Merge pull request #9 from Zipstack/feature/prompt-studio-summarize

Prompt Studio - Summarize, Output Analyzer and other UI Improvements

.gitignore

@@ -607,6 +607,9 @@ $RECYCLE.BIN/
 backend/plugins/authentication/*
 !backend/plugins/authentication/auth_sample
 
+# Processor Plugins
+backend/plugins/processor/*
+
 # Tool registry
 unstract/tool-registry/src/unstract/tool_registry/*.json
 unstract/tool-registry/tests/*.yaml
@@ -627,5 +630,3 @@ tools/*/sdks/
 tools/*/data_dir/
 
 docker/workflow_data/
-
-

backend/adapter_processor/migrations/0005_alter_adapterinstance_adapter_type.py

@@ -0,0 +1,27 @@
+# Generated by Django 4.2.1 on 2024-02-28 09:03
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("adapter_processor", "0004_alter_adapterinstance_adapter_type"),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name="adapterinstance",
+            name="adapter_type",
+            field=models.CharField(
+                choices=[
+                    ("UNKNOWN", "UNKNOWN"),
+                    ("LLM", "LLM"),
+                    ("EMBEDDING", "EMBEDDING"),
+                    ("VECTOR_DB", "VECTOR_DB"),
+                    ("OCR", "OCR"),
+                    ("X2TEXT", "X2TEXT"),
+                ],
+                db_comment="Type of adapter LLM/EMBEDDING/VECTOR_DB",
+            ),
+        ),
+    ]

backend/file_management/file_management_helper.py

@@ -195,8 +195,8 @@ def fetch_file_contents(
         elif file_content_type == "text/plain":
             with fs.open(file_path, "r") as file:
                 FileManagerHelper.logger.info(f"Reading text file: {file_path}")
-                encoded_string = base64.b64encode(file.read())
-                return encoded_string
+                text_content = file.read()
+                return text_content
         else:
             raise InvalidFileType
 
@@ -246,9 +246,13 @@ def handle_sub_directory_for_tenants(
             raise OrgIdNotValid()
         base_path = Path(settings.PROMPT_STUDIO_FILE_PATH)
         file_path: Path = base_path / org_id / user_id / tool_id
+        extract_file_path: Path = Path(file_path, "extract")
+        summarize_file_path: Path = Path(file_path, "summarize")
         if is_create:
             try:
                 os.makedirs(file_path, exist_ok=True)
+                os.makedirs(extract_file_path, exist_ok=True)
+                os.makedirs(summarize_file_path, exist_ok=True)
             except OSError as e:
                 FileManagerHelper.logger.error(
                     f"Error while creating {file_path}: {e}"

backend/file_management/views.py

@@ -121,7 +121,6 @@ def upload(self, request: HttpRequest) -> Response:
 
     @action(detail=True, methods=["post"])
     def upload_for_ide(self, request: HttpRequest) -> Response:
-        print(request.data)
         serializer = FileUploadIdeSerializer(data=request.data)
         serializer.is_valid(raise_exception=True)
         uploaded_files: Any = serializer.validated_data.get("file")

backend/prompt_studio/processor_loader.py

@@ -0,0 +1,76 @@
+import logging
+import os
+from importlib import import_module
+from typing import Any
+
+from django.apps import apps
+
+logger = logging.getLogger(__name__)
+
+
+class ProcessorConfig:
+    """Loader config for processor plugins."""
+
+    PLUGINS_APP = "plugins"
+    PLUGIN_DIR = "processor"
+    MODULE = "module"
+    METADATA = "metadata"
+    METADATA_NAME = "name"
+    METADATA_SERVICE_CLASS = "service_class"
+    METADATA_IS_ACTIVE = "is_active"
+
+
+def load_plugins() -> list[Any]:
+    """Iterate through the processor plugins and register them."""
+    plugins_app = apps.get_app_config(ProcessorConfig.PLUGINS_APP)
+    package_path = plugins_app.module.__package__
+    processor_dir = os.path.join(plugins_app.path, ProcessorConfig.PLUGIN_DIR)
+    processor_package_path = f"{package_path}.{ProcessorConfig.PLUGIN_DIR}"
+    processor_plugins: list[Any] = []
+
+    for item in os.listdir(processor_dir):
+        # Loads a plugin if it is in a directory.
+        if os.path.isdir(os.path.join(processor_dir, item)):
+            processor_module_name = item
+        # Loads a plugin if it is a shared library.
+        # Module name is extracted from shared library name.
+        # `processor.platform_architecture.so` will be file name and
+        # `processor` will be the module name.
+        elif item.endswith(".so"):
+            processor_module_name = item.split(".")[0]
+        else:
+            continue
+        try:
+            full_module_path = (
+                f"{processor_package_path}.{processor_module_name}"
+            )
+            module = import_module(full_module_path)
+            metadata = getattr(module, ProcessorConfig.METADATA, {})
+
+            if metadata.get(ProcessorConfig.METADATA_IS_ACTIVE, False):
+                processor_plugins.append(
+                    {
+                        ProcessorConfig.MODULE: module,
+                        ProcessorConfig.METADATA: module.metadata,
+                    }
+                )
+                logger.info(
+                    "Loaded processor plugin: %s, is_active: %s",
+                    module.metadata[ProcessorConfig.METADATA_NAME],
+                    module.metadata[ProcessorConfig.METADATA_IS_ACTIVE],
+                )
+            else:
+                logger.info(
+                    "Processor plugin %s is not active.",
+                    processor_module_name,
+                )
+        except ModuleNotFoundError as exception:
+            logger.error(
+                "Error while importing processor plugin: %s",
+                exception,
+            )
+
+    if len(processor_plugins) == 0:
+        logger.info("No processor plugins found.")
+
+    return processor_plugins

backend/prompt_studio/prompt_studio_core/constants.py

@@ -73,6 +73,8 @@ class ToolStudioPromptKeys:
     EVAL_SETTINGS_EVALUATE = "evaluate"
     EVAL_SETTINGS_MONITOR_LLM = "monitor_llm"
     EVAL_SETTINGS_EXCLUDE_FAILED = "exclude_failed"
+    SUMMARIZE = "summarize"
+    SUMMARIZED_RESULT = "summarized_result"
 
 
 class LogLevels:

backend/prompt_studio/prompt_studio_core/migrations/0004_customtool_summarize_as_source_and_more.py

@@ -0,0 +1,42 @@
+# Generated by Django 4.2.1 on 2024-02-27 05:43
+
+import django.db.models.deletion
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("prompt_profile_manager", "0006_alter_profilemanager_x2text"),
+        ("prompt_studio_core", "0003_merge_20240125_1501"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="customtool",
+            name="summarize_as_source",
+            field=models.BooleanField(default=True),
+        ),
+        migrations.AddField(
+            model_name="customtool",
+            name="summarize_context",
+            field=models.BooleanField(default=True),
+        ),
+        migrations.AddField(
+            model_name="customtool",
+            name="summarize_llm_profile",
+            field=models.ForeignKey(
+                blank=True,
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                related_name="summarize_llm_profile",
+                to="prompt_profile_manager.profilemanager",
+            ),
+        ),
+        migrations.AddField(
+            model_name="customtool",
+            name="summarize_prompt",
+            field=models.TextField(
+                blank=True, db_comment="Field to store the summarize prompt"
+            ),
+        ),
+    ]

backend/prompt_studio/prompt_studio_core/migrations/0005_alter_customtool_default_profile_and_more.py

@@ -0,0 +1,53 @@
+# Generated by Django 4.2.1 on 2024-02-28 09:03
+
+import django.db.models.deletion
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("prompt_profile_manager", "0006_alter_profilemanager_x2text"),
+        ("prompt_studio_core", "0004_customtool_summarize_as_source_and_more"),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name="customtool",
+            name="default_profile",
+            field=models.ForeignKey(
+                blank=True,
+                db_comment="Default LLM Profile used in prompt",
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                related_name="default_profile",
+                to="prompt_profile_manager.profilemanager",
+            ),
+        ),
+        migrations.AlterField(
+            model_name="customtool",
+            name="summarize_as_source",
+            field=models.BooleanField(
+                db_comment="Flag to use summarized content as source",
+                default=True,
+            ),
+        ),
+        migrations.AlterField(
+            model_name="customtool",
+            name="summarize_context",
+            field=models.BooleanField(
+                db_comment="Flag to summarize content", default=True
+            ),
+        ),
+        migrations.AlterField(
+            model_name="customtool",
+            name="summarize_llm_profile",
+            field=models.ForeignKey(
+                blank=True,
+                db_comment="LLM Profile used for summarize",
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                related_name="summarize_llm_profile",
+                to="prompt_profile_manager.profilemanager",
+            ),
+        ),
+    ]

backend/prompt_studio/prompt_studio_core/models.py

@@ -44,6 +44,26 @@ class CustomTool(BaseModel):
         related_name="default_profile",
         null=True,
         blank=True,
+        db_comment="Default LLM Profile used in prompt",
+    )
+    summarize_llm_profile = models.ForeignKey(
+        ProfileManager,
+        on_delete=models.SET_NULL,
+        related_name="summarize_llm_profile",
+        null=True,
+        blank=True,
+        db_comment="LLM Profile used for summarize",
+    )
+    summarize_context = models.BooleanField(
+        default=True, db_comment="Flag to summarize content"
+    )
+    summarize_as_source = models.BooleanField(
+        default=True, db_comment="Flag to use summarized content as source"
+    )
+    summarize_prompt = models.TextField(
+        blank=True,
+        db_comment="Field to store the summarize prompt",
+        unique=False,
     )
     prompt_grammer = models.JSONField(
         null=True, blank=True, db_comment="Synonymous words used in prompt"