feat(checker): add tesseract checker (#3570)

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

Author: ffontaine

cve_bin_tool/checkers/__init__.py

@@ -310,6 +310,7 @@
     "tcpdump",
     "tcpreplay",
     "terminology",
+    "tesseract",
     "thrift",
     "thttpd",
     "thunderbird",

cve_bin_tool/checkers/tesseract.py

@@ -0,0 +1,24 @@
+# Copyright (C) 2023 Orange
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+
+"""
+CVE checker for tesseract
+
+https://www.cvedetails.com/product/62309/Tesseract-Project-Tesseract.html?vendor_id=21055
+https://www.cvedetails.com/product/97901/Tesseract-Ocr-Project-Tesseract-Ocr.html?vendor_id=24900
+
+"""
+from __future__ import annotations
+
+from cve_bin_tool.checkers import Checker
+
+
+class TesseractChecker(Checker):
+    CONTAINS_PATTERNS: list[str] = []
+    FILENAME_PATTERNS: list[str] = []
+    VERSION_PATTERNS = [r"tesseract ([0-9]+\.[0-9]+\.[0-9]+)"]
+    VENDOR_PRODUCT = [
+        ("tesseract_project", "tesseract"),
+        ("tesseract_ocr_project", "tesseract_ocr"),
+    ]

test/condensed-downloads/libtesseract4_4.0.0-2_amd64.deb.tar.gz

@@ -0,0 +1,26 @@
+# Copyright (C) 2023 Orange
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+mapping_test_data = [
+    {"product": "tesseract", "version": "4.0.0", "version_strings": ["tesseract 4.0.0"]}
+]
+package_test_data = [
+    {
+        "url": "http://rpmfind.net/linux/fedora/linux/development/rawhide/Everything/aarch64/os/Packages/t/",
+        "package_name": "tesseract-5.3.3-1.fc40.aarch64.rpm",
+        "product": "tesseract",
+        "version": "5.3.3",
+    },
+    {
+        "url": "http://ftp.fr.debian.org/debian/pool/main/t/tesseract/",
+        "package_name": "libtesseract4_4.0.0-2_amd64.deb",
+        "product": "tesseract",
+        "version": "4.0.0",
+    },
+    {
+        "url": "https://downloads.openwrt.org/releases/21.02.0/packages/x86_64/packages/",
+        "package_name": "tesseract_4.0.0-2_x86_64.ipk",
+        "product": "tesseract",
+        "version": "4.0.0",
+    },
+]