Update BaseNpmLockHandler to handle latest pacakge-lock.json file

src/packagedcode/npm.py

@@ -715,7 +715,15 @@ def parse(cls, location, package_only=False):
                 else:
                     name = dep
             ns, _ , name = name.rpartition('/')
-            version = dep_data.get('version')
+            version_string = dep_data.get('version')
+            version_info = parse_npm_version(version_string)
+
+            extra_data = {}
+            if version_info['type'] != "semver":
+                extra_data['version_type'] = version_info['type'],
+                extra_data['url'] = version_info['url']
+
+            version = get_version(version_info)
 
             dep_purl = PackageURL(
                 type=cls.default_package_type,
@@ -733,6 +741,8 @@ def parse(cls, location, package_only=False):
                 is_pinned=True,
                 is_direct=False,
             )
+            if extra_data:
+                dependency.extra_data = extra_data
 
             # URLs and checksums
             misc = get_urls(ns, name, version)
@@ -1878,6 +1888,46 @@ def deps_mapper(deps, package, field_name, is_direct=True):
 
     return package
 
+def parse_npm_version(version_string):
+    version_string = version_string.strip()
+
+    if version_string.startswith("git+"):
+        type = 'git'
+
+    elif version_string.startswith("https:") or version_string.startswith("http:"):
+        type = 'remote-tarball'
+
+    elif version_string.startswith("file:"):
+        type = 'local'
+
+    elif re.match(r"^\d+\.\d+\.\d+", version_string):
+        type = "semver"
+
+    else:
+        type = 'unknown'
+
+    return {'type': type, 'url': version_string}
+
+def get_version(version_info):
+    type = version_info['type']
+    version_string = version_info['url']
+
+    version = None
+
+    if type == 'semver':
+        version = version_string
+
+    elif type == 'remote-tarball':
+        version_match = re.search(r'[-_](\d+\.\d+\.\d+[^/]*)', version_string)
+        if version_match:
+            version = version_match.group(1)
+            version = re.sub(r'\.tgz$', '', version)
+
+    elif type == 'git':
+        if '#' in version_string:
+            _, _, version = version_string.rpartition('#')
+
+    return version
 
 person_parser = re.compile(
     r'^(?P<name>[^\(<]+)'