parsagon 0.9.6__tar.gz → 0.10.1__tar.gz

{parsagon-0.9.6 → parsagon-0.10.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: parsagon
-Version: 0.9.6
+Version: 0.10.1
 Summary: Allows you to create browser automations with natural language
 Author-email: Sandy Suh <sandy@parsagon.io>
 Project-URL: Homepage, https://parsagon.io
@@ -8,7 +8,22 @@ Classifier: Programming Language :: Python :: 3
 Classifier: Operating System :: OS Independent
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
+Requires-Dist: selenium==4.9.1
+Requires-Dist: lxml==4.9.2
+Requires-Dist: httpx==0.24.1
+Requires-Dist: halo==0.0.31
+Requires-Dist: pandas==1.4.2
+Requires-Dist: PyVirtualDisplay==3.0
+Requires-Dist: selenium-wire==5.1.0
+Requires-Dist: cssselect==1.1.0
+Requires-Dist: undetected-chromedriver==3.5.2
+Requires-Dist: webdriver-manager==4.0.0
+Requires-Dist: jsonpath-ng==1.5.3
+Requires-Dist: usaddress==0.5.10
+Requires-Dist: simplejson==3.19.1
 Provides-Extra: dev
+Requires-Dist: pytest==7.3.2; extra == "dev"
+Requires-Dist: pytest-mock==3.11.1; extra == "dev"
 
 # parsagon
 

{parsagon-0.9.6 → parsagon-0.10.1}/pyproject.toml

@@ -16,7 +16,7 @@ line-length = 120
 
 [project]
 name = "parsagon"
-version = "0.9.6"
+version = "0.10.1"
 description = "Allows you to create browser automations with natural language"
 readme = "README.md"
 requires-python = ">=3.8"

parsagon-0.10.1/src/parsagon/__init__.py

@@ -0,0 +1 @@
+from parsagon.main import create, update, detail, run, delete, get_product, get_review_article, get_article_list

{parsagon-0.9.6 → parsagon-0.10.1}/src/parsagon/api.py

@@ -58,7 +58,7 @@ def get_program_sketches(description):
     """
     Gets a program sketches (full and abridged) from a description.
     :param description: Description in natural language that will be used to generate the scraping program.
-    :return: A dict with keys "full" and "abridged" for the respective program ASTs.
+    :return: A dict with keys "full", "abridged", and "pseudocode" for the respective program ASTs and pseudocode.
     """
     return _api_call(httpx.post, "/transformers/get-program-sketch/", json={"description": description})
 
@@ -197,3 +197,7 @@ def get_run(run_id):
         httpx.get,
         f"/pipelines/runs/{run_id}/",
     )
+
+
+def poll_data(url, page_type):
+    return _api_call(httpx.post, "/extract/", json={"url": url, "page_type": page_type})

{parsagon-0.9.6 → parsagon-0.10.1}/src/parsagon/main.py

@@ -17,6 +17,7 @@ from parsagon.api import (
     get_pipelines,
     get_pipeline_code,
     get_run,
+    poll_data,
     APIException,
 )
 from parsagon.exceptions import ParsagonException
@@ -189,10 +190,11 @@ def create(task=None, program_name=None, headless=False, infer=False, verbose=Fa
 
     logger.info("Analyzing task description...")
     program_sketches = get_program_sketches(task)
-    logger.info("Created a program based on task description. Now demonstrating what the program does:\n")
 
     full_program = program_sketches["full"]
     abridged_program = program_sketches["abridged"]
+    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
 
@@ -250,6 +252,14 @@ def update(program_name, variables={}, headless=False, infer=False, replace=Fals
     executor = Executor(headless=headless, infer=infer)
     executor.execute(abridged_program)
 
+    while True:
+        program_name_input = input(f"Type \"{program_name}\" to update this program, or press enter without typing a name to CANCEL: ")
+        if not program_name_input:
+            logger.info("Canceled update.")
+            return
+        if program_name_input == program_name:
+            break
+
     pipeline_id = pipeline["id"]
     try:
         for call_id, custom_function in executor.custom_functions.items():
@@ -343,3 +353,27 @@ def setup(verbose=False):
         logger.error("\nCancelled operation.")
         return
     logger.info("Setup complete.")
+
+
+def _get_data(url, page_type, timeout):
+    start_time = time.time()
+    with Halo(text="Extracting data...", spinner="dots"):
+        while time.time() - start_time <= timeout:
+            result = poll_data(url, page_type)
+            if result["done"]:
+                return result["result"]
+            time.sleep(15)
+    logger.info("No data found")
+    return None
+
+
+def get_product(url, timeout=300):
+    return _get_data(url, "PRODUCT_DETAIL", timeout)
+
+
+def get_review_article(url, timeout=300):
+    return _get_data(url, "REVIEW_ARTICLE_DETAIL", timeout)
+
+
+def get_article_list(url, timeout=300):
+    return _get_data(url, "ARTICLE_LIST", timeout)

{parsagon-0.9.6 → parsagon-0.10.1}/src/parsagon.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: parsagon
-Version: 0.9.6
+Version: 0.10.1
 Summary: Allows you to create browser automations with natural language
 Author-email: Sandy Suh <sandy@parsagon.io>
 Project-URL: Homepage, https://parsagon.io
@@ -8,7 +8,22 @@ Classifier: Programming Language :: Python :: 3
 Classifier: Operating System :: OS Independent
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
+Requires-Dist: selenium==4.9.1
+Requires-Dist: lxml==4.9.2
+Requires-Dist: httpx==0.24.1
+Requires-Dist: halo==0.0.31
+Requires-Dist: pandas==1.4.2
+Requires-Dist: PyVirtualDisplay==3.0
+Requires-Dist: selenium-wire==5.1.0
+Requires-Dist: cssselect==1.1.0
+Requires-Dist: undetected-chromedriver==3.5.2
+Requires-Dist: webdriver-manager==4.0.0
+Requires-Dist: jsonpath-ng==1.5.3
+Requires-Dist: usaddress==0.5.10
+Requires-Dist: simplejson==3.19.1
 Provides-Extra: dev
+Requires-Dist: pytest==7.3.2; extra == "dev"
+Requires-Dist: pytest-mock==3.11.1; extra == "dev"
 
 # parsagon
 

parsagon-0.9.6/src/parsagon/__init__.py

@@ -1 +0,0 @@
-from parsagon.main import create, detail, run, delete