def test_construct_version_query_4():
    """Test the GraphPopulator.construct_version_query() class method."""
    input_json = {
        "version": "0.4.59",
        "package": "access_points",
        "ecosystem": "pypi",
        "analyses": {
            "metadata": {
                "details": [{
                    "description": "Some description here",
                    "declared_license": "GPL and\nv2.0"
                }]
            }
        }
    }
    q = GraphPopulator.construct_version_query(input_json)
    logger.info(q)

    assert "access_points" in q
    assert "0.4.59" in q
    assert "pypi" in q

    input_json = {
        "version": "deb579d6e030503f430978ee229008b9bc912d40",
        "package": "github.com/gorilla/mux",
        "ecosystem": "go",
        "analyses": {
            "source_licenses": {
                "status": "success",
                "summary": {
                    "sure_licenses": ["BSD-Modified"]
                }
            },
            "metadata": {
                "details": [{
                    "code_repository": {
                        "type": "git",
                        "url": "https://github.com/gorilla/mux"
                    },
                    "dependencies": [],
                    "ecosystem":
                    "gofedlib",
                    "name":
                    "github.com/gorilla/mux",
                    "version":
                    "deb579d6e030503f430978ee229008b9bc912d40"
                }]
            }
        }
    }
    q = GraphPopulator.construct_version_query(input_json)

    assert "'declared_licenses'" in q
    assert "'licenses'" in q
    assert "BSD-Modified" in q
 def create_pv_nodes(self):
     """Create Package and Version nodes, if needed."""
     nodes = [
     ]  # return (e, p, v) tuples of created/existing nodes; for easier testing
     affected_pkgs = {}
     all_epvs_created = True
     for pv_dict in self._cve_dict.get('affected'):
         epv_dict = pv_dict.copy()
         epv_dict['ecosystem'] = self._cve_dict.get('ecosystem')
         query = GraphPopulator.construct_graph_nodes(epv_dict)
         latest_version = "-1"
         # Fetch the value of the latest_version from the query created
         if "latest_version" in query:
             data = query.split("\'latest_version\'")[1].split(");")[0]
             latest_version = data.replace(",", "").strip().replace("'", "")
         success, json_response = BayesianGraph.execute(query)
         e = epv_dict.get('ecosystem')
         p = epv_dict.get('name')
         v = epv_dict.get('version')
         if p not in affected_pkgs:
             tmp = {"ecosystem": e, "latest_version": latest_version}
             affected_pkgs[p] = tmp
         if not success:
             logger.error(
                 'CVEIngestionError - Error creating nodes for {e}/{p}/{v}: {r}'
                 .format(e=e, p=p, v=v, r=str(json_response)))
             all_epvs_created = False
         else:
             nodes.append((e, p, v))
     return nodes, all_epvs_created, affected_pkgs
Beispiel #3
0
 def create_pv_nodes(self):
     """Create Package and Version nodes, if needed."""
     nodes = []  # return (e, p, v) tuples of created/existing nodes; for easier testing
     affected_pkgs = {}
     all_epvs_created = True
     for pv_dict in self._cve_dict.get('affected'):
         epv_dict = pv_dict.copy()
         epv_dict['ecosystem'] = self._cve_dict.get('ecosystem')
         query, bindings = GraphPopulator.construct_graph_nodes(epv_dict)
         # Fetch the value of the latest_version from the query created
         latest_version = -1
         if "latest" in bindings:
             latest_version = bindings['latest']
         success, json_response = BayesianGraph.execute(self.prepare_payload(query, bindings))
         e = epv_dict.get('ecosystem')
         p = epv_dict.get('name')
         v = epv_dict.get('version')
         if p not in affected_pkgs:
             tmp = {
                 "ecosystem": e,
                 "latest_version": latest_version
             }
             affected_pkgs[p] = tmp
         if not success:
             logger.error('CVEIngestionError - Error creating nodes for {e}/{p}/{v}: {r}'.format(
                 e=e, p=p, v=v, r=str(json_response))
             )
             all_epvs_created = False
         else:
             nodes.append((e, p, v))
     return nodes, all_epvs_created, affected_pkgs
    def create_pv_nodes(self):
        """Create Package and Version nodes, if needed."""
        nodes = [
        ]  # return (e, p, v) tuples of created/existing nodes; for easier testing
        affected_pkgs = {}
        all_epvs_created = True
        p = self._cve_dict.get('package')
        e = self._cve_dict.get('ecosystem')
        epv_dict = {
            "ecosystem": self._cve_dict.get('ecosystem'),
            "name": self._cve_dict.get('package')
        }
        latest_version = ""
        for ver in self._cve_dict.get('affected'):
            epv_dict['version'] = ver
            query = GraphPopulator.construct_graph_nodes(epv_dict)
            success, json_response = BayesianGraph.execute(query)
            # Fetch the value of the latest_version from the query create
            if not latest_version and "latest_version" in query:
                data = query.split("\'latest_version\'")[1].split(");")[0]
                latest_version = data.replace(",", "").strip().replace("'", "")

            if not success:
                logger.error(
                    'CVEIngestionError - Error creating nodes for {e}/{p}/{v}: {r}'
                    .format(e=e, p=p, v=ver, r=str(json_response)))
                all_epvs_created = False
            else:
                nodes.append((e, p, ver))

        if p not in affected_pkgs:
            affected_pkg = {"ecosystem": e, "latest_version": latest_version}
            affected_pkgs[p] = affected_pkg

        # To create the latest version node if not present
        if latest_version and latest_version != "-1":
            epv_dict['version'] = latest_version
            query = GraphPopulator.construct_graph_nodes(epv_dict)
            BayesianGraph.execute(query)
        return nodes, all_epvs_created, affected_pkgs
def test_construct_version_query_2():
    """Test the GraphPopulator.construct_version_query() class method."""
    input_json = {
        "version": "0.4.59",
        "package": "access_points",
        "ecosystem": "pypi",
        "analyses": {
            "metadata": {
                "details": [{
                    "description": "Some description here",
                    "declared_licenses": ["GPL v3", "APL v2.0"]
                }]
            },
            "github_details": {},
            'libraries_io': {},
            'source_licenses': {},
            'security_issues': {
                "details": [{
                    "id": "CEV-007",
                    "cvss": {
                        "score": 9.7
                    }
                }]
            },
            'code_metrics': {
                "details": {
                    "languages": [{
                        "metrics": {
                            "functions": {
                                'average_cyclomatic_complexity': 3
                            }
                        }
                    }]
                }
            },
            'redhat_downstream': {
                "summary": {
                    "all_rhsm_product_names": ["access_points_rh"]
                }
            }
        }
    }
    q = GraphPopulator.construct_version_query(input_json)
    logger.info(q)

    assert "access_points" in q
    assert "0.4.59" in q
    assert "pypi" in q
def test_construct_version_query_1():
    """Test the GraphPopulator.construct_version_query() class method."""
    input_json = {
        "version": "0.4.59",
        "package": "access_points",
        "ecosystem": "pypi",
        "analyses": {
            "metadata": {}
        }
    }
    q = GraphPopulator.construct_version_query(input_json)
    logger.info(q)

    assert "access_points" in q
    assert "0.4.59" in q
    assert "pypi" in q
    assert "addVertex" in q
    assert "drop()" not in q
Beispiel #7
0
def test_construct_version_query_3():
    """Test the GraphPopulator.construct_version_query() class method."""
    input_json = {
        "version": "0.4.59",
        "package": "access_points",
        "ecosystem": "pypi",
        "analyses": {
            "metadata": {"details": [
                {"description": "Some description here",
                 "declared_license": "GPL \nv2.0"}
            ]}
        }
    }
    q = GraphPopulator.construct_version_query(input_json)
    logger.info(q)

    assert "access_points" in q
    assert "0.4.59" in q
    assert "pypi" in q
Beispiel #8
0
 def create_pv_nodes(self):
     """Create Package and Version nodes, if needed."""
     nodes = [
     ]  # return (e, p, v) tuples of created/existing nodes; for easier testing
     all_epvs_created = True
     for pv_dict in self._cve_dict.get('affected'):
         epv_dict = pv_dict.copy()
         epv_dict['ecosystem'] = self._cve_dict.get('ecosystem')
         query = GraphPopulator.construct_graph_nodes(epv_dict)
         success, json_response = BayesianGraph.execute(query)
         e = epv_dict.get('ecosystem')
         p = epv_dict.get('name')
         v = epv_dict.get('version')
         if not success:
             logger.error(
                 'CVEIngestionError - Error creating nodes for {e}/{p}/{v}: {r}'
                 .format(e=e, p=p, v=v, r=str(json_response)))
             all_epvs_created = False
         else:
             nodes.append((e, p, v))
     return nodes, all_epvs_created
Beispiel #9
0
def create_graph_nodes(list_epv):
    """Create blank graph nodes given an EPV."""
    count_blank_epvs_created = 0
    success_epvs = []
    failure_epvs = []

    for item in list_epv:
        str_gremlin, bindings = GraphPopulator.construct_graph_nodes(item)
        epv = item.get('ecosystem') + ":" + item.get('name') + ":" + item.get(
            'version')

        if str_gremlin:
            payload = {'gremlin': str_gremlin, "bindings": bindings}
            print(json.dumps(payload))
            try:
                result = requests.post(config.GREMLIN_SERVER_URL_REST,
                                       data=json.dumps(payload),
                                       timeout=30)
                resp = result.json()
                print(json.dumps(resp))

                if resp['status']['code'] == 200:
                    count_blank_epvs_created += 1
                    success_epvs.append(epv)
            except Exception as e:  # pragma: no cover
                logger.error(e)
                failure_json = {epv: e}
                failure_epvs.append(failure_json)

    status = "Success"
    if count_blank_epvs_created == 0:
        status = "Failure"

    response = {
        "epv_nodes_created": count_blank_epvs_created,
        "success_list": success_epvs,
        "failure_list": failure_epvs,
        "status": status
    }
    return response
Beispiel #10
0
def _import_keys_from_s3_http(data_source, epv_list):
    # TODO: reduce cyclomatic complexity
    logger.debug("Begin import...")
    report = {'status': 'Success', 'message': 'The import finished successfully!'}
    count_imported_EPVs = 0
    last_imported_EPV = None
    epv = []
    for epv_key in epv_list:
        for key, contents in epv_key.items():
            if len(contents.get('pkg_list_keys')) == 0 and len(contents.get('ver_list_keys')) == 0:
                report['message'] = 'Nothing to be imported! No data found on S3 to be imported!'
                continue
            pkg_ecosystem = contents.get('ecosystem')
            pkg_name = contents.get('package')
            pkg_version = contents.get('version') or ''
            pkg_source = contents.get('source_repo', pkg_ecosystem)

            obj = {
                'ecosystem': pkg_ecosystem,
                'package': pkg_name,
                'version': pkg_version,
                'source_repo': pkg_source}

            latest_version = get_latest_versions_for_ep(pkg_ecosystem, pkg_name)
            latest_epv_list = [{
                'ecosystem': pkg_ecosystem,
                'name': pkg_name,
                'version': latest_version
            }]
            create_graph_nodes(latest_epv_list)

            try:
                # Check other Version level information and add it to common object
                if len(contents.get('ver_list_keys')) > 0:
                    first_key = contents['ver_key_prefix'] + '.json'
                    first_obj = _first_key_info(data_source, first_key, config.AWS_EPV_BUCKET)
                    first_obj['latest_version'] = latest_version
                    obj.update(first_obj)
                    ver_obj = _other_key_info(data_source, contents.get('ver_list_keys'),
                                              config.AWS_EPV_BUCKET)
                    if 'analyses' in obj:
                        obj.get('analyses', {}).update(ver_obj['analyses'])
                    else:
                        obj.update(ver_obj)

                # Check Package related information and add it to package object
                if len(contents.get('pkg_list_keys')) > 0:
                    pkg_obj = _other_key_info(data_source, contents.get('pkg_list_keys'),
                                              config.AWS_PKG_BUCKET)
                    if 'analyses' in obj:
                        obj.get('analyses', {}).update(pkg_obj['analyses'])
                    else:
                        obj.update(pkg_obj)

                # Create Gremlin Query
                str_gremlin = GraphPopulator.create_query_string(obj)

                if str_gremlin:
                    # Fire Gremlin HTTP query now
                    epv_full = pkg_ecosystem + ":" + pkg_name + ":" + pkg_version
                    logger.info("Ingestion initialized for EPV - %s" % epv_full)
                    epv.append(epv_full)
                    payload = {'gremlin': str_gremlin}
                    response = requests.post(config.GREMLIN_SERVER_URL_REST,
                                             data=json.dumps(payload), timeout=30)
                    resp = response.json()

                    if resp['status']['code'] == 200:
                        count_imported_EPVs += 1
                        last_imported_EPV = (obj.get('ecosystem') + ":" + obj.get('package') +
                                             ":" + obj.get('version'))

                        # update first key with graph synced tag
                        logger.info("Mark as synced in RDS %s" % last_imported_EPV)
                        if not config.AWS_S3_IS_LOCAL:  # pragma: no cover
                            PostgresHandler().mark_epv_synced(
                                obj.get('ecosystem'),
                                obj.get('package'),
                                obj.get('version')
                            )

            except Exception as e:  # pragma: no cover
                logger.error(e)
                msg = _get_exception_msg("The import failed", e)
                report['status'] = 'Failure'
                report['message'] = msg
                report['epv'] = epv_key

    report['epv'] = epv_list
    report['count_imported_EPVs'] = count_imported_EPVs
    if count_imported_EPVs == 0 and report['status'] == 'Success':
        report['message'] = 'Nothing to be synced to Graph!'
    report['last_imported_EPV'] = last_imported_EPV

    return report
Beispiel #11
0
    def create_pv_nodes(self):
        """Create Package and Version nodes, if needed."""
        nodes = []  # return (e, p, v) tuples of created/existing nodes; for easier testing
        affected_pkgs = {}
        all_epvs_created = True
        p = self._snyk_pkg_data.get('package')
        e = self._snyk_pkg_data.get('ecosystem')
        latest_version = self._snyk_pkg_data.get('latest_version')
        latest_non_cve_version = ''
        epv_dict = {
            "ecosystem": e,
            "name": p,
            "latest_version": latest_version
        }
        if latest_version not in self._snyk_pkg_data.get('affected'):
            logger.info("Latest version is not affected {}".format(p))
            latest_non_cve_version = latest_version
        else:
            logger.info("Latest version is affected {p} {v}".format(p=p, v=latest_version))

        if e == 'golang':
            itr_list = self._snyk_pkg_data.get('all_ver')
            epv_dict['gh_link'] = self._snyk_pkg_data.get('gh_link')
            epv_dict['license'] = self._snyk_pkg_data.get('license')
        else:
            itr_list = self._snyk_pkg_data.get('affected')

        for ver in itr_list:
            epv_dict['version'] = ver
            query = GraphPopulator.construct_graph_nodes(epv_dict)
            success, json_response = BayesianGraph.execute(query)
            # Fetch the value of the latest_version from the query create
            if not latest_version and "latest_version" in query:
                data = query.split("\'latest_version\'")[1].split(");")[0]
                latest_version = data.replace(",", "").strip().replace("'", "")

            if not success:
                logger.error('CVEIngestionError - Error creating nodes for {e}/{p}/{v}: {r}'.format(
                    e=e, p=p, v=ver, r=str(json_response))
                )
                all_epvs_created = False
            else:
                nodes.append((e, p, ver))

        # To create the latest version node if not present
        if latest_version and latest_version != "-1" and e != "golang":
            epv_dict['version'] = latest_version
            logger.info("Creating latest version node {e} {p} {v}".format(e=epv_dict['ecosystem'],
                                                                          p=epv_dict['name'],
                                                                          v=epv_dict['version']))
            query = GraphPopulator.construct_graph_nodes(epv_dict)
            BayesianGraph.execute(query)

        res = ""
        if latest_non_cve_version:
            res = update_non_cve_on_pkg(e, p, latest_non_cve_version)

        if p not in affected_pkgs and res != "Success":
            affected_pkg = {
                "ecosystem": e,
                "latest_version": latest_version
            }
            affected_pkgs[p] = affected_pkg
        return nodes, all_epvs_created, affected_pkgs
Beispiel #12
0
def handle_properties(ecosystem, package, version):
    """
    Handle (update/delete) properties associated with given EPV.

    Update replaces properties with the same name.

    Expects JSON payload in following format:
    {
        "properties": [
            {
                "name": "cve_ids",
                "value": "CVE-3005-0001:10"
            }
        ]
    }

    "value" can be omitted in DELETE requests.

    :param ecosystem: str, ecosystem
    :param package: str, package name
    :param version: str, package version
    :return: 200 on success, 400 on failure
    """
    # TODO: reduce cyclomatic complexity
    input_json = request.get_json()
    properties = input_json.get('properties')

    error = flask.jsonify({'error': 'invalid input'})
    if not properties:
        return error, 400

    input_json = {
        k: GraphPopulator.sanitize_text_for_query(str(v))
        for k, v in input_json.items()
    }

    if request.method == 'PUT':
        if [
                x for x in properties
                if not x.get('name') or x.get('value') is None
        ]:
            return error, 400

    log_msg = '[{m}] Updating properties for {e}/{p}/{v} with payload {b}'
    current_app.logger.info(
        log_msg.format(m=request.method,
                       e=ecosystem,
                       p=package,
                       v=version,
                       b=input_json))

    query_statement = "g.V()" \
                      ".has('pecosystem','{ecosystem}')" \
                      ".has('pname','{pkg_name}')" \
                      ".has('version','{version}')".format(ecosystem=ecosystem,
                                                           pkg_name=package,
                                                           version=version)
    statement = ''

    if request.method in ('DELETE', 'PUT'):
        # build "delete" part of the statement
        drop_str = ""
        for prop in properties:
            drop_str += query_statement
            drop_str += ".properties('{property}').drop().iterate();".format(
                property=prop['name'])
        statement += drop_str

    if request.method == 'PUT':
        # build "add" part of the statement
        add_str = ""
        for prop in properties:
            add_str += ".property('{property}','{value}')".format(
                property=prop['name'], value=prop['value'])
        statement += query_statement + add_str + ';'

    current_app.logger.info('Gremlin statement: {s}'.format(s=statement))
    success, response_json = BayesianGraph.execute(statement)
    if not success:
        current_app.logger.error(
            "Failed to update properties for {e}/{p}/{v}".format(e=ecosystem,
                                                                 p=package,
                                                                 v=version))
        return flask.jsonify(response_json), 400

    return flask.jsonify(response_json), 200
def test_construct_package_query():
    """Test the GraphPopulator.construct_package_query() class method."""
    input_json = {
        "version": "0.4.59",
        "package": "access_points",
        "ecosystem": "pypi",
        "analyses": {
            "metadata": {
                "details": [{
                    "description": "Some description here"
                }]
            },
            "github_details": {},
            'libraries_io': {
                'schema': {
                    'version': '2-0-0'
                },
                'details': {
                    'releases': {
                        'count': 2,
                        'recent': [{
                            "published_at": "2016-09-09"
                        }],
                        "published_at": "2016-09-09"
                    }
                }
            }
        }
    }
    str_package, prp_package = GraphPopulator.construct_package_query(
        input_json)
    logger.info(str_package)
    logger.info(prp_package)

    assert 'access_points' in str_package
    assert "pypi" in str_package

    input_json = {
        "version": "0.4.59",
        "package": "access_points",
        "ecosystem": "pypi",
        "analyses": {
            "metadata": {
                "details": [{
                    "description": "Some description here"
                }]
            },
            "github_details": {},
            'libraries_io': {}
        }
    }
    str_package, prp_package = GraphPopulator.construct_package_query(
        input_json)
    logger.info(str_package)
    logger.info(prp_package)

    assert 'access_points' in str_package
    assert "pypi" in str_package

    input_json = {
        "version": "0.4.59",
        "package": "access_points",
        "ecosystem": "pypi",
        "analyses": {
            "metadata": {
                "details": [{
                    "description": "Some description here"
                }]
            },
            "github_details": {},
            'libraries_io': {
                'schema': {
                    'version': '1-0-0'
                },
                'details': {
                    'releases': {
                        'count': 2,
                        'recent': [{
                            "published_at": "2016-09-09"
                        }],
                        'latest': {
                            'recent': {
                                "0.4.59": "2016-09-09"
                            }
                        },
                        "published_at": "2016-09-09"
                    }
                }
            }
        }
    }
    str_package, prp_package = GraphPopulator.construct_package_query(
        input_json)
    logger.info(str_package)
    logger.info(prp_package)

    assert 'access_points' in str_package
    assert "pypi" in str_package