuild_from_filesystem(project.id, "post", on_progress: report)
report.(1.0, "Post rebuild complete")
%{project_id: project.id, counts: %{posts: length(posts)}}
end)

View File

@@ -17,7 +17,11 @@ defmodule BDS.Desktop.ShellLive.MiscEditor do
def rerun(socket) do
case meta(socket.assigns) do
%{action: action} when is_binary(action) -> {:command, action}
_other -> {:noop, socket}
_other ->
case misc_route_action(socket.assigns.current_tab.type) do
nil -> {:noop, socket}
action -> {:command, action}
end
end
end
@@ -464,7 +468,13 @@ defmodule BDS.Desktop.ShellLive.MiscEditor do
end
end
defp metadata_diff_repairable_tab?(tab_id), do: tab_id in ["posts", "media", "scripts", "templates", "project"]
defp metadata_diff_repairable_tab?(tab_id), do: tab_id in ["posts", "media", "scripts", "templates", "project", "embeddings"]
defp misc_route_action(:site_validation), do: "validate_site"
defp misc_route_action(:metadata_diff), do: "metadata_diff"
defp misc_route_action(:translation_validation), do: "validate_translations"
defp misc_route_action(:find_duplicates), do: "find_duplicates"
defp misc_route_action(_route), do: nil
defp format_metadata_diff_value(nil), do: "-"
defp format_metadata_diff_value(""), do: "-"

View File

@@ -20,6 +20,14 @@ defmodule BDS.Embeddings do
def index_path(project_id), do: Index.path(project_id)
def reindex_all(project_id), do: rebuild_project(project_id)
def refresh_snapshot(project_id) when is_binary(project_id) do
if enabled_for_project?(project_id) do
:ok = rebuild_snapshot(project_id)
end
:ok
end
def get_indexing_progress(project_id) when is_binary(project_id) do
indexed =
Repo.one(
@@ -105,7 +113,15 @@ defmodule BDS.Embeddings do
if differences == [] do
[]
else
[%{entity_type: "embedding", entity_id: post.id, differences: differences}]
[
%{
entity_type: "embedding",
entity_id: post.id,
label: post.title || post.slug || post.id,
meta_label: Persistence.timestamp_to_iso8601(post.created_at),
differences: differences
}
]
end
end)
else

View File

@@ -664,6 +664,8 @@ defmodule BDS.Maintenance do
{:db_to_file, "script"} -> BDS.Scripts.sync_published_script_file(entity_id)
{:file_to_db, "template"} -> BDS.Templates.sync_template_from_file(entity_id)
{:db_to_file, "template"} -> BDS.Templates.sync_published_template_file(entity_id)
{:file_to_db, "embedding"} -> BDS.Embeddings.sync_post(entity_id)
{:db_to_file, "embedding"} -> BDS.Embeddings.refresh_snapshot(project_id)
_other -> {:error, :unsupported}
end
end

View File

@@ -99,6 +99,30 @@ defmodule BDS.Desktop.ShellCommandsTest do
assert is_map(completed.result.payload.summary)
end
test "rebuild_posts_from_files rebuilds embeddings for published posts when semantic similarity is enabled", %{project: project} do
assert {:ok, _metadata} =
BDS.Metadata.update_project_metadata(project.id, %{semantic_similarity_enabled: true})
assert {:ok, post} =
BDS.Posts.create_post(%{
project_id: project.id,
title: "Filesystem Embedding Source",
content: "space rocket orbit mission galaxy",
language: "en"
})
assert {:ok, published_post} = BDS.Posts.publish_post(post.id)
assert BDS.Repo.get_by(BDS.Embeddings.Key, project_id: project.id, post_id: published_post.id) != nil
BDS.Repo.delete_all(BDS.Embeddings.Key)
assert {:ok, result} = ShellCommands.execute("rebuild_posts_from_files")
completed = wait_for_task(result.task_id, &(&1.status == :completed))
assert completed.group_name == "Maintenance"
assert BDS.Repo.get_by(BDS.Embeddings.Key, project_id: project.id, post_id: published_post.id) != nil
end
test "repair_metadata_diff exposes live in-task progress from the repair worker", %{project: project} do
original = Application.get_env(:bds, :tasks, [])

View File

@@ -370,6 +370,34 @@ defmodule BDS.Desktop.ShellLiveTest do
assert html =~ ~s(class="tab active transient")
end
test "metadata diff refresh reruns after workbench session restore", %{project: project} do
:ok = BDS.Tasks.clear_finished()
{:ok, view, _html} = live_isolated(build_conn(), BDS.Desktop.ShellLive)
session_payload =
Workbench.new()
|> Workbench.open_tab(:metadata_diff, "metadata_diff", :pin)
|> Session.serialize()
html = render_hook(view, "restore_workbench_session", %{"session" => session_payload})
assert html =~ ~s(data-tab-type="metadata_diff")
existing_ids = MapSet.new(Enum.map(BDS.Tasks.list_tasks(), & &1.id))
_html =
view
|> element("button[phx-click='rerun_misc_editor']")
|> render_click()
refresh_task = new_task!(existing_ids, "Metadata Diff")
assert refresh_task.group_name == "Maintenance"
completed_task!(refresh_task.id)
send(view.pid, :refresh_task_status)
assert render(view) =~ project.name
end
test "shell live renders the legacy git activity badge from remote behind count" do
Application.put_env(:bds, :git_remote_state_provider, fn _project_id, _opts ->
{:ok, %{local_branch: "main", upstream_branch: "origin/main", has_upstream: true, ahead: 0, behind: 7}}
@@ -1163,6 +1191,59 @@ defmodule BDS.Desktop.ShellLiveTest do
refute orphan_relative_path in Enum.map(diff.orphan_reports, & &1.file_path)
end
test "metadata diff embeddings tab exposes repair actions and clears embedding drift", %{project: project} do
:ok = BDS.Tasks.clear_finished()
assert {:ok, _metadata} =
Metadata.update_project_metadata(project.id, %{semantic_similarity_enabled: true})
assert {:ok, post} =
Posts.create_post(%{
project_id: project.id,
title: "Embedding Drift",
content: "space rocket orbit mission galaxy",
language: "en"
})
assert {:ok, published_post} = Posts.publish_post(post.id)
assert {:ok, _indexed} = BDS.Embeddings.index_unindexed(project.id)
Repo.delete_all(BDS.Embeddings.Key)
{:ok, view, _html} = live_isolated(build_conn(), BDS.Desktop.ShellLive)
assert {:ok, queued} = BDS.Desktop.ShellCommands.execute("metadata_diff")
completed_task!(queued.task_id)
send(view.pid, :refresh_task_status)
html =
view
|> element("[data-testid='metadata-diff-tab'][data-entity-tab='embeddings']")
|> render_click()
assert html =~ "content_hash"
assert html =~ ~s(data-testid="metadata-diff-repair-button")
existing_ids = MapSet.new(Enum.map(BDS.Tasks.list_tasks(), & &1.id))
html =
view
|> element("[data-testid='metadata-diff-repair-button'][data-direction='file_to_db'][data-field='content_hash']")
|> render_click()
assert html =~ "Repair Metadata Diff"
repair_task = new_task!(existing_ids, "Repair Metadata Diff")
completed_task!(repair_task.id)
send(view.pid, :refresh_task_status)
_html = render(view)
assert Repo.get_by(BDS.Embeddings.Key, project_id: project.id, post_id: published_post.id) != nil
assert {:ok, diff} = BDS.Maintenance.metadata_diff(project.id)
refute Enum.any?(diff.diff_reports, &(&1.entity_type == "embedding" and &1.entity_id == published_post.id))
end
test "post tabs render a real editor and drive save publish discard flows", %{project: project} do
assert {:ok, _tag} = Tags.create_tag(%{project_id: project.id, name: "alpha", color: "#112233"})
assert {:ok, _tag} = Tags.create_tag(%{project_id: project.id, name: "beta", color: "#445566"})