bbot 2.1.2.5238rc0__py3-none-any.whl → 2.1.2.5240rc0__py3-none-any.whl

bbot/__init__.py

@@ -1,4 +1,4 @@
 # version placeholder (replaced by poetry-dynamic-versioning)
-__version__ = "v2.1.2.5238rc"
+__version__ = "v2.1.2.5240rc"
 
 from .scanner import Scanner, Preset

bbot/modules/extractous.py

@@ -81,6 +81,12 @@ class extractous(BaseModule):
     async def handle_event(self, event):
         file_path = event.data["path"]
         content = await self.scan.helpers.run_in_executor_mp(extract_text, file_path)
+        if isinstance(content, tuple):
+            error, traceback = content
+            self.error(f"Error extracting text from {file_path}: {error}")
+            self.trace(traceback)
+            return
+
         if content:
             raw_text_event = self.make_event(
                 content,
@@ -99,49 +105,18 @@ def extract_text(file_path):
     :return: ASCII-encoded plaintext extracted from the document.
     """
 
-    extractable_file_types = [
-        ".csv",
-        ".eml",
-        ".msg",
-        ".epub",
-        ".xlsx",
-        ".xls",
-        ".html",
-        ".htm",
-        ".md",
-        ".org",
-        ".odt",
-        ".pdf",
-        ".txt",
-        ".text",
-        ".log",
-        ".ppt",
-        ".pptx",
-        ".rst",
-        ".rtf",
-        ".tsv",
-        ".doc",
-        ".docx",
-        ".xml",
-    ]
-
-    # If the file can be extracted with extractous use its partition function or try and read it
-    if any(file_path.lower().endswith(file_type) for file_type in extractable_file_types):
-        try:
-            extractor = Extractor()
-            reader = extractor.extract_file(str(file_path))
+    try:
+        extractor = Extractor()
+        reader, metadata = extractor.extract_file(str(file_path))
 
-            result = ""
+        result = ""
+        buffer = reader.read(4096)
+        while len(buffer) > 0:
+            result += buffer.decode("utf-8")
             buffer = reader.read(4096)
-            while len(buffer) > 0:
-                result += buffer.decode("utf-8")
-                buffer = reader.read(4096)
 
-            return result.strip()
+        return result.strip()
+    except Exception as e:
+        import traceback
 
-        except Exception:
-            with open(file_path, "rb") as file:
-                return file.read().decode("utf-8", errors="ignore")
-    else:
-        with open(file_path, "rb") as file:
-            return file.read().decode("utf-8", errors="ignore")
+        return (str(e), traceback.format_exc())

{bbot-2.1.2.5238rc0.dist-info → bbot-2.1.2.5240rc0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: bbot
-Version: 2.1.2.5238rc0
+Version: 2.1.2.5240rc0
 Summary: OSINT automation for hackers.
 Home-page: https://github.com/blacklanternsecurity/bbot
 License: GPL-3.0

{bbot-2.1.2.5238rc0.dist-info → bbot-2.1.2.5240rc0.dist-info}/RECORD

@@ -1,4 +1,4 @@
-bbot/__init__.py,sha256=JVjtsonS_He_khDQUwAqECN1ZBEtf2leD_lCLKF4TSc,130
+bbot/__init__.py,sha256=9oJhyJvXyXbI4jsxhLDiobby05_5HUsSxccd9M__JTc,130
 bbot/cli.py,sha256=7S3a4eB-Dl8yodc5WC-927Z30CNlLl9EXimGvIVypJo,10434
 bbot/core/__init__.py,sha256=l255GJE_DvUnWvrRb0J5lG-iMztJ8zVvoweDOfegGtI,46
 bbot/core/config/__init__.py,sha256=zYNw2Me6tsEr8hOOkLb4BQ97GB7Kis2k--G81S8vofU,342
@@ -96,7 +96,7 @@ bbot/modules/docker_pull.py,sha256=Dp8de9UCCELcozwmZphA3lMh8qZaXyDo2kfwG45Wm3w,9
 bbot/modules/dockerhub.py,sha256=ruvTP8Uz5LEuX-_SrKDzByvSNtd1ofZbX-lRTeKUB24,3491
 bbot/modules/dotnetnuke.py,sha256=XZysDA99ahQSLXR8RPROlmUwDxqrxvBFvscZMYBmsmc,10539
 bbot/modules/emailformat.py,sha256=RLPJW-xitYB-VT4Lp08qVzFkXx_kMyV_035JT_Yf4fM,1082
-bbot/modules/extractous.py,sha256=yPIM6UHYExGPNVDt8x_jE-UxRl_JbDrThFguIfBUuuY,5129
+bbot/modules/extractous.py,sha256=22pc62jOs_4E3uZtBjUbUksLTadNr9sp_4r0-eU1un8,4551
 bbot/modules/ffuf_shortnames.py,sha256=9Kh0kJsw7XXpXmCkiB5eAhG4h9rSo8Y-mB3p0EDa_l0,12624
 bbot/modules/filedownload.py,sha256=1prC84wAQO-W1HstitKPQ0-eYEApjzFn3RHFa9oaqLc,8185
 bbot/modules/fingerprintx.py,sha256=rdlR9d64AntAhbS_eJzh8bZCeLPTJPSKdkdKdhH_qAo,3269
@@ -404,8 +404,8 @@ bbot/wordlists/raft-small-extensions-lowercase_CLEANED.txt,sha256=ruUQwVfia1_m2u
 bbot/wordlists/top_open_ports_nmap.txt,sha256=LmdFYkfapSxn1pVuQC2LkOIY2hMLgG-Xts7DVtYzweM,42727
 bbot/wordlists/valid_url_schemes.txt,sha256=VciB-ww0y-O8Ii1wpTR6rJzGDiC2r-dhVsIJApS1ZYU,3309
 bbot/wordlists/wordninja_dns.txt.gz,sha256=DYHvvfW0TvzrVwyprqODAk4tGOxv5ezNmCPSdPuDUnQ,570241
-bbot-2.1.2.5238rc0.dist-info/LICENSE,sha256=GzeCzK17hhQQDNow0_r0L8OfLpeTKQjFQwBQU7ZUymg,32473
-bbot-2.1.2.5238rc0.dist-info/METADATA,sha256=HSSIhK3Gqj3mxD3gQoYAgdDlJInB3EdQUIy8eBPmuxQ,17109
-bbot-2.1.2.5238rc0.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-bbot-2.1.2.5238rc0.dist-info/entry_points.txt,sha256=cWjvcU_lLrzzJgjcjF7yeGuRA_eDS8pQ-kmPUAyOBfo,38
-bbot-2.1.2.5238rc0.dist-info/RECORD,,
+bbot-2.1.2.5240rc0.dist-info/LICENSE,sha256=GzeCzK17hhQQDNow0_r0L8OfLpeTKQjFQwBQU7ZUymg,32473
+bbot-2.1.2.5240rc0.dist-info/METADATA,sha256=-25uLYiERzYogB42LWOMPVT4217SpPPlAt7yv00ju5I,17109
+bbot-2.1.2.5240rc0.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+bbot-2.1.2.5240rc0.dist-info/entry_points.txt,sha256=cWjvcU_lLrzzJgjcjF7yeGuRA_eDS8pQ-kmPUAyOBfo,38
+bbot-2.1.2.5240rc0.dist-info/RECORD,,