Merge branch 'develop' into optic-1353

.github/CODEOWNERS

@@ -1,4 +1,7 @@
 .github/          @HumanSignal/devops
+deploy/           @HumanSignal/devops
+Dockerfile*       @HumanSignal/devops
+docker-compose*   @HumanSignal/devops
 docs/.npmignore   @hlomzik @Gondragos @nicholasrq
 docs/package.json @hlomzik @Gondragos @nicholasrq
 docs/yarn.lock    @hlomzik @Gondragos @nicholasrq

.github/workflows/test_conda.yml

@@ -40,7 +40,7 @@ jobs:
           ref: ${{ inputs.head_sha }}
 
       - name: Setup Conda
-        uses: conda-incubator/setup-miniconda@v3.0.4
+        uses: conda-incubator/setup-miniconda@v3.1.0
         with:
           auto-update-conda: true
           python-version: ${{ matrix.python-version }}

.github/workflows/tests.yml

@@ -163,7 +163,7 @@ jobs:
 
       - name: Upload coverage to Codecov
         if: ${{ github.event.pull_request.head.repo.fork == false && github.event.pull_request.user.login != 'dependabot[bot]' }}
-        uses: codecov/codecov-action@v4.6.0
+        uses: codecov/codecov-action@v5.0.2
         with:
           name: codecov-python-${{ matrix.python-version }}
           flags: pytests

Makefile

@@ -1,18 +1,18 @@
 # Run Django dev server with Sqlite
 run-dev:
-	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio python label_studio/manage.py runserver
+	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio poetry run python label_studio/manage.py runserver
 
 # Run Django dev migrations with Sqlite
 migrate-dev:
-	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio python label_studio/manage.py migrate
+	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio poetry run python label_studio/manage.py migrate
 
 # Run Django dev make migrations with Sqlite
 makemigrations-dev:
-	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio python label_studio/manage.py makemigrations
+	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio poetry run python label_studio/manage.py makemigrations
 
 # Run Django dev shell environment with Sqlite
 shell-dev:
-	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio python label_studio/manage.py shell_plus
+	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio poetry run python label_studio/manage.py shell_plus
 
 env-dev-setup:
 	if [ ! -f .env ]; then \
@@ -72,7 +72,7 @@ docker-testing-shell:
 
 # Update urls
 update-urls:
-	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio python label_studio/manage.py show_urls --format pretty-json > ./label_studio/core/all_urls.json
+	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio poetry run python label_studio/manage.py show_urls --format pretty-json > ./label_studio/core/all_urls.json
 
 # Format changed files on branch
 fmt:
@@ -96,4 +96,4 @@ configure-hooks:
 
 # Generate swagger.json
 generate-swagger:
-	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio python label_studio/manage.py generate_swagger swagger.json
+	DJANGO_DB=sqlite LOG_DIR=tmp DEBUG=true LOG_LEVEL=DEBUG DJANGO_SETTINGS_MODULE=core.settings.label_studio poetry run python label_studio/manage.py generate_swagger swagger.json

docs/scripts/index.js

@@ -1,7 +1,9 @@
 var breadcrumb = require("./breadcrumb")(hexo);
 var removeContent = require("./removeContent")(hexo);
 var replaceContent = require("./replaceContent")(hexo);
+var tableOfContent = require("./tableOfContent")(hexo);
 
 hexo.extend.helper.register("breadcrumb", breadcrumb, { async: true });
 hexo.extend.helper.register("removeContent", removeContent, { async: true });
 hexo.extend.helper.register("replaceContent", replaceContent, { async: true });
+hexo.extend.helper.register("tableOfContent", tableOfContent, { async: true });

docs/scripts/tableOfContent.js

@@ -0,0 +1,24 @@
+const { tocObj } = require('hexo-util');
+
+module.exports = function () {
+  return function tableOfContent(str, options = {}) {
+
+      options = Object.assign({
+        min_depth: 1,
+        max_depth: 6,
+      }, options);
+
+      const data = tocObj(str, { min_depth: options.min_depth, max_depth: options.max_depth });
+
+
+      if (!data.length) return '';
+
+      let html = '';
+
+      data.forEach(item => {
+        html += `<li class="toc-list-level-${item.level}"><a href="#${item.id}">${item.text}</a></li>`
+      });
+
+      return `<ol class="toc-list">${html}</ol>`;
+  };
+};

docs/source/templates/generative-pairwise-human-preference.md

@@ -15,7 +15,7 @@ This project will help you to get up your LLM to the ChatGPT quality level throu
 
 Through ranking multiple responses based on quality, you can train a reward model that effectively captures human preferences. This reward model plays a crucial role in Reinforcement Learning, optimizing the performance of the fine-tuned foundational model.
 
-### Further Reading and Resources
+#### Further Reading and Resources
 
 - [Gathering Human Feedback Tutorial](https://github.com/heartexlabs/RLHF/blob/master/tutorials/RLHF_with_Custom_Datasets.ipynb) A Jupyter Notebook tutorial that will guide you through the step-by-step process of collecting comparison data, establishing human preferences, and incorporating this feedback into the reward model training.
 - [RLHF Resources](https://github.com/heartexlabs/RLHF): A collection of links, tutorials and best practices on how collect data and build an end-to-end Reinforcement Learning from Human Feedback (RLHF) system to fine-tune Generative AI models.

docs/source/templates/index.ejs

@@ -51,6 +51,6 @@ cards:
   url: "/templates/gallery_generative_ai.html"
 - title: LLM Evaluations
   categories: generative ai, llm
-  image: "/images/templates/generative-pairwise-human-preference.png"
+  image: "/images/templates/response-moderation.png"
   url: "/templates/gallery_llm_evals.html"
 ---

docs/themes/v2/layout/page.ejs

@@ -1,11 +1,10 @@
 <% const isEnterpriseTheme = theme.tier === "enterprise" %>
-<% const pageContainsHeadings = page.content.includes("<h2") %>
 <% const tocPageTypes = page.type === "guide" || page.type === "blog" || page.type === "templates" || page.type === "tags" %>
 
-<% const showToc = pageContainsHeadings && tocPageTypes %>
-
 <% const formattedContent = removeContent(page.content) %>
 <% const tocContent = replaceContent(formattedContent) %>
+<% const pageHasHeading = /<h[1-6]\b[^>]*>/i.test(formattedContent); %>
+<% const showToc = tocPageTypes && pageHasHeading %>
 
 <div class="content-grid">
   <div class="content-markdown">
@@ -21,7 +20,7 @@
     <% if (showToc) { %>
       <div class="toc">
         <%- partial("component/text", {text: "In this article", tag: "h3", size: "Eyebrow"}) %>
-        <%- toc(tocContent, {max_depth: 3, list_number: false, class: "toc-list"}) %>
+        <%- tableOfContent(tocContent, {max_depth: 3}) %>
       </div>
     <% } %>
     <div class="toc-enterprise-cta">

docs/themes/v2/source/css/styles.css

@@ -1223,6 +1223,10 @@ ul {
   margin-top: 1em;
 }
 
+.toc-list-level-2~.toc-list-level-3 {
+  padding-left: 1em;
+}
+
 .toc-list-child {
   list-style: none;
   padding-left: 0.825em;

docs/yarn.lock

@@ -1674,9 +1674,9 @@ create-error-class@^3.0.2:
     capture-stack-trace "^1.0.0"
 
 cross-spawn@^7.0.0, cross-spawn@^7.0.3:
-  version "7.0.3"
-  resolved "https://registry.yarnpkg.com/cross-spawn/-/cross-spawn-7.0.3.tgz#f73a85b9d5d41d045551c177e2882d4ac85728a6"
-  integrity sha512-iRDPJKUPVEND7dHPO8rkbOnPpyDygcDFtWjpeWNCgy8WP2rXcxXL8TskReQl6OrB2G7+UJrags1q15Fudc7G6w==
+  version "7.0.5"
+  resolved "https://registry.yarnpkg.com/cross-spawn/-/cross-spawn-7.0.5.tgz#910aac880ff5243da96b728bc6521a5f6c2f2f82"
+  integrity sha512-ZVJrKKYunU38/76t0RMOulHOnUcbU9GbpWKAOZ0mhjr7CX6FVrH+4FrAapSOekrgFQ3f/8gwMEuIft0aKq6Hug==
   dependencies:
     path-key "^3.1.0"
     shebang-command "^2.0.0"

label_studio/core/feature_flags/base.py

@@ -8,13 +8,16 @@
 from ldclient.integrations import Files, Redis
 
 from label_studio.core.current_request import get_current_request
+from label_studio.core.utils.common import load_func
 from label_studio.core.utils.io import find_node
 from label_studio.core.utils.params import get_all_env_with_prefix, get_bool_env
 
 from .stale_feature_flags import STALE_FEATURE_FLAGS
 
 logger = logging.getLogger(__name__)
 
+get_user_repr = load_func(settings.FEATURE_FLAGS_GET_USER_REPR)
+
 
 def get_feature_file_path():
     package_name = 'label_studio' if settings.VERSION_EDITION == 'Community' else 'label_studio_enterprise'
@@ -66,19 +69,6 @@ def get_feature_file_path():
     client = ldclient.get()
 
 
-def _get_user_repr(user):
-    """Turn user object into dict with required properties"""
-    if user.is_anonymous:
-        return {'key': str(user), 'custom': {'organization': None}}
-    user_data = {'email': user.email}
-    user_data['key'] = user_data['email']
-    if user.active_organization is not None:
-        user_data['custom'] = {'organization': user.active_organization.created_by.email}
-    else:
-        user_data['custom'] = {'organization': None}
-    return user_data
-
-
 def flag_set(feature_flag, user=None, override_system_default=None):
     """Use this method to check whether this flag is set ON to the current user, to split the logic on backend
     For example,
@@ -114,7 +104,7 @@ def flag_set(feature_flag, user=None, override_system_default=None):
         system_default = override_system_default
     else:
         system_default = settings.FEATURE_FLAGS_DEFAULT_VALUE
-    user_dict = _get_user_repr(user)
+    user_dict = get_user_repr(user)
     return client.variation(feature_flag, user_dict, system_default)
 
 
@@ -123,7 +113,7 @@ def all_flags(user):
     More on https://docs.launchdarkly.com/sdk/features/bootstrapping#javascript
     stale_feature_flags will override any client configuration
     """
-    user_dict = _get_user_repr(user)
+    user_dict = get_user_repr(user)
     logger.debug(f'Resolve all flags state for user {user_dict}')
     state = client.all_flags_state(user_dict)
     flags = state.to_json_dict()

label_studio/core/feature_flags/stale_feature_flags.py

@@ -24,7 +24,6 @@
     'fflag_feat_front_lsdv_4661_full_uri_resolve_15032023_short': True,
     'fflag_fix_back_LSDV_4748_annotate_task_number_14032023_short': True,
     'fflag_feat_front_lsdv_4659_skipduplicates_060323_short': True,
-    'fflag_fix_back_LSDV_961_project_list_09022023_short': True,
     'fflag_fix_font_lsdv_1148_hotkeys_namespaces_01022023_short': True,
     'fflag_fix_back_lsdv_1044_check_annotations_24012023_short': False,
     'fflag_feat_front_dev_4081_magic_wand_tool': True,
@@ -35,7 +34,6 @@
     'fflag_feat_back_dev_3756_queue_enrollment_min_short': False,
     'fflag_fix_front_dev_3666_max_usages_on_region_creation_171122_short': True,
     'fflag_fix_front_dev_3617_taxonomy_memory_leaks_fix': True,
-    'ff_back_dev_2915_storage_nginx_proxy_26092022_short': True,
     'fflag_fix_front_dev_3350_restrict_drawing_area_short': True,
     'fflag_feat_front_dev_2984_dm_draggable_columns_short': True,
     'fflag_feat_front_dev_2461_audio_paragraphs_seek_chunk_position_short': True,

label_studio/core/feature_flags/utils.py

@@ -0,0 +1,11 @@
+def get_user_repr(user):
+    """Turn user object into dict with required properties"""
+    if user.is_anonymous:
+        return {'key': str(user), 'custom': {'organization': None}}
+    user_data = {'email': user.email}
+    user_data['key'] = user_data['email']
+    if user.active_organization is not None:
+        user_data['custom'] = {'organization': user.active_organization.created_by.email}
+    else:
+        user_data['custom'] = {'organization': None}
+    return user_data

label_studio/core/label_config.py

@@ -261,25 +261,26 @@ def generate_sample_task_without_check(label_config, mode='upload', secure_mode=
     # make examples pretty
     examples = data_examples(mode=mode)
 
-    # iterate over xml tree and find values with '$'
+    # iterate over xml tree and find elements with 'value' or 'valueList' attributes
     task = {}
-    parent = xml.findall('.//*[@value]')  # take all tags with value attribute
+    # Include both 'value' and 'valueList' attributes in the search
+    parent = xml.findall('.//*[@value]') + xml.findall('.//*[@valueList]')
     for p in parent:
-
-        # Make sure it is a real object tag, extract data placeholder key
-        value = p.get('value')
+        # Extract data placeholder key
+        value = p.get('value') or p.get('valueList')
         if not value or not value.startswith('$'):
             continue
         value = value[1:]
+        is_value_list = 'valueList' in p.attrib  # Check if the attribute is 'valueList'
 
         # detect secured mode - objects served as URLs
         value_type = p.get('valueType') or p.get('valuetype')
         only_urls = secure_mode or value_type == 'url'
 
         example_from_field_name = examples.get('$' + value)
         if example_from_field_name:
-            # try to get example by variable name
-            task[value] = example_from_field_name
+            # Get example by variable name
+            task[value] = [example_from_field_name] if is_value_list else example_from_field_name
 
         elif value == 'video' and p.tag == 'HyperText':
             task[value] = examples.get('$videoHack')
@@ -297,7 +298,6 @@ def generate_sample_task_without_check(label_config, mode='upload', secure_mode=
                 task[value] = []
                 for item in examples[p.tag]:
                     task[value].append({name_key: item['author'], text_key: item['text']})
-
         elif p.tag == 'TimeSeries':
             # TimeSeries special case - generate signals on-the-fly
             time_column = p.get('timeColumn')
@@ -334,10 +334,11 @@ def generate_sample_task_without_check(label_config, mode='upload', secure_mode=
             else:
                 task[value] = examples['Choices']
         else:
-            # patch for valueType="url"
+            # Patch for valueType="url"
             examples['Text'] = examples['TextUrl'] if only_urls else examples['TextRaw']
-            # not found by name, try get example by type
-            task[value] = examples.get(p.tag, 'Something')
+            # Not found by name, try to get example by type
+            example_value = examples.get(p.tag, 'Something')
+            task[value] = [example_value] if is_value_list else example_value
 
         # support for Repeater tag
         if '[' in value:

label_studio/core/settings/base.py

@@ -572,6 +572,7 @@
 STORAGE_PERMISSION = 'io_storages.permissions.StoragePermission'
 PROJECT_IMPORT_PERMISSION = 'projects.permissions.ProjectImportPermission'
 DELETE_TASKS_ANNOTATIONS_POSTPROCESS = None
+FEATURE_FLAGS_GET_USER_REPR = 'core.feature_flags.utils.get_user_repr'
 
 
 def project_delete(project):

label_studio/core/static/css/login.css

@@ -181,7 +181,8 @@ html, body {
   font-size: .875rem;
 }
 .login_page_new_ui form input,
-.login_page_new_ui form button {
+.login_page_new_ui form button,
+.login_page_new_ui form select {
   width: 100%;
   font-family: 'Hellix', sans-serif;
 }

label_studio/core/static/css/uikit.css

@@ -227,4 +227,3 @@ ul.toggle li.active {
   box-shadow: none;
   outline: none;
 }
-