[PATCH] cve-update-nvd2-native: new CVE database fetcher


Marta Rybczynska
 

Add new fetcher for the NVD database using the 2.0 API [1].
The implementation changes as little as possible, keeping the current
database format (but using a different database file for the transition
period), with a notable exception of not using the META table.

Minor changes that could be visible:
- the database starts in 1999 instead of 2002
- the complete fetch is longer (30 minutes typically)

[1] https://nvd.nist.gov/developers/vulnerabilities

Signed-off-by: Marta Rybczynska <marta.rybczynska@...>
---
meta/classes/cve-check.bbclass | 4 +-
.../meta/cve-update-nvd2-native.bb | 333 ++++++++++++++++++
2 files changed, 335 insertions(+), 2 deletions(-)
create mode 100644 meta/recipes-core/meta/cve-update-nvd2-native.bb

diff --git a/meta/classes/cve-check.bbclass b/meta/classes/cve-check.bbclass
index 41fdf8363f..1f16683c32 100644
--- a/meta/classes/cve-check.bbclass
+++ b/meta/classes/cve-check.bbclass
@@ -32,7 +32,7 @@ CVE_PRODUCT ??= "${BPN}"
CVE_VERSION ??= "${PV}"

CVE_CHECK_DB_DIR ?= "${DL_DIR}/CVE_CHECK"
-CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_1.1.db"
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
CVE_CHECK_DB_FILE_LOCK ?= "${CVE_CHECK_DB_FILE}.lock"

CVE_CHECK_LOG ?= "${T}/cve.log"
@@ -161,7 +161,7 @@ python do_cve_check () {
}

addtask cve_check before do_build
-do_cve_check[depends] = "cve-update-db-native:do_fetch"
+do_cve_check[depends] = "cve-update-nvd2-native:do_fetch"
do_cve_check[nostamp] = "1"

python cve_check_cleanup () {
diff --git a/meta/recipes-core/meta/cve-update-nvd2-native.bb b/meta/recipes-core/meta/cve-update-nvd2-native.bb
new file mode 100644
index 0000000000..1c14481c21
--- /dev/null
+++ b/meta/recipes-core/meta/cve-update-nvd2-native.bb
@@ -0,0 +1,333 @@
+SUMMARY = "Updates the NVD CVE database"
+LICENSE = "MIT"
+
+# Important note:
+# This product uses the NVD API but is not endorsed or certified by the NVD.
+
+INHIBIT_DEFAULT_DEPS = "1"
+
+inherit native
+
+deltask do_unpack
+deltask do_patch
+deltask do_configure
+deltask do_compile
+deltask do_install
+deltask do_populate_sysroot
+
+NVDCVE_URL ?= "https://services.nvd.nist.gov/rest/json/cves/2.0"
+
+# CVE database update interval, in seconds. By default: once a day (24*60*60).
+# Use 0 to force the update
+# Use a negative value to skip the update
+CVE_DB_UPDATE_INTERVAL ?= "86400"
+
+# Timeout for blocking socket operations, such as the connection attempt.
+CVE_SOCKET_TIMEOUT ?= "60"
+
+CVE_DB_TEMP_FILE ?= "${CVE_CHECK_DB_DIR}/temp_nvdcve_2.db"
+
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
+
+python () {
+ if not bb.data.inherits_class("cve-check", d):
+ raise bb.parse.SkipRecipe("Skip recipe when cve-check class is not loaded.")
+}
+
+python do_fetch() {
+ """
+ Update NVD database with API 2.0
+ """
+ import bb.utils
+ import bb.progress
+ import shutil
+
+ bb.utils.export_proxies(d)
+
+ db_file = d.getVar("CVE_CHECK_DB_FILE")
+ db_dir = os.path.dirname(db_file)
+ db_tmp_file = d.getVar("CVE_DB_TEMP_FILE")
+
+ cleanup_db_download(db_file, db_tmp_file)
+ # By default let's update the whole database (since time 0)
+ database_time = 0
+
+ # The NVD database changes once a day, so no need to update more frequently
+ # Allow the user to force-update
+ try:
+ import time
+ update_interval = int(d.getVar("CVE_DB_UPDATE_INTERVAL"))
+ if update_interval < 0:
+ bb.note("CVE database update skipped")
+ return
+ if time.time() - os.path.getmtime(db_file) < update_interval:
+ bb.note("CVE database recently updated, skipping")
+ return
+ database_time = os.path.getmtime(db_file)
+
+ except OSError:
+ pass
+
+ bb.utils.mkdirhier(db_dir)
+ if os.path.exists(db_file):
+ shutil.copy2(db_file, db_tmp_file)
+
+ if update_db_file(db_tmp_file, d, database_time) == True:
+ # Update downloaded correctly, can swap files
+ shutil.move(db_tmp_file, db_file)
+ else:
+ # Update failed, do not modify the database
+ bb.warn("CVE database update failed")
+ os.remove(db_tmp_file)
+}
+
+do_fetch[lockfiles] += "${CVE_CHECK_DB_FILE_LOCK}"
+do_fetch[file-checksums] = ""
+do_fetch[vardeps] = ""
+
+def cleanup_db_download(db_file, db_tmp_file):
+ """
+ Cleanup the download space from possible failed downloads
+ """
+
+ # Clean up the updates done on the main file
+ # Remove it only if a journal file exists - it means a complete re-download
+ if os.path.exists("{0}-journal".format(db_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_file))
+
+ if os.path.exists(db_file):
+ os.remove(db_file)
+
+ # Clean-up the temporary file downloads, we can remove both journal
+ # and the temporary database
+ if os.path.exists("{0}-journal".format(db_tmp_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_tmp_file))
+
+ if os.path.exists(db_tmp_file):
+ os.remove(db_tmp_file)
+
+def nvd_request_next(url, api_key, args):
+ """
+ Request next part of the NVD dabase
+ """
+
+ import urllib.request
+ import urllib.parse
+ import gzip
+
+ headers = {}
+ if api_key:
+ headers['apiKey'] = api_key
+
+ data = urllib.parse.urlencode(args)
+
+ full_request = url + '?' + data
+
+ for attempt in range(3):
+ try:
+ r = urllib.request.urlopen(full_request)
+
+ if (r.headers['content-encoding'] == 'gzip'):
+ buf = r.read()
+ raw_data = gzip.decompress(buf)
+ else:
+ raw_data = r.read().decode("utf-8")
+
+ r.close()
+
+ except UnicodeDecodeError:
+ # Received garbage, retry
+ bb.debug(2, "CVE database: received malformed data, retrying (request: %s)" %(full_request))
+ pass
+ except http.client.IncompleteRead:
+ # Read incomplete, let's try again
+ bb.debug(2, "CVE database: received incomplete data, retrying (request: %s)" %(full_request))
+ pass
+ else:
+ return raw_data
+ else:
+ # We failed at all attempts
+ return None
+
+def update_db_file(db_tmp_file, d, database_time):
+ """
+ Update the given database file
+ """
+ import bb.utils, bb.progress
+ import datetime
+ import sqlite3
+ import json
+
+ # Connect to database
+ conn = sqlite3.connect(db_tmp_file)
+ initialize_db(conn)
+
+ req_args = {'startIndex' : 0}
+
+ # The maximum range for time is 120 days
+ # Force a complete update if our range is longer
+ if (database_time != 0):
+ database_date = datetime.datetime.combine(datetime.date.fromtimestamp(database_time), datetime.time())
+ today_date = datetime.datetime.combine(datetime.date.today(), datetime.time())
+ delta = today_date - database_date
+ if delta.days < 120:
+ bb.debug(2, "CVE database: performing partial update")
+ req_args['lastModStartDate'] = database_date.isoformat()
+ req_args['lastModEndDate'] = today_date.isoformat()
+ else:
+ bb.note("CVE database: file too old, forcing a full update")
+
+ with bb.progress.ProgressHandler(d) as ph, open(os.path.join(d.getVar("TMPDIR"), 'cve_check'), 'a') as cve_f:
+
+ bb.debug(2, "Updating entries")
+ index = 0
+ url = d.getVar("NVDCVE_URL")
+ while True:
+ req_args['startIndex'] = index
+ raw_data = nvd_request_next(url, None, req_args)
+ if raw_data is None:
+ # We haven't managed to download data
+ return False
+
+ data = json.loads(raw_data)
+
+ index = data["startIndex"]
+ total = data["totalResults"]
+ per_page = data["resultsPerPage"]
+
+ for cve in data["vulnerabilities"]:
+ update_db(conn, cve)
+
+ index += per_page
+ ph.update((float(index) / (total+1)) * 100)
+ if index >= total:
+ break
+
+ # Recommended by NVD
+ time.sleep(6)
+
+ # Update success, set the date to cve_check file.
+ cve_f.write('CVE database update : %s\n\n' % datetime.date.today())
+
+ conn.commit()
+ conn.close()
+ return True
+
+def initialize_db(conn):
+ with conn:
+ c = conn.cursor()
+
+ c.execute("CREATE TABLE IF NOT EXISTS META (YEAR INTEGER UNIQUE, DATE TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS NVD (ID TEXT UNIQUE, SUMMARY TEXT, \
+ SCOREV2 TEXT, SCOREV3 TEXT, MODIFIED INTEGER, VECTOR TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS PRODUCTS (ID TEXT, \
+ VENDOR TEXT, PRODUCT TEXT, VERSION_START TEXT, OPERATOR_START TEXT, \
+ VERSION_END TEXT, OPERATOR_END TEXT)")
+ c.execute("CREATE INDEX IF NOT EXISTS PRODUCT_ID_IDX on PRODUCTS(ID);")
+
+ c.close()
+
+def parse_node_and_insert(conn, node, cveId):
+
+ def cpe_generator():
+ for cpe in node.get('cpeMatch', ()):
+ if not cpe['vulnerable']:
+ return
+ cpe23 = cpe.get('criteria')
+ if not cpe23:
+ return
+ cpe23 = cpe23.split(':')
+ if len(cpe23) < 6:
+ return
+ vendor = cpe23[3]
+ product = cpe23[4]
+ version = cpe23[5]
+
+ if cpe23[6] == '*' or cpe23[6] == '-':
+ version_suffix = ""
+ else:
+ version_suffix = "_" + cpe23[6]
+
+ if version != '*' and version != '-':
+ # Version is defined, this is a '=' match
+ yield [cveId, vendor, product, version + version_suffix, '=', '', '']
+ elif version == '-':
+ # no version information is available
+ yield [cveId, vendor, product, version, '', '', '']
+ else:
+ # Parse start version, end version and operators
+ op_start = ''
+ op_end = ''
+ v_start = ''
+ v_end = ''
+
+ if 'versionStartIncluding' in cpe:
+ op_start = '>='
+ v_start = cpe['versionStartIncluding']
+
+ if 'versionStartExcluding' in cpe:
+ op_start = '>'
+ v_start = cpe['versionStartExcluding']
+
+ if 'versionEndIncluding' in cpe:
+ op_end = '<='
+ v_end = cpe['versionEndIncluding']
+
+ if 'versionEndExcluding' in cpe:
+ op_end = '<'
+ v_end = cpe['versionEndExcluding']
+
+ if op_start or op_end or v_start or v_end:
+ yield [cveId, vendor, product, v_start, op_start, v_end, op_end]
+ else:
+ # This is no version information, expressed differently.
+ # Save processing by representing as -.
+ yield [cveId, vendor, product, '-', '', '', '']
+
+ conn.executemany("insert into PRODUCTS values (?, ?, ?, ?, ?, ?, ?)", cpe_generator()).close()
+
+def update_db(conn, elt):
+ """
+ Update a single entry in the on-disk database
+ """
+
+ accessVector = None
+ cveId = elt['cve']['id']
+ if elt['cve']['vulnStatus'] == "Rejected":
+ return
+ cveDesc = ""
+ for desc in elt['cve']['descriptions']:
+ if desc['lang'] == 'en':
+ cveDesc = desc['value']
+ date = elt['cve']['lastModified']
+ try:
+ accessVector = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['accessVector']
+ cvssv2 = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['baseScore']
+ except KeyError:
+ cvssv2 = 0.0
+ try:
+ accessVector = accessVector or elt['impact']['baseMetricV3']['cvssV3']['attackVector']
+ cvssv3 = elt['impact']['baseMetricV3']['cvssV3']['baseScore']
+ except KeyError:
+ accessVector = accessVector or "UNKNOWN"
+ cvssv3 = 0.0
+
+ conn.execute("insert or replace into NVD values (?, ?, ?, ?, ?, ?)",
+ [cveId, cveDesc, cvssv2, cvssv3, date, accessVector]).close()
+
+ try:
+ configurations = elt['cve']['configurations'][0]['nodes']
+ for config in configurations:
+ parse_node_and_insert(conn, config, cveId)
+ except KeyError:
+ bb.debug(2, "Entry without a configuration")
+
+do_fetch[nostamp] = "1"
+
+EXCLUDE_FROM_WORLD = "1"
--
2.35.1


Marta Rybczynska
 



On Wed, 29 Mar 2023, 12:03 Marta Rybczynska via lists.openembedded.org, <rybczynska=gmail.com@...> wrote:
Add new fetcher for the NVD database using the 2.0 API [1].
The implementation changes as little as possible, keeping the current
database format (but using a different database file for the transition
period), with a notable exception of not using the META table.

Minor changes that could be visible:
- the database starts in 1999 instead of 2002
- the complete fetch is longer (30 minutes typically)

Ross, Richard,
This is the most up-to-date version with urllib and all download errors I managed to run into.

Kind regards 
Marta
 


Alexandre Belloni
 

Hello Marta,

This fails oe-selftest with:

ERROR: cve-update-nvd2-native-1.0-r0 do_fetch: QA Issue: Recipe cve-update-nvd2-native in /home/pokybuild/yocto-worker/oe-selftest-debian/build/meta/recipes-core/meta/cve-update-nvd2-native.bb does not have an assigned maintainer. Please add an entry into meta/conf/distro/include/maintainers.inc. [missing-metadata]
ERROR: cve-update-nvd2-native-1.0-r0 do_fetch: Fatal QA errors were found, failing task.

On 29/03/2023 12:02:56+0200, Marta Rybczynska wrote:
Add new fetcher for the NVD database using the 2.0 API [1].
The implementation changes as little as possible, keeping the current
database format (but using a different database file for the transition
period), with a notable exception of not using the META table.

Minor changes that could be visible:
- the database starts in 1999 instead of 2002
- the complete fetch is longer (30 minutes typically)

[1] https://nvd.nist.gov/developers/vulnerabilities

Signed-off-by: Marta Rybczynska <marta.rybczynska@...>
---
meta/classes/cve-check.bbclass | 4 +-
.../meta/cve-update-nvd2-native.bb | 333 ++++++++++++++++++
2 files changed, 335 insertions(+), 2 deletions(-)
create mode 100644 meta/recipes-core/meta/cve-update-nvd2-native.bb

diff --git a/meta/classes/cve-check.bbclass b/meta/classes/cve-check.bbclass
index 41fdf8363f..1f16683c32 100644
--- a/meta/classes/cve-check.bbclass
+++ b/meta/classes/cve-check.bbclass
@@ -32,7 +32,7 @@ CVE_PRODUCT ??= "${BPN}"
CVE_VERSION ??= "${PV}"

CVE_CHECK_DB_DIR ?= "${DL_DIR}/CVE_CHECK"
-CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_1.1.db"
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
CVE_CHECK_DB_FILE_LOCK ?= "${CVE_CHECK_DB_FILE}.lock"

CVE_CHECK_LOG ?= "${T}/cve.log"
@@ -161,7 +161,7 @@ python do_cve_check () {
}

addtask cve_check before do_build
-do_cve_check[depends] = "cve-update-db-native:do_fetch"
+do_cve_check[depends] = "cve-update-nvd2-native:do_fetch"
do_cve_check[nostamp] = "1"

python cve_check_cleanup () {
diff --git a/meta/recipes-core/meta/cve-update-nvd2-native.bb b/meta/recipes-core/meta/cve-update-nvd2-native.bb
new file mode 100644
index 0000000000..1c14481c21
--- /dev/null
+++ b/meta/recipes-core/meta/cve-update-nvd2-native.bb
@@ -0,0 +1,333 @@
+SUMMARY = "Updates the NVD CVE database"
+LICENSE = "MIT"
+
+# Important note:
+# This product uses the NVD API but is not endorsed or certified by the NVD.
+
+INHIBIT_DEFAULT_DEPS = "1"
+
+inherit native
+
+deltask do_unpack
+deltask do_patch
+deltask do_configure
+deltask do_compile
+deltask do_install
+deltask do_populate_sysroot
+
+NVDCVE_URL ?= "https://services.nvd.nist.gov/rest/json/cves/2.0"
+
+# CVE database update interval, in seconds. By default: once a day (24*60*60).
+# Use 0 to force the update
+# Use a negative value to skip the update
+CVE_DB_UPDATE_INTERVAL ?= "86400"
+
+# Timeout for blocking socket operations, such as the connection attempt.
+CVE_SOCKET_TIMEOUT ?= "60"
+
+CVE_DB_TEMP_FILE ?= "${CVE_CHECK_DB_DIR}/temp_nvdcve_2.db"
+
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
+
+python () {
+ if not bb.data.inherits_class("cve-check", d):
+ raise bb.parse.SkipRecipe("Skip recipe when cve-check class is not loaded.")
+}
+
+python do_fetch() {
+ """
+ Update NVD database with API 2.0
+ """
+ import bb.utils
+ import bb.progress
+ import shutil
+
+ bb.utils.export_proxies(d)
+
+ db_file = d.getVar("CVE_CHECK_DB_FILE")
+ db_dir = os.path.dirname(db_file)
+ db_tmp_file = d.getVar("CVE_DB_TEMP_FILE")
+
+ cleanup_db_download(db_file, db_tmp_file)
+ # By default let's update the whole database (since time 0)
+ database_time = 0
+
+ # The NVD database changes once a day, so no need to update more frequently
+ # Allow the user to force-update
+ try:
+ import time
+ update_interval = int(d.getVar("CVE_DB_UPDATE_INTERVAL"))
+ if update_interval < 0:
+ bb.note("CVE database update skipped")
+ return
+ if time.time() - os.path.getmtime(db_file) < update_interval:
+ bb.note("CVE database recently updated, skipping")
+ return
+ database_time = os.path.getmtime(db_file)
+
+ except OSError:
+ pass
+
+ bb.utils.mkdirhier(db_dir)
+ if os.path.exists(db_file):
+ shutil.copy2(db_file, db_tmp_file)
+
+ if update_db_file(db_tmp_file, d, database_time) == True:
+ # Update downloaded correctly, can swap files
+ shutil.move(db_tmp_file, db_file)
+ else:
+ # Update failed, do not modify the database
+ bb.warn("CVE database update failed")
+ os.remove(db_tmp_file)
+}
+
+do_fetch[lockfiles] += "${CVE_CHECK_DB_FILE_LOCK}"
+do_fetch[file-checksums] = ""
+do_fetch[vardeps] = ""
+
+def cleanup_db_download(db_file, db_tmp_file):
+ """
+ Cleanup the download space from possible failed downloads
+ """
+
+ # Clean up the updates done on the main file
+ # Remove it only if a journal file exists - it means a complete re-download
+ if os.path.exists("{0}-journal".format(db_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_file))
+
+ if os.path.exists(db_file):
+ os.remove(db_file)
+
+ # Clean-up the temporary file downloads, we can remove both journal
+ # and the temporary database
+ if os.path.exists("{0}-journal".format(db_tmp_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_tmp_file))
+
+ if os.path.exists(db_tmp_file):
+ os.remove(db_tmp_file)
+
+def nvd_request_next(url, api_key, args):
+ """
+ Request next part of the NVD dabase
+ """
+
+ import urllib.request
+ import urllib.parse
+ import gzip
+
+ headers = {}
+ if api_key:
+ headers['apiKey'] = api_key
+
+ data = urllib.parse.urlencode(args)
+
+ full_request = url + '?' + data
+
+ for attempt in range(3):
+ try:
+ r = urllib.request.urlopen(full_request)
+
+ if (r.headers['content-encoding'] == 'gzip'):
+ buf = r.read()
+ raw_data = gzip.decompress(buf)
+ else:
+ raw_data = r.read().decode("utf-8")
+
+ r.close()
+
+ except UnicodeDecodeError:
+ # Received garbage, retry
+ bb.debug(2, "CVE database: received malformed data, retrying (request: %s)" %(full_request))
+ pass
+ except http.client.IncompleteRead:
+ # Read incomplete, let's try again
+ bb.debug(2, "CVE database: received incomplete data, retrying (request: %s)" %(full_request))
+ pass
+ else:
+ return raw_data
+ else:
+ # We failed at all attempts
+ return None
+
+def update_db_file(db_tmp_file, d, database_time):
+ """
+ Update the given database file
+ """
+ import bb.utils, bb.progress
+ import datetime
+ import sqlite3
+ import json
+
+ # Connect to database
+ conn = sqlite3.connect(db_tmp_file)
+ initialize_db(conn)
+
+ req_args = {'startIndex' : 0}
+
+ # The maximum range for time is 120 days
+ # Force a complete update if our range is longer
+ if (database_time != 0):
+ database_date = datetime.datetime.combine(datetime.date.fromtimestamp(database_time), datetime.time())
+ today_date = datetime.datetime.combine(datetime.date.today(), datetime.time())
+ delta = today_date - database_date
+ if delta.days < 120:
+ bb.debug(2, "CVE database: performing partial update")
+ req_args['lastModStartDate'] = database_date.isoformat()
+ req_args['lastModEndDate'] = today_date.isoformat()
+ else:
+ bb.note("CVE database: file too old, forcing a full update")
+
+ with bb.progress.ProgressHandler(d) as ph, open(os.path.join(d.getVar("TMPDIR"), 'cve_check'), 'a') as cve_f:
+
+ bb.debug(2, "Updating entries")
+ index = 0
+ url = d.getVar("NVDCVE_URL")
+ while True:
+ req_args['startIndex'] = index
+ raw_data = nvd_request_next(url, None, req_args)
+ if raw_data is None:
+ # We haven't managed to download data
+ return False
+
+ data = json.loads(raw_data)
+
+ index = data["startIndex"]
+ total = data["totalResults"]
+ per_page = data["resultsPerPage"]
+
+ for cve in data["vulnerabilities"]:
+ update_db(conn, cve)
+
+ index += per_page
+ ph.update((float(index) / (total+1)) * 100)
+ if index >= total:
+ break
+
+ # Recommended by NVD
+ time.sleep(6)
+
+ # Update success, set the date to cve_check file.
+ cve_f.write('CVE database update : %s\n\n' % datetime.date.today())
+
+ conn.commit()
+ conn.close()
+ return True
+
+def initialize_db(conn):
+ with conn:
+ c = conn.cursor()
+
+ c.execute("CREATE TABLE IF NOT EXISTS META (YEAR INTEGER UNIQUE, DATE TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS NVD (ID TEXT UNIQUE, SUMMARY TEXT, \
+ SCOREV2 TEXT, SCOREV3 TEXT, MODIFIED INTEGER, VECTOR TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS PRODUCTS (ID TEXT, \
+ VENDOR TEXT, PRODUCT TEXT, VERSION_START TEXT, OPERATOR_START TEXT, \
+ VERSION_END TEXT, OPERATOR_END TEXT)")
+ c.execute("CREATE INDEX IF NOT EXISTS PRODUCT_ID_IDX on PRODUCTS(ID);")
+
+ c.close()
+
+def parse_node_and_insert(conn, node, cveId):
+
+ def cpe_generator():
+ for cpe in node.get('cpeMatch', ()):
+ if not cpe['vulnerable']:
+ return
+ cpe23 = cpe.get('criteria')
+ if not cpe23:
+ return
+ cpe23 = cpe23.split(':')
+ if len(cpe23) < 6:
+ return
+ vendor = cpe23[3]
+ product = cpe23[4]
+ version = cpe23[5]
+
+ if cpe23[6] == '*' or cpe23[6] == '-':
+ version_suffix = ""
+ else:
+ version_suffix = "_" + cpe23[6]
+
+ if version != '*' and version != '-':
+ # Version is defined, this is a '=' match
+ yield [cveId, vendor, product, version + version_suffix, '=', '', '']
+ elif version == '-':
+ # no version information is available
+ yield [cveId, vendor, product, version, '', '', '']
+ else:
+ # Parse start version, end version and operators
+ op_start = ''
+ op_end = ''
+ v_start = ''
+ v_end = ''
+
+ if 'versionStartIncluding' in cpe:
+ op_start = '>='
+ v_start = cpe['versionStartIncluding']
+
+ if 'versionStartExcluding' in cpe:
+ op_start = '>'
+ v_start = cpe['versionStartExcluding']
+
+ if 'versionEndIncluding' in cpe:
+ op_end = '<='
+ v_end = cpe['versionEndIncluding']
+
+ if 'versionEndExcluding' in cpe:
+ op_end = '<'
+ v_end = cpe['versionEndExcluding']
+
+ if op_start or op_end or v_start or v_end:
+ yield [cveId, vendor, product, v_start, op_start, v_end, op_end]
+ else:
+ # This is no version information, expressed differently.
+ # Save processing by representing as -.
+ yield [cveId, vendor, product, '-', '', '', '']
+
+ conn.executemany("insert into PRODUCTS values (?, ?, ?, ?, ?, ?, ?)", cpe_generator()).close()
+
+def update_db(conn, elt):
+ """
+ Update a single entry in the on-disk database
+ """
+
+ accessVector = None
+ cveId = elt['cve']['id']
+ if elt['cve']['vulnStatus'] == "Rejected":
+ return
+ cveDesc = ""
+ for desc in elt['cve']['descriptions']:
+ if desc['lang'] == 'en':
+ cveDesc = desc['value']
+ date = elt['cve']['lastModified']
+ try:
+ accessVector = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['accessVector']
+ cvssv2 = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['baseScore']
+ except KeyError:
+ cvssv2 = 0.0
+ try:
+ accessVector = accessVector or elt['impact']['baseMetricV3']['cvssV3']['attackVector']
+ cvssv3 = elt['impact']['baseMetricV3']['cvssV3']['baseScore']
+ except KeyError:
+ accessVector = accessVector or "UNKNOWN"
+ cvssv3 = 0.0
+
+ conn.execute("insert or replace into NVD values (?, ?, ?, ?, ?, ?)",
+ [cveId, cveDesc, cvssv2, cvssv3, date, accessVector]).close()
+
+ try:
+ configurations = elt['cve']['configurations'][0]['nodes']
+ for config in configurations:
+ parse_node_and_insert(conn, config, cveId)
+ except KeyError:
+ bb.debug(2, "Entry without a configuration")
+
+do_fetch[nostamp] = "1"
+
+EXCLUDE_FROM_WORLD = "1"
--
2.35.1


--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com


Alexandre Belloni
 

On 31/03/2023 23:40:58+0200, Alexandre Belloni via lists.openembedded.org wrote:
Hello Marta,

This fails oe-selftest with:

ERROR: cve-update-nvd2-native-1.0-r0 do_fetch: QA Issue: Recipe cve-update-nvd2-native in /home/pokybuild/yocto-worker/oe-selftest-debian/build/meta/recipes-core/meta/cve-update-nvd2-native.bb does not have an assigned maintainer. Please add an entry into meta/conf/distro/include/maintainers.inc. [missing-metadata]
ERROR: cve-update-nvd2-native-1.0-r0 do_fetch: Fatal QA errors were found, failing task.
We have discussed this with Richard, this is from another patch in my
tree that is not going to be applied. No action on your part is needed.

On 29/03/2023 12:02:56+0200, Marta Rybczynska wrote:
Add new fetcher for the NVD database using the 2.0 API [1].
The implementation changes as little as possible, keeping the current
database format (but using a different database file for the transition
period), with a notable exception of not using the META table.

Minor changes that could be visible:
- the database starts in 1999 instead of 2002
- the complete fetch is longer (30 minutes typically)

[1] https://nvd.nist.gov/developers/vulnerabilities

Signed-off-by: Marta Rybczynska <marta.rybczynska@...>
---
meta/classes/cve-check.bbclass | 4 +-
.../meta/cve-update-nvd2-native.bb | 333 ++++++++++++++++++
2 files changed, 335 insertions(+), 2 deletions(-)
create mode 100644 meta/recipes-core/meta/cve-update-nvd2-native.bb

diff --git a/meta/classes/cve-check.bbclass b/meta/classes/cve-check.bbclass
index 41fdf8363f..1f16683c32 100644
--- a/meta/classes/cve-check.bbclass
+++ b/meta/classes/cve-check.bbclass
@@ -32,7 +32,7 @@ CVE_PRODUCT ??= "${BPN}"
CVE_VERSION ??= "${PV}"

CVE_CHECK_DB_DIR ?= "${DL_DIR}/CVE_CHECK"
-CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_1.1.db"
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
CVE_CHECK_DB_FILE_LOCK ?= "${CVE_CHECK_DB_FILE}.lock"

CVE_CHECK_LOG ?= "${T}/cve.log"
@@ -161,7 +161,7 @@ python do_cve_check () {
}

addtask cve_check before do_build
-do_cve_check[depends] = "cve-update-db-native:do_fetch"
+do_cve_check[depends] = "cve-update-nvd2-native:do_fetch"
do_cve_check[nostamp] = "1"

python cve_check_cleanup () {
diff --git a/meta/recipes-core/meta/cve-update-nvd2-native.bb b/meta/recipes-core/meta/cve-update-nvd2-native.bb
new file mode 100644
index 0000000000..1c14481c21
--- /dev/null
+++ b/meta/recipes-core/meta/cve-update-nvd2-native.bb
@@ -0,0 +1,333 @@
+SUMMARY = "Updates the NVD CVE database"
+LICENSE = "MIT"
+
+# Important note:
+# This product uses the NVD API but is not endorsed or certified by the NVD.
+
+INHIBIT_DEFAULT_DEPS = "1"
+
+inherit native
+
+deltask do_unpack
+deltask do_patch
+deltask do_configure
+deltask do_compile
+deltask do_install
+deltask do_populate_sysroot
+
+NVDCVE_URL ?= "https://services.nvd.nist.gov/rest/json/cves/2.0"
+
+# CVE database update interval, in seconds. By default: once a day (24*60*60).
+# Use 0 to force the update
+# Use a negative value to skip the update
+CVE_DB_UPDATE_INTERVAL ?= "86400"
+
+# Timeout for blocking socket operations, such as the connection attempt.
+CVE_SOCKET_TIMEOUT ?= "60"
+
+CVE_DB_TEMP_FILE ?= "${CVE_CHECK_DB_DIR}/temp_nvdcve_2.db"
+
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
+
+python () {
+ if not bb.data.inherits_class("cve-check", d):
+ raise bb.parse.SkipRecipe("Skip recipe when cve-check class is not loaded.")
+}
+
+python do_fetch() {
+ """
+ Update NVD database with API 2.0
+ """
+ import bb.utils
+ import bb.progress
+ import shutil
+
+ bb.utils.export_proxies(d)
+
+ db_file = d.getVar("CVE_CHECK_DB_FILE")
+ db_dir = os.path.dirname(db_file)
+ db_tmp_file = d.getVar("CVE_DB_TEMP_FILE")
+
+ cleanup_db_download(db_file, db_tmp_file)
+ # By default let's update the whole database (since time 0)
+ database_time = 0
+
+ # The NVD database changes once a day, so no need to update more frequently
+ # Allow the user to force-update
+ try:
+ import time
+ update_interval = int(d.getVar("CVE_DB_UPDATE_INTERVAL"))
+ if update_interval < 0:
+ bb.note("CVE database update skipped")
+ return
+ if time.time() - os.path.getmtime(db_file) < update_interval:
+ bb.note("CVE database recently updated, skipping")
+ return
+ database_time = os.path.getmtime(db_file)
+
+ except OSError:
+ pass
+
+ bb.utils.mkdirhier(db_dir)
+ if os.path.exists(db_file):
+ shutil.copy2(db_file, db_tmp_file)
+
+ if update_db_file(db_tmp_file, d, database_time) == True:
+ # Update downloaded correctly, can swap files
+ shutil.move(db_tmp_file, db_file)
+ else:
+ # Update failed, do not modify the database
+ bb.warn("CVE database update failed")
+ os.remove(db_tmp_file)
+}
+
+do_fetch[lockfiles] += "${CVE_CHECK_DB_FILE_LOCK}"
+do_fetch[file-checksums] = ""
+do_fetch[vardeps] = ""
+
+def cleanup_db_download(db_file, db_tmp_file):
+ """
+ Cleanup the download space from possible failed downloads
+ """
+
+ # Clean up the updates done on the main file
+ # Remove it only if a journal file exists - it means a complete re-download
+ if os.path.exists("{0}-journal".format(db_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_file))
+
+ if os.path.exists(db_file):
+ os.remove(db_file)
+
+ # Clean-up the temporary file downloads, we can remove both journal
+ # and the temporary database
+ if os.path.exists("{0}-journal".format(db_tmp_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_tmp_file))
+
+ if os.path.exists(db_tmp_file):
+ os.remove(db_tmp_file)
+
+def nvd_request_next(url, api_key, args):
+ """
+ Request next part of the NVD dabase
+ """
+
+ import urllib.request
+ import urllib.parse
+ import gzip
+
+ headers = {}
+ if api_key:
+ headers['apiKey'] = api_key
+
+ data = urllib.parse.urlencode(args)
+
+ full_request = url + '?' + data
+
+ for attempt in range(3):
+ try:
+ r = urllib.request.urlopen(full_request)
+
+ if (r.headers['content-encoding'] == 'gzip'):
+ buf = r.read()
+ raw_data = gzip.decompress(buf)
+ else:
+ raw_data = r.read().decode("utf-8")
+
+ r.close()
+
+ except UnicodeDecodeError:
+ # Received garbage, retry
+ bb.debug(2, "CVE database: received malformed data, retrying (request: %s)" %(full_request))
+ pass
+ except http.client.IncompleteRead:
+ # Read incomplete, let's try again
+ bb.debug(2, "CVE database: received incomplete data, retrying (request: %s)" %(full_request))
+ pass
+ else:
+ return raw_data
+ else:
+ # We failed at all attempts
+ return None
+
+def update_db_file(db_tmp_file, d, database_time):
+ """
+ Update the given database file
+ """
+ import bb.utils, bb.progress
+ import datetime
+ import sqlite3
+ import json
+
+ # Connect to database
+ conn = sqlite3.connect(db_tmp_file)
+ initialize_db(conn)
+
+ req_args = {'startIndex' : 0}
+
+ # The maximum range for time is 120 days
+ # Force a complete update if our range is longer
+ if (database_time != 0):
+ database_date = datetime.datetime.combine(datetime.date.fromtimestamp(database_time), datetime.time())
+ today_date = datetime.datetime.combine(datetime.date.today(), datetime.time())
+ delta = today_date - database_date
+ if delta.days < 120:
+ bb.debug(2, "CVE database: performing partial update")
+ req_args['lastModStartDate'] = database_date.isoformat()
+ req_args['lastModEndDate'] = today_date.isoformat()
+ else:
+ bb.note("CVE database: file too old, forcing a full update")
+
+ with bb.progress.ProgressHandler(d) as ph, open(os.path.join(d.getVar("TMPDIR"), 'cve_check'), 'a') as cve_f:
+
+ bb.debug(2, "Updating entries")
+ index = 0
+ url = d.getVar("NVDCVE_URL")
+ while True:
+ req_args['startIndex'] = index
+ raw_data = nvd_request_next(url, None, req_args)
+ if raw_data is None:
+ # We haven't managed to download data
+ return False
+
+ data = json.loads(raw_data)
+
+ index = data["startIndex"]
+ total = data["totalResults"]
+ per_page = data["resultsPerPage"]
+
+ for cve in data["vulnerabilities"]:
+ update_db(conn, cve)
+
+ index += per_page
+ ph.update((float(index) / (total+1)) * 100)
+ if index >= total:
+ break
+
+ # Recommended by NVD
+ time.sleep(6)
+
+ # Update success, set the date to cve_check file.
+ cve_f.write('CVE database update : %s\n\n' % datetime.date.today())
+
+ conn.commit()
+ conn.close()
+ return True
+
+def initialize_db(conn):
+ with conn:
+ c = conn.cursor()
+
+ c.execute("CREATE TABLE IF NOT EXISTS META (YEAR INTEGER UNIQUE, DATE TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS NVD (ID TEXT UNIQUE, SUMMARY TEXT, \
+ SCOREV2 TEXT, SCOREV3 TEXT, MODIFIED INTEGER, VECTOR TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS PRODUCTS (ID TEXT, \
+ VENDOR TEXT, PRODUCT TEXT, VERSION_START TEXT, OPERATOR_START TEXT, \
+ VERSION_END TEXT, OPERATOR_END TEXT)")
+ c.execute("CREATE INDEX IF NOT EXISTS PRODUCT_ID_IDX on PRODUCTS(ID);")
+
+ c.close()
+
+def parse_node_and_insert(conn, node, cveId):
+
+ def cpe_generator():
+ for cpe in node.get('cpeMatch', ()):
+ if not cpe['vulnerable']:
+ return
+ cpe23 = cpe.get('criteria')
+ if not cpe23:
+ return
+ cpe23 = cpe23.split(':')
+ if len(cpe23) < 6:
+ return
+ vendor = cpe23[3]
+ product = cpe23[4]
+ version = cpe23[5]
+
+ if cpe23[6] == '*' or cpe23[6] == '-':
+ version_suffix = ""
+ else:
+ version_suffix = "_" + cpe23[6]
+
+ if version != '*' and version != '-':
+ # Version is defined, this is a '=' match
+ yield [cveId, vendor, product, version + version_suffix, '=', '', '']
+ elif version == '-':
+ # no version information is available
+ yield [cveId, vendor, product, version, '', '', '']
+ else:
+ # Parse start version, end version and operators
+ op_start = ''
+ op_end = ''
+ v_start = ''
+ v_end = ''
+
+ if 'versionStartIncluding' in cpe:
+ op_start = '>='
+ v_start = cpe['versionStartIncluding']
+
+ if 'versionStartExcluding' in cpe:
+ op_start = '>'
+ v_start = cpe['versionStartExcluding']
+
+ if 'versionEndIncluding' in cpe:
+ op_end = '<='
+ v_end = cpe['versionEndIncluding']
+
+ if 'versionEndExcluding' in cpe:
+ op_end = '<'
+ v_end = cpe['versionEndExcluding']
+
+ if op_start or op_end or v_start or v_end:
+ yield [cveId, vendor, product, v_start, op_start, v_end, op_end]
+ else:
+ # This is no version information, expressed differently.
+ # Save processing by representing as -.
+ yield [cveId, vendor, product, '-', '', '', '']
+
+ conn.executemany("insert into PRODUCTS values (?, ?, ?, ?, ?, ?, ?)", cpe_generator()).close()
+
+def update_db(conn, elt):
+ """
+ Update a single entry in the on-disk database
+ """
+
+ accessVector = None
+ cveId = elt['cve']['id']
+ if elt['cve']['vulnStatus'] == "Rejected":
+ return
+ cveDesc = ""
+ for desc in elt['cve']['descriptions']:
+ if desc['lang'] == 'en':
+ cveDesc = desc['value']
+ date = elt['cve']['lastModified']
+ try:
+ accessVector = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['accessVector']
+ cvssv2 = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['baseScore']
+ except KeyError:
+ cvssv2 = 0.0
+ try:
+ accessVector = accessVector or elt['impact']['baseMetricV3']['cvssV3']['attackVector']
+ cvssv3 = elt['impact']['baseMetricV3']['cvssV3']['baseScore']
+ except KeyError:
+ accessVector = accessVector or "UNKNOWN"
+ cvssv3 = 0.0
+
+ conn.execute("insert or replace into NVD values (?, ?, ?, ?, ?, ?)",
+ [cveId, cveDesc, cvssv2, cvssv3, date, accessVector]).close()
+
+ try:
+ configurations = elt['cve']['configurations'][0]['nodes']
+ for config in configurations:
+ parse_node_and_insert(conn, config, cveId)
+ except KeyError:
+ bb.debug(2, "Entry without a configuration")
+
+do_fetch[nostamp] = "1"
+
+EXCLUDE_FROM_WORLD = "1"
--
2.35.1



--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com



--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com


Steve Sakoman
 

Hi Marta,

Is this safe to backport to the stable branches, or should I let it
"age" in master for a while?

Thanks!

Steve

On Wed, Mar 29, 2023 at 12:03 AM Marta Rybczynska <rybczynska@...> wrote:

Add new fetcher for the NVD database using the 2.0 API [1].
The implementation changes as little as possible, keeping the current
database format (but using a different database file for the transition
period), with a notable exception of not using the META table.

Minor changes that could be visible:
- the database starts in 1999 instead of 2002
- the complete fetch is longer (30 minutes typically)

[1] https://nvd.nist.gov/developers/vulnerabilities

Signed-off-by: Marta Rybczynska <marta.rybczynska@...>
---
meta/classes/cve-check.bbclass | 4 +-
.../meta/cve-update-nvd2-native.bb | 333 ++++++++++++++++++
2 files changed, 335 insertions(+), 2 deletions(-)
create mode 100644 meta/recipes-core/meta/cve-update-nvd2-native.bb

diff --git a/meta/classes/cve-check.bbclass b/meta/classes/cve-check.bbclass
index 41fdf8363f..1f16683c32 100644
--- a/meta/classes/cve-check.bbclass
+++ b/meta/classes/cve-check.bbclass
@@ -32,7 +32,7 @@ CVE_PRODUCT ??= "${BPN}"
CVE_VERSION ??= "${PV}"

CVE_CHECK_DB_DIR ?= "${DL_DIR}/CVE_CHECK"
-CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_1.1.db"
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
CVE_CHECK_DB_FILE_LOCK ?= "${CVE_CHECK_DB_FILE}.lock"

CVE_CHECK_LOG ?= "${T}/cve.log"
@@ -161,7 +161,7 @@ python do_cve_check () {
}

addtask cve_check before do_build
-do_cve_check[depends] = "cve-update-db-native:do_fetch"
+do_cve_check[depends] = "cve-update-nvd2-native:do_fetch"
do_cve_check[nostamp] = "1"

python cve_check_cleanup () {
diff --git a/meta/recipes-core/meta/cve-update-nvd2-native.bb b/meta/recipes-core/meta/cve-update-nvd2-native.bb
new file mode 100644
index 0000000000..1c14481c21
--- /dev/null
+++ b/meta/recipes-core/meta/cve-update-nvd2-native.bb
@@ -0,0 +1,333 @@
+SUMMARY = "Updates the NVD CVE database"
+LICENSE = "MIT"
+
+# Important note:
+# This product uses the NVD API but is not endorsed or certified by the NVD.
+
+INHIBIT_DEFAULT_DEPS = "1"
+
+inherit native
+
+deltask do_unpack
+deltask do_patch
+deltask do_configure
+deltask do_compile
+deltask do_install
+deltask do_populate_sysroot
+
+NVDCVE_URL ?= "https://services.nvd.nist.gov/rest/json/cves/2.0"
+
+# CVE database update interval, in seconds. By default: once a day (24*60*60).
+# Use 0 to force the update
+# Use a negative value to skip the update
+CVE_DB_UPDATE_INTERVAL ?= "86400"
+
+# Timeout for blocking socket operations, such as the connection attempt.
+CVE_SOCKET_TIMEOUT ?= "60"
+
+CVE_DB_TEMP_FILE ?= "${CVE_CHECK_DB_DIR}/temp_nvdcve_2.db"
+
+CVE_CHECK_DB_FILE ?= "${CVE_CHECK_DB_DIR}/nvdcve_2.db"
+
+python () {
+ if not bb.data.inherits_class("cve-check", d):
+ raise bb.parse.SkipRecipe("Skip recipe when cve-check class is not loaded.")
+}
+
+python do_fetch() {
+ """
+ Update NVD database with API 2.0
+ """
+ import bb.utils
+ import bb.progress
+ import shutil
+
+ bb.utils.export_proxies(d)
+
+ db_file = d.getVar("CVE_CHECK_DB_FILE")
+ db_dir = os.path.dirname(db_file)
+ db_tmp_file = d.getVar("CVE_DB_TEMP_FILE")
+
+ cleanup_db_download(db_file, db_tmp_file)
+ # By default let's update the whole database (since time 0)
+ database_time = 0
+
+ # The NVD database changes once a day, so no need to update more frequently
+ # Allow the user to force-update
+ try:
+ import time
+ update_interval = int(d.getVar("CVE_DB_UPDATE_INTERVAL"))
+ if update_interval < 0:
+ bb.note("CVE database update skipped")
+ return
+ if time.time() - os.path.getmtime(db_file) < update_interval:
+ bb.note("CVE database recently updated, skipping")
+ return
+ database_time = os.path.getmtime(db_file)
+
+ except OSError:
+ pass
+
+ bb.utils.mkdirhier(db_dir)
+ if os.path.exists(db_file):
+ shutil.copy2(db_file, db_tmp_file)
+
+ if update_db_file(db_tmp_file, d, database_time) == True:
+ # Update downloaded correctly, can swap files
+ shutil.move(db_tmp_file, db_file)
+ else:
+ # Update failed, do not modify the database
+ bb.warn("CVE database update failed")
+ os.remove(db_tmp_file)
+}
+
+do_fetch[lockfiles] += "${CVE_CHECK_DB_FILE_LOCK}"
+do_fetch[file-checksums] = ""
+do_fetch[vardeps] = ""
+
+def cleanup_db_download(db_file, db_tmp_file):
+ """
+ Cleanup the download space from possible failed downloads
+ """
+
+ # Clean up the updates done on the main file
+ # Remove it only if a journal file exists - it means a complete re-download
+ if os.path.exists("{0}-journal".format(db_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_file))
+
+ if os.path.exists(db_file):
+ os.remove(db_file)
+
+ # Clean-up the temporary file downloads, we can remove both journal
+ # and the temporary database
+ if os.path.exists("{0}-journal".format(db_tmp_file)):
+ # If a journal is present the last update might have been interrupted. In that case,
+ # just wipe any leftovers and force the DB to be recreated.
+ os.remove("{0}-journal".format(db_tmp_file))
+
+ if os.path.exists(db_tmp_file):
+ os.remove(db_tmp_file)
+
+def nvd_request_next(url, api_key, args):
+ """
+ Request next part of the NVD dabase
+ """
+
+ import urllib.request
+ import urllib.parse
+ import gzip
+
+ headers = {}
+ if api_key:
+ headers['apiKey'] = api_key
+
+ data = urllib.parse.urlencode(args)
+
+ full_request = url + '?' + data
+
+ for attempt in range(3):
+ try:
+ r = urllib.request.urlopen(full_request)
+
+ if (r.headers['content-encoding'] == 'gzip'):
+ buf = r.read()
+ raw_data = gzip.decompress(buf)
+ else:
+ raw_data = r.read().decode("utf-8")
+
+ r.close()
+
+ except UnicodeDecodeError:
+ # Received garbage, retry
+ bb.debug(2, "CVE database: received malformed data, retrying (request: %s)" %(full_request))
+ pass
+ except http.client.IncompleteRead:
+ # Read incomplete, let's try again
+ bb.debug(2, "CVE database: received incomplete data, retrying (request: %s)" %(full_request))
+ pass
+ else:
+ return raw_data
+ else:
+ # We failed at all attempts
+ return None
+
+def update_db_file(db_tmp_file, d, database_time):
+ """
+ Update the given database file
+ """
+ import bb.utils, bb.progress
+ import datetime
+ import sqlite3
+ import json
+
+ # Connect to database
+ conn = sqlite3.connect(db_tmp_file)
+ initialize_db(conn)
+
+ req_args = {'startIndex' : 0}
+
+ # The maximum range for time is 120 days
+ # Force a complete update if our range is longer
+ if (database_time != 0):
+ database_date = datetime.datetime.combine(datetime.date.fromtimestamp(database_time), datetime.time())
+ today_date = datetime.datetime.combine(datetime.date.today(), datetime.time())
+ delta = today_date - database_date
+ if delta.days < 120:
+ bb.debug(2, "CVE database: performing partial update")
+ req_args['lastModStartDate'] = database_date.isoformat()
+ req_args['lastModEndDate'] = today_date.isoformat()
+ else:
+ bb.note("CVE database: file too old, forcing a full update")
+
+ with bb.progress.ProgressHandler(d) as ph, open(os.path.join(d.getVar("TMPDIR"), 'cve_check'), 'a') as cve_f:
+
+ bb.debug(2, "Updating entries")
+ index = 0
+ url = d.getVar("NVDCVE_URL")
+ while True:
+ req_args['startIndex'] = index
+ raw_data = nvd_request_next(url, None, req_args)
+ if raw_data is None:
+ # We haven't managed to download data
+ return False
+
+ data = json.loads(raw_data)
+
+ index = data["startIndex"]
+ total = data["totalResults"]
+ per_page = data["resultsPerPage"]
+
+ for cve in data["vulnerabilities"]:
+ update_db(conn, cve)
+
+ index += per_page
+ ph.update((float(index) / (total+1)) * 100)
+ if index >= total:
+ break
+
+ # Recommended by NVD
+ time.sleep(6)
+
+ # Update success, set the date to cve_check file.
+ cve_f.write('CVE database update : %s\n\n' % datetime.date.today())
+
+ conn.commit()
+ conn.close()
+ return True
+
+def initialize_db(conn):
+ with conn:
+ c = conn.cursor()
+
+ c.execute("CREATE TABLE IF NOT EXISTS META (YEAR INTEGER UNIQUE, DATE TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS NVD (ID TEXT UNIQUE, SUMMARY TEXT, \
+ SCOREV2 TEXT, SCOREV3 TEXT, MODIFIED INTEGER, VECTOR TEXT)")
+
+ c.execute("CREATE TABLE IF NOT EXISTS PRODUCTS (ID TEXT, \
+ VENDOR TEXT, PRODUCT TEXT, VERSION_START TEXT, OPERATOR_START TEXT, \
+ VERSION_END TEXT, OPERATOR_END TEXT)")
+ c.execute("CREATE INDEX IF NOT EXISTS PRODUCT_ID_IDX on PRODUCTS(ID);")
+
+ c.close()
+
+def parse_node_and_insert(conn, node, cveId):
+
+ def cpe_generator():
+ for cpe in node.get('cpeMatch', ()):
+ if not cpe['vulnerable']:
+ return
+ cpe23 = cpe.get('criteria')
+ if not cpe23:
+ return
+ cpe23 = cpe23.split(':')
+ if len(cpe23) < 6:
+ return
+ vendor = cpe23[3]
+ product = cpe23[4]
+ version = cpe23[5]
+
+ if cpe23[6] == '*' or cpe23[6] == '-':
+ version_suffix = ""
+ else:
+ version_suffix = "_" + cpe23[6]
+
+ if version != '*' and version != '-':
+ # Version is defined, this is a '=' match
+ yield [cveId, vendor, product, version + version_suffix, '=', '', '']
+ elif version == '-':
+ # no version information is available
+ yield [cveId, vendor, product, version, '', '', '']
+ else:
+ # Parse start version, end version and operators
+ op_start = ''
+ op_end = ''
+ v_start = ''
+ v_end = ''
+
+ if 'versionStartIncluding' in cpe:
+ op_start = '>='
+ v_start = cpe['versionStartIncluding']
+
+ if 'versionStartExcluding' in cpe:
+ op_start = '>'
+ v_start = cpe['versionStartExcluding']
+
+ if 'versionEndIncluding' in cpe:
+ op_end = '<='
+ v_end = cpe['versionEndIncluding']
+
+ if 'versionEndExcluding' in cpe:
+ op_end = '<'
+ v_end = cpe['versionEndExcluding']
+
+ if op_start or op_end or v_start or v_end:
+ yield [cveId, vendor, product, v_start, op_start, v_end, op_end]
+ else:
+ # This is no version information, expressed differently.
+ # Save processing by representing as -.
+ yield [cveId, vendor, product, '-', '', '', '']
+
+ conn.executemany("insert into PRODUCTS values (?, ?, ?, ?, ?, ?, ?)", cpe_generator()).close()
+
+def update_db(conn, elt):
+ """
+ Update a single entry in the on-disk database
+ """
+
+ accessVector = None
+ cveId = elt['cve']['id']
+ if elt['cve']['vulnStatus'] == "Rejected":
+ return
+ cveDesc = ""
+ for desc in elt['cve']['descriptions']:
+ if desc['lang'] == 'en':
+ cveDesc = desc['value']
+ date = elt['cve']['lastModified']
+ try:
+ accessVector = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['accessVector']
+ cvssv2 = elt['cve']['metrics']['cvssMetricV2'][0]['cvssData']['baseScore']
+ except KeyError:
+ cvssv2 = 0.0
+ try:
+ accessVector = accessVector or elt['impact']['baseMetricV3']['cvssV3']['attackVector']
+ cvssv3 = elt['impact']['baseMetricV3']['cvssV3']['baseScore']
+ except KeyError:
+ accessVector = accessVector or "UNKNOWN"
+ cvssv3 = 0.0
+
+ conn.execute("insert or replace into NVD values (?, ?, ?, ?, ?, ?)",
+ [cveId, cveDesc, cvssv2, cvssv3, date, accessVector]).close()
+
+ try:
+ configurations = elt['cve']['configurations'][0]['nodes']
+ for config in configurations:
+ parse_node_and_insert(conn, config, cveId)
+ except KeyError:
+ bb.debug(2, "Entry without a configuration")
+
+do_fetch[nostamp] = "1"
+
+EXCLUDE_FROM_WORLD = "1"
--
2.35.1




Marta Rybczynska
 



On Wed, Apr 5, 2023 at 5:55 PM Steve Sakoman <steve@...> wrote:
Hi Marta,

Is this safe to backport to the stable branches, or should I let it
"age" in master for a while?


Hi Steve,
I vote to let it age for a little moment. In the meantime I'm trying to figure out when exactly NVD will turn off the old database.

Kind regards,
Marta


Steve Sakoman
 

On Wed, Apr 5, 2023 at 8:43 AM Marta Rybczynska <rybczynska@...> wrote:



On Wed, Apr 5, 2023 at 5:55 PM Steve Sakoman <steve@...> wrote:

Hi Marta,

Is this safe to backport to the stable branches, or should I let it
"age" in master for a while?
Hi Steve,
I vote to let it age for a little moment. In the meantime I'm trying to figure out when exactly NVD will turn off the old database.
OK, will do!

Steve


Marta Rybczynska
 



On Wed, Apr 5, 2023 at 8:44 PM Steve Sakoman <steve@...> wrote:
On Wed, Apr 5, 2023 at 8:43 AM Marta Rybczynska <rybczynska@...> wrote:
>
>
>
> On Wed, Apr 5, 2023 at 5:55 PM Steve Sakoman <steve@...> wrote:
>>
>> Hi Marta,
>>
>> Is this safe to backport to the stable branches, or should I let it
>> "age" in master for a while?
>>
>
> Hi Steve,
> I vote to let it age for a little moment. In the meantime I'm trying to figure out when exactly NVD will turn off the old database.

OK, will do!

I have received a confirmation from the NVD that they are currently planning to turn off teh old feed in *September 2023*.
We have time to stabilize then.

Kind regards,
Marta