Use artifact in context

* Pass the artifact to the location instead of just file name
* Load the snippet with the initialization of the location

Signed-off-by: Eric Brown <[email protected]>

precli/core/location.py

@@ -5,17 +5,12 @@
 class Location:
     def __init__(
         self,
-        file_name: str = None,
-        url: str = None,
         node: Node = None,
         start_line: int = 0,
         end_line: int = -1,
         start_column: int = 1,
         end_column: int = -1,
-        snippet: str = None,
     ):
-        self._file_name = file_name
-        self._url = url
         if node is not None:
             self._start_line = node.start_point[0] + 1
             self._start_column = node.start_point[1]
@@ -27,37 +22,6 @@ def __init__(
             self._start_column = start_column
             # TODO: default to end of line
             self._end_column = end_column
-        self._snippet = snippet
-
-    @property
-    def file_name(self) -> str:
-        """
-        Name of the file.
-
-        :return: file name
-        :rtype: str
-        """
-        return self._file_name
-
-    @property
-    def url(self) -> str:
-        """
-        If the original target was given as a URL, this
-        property will return that address.
-
-        :return: URL
-        :rtype: str
-        """
-        return self._url
-
-    @url.setter
-    def url(self, url: str):
-        """
-        Set the file location as a URL
-
-        :param str url: file network location
-        """
-        self._url = url
 
     @property
     def start_line(self) -> int:
@@ -98,22 +62,3 @@ def end_column(self) -> int:
         :rtype: int
         """
         return self._end_column
-
-    @property
-    def snippet(self) -> str:
-        """
-        Snippet of context of the code.
-
-        :return: snippet of context
-        :rtype: str
-        """
-        return self._snippet
-
-    @snippet.setter
-    def snippet(self, snippet):
-        """
-        Set the code context snippet.
-
-        :param str snippet: context snippet
-        """
-        self._snippet = snippet

precli/core/result.py

@@ -1,7 +1,9 @@
-# Copyright 2023 Secure Saurce LLC
+# Copyright 2024 Secure Saurce LLC
+from precli.core.artifact import Artifact
 from precli.core.fix import Fix
 from precli.core.kind import Kind
 from precli.core.level import Level
+from precli.core.linecache import LineCache
 from precli.core.location import Location
 from precli.core.suppression import Suppression
 from precli.rules import Rule
@@ -11,14 +13,17 @@ class Result:
     def __init__(
         self,
         rule_id: str,
+        artifact: Artifact = None,
         kind: Kind = Kind.FAIL,
         level: Level = None,
         location: Location = None,
         message: str = None,
         fixes: list[Fix] = None,
         suppression: Suppression = None,
+        snippet: str = None,
     ):
         self._rule_id = rule_id
+        self._artifact = artifact
         self._kind = kind
         default_config = Rule.get_by_id(self._rule_id).default_config
         self._rank = default_config.rank
@@ -34,6 +39,17 @@ def __init__(
         self._fixes = fixes if fixes is not None else []
         self._suppression = suppression
 
+        if snippet is not None:
+            self._snippet = snippet
+        else:
+            linecache = LineCache(
+                artifact.file_name,
+                artifact.contents.decode(),
+            )
+            self._snippet = ""
+            for i in range(location.start_line - 1, location.end_line + 2):
+                self._snippet += linecache.getline(i)
+
     @property
     def rule_id(self) -> str:
         """
@@ -48,24 +64,14 @@ def rule_id(self) -> str:
         return self._rule_id
 
     @property
-    def source_language(self) -> str:
+    def artifact(self) -> Artifact:
         """
-        The source language.
+        Artifact, typically the file.
 
-        :return: language of the source code
-        :rtype: str
+        :return: the artifact
+        :rtype: Artifact
         """
-        match (self._rule_id[:2]):
-            case "GO":
-                return "go"
-            case "JV":
-                return "java"
-            case "PY":
-                return "python"
-            case "RB":
-                return "ruby"
-            case "RS":
-                return "rust"
+        return self._artifact
 
     @property
     def location(self) -> Location:
@@ -159,3 +165,13 @@ def suppression(self, suppression):
         :param Suppression suppression: suppression
         """
         self._suppression = suppression
+
+    @property
+    def snippet(self) -> str:
+        """
+        Snippet of context of the code.
+
+        :return: snippet of context
+        :rtype: str
+        """
+        return self._snippet

precli/parsers/__init__.py

@@ -6,7 +6,6 @@
 from tree_sitter import Node
 
 from precli.core.artifact import Artifact
-from precli.core.linecache import LineCache
 from precli.core.location import Location
 from precli.core.result import Result
 from precli.core.suppression import Suppression
@@ -78,22 +77,14 @@ def parse(self, artifact: Artifact) -> list[Result]:
         :rtype: list
         """
         self.results = []
-        self.context = {"file_name": artifact.file_name}
+        self.context = {"artifact": artifact}
         if artifact.contents is None:
             with open(artifact.file_name, "rb") as fdata:
                 artifact.contents = fdata.read()
         tree = self.tree_sitter_parser.parse(artifact.contents)
         self.visit([tree.root_node])
 
-        linecache = LineCache(artifact.file_name, artifact.contents.decode())
-
         for result in self.results:
-            start = result.location.start_line - 1
-            stop = result.location.end_line + 2
-            result.location.snippet = ""
-            for i in range(start, stop):
-                result.location.snippet += linecache.getline(i)
-
             suppression = self.suppressions.get(result.location.start_line)
             if suppression and result.rule_id in suppression.rules:
                 result.suppression = suppression
@@ -161,7 +152,7 @@ def visit_ERROR(self, nodes: list[Node]):
         raise SyntaxError(
             "Syntax error while parsing file.",
             (
-                self.context["file_name"],
+                self.context["artifact"].file_name,
                 err_node.start_point[0] + 1,
                 err_node.start_point[1] + 1,
                 err_node.text.decode(errors="ignore"),

precli/renderers/detailed.py

@@ -36,10 +36,10 @@ def render(self, results: list[Result], metrics: Metrics):
                     emoji = ":information-emoji: "
                     style = "blue"
 
-            if result.location.url is not None:
-                file_name = result.location.url
+            if result.artifact.uri is not None:
+                file_name = result.artifact.uri
             else:
-                file_name = result.location.file_name
+                file_name = result.artifact.file_name
 
             self.console.print(
                 f"{emoji} {result.level.name.title()} on line "
@@ -59,8 +59,8 @@ def render(self, results: list[Result], metrics: Metrics):
 
             line_offset = result.location.start_line - 2
             code = syntax.Syntax(
-                result.location.snippet,
-                result.source_language,
+                result.snippet,
+                result.artifact.language,
                 line_numbers=True,
                 start_line=line_offset + 1,
                 line_range=(
@@ -96,13 +96,13 @@ def render(self, results: list[Result], metrics: Metrics):
                     before = 0
                 else:
                     line_before = linecache.getline(
-                        filename=result.location.file_name,
+                        filename=result.artifact.file_name,
                         lineno=start_line - 1,
                     )
                     before = 1
 
                 code = linecache.getline(
-                    filename=result.location.file_name,
+                    filename=result.artifact.file_name,
                     lineno=start_line,
                 )
 
@@ -111,7 +111,7 @@ def render(self, results: list[Result], metrics: Metrics):
                     after = 0
                 else:
                     line_after = linecache.getline(
-                        filename=result.location.file_name,
+                        filename=result.artifact.file_name,
                         lineno=start_line + 1,
                     )
                     after = 1
@@ -127,7 +127,7 @@ def render(self, results: list[Result], metrics: Metrics):
 
                 code = syntax.Syntax(
                     code,
-                    result.source_language,
+                    result.artifact.language,
                     line_numbers=True,
                     line_range=(start_line - before, end_line + after),
                     highlight_lines=highlight_lines,

precli/renderers/json.py

@@ -19,10 +19,10 @@ def render(self, results: list[Result], metrics: Metrics):
         for result in results:
             rule = Rule.get_by_id(result.rule_id)
 
-            if result.location.url is not None:
-                file_name = result.location.url
+            if result.artifact.uri is not None:
+                file_name = result.artifact.uri
             else:
-                file_name = result.location.file_name
+                file_name = result.artifact.file_name
 
             results_json["results"].append(
                 {

precli/renderers/plain.py

@@ -35,17 +35,17 @@ def render(self, results: list[Result], metrics: Metrics):
                 f"{rule.id}: {rule.cwe.name}",
             )
 
-            if result.location.url is not None:
-                file_name = result.location.url
+            if result.artifact.uri is not None:
+                file_name = result.artifact.uri
             else:
-                file_name = result.location.file_name
+                file_name = result.artifact.file_name
 
             # TODO(ericwb): replace hardcoded <module> with actual scope
             self.console.print(
                 f'  File "{file_name}", line '
                 f"{result.location.start_line}, in <module>",
             )
-            code_lines = result.location.snippet.splitlines(keepends=True)
+            code_lines = result.snippet.splitlines(keepends=True)
             code_line = code_lines[1] if len(code_lines) > 1 else code_lines[0]
             underline_width = (
                 result.location.end_column - result.location.start_column

precli/rules/go/stdlib/crypto_weak_cipher.py

@@ -1,4 +1,4 @@
-# Copyright 2023 Secure Saurce LLC
+# Copyright 2024 Secure Saurce LLC
 r"""
 ==================================================================
 Use of a Broken or Risky Cryptographic Algorithm in Crypto Package
@@ -158,10 +158,8 @@ def analyze(self, context: dict, **kwargs: dict) -> Result:
             )
             return Result(
                 rule_id=self.id,
-                location=Location(
-                    file_name=context["file_name"],
-                    node=call.function_node,
-                ),
+                artifact=context["artifact"],
+                location=Location(node=call.function_node),
                 level=Level.ERROR,
                 message=self.message.format(call.name),
                 fixes=fixes,

precli/rules/go/stdlib/crypto_weak_hash.py

@@ -1,4 +1,4 @@
-# Copyright 2023 Secure Saurce LLC
+# Copyright 2024 Secure Saurce LLC
 r"""
 =========================================
 Reversible One Way Hash in Crypto Package
@@ -106,10 +106,8 @@ def analyze(self, context: dict, **kwargs: dict) -> Result:
             )
             return Result(
                 rule_id=self.id,
-                location=Location(
-                    file_name=context["file_name"],
-                    node=call.function_node,
-                ),
+                artifact=context["artifact"],
+                location=Location(node=call.function_node),
                 level=Level.ERROR,
                 message=self.message.format(call.name_qualified),
                 fixes=fixes,
@@ -126,10 +124,8 @@ def analyze(self, context: dict, **kwargs: dict) -> Result:
             )
             return Result(
                 rule_id=self.id,
-                location=Location(
-                    file_name=context["file_name"],
-                    node=call.function_node,
-                ),
+                artifact=context["artifact"],
+                location=Location(node=call.function_node),
                 level=Level.ERROR,
                 message=self.message.format(call.name_qualified),
                 fixes=fixes,

precli/rules/go/stdlib/crypto_weak_key.py

@@ -1,4 +1,4 @@
-# Copyright 2023 Secure Saurce LLC
+# Copyright 2024 Secure Saurce LLC
 r"""
 ================================================================
 Inadequate Encryption Strength Using Weak Keys in Crypto Package
@@ -125,10 +125,8 @@ def analyze(self, context: dict, **kwargs: dict) -> Result:
 
                 return Result(
                     rule_id=self.id,
-                    location=Location(
-                        file_name=context["file_name"],
-                        node=argument.identifier_node,
-                    ),
+                    artifact=context["artifact"],
+                    location=Location(node=argument.identifier_node),
                     level=Level.ERROR,
                     message=self.message.format("DSA", 2048),
                     fixes=fixes,
@@ -147,10 +145,8 @@ def analyze(self, context: dict, **kwargs: dict) -> Result:
 
                 return Result(
                     rule_id=self.id,
-                    location=Location(
-                        file_name=context["file_name"],
-                        node=argument.node,
-                    ),
+                    artifact=context["artifact"],
+                    location=Location(node=argument.node),
                     level=Level.ERROR if bits <= 1024 else Level.WARNING,
                     message=self.message.format("RSA", 2048),
                     fixes=fixes,