Merge pull request #10 from DavidCruciani/gharchive-inclusion

gharchive process

bin/finder.py

@@ -14,7 +14,7 @@ import json
 import sys
 import argparse
 
-from git_vuln_finder import find, find_vuln, summary
+from git_vuln_finder import find, find_event
 
 
 def main():
@@ -45,24 +45,36 @@ def main():
     parser.add_argument(
         "-t", help="Include tags matching a specific commit", action="store_true"
     )
+    parser.add_argument(
+        "-gh", help="special option for gharchive, pass a file containing a PushEvent in JSON format"
+    )
     args = parser.parse_args()
 
     if args.p not in ["vulnpatterns", "cryptopatterns", "cpatterns", "all"]:
         parser.print_usage()
         parser.exit()
 
-    if not args.r:
+    if not args.r and not args.gh:
         parser.print_usage()
         parser.exit()
 
-    # Launch the process
-    all_potential_vulnerabilities, all_cve_found, found = find(
-        args.r,
-        tags_matching=args.t,
-        commit_state=args.s,
-        verbose=args.v,
-        defaultpattern=args.p,
-    )
+    if args.gh:
+        with open(args.gh, "r") as read_file:
+            event = json.load(read_file)
+
+        for element in event:
+            for i in range(0,len(element["payload"]["commits"])):
+                all_potential_vulnerabilities, all_cve_found, found = find_event(element["payload"]["commits"][i], element)
+
+    else:
+        # Launch the process
+        all_potential_vulnerabilities, all_cve_found, found = find(
+            args.r,
+            tags_matching=args.t,
+            commit_state=args.s,
+            verbose=args.v,
+            defaultpattern=args.p,
+        )
 
     # Output the result as json. Can be piped to another software.
     if not args.c:

git_vuln_finder/__init__.py

@@ -1,6 +1,9 @@
 from git_vuln_finder.pattern import build_pattern
 from git_vuln_finder.pattern import get_patterns
 from git_vuln_finder.vulnerability import find_vuln
+from git_vuln_finder.vulnerability import find_vuln_event
 from git_vuln_finder.vulnerability import summary
+from git_vuln_finder.vulnerability import summary_event
 from git_vuln_finder.vulnerability import extract_cve
 from git_vuln_finder.run import find
+from git_vuln_finder.run import find_event

git_vuln_finder/run.py

@@ -12,7 +12,7 @@
 import sys
 import git
 import typing
-from git_vuln_finder import get_patterns, find_vuln, summary
+from git_vuln_finder import get_patterns, find_vuln, find_vuln_event, summary, summary_event
 
 
 def find(
@@ -95,3 +95,32 @@ def find(
                         found += 1
 
         return all_potential_vulnerabilities, all_cve_found, found
+
+def find_event(commit, element):
+    # Initialization of the variables for the results
+    found = 0
+    all_potential_vulnerabilities = {}
+    all_cve_found = set()
+
+    # Initialization of the patterns
+    patterns = get_patterns()
+    vulnpatterns = patterns["en"]["medium"]["vuln"]
+    cryptopatterns = patterns["en"]["medium"]["crypto"]
+    cpatterns = patterns["en"]["medium"]["c"]
+
+    defaultpattern = [vulnpatterns, cryptopatterns, cpatterns]
+    
+    for p in defaultpattern:
+        ret = find_vuln_event(commit["message"], pattern=p)
+        if ret:
+            potential_vulnerabilities, cve_found = summary_event(
+                commit,
+                p,
+                element,
+                vuln_match=ret["match"]
+            )
+            all_potential_vulnerabilities.update(potential_vulnerabilities)
+            all_cve_found.update(cve_found)
+            found += 1
+
+    return all_potential_vulnerabilities, all_cve_found, found

git_vuln_finder/vulnerability.py

@@ -31,6 +31,23 @@ def find_vuln(commit, pattern, verbose=False):
     else:
         return None
 
+def find_vuln_event(commit_msg, pattern, verbose=False):
+    """Find a potential vulnerability from a commit message thanks to a regex
+    pattern.
+    """
+    m = pattern.search(commit_msg)
+    if m:
+        if verbose:
+            print("Match found: {}".format(m.group(0)), file=sys.stderr)
+            print(commit_msg, file=sys.stderr)
+            print("---", file=sys.stderr)
+        ret = {}
+        ret["commit"] = commit_msg
+        ret["match"] = m.groups()
+        return ret
+    else:
+        return None
+
 
 def summary(
     repo,
@@ -103,6 +120,38 @@ def summary(
     return rcommit.hexsha, potential_vulnerabilities, cve_found
 
 
+def summary_event(
+    commit,
+    pattern,
+    element,
+    vuln_match=None,
+    commit_state="under-review"
+):
+    potential_vulnerabilities = {}
+    
+    cve, cve_found = extract_cve(commit["message"])
+
+    potential_vulnerabilities[commit["sha"]] = {}
+    potential_vulnerabilities[commit["sha"]]["repo_name"] = element["repo"]["name"]
+    potential_vulnerabilities[commit["sha"]]["message"] = commit["message"]
+    potential_vulnerabilities[commit["sha"]]["language"] = langdetect(commit["message"])
+    potential_vulnerabilities[commit["sha"]]["commit-id"] = commit["sha"]
+    potential_vulnerabilities[commit["sha"]]["author"] = commit["author"]["name"]
+    potential_vulnerabilities[commit["sha"]]["author-email"] = commit["author"]["email"]
+    potential_vulnerabilities[commit["sha"]]["authored_date"] = element["created_at"]
+    potential_vulnerabilities[commit["sha"]]["branches"] = element["payload"]["ref"]
+    potential_vulnerabilities[commit["sha"]]["pattern-selected"] = pattern.pattern
+    potential_vulnerabilities[commit["sha"]]["pattern-matches"] = vuln_match
+    potential_vulnerabilities[commit["sha"]]["origin-github-api"] = commit["url"]
+    if cve:
+        potential_vulnerabilities[commit["sha"]]["cve"] = cve
+        potential_vulnerabilities[commit["sha"]]["state"] = "cve-assigned"
+    else:
+        potential_vulnerabilities[commit["sha"]]["state"] = commit_state
+
+    return potential_vulnerabilities, cve_found
+
+
 def extract_cve(commit):
     cve_found = set()
     cve_find = re.compile(r"CVE-[1-2]\d{1,4}-\d{1,7}", re.IGNORECASE)

tests/gharchive_test.json

@@ -0,0 +1,48 @@
+[
+   {
+    "id": "19351512310",
+    "type": "PushEvent",
+    "actor": {
+      "id": 32466128,
+      "login": "DavidCruciani",
+      "display_login": "DavidCruciani",
+      "gravatar_id": "",
+      "url": "https://api.github.com/users/DavidCruciani",
+      "avatar_url": "https://avatars.githubusercontent.com/u/32466128?"
+    },
+    "repo": {
+      "id": 424660123,
+      "name": "ail-project/ail-feeder-gharchive",
+      "url": "https://api.github.com/repos/ail-project/ail-feeder-gharchive"
+    },
+    "payload": {
+      "push_id": 8628652926,
+      "size": 1,
+      "distinct_size": 1,
+      "ref": "refs/heads/main",
+      "head": "910ed71a2819546a3f3bcce1ebb9e3984a8c8d86",
+      "before": "40a9ef5dc6b2add5184a0a58401bfe9058faa8df",
+      "commits": [
+        {
+          "sha": "910ed71a2819546a3f3bcce1ebb9e3984a8c8d86",
+          "author": {
+            "email": "da.cruciani@laposte.net",
+            "name": "David Cruciani"
+          },
+          "message": "chg: [feeder] case sensitive",
+          "distinct": true,
+          "url": "https://api.github.com/repos/ail-project/ail-feeder-gharchive/commits/910ed71a2819546a3f3bcce1ebb9e3984a8c8d86"
+        }
+      ]
+    },
+    "public": true,
+    "created_at": "2021-12-15T16:06:43Z",
+    "org": {
+      "id": 62389074,
+      "login": "ail-project",
+      "gravatar_id": "",
+      "url": "https://api.github.com/orgs/ail-project",
+      "avatar_url": "https://avatars.githubusercontent.com/u/62389074?"
+    }
+  }
+]