✨ Fix on_download and on_pdf

harambe/core.py

@@ -175,7 +175,7 @@ async def capture_download(
 
         res = await asyncio.gather(
             *[
-                o.on_download(download.suggested_filename, content)
+                o.on_download(download.url, download.suggested_filename, content)
                 for o in self._observers
             ]
         )
@@ -189,9 +189,9 @@ async def capture_pdf(
         from the observer to transform to a usable URL
         """
         pdf_content = await self.page.pdf()
-        file_name = f"{self.page.url}-screen.pdf"
+        file_name = f"reworkd_page_snapshot.pdf"
         res = await asyncio.gather(
-            *[o.on_download(file_name, pdf_content) for o in self._observers]
+            *[o.on_download(self.page.url, file_name, pdf_content) for o in self._observers]
         )
         return res[0]
 

harambe/observer.py

@@ -1,5 +1,6 @@
 from abc import abstractmethod
 from typing import Any, Dict, List, Protocol, Tuple, runtime_checkable, TypedDict
+from urllib.parse import quote
 
 from harambe.tracker import FileDataTracker
 from harambe.types import URL, Context, Stage
@@ -31,7 +32,7 @@ async def on_queue_url(self, url: URL, context: Dict[str, Any]) -> None:
     async def on_download(self, download_url: str, filename: str, content: bytes) -> "DownloadMeta":
         print(f"Downloading file: {filename}")  # TODO: use logger
         return {
-            "url": f"{download_url}/{filename}",
+            "url": f"{download_url}/{quote(filename)}",
             "filename": filename,
         }
 
@@ -48,7 +49,7 @@ async def on_queue_url(self, url: URL, context: Dict[str, Any]) -> None:
 
     async def on_download(self, download_url: str, filename: str, content: bytes) -> "DownloadMeta":
         data = {
-            "url": f"{download_url}/{filename}",
+            "url": f"{download_url}/{quote(filename)}",
             "filename": filename,
         }
         self._tracker.save_data(data)
@@ -69,7 +70,7 @@ async def on_queue_url(self, url: URL, context: Dict[str, Any]) -> None:
 
     async def on_download(self, download_url: str, filename: str, content: bytes) -> "DownloadMeta":
         data = {
-            "url": f"{download_url}/{filename}",
+            "url": f"{download_url}/{quote(filename)}",
             "filename": filename,
         }
         self._files.append((filename, content))

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "harambe-sdk"
-version = "0.8.2"
+version = "0.8.3"
 description = "Data extraction SDK for Playwright 🐒🍌"
 authors = ["awtkns <[email protected]>"]
 readme = "README.md"