tinybird 0.0.1.dev217__tar.gz → 0.0.1.dev221__tar.gz

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: tinybird
-Version: 0.0.1.dev217
+Version: 0.0.1.dev221
 Summary: Tinybird Command Line Tool
 Home-page: https://www.tinybird.co/docs/forward/commands
 Author: Tinybird

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/tinybird/ch_utils/constants.py

@@ -242,3 +242,16 @@ FORBIDDEN_SQL_KEYWORDS = {
 
 CH_SETTINGS_JOIN_ALGORITHM_HASH = "hash"  # uses 'hash' by default, https://clickhouse.com/docs/en/operations/settings/settings/#settings-join_algorithm
 CH_SETTINGS_JOIN_ALGORITHM_AUTO = "auto,hash"
+
+VALID_QUERY_FORMATS = (
+    "JSON",
+    "CSV",
+    "CSVWithNames",
+    "TSV",
+    "TSVWithNames",
+    "PrettyCompact",
+    "JSONEachRow",
+    "Parquet",
+    "JSONStrings",
+    "Prometheus",
+)

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/tinybird/datafile/common.py

@@ -1710,6 +1710,8 @@ def parse(
             "s3_access_key": assign_var("s3_access_key"),
             "s3_secret": assign_var("s3_secret"),
             "gcs_service_account_credentials_json": assign_var_json("gcs_service_account_credentials_json"),
+            "gcs_hmac_access_id": assign_var("gcs_hmac_access_id"),
+            "gcs_hmac_secret": assign_var("gcs_hmac_secret"),
             "include": include,
         },
     }

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/tinybird/prompts.py

@@ -1,3 +1,4 @@
+from datetime import date
 from typing import Optional
 
 general_functions = [
@@ -453,6 +454,8 @@ Use the following format to generate the response and do not wrap it in any othe
 
 
 def mock_prompt(rows: int, feedback: str = "") -> str:
+    today = date.today().isoformat()
+
     if feedback:
         feedback = f"""In case the <feedback> tag is present and not empty,
 it means there was a previous attempt to generate the resources and the system provided feedback about the previous response.
@@ -495,6 +498,10 @@ FROM numbers({rows})
 - The query MUST return a sample of EXACTLY {rows} rows.
 - The query MUST be valid for clickhouse and Tinybird.
 - FROM numbers({rows}) part is mandatory.
+- If json paths are present (e.g. `userAgent` String `json:$.userAgent`), rely on the json paths to generate the sample record.
+- If the schema has nested json paths (e.g. `json:$.location.country`), generate nested JSON objects accordingly.
+- When working with dates, take into account that the current date is {today}.
+- Use recent dates to avoid generating dates that are too far in the past.
 - Do NOT include ```clickhouse or ```sql or any other wrapping text to the sql query.
 - Do NOT use any of these functions: elementAt
 - Do NOT add a semicolon at the end of the query
@@ -539,58 +546,55 @@ SELECT
   arrayMap(x -> concat('item_', toString(x)), range(1, rand() % 5 + 1)) AS items
 FROM numbers(ROWS)
 
-## Example schema with a nested JSON field:
+## Example schema with nested JSON paths:
 
 ### Schema:
 
 SCHEMA >
-    `request_id` String `json:$.request_id`,
     `timestamp` DateTime `json:$.timestamp`,
-    `model` String `json:$.request.model`,
-    `temperature` Float32 `json:$.request.options.temperature`,
-    `max_tokens` UInt32 `json:$.request.options.max_tokens`,
-    `stream` UInt8 `json:$.request.options.stream`
-
-### Desired final output of the query:
-
-Note that the important part is generating the nested fields:
-json:$.request.options.max_tokens > this means that the max_tokens field is nested inside the options field inside the request field.
-
+    `location_country` String `json:$.location.country`,
+    `location_region` String `json:$.location.region`,
+    `location_city` String `json:$.location.city`,
+    `location_latitude` String `json:$.location.latitude`,
+    `location_longitude` String `json:$.location.longitude`
+
+### Important: Understanding JSON paths
+When you see json paths like `json:$.location.country`, it means the data should be structured as nested JSON:
+- `json:$.location.country` → the country field is nested inside the location object
+- `json:$.location.region` → the region field is nested inside the location object
+
+### Desired final output structure:
 {{
-  "request_id": "req_abc123",
   "timestamp": "2024-11-30T10:30:00.000Z",
-  "request": {{
-    "model": "gpt-4",
-    "options": {{
-      "temperature": 0.7,
-      "max_tokens": 1000,
-      "stream": false
-    }}
+  "location": {{
+    "country": "United States",
+    "region": "California", 
+    "city": "San Francisco",
+    "latitude": "37.7749",
+    "longitude": "-122.4194"
   }}
 }}
 
-### Example SQL output with nested fields:
+### Example SQL output for nested JSON paths:
 
 SELECT
-    request_id,
     timestamp,
     CAST(concat('{{
-        "model": "', model, '",
-        "options": {{
-            "temperature": ', temperature, ',
-            "max_tokens": ', max_tokens, ',
-            "stream": ', IF(stream = 1, 'true', 'false'), '
-        }}
-    }}'), 'JSON') AS request
+        "country": "', country, '",
+        "region": "', region, '",
+        "city": "', city, '",
+        "latitude": "', latitude, '",
+        "longitude": "', longitude, '"
+    }}'), 'JSON') AS location
 FROM
 (
     SELECT
-        concat('req_', lower(hex(randomString(6)))) AS request_id,
-        (now() - toIntervalDay(rand() % 30)) + toIntervalSecond(rand() % 86400) AS timestamp,
-        ['gpt-4', 'gpt-3.5-turbo', 'gpt-4-turbo'][(rand() % 3) + 1] AS model,
-        round(rand() / 10, 2) AS temperature,
-        500 + (rand() % 2500) AS max_tokens,
-        rand() % 2 AS stream
+        now() - rand() % 86400 AS timestamp,
+        ['United States', 'Canada', 'United Kingdom', 'Germany', 'France'][(rand() % 5) + 1] AS country,
+        ['California', 'Texas', 'New York', 'Ontario', 'London'][(rand() % 5) + 1] AS region,
+        ['San Francisco', 'Los Angeles', 'New York', 'Toronto', 'London'][(rand() % 5) + 1] AS city,
+        toString(round(rand() * 180 - 90, 4)) AS latitude,
+        toString(round(rand() * 360 - 180, 4)) AS longitude
     FROM numbers(ROWS)
 )
 </more_examples>
@@ -811,14 +815,17 @@ sql_instructions = """
     - Do NOT use CREATE TABLE, INSERT INTO, CREATE DATABASE, etc.
     - Use ONLY SELECT statements in the SQL section.
     - INSERT INTO is not supported in SQL section.
-    - General functions supported are: {general_functions}
-    - Character insensitive functions supported are: {general_functions_insensitive}
-    - Aggregate functions supported are: {aggregate_functions}
-    - Do not use any function that is not present in the list of general functions, character insensitive functions and aggregate functions.
-    - If the function is not present in the list, the sql query will fail, so avoid at all costs to use any function that is not present in the list.
-    - When aliasing a column, use first the column name and then the alias.
-    - General functions and aggregate functions are case sensitive.
-    - Character insensitive functions are case insensitive.
+    - ClickHouse functions supported are:
+        - General functions supported are: {general_functions}
+        - Character insensitive functions supported are: {general_functions_insensitive}
+        - Aggregate functions supported are: {aggregate_functions}
+    - How to use ClickHouse supported functions:
+        - When using functions try always ClickHouse functions first, then SQL functions.
+        - Do not use any ClickHouse function that is not present in the list of general functions, character insensitive functions and aggregate functions.
+        - If the function is not present in the list, the sql query will fail, so avoid at all costs to use any function that is not present in the list.
+        - When aliasing a column, use first the column name and then the alias.
+        - General functions and aggregate functions are case sensitive.
+        - Character insensitive functions are case insensitive.
     - Parameters are never quoted in any case.
     - Use the following syntax in the SQL section for the iceberg table function: iceberg('s3://bucket/path/to/table', {{{{tb_secret('aws_access_key_id')}}}}, {{{{tb_secret('aws_secret_access_key')}}}})
     - Use the following syntax in the SQL section for the postgres table function: postgresql('host:port', 'database', 'table', {{{{tb_secret('db_username')}}}}, {{{{tb_secret('db_password')}}}}), 'schema')

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/tinybird/tb/__cli__.py

@@ -4,5 +4,5 @@ __description__ = 'Tinybird Command Line Tool'
 __url__ = 'https://www.tinybird.co/docs/forward/commands'
 __author__ = 'Tinybird'
 __author_email__ = 'support@tinybird.co'
-__version__ = '0.0.1.dev217'
-__revision__ = '8183690'
+__version__ = '0.0.1.dev221'
+__revision__ = '8f74332'

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/tinybird/tb/modules/create.py

@@ -101,7 +101,7 @@ async def create(
         if prompt:
             prompt_result = await create_resources_from_prompt(tb_client, user_token, prompt, project)
             result.extend(prompt_result)
-            readme_path = Path(root_folder) / "README.md"
+            readme_path = folder_path / "README.md"
             if readme_path.exists():
                 click.echo(FeedbackManager.highlight(message="\n» Updating project description..."))
             else:

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/tinybird/tb/modules/workspace.py

@@ -288,14 +288,25 @@ async def clear_workspace() -> None:
 
     ws = next((ws for ws in local_workspaces if ws["name"] == ws_name), None)
 
+    if not ws:
+        raise CLIWorkspaceException(FeedbackManager.error(message=f"Workspace '{ws_name}' not found."))
+
+    requests.delete(f"{TB_LOCAL_ADDRESS}/v1/workspaces/{ws['id']}?token={user_token}&hard_delete_confirmation=yes")  # noqa: ASYNC210
+    user_workspaces = await user_client.user_workspaces(version="v1")
+    ws = next((ws for ws in user_workspaces["workspaces"] if ws["name"] == ws_name), None)
+
     if ws:
-        requests.delete(f"{TB_LOCAL_ADDRESS}/v1/workspaces/{ws['id']}?token={user_token}&hard_delete_confirmation=yes")  # noqa: ASYNC210
-        ws = None
+        raise CLIWorkspaceException(
+            FeedbackManager.error(message=f"Workspace '{ws_name}' was not cleared properly. Please try again.")
+        )
+
+    await user_client.create_workspace(ws_name, assign_to_organization_id=user_org_id, version="v1")
+    user_workspaces = requests.get(f"{TB_LOCAL_ADDRESS}/v1/user/workspaces?token={admin_token}").json()  # noqa: ASYNC210
+    ws = next((ws for ws in user_workspaces["workspaces"] if ws["name"] == ws_name), None)
 
     if not ws:
-        await user_client.create_workspace(ws_name, assign_to_organization_id=user_org_id, version="v1")
-        user_workspaces = requests.get(f"{TB_LOCAL_ADDRESS}/v1/user/workspaces?token={admin_token}").json()  # noqa: ASYNC210
-        ws = next((ws for ws in user_workspaces["workspaces"] if ws["name"] == ws_name), None)
+        raise CLIWorkspaceException(
+            FeedbackManager.error(message=f"Workspace '{ws_name}' was not cleared properly. Please try again.")
+        )
 
-    if ws:
-        click.echo(FeedbackManager.success(message=f"\n✓ Workspace '{ws_name}' cleared"))
+    click.echo(FeedbackManager.success(message=f"✓ Workspace '{ws_name}' cleared"))

{tinybird-0.0.1.dev217 → tinybird-0.0.1.dev221}/tinybird.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: tinybird
-Version: 0.0.1.dev217
+Version: 0.0.1.dev221
 Summary: Tinybird Command Line Tool
 Home-page: https://www.tinybird.co/docs/forward/commands
 Author: Tinybird