Fix issue where setting baml_log would be overwritten (#1732)

<!-- ELLIPSIS_HIDDEN -->


> [!IMPORTANT]
> Fixes issue with `baml_log` being overwritten by ensuring correct
environment variable handling and updates logging initialization and
tests.
> 
>   - **Behavior**:
> - Fix issue where `baml_log` setting was being overwritten by ensuring
it is set correctly from environment variables in
`baml-runtime/src/lib.rs` and `language_client_python/src/lib.rs`.
> - Remove default setting of `BAML_LOG` to "info" in
`baml_py/__init__.py`.
>   - **Logging**:
> - Add `init_debug_logger()` function in
`language_client_python/src/lib.rs` to initialize a debug logger with
`BAML_INTERNAL_LOG`.
> - Update tests in `test_logger.py` to verify logger initialization and
behavior with different log levels.
>   - **Misc**:
> - Remove unnecessary `load_dotenv()` calls from various test files.
> - Add `test-dotenv` file to set `BAML_LOG=warn` for testing purposes.
> 
> <sup>This description was created by </sup>[<img alt="Ellipsis"
src="https://img.shields.io/badge/Ellipsis-blue?color=175173">](https://www.ellipsis.dev?ref=BoundaryML%2Fbaml&utm_source=github&utm_medium=referral)<sup>
for 3fbf668. It will automatically
update as commits are pushed.</sup>


<!-- ELLIPSIS_HIDDEN -->

Author: aaronvg

engine/baml-runtime/src/lib.rs

@@ -180,7 +180,6 @@ impl BamlRuntime {
             .map(|(k, v)| (k.as_ref().to_string(), v.as_ref().to_string()))
             .collect();
         baml_log::set_from_env(&copy)?;
-
         Ok(BamlRuntime {
             inner: InternalBamlRuntime::from_file_content(root_path, files)?,
             tracer: BamlTracer::new(None, env_vars.into_iter())?.into(),

engine/language_client_codegen/src/python/templates/__init__.py.j2

@@ -25,7 +25,6 @@ with EnsureBamlPyImport(__version__) as e:
   from . import partial_types
   from . import config
   from .config import reset_baml_env_vars
-
   {% if default_client_mode == GeneratorDefaultClientMode::Async %}
   from .async_client import b
   {% else %}
@@ -38,5 +37,6 @@ __all__ = [
   "tracing",
   "types",
   "reset_baml_env_vars",
+  "config",
   "logging",
 ]

engine/language_client_python/python_src/baml_py/__init__.py

@@ -1,10 +1,5 @@
 # BAML Python API: provides the Python API for the BAML runtime.
 
-if __name__ == "baml_py":
-    import os
-
-    if "BAML_LOG" not in os.environ:
-        os.environ["BAML_LOG"] = "info"
 
 # Re-export the pyo3 API
 from .baml_py import (

engine/language_client_python/src/lib.rs

@@ -86,6 +86,15 @@ fn baml_py(m: Bound<'_, PyModule>) -> PyResult<()> {
 
     // Initialize the logger
     baml_log::init().map_err(errors::BamlError::from_anyhow)?;
-
+    init_debug_logger();
     Ok(())
 }
+
+fn init_debug_logger() {
+    // Regular formatting
+    if let Err(e) =
+        env_logger::try_init_from_env(env_logger::Env::new().filter("BAML_INTERNAL_LOG"))
+    {
+        println!("Failed to initialize BAML DEBUG logger: {:#}", e);
+    }
+}

integ-tests/python/baml_client/__init__.py

@@ -1,79 +1,78 @@
 import time
 import threading
 import asyncio
-from dotenv import load_dotenv
 from http.server import BaseHTTPRequestHandler, HTTPServer
 import json
 import os
 from baml_client.tracing import trace, flush
 
-load_dotenv()
-os.environ['BOUNDARY_BASE_URL'] = 'http://localhost:4040'
+os.environ["BOUNDARY_BASE_URL"] = "http://localhost:4040"
 
-class TraceRequestHandler(BaseHTTPRequestHandler):
 
+class TraceRequestHandler(BaseHTTPRequestHandler):
     @staticmethod
     def run_forever():
-      address = ('', int(os.environ['BOUNDARY_BASE_URL'].rsplit(':', maxsplit=1)[1]))
-      httpd = HTTPServer(address, TraceRequestHandler)
-      print(f'Starting BAML event logger on {address}')
-      httpd.serve_forever()
-
+        address = ("", int(os.environ["BOUNDARY_BASE_URL"].rsplit(":", maxsplit=1)[1]))
+        httpd = HTTPServer(address, TraceRequestHandler)
+        print(f"Starting BAML event logger on {address}")
+        httpd.serve_forever()
 
     def do_GET(self):
-      print(f"Received GET request: {self.path}")
-      # self.send_response(200)
-      # self.send_header('Content-type', 'text/html')
-      # #self.send_header('Content-type', 'application/json')
-      # self.end_headers()
-      # self.wfile.write('Hello, BAML!'.encode('utf-8'))
-      self.send_error(404, "File not found")
+        print(f"Received GET request: {self.path}")
+        # self.send_response(200)
+        # self.send_header('Content-type', 'text/html')
+        # #self.send_header('Content-type', 'application/json')
+        # self.end_headers()
+        # self.wfile.write('Hello, BAML!'.encode('utf-8'))
+        self.send_error(404, "File not found")
 
     def do_POST(self):
-      print(f"Received POST request: {self.path}")
-      self.send_error(404, "File not found")
-      return
-      if self.path == '/log/v2':
-          content_length = int(self.headers['Content-Length'])  # Get the size of data
-          post_data = self.rfile.read(content_length)  # Read the data
-          data = json.loads(post_data.decode('utf-8'))  # Decode it to string
-
-          print(f"Received event log for {data}")
-          print(json.dumps(data, indent=2))
-
-          self.send_response(200)
-          self.send_header('Content-type', 'application/json')
-          self.end_headers()
-          response = json.dumps({'message': 'Log received'})
-          self.wfile.write(response.encode('utf-8'))  # Send response back to client
-      else:
-          self.send_error(404, "File not found")
+        print(f"Received POST request: {self.path}")
+        self.send_error(404, "File not found")
+        return
+        if self.path == "/log/v2":
+            content_length = int(self.headers["Content-Length"])  # Get the size of data
+            post_data = self.rfile.read(content_length)  # Read the data
+            data = json.loads(post_data.decode("utf-8"))  # Decode it to string
+
+            print(f"Received event log for {data}")
+            print(json.dumps(data, indent=2))
+
+            self.send_response(200)
+            self.send_header("Content-type", "application/json")
+            self.end_headers()
+            response = json.dumps({"message": "Log received"})
+            self.wfile.write(response.encode("utf-8"))  # Send response back to client
+        else:
+            self.send_error(404, "File not found")
+
 
 async def main():
-  print('Hello, BAML!')
-  #TraceRequestHandler.run_forever()
-  t = threading.Thread(target=TraceRequestHandler.run_forever, daemon=True)
-  t.start()
-
-  print('waiting for server to start')
-  time.sleep(5)
-  print('server has started ( i think )')
-
-  @trace
-  def sync_dummy_fn():
-    time.sleep(.05)
-
-  sync_dummy_fn()
-
-  print('before flush')
-  t_flush = threading.Thread(target=flush)
-  print('wtf?')
-  t_flush.start()
-  print('after start before join')
-  t_flush.join(timeout=5)
-  print('after flush')
-  
-  assert False
-
-if __name__ == '__main__':
-  asyncio.run(main())
+    print("Hello, BAML!")
+    # TraceRequestHandler.run_forever()
+    t = threading.Thread(target=TraceRequestHandler.run_forever, daemon=True)
+    t.start()
+
+    print("waiting for server to start")
+    time.sleep(5)
+    print("server has started ( i think )")
+
+    @trace
+    def sync_dummy_fn():
+        time.sleep(0.05)
+
+    sync_dummy_fn()
+
+    print("before flush")
+    t_flush = threading.Thread(target=flush)
+    print("wtf?")
+    t_flush.start()
+    print("after start before join")
+    t_flush.join(timeout=5)
+    print("after flush")
+
+    assert False
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

integ-tests/python/baml_example_tracing.py

@@ -0,0 +1,3 @@
+# assume tests run from this directory.
+# Dont change this. It will get picked up by test_logger.py
+BAML_LOG=warn

integ-tests/python/test-dotenv

@@ -1,6 +1,5 @@
 from baml_py.errors import BamlInvalidArgumentError, BamlError, BamlClientError
 import pytest
-import dotenv
 from openai.types.chat import ChatCompletion
 
 from ..baml_client import b
@@ -10,8 +9,6 @@
 import sys
 import asyncio
 
-dotenv.load_dotenv()
-
 
 def function_span_count():
     return Collector.__function_span_count()  # type: ignore