First step in making gpt-engineer learn. Rename main_prompt -> prompt (#381)

* First step in collecting learnings

* Rename prompts

* remove requirements, use pip install -e . instead

* Add requirements

* Fix tests

gpt_engineer/collect.py

@@ -0,0 +1,85 @@
+import hashlib
+import json
+import os
+import random
+import tempfile
+
+from dataclasses import dataclass
+from pathlib import Path
+
+from dataclasses_json import dataclass_json
+
+from gpt_engineer import steps
+from gpt_engineer.db import DBs
+from gpt_engineer.steps import Step
+
+
+@dataclass_json
+@dataclass
+class Learning:
+    model: str
+    temperature: float
+    steps: str
+    steps_file_hash: str
+    prompt: str
+    feedback: str | None
+    session: str
+    version: str = "0.1"
+
+
+def steps_file_hash():
+    with open(steps.__file__, "r") as f:
+        content = f.read()
+        return hashlib.sha256(content.encode("utf-8"), usedforsecurity=False).hexdigest()
+
+
+def extract_learning(
+    model: str, temperature: float, steps: list[Step], dbs: DBs
+) -> Learning:
+    learning = Learning(
+        prompt=dbs.input["prompt"],
+        model=model,
+        temperature=temperature,
+        steps=json.dumps([step.__name__ for step in steps]),
+        steps_file_hash=steps_file_hash(),
+        feedback=dbs.input.get("feedback"),
+        session=get_session(),
+    )
+    return learning
+
+
+def send_learnings(learning: Learning):
+    import rudderstack.analytics as rudder_analytics
+
+    rudder_analytics.write_key = "2Re4kqwL61GDp7S8ewe6K5dbogG"
+    rudder_analytics.dataPlaneUrl = "https://gptengineerezm.dataplane.rudderstack.com"
+
+    rudder_analytics.track(
+        user_id=learning.session,
+        event="learning",
+        properties=learning.to_dict(),  # type: ignore
+    )
+
+
+def get_session():
+    path = Path(tempfile.gettempdir()) / "gpt_engineer_user_id.txt"
+
+    try:
+        if path.exists():
+            user_id = path.read_text()
+        else:
+            # random uuid:
+            user_id = str(random.randint(0, 2**32))
+            path.write_text(user_id)
+        return user_id
+    except IOError:
+        return "ephemeral_" + str(random.randint(0, 2**32))
+
+
+def collect_learnings(model: str, temperature: float, steps: list[Step], dbs: DBs):
+    if os.environ.get("COLLECT_LEARNINGS_OPT_OUT") in ["true", "1"]:
+        print("COLLECT_LEARNINGS_OPT_OUT is set to true, not collecting learning")
+        return
+
+    learnings = extract_learning(model, temperature, steps, dbs)
+    send_learnings(learnings)

gpt_engineer/db.py

@@ -22,6 +22,12 @@ class DB:
         with full_path.open("r", encoding="utf-8") as f:
             return f.read()
 
+    def get(self, key, default=None):
+        try:
+            return self[key]
+        except KeyError:
+            return default
+
     def __setitem__(self, key, val):
         full_path = self.path / key
         full_path.parent.mkdir(parents=True, exist_ok=True)

gpt_engineer/main.py

@@ -8,6 +8,7 @@ import typer
 
 from gpt_engineer import steps
 from gpt_engineer.ai import AI
+from gpt_engineer.collect import collect_learnings
 from gpt_engineer.db import DB, DBs
 from gpt_engineer.steps import STEPS
 
@@ -56,10 +57,13 @@ def main(
         preprompts=DB(Path(__file__).parent / "preprompts"),
     )
 
-    for step in STEPS[steps_config]:
+    steps = STEPS[steps_config]
+    for step in steps:
         messages = step(ai, dbs)
         dbs.logs[step.__name__] = json.dumps(messages)
 
+    collect_learnings(model, temperature, steps, dbs)
+
 
 if __name__ == "__main__":
     app()

gpt_engineer/steps.py

@@ -5,6 +5,8 @@ import subprocess
 from enum import Enum
 from typing import Callable, List, TypeVar
 
+from termcolor import colored
+
 from gpt_engineer.ai import AI
 from gpt_engineer.chat_to_files import to_files
 from gpt_engineer.db import DBs
@@ -19,12 +21,24 @@ def setup_sys_prompt(dbs):
 Step = TypeVar("Step", bound=Callable[[AI, DBs], List[dict]])
 
 
+def get_prompt(dbs):
+    """While we migrate we have this fallback getter"""
+    assert (
+        "prompt" in dbs.input or "main_prompt" in dbs.input
+    ), "Please put your prompt in the file `prompt` in the project directory"
+
+    if "prompt" not in dbs.input:
+        print(
+            colored("Please put the prompt in the file `prompt`, not `main_prompt", "red")
+        )
+        print()
+
+    return dbs.input.get("prompt", dbs.input["main_prompt"])
+
+
 def simple_gen(ai: AI, dbs: DBs):
     """Run the AI on the main prompt and save the results"""
-    messages = ai.start(
-        setup_sys_prompt(dbs),
-        dbs.input["main_prompt"],
-    )
+    messages = ai.start(setup_sys_prompt(dbs), get_prompt(dbs))
     to_files(messages[-1]["content"], dbs.workspace)
     return messages
 
@@ -34,22 +48,31 @@ def clarify(ai: AI, dbs: DBs):
     Ask the user if they want to clarify anything and save the results to the workspace
     """
     messages = [ai.fsystem(dbs.preprompts["qa"])]
-    user = dbs.input["main_prompt"]
+    user_input = get_prompt(dbs)
     while True:
-        messages = ai.next(messages, user)
+        messages = ai.next(messages, user_input)
 
         if messages[-1]["content"].strip().lower().startswith("no"):
-            print(" Nothing more to clarify.")
+            print("Nothing more to clarify.")
             break
 
         print()
-        user = input('(answer in text, or "c" to move on)\n')
+        user_input = input('(answer in text, or "c" to move on)\n')
         print()
 
-        if not user or user == "c":
-            break
+        if not user_input or user_input == "c":
+            print("(letting gpt-engineer make its own assumptions)")
+            print()
+            messages = ai.next(
+                messages,
+                ai.fuser(
+                    "Make your own assumptions and state them explicitly before starting"
+                ),
+            )
+            print()
+            return messages
 
-        user += (
+        user_input += (
             "\n\n"
             "Is anything else unclear? If yes, only answer in the form:\n"
             "{remaining unclear areas} remaining questions.\n"
@@ -68,7 +91,7 @@ def gen_spec(ai: AI, dbs: DBs):
     """
     messages = [
         ai.fsystem(setup_sys_prompt(dbs)),
-        ai.fsystem(f"Instructions: {dbs.input['main_prompt']}"),
+        ai.fsystem(f"Instructions: {dbs.input['prompt']}"),
     ]
 
     messages = ai.next(messages, dbs.preprompts["spec"])
@@ -105,7 +128,7 @@ def gen_unit_tests(ai: AI, dbs: DBs):
     """
     messages = [
         ai.fsystem(setup_sys_prompt(dbs)),
-        ai.fuser(f"Instructions: {dbs.input['main_prompt']}"),
+        ai.fuser(f"Instructions: {dbs.input['prompt']}"),
         ai.fuser(f"Specification:\n\n{dbs.memory['specification']}"),
     ]
 
@@ -136,7 +159,7 @@ def gen_code(ai: AI, dbs: DBs):
 
     messages = [
         ai.fsystem(setup_sys_prompt(dbs)),
-        ai.fuser(f"Instructions: {dbs.input['main_prompt']}"),
+        ai.fuser(f"Instructions: {dbs.input['prompt']}"),
         ai.fuser(f"Specification:\n\n{dbs.memory['specification']}"),
         ai.fuser(f"Unit tests:\n\n{dbs.memory['unit_tests']}"),
     ]
@@ -200,7 +223,7 @@ def gen_entrypoint(ai, dbs):
 def use_feedback(ai: AI, dbs: DBs):
     messages = [
         ai.fsystem(setup_sys_prompt(dbs)),
-        ai.fuser(f"Instructions: {dbs.input['main_prompt']}"),
+        ai.fuser(f"Instructions: {dbs.input['prompt']}"),
         ai.fassistant(dbs.workspace["all_output.txt"]),
         ai.fsystem(dbs.preprompts["use_feedback"]),
     ]
@@ -213,7 +236,7 @@ def fix_code(ai: AI, dbs: DBs):
     code_output = json.loads(dbs.logs[gen_code.__name__])[-1]["content"]
     messages = [
         ai.fsystem(setup_sys_prompt(dbs)),
-        ai.fuser(f"Instructions: {dbs.input['main_prompt']}"),
+        ai.fuser(f"Instructions: {dbs.input['prompt']}"),
         ai.fuser(code_output),
         ai.fsystem(dbs.preprompts["fix_code"]),
     ]