feat: add pyperscan support

hyperscan will run simultaneously all version checkers on a file which
reduce processing time.

pyperscan package is used instead of the most well-known hyperscan
package as pyperscan allows to add a tag for each pattern. This feature
will allow to retrieve easily the checker associated to the matched
pattern.

On my local machine, running a scan on an embedded firmware takes 220
seconds with pyperscan instead of 326 seconds.

However, pyperscan is slower on a single file and unsupported on
Windows, so add a --pyperscan option (disabled by default)

Fix #2485

Signed-off-by: Fabrice Fontaine <[email protected]>

Author: ffontaine

.github/actions/spelling/allow.txt

@@ -303,6 +303,7 @@ https
 hunspell
 hur
 hwloc
+hyperscan
 i
 icecast
 icu
@@ -606,6 +607,7 @@ pybabel
 pycon
 pycqa
 pypa
+pyperscan
 pypi
 pytest
 pythex

README.md

@@ -541,6 +541,7 @@ Checkers:
   <a href="https://github.com/intel/cve-bin-tool/blob/main/doc/MANUAL.md#-s-skips---skips-skips">-s SKIPS, --skips SKIPS</a>
                         comma-separated list of checkers to disable
   <a href="https://github.com/intel/cve-bin-tool/blob/main/doc/MANUAL.md#-r-checkers---runs-checkers">-r RUNS, --runs RUNS</a>  comma-separated list of checkers to enable
+  <a href="https://github.com/intel/cve-bin-tool/blob/main/doc/MANUAL.md#--pyperscan">--pyperscan</a>  use pyperscan for binary checkers (unsupported on Windows)
 
 Database Management:
   --import-json IMPORT_JSON

cve_bin_tool/cli.py

@@ -488,6 +488,12 @@ def main(argv=None):
         help="comma-separated list of checkers to enable",
         default="",
     )
+    checker_group.add_argument(
+        "--pyperscan",
+        action="store_true",
+        help="Use pyperscan for binary checkers (unsupported on Windows)",
+        default=False,
+    )
 
     database_group = parser.add_argument_group("Database Management")
     database_group.add_argument(
@@ -1126,6 +1132,7 @@ def main(argv=None):
                 validate=not args["disable_validation_check"],
                 sources=enabled_sources,
                 no_scan=args["no_scan"],
+                pyperscan=args["pyperscan"],
             )
             version_scanner.remove_skiplist(skips)
             LOGGER.info(f"Number of checkers: {version_scanner.number_of_checkers()}")

cve_bin_tool/version_scanner.py

@@ -8,7 +8,7 @@
 from pathlib import Path
 from typing import Iterator
 
-from cve_bin_tool.checkers import BUILTIN_CHECKERS, Checker
+from cve_bin_tool.checkers import BUILTIN_CHECKERS, Checker, VersionMatchInfo
 from cve_bin_tool.cvedb import CVEDB
 from cve_bin_tool.egg_updater import IS_DEVELOP, update_egg
 from cve_bin_tool.error_handler import ErrorMode
@@ -19,6 +19,9 @@
 from cve_bin_tool.strings import parse_strings
 from cve_bin_tool.util import DirWalk, ProductInfo, ScanInfo, inpath
 
+if sys.platform != "win32":
+    from pyperscan import Pattern, Scan, StreamDatabase
+
 if sys.version_info >= (3, 10):
     from importlib import metadata as importlib_metadata
 else:
@@ -45,6 +48,7 @@ def __init__(
         validate: bool = True,
         sources=None,
         no_scan=False,
+        pyperscan=False,
     ):
         self.no_scan = no_scan
         self.logger = logger or LOGGER.getChild(self.__class__.__name__)
@@ -73,6 +77,12 @@ def __init__(
         self.logger.info("Checkers loaded: %s" % (", ".join(self.checkers.keys())))
         self.language_checkers = valid_files
         self.language_checkers_names = self.available_language_checkers()
+        if sys.platform == "win32" and pyperscan:
+            self.logger.error("pyperscan unsupported on Windows")
+            self.pyperscan = False
+        else:
+            self.pyperscan = pyperscan
+        self.pyperscan_db = None
 
     @classmethod
     def load_checkers(cls) -> dict[str, type[Checker]]:
@@ -276,36 +286,79 @@ def scan_file(self, filename: str) -> Iterator[ScanInfo]:
 
         yield from self.run_checkers(filename, lines)
 
+    def build_pyperscan_database(self, checkers: Checker) -> None:
+        # Database built only once to improve performance
+        if self.pyperscan_db is None:
+            patterns = []
+            for dummy_checker_name, checker in self.checkers.items():
+                checker = checker()
+                checker.dummy_checker_name = dummy_checker_name
+                for pattern in checker.VERSION_PATTERNS + checker.CONTAINS_PATTERNS:
+                    patterns.append(Pattern(pattern.pattern.encode(), tag=checker))
+            if patterns:
+                self.pyperscan_db = StreamDatabase(*patterns)
+
+    @staticmethod
+    def pyperscan_match(
+        pyperscan_matches: dict, checker: Checker, offset: int, end: int
+    ) -> Scan:
+        pyperscan_matches[checker.dummy_checker_name] = checker, offset, end
+        return Scan.Continue
+
     def run_checkers(self, filename: str, lines: str) -> Iterator[ScanInfo]:
         """process a Set of checker objects, run them on file lines,
         and yield information about detected products and versions.
         It uses logging to provide debug and error information along the way."""
         LOGGER.info(f"filename = {filename}")
-        # tko
-        for dummy_checker_name, checker in self.checkers.items():
-            checker = checker()
-            version_results = checker.get_versions(lines, filename)
-
-            if version_results.matched_filename or version_results.matched_contains:
-                for version in version_results.versions:
-                    if version == "UNKNOWN":
-                        file_path = "".join(self.file_stack)
-                        self.logger.debug(
-                            f"{dummy_checker_name} was detected with version UNKNOWN in file {file_path}"
-                        )
-                    else:
-                        file_path = "".join(self.file_stack)
-                        self.logger.debug(
-                            f"{file_path} matched {dummy_checker_name} {version} ({version_results.matched_filename=}, {version_results.matched_contains=})"
-                        )
-                        for vendor, product in checker.VENDOR_PRODUCT:
-                            yield ScanInfo(
-                                ProductInfo(vendor, product, version),
-                                file_path,
-                            )
+        if self.pyperscan:
+            self.build_pyperscan_database(self.checkers)
+
+            pyperscan_matches = dict()
+            scanner = self.pyperscan_db.build(pyperscan_matches, self.pyperscan_match)
+            scanner.scan(lines.encode())
+
+            for dummy_checker_name, (checker, offset, end) in pyperscan_matches.items():
+                # Confirm pyperscan match with get_versions as pyperscan doesn't support
+                # group capture. SOM_LEFTMOST is not enabled (offset is always 0)
+                version_results = checker.get_versions(lines[offset:end], filename)
+                yield from self.parse_version_match(
+                    dummy_checker_name, checker, version_results
+                )
+        else:
+            # tko
+            for dummy_checker_name, checker in self.checkers.items():
+                checker = checker()
+                version_results = checker.get_versions(lines, filename)
+                yield from self.parse_version_match(
+                    dummy_checker_name, checker, version_results
+                )
 
         self.logger.debug(f"Done scanning file: {filename}")
 
+    def parse_version_match(
+        self,
+        dummy_checker_name: str,
+        checker: Checker,
+        version_results: VersionMatchInfo,
+    ):
+        if version_results.matched_filename or version_results.matched_contains:
+            for version in version_results.versions:
+                if version == "UNKNOWN":
+                    file_path = "".join(self.file_stack)
+                    self.logger.debug(
+                        f"{dummy_checker_name} was detected with version UNKNOWN in file {file_path}"
+                    )
+                else:
+                    file_path = "".join(self.file_stack)
+                    self.logger.debug(
+                        f"{file_path} matched {dummy_checker_name} {version} ({version_results.matched_filename=}, {version_results.matched_contains=})"
+                    )
+                    for vendor, product in checker.VENDOR_PRODUCT:
+                        yield ScanInfo(
+                            ProductInfo(vendor, product, version),
+                            file_path,
+                        )
+
     @staticmethod
     def clean_file_path(filepath: str) -> str:
         """Returns a cleaner filepath by removing temp path from filepath"""

doc/MANUAL.md

@@ -37,6 +37,7 @@
   - [Checkers Arguments](#checkers-arguments)
     - [-s SKIPS, --skips SKIPS](#-s-skips---skips-skips)
     - [-r CHECKERS, --runs CHECKERS](#-r-checkers---runs-checkers)
+    - [--pyperscan](#--pyperscan)
   - [Input Arguments](#input-arguments)
     - [directory (positional argument)](#directory-positional-argument)
     - [-i INPUT\_FILE, --input-file INPUT\_FILE](#-i-input_file---input-file-input_file)
@@ -214,6 +215,7 @@ which is useful if you're trying the latest code from
       -s SKIPS, --skips SKIPS
                             comma-separated list of checkers to disable
       -r RUNS, --runs RUNS  comma-separated list of checkers to enable
+      --pyperscan           use pyperscan for binary checkers (unsupported on Windows)
 
     Database Management:
       --import-json IMPORT_JSON
@@ -887,6 +889,17 @@ This option allows one to skip (disable) a comma-separated list of checkers and
 
 This option allows one to enable a comma-separated list of checkers.
 
+### --pyperscan
+
+The pyperscan flag enables pyperscan support in the CVE Bin Tool. [pyperscan](https://github.com/vlaci/pyperscan) is an opinionated Python binding for [Hyperscan](https://www.hyperscan.io) focusing on easy of use and safety.
+
+When pyperscan flag is enabled, the tool leverages on Hyperscan High-performance regular expression matching library to runs simultaneously all binary version checkers on a file which significantly reduce processing time.
+
+pyperscan package is used instead of the most well-known hyperscan package as pyperscan allows to add a tag for each pattern. This feature allows to easily retrieve the binary checker associated to the matched pattern.
+
+> **Note**: pyperscan is unsupported on Windows.
+
+
 ## Input Arguments
 
 ### directory (positional argument)

requirements.csv

@@ -26,3 +26,4 @@ the_purl_authors_not_in_db,packageurl-python
 h2non,filetype
 python,setuptools
 jaraco,zipp
+vlaci_not_in_db,pyperscan

requirements.txt

@@ -14,6 +14,7 @@ lib4vex>=0.2.0
 packageurl-python
 packaging>=22.0
 plotly
+pyperscan; sys_platform != 'win32' # pyperscan unsupported on Windows
 python-gnupg
 pyyaml>=5.4
 requests>=2.32.2