Compare commits

..
Author SHA1 Message Date
awesomerobot 522c1aabb8 wip experiment with /filter route 2024-12-20 15:55:49 -05:00
Discourse Translator Bot a4033e2af9 Update translations (#1032) 2024-12-18 15:19:47 +01:00
Martin Brennan f35db8068b DEV: Change to use DPageSubheader (#1033)
Previously was AdminPageSubheader until https://github.com/discourse/discourse/pull/30146
2024-12-18 17:39:31 +10:00
Sam fae2d5ff2c FEATURE: link correctly to filters to assist in debugging spam (#1031)
- Add spam_score_type to AiSpamSerializer for better integration with reviewables.
- Introduce a custom filter for detecting AI spam false negatives in moderation workflows.
- Refactor spam report generation to improve identification of false negatives.
- Add tests to verify the custom filter and its behavior.
- Introduce links for all spam counts in report
2024-12-17 11:02:18 +11:00
Keegan George 90ce942108 FEATURE: Add periodic problem checks for each LLM in use (#1020)
This feature adds a periodic problem check which periodically checks for issues with LLMs that are in use. Periodically, we will run a test to see if the in use LLMs are still operational. If it is not, the LLM with the problem is surfaced to the admin so they can easily go and update the configuration.
2024-12-16 15:00:05 -08:00
Mark VanLandingham 24b107881a FEATURE: Unavailable state for semantic search when sort is not Relevant (#1030)
This commit adds an "unavailable" state for the AI semantic search toggle. Currently the AI toggle disappears when the sort by is anything but Relevance which makes the UI confusing for users looking for AI results. This should help!
2024-12-16 14:30:11 -06:00
Roman Rizzi 534b0df391 REFACTOR: Separation of concerns for embedding generation. (#1027)
In a previous refactor, we moved the responsibility of querying and storing embeddings into the `Schema` class. Now, it's time for embedding generation.

The motivation behind these changes is to isolate vector characteristics in simple objects to later replace them with a DB-backed version, similar to what we did with LLM configs.
2024-12-16 09:55:39 -03:00
Martin Brennan 222e2cf4f9 UX: Use new DStatTiles reusable component from core (#1025)
For the Spam and Usage tabs in admin
2024-12-16 16:48:46 +10:00
Roman RizziandKeegan George 94b85ece80 FIX: Make sure gists are atleast five minutes old before updating them (#1029)
* FIX: Make sure gists are atleast five minutes old before updating them

* Update app/jobs/regular/fast_track_topic_gist.rb

Co-authored-by: Keegan George <[email protected]>

---------

Co-authored-by: Keegan George <[email protected]>
2024-12-13 19:36:34 -03:00
Roman Rizzi 1c40a698ca FIX: get strategy version through vector_rep (#1028) 2024-12-13 18:49:18 -03:00
David Taylor fae1fbc796 DEV: Correct i18n lint violation (#1026) 2024-12-13 16:01:04 +00:00
Roman Rizzi eae527f99d REFACTOR: A Simpler way of interacting with embeddings tables. (#1023)
* REFACTOR: A Simpler way of interacting with embeddings' tables.

This change adds a new abstraction called `Schema`, which acts as a repository that supports the same DB features `VectorRepresentation::Base` has, with the exception that removes the need to have duplicated methods per embeddings table.

It is also a bit more flexible when performing a similarity search because you can pass it a block that gives you access to the builder, allowing you to add multiple joins/where conditions.
2024-12-13 10:15:21 -03:00
Roman Rizzi 97ec2c5ff4 FEATURE: Show gists everywhere except suggested/related (#995) 2024-12-12 12:29:35 -03:00
Krzysztof Kotlarek 04c4ff8cf0 UX: No admin header for edit personas tools or llms (#1021)
In this PR, we added functionality to hide the admin header for edit/new actions - https://github.com/discourse/discourse/pull/30175

To make it work properly, we have to rename `show` to `edit` which is also a more accurate name.
2024-12-12 10:48:58 +11:00
Sam 47c1ea337e FIX: allow scanning of trashed posts and deleted users for test (#1024)
When a post is trashed we should still be allowed to scan it
post.topic will be nil for a trashed topic even if post is trashed
2024-12-12 10:26:05 +11:00
47f5da7e42 FEATURE: Add AI-powered spam detection for new user posts (#1004)
This introduces a comprehensive spam detection system that uses LLM models
to automatically identify and flag potential spam posts. The system is
designed to be both powerful and configurable while preventing false positives.

Key Features:
* Automatically scans first 3 posts from new users (TL0/TL1)
* Creates dedicated AI flagging user to distinguish from system flags
* Tracks false positives/negatives for quality monitoring
* Supports custom instructions to fine-tune detection
* Includes test interface for trying detection on any post

Technical Implementation:
* New database tables:
  - ai_spam_logs: Stores scan history and results
  - ai_moderation_settings: Stores LLM config and custom instructions
* Rate limiting and safeguards:
  - Minimum 10-minute delay between rescans
  - Only scans significant edits (>10 char difference)
  - Maximum 3 scans per post
  - 24-hour maximum age for scannable posts
* Admin UI features:
  - Real-time testing capabilities
  - 7-day statistics dashboard
  - Configurable LLM model selection
  - Custom instruction support

Security and Performance:
* Respects trust levels - only scans TL0/TL1 users
* Skips private messages entirely
* Stops scanning users after 3 successful public posts
* Includes comprehensive test coverage
* Maintains audit log of all scan attempts


---------

Co-authored-by: Keegan George <[email protected]>
Co-authored-by: Martin Brennan <[email protected]>
2024-12-12 09:17:25 +11:00
Martin Brennan ae80494448 UX: Improve rough edges of AI usage page (#1014)
* UX: Improve rough edges of AI usage page

* Ensure all text uses I18n
* Change from <button> usage to <DButton>
* Use <AdminConfigAreaCard> in place of custom card styles
* Format numbers nicely using our number format helper,
  show full values on hover using title attr
* Ensure 0 is always shown for counters, instead of being blank

* FEATURE: Load usage data after page load

Use ConditionalLoadingSpinner to hide load of usage
data, this prevents us hanging on page load with a white
screen.

* UX: Split users table, and add empty placeholders and page subheader

* DEV: Test fix
2024-12-12 08:55:24 +11:00
Keegan George a4440c507b UX: Make sentiment trends more readable (#1018)
Instead of a stacked chart showing a separate series for positive and negative, this PR introduces a simplification to the overall sentiment dashboard. It comprises the sentiment into a single series of the difference between `positive - negative` instead. This should allow for the data to be more easy to scan and look for trends
2024-12-11 09:13:18 -08:00
Roman Rizzi 5fc7a730ef FIX: Triage rule should append selected tags instead of replacing them (#1022) 2024-12-11 11:19:44 -03:00
Discourse Translator Bot 34f43f398d Update translations (#1015) 2024-12-10 20:16:44 +01:00
Roman Rizzi 6da35d8e66 FIX: Gemini inference client was missing #instance (#1019) 2024-12-10 15:42:31 -03:00
Keegan George 700e9de073 Revert "UX: Make sentiment trends more readable in time series data (#1013)" (#1016)
This reverts commit 375dd702b2.
2024-12-10 08:15:27 -08:00
Keegan George 375dd702b2 UX: Make sentiment trends more readable in time series data (#1013)
Instead of a stacked chart showing a separate series for positive and negative, this PR introduces a simplification to the overall sentiment dashboard. It comprises the sentiment into a single series of the difference between `positive - negative` instead. This should allow for the data to be more easy to scan and look for trends.
2024-12-10 07:22:41 -08:00
Sam 7ca21cc329 FEATURE: first class support for OpenRouter (#1011)
* FEATURE: first class support for OpenRouter

This new implementation supports picking quantization and provider pref

Also:

- Improve logging for summary generation
- Improve error message when contacting LLMs fails

* Better support for full screen artifacts on iPad

Support back button to close full screen
2024-12-10 05:59:19 +11:00
Kris 51e0a96e51 UX: ensure composer popup is always above AI icons (#1012) 2024-12-09 13:03:27 -05:00
Discourse Translator Bot 6ce14a778b Update translations (#1009) 2024-12-06 18:05:02 +01:00
Roman Rizzi 085dde7042 FEATURE: Select stop sequences from triage script (#1010) 2024-12-06 11:13:47 -03:00
Roman Rizzi 7ebbcd2de3 FIX: Make sure prompt uploads get included in the prompt when triaging (#1008) 2024-12-05 21:04:35 -03:00
Sam a55216773a FEATURE: Amazon Nova support via bedrock (#997)
Refactor dialect selection and add Nova API support

    Change dialect selection to use llm_model object instead of just provider name
    Add support for Amazon Bedrock's Nova API with native tools
    Implement Nova-specific message processing and formatting
    Update specs for Nova and AWS Bedrock endpoints
    Enhance AWS Bedrock support to handle Nova models
    Fix Gemini beta API detection logic
2024-12-06 07:45:58 +11:00
Kelv 5e87a50202 DEV: Update more deprecated Font Awesome icon names (#1005)
* DEV: Update more deprecated Font Awesome icon names

* update to trash-can
2024-12-06 07:45:05 +11:00
Kris 70fae2b699 FIX: close link in shared conversation model (#1007) 2024-12-05 15:19:36 -05:00
Rafael dos Santos Silva 5ba274f7a5 FIX: Change how we filter date for emotion on /filter (#1006) 2024-12-05 14:10:12 -03:00
Roman Rizzi 4ba74511c2 FIX: Make sure limits are updated and applied on each step (#1002) 2024-12-05 10:31:39 -03:00
Kris 50f61925eb FIX: typo in topic gist for escapedExcerpt (#1003) 2024-12-05 08:03:08 -05:00
Kris 625fd06744 UX: improve layout of inline title editing buttons (#1000) 2024-12-04 17:04:26 -05:00
Roman Rizzi b32b1cf241 FIX: Add a digest check to avoid repeteadly generating embeddings (bulk) (#1001) 2024-12-04 17:47:28 -03:00
Keegan George d6beac48f8 DEV: Improve explain suggestion footnote replacement (#999)
Previously, when clicking add footnote on an explain suggestion it would replace the selected word by finding the first occurrence of the word. This results in issues when there are more than one occurrences of a word in a post. This is not trivial to solve, so this PR instead prevents incorrect text replacements by only allowing the replacement if it's unique. We use the same logic here that we use to determine if something can be fast edited.

In this PR we also update tests for post helper explain suggestions. For a while, we haven't had tests here due to streaming/timing issues, we've been skipping our system specs. In this PR, we add acceptance tests to handle this which gives us improved ability to publish message bus updates in the testing environment so that it can be better tested without issues.
2024-12-04 11:41:34 -08:00
Rafael dos Santos Silva 938d4c018c FIX: More resilient sentiment backfill query (#998) 2024-12-04 12:10:31 -03:00
Discourse Translator Bot e7c2cd861a Update translations (#994) 2024-12-04 11:59:37 +01:00
Kelv f8a81a0beb DEV: Update deprecated Font Awesome icon names (#996) 2024-12-04 03:18:08 +01:00
KrisandKeegan George 8203bdfbc9 UX: move topic summary from DMenu to DModal (#992)
Co-authored-by: Keegan George <[email protected]>
2024-12-03 13:30:15 -05:00
Roman Rizzi ce6a2eca21 FEATURE: Backfill posts sentiment. (#982)
* FEATURE: Backfill posts sentiment.

It adds a scheduled job to backfill posts' sentiment, similar to our existing rake task, but with two settings to control the batch size and posts' max-age.

* Make sure model_name order is consistent.
2024-12-03 10:27:03 -03:00
Sam 7c65dd171f FIX: regression, no longer sending examples to AI helper (#993)
For a while now we have not been sending the examples to AI
helper, which can lead to inconsistent results.

Note: this also means that in non English we did not send
English results, so this may end up reducing performance

That said first thing we need to do is fix the regression.
2024-12-03 16:03:46 +11:00
Rafael dos Santos Silva e3f5e86dc5 FIX: AI Automation scripts were broken when using seeded models (#991) 2024-12-02 19:07:05 -03:00
Tobias Eigen fcf2350652 changed "contact this model to "use this model" server.en.yml (#989)
translators gave feedback that the phrase "contact this model" is confusing. changed it to "use this model".
2024-12-03 07:26:54 +11:00
Keegan George fc88bb08ab FIX: Tag suggester is suggesting already assigned tags (#990)
This PR fixes an issue where the tag suggester for edit title topic area was suggesting tags that are already assigned on a post. It also updates the amount of suggested tags to 7 so that there is still a decent amount of tags suggested when tags are already assigned.
2024-12-03 07:25:04 +11:00
Sam 117c06220e FEATURE: allow artifacts to be updated (#980)
Add support for versioned artifacts with improved diff handling

* Add versioned artifacts support allowing artifacts to be updated and tracked
  - New `ai_artifact_versions` table to store version history
  - Support for updating artifacts through a new `UpdateArtifact` tool
  - Add version-aware artifact rendering in posts
  - Include change descriptions for version tracking

* Enhance artifact rendering and security
  - Add support for module-type scripts and external JS dependencies
  - Expand CSP to allow trusted CDN sources (unpkg, cdnjs, jsdelivr, googleapis)
  - Improve JavaScript handling in artifacts

* Implement robust diff handling system (this is dormant but ready to use once LLMs catch up)
  - Add new DiffUtils module for applying changes to artifacts
  - Support for unified diff format with multiple hunks
  - Intelligent handling of whitespace and line endings
  - Comprehensive error handling for diff operations

* Update routes and UI components
  - Add versioned artifact routes
  - Update markdown processing for versioned artifacts

Also

- Tweaks summary prompt
- Improves upload support in custom tool to also provide urls
2024-12-03 07:23:31 +11:00
Rafael dos Santos Silva 0ac18d157b FEATURE: Adjustments to gist summaries (#988)
- makes visible to everyone by default
- backfills gists before full summaries
- adds configurable max age setting to backfill job
2024-12-02 15:22:35 -03:00
Keegan George 50ad5415ff UX: no-text style being incorrectly added to tag suggester (#987)
In the tag suggester menu we use `DButton` as a wrapper element and use the `discourseTag` helper to render the text inside the element. So visually there is text content inside the button. However, since `DButton` assumes that no `label`/`translatedLabel` inside an element means `.no-text` CSS style should be applied to the button's element, it was resulting in some incorrect styling being applied to this menu. This PR resolves that by programmatically adding the tag as a `translatedLabel` and then visually hiding it with CSS.
2024-12-02 10:11:04 -08:00
Tobias Eigen c53363406d changed "automatic auto" to "auto" in client.en.yml (#986)
The description of the automatic captions dialog prompt included "Would you like to enable automatic auto captions on image uploads." I removed the second "auto" at the suggestion of translators.
2024-12-02 09:32:08 -08:00
Rafael dos Santos Silva 3828370679 DEV: Cleanup deprecations (#952) 2024-12-02 14:18:03 -03:00
298 changed files with 10941 additions and 3710 deletions
@@ -1,6 +1,6 @@
import DiscourseRoute from "discourse/routes/discourse";
export default class AdminPluginsShowDiscourseAiLlmsShow extends DiscourseRoute {
export default class AdminPluginsShowDiscourseAiLlmsEdit extends DiscourseRoute {
async model(params) {
const allLlms = this.modelFor("adminPlugins.show.discourse-ai-llms");
const id = parseInt(params.id, 10);
@@ -1,6 +1,6 @@
import DiscourseRoute from "discourse/routes/discourse";
export default class AdminPluginsShowDiscourseAiPersonasShow extends DiscourseRoute {
export default class AdminPluginsShowDiscourseAiPersonasEdit extends DiscourseRoute {
async model(params) {
const allPersonas = this.modelFor(
"adminPlugins.show.discourse-ai-personas"
@@ -0,0 +1,11 @@
import { service } from "@ember/service";
import { ajax } from "discourse/lib/ajax";
import DiscourseRoute from "discourse/routes/discourse";
export default class DiscourseAiSpamRoute extends DiscourseRoute {
@service store;
model() {
return ajax("/admin/plugins/discourse-ai/ai-spam.json");
}
}
@@ -1,6 +1,6 @@
import DiscourseRoute from "discourse/routes/discourse";
export default class DiscourseAiToolsShowRoute extends DiscourseRoute {
export default class DiscourseAiToolsEditRoute extends DiscourseRoute {
async model(params) {
const allTools = this.modelFor("adminPlugins.show.discourse-ai-tools");
const id = parseInt(params.id, 10);
@@ -0,0 +1 @@
<AiSpam @model={{this.model}} />
@@ -30,7 +30,10 @@ module DiscourseAi
}
end
def show
def new
end
def edit
llm_model = LlmModel.find(params[:id])
render json: LlmModelSerializer.new(llm_model)
end
@@ -5,7 +5,7 @@ module DiscourseAi
class AiPersonasController < ::Admin::AdminController
requires_plugin ::DiscourseAi::PLUGIN_NAME
before_action :find_ai_persona, only: %i[show update destroy create_user]
before_action :find_ai_persona, only: %i[edit update destroy create_user]
def index
ai_personas =
@@ -33,7 +33,10 @@ module DiscourseAi
render json: { ai_personas: ai_personas, meta: { tools: tools, llms: llms } }
end
def show
def new
end
def edit
render json: LocalizedAiPersonaSerializer.new(@ai_persona)
end
@@ -0,0 +1,112 @@
# frozen_string_literal: true
module DiscourseAi
module Admin
class AiSpamController < ::Admin::AdminController
requires_plugin "discourse-ai"
def show
render json: AiSpamSerializer.new(spam_config, root: false)
end
def update
updated_params = {}
if allowed_params.key?(:llm_model_id)
llm_model_id = updated_params[:llm_model_id] = allowed_params[:llm_model_id]
if llm_model_id.to_i < 0 &&
!SiteSetting.ai_spam_detection_model_allowed_seeded_models_map.include?(
"custom:#{llm_model_id}",
)
return(
render_json_error(
I18n.t("discourse_ai.llm.configuration.invalid_seeded_model"),
status: 422,
)
)
end
end
updated_params[:data] = {
custom_instructions: allowed_params[:custom_instructions],
} if allowed_params.key?(:custom_instructions)
if updated_params.present?
# not using upsert cause we will not get the correct validation errors
if AiModerationSetting.spam
AiModerationSetting.spam.update!(updated_params)
else
AiModerationSetting.create!(updated_params.merge(setting_type: :spam))
end
end
is_enabled = ActiveModel::Type::Boolean.new.cast(allowed_params[:is_enabled])
if allowed_params.key?(:is_enabled)
if is_enabled && !AiModerationSetting.spam&.llm_model_id
return(
render_json_error(
I18n.t("discourse_ai.llm.configuration.must_select_model"),
status: 422,
)
)
end
SiteSetting.ai_spam_detection_enabled = is_enabled
end
render json: AiSpamSerializer.new(spam_config, root: false)
end
def test
url = params[:post_url].to_s
post = nil
if url.match?(/^\d+$/)
post_id = url.to_i
post = Post.find_by(id: post_id)
end
route = UrlHelper.rails_route_from_url(url) if !post
if route
if route[:controller] == "topics"
post_number = route[:post_number] || 1
post = Post.with_deleted.find_by(post_number: post_number, topic_id: route[:topic_id])
end
end
raise Discourse::NotFound if !post
result =
DiscourseAi::AiModeration::SpamScanner.test_post(
post,
custom_instructions: params[:custom_instructions],
llm_id: params[:llm_id],
)
render json: result
end
private
def allowed_params
params.permit(:is_enabled, :llm_model_id, :custom_instructions)
end
def spam_config
spam_config = {
enabled: SiteSetting.ai_spam_detection_enabled,
settings: AiModerationSetting.spam,
}
spam_config[:stats] = DiscourseAi::AiModeration::SpamReport.generate(min_date: 1.week.ago)
if spam_config[:stats].scanned_count > 0
spam_config[
:flagging_username
] = DiscourseAi::AiModeration::SpamScanner.flagging_user&.username
end
spam_config
end
end
end
end
@@ -5,14 +5,17 @@ module DiscourseAi
class AiToolsController < ::Admin::AdminController
requires_plugin ::DiscourseAi::PLUGIN_NAME
before_action :find_ai_tool, only: %i[test show update destroy]
before_action :find_ai_tool, only: %i[test edit update destroy]
def index
ai_tools = AiTool.all
render_serialized({ ai_tools: ai_tools }, AiCustomToolListSerializer, root: false)
end
def show
def new
end
def edit
render_serialized(@ai_tool, AiCustomToolSerializer)
end
@@ -6,6 +6,9 @@ module DiscourseAi
requires_plugin "discourse-ai"
def show
end
def report
render json: AiUsageSerializer.new(create_report, root: false)
end
@@ -19,22 +19,33 @@ module DiscourseAi
raise Discourse::NotFound if !guardian.can_see?(post)
end
name = artifact.name
if params[:version].present?
artifact = artifact.versions.find_by(version_number: params[:version])
raise Discourse::NotFound if !artifact
end
js = artifact.js || ""
if !js.match?(%r{\A\s*<script.*</script>}mi)
mod = ""
mod = " type=\"module\"" if js.match?(/\A\s*import.*/)
js = "<script#{mod}>\n#{js}\n</script>"
end
# Prepare the inner (untrusted) HTML document
untrusted_html = <<~HTML
<!DOCTYPE html>
<html>
<head>
<meta charset="UTF-8">
<title>#{ERB::Util.html_escape(artifact.name)}</title>
<title>#{ERB::Util.html_escape(name)}</title>
<style>
#{artifact.css}
</style>
</head>
<body>
#{artifact.html}
<script>
#{artifact.js}
</script>
#{js}
</body>
</html>
HTML
@@ -45,7 +56,7 @@ module DiscourseAi
<html>
<head>
<meta charset="UTF-8">
<title>#{ERB::Util.html_escape(artifact.name)}</title>
<title>#{ERB::Util.html_escape(name)}</title>
<style>
html, body, iframe {
margin: 0;
@@ -67,7 +78,9 @@ module DiscourseAi
HTML
response.headers.delete("X-Frame-Options")
response.headers["Content-Security-Policy"] = "script-src 'unsafe-inline';"
response.headers[
"Content-Security-Policy"
] = "script-src 'self' 'unsafe-inline' 'wasm-unsafe-eval' https://unpkg.com https://cdnjs.cloudflare.com https://ajax.googleapis.com https://cdn.jsdelivr.net;"
response.headers["X-Robots-Tag"] = "noindex"
# Render the content
+13
View File
@@ -0,0 +1,13 @@
# frozen_string_literal: true
module Jobs
class AiSpamScan < ::Jobs::Base
def execute(args)
return if !args[:post_id]
post = Post.find_by(id: args[:post_id])
return if !post
DiscourseAi::AiModeration::SpamScanner.perform_scan(post)
end
end
end
+1 -3
View File
@@ -18,9 +18,7 @@ module ::Jobs
target = target_type.constantize.find_by(id: target_id)
return if !target
truncation = DiscourseAi::Embeddings::Strategies::Truncation.new
vector_rep =
DiscourseAi::Embeddings::VectorRepresentations::Base.current_representation(truncation)
vector_rep = DiscourseAi::Embeddings::VectorRepresentations::Base.current_representation
tokenizer = vector_rep.tokenizer
chunk_tokens = target.rag_chunk_tokens
-17
View File
@@ -1,17 +0,0 @@
# frozen_string_literal: true
module Jobs
class EvaluatePostUploads < ::Jobs::Base
def execute(args)
return unless SiteSetting.ai_nsfw_detection_enabled
return if (post_id = args[:post_id]).blank?
post = Post.includes(:uploads).find_by_id(post_id)
return if post.nil? || post.uploads.empty?
return if post.uploads.none? { |u| FileHelper.is_supported_image?(u.url) }
DiscourseAi::PostClassificator.new(DiscourseAi::Nsfw::Classification.new).classify!(post)
end
end
end
+1 -1
View File
@@ -14,7 +14,7 @@ module ::Jobs
summarizer = DiscourseAi::Summarization.topic_gist(topic)
gist = summarizer.existing_summary
return if gist.present? && !gist.outdated
return if gist.present? && (!gist.outdated || gist.created_at >= 5.minutes.ago)
summarizer.force_summarize(Discourse.system_user)
end
+1 -5
View File
@@ -16,11 +16,7 @@ module Jobs
return if topic.private_message? && !SiteSetting.ai_embeddings_generate_for_pms
return if post.raw.blank?
strategy = DiscourseAi::Embeddings::Strategies::Truncation.new
vector_rep =
DiscourseAi::Embeddings::VectorRepresentations::Base.current_representation(strategy)
vector_rep.generate_representation_from(target)
DiscourseAi::Embeddings::Vector.instance.generate_representation_from(target)
end
end
end
+2 -4
View File
@@ -8,13 +8,11 @@ module ::Jobs
def execute(args)
return if (fragments = RagDocumentFragment.where(id: args[:fragment_ids].to_a)).empty?
truncation = DiscourseAi::Embeddings::Strategies::Truncation.new
vector_rep =
DiscourseAi::Embeddings::VectorRepresentations::Base.current_representation(truncation)
vector = DiscourseAi::Embeddings::Vector.instance
# generate_representation_from checks compares the digest value to make sure
# the embedding is only generated once per fragment unless something changes.
fragments.map { |fragment| vector_rep.generate_representation_from(fragment) }
fragments.map { |fragment| vector.generate_representation_from(fragment) }
last_fragment = fragments.last
target = last_fragment.target
@@ -1,18 +0,0 @@
# frozen_string_literal: true
module ::Jobs
class ToxicityClassifyChatMessage < ::Jobs::Base
def execute(args)
return unless SiteSetting.ai_toxicity_enabled
return if (chat_message_id = args[:chat_message_id]).blank?
chat_message = ::Chat::Message.find_by(id: chat_message_id)
return if chat_message&.message.blank?
DiscourseAi::ChatMessageClassificator.new(
DiscourseAi::Toxicity::ToxicityClassification.new,
).classify!(chat_message)
end
end
end
@@ -1,19 +0,0 @@
# frozen_string_literal: true
module ::Jobs
class ToxicityClassifyPost < ::Jobs::Base
def execute(args)
return unless SiteSetting.ai_toxicity_enabled
post_id = args[:post_id]
return if post_id.blank?
post = Post.find_by(id: post_id, post_type: Post.types[:regular])
return if post&.raw.blank?
DiscourseAi::PostClassificator.new(
DiscourseAi::Toxicity::ToxicityClassification.new,
).classify!(post)
end
end
end
+36 -37
View File
@@ -20,10 +20,9 @@ module Jobs
rebaked = 0
strategy = DiscourseAi::Embeddings::Strategies::Truncation.new
vector_rep =
DiscourseAi::Embeddings::VectorRepresentations::Base.current_representation(strategy)
table_name = vector_rep.topic_table_name
vector = DiscourseAi::Embeddings::Vector.instance
vector_def = vector.vdef
table_name = DiscourseAi::Embeddings::Schema::TOPICS_TABLE
topics =
Topic
@@ -31,21 +30,20 @@ module Jobs
.where(archetype: Archetype.default)
.where(deleted_at: nil)
.order("topics.bumped_at DESC")
.limit(limit - rebaked)
rebaked += populate_topic_embeddings(vector_rep, topics)
rebaked += populate_topic_embeddings(vector, topics.limit(limit - rebaked))
return if rebaked >= limit
# Then, we'll try to backfill embeddings for topics that have outdated
# embeddings, be it model or strategy version
relation = topics.where(<<~SQL)
#{table_name}.model_version < #{vector_rep.version}
relation = topics.where(<<~SQL).limit(limit - rebaked)
#{table_name}.model_version < #{vector_def.version}
OR
#{table_name}.strategy_version < #{strategy.version}
#{table_name}.strategy_version < #{vector_def.strategy_version}
SQL
rebaked += populate_topic_embeddings(vector_rep, relation)
rebaked += populate_topic_embeddings(vector, relation)
return if rebaked >= limit
@@ -57,28 +55,30 @@ module Jobs
.where("#{table_name}.updated_at < topics.updated_at")
.limit((limit - rebaked) / 10)
populate_topic_embeddings(vector_rep, relation, force: true)
populate_topic_embeddings(vector, relation, force: true)
return if rebaked >= limit
return unless SiteSetting.ai_embeddings_per_post_enabled
# Now for posts
table_name = vector_rep.post_table_name
table_name = DiscourseAi::Embeddings::Schema::POSTS_TABLE
posts_batch_size = 1000
posts =
Post
.joins("LEFT JOIN #{table_name} ON #{table_name}.post_id = posts.id")
.where(deleted_at: nil)
.where(post_type: Post.types[:regular])
.limit(limit - rebaked)
# First, we'll try to backfill embeddings for posts that have none
posts
.where("#{table_name}.post_id IS NULL")
.find_in_batches do |batch|
vector_rep.gen_bulk_reprensentations(batch)
rebaked += batch.size
.limit(limit - rebaked)
.pluck(:id)
.each_slice(posts_batch_size) do |batch|
vector.gen_bulk_reprensentations(Post.where(id: batch))
rebaked += batch.length
end
return if rebaked >= limit
@@ -87,48 +87,47 @@ module Jobs
# embeddings, be it model or strategy version
posts
.where(<<~SQL)
#{table_name}.model_version < #{vector_rep.version}
#{table_name}.model_version < #{vector_def.version}
OR
#{table_name}.strategy_version < #{strategy.version}
#{table_name}.strategy_version < #{vector_def.strategy_version}
SQL
.find_in_batches do |batch|
vector_rep.gen_bulk_reprensentations(batch)
rebaked += batch.size
.limit(limit - rebaked)
.pluck(:id)
.each_slice(posts_batch_size) do |batch|
vector.gen_bulk_reprensentations(Post.where(id: batch))
rebaked += batch.length
end
return if rebaked >= limit
# Finally, we'll try to backfill embeddings for posts that have outdated
# embeddings due to edits. Here we only do 10% of the limit
posts_batch_size = 1000
outdated_post_ids =
posts
.where("#{table_name}.updated_at < ?", 7.days.ago)
.order("random()")
.limit((limit - rebaked) / 10)
.pluck(:id)
outdated_post_ids.each_slice(posts_batch_size) do |batch|
vector_rep.gen_bulk_reprensentations(Post.where(id: batch))
rebaked += batch.length
end
posts
.where("#{table_name}.updated_at < ?", 7.days.ago)
.order("random()")
.limit((limit - rebaked) / 10)
.pluck(:id)
.each_slice(posts_batch_size) do |batch|
vector.gen_bulk_reprensentations(Post.where(id: batch))
rebaked += batch.length
end
rebaked
end
private
def populate_topic_embeddings(vector_rep, topics, force: false)
def populate_topic_embeddings(vector, topics, force: false)
done = 0
topics = topics.where("#{vector_rep.topic_table_name}.topic_id IS NULL") if !force
topics =
topics.where("#{DiscourseAi::Embeddings::Schema::TOPICS_TABLE}.topic_id IS NULL") if !force
ids = topics.pluck("topics.id")
batch_size = 1000
ids.each_slice(batch_size) do |batch|
vector_rep.gen_bulk_reprensentations(Topic.where(id: batch).order("topics.bumped_at DESC"))
vector.gen_bulk_reprensentations(Topic.where(id: batch).order("topics.bumped_at DESC"))
done += batch.length
end
+30
View File
@@ -0,0 +1,30 @@
# frozen_string_literal: true
module Jobs
class SentimentBackfill < ::Jobs::Scheduled
every 5.minutes
cluster_concurrency 1
def execute(_args)
return if !SiteSetting.ai_sentiment_enabled
base_budget = SiteSetting.ai_sentiment_backfill_maximum_posts_per_hour
return if base_budget.zero?
# Split budget in 12 intervals, but make sure is at least one.
#
# This is not exact as we don't have a way of tracking how many
# posts we classified in the current hour, but it's a good enough approximation.
limit_per_job = [base_budget, 12].max / 12
classificator = DiscourseAi::Sentiment::PostClassification.new
return if !classificator.has_classifiers?
posts =
DiscourseAi::Sentiment::PostClassification.backfill_query(
max_age_days: SiteSetting.ai_sentiment_backfill_post_max_age_days,
).limit(limit_per_job)
classificator.bulk_classify!(posts)
end
end
end
+20 -9
View File
@@ -12,22 +12,27 @@ module ::Jobs
system_user = Discourse.system_user
if SiteSetting.ai_summary_gists_enabled
gist_t = AiSummary.summary_types[:gist]
backfill_candidates(gist_t)
.limit(current_budget(gist_t))
.each do |topic|
DiscourseAi::Summarization.topic_gist(topic).force_summarize(system_user)
end
end
complete_t = AiSummary.summary_types[:complete]
backfill_candidates(complete_t)
.limit(current_budget(complete_t))
.each do |topic|
DiscourseAi::Summarization.topic_summary(topic).force_summarize(system_user)
end
return unless SiteSetting.ai_summary_gists_enabled
gist_t = AiSummary.summary_types[:gist]
backfill_candidates(gist_t)
.limit(current_budget(gist_t))
.each { |topic| DiscourseAi::Summarization.topic_gist(topic).force_summarize(system_user) }
end
def backfill_candidates(summary_type)
max_age_days = SiteSetting.ai_summary_backfill_topic_max_age_days
Topic
.where("topics.word_count >= ?", SiteSetting.ai_summary_backfill_minimum_word_count)
.joins(<<~SQL)
@@ -36,9 +41,15 @@ module ::Jobs
ais.target_type = 'Topic' AND
ais.summary_type = '#{summary_type}'
SQL
.where("topics.created_at > current_timestamp - INTERVAL '#{max_age_days.to_i} DAY'")
.where(
"ais.id IS NULL OR UPPER(ais.content_range) < topics.highest_post_number + 1",
) # (1..1) gets stored ad (1..2).
<<~SQL, # (1..1) gets stored ad (1..2).
ais.id IS NULL OR (
UPPER(ais.content_range) < topics.highest_post_number + 1
AND ais.created_at < (current_timestamp - INTERVAL '5 minutes')
)
SQL
)
.order("ais.created_at DESC NULLS FIRST, topics.last_posted_at DESC")
end
+1
View File
@@ -15,6 +15,7 @@ class AiApiAuditLog < ActiveRecord::Base
Ollama = 7
SambaNova = 8
Mistral = 9
OpenRouter = 10
end
def next_log_id
+42 -5
View File
@@ -1,23 +1,29 @@
# frozen_string_literal: true
class AiArtifact < ActiveRecord::Base
has_many :versions, class_name: "AiArtifactVersion", dependent: :destroy
belongs_to :user
belongs_to :post
validates :html, length: { maximum: 65_535 }
validates :css, length: { maximum: 65_535 }
validates :js, length: { maximum: 65_535 }
def self.iframe_for(id)
def self.iframe_for(id, version = nil)
<<~HTML
<div class='ai-artifact'>
<iframe src='#{url(id)}' frameborder="0" height="100%" width="100%"></iframe>
<a href='#{url(id)}' target='_blank'>#{I18n.t("discourse_ai.ai_artifact.link")}</a>
<iframe src='#{url(id, version)}' frameborder="0" height="100%" width="100%"></iframe>
<a href='#{url(id, version)}' target='_blank'>#{I18n.t("discourse_ai.ai_artifact.link")}</a>
</div>
HTML
end
def self.url(id)
Discourse.base_url + "/discourse-ai/ai-bot/artifacts/#{id}"
def self.url(id, version = nil)
url = Discourse.base_url + "/discourse-ai/ai-bot/artifacts/#{id}"
if version
"#{url}/#{version}"
else
url
end
end
def self.share_publicly(id:, post:)
@@ -33,6 +39,37 @@ class AiArtifact < ActiveRecord::Base
def url
self.class.url(id)
end
def apply_diff(html_diff: nil, css_diff: nil, js_diff: nil, change_description: nil)
differ = DiscourseAi::Utils::DiffUtils
html = html_diff ? differ.apply_hunk(self.html, html_diff) : self.html
css = css_diff ? differ.apply_hunk(self.css, css_diff) : self.css
js = js_diff ? differ.apply_hunk(self.js, js_diff) : self.js
create_new_version(html: html, css: css, js: js, change_description: change_description)
end
def create_new_version(html: nil, css: nil, js: nil, change_description: nil)
latest_version = versions.order(version_number: :desc).first
new_version_number = latest_version ? latest_version.version_number + 1 : 1
version = nil
transaction do
# Create the version record
version =
versions.create!(
version_number: new_version_number,
html: html || self.html,
css: css || self.css,
js: js || self.js,
change_description: change_description,
)
save!
end
version
end
end
# == Schema Information
+27
View File
@@ -0,0 +1,27 @@
# frozen_string_literal: true
class AiArtifactVersion < ActiveRecord::Base
belongs_to :ai_artifact
validates :html, length: { maximum: 65_535 }
validates :css, length: { maximum: 65_535 }
validates :js, length: { maximum: 65_535 }
end
# == Schema Information
#
# Table name: ai_artifact_versions
#
# id :bigint not null, primary key
# ai_artifact_id :bigint not null
# version_number :integer not null
# html :string(65535)
# css :string(65535)
# js :string(65535)
# metadata :jsonb
# change_description :string
# created_at :datetime not null
# updated_at :datetime not null
#
# Indexes
#
# index_ai_artifact_versions_on_ai_artifact_id_and_version_number (ai_artifact_id,version_number) UNIQUE
#
+32
View File
@@ -0,0 +1,32 @@
# frozen_string_literal: true
class AiModerationSetting < ActiveRecord::Base
belongs_to :llm_model
validates :llm_model_id, presence: true
validates :setting_type, presence: true
validates :setting_type, uniqueness: true
def self.spam
find_by(setting_type: :spam)
end
def custom_instructions
data["custom_instructions"]
end
end
# == Schema Information
#
# Table name: ai_moderation_settings
#
# id :bigint not null, primary key
# setting_type :enum not null
# data :jsonb
# llm_model_id :bigint not null
# created_at :datetime not null
# updated_at :datetime not null
#
# Indexes
#
# index_ai_moderation_settings_on_setting_type (setting_type) UNIQUE
#
+26
View File
@@ -0,0 +1,26 @@
# frozen_string_literal: true
class AiSpamLog < ActiveRecord::Base
belongs_to :post
belongs_to :llm_model
belongs_to :ai_api_audit_log
belongs_to :reviewable
end
# == Schema Information
#
# Table name: ai_spam_logs
#
# id :bigint not null, primary key
# post_id :bigint not null
# llm_model_id :bigint not null
# ai_api_audit_log_id :bigint
# reviewable_id :bigint
# is_spam :boolean not null
# payload :string(20000) default(""), not null
# created_at :datetime not null
# updated_at :datetime not null
#
# Indexes
#
# index_ai_spam_logs_on_post_id (post_id)
#

Some files were not shown because too many files have changed in this diff Show More