fix(gitlab): stabilize report data

Author: lelia

socketsecurity/core/__init__.py

@@ -1431,17 +1431,38 @@ def get_repo_info(self, repo_slug: str, default_branch: str = "socket-default-br
 
         return response.data
 
-    def get_head_scan_for_repo(self, repo_slug: str) -> str:
+    def get_head_scan_for_repo(
+            self,
+            repo_slug: str,
+            workspace: Optional[str] = None,
+            scan_type: Optional[str] = None,
+    ) -> Optional[str]:
         """
         Gets the head scan ID for a repository.
 
         Args:
             repo_slug: Repository slug to get head scan for
+            workspace: Socket workspace the scan belongs to, if any
+            scan_type: Socket scan type to match, if any
 
         Returns:
             Head scan ID if it exists, None otherwise
         """
         repo_info = self.get_repo_info(repo_slug)
+        if workspace:
+            query_params = {
+                "repo": repo_slug,
+                "workspace": workspace,
+                "branch": repo_info.default_branch,
+                "sort": "created_at",
+                "direction": "desc",
+                "per_page": 1,
+            }
+            if scan_type:
+                query_params["scan_type"] = scan_type
+            response = self.sdk.fullscans.get(self.config.org_slug, query_params)
+            results = response.get("results") if isinstance(response, dict) else None
+            return results[0].get("id") if results else None
         return repo_info.head_full_scan_id if repo_info.head_full_scan_id else None
 
     def get_full_scan_id_by_commit(
@@ -1528,7 +1549,11 @@ def resolve_base_full_scan_id(self, params: FullScanParams) -> Optional[str]:
             return scan_id
 
         try:
-            return self.get_head_scan_for_repo(params.repo)
+            return self.get_head_scan_for_repo(
+                params.repo,
+                workspace=params.workspace,
+                scan_type=params.scan_type,
+            )
         except APIResourceNotFound:
             return None
 

socketsecurity/core/classes.py

@@ -153,18 +153,16 @@ def from_socket_artifact(cls, data: dict) -> "Package":
         Returns:
             New Package instance
         """
-        purl = f"{data['type']}/"
-        namespace = data.get("namespace")
-        if namespace:
-            purl += f"{namespace}@"
-        purl += f"{data['name']}@{data['version']}"
-        base_url = "https://socket.dev"
-        url = f"{base_url}/{data['type']}/package/{namespace or ''}{data['name']}/overview/{data['version']}"
+        package_type = getattr(data["type"], "value", data["type"])
+        namespace = (data.get("namespace") or "").strip("/")
+        package_path = "/".join(part for part in (namespace, data["name"]) if part)
+        purl = f"{package_type}/{package_path}@{data['version']}"
+        url = f"https://socket.dev/{package_type}/package/{package_path}/overview/{data['version']}"
         return cls(
             id=data["id"],
             name=data["name"],
             version=data["version"],
-            type=data["type"],
+            type=package_type,
             release=data.get("release"),
             diffType=data.get("diffType"),
             score=data["score"],
@@ -179,7 +177,7 @@ def from_socket_artifact(cls, data: dict) -> "Package":
             artifact=data.get("artifact"),
             purl=purl,
             url=url,
-            namespace=namespace
+            namespace=namespace or None
         )
 
     @classmethod

socketsecurity/core/messages.py

@@ -5,6 +5,7 @@
 import uuid
 from datetime import datetime, timezone
 from pathlib import Path
+
 from mdutils import MdUtils
 from prettytable import PrettyTable
 
@@ -655,25 +656,31 @@ def extract_identifiers_gitlab(alert: Issue) -> list:
             "url": alert.url if hasattr(alert, 'url') and alert.url else None
         })
 
-        # Extract CVE identifiers from props
-        if hasattr(alert, 'props') and alert.props:
-            if 'cve' in alert.props:
-                cves = alert.props['cve']
-                if isinstance(cves, list):
-                    for cve in cves:
-                        identifiers.append({
-                            "type": "cve",
-                            "name": cve,
-                            "value": cve,
-                            "url": f"https://cve.mitre.org/cgi-bin/cvename.cgi?name={cve}"
-                        })
-                elif isinstance(cves, str):
-                    identifiers.append({
-                        "type": "cve",
-                        "name": cves,
-                        "value": cves,
-                        "url": f"https://cve.mitre.org/cgi-bin/cvename.cgi?name={cves}"
-                    })
+        props = getattr(alert, "props", None) or {}
+        identifier_fields = (
+            ("cveId", "cve", "https://nvd.nist.gov/vuln/detail/"),
+            ("cve", "cve", "https://nvd.nist.gov/vuln/detail/"),
+            ("ghsaId", "ghsa", "https://github.com/advisories/"),
+        )
+        seen = set()
+        for field, identifier_type, url_prefix in identifier_fields:
+            values = props.get(field, [])
+            if isinstance(values, str):
+                values = [values]
+            for value in values or []:
+                if not isinstance(value, str) or not value.strip():
+                    continue
+                value = value.strip()
+                identifier_key = (identifier_type, value.upper())
+                if identifier_key in seen:
+                    continue
+                seen.add(identifier_key)
+                identifiers.append({
+                    "type": identifier_type,
+                    "name": value,
+                    "value": value,
+                    "url": f"{url_prefix}{value}"
+                })
 
         return identifiers
 

tests/core/test_package_and_alerts.py

@@ -1,8 +1,9 @@
-from dataclasses import dataclass
+from dataclasses import asdict, dataclass
 from unittest.mock import Mock
 
 import pytest
 from socketdev import socketdev
+from socketdev.fullscans import SocketArtifact
 
 from socketsecurity.core import Core, _humanize_alert_type
 from socketsecurity.core.classes import Issue, Package
@@ -104,6 +105,27 @@ def test_create_packages_dict_basic(self, core):
         assert pkg.version == "1.0.0"
         assert pkg.transitives == 0
 
+    def test_full_scan_package_normalizes_enum_type_and_namespace_url(self):
+        artifact = SocketArtifact.from_dict({
+            "id": "pkg:maven/com.arenko/trading-core@1.2.3",
+            "type": "maven",
+            "namespace": "com.arenko",
+            "name": "trading-core",
+            "version": "1.2.3",
+            "direct": True,
+            "topLevelAncestors": [],
+            "manifestFiles": [{"file": "pom.xml"}],
+            "alerts": [],
+        })
+
+        package = Package.from_socket_artifact(asdict(artifact))
+
+        assert package.type == "maven"
+        assert package.purl == "maven/com.arenko/trading-core@1.2.3"
+        assert package.url == (
+            "https://socket.dev/maven/package/com.arenko/trading-core/overview/1.2.3"
+        )
+
     def test_create_packages_dict_with_transitives(self, core):
         """Test package dictionary creation with transitive dependencies"""
         mock_artifacts = [
@@ -340,4 +362,3 @@ def test_empty_input_returns_empty_string(self):
     def test_handles_acronyms_conservatively(self):
         """Adjacent capitals are kept together: SQLInjection -> 'SQL Injection'."""
         assert _humanize_alert_type("SQLInjection") == "SQL Injection"
-

tests/core/test_sdk_methods.py

@@ -63,6 +63,35 @@ def test_get_head_scan_for_repo_no_head(core, mock_sdk_with_responses):
     head_scan_id = core.get_head_scan_for_repo("no-head")
     assert head_scan_id is None
 
+
+def test_get_head_scan_for_repo_scopes_workspace_to_default_branch(
+        core, mock_sdk_with_responses
+):
+    mock_sdk_with_responses.fullscans.get.return_value = {
+        "results": [{"id": "workspace-head"}],
+        "nextPage": None,
+    }
+
+    head_scan_id = core.get_head_scan_for_repo(
+        "test",
+        workspace="customer-a",
+        scan_type="socket_tier1",
+    )
+
+    assert head_scan_id == "workspace-head"
+    mock_sdk_with_responses.fullscans.get.assert_called_once_with(
+        core.config.org_slug,
+        {
+            "repo": "test",
+            "workspace": "customer-a",
+            "branch": "main",
+            "sort": "created_at",
+            "direction": "desc",
+            "per_page": 1,
+            "scan_type": "socket_tier1",
+        },
+    )
+
 def test_get_full_scan_id_by_commit(core, mock_sdk_with_responses):
     """Looks up the newest full scan for a repo + commit via the list endpoint"""
     mock_sdk_with_responses.fullscans.get.return_value = {
@@ -126,6 +155,29 @@ def test_resolve_base_full_scan_id_defaults_to_head_scan(core):
     """Without base overrides the repository head scan is the baseline"""
     assert core.resolve_base_full_scan_id(make_full_scan_params()) == "head"
 
+
+def test_resolve_base_full_scan_id_scopes_head_to_workspace(core):
+    core.sdk.fullscans.get.return_value = {
+        "results": [{"id": "workspace-head"}],
+        "nextPage": None,
+    }
+
+    params = make_full_scan_params(workspace="customer-a", scan_type="socket_tier1")
+
+    assert core.resolve_base_full_scan_id(params) == "workspace-head"
+    core.sdk.fullscans.get.assert_called_once_with(
+        core.config.org_slug,
+        {
+            "repo": "test",
+            "workspace": "customer-a",
+            "branch": "main",
+            "sort": "created_at",
+            "direction": "desc",
+            "per_page": 1,
+            "scan_type": "socket_tier1",
+        },
+    )
+
 def test_resolve_base_full_scan_id_uses_base_scan_id(core):
     """--base-scan-id is used verbatim, without touching the repo endpoint"""
     core.cli_config = make_cli_config("--base-scan-id", "explicit-base")

tests/unit/test_gitlab_format.py

@@ -1,8 +1,7 @@
 import re
 
-import pytest
-from socketsecurity.core.messages import Messages
 from socketsecurity.core.classes import Diff, Issue
+from socketsecurity.core.messages import Messages
 
 
 class TestGitLabFormat:
@@ -87,7 +86,10 @@ def test_identifier_extraction_with_cve(self):
             type="vulnerability",
             severity="critical",
             title="Known CVE",
-            props={"cve": ["CVE-2024-5678", "CVE-2024-9012"]},
+            props={
+                "cveId": ["CVE-2024-5678", "CVE-2024-9012"],
+                "ghsaId": "GHSA-1234-5678-9012",
+            },
             pkg_type="npm",
             key="test-key",
             purl="pkg:npm/vulnerable-pkg@2.0.0"
@@ -97,15 +99,17 @@ def test_identifier_extraction_with_cve(self):
         report = Messages.create_security_comment_gitlab(diff)
         vuln = report["vulnerabilities"][0]
 
-        # Should have socket_alert identifier + 2 CVE identifiers
-        assert len(vuln["identifiers"]) >= 3
+        # Should have socket_alert identifier + CVE and GHSA identifiers
+        assert len(vuln["identifiers"]) == 4
         cve_identifiers = [i for i in vuln["identifiers"] if i["type"] == "cve"]
         assert len(cve_identifiers) == 2
         assert any(i["value"] == "CVE-2024-5678" for i in cve_identifiers)
         assert any(i["value"] == "CVE-2024-9012" for i in cve_identifiers)
+        ghsa_identifiers = [i for i in vuln["identifiers"] if i["type"] == "ghsa"]
+        assert ghsa_identifiers[0]["value"] == "GHSA-1234-5678-9012"
 
     def test_identifier_extraction_with_single_cve_string(self):
-        """Test single CVE identifier as string"""
+        """Legacy CVE property remains supported"""
         diff = Diff()
         diff.id = "test-scan-id"
         diff.diff_url = "https://socket.dev/test"
@@ -130,6 +134,23 @@ def test_identifier_extraction_with_single_cve_string(self):
         assert len(cve_identifiers) == 1
         assert cve_identifiers[0]["value"] == "CVE-2024-1111"
 
+    def test_identifier_extraction_deduplicates_legacy_and_current_cve_fields(self):
+        issue = Issue(
+            pkg_name="vulnerable-pkg",
+            pkg_version="2.0.0",
+            type="vulnerability",
+            severity="high",
+            title="Duplicate CVE",
+            props={"cve": "CVE-2024-1111", "cveId": "CVE-2024-1111"},
+            pkg_type="npm",
+            key="test-key",
+            purl="pkg:npm/vulnerable-pkg@2.0.0",
+        )
+
+        identifiers = Messages.extract_identifiers_gitlab(issue)
+
+        assert [item["value"] for item in identifiers].count("CVE-2024-1111") == 1
+
     def test_dependency_chain_handling_transitive(self):
         """Test transitive dependency path is captured"""
         diff = Diff()