raise filesystem events from filedownload

bbot/modules/filedownload.py

@@ -13,7 +13,7 @@ class filedownload(BaseModule):
     """
 
     watched_events = ["URL_UNVERIFIED", "HTTP_RESPONSE"]
-    produced_events = []
+    produced_events = ["FILESYSTEM"]
     flags = ["active", "safe", "web-basic", "web-thorough"]
     meta = {
         "description": "Download common filetypes such as PDF, DOCX, PPTX, etc.",
@@ -122,22 +122,28 @@ async def handle_event(self, event):
             extension_matches = any(url_lower.endswith(f".{e}") for e in self.extensions)
             filedownload_requested = "filedownload" in event.tags
             if extension_matches or filedownload_requested:
-                await self.download_file(event.data)
+                await self.download_file(event.data, source_event=event)
         elif event.type == "HTTP_RESPONSE":
             headers = event.data.get("header", {})
             content_type = headers.get("content_type", "")
             if content_type:
                 url = event.data["url"]
-                await self.download_file(url, content_type=content_type)
+                await self.download_file(url, content_type=content_type, source_event=event)
 
-    async def download_file(self, url, content_type=None):
+    async def download_file(self, url, content_type=None, source_event=None):
         orig_filename, file_destination, base_url = self.make_filename(url, content_type=content_type)
         if orig_filename is None:
             return
         result = await self.helpers.download(url, warn=False, filename=file_destination, max_size=self.max_filesize)
         if result:
             self.info(f'Found "{orig_filename}" at "{base_url}", downloaded to {file_destination}')
             self.files_downloaded += 1
+            if source_event:
+                file_event = self.make_event(
+                    {"path": str(file_destination)}, "FILESYSTEM", tags=["filedownload", "file"], source=source_event
+                )
+                file_event.scope_distance = source_event.scope_distance
+                await self.emit_event(file_event)
         self.urls_downloaded.add(hash(url))
 
     def make_filename(self, url, content_type=None):

bbot/test/test_step_2/module_tests/test_module_filedownload.py

@@ -1,3 +1,4 @@
+from pathlib import Path
 from .base import ModuleTestBase
 
 
@@ -39,19 +40,20 @@ async def setup_after_prep(self, module_test):
         )
 
     def check(self, module_test, events):
+        filesystem_events = [e for e in events if e.type == "FILESYSTEM"]
         download_dir = module_test.scan.home / "filedownload"
 
         # text file
-        text_files = list(download_dir.glob("*test-file.txt"))
-        assert len(text_files) == 1, f"No text file found at {download_dir}"
-        file = text_files[0]
+        text_file_event = [e for e in filesystem_events if "test-file.txt" in e.data["path"]]
+        assert 1 == len(text_file_event), f"No text file found at {download_dir}"
+        file = Path(text_file_event[0].data["path"])
         assert file.is_file(), f"File not found at {file}"
         assert open(file).read() == "juicy stuff", f"File at {file} does not contain the correct content"
 
         # PDF file (no extension)
-        pdf_files = list(download_dir.glob("*test-pdf.pdf"))
-        assert len(pdf_files) == 1, f"No PDF file found at {download_dir}"
-        file = pdf_files[0]
+        pdf_file_event = [e for e in filesystem_events if "test-pdf.pdf" in e.data["path"]]
+        assert 1 == len(pdf_file_event), f"No PDF file found at {download_dir}"
+        file = Path(pdf_file_event[0].data["path"])
         assert file.is_file(), f"File not found at {file}"
         assert open(file).read() == self.pdf_data, f"File at {file} does not contain the correct content"