parsagon 0.10.20__py3-none-any.whl → 0.10.22__py3-none-any.whl

parsagon/api.py

@@ -135,11 +135,11 @@ def get_bool_about_data(data, question):
     return data["result"]
 
 
-def create_pipeline(name, description, program_sketch, pseudocode):
+def create_pipeline(name, description, program_sketch, pseudocode, secrets):
     return _api_call(
         httpx.post,
         "/pipelines/",
-        json={"name": name, "description": description, "program_sketch": program_sketch, "pseudocode": pseudocode},
+        json={"name": name, "description": description, "program_sketch": program_sketch, "pseudocode": pseudocode, "secrets": secrets},
     )
 
 
@@ -192,11 +192,19 @@ def get_pipeline_code(pipeline_name, variables, headless):
         )
 
 
-def create_pipeline_run(pipeline_id, variables):
+def create_pipeline_run(pipeline_id, variables, is_local):
     return _api_call(
         httpx.post,
         f"/pipelines/{pipeline_id}/runs/",
-        json={"variables": variables},
+        json={"variables": variables, "is_local": is_local},
+    )
+
+
+def update_pipeline_run(run_id, data):
+    return _api_call(
+        httpx.patch,
+        f"/pipelines/runs/{run_id}/",
+        json=data,
     )
 
 

parsagon/exceptions.py

@@ -24,3 +24,8 @@ class ProgramNotFoundException(ParsagonException):
 
     def to_string(self, verbose):
         return f"A program with name {self.program} does not exist."
+
+
+class RunFailedException(ParsagonException):
+    """Raised when a run fails."""
+    pass

parsagon/main.py

@@ -1,9 +1,11 @@
 import argparse
+import datetime
 import json
 import logging
 import logging.config
 import psutil
 import time
+import traceback
 
 from halo import Halo
 from tqdm import tqdm
@@ -15,15 +17,16 @@ from parsagon.api import (
     create_custom_function,
     add_examples_to_custom_function,
     create_pipeline_run,
+    update_pipeline_run,
     get_pipeline,
     get_pipelines,
     get_pipeline_code,
     get_run,
     poll_data,
-    APIException,
 )
-from parsagon.exceptions import ParsagonException
+from parsagon.exceptions import ParsagonException, APIException, RunFailedException
 from parsagon.executor import Executor, custom_functions_to_descriptions
+from parsagon.secrets import extract_secrets
 from parsagon.settings import get_api_key, get_settings, clear_settings, save_setting, get_logging_config
 
 logger = logging.getLogger(__name__)
@@ -139,6 +142,11 @@ def get_args():
         action="store_true",
         help="run the program in the cloud",
     )
+    parser_run.add_argument(
+        "--output_log",
+        action="store_true",
+        help="output log data from the run",
+    )
     parser_run.set_defaults(func=run)
 
     # Delete
@@ -185,12 +193,15 @@ def main():
 
 
 def create(task=None, program_name=None, headless=False, infer=False, verbose=False):
+    configure_logging(verbose)
+
     if task:
         logger.info("Launched with task description:\n%s", task)
     else:
         task = input("Type what you want to do: ")
 
     logger.info("Analyzing task description...")
+    task, secrets = extract_secrets(task)
     program_sketches = get_program_sketches(task)
 
     full_program = program_sketches["full"]
@@ -198,7 +209,8 @@ def create(task=None, program_name=None, headless=False, infer=False, verbose=Fa
     pseudocode = program_sketches["pseudocode"]
     logger.info(f"Created a program based on task description. Program does the following:\n\n{pseudocode}\n\nNow executing the program to identify web elements to be scraped:\n")
     logger.debug("Program:\n%s", abridged_program)
-    abridged_program += "\n\noutput = func()\nprint(f'Program finished and returned a value of:\\n{output}\\n')\n"  # Make the program runnable
+    args = ", ".join(f"{k}={repr(v)}" for k, v in secrets.items())
+    abridged_program += f"\n\noutput = func({args})" + "\nprint(f'Program finished and returned a value of:\\n{output}\\n')\n"  # Make the program runnable
 
     # Execute the abridged program to gather examples
     executor = Executor(headless=headless, infer=infer)
@@ -211,7 +223,7 @@ def create(task=None, program_name=None, headless=False, infer=False, verbose=Fa
         if program_name:
             logger.info(f"Saving program as {program_name}")
             try:
-                pipeline = create_pipeline(program_name, task, full_program, pseudocode)
+                pipeline = create_pipeline(program_name, task, full_program, pseudocode, secrets)
             except APIException as e:
                 if isinstance(e.value, list) and "Pipeline with name already exists" in e.value:
                     logger.info("A program with this name already exists. Please choose another name.")
@@ -243,6 +255,8 @@ def create(task=None, program_name=None, headless=False, infer=False, verbose=Fa
 
 
 def update(program_name, variables={}, headless=False, infer=False, replace=False, verbose=False):
+    configure_logging(verbose)
+
     pipeline = get_pipeline(program_name)
     abridged_program = pipeline["abridged_sketch"]
     # Make the program runnable
@@ -274,8 +288,7 @@ def update(program_name, variables={}, headless=False, infer=False, replace=Fals
             add_examples_to_custom_function(pipeline_id, call_id, custom_function, replace)
         logger.info(f"Saved.")
     except Exception as e:
-        print(e)
-        logger.info(f"An error occurred while saving the program. The program was not updated.")
+        logger.error(f"An error occurred while saving the program. The program was not updated.")
 
 
 def detail(program_name=None, verbose=False):
@@ -289,96 +302,165 @@ def detail(program_name=None, verbose=False):
         )
 
 
-def run(program_name, variables={}, headless=False, remote=False, verbose=False):
+def run(program_name, variables={}, headless=False, remote=False, output_log=False, verbose=False):
     """
     Executes pipeline code
     """
+    configure_logging(verbose)
+
     if headless and remote:
         raise ParsagonException("Cannot run a program remotely in headless mode")
 
+    logger.info("Preparing to run program %s", program_name)
+    pipeline_id = get_pipeline(program_name)["id"]
+
     if remote:
-        pipeline_id = get_pipeline(program_name)["id"]
-        result = create_pipeline_run(pipeline_id, variables)
+        result = create_pipeline_run(pipeline_id, variables, False)
         with Halo(text="Program running remotely...", spinner="dots"):
             while True:
                 run = get_run(result["id"])
                 status = run["status"]
+
+                if output_log and status in ("FINISHED", "ERROR"):
+                    return {k: v for k, v in run.items() if k in ("output", "status", "log", "warnings", "error")}
+
                 if status == "FINISHED":
+                    if verbose:
+                        logger.info(run["log"])
+                        for warning in run["warnings"]:
+                            logger.warning(warning)
                     logger.info("Program finished running.")
                     return run["output"]
                 elif status == "ERROR":
                     raise ParsagonException(f"Program failed to run: {run['error']}")
                 elif status == "CANCELED":
                     raise ParsagonException("Program execution was canceled")
+
                 time.sleep(5)
 
-    logger.info("Preparing to run program %s", program_name)
+    run = create_pipeline_run(pipeline_id, variables, True)
     code = get_pipeline_code(program_name, variables, headless)["code"]
+    start_time = datetime.datetime.now(datetime.timezone.utc).isoformat()
+    run_data = {"start_time": start_time}
 
     logger.info("Running program...")
     globals_locals = {"PARSAGON_API_KEY": get_api_key()}
     try:
         exec(code, globals_locals, globals_locals)
+        run_data["status"] = "FINISHED"
+    except:
+        run_data["status"] = "ERROR"
+        run_data["error"] = str(traceback.format_exc())
+        if not output_log:
+            raise
     finally:
+        end_time = datetime.datetime.now(datetime.timezone.utc).isoformat()
+        run_data["end_time"] = end_time
         if "driver" in globals_locals:
             globals_locals["driver"].quit()
         if "display" in globals_locals:
             globals_locals["display"].stop()
+        if "parsagon_log" in globals_locals:
+            run_data["log"] = "\n".join(globals_locals["parsagon_log"])
+            logger.info(run_data["log"])
+        if "parsagon_warnings" in globals_locals:
+            run_data["warnings"] = globals_locals["parsagon_warnings"]
         for proc in psutil.process_iter():
             try:
                 if proc.name() == "chromedriver":
                     proc.kill()
             except psutil.NoSuchProcess:
                 continue
+        run = update_pipeline_run(run["id"], run_data)
     logger.info("Done.")
+    if output_log:
+        if "error" not in run_data:
+            run["output"] = globals_locals["output"]
+        return {k: v for k, v in run.items() if k in ("output", "status", "log", "warnings", "error")}
     return globals_locals["output"]
 
 
-def batch_runs(batch_name, program_name, runs=[], headless=False, ignore_errors=False, error_value=None):
+def batch_runs(batch_name, program_name, runs=[], headless=False, ignore_errors=False, error_value=None, rerun_warnings=False, rerun_warning_types=[], rerun_errors=False, verbose=False):
+    configure_logging(verbose)
+
     save_file = f"{batch_name}.json"
     try:
         with open(save_file) as f:
-            results = json.load(f)
+            outputs = json.load(f)
     except FileNotFoundError:
-        results = []
-    num_initial_results = len(results)
+        outputs = []
+    metadata_file = f"{batch_name}_metadata.json"
+    try:
+        with open(metadata_file) as f:
+            metadata = json.load(f)
+    except FileNotFoundError:
+        metadata = []
+
+    num_initial_results = len(outputs)
     pbar = tqdm(runs)
     default_desc = f'Running program "{program_name}"'
     pbar.set_description(default_desc)
     error = None
-    error_variables = None
+    variables = None
     try:
         for i, variables in enumerate(pbar):
             if i < num_initial_results:
-                continue
+                if rerun_errors and metadata[i]["status"] == "ERROR":
+                    pass
+                elif rerun_warnings and metadata[i]["warnings"]:
+                    if not rerun_warning_types or any(warning["type"] in rerun_warning_types for warning in metadata[i]["warnings"]):
+                        pass
+                    else:
+                        continue
+                else:
+                    continue
             for j in range(3):
-                try:
-                    results.append(run(program_name, variables, headless))
+                result = run(program_name, variables, headless, output_log=True)
+                if result["status"] != "ERROR":
+                    output = result.pop("output")
+                    if i < num_initial_results:
+                        outputs[i] = output
+                        metadata[i] = result
+                    else:
+                        outputs.append(output)
+                        metadata.append(result)
                     break
-                except Exception as e:
-                    error = e
-                    error_variables = variables
+                else:
+                    error = result["error"].strip().split("\n")[-1]
                     if j < 2:
-                        pbar.set_description(f"An error occurred: {e} - Waiting 60s before retrying (Attempt {j+2}/3)")
+                        pbar.set_description(f"An error occurred: {error} - Waiting 60s before retrying (Attempt {j+2}/3)")
                         time.sleep(60)
                         pbar.set_description(default_desc)
                         error = None
-                        error_variables = None
                         continue
                     else:
                         if ignore_errors:
                             error = None
-                            error_variables = None
-                            results.append(error_value)
+                            if i < num_initial_results:
+                                outputs[i] = error_value
+                            else:
+                                outputs.append(error_value)
                             break
                         else:
-                            raise
+                            raise RunFailedException
+    except RunFailedException:
+        logger.error(f"Unresolvable error occurred on run with variables {variables}: {error} - Data has been saved to {save_file}. Rerun your command to resume.")
     except Exception as e:
-        logger.error(f"Unresolvable error occurred on run with variables {error_variables}: {error} - Data has been saved to {save_file}. Rerun your command to resume.")
+        error = str(e)
+        logger.error(f"Unresolvable error occurred while looping over runs: {error} - Data has been saved to {save_file}. Rerun your command to resume.")
     finally:
         with open(save_file, "w") as f:
-            json.dump(results, f)
-    return None if error else results
+            json.dump(outputs, f)
+        with open(metadata_file, "w") as f:
+            json.dump(metadata, f)
+    num_warnings = 0
+    num_runs_with_warnings = 0
+    for m in metadata:
+        if m["warnings"]:
+            num_warnings += len(m["warnings"])
+            num_runs_with_warnings += 1
+    logger.info(f"\nSummary: {len(outputs)} runs made; {num_warnings} warnings encountered across {num_runs_with_warnings} runs. See {metadata_file} for logs.\n")
+    return None if error else outputs
 
 
 def delete(program_name, verbose=False, confirm_with_user=False):

parsagon/secrets.py

@@ -0,0 +1,15 @@
+import ast
+import re
+
+
+def extract_secrets(task):
+    secrets = {}
+    matches = list(re.finditer(r'\{\s*(?P<var>[A-Za-z_]+)\s*:\s*(?P<value>"([^"]|\\")*")\}', task))
+    for match in matches:
+        var_name = match.group("var")
+        if not var_name.startswith("SECRET"):
+            continue
+        new_match = re.sub(r'\{([A-Za-z_]+):\s*"([^"]|\\")*"\}', '{\\1: "******"}', match.group(0))
+        task = task.replace(match.group(0), new_match)
+        secrets[match.group(1)] = ast.literal_eval(match.group(2))
+    return task, secrets

parsagon/tests/test_secrets.py

@@ -0,0 +1,53 @@
+import pytest
+
+from parsagon.secrets import extract_secrets
+
+
+def test_non_secrets_are_not_extracted():
+    """
+    Non-secrets should not be extracted from task descriptions.
+    """
+    task = 'Go to https://example.com. Type {username: "myusername"} in the username field'
+    task, secrets = extract_secrets(task)
+    assert secrets == {}
+    assert task == 'Go to https://example.com. Type {username: "myusername"} in the username field'
+
+
+def test_secret_is_extracted():
+    """
+    A secret should be extracted and replaced in a task description.
+    """
+    task = 'Go to https://example.com. Type {SECRET_PASSWORD: "mypassword"} in the password field'
+    task, secrets = extract_secrets(task)
+    assert secrets == {"SECRET_PASSWORD": "mypassword"}
+    assert task == 'Go to https://example.com. Type {SECRET_PASSWORD: "******"} in the password field'
+
+
+def test_secret_with_quotes_is_extracted():
+    """
+    A secret with quotes in its value should be extracted and replaced in a task description.
+    """
+    task = 'Go to https://example.com. Type {SECRET_PASSWORD: "mypassword\\"?!1"} in the password field'
+    task, secrets = extract_secrets(task)
+    assert secrets == {"SECRET_PASSWORD": 'mypassword"?!1'}
+    assert task == 'Go to https://example.com. Type {SECRET_PASSWORD: "******"} in the password field'
+
+
+def test_multiple_secrets_are_extracted():
+    """
+    Multiple secrets should be extracted and replaced in the same task description.
+    """
+    task = 'Go to https://example.com. Type {SECRET_PASSWORD: "mypassword"} in the password field. Type {SECRET_ADDRESS: "myaddress"} in the address field'
+    task, secrets = extract_secrets(task)
+    assert secrets == {"SECRET_PASSWORD": "mypassword", "SECRET_ADDRESS": "myaddress"}
+    assert task == 'Go to https://example.com. Type {SECRET_PASSWORD: "******"} in the password field. Type {SECRET_ADDRESS: "******"} in the address field'
+
+
+def test_secrets_mixed_with_non_secrets_are_extracted():
+    """
+    Multiple secrets should be extracted and replaced in the same task description, and non-secrets should remain the same.
+    """
+    task = 'Go to https://example.com. Type {USERNAME: "myusername"} in the username field. Type {SECRET_PASSWORD: "mypassword"} in the password field. Type {SECRET_ADDRESS: "myaddress"} in the address field'
+    task, secrets = extract_secrets(task)
+    assert secrets == {"SECRET_PASSWORD": "mypassword", "SECRET_ADDRESS": "myaddress"}
+    assert task == 'Go to https://example.com. Type {USERNAME: "myusername"} in the username field. Type {SECRET_PASSWORD: "******"} in the password field. Type {SECRET_ADDRESS: "******"} in the address field'

{parsagon-0.10.20.dist-info → parsagon-0.10.22.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: parsagon
-Version: 0.10.20
+Version: 0.10.22
 Summary: Allows you to create browser automations with natural language
 Author-email: Sandy Suh <sandy@parsagon.io>
 Project-URL: Homepage, https://parsagon.io

{parsagon-0.10.20.dist-info → parsagon-0.10.22.dist-info}/RECORD

@@ -1,10 +1,11 @@
 parsagon/__init__.py,sha256=n4-wiFVVuyW_KOJeNiycggAg9BTa5bbBIVpD_DkdOO4,125
-parsagon/api.py,sha256=eJULOzTyWqA4Mio7tH9PszwTrZyxRBI0uO9t1h3R7rw,6634
+parsagon/api.py,sha256=nDTDe0LdDTn1hSXbgqd8j1qxe_3xWm3wZXhrTsmbwOE,6842
 parsagon/custom_function.py,sha256=oEj28qItaHUnsvLIHD7kg5QL3J3aO6rW6xKKP-H-Drs,770
-parsagon/exceptions.py,sha256=NYpFaSLZplBTv9fov_1LKPzDPIqb7Ffe7IunnjntxvA,819
+parsagon/exceptions.py,sha256=tG1vnpmUN1GdJ1GSpe1MaWH3zWmFLZCwtOfEGu8qPP0,910
 parsagon/executor.py,sha256=e_e9p5eLvf7wYHk1BNJf0j_qt0H17BfivPb8CoOKMHE,22791
 parsagon/highlights.js,sha256=2UDfUApblU9xtGgTLCq4X7rHRV0wcqDSSFZPmJS6fJg,16643
-parsagon/main.py,sha256=yQbIzqJ7Ea6XZJ3Eolx2lTh7Di87qPJvDI0WRNeoX14,14736
+parsagon/main.py,sha256=mHmeXPUskTXyxJvuDnmOKF_MXkaOXB2oYYu5VOAE8s4,18344
+parsagon/secrets.py,sha256=72dr-6q1q2ATBkE75fT18tcvwDM-4nymTb9NDVwjHTE,545
 parsagon/settings.py,sha256=s5_MsDMFM5tB8U8tfHaFnKibCoEqPnAu8b_ueg07Ftw,2947
 parsagon/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 parsagon/tests/api_mocks.py,sha256=M8xhiyPa1dI8Vx-odDk7ETopfFAfcjfAf-ApmSqgvfw,3127
@@ -13,8 +14,9 @@ parsagon/tests/conftest.py,sha256=KMlHohc0QT77HzumraIojzKeqroyxarnaT6naJDNvEc,42
 parsagon/tests/test_executor.py,sha256=n3cmh84r74siSeJqUeAIwjjnNzDVPEdxcvYAeJ4hNX8,645
 parsagon/tests/test_invalid_args.py,sha256=kOjMpbZvviR1CwvXReteZMxBvuhq_rOv5Tm1muBSzNk,676
 parsagon/tests/test_pipeline_operations.py,sha256=TpBKCuRA8LHYWx3PD_k9mYCSsA_9SZjrOX-rS4mE8XE,1089
-parsagon-0.10.20.dist-info/METADATA,sha256=vA0gDtSRv8FeLScDZJHTNQjSgBw2MaAetLcrDJ9bIaw,2410
-parsagon-0.10.20.dist-info/WHEEL,sha256=yQN5g4mg4AybRjkgi-9yy4iQEFibGQmlz78Pik5Or-A,92
-parsagon-0.10.20.dist-info/entry_points.txt,sha256=I1UlPUb4oY2k9idkI8kvdkEcrjKGRSOl5pMbA6uu6kw,48
-parsagon-0.10.20.dist-info/top_level.txt,sha256=ih5uYQzW4qjhRKppys-WiHLIbXVZ99YdqDcfAtlcQwk,9
-parsagon-0.10.20.dist-info/RECORD,,
+parsagon/tests/test_secrets.py,sha256=Ctsscl2tmMTZcFAy5dnyqUlgTov2UharZgLpbRCLdEg,2662
+parsagon-0.10.22.dist-info/METADATA,sha256=WMKXNXXNse8ftQwxVvwnV9LTuur6NmF2KCOJr9C8yZI,2410
+parsagon-0.10.22.dist-info/WHEEL,sha256=Xo9-1PvkuimrydujYJAjF7pCkriuXBpUPEjma1nZyJ0,92
+parsagon-0.10.22.dist-info/entry_points.txt,sha256=I1UlPUb4oY2k9idkI8kvdkEcrjKGRSOl5pMbA6uu6kw,48
+parsagon-0.10.22.dist-info/top_level.txt,sha256=ih5uYQzW4qjhRKppys-WiHLIbXVZ99YdqDcfAtlcQwk,9
+parsagon-0.10.22.dist-info/RECORD,,

{parsagon-0.10.20.dist-info → parsagon-0.10.22.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: bdist_wheel (0.41.2)
+Generator: bdist_wheel (0.41.3)
 Root-Is-Purelib: true
 Tag: py3-none-any