parsagon 0.14.25__py3-none-any.whl → 0.14.27__py3-none-any.whl

parsagon/api.py

@@ -235,7 +235,19 @@ def get_pipelines():
     return _api_call(httpx.get, f"/pipelines/")
 
 
-def get_pipeline_code(pipeline_name, variables, headless, use_uc, optimize, use_proxy):
+def get_pipeline_code(pipeline_name, variables, headless, use_uc, optimize, use_proxy, pipeline_id=None):
+    if pipeline_id:
+        return _api_call(
+            httpx.post,
+            f"/pipelines/{pipeline_id}/code/",
+            json={
+                "variables": variables,
+                "headless": headless,
+                "use_uc": use_uc,
+                "optimize": optimize,
+                "use_proxy": use_proxy,
+            },
+        )
     escaped_pipeline_name = double_quote(pipeline_name)
     with RaiseProgramNotFound(pipeline_name):
         return _api_call(

parsagon/executor.py

@@ -1,13 +1,16 @@
 from collections import defaultdict
 import copy
+import glob
 import json
 import logging
+import os
 from pathlib import Path
 import psutil
 import time
 from urllib.parse import urljoin
 
 import lxml.html
+from pypdf import PdfReader
 from pyvirtualdisplay import Display
 import undetected_chromedriver as uc
 from selenium import webdriver
@@ -92,12 +95,30 @@ class Executor:
             chrome_options.add_argument("--start-maximized")
             for option in options:
                 chrome_options.add_argument(option)
+            chrome_options.add_experimental_option(
+                "prefs",
+                {
+                    "download.default_directory": os.getcwd(),
+                    "download.prompt_for_download": False,
+                    "download.directory_upgrade": True,
+                    "plugins.always_open_pdf_externally": True,
+                },
+            )
             self.driver = uc.Chrome(driver_executable_path=driver_executable_path, options=chrome_options)
         else:
             chrome_options = webdriver.ChromeOptions()
             chrome_options.add_argument("--start-maximized")
             for option in options:
                 chrome_options.add_argument(option)
+            chrome_options.add_experimental_option(
+                "prefs",
+                {
+                    "download.default_directory": os.getcwd(),
+                    "download.prompt_for_download": False,
+                    "download.directory_upgrade": True,
+                    "plugins.always_open_pdf_externally": True,
+                },
+            )
             self.driver = webdriver.Chrome(service=ChromeService(driver_executable_path), options=chrome_options)
         if page_load_timeout:
             self.driver.set_page_load_timeout(page_load_timeout)
@@ -124,6 +145,7 @@ class Executor:
             "get_str_about_data": get_str_about_data,
             "get_bool_about_data": get_bool_about_data,
             "get_json_about_data": get_json_about_data,
+            "get_pdf_text": self.get_pdf_text,
         }
         self.custom_functions = {}
         self.infer = infer
@@ -617,6 +639,19 @@ class Executor:
         self.add_custom_function(call_id, custom_function)
         return scraped_data
 
+    def get_pdf_text(self, url):
+        window_id = self.goto(url)
+        self.close_window(window_id)
+        files = glob.glob("*")
+        most_recent_file = max(files, key=os.path.getmtime)
+        reader = PdfReader(most_recent_file)
+        text = ""
+        for page in reader.pages:
+            text += page.extract_text(extraction_mode="layout", layout_mode_space_vertically=False)
+            text += "\n"
+        os.remove(most_recent_file)
+        return text
+
     def execute(self, code):
         loc = {}
         try:

parsagon/runs.py

@@ -22,6 +22,7 @@ logger = logging.getLogger(__name__)
 
 def run(
     program_name,
+    program_id=None,
     variables={},
     headless=False,
     remote=False,
@@ -42,7 +43,7 @@ def run(
         raise ParsagonException("Variables must be a dictionary")
 
     logger.info("Preparing to run program %s", program_name)
-    pipeline_id = get_pipeline(program_name)["id"]
+    pipeline_id = program_id or get_pipeline(program_name)["id"]
 
     if remote:
         result = create_pipeline_run(pipeline_id, variables, False)
@@ -81,7 +82,9 @@ def run(
                 time.sleep(5)
 
     run = create_pipeline_run(pipeline_id, variables, True)
-    code = get_pipeline_code(program_name, variables, headless, undetected, optimize, use_proxy)["code"]
+    code = get_pipeline_code(
+        program_name, variables, headless, undetected, optimize, use_proxy, pipeline_id=program_id
+    )["code"]
     start_time = datetime.datetime.now(datetime.timezone.utc).isoformat()
     run_data = {"start_time": start_time}
 

{parsagon-0.14.25.dist-info → parsagon-0.14.27.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: parsagon
-Version: 0.14.25
+Version: 0.14.27
 Summary: Allows you to create browser automations with natural language
 Author-email: Sandy Suh <sandy@parsagon.io>
 Project-URL: Homepage, https://parsagon.io
@@ -16,10 +16,11 @@ Requires-Dist: rich ==13.6.0
 Requires-Dist: PyVirtualDisplay ==3.0
 Requires-Dist: selenium-wire ==5.1.0
 Requires-Dist: cssselect ==1.1.0
-Requires-Dist: undetected-chromedriver ==3.5.3
+Requires-Dist: undetected-chromedriver ==3.5.5
 Requires-Dist: webdriver-manager ==4.0.1
 Requires-Dist: jsonpath-ng ==1.5.3
 Requires-Dist: simplejson ==3.19.1
+Requires-Dist: pypdf ==4.2.0
 Provides-Extra: dev
 Requires-Dist: pytest ==7.3.2 ; extra == 'dev'
 Requires-Dist: pytest-mock ==3.11.1 ; extra == 'dev'

{parsagon-0.14.25.dist-info → parsagon-0.14.27.dist-info}/RECORD

@@ -1,16 +1,16 @@
 parsagon/__init__.py,sha256=_S5MlYHBViB4iY96_UbGo3mfGndE8MmFmb1EORJppK8,452
-parsagon/api.py,sha256=z52b5zBMTaIpWw9tu5-fF0Ngt58Vtm5TFnreUm8O_XU,9834
+parsagon/api.py,sha256=DpnHmbP61je3qeX9_6mUFt44Ps0f4TwaSRUsWVueAIM,10203
 parsagon/assistant.py,sha256=V3NL6UdDqe74W_X3wPQ1qwFuJRvHzitXtOt-XSCXvds,4065
 parsagon/create.py,sha256=igrJN8jVP5p3P9-dPSabrHGOQOlG2Flx7ep4SqEDB14,4373
 parsagon/custom_function.py,sha256=oEj28qItaHUnsvLIHD7kg5QL3J3aO6rW6xKKP-H-Drs,770
 parsagon/edit.py,sha256=aLeAcp1hFHm29qh6do21cfYykK76CsHC4tc9Uh2_2bk,3006
 parsagon/exceptions.py,sha256=tG1vnpmUN1GdJ1GSpe1MaWH3zWmFLZCwtOfEGu8qPP0,910
-parsagon/executor.py,sha256=9KVsw_Hh4NEwIvWau8MnkAQ2mK83nWrFy1B3iAGzm9Q,25620
+parsagon/executor.py,sha256=Vijqx5fbYmglC1EFxMp4K0oqDTOmCy0ruBxZ784X0Uk,26932
 parsagon/gui_entry.py,sha256=bqG9K0CArXWWwDGoT8aV17YLNM8MfjSf6SJ_B3QbNeA,671
 parsagon/highlights.js,sha256=2UDfUApblU9xtGgTLCq4X7rHRV0wcqDSSFZPmJS6fJg,16643
 parsagon/main.py,sha256=WPxrT1ZyPe6wNSFP7GnnKIGisgPM2dFieHP2TC5TGVQ,9467
 parsagon/print.py,sha256=-7iVKil0W9e8zX1EJMcdlqNdfpmfPxKTBtZfwzWpGYU,4106
-parsagon/runs.py,sha256=XXbk3eUJjBBETBq6AUyBrNPIR18svCRTSCSnhN2GpOc,8560
+parsagon/runs.py,sha256=gi36oak3RxKo1BxB0Bc6GV8K06vrrBrEXu3NYkIVPwY,8633
 parsagon/secrets.py,sha256=72dr-6q1q2ATBkE75fT18tcvwDM-4nymTb9NDVwjHTE,545
 parsagon/settings.py,sha256=ejd9wGCOEB68n8a0M2SyzlmlwPofnzMKaxpBqYhcs34,3552
 parsagon/gui/__init__.py,sha256=ZvKZfcchSnBh__UY_XeMnQJACd1EG99ix3_SfSHth8g,68
@@ -25,8 +25,8 @@ parsagon/tests/test_invalid_args.py,sha256=TAFdHGy92lUxjljPrtODOuEGVss6rn-F5GvEK
 parsagon/tests/test_pipeline_operations.py,sha256=aEwZNtIwOl9X7jdLDLB4YEdgMp7_x8PXCINAE7RT4NY,805
 parsagon/tests/test_print.py,sha256=BG7f55YDBoL0S7k291-so_Gje_hUAQOkB-jh-bEYsJY,198
 parsagon/tests/test_secrets.py,sha256=Ctsscl2tmMTZcFAy5dnyqUlgTov2UharZgLpbRCLdEg,2662
-parsagon-0.14.25.dist-info/METADATA,sha256=SSWWJVRIrKeqzti57XG6tZdmgAJzegU8zXTt_5pBnu4,2432
-parsagon-0.14.25.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-parsagon-0.14.25.dist-info/entry_points.txt,sha256=I1UlPUb4oY2k9idkI8kvdkEcrjKGRSOl5pMbA6uu6kw,48
-parsagon-0.14.25.dist-info/top_level.txt,sha256=ih5uYQzW4qjhRKppys-WiHLIbXVZ99YdqDcfAtlcQwk,9
-parsagon-0.14.25.dist-info/RECORD,,
+parsagon-0.14.27.dist-info/METADATA,sha256=zVcojJBz_ObyKy4oxzOdWZqP51O7MqhrKJYSQMz-_Ns,2461
+parsagon-0.14.27.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+parsagon-0.14.27.dist-info/entry_points.txt,sha256=I1UlPUb4oY2k9idkI8kvdkEcrjKGRSOl5pMbA6uu6kw,48
+parsagon-0.14.27.dist-info/top_level.txt,sha256=ih5uYQzW4qjhRKppys-WiHLIbXVZ99YdqDcfAtlcQwk,9
+parsagon-0.14.27.dist-info/RECORD,,