initial commit ai-search task

chatbot/services/search/vocabularies.py

@@ -11,56 +11,57 @@
 
 import logging
 
-from django.core.cache import cache
-
-from chatbot.models.company_models import Company
 from chatbot.models.enums import EntityStatus, FileTypeChoices
 from chatbot.services.search.config import (
     VOCAB_AUTO,
     VOCAB_CANDIDATES,
     VOCAB_FULL,
     get_search_llm_setting,
 )
+from chatbot.utils.company_cache import get_all_companies
 
 logger = logging.getLogger('django')
 
-CACHE_TTL_SECONDS = 300
-ORG_CACHE_KEY = 'ai_search:org_vocab:{scope}'
-
 
 def organization_vocabulary(scope=None):
     """
-    ``{slug: [display name]}`` for every active organization, cached per scope.
+    ``{slug: [display name]}`` for every active organization.
 
-    ``scope`` is unused today (search is global) but keeps the cache key ready
-    for tenant-scoped search, so an unscoped cache can't leak across tenants later.
+    The organization list comes from Vishwa's Redis-backed company cache
+    (`get_all_companies`) instead of SEARCH_FILTER_ORGANIZATIONS. When the Redis
+    cache is cold, that helper hydrates it from the database.
     """
-    key = ORG_CACHE_KEY.format(scope=scope or 'global')
-    cached = cache.get(key)
-    if cached is not None:
-        return cached
-
-    rows = (
-        Company.objects
-        .filter(status=EntityStatus.ACTIVE)
-        .exclude(slug__isnull=True).exclude(slug='')
-        .values_list('slug', 'name')
-    )
-    # Skip inactive orgs: suggesting one would offer a filter that matches nothing.
-    vocabulary = {slug: [name] for slug, name in rows if slug}
-
-    cache.set(key, vocabulary, CACHE_TTL_SECONDS)
-    return vocabulary
+    return {
+        company.slug: [company.name]
+        for company in get_all_companies()
+        if (
+            company.slug
+            and company.status == EntityStatus.ACTIVE
+        )
+    }
 
 
 def file_type_vocabulary():
     """``{mime: [label, ext, .ext]}`` from the static FileTypeChoices enum."""
+    vector_type_aliases = {
+        FileTypeChoices.CSV.value: ["project_task"],
+        FileTypeChoices.XLS.value: ["xlsx_rag_optimized"],
+        FileTypeChoices.XLSX.value: ["xlsx_rag_optimized"],
+        FileTypeChoices.TXT.value: ["text", "markdown"],
+    }
     extensions = FileTypeChoices.get_extension_mapping()
     vocabulary = {}
     for choice in FileTypeChoices:
         dotted = extensions.get(choice, '')
         bare = dotted.lstrip('.')
-        vocabulary[choice.value] = [str(choice.label), bare, dotted]
+        label = str(choice.label)
+        aliases = [label, bare, dotted]
+        aliases.extend(vector_type_aliases.get(choice.value, []))
+        if label:
+            aliases.append(f"{label}s")
+        if bare:
+            aliases.append(f"{bare}s")
+        vocabulary[choice.value] = list(dict.fromkeys(aliases))
     return vocabulary
 
 

chatbot/utils/chat_query_handler.py

@@ -126,7 +126,7 @@ def query_database_with_metadata(
     file_type: List[str] = None,
     exclude_organizations: List[str] = None,
     exclude_file_type: List[str] = None,
-    any_of: List[Dict[str, Any]] = None
+    any_of: List[Dict[str, Any]] = None,
 ):
     """
     Query vector database with metadata filters for media search v2.
@@ -152,7 +152,7 @@ def query_database_with_metadata(
         "filter_score": filter_score,
         "detail_filter_score": detail_filter_score
     }
-    
+
     # Add query only if provided
     if query:
         data["query"] = query