Merge pull request #131 from Zeecka/sentry_workers

fix 109

aperisolve/analyzers/base_analyzer.py

@@ -161,6 +161,7 @@ class SubprocessAnalyzer(ABC):
             self.update_result(result)
         except Exception as e:
             self.update_result({"status": "error", "error": str(e)})
+            raise
 
     def is_error(self, returncode: int, stdout: str, stderr: str, zip_exist: bool) -> bool:
         """Check if the result is an error."""

aperisolve/analyzers/decomposer.py

@@ -20,73 +20,70 @@ class DecomposerAnalyzer(SubprocessAnalyzer):
 
     def get_results(self, _: Optional[str] = None) -> dict[str, Any]:
         """Analyze an image submission using bits decomposition."""
-        try:
+        img = Image.open(self.input_img)
-            img = Image.open(self.input_img)
+        converted = False
-            converted = False
 
-            if img.mode == "P":
+        if img.mode == "P":
-                img = img.convert("RGB")
+            img = img.convert("RGB")
-                converted = True
+            converted = True
 
-            img_np = np.array(img)
+        img_np = np.array(img)
 
-            # Handle grayscale, RGB, RGBA, etc.
+        # Handle grayscale, RGB, RGBA, etc.
-            if len(img_np.shape) == 2:
+        if len(img_np.shape) == 2:
-                channels = 1
+            channels = 1
-            else:
+        else:
-                channels = img_np.shape[2]
+            channels = img_np.shape[2]
 
-            if channels > 1:
+        if channels > 1:
-                channel_names = ["Red", "Green", "Blue", "Alpha"]
+            channel_names = ["Red", "Green", "Blue", "Alpha"]
-            else:
+        else:
-                channel_names = ["Grayscale"]
+            channel_names = ["Grayscale"]
 
-            image_json = {}
+        image_json = {}
 
-            # Add Superimposed RGB bit planes
+        # Add Superimposed RGB bit planes
-            if channels >= 3:
+        if channels >= 3:
-                superimposed_json = []
+            superimposed_json = []
-                for bit in range(8):
+            for bit in range(8):
-                    bit_mask = 1 << bit
+                bit_mask = 1 << bit
-                    rgb_planes = [((img_np[..., c] & bit_mask) >> bit) * 255 for c in range(3)]
+                rgb_planes = [((img_np[..., c] & bit_mask) >> bit) * 255 for c in range(3)]
-                    rgb_image = np.stack(rgb_planes, axis=-1).astype(np.uint8)
+                rgb_image = np.stack(rgb_planes, axis=-1).astype(np.uint8)
-                    rgb_img = Image.fromarray(rgb_image, mode="RGB")
+                rgb_img = Image.fromarray(rgb_image, mode="RGB")
-                    img_name = f"superimposed_bit_{bit}.png"
+                img_name = f"superimposed_bit_{bit}.png"
-                    out_path = self.output_dir / img_name
+                out_path = self.output_dir / img_name
-                    dl_path = Path(self.output_dir.name) / img_name
+                dl_path = Path(self.output_dir.name) / img_name
-                    superimposed_json.append("/image/" + str(dl_path))
+                superimposed_json.append("/image/" + str(dl_path))
-                    rgb_img.save(out_path)
+                rgb_img.save(out_path)
 
-                image_json["Superimposed"] = superimposed_json
+            image_json["Superimposed"] = superimposed_json
 
-            # Process individual channels
+        # Process individual channels
-            for c in range(channels):
+        for c in range(channels):
-                channel_data = img_np[..., c] if channels > 1 else img_np
+            channel_data = img_np[..., c] if channels > 1 else img_np
-                channel_label = channel_names[c]
+            channel_label = channel_names[c]
-                channel_json = []
+            channel_json = []
-                for bit in range(8):
+            for bit in range(8):
-                    bit_mask = 1 << bit
+                bit_mask = 1 << bit
-                    bit_plane = ((channel_data & bit_mask) >> bit) * 255
+                bit_plane = ((channel_data & bit_mask) >> bit) * 255
-                    bit_img = Image.fromarray(bit_plane.astype(np.uint8), mode="L")
+                bit_img = Image.fromarray(bit_plane.astype(np.uint8), mode="L")
-                    img_name = f"{channel_label}_bit_{bit}.png"
+                img_name = f"{channel_label}_bit_{bit}.png"
-                    out_path = self.output_dir / img_name
+                out_path = self.output_dir / img_name
-                    dl_path = Path(self.output_dir.name) / img_name
+                dl_path = Path(self.output_dir.name) / img_name
-                    channel_json.append("/image/" + str(dl_path))
+                channel_json.append("/image/" + str(dl_path))
-                    bit_img.save(out_path)
+                bit_img.save(out_path)
-                image_json[channel_label] = channel_json
+            image_json[channel_label] = channel_json
 
-            output = {
+        output = {
-                "status": "ok",
+            "status": "ok",
-                "images": image_json,
+            "images": image_json,
-            }
+        }
-            if converted:
+        if converted:
-                output["note"] = (
+            output["note"] = (
-                    "Image contains a color palette and was converted to RGB for processing."
+                "Image contains a color palette and was converted to RGB for processing."
-                )
+            )
 
-            return output
+        return output
-        except Exception as e:
-            return {"status": "error", "error": str(e)}
 
 
 def analyze_decomposer(input_img: Path, output_dir: Path) -> None:

aperisolve/analyzers/image_resize.py

@@ -115,53 +115,47 @@ class ResizeAnalyzer(SubprocessAnalyzer):
         """
         logs = []
 
-        try:
+        with self.input_img.open("rb") as image:
-            with self.input_img.open("rb") as image:
+            img_bytes = bytearray(image.read())
-                img_bytes = bytearray(image.read())
+            try:
-                try:
+                self.png = PNG.from_bytearray(img_bytes)
-                    self.png = PNG.from_bytearray(img_bytes)
+            except ValueError:
-                except ValueError:
-                    return {
-                        "status": "error",
-                        "error": "IHDR chunk is not the first chunk, or PNG has invalid structure.",
-                    }
-
-            # Already correct
-            if self.png.crc == self.png.compute_crc():
-                return {
-                    "status": "ok",
-                    "note": f"PNG is already valid with dimensions "
-                    f"{self.png.width}x{self.png.height} and crc 0x{self.png.crc:08x}.",
-                }
-
-            saved_img_urls = []
-
-            # No match found
-            if not (db_matches := self._lookup_crc()) and not (
-                db_matches := self._search_height_crc()
-            ):
                 return {
                     "status": "error",
-                    "error": "Failure: No matching dimensions found.",
+                    "error": "IHDR chunk is not the first chunk, or PNG has invalid structure.",
                 }
 
-            # Match found, saving image(s)
+        # Already correct
-            logs.append(f"Target CRC found: 0x{self.png.crc:08x}")
+        if self.png.crc == self.png.compute_crc():
-            self.output_dir.mkdir(parents=True, exist_ok=True)
-            for png_match in db_matches:
-                img_name = self._write_recovered_png(img_bytes, png_match)
-                logs.append(f"Image saved: {img_name}")
-                # Append the formatted URL to our list
-                saved_img_urls.append("/image/" + str(Path(self.output_dir.name) / img_name))
-
             return {
                 "status": "ok",
-                "output": logs,
+                "note": f"PNG is already valid with dimensions "
-                "png_images": saved_img_urls,
+                f"{self.png.width}x{self.png.height} and crc 0x{self.png.crc:08x}.",
             }
 
-        except Exception as e:
+        saved_img_urls = []
-            return {"status": "error", "error": str(e)}
+
+        # No match found
+        if not (db_matches := self._lookup_crc()) and not (db_matches := self._search_height_crc()):
+            return {
+                "status": "error",
+                "error": "Failure: No matching dimensions found.",
+            }
+
+        # Match found, saving image(s)
+        logs.append(f"Target CRC found: 0x{self.png.crc:08x}")
+        self.output_dir.mkdir(parents=True, exist_ok=True)
+        for png_match in db_matches:
+            img_name = self._write_recovered_png(img_bytes, png_match)
+            logs.append(f"Image saved: {img_name}")
+            # Append the formatted URL to our list
+            saved_img_urls.append("/image/" + str(Path(self.output_dir.name) / img_name))
+
+        return {
+            "status": "ok",
+            "output": logs,
+            "png_images": saved_img_urls,
+        }
 
 
 def analyze_image_resize(input_img: Path, output_dir: Path) -> None:

aperisolve/analyzers/openstego.py

@@ -44,6 +44,7 @@ class OpenStegoAnalyzer(SubprocessAnalyzer):
             self.update_result(result)
         except Exception as e:
             self.update_result({"status": "error", "error": str(e)})
+            raise
 
     def process_output(self, stdout: str, stderr: str) -> str | list[str] | dict[str, str]:
         """Process the stdout/stderr."""

aperisolve/workers.py

@@ -77,12 +77,26 @@ def analyze_image(submission_hash: str) -> None:
 
             def run_analyzer(analyzer_func: Any, *args: Any) -> None:
                 """Run an analyzer function in a separate thread."""
+                analyzer_name = analyzer_func.__name__.replace("analyze_", "")
                 try:
                     analyzer_func(*args)
                 except Exception as e:
-                    sentry_sdk.capture_exception(e)
+                    print(f"Error in {analyzer_name}: {e}")
-                    print(f"Error in {analyzer_func.__name__}: {e}")
+                    with sentry_sdk.push_scope() as scope:
-                    sentry_sdk.capture_exception(e)
+                        scope.set_tag("analyzer", analyzer_name)
+                        scope.set_tag("submission_hash", submission_hash)
+                        scope.set_context(
+                            "analyzer_info",
+                            {
+                                "tool": analyzer_name,
+                                "image_path": str(img_path),
+                                "result_path": str(result_path),
+                                "filename": submission.filename,
+                                "deep_analysis": submission.deep_analysis,
+                            },
+                        )
+                        scope.fingerprint = ["analyzer-error", analyzer_name]
+                        sentry_sdk.capture_exception(e)
 
             analyzers = [
                 (analyze_binwalk, img_path, result_path),