More cleaning

Author: lmfarley10

.gitignore

@@ -57,18 +57,24 @@ __pycache__/
 *.webp
 *.ico
 
+*.docx 
+
 # Config
 *.ini
 
 # Binaries
 *.xlsx
 *.zip
+/rest/nl2sql-trust/.coverage
+
+# WebSDK should be included in VBCS Applications
+/oda/websdk-apache-deploy/
 
 # scan file
 .scannerwork/
 
 # Files with possible sensitive info 
-clientApp/ConfigFile.properties
+#clientApp/ConfigFile.properties # cleaned sensitive info
 clientApp/autoprompt/ConfigFile.properties
 clientApp/autoprompt/semantic_suggest_runbook.md 
 
@@ -129,9 +135,9 @@ mkdocs/docs/deployment/VBCS.md
 /promotion/
 
 # Random IPs 
-/rest/nl2sql-trust/conf/dev_trust_config.json
+/rest/nl2sql-trust/conf/dev_trust_config.json 
 /rest/nl2sql-trust/conf/metadata_v2.json
-/rest/nl2sql-trust/conf/trust_config.json
+#/rest/nl2sql-trust/conf/trust_config.json # added placeholders for sensitive info
 
 # Sensitive info 
 /rest/nl2sql-trust/helpers/config_json_helper.py
@@ -145,7 +151,7 @@ mkdocs/docs/deployment/VBCS.md
 
 # Reference to OCIDs
 /rest/nl2sql-trust/helpers/llm_helper.py
-/rest/nl2sql-trust/helpers/oci_helper_boostrap.py
+#/rest/nl2sql-trust/helpers/oci_helper_boostrap.py # cleaned up sensitive info 
 /rest/nl2sql-trust/helpers/oci_helper_json.py
 
 /rest/nl2sql-trust/helpers/test_ca_creation.json 
@@ -164,8 +170,8 @@ mkdocs/docs/deployment/VBCS.md
 
 /rest/wallet/nl2sql-test
 
-# example ips in comments 
-/rest/nl2sql-trust/nl2sql_service.py
+# Replaced given ips with placeholders
+#/rest/nl2sql-trust/nl2sql_service.py
 
 # Reference to internal db 
 /rest/nl2sql-trust/README
@@ -174,7 +180,7 @@ mkdocs/docs/deployment/VBCS.md
 /rest/script/devops
 
 # docker artifactory info 
-/rest/script/docker/artifacts 
+/rest/script/docker/artifacts/
 
 # oracle instant client reference?
 /rest/script/docker/oracle-instantclient-23ai-ol8.repo
@@ -243,7 +249,7 @@ terraform/example/vm_trust_cloud_init/terraform.tfvars
 terraform/NLSQL_v1
 
 # Secrets and binaries 
-terraform/NLSQL_V2/Deployment/config
+terraform/NLSQL_V2/Deployment/config/files
 
 # ocids 
 terraform/NLSQL_V2/Deployment/modules/instances/gpu/main.tf
@@ -252,8 +258,11 @@ terraform/NLSQL_V2/Deployment/modules/instances/gpu/main.tf
 # empty dir 
 terraform/NLSQL_V2/Deployment/scripts/
 
-# are these archives necessary? 
-vbcs_oda_archives
+# VBCS Applications 
+# cleaned up main vbcs-askdata app. Will need to provide instructions for certain inputs e.g. idcs endpoint, api gateway endpoints, etc.
+# The app is dependent on the images for the app. We can't include the images in the repo. Maybe we'd want to create an associated livelab for this section?
+# Still need to bloat and clean zip files for dependent VBCS apps
+# vbcs_oda_archives 
 
 # wallet 
 /wallet

clientApp/ConfigFile.properties

@@ -0,0 +1,70 @@
+[Logging]
+#DEBUG, INFO, WARNING, ERROR, CRITICAL
+file.level=DEBUG
+console.level=DEBUG
+logs.path=./logs
+[FeatureFlags]
+feature.explain=false
+feature.intent=false
+feature.llmgraphcheck=false
+[security]
+anonymous.flag=true
+[DatabaseSection]
+#database.user=ttech
+database.user=nl2sql_app_data
+database.password=<client-database-pw>
+database.dsn=<client-database-dsn>
+database.config=<client-db-path>
+database.walletpsswd=<client-db-wallet-pw>
+database.debug=Y
+database.rbac=N
+[KeySection]
+key.llm=notused
+key.searchapi=notused
+token=notused
+[RedisSection]
+redis.url=<rediss://xyz.redis.us-chicago-1.oci.oraclecloud.com>:6379
+url=<xyz>.redis.us-chicago-1.oci.oraclecloud.com
+port=6379
+[GenAISQLGenerator]
+sql.dialect=Oracle
+#Scenario3
+[QueryResult]
+max.resultset=50
+filter.upn=notused
+filter.ignoreupntablelist=notused
+[vbcs]
+endpoint.url=<vbcs-endpoint-url>
+graph_app.url=<graph-app-url>
+idata_app.url=<table-graph-url>
+[OCI]
+#Sao Paulo
+#serviceendpoint.url=https://inference.generativeai.sa-saopaulo-1.oci.oraclecloud.com
+#Scenario2
+# DAC or DS or GAI
+serviceendpoint.active=DS
+serviceendpoint.ds_endpt=<service-endpoint-url>
+serviceendpoint.ds_model=<service-endpoint-ds-model>
+serviceendpoint.url=https://inference.generativeai.us-chicago-1.oci.oraclecloud.com
+serviceendpoint.ocid=<service-endpoint-ocid>
+serviceendpoint.model=<service-endpoint-model>
+serviceendpoint.dac_url=https://inference.generativeai.us-chicago-1.oci.oraclecloud.com
+serviceendpoint.dac_ocid=<comp-ocid>
+serviceendpoint.dac_endpt=<dac-endpoint-ocid>
+serviceendpoint.model_embed=cohere.embed-english-v3.0
+serviceendpoint.llm_name=notused
+#serviceendpoint.llm_name=OCI-GenAI-LLAMA405bB
+[METADATA]
+basepath=./metadata
+default=metadata.sql
+schema.ddl=metadata.sql
+file.embdgs=notused
+file.col_embdgs=notused
+librarymatch.threshold=0.80
+librarymatch.upperthreshold=1.0
+[DEFAULT] # trust db connection
+user=xxxxxx
+password=xxxxx
+dsn= xxxxx
+wallet_location=xxxx
+wallet_password=xxxxx

rest/nl2sql-trust/conf/trust_config.json

@@ -0,0 +1,72 @@
+{
+    "app": {
+        "db_datetime_format_nospace": "DDMONYYYY_hh24miss",
+        "datetime_format_nospace": "%d%b%Y_%H%M%S"
+    },
+    "oci": {
+        "tenancy_ocid": "<tenancy-ocid>",
+        "compartment_ocid": "<comp-ocid>",
+        "namespace": "<tenancy-namespace>",
+        "region": "us-chicago-1",
+        "ai": {
+            "url": "https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
+            "model_embed": "cohere.embed-english-v3.0",
+            "dac_endpoint": "https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
+            "dac_endpoint_model": "<dac-endpoint-ocid>"
+        },
+        "engine": {
+            "get_sql_url": "http://<your-ip>:8002/getsql",
+            "llm_prompt_url": "http://<your-ip>:8002/getprompt"
+        },
+        "vault": {
+            "db_secret_ocid": "<db-vault-secret-ocid>",
+            "region": "<your-vault-region>"
+        },
+        "finetune": {
+            "max_regression": 100,
+            "max_corrected": 100,
+            "keep_local_jsonl": true,
+            "base_model_id": "<base-model-ocid>",
+            "unit_count": 2,
+            "unit_shape": "LARGE_GENERIC",
+            "dataset_type": "OBJECT_STORAGE",
+            "training_dataset": {
+                "early_stopping_patience": 6,
+                "early_stopping_threshold": 0.01,
+                "learning_rate": 5e-05,
+                "log_model_metrics_interval_in_steps": 10,
+                "lora_alpha": 32,
+                "lora_dropout": 0.1,
+                "lora_r": 8,
+                "total_training_epochs": 3,
+                "training_batch_size": 8,
+                "training_config_type": "LORA_TRAINING_CONFIG"
+            }
+        },
+        "os": {
+            "bucket": "<your-bucket-name>",
+            "finetune": {
+                "os_path": "finetune/jsonl/",
+                "local_path": "./upload"
+            },
+            "metadata": {
+                "file_name": "metadata_v2.json",
+                "os_path": "config/",
+                "local_path": "./conf"
+            }
+        }
+    },
+    "database": {
+        "wallet": "wallet/nl2sql-test",
+        "os_wallet": "wallet/nl2sqltest_wallet.zip",
+        "user": "admin",
+        "pwd": "<password>",
+        "dns": "<dns>",
+        "datetime_format": "DD-MON-YYYY hh24:mi:ss",
+        "date_format": "DD-MON-YYYY"
+    },
+    "metrics": {
+        "start_date": "10-FEB-2025",
+        "python_format": "%d-%b-%Y"
+    }
+}