@trustify-da/trustify-da-javascript-client 0.3.0-ea.e12bc82 → 0.3.0-ea.e645720

package/dist/src/providers/python_pip.js

@@ -1,16 +1,18 @@
 import fs from 'node:fs';
-import { EOL } from 'os';
 import { PackageURL } from 'packageurl-js';
+import { readLicenseFile } from '../license/license_utils.js';
 import Sbom from '../sbom.js';
 import { environmentVariableIsPopulated, getCustom, getCustomPath, invokeCommand } from "../tools.js";
 import Python_controller from './python_controller.js';
-export default { isSupported, validateLockFile, provideComponent, provideStack };
-/** @typedef {{name: string, version: string, dependencies: DependencyEntry[]}} DependencyEntry */
+import { getParser, getIgnoreQuery, getPinnedVersionQuery } from './requirements_parser.js';
+export default { isSupported, validateLockFile, provideComponent, provideStack, readLicenseFromManifest };
+/** @typedef {{name: string, version: string, dependencies: DependencyEntry[], hashes?: Array<{alg: string, content: string}>}} DependencyEntry */
 /**
  * @type {string} ecosystem for python-pip is 'pip'
  * @private
  */
 const ecosystem = 'pip';
+const NO_SCOPE = undefined;
 /**
  * @param {string} manifestName - the subject manifest name-type
  * @returns {boolean} - return true if `requirements.txt` is the manifest name-type
@@ -18,6 +20,13 @@ const ecosystem = 'pip';
 function isSupported(manifestName) {
     return 'requirements.txt' === manifestName;
 }
+/**
+ * Python requirements.txt has no standard license field
+ * @param {string} manifestPath - path to requirements.txt
+ * @returns {string|null}
+ */
+// eslint-disable-next-line no-unused-vars
+function readLicenseFromManifest(manifestPath) { return readLicenseFile(manifestPath); }
 /**
  * @param {string} manifestDir - the directory where the manifest lies
  */
@@ -26,12 +35,12 @@ function validateLockFile() { return true; }
  * Provide content and content type for python-pip stack analysis.
  * @param {string} manifest - the manifest path or name
  * @param {{}} [opts={}] - optional various options to pass along the application
- * @returns {Provided}
+ * @returns {Promise<Provided>}
  */
-function provideStack(manifest, opts = {}) {
+async function provideStack(manifest, opts = {}) {
     return {
         ecosystem,
-        content: createSbomStackAnalysis(manifest, opts),
+        content: await createSbomStackAnalysis(manifest, opts),
         contentType: 'application/vnd.cyclonedx+json'
     };
 }
@@ -39,16 +48,15 @@ function provideStack(manifest, opts = {}) {
  * Provide content and content type for python-pip component analysis.
  * @param {string} manifest - path to requirements.txt for component report
  * @param {{}} [opts={}] - optional various options to pass along the application
- * @returns {Provided}
+ * @returns {Promise<Provided>}
  */
-function provideComponent(manifest, opts = {}) {
+async function provideComponent(manifest, opts = {}) {
     return {
         ecosystem,
-        content: getSbomForComponentAnalysis(manifest, opts),
+        content: await getSbomForComponentAnalysis(manifest, opts),
         contentType: 'application/vnd.cyclonedx+json'
     };
 }
-/** @typedef {{name: string, , version: string, dependencies: DependencyEntry[]}} DependencyEntry */
 /**
  *
  * @param {PackageURL}source
@@ -58,7 +66,7 @@ function provideComponent(manifest, opts = {}) {
  */
 function addAllDependencies(source, dep, sbom) {
     let targetPurl = toPurl(dep["name"], dep["version"]);
-    sbom.addDependency(source, targetPurl);
+    sbom.addDependency(source, targetPurl, NO_SCOPE, dep["hashes"]);
     let directDeps = dep["dependencies"];
     if (directDeps !== undefined && directDeps.length > 0) {
         directDeps.forEach((dependency) => { addAllDependencies(toPurl(dep["name"], dep["version"]), dependency, sbom); });
@@ -66,49 +74,34 @@ function addAllDependencies(source, dep, sbom) {
 }
 /**
  *
- * @param nameVersion
- * @return {string}
- */
-function splitToNameVersion(nameVersion) {
-    let result = [];
-    if (nameVersion.includes("==")) {
-        return nameVersion.split("==");
-    }
-    const regex = /[^\w\s-_]/g;
-    let endIndex = nameVersion.search(regex);
-    if (endIndex === -1) {
-        return [nameVersion.trim()];
-    }
-    result.push(nameVersion.substring(0, endIndex).trim());
-    return result;
-}
-/**
- *
- * @param {string} requirementTxtContent
- * @return {PackageURL []}
+ * @param {string} manifest - path to requirements.txt
+ * @return {Promise<PackageURL[]>}
  */
-function getIgnoredDependencies(requirementTxtContent) {
-    let requirementsLines = requirementTxtContent.split(EOL);
-    return requirementsLines
-        .filter(line => line.includes("#exhortignore") || line.includes("# exhortignore"))
-        .map((line) => line.substring(0, line.indexOf("#")).trim())
-        .map((name) => {
-        let nameVersion = splitToNameVersion(name);
-        if (nameVersion.length === 2) {
-            return toPurl(nameVersion[0], nameVersion[1]);
-        }
-        return toPurl(nameVersion[0], undefined);
+async function getIgnoredDependencies(manifest) {
+    const [parser, ignoreQuery, pinnedVersionQuery] = await Promise.all([
+        getParser(), getIgnoreQuery(), getPinnedVersionQuery()
+    ]);
+    const content = fs.readFileSync(manifest).toString();
+    const tree = parser.parse(content);
+    return ignoreQuery.matches(tree.rootNode).map(match => {
+        const reqNode = match.captures.find(c => c.name === 'req').node;
+        const name = match.captures.find(c => c.name === 'name').node.text;
+        const versionMatches = pinnedVersionQuery.matches(reqNode);
+        const version = versionMatches.length > 0
+            ? versionMatches[0].captures.find(c => c.name === 'version').node.text
+            : undefined;
+        return toPurl(name, version);
     });
 }
 /**
  *
- * @param {string} requirementTxtContent content of requirments.txt in string
+ * @param {string} manifest - path to requirements.txt
  * @param {Sbom} sbom object to filter out from it exhortignore dependencies.
  * @param {{Object}} opts - various options and settings for the application
  * @private
  */
-function handleIgnoredDependencies(requirementTxtContent, sbom, opts = {}) {
-    let ignoredDeps = getIgnoredDependencies(requirementTxtContent);
+async function handleIgnoredDependencies(manifest, sbom, opts = {}) {
+    let ignoredDeps = await getIgnoredDependencies(manifest);
     let matchManifestVersions = getCustom("MATCH_MANIFEST_VERSIONS", "true", opts);
     if (matchManifestVersions === "true") {
         const ignoredDepsVersion = ignoredDeps.filter(dep => dep.version !== undefined);
@@ -172,22 +165,22 @@ const DEFAULT_PIP_ROOT_COMPONENT_VERSION = "0.0.0";
  * Create sbom json string out of a manifest path for stack analysis.
  * @param {string} manifest - path for requirements.txt
  * @param {{}} [opts={}] - optional various options to pass along the application
- * @returns {string} the sbom json string content
+ * @returns {Promise<string>} the sbom json string content
  * @private
  */
-function createSbomStackAnalysis(manifest, opts = {}) {
+async function createSbomStackAnalysis(manifest, opts = {}) {
     let binaries = {};
     let createVirtualPythonEnv = handlePythonEnvironment(binaries, opts);
     let pythonController = new Python_controller(createVirtualPythonEnv === "false", binaries.pip, binaries.python, manifest, opts);
-    let dependencies = pythonController.getDependencies(true);
+    let dependencies = await pythonController.getDependencies(true);
     let sbom = new Sbom();
     const rootPurl = toPurl(DEFAULT_PIP_ROOT_COMPONENT_NAME, DEFAULT_PIP_ROOT_COMPONENT_VERSION);
-    sbom.addRoot(rootPurl);
+    const license = readLicenseFromManifest(manifest);
+    sbom.addRoot(rootPurl, license);
     dependencies.forEach(dep => {
         addAllDependencies(rootPurl, dep, sbom);
     });
-    let requirementTxtContent = fs.readFileSync(manifest).toString();
-    handleIgnoredDependencies(requirementTxtContent, sbom, opts);
+    await handleIgnoredDependencies(manifest, sbom, opts);
     // In python there is no root component, then we must remove the dummy root we added, so the sbom json will be accepted by the DA backend
     // sbom.removeRootComponent()
     return sbom.getAsJsonString(opts);
@@ -196,22 +189,22 @@ function createSbomStackAnalysis(manifest, opts = {}) {
  * Create a sbom json string out of a manifest content for component analysis
  * @param {string} manifest - path to requirements.txt
  * @param {{}} [opts={}] - optional various options to pass along the application
- * @returns {string} the sbom json string content
+ * @returns {Promise<string>} the sbom json string content
  * @private
  */
-function getSbomForComponentAnalysis(manifest, opts = {}) {
+async function getSbomForComponentAnalysis(manifest, opts = {}) {
     let binaries = {};
     let createVirtualPythonEnv = handlePythonEnvironment(binaries, opts);
     let pythonController = new Python_controller(createVirtualPythonEnv === "false", binaries.pip, binaries.python, manifest, opts);
-    let dependencies = pythonController.getDependencies(false);
+    let dependencies = await pythonController.getDependencies(false);
     let sbom = new Sbom();
     const rootPurl = toPurl(DEFAULT_PIP_ROOT_COMPONENT_NAME, DEFAULT_PIP_ROOT_COMPONENT_VERSION);
-    sbom.addRoot(rootPurl);
+    const license = readLicenseFromManifest(manifest);
+    sbom.addRoot(rootPurl, license);
     dependencies.forEach(dep => {
-        sbom.addDependency(rootPurl, toPurl(dep.name, dep.version));
+        sbom.addDependency(rootPurl, toPurl(dep.name, dep.version), NO_SCOPE, dep.hashes);
     });
-    let requirementTxtContent = fs.readFileSync(manifest).toString();
-    handleIgnoredDependencies(requirementTxtContent, sbom, opts);
+    await handleIgnoredDependencies(manifest, sbom, opts);
     // In python there is no root component, then we must remove the dummy root we added, so the sbom json will be accepted by the DA backend
     // sbom.removeRootComponent()
     return sbom.getAsJsonString(opts);

package/dist/src/providers/python_pip_pyproject.d.ts

@@ -0,0 +1,61 @@
+/**
+ * Python provider for pyproject.toml files using PEP 621 format without a lock file.
+ * Uses `pip install --dry-run --ignore-installed --report` to resolve the full dependency tree.
+ * Acts as the fallback provider when no lock file (uv.lock/poetry.lock) is found.
+ */
+export default class Python_pip_pyproject extends Base_pyproject {
+    /**
+     * Always returns true — pip provider is the fallback when no lock file is found.
+     * @param {string} manifestDir
+     * @param {{}} [opts={}]
+     * @returns {boolean}
+     */
+    validateLockFile(manifestDir: string, opts?: {}): boolean;
+    /**
+     * Get pip report output from env var override or by running pip.
+     * @param {string} manifestDir - directory containing pyproject.toml
+     * @param {{}} [opts={}]
+     * @returns {string} pip report JSON string
+     */
+    _getPipReportOutput(manifestDir: string, opts?: {}): string;
+    /**
+     * Parse pip report JSON and build dependency graph.
+     * @param {string} reportJson - pip report JSON string
+     * @returns {{directDeps: string[], graph: Map<string, {name: string, version: string, children: string[]}>}}
+     */
+    _parsePipReport(reportJson: string): {
+        directDeps: string[];
+        graph: Map<string, {
+            name: string;
+            version: string;
+            children: string[];
+        }>;
+    };
+    /**
+     * Check if a requires_dist entry is an extras-only dependency.
+     * @param {string} req - e.g. "PySocks!=1.5.7,>=1.5.6; extra == \"socks\""
+     * @returns {boolean}
+     */
+    _hasExtraMarker(req: string): boolean;
+    /**
+     * Extract package name from a requires_dist entry.
+     * @param {string} req - e.g. "charset_normalizer<4,>=2"
+     * @returns {string|null}
+     */
+    _extractDepName(req: string): string | null;
+    /**
+     * Resolve dependencies using pip install --dry-run --report.
+     * @param {string} manifestDir
+     * @param {string} _workspaceDir - unused (pip resolves from manifest directory)
+     * @param {object} parsed - parsed pyproject.toml
+     * @param {{}} [opts={}]
+     * @returns {Promise<{directDeps: string[], graph: Map}>}
+     */
+    _getDependencyData(manifestDir: string, _workspaceDir: string, parsed: object, opts?: {}): Promise<{
+        directDeps: string[];
+        graph: Map<any, any>;
+    }>;
+    _findEggInfoDirs(dir: any): string[];
+    _cleanupEggInfo(dir: any, existing: any): void;
+}
+import Base_pyproject from './base_pyproject.js';

package/dist/src/providers/python_pip_pyproject.js

@@ -0,0 +1,144 @@
+import fs from 'node:fs';
+import path from 'node:path';
+import { environmentVariableIsPopulated, getCustomPath, invokeCommand } from '../tools.js';
+import Base_pyproject from './base_pyproject.js';
+/**
+ * Python provider for pyproject.toml files using PEP 621 format without a lock file.
+ * Uses `pip install --dry-run --ignore-installed --report` to resolve the full dependency tree.
+ * Acts as the fallback provider when no lock file (uv.lock/poetry.lock) is found.
+ */
+export default class Python_pip_pyproject extends Base_pyproject {
+    /** @returns {string} */
+    _lockFileName() {
+        return '.pip-lock-nonexistent';
+    }
+    /** @returns {string} */
+    _cmdName() {
+        return 'pip';
+    }
+    /**
+     * Always returns true — pip provider is the fallback when no lock file is found.
+     * @param {string} manifestDir
+     * @param {{}} [opts={}]
+     * @returns {boolean}
+     */
+    // eslint-disable-next-line no-unused-vars
+    validateLockFile(manifestDir, opts = {}) {
+        return true;
+    }
+    /**
+     * Get pip report output from env var override or by running pip.
+     * @param {string} manifestDir - directory containing pyproject.toml
+     * @param {{}} [opts={}]
+     * @returns {string} pip report JSON string
+     */
+    _getPipReportOutput(manifestDir, opts) {
+        if (environmentVariableIsPopulated('TRUSTIFY_DA_PIP_REPORT')) {
+            return Buffer.from(process.env['TRUSTIFY_DA_PIP_REPORT'], 'base64').toString('ascii');
+        }
+        let pipBin = getCustomPath('pip3', opts);
+        try {
+            invokeCommand(pipBin, ['--version']);
+        }
+        catch {
+            pipBin = getCustomPath('pip', opts);
+        }
+        let eggInfoDirs = this._findEggInfoDirs(manifestDir);
+        let result = invokeCommand(pipBin, [
+            'install', '--dry-run', '--ignore-installed', '--quiet', '--report', '-', '.'
+        ], { cwd: manifestDir }).toString();
+        this._cleanupEggInfo(manifestDir, eggInfoDirs);
+        return result;
+    }
+    /**
+     * Parse pip report JSON and build dependency graph.
+     * @param {string} reportJson - pip report JSON string
+     * @returns {{directDeps: string[], graph: Map<string, {name: string, version: string, children: string[]}>}}
+     */
+    _parsePipReport(reportJson) {
+        let report = JSON.parse(reportJson);
+        let packages = report.install || [];
+        let rootEntry = packages.find(p => p.download_info?.dir_info !== undefined);
+        let rootRequires = rootEntry?.metadata?.requires_dist || [];
+        let directDepNames = new Set();
+        for (let req of rootRequires) {
+            if (this._hasExtraMarker(req)) {
+                continue;
+            }
+            let name = this._extractDepName(req);
+            if (name) {
+                directDepNames.add(this._canonicalize(name));
+            }
+        }
+        let graph = new Map();
+        let nonRootPackages = packages.filter(p => p !== rootEntry);
+        for (let pkg of nonRootPackages) {
+            let name = pkg.metadata.name;
+            let version = pkg.metadata.version;
+            let key = this._canonicalize(name);
+            graph.set(key, { name, version, children: [] });
+        }
+        for (let pkg of nonRootPackages) {
+            let key = this._canonicalize(pkg.metadata.name);
+            let entry = graph.get(key);
+            let requires = pkg.metadata.requires_dist || [];
+            for (let req of requires) {
+                let depName = this._extractDepName(req);
+                if (!depName) {
+                    continue;
+                }
+                let depKey = this._canonicalize(depName);
+                if (graph.has(depKey)) {
+                    entry.children.push(depKey);
+                }
+            }
+        }
+        let directDeps = [...directDepNames].filter(key => graph.has(key));
+        return { directDeps, graph };
+    }
+    /**
+     * Check if a requires_dist entry is an extras-only dependency.
+     * @param {string} req - e.g. "PySocks!=1.5.7,>=1.5.6; extra == \"socks\""
+     * @returns {boolean}
+     */
+    _hasExtraMarker(req) {
+        return /;\s*.*extra\s*==/.test(req);
+    }
+    /**
+     * Extract package name from a requires_dist entry.
+     * @param {string} req - e.g. "charset_normalizer<4,>=2"
+     * @returns {string|null}
+     */
+    _extractDepName(req) {
+        let match = req.match(/^([A-Za-z0-9]([A-Za-z0-9._-]*[A-Za-z0-9])?)/);
+        return match ? match[1] : null;
+    }
+    /**
+     * Resolve dependencies using pip install --dry-run --report.
+     * @param {string} manifestDir
+     * @param {string} _workspaceDir - unused (pip resolves from manifest directory)
+     * @param {object} parsed - parsed pyproject.toml
+     * @param {{}} [opts={}]
+     * @returns {Promise<{directDeps: string[], graph: Map}>}
+     */
+    // eslint-disable-next-line no-unused-vars
+    async _getDependencyData(manifestDir, _workspaceDir, parsed, opts) {
+        let reportOutput = this._getPipReportOutput(manifestDir, opts);
+        return this._parsePipReport(reportOutput);
+    }
+    _findEggInfoDirs(dir) {
+        try {
+            return fs.readdirSync(dir).filter(f => f.endsWith('.egg-info'));
+        }
+        catch {
+            return [];
+        }
+    }
+    _cleanupEggInfo(dir, existing) {
+        for (let entry of this._findEggInfoDirs(dir)) {
+            if (!existing.includes(entry)) {
+                fs.rmSync(path.join(dir, entry), { recursive: true, force: true });
+            }
+        }
+    }
+}

package/dist/src/providers/python_poetry.d.ts

@@ -0,0 +1,75 @@
+export default class Python_poetry extends Base_pyproject {
+    /**
+     * @param {string} manifestDir
+     * @param {string} _workspaceDir - unused (poetry has no workspace support)
+     * @param {object} parsed - parsed pyproject.toml
+     * @param {Object} opts
+     * @returns {Promise<{directDeps: string[], graph: Map<string, {name: string, version: string, children: string[]}>}>}
+     */
+    _getDependencyData(manifestDir: string, _workspaceDir: string, parsed: object, opts: any): Promise<{
+        directDeps: string[];
+        graph: Map<string, {
+            name: string;
+            version: string;
+            children: string[];
+        }>;
+    }>;
+    /**
+     * Get poetry show --tree output.
+     * @param {string} manifestDir
+     * @param {boolean} hasDevGroup
+     * @param {Object} opts
+     * @returns {string}
+     */
+    _getPoetryShowTreeOutput(manifestDir: string, hasDevGroup: boolean, opts: any): string;
+    /**
+     * Get poetry show --all output (flat list with resolved versions).
+     * @param {string} manifestDir
+     * @param {Object} opts
+     * @returns {string}
+     */
+    _getPoetryShowAllOutput(manifestDir: string, opts: any): string;
+    /**
+     * Parse poetry show --all output into a version map.
+     * Lines look like: "name         (!) 1.2.3  Description text..."
+     * or:              "name             1.2.3  Description text..."
+     * @param {string} output
+     * @returns {Map<string, string>} canonical name -> version
+     */
+    _parsePoetryShowAll(output: string): Map<string, string>;
+    /**
+     * Collects PEP 508 marker expressions for direct and transitive deps.
+     * Direct markers come from pyproject.toml dependency strings, e.g.:
+     *   "pywin32>=311 ; sys_platform == 'win32'"  → directMarkers['pywin32'] = "sys_platform == 'win32'"
+     * Transitive markers come from poetry.lock [package.dependencies] entries, e.g.:
+     *   colorama = {version = "*", markers = "sys_platform == 'win32'"}
+     *   → transitiveMarkers['click']['colorama'] = "sys_platform == 'win32'"
+     * @param {string|null} lockDir
+     * @param {object} parsed - parsed pyproject.toml
+     * @returns {{directMarkers: Map<string, string>, transitiveMarkers: Map<string, Map<string, string>>}}
+     */
+    _extractMarkerData(lockDir: string | null, parsed: object): {
+        directMarkers: Map<string, string>;
+        transitiveMarkers: Map<string, Map<string, string>>;
+    };
+    /**
+     * Parse poetry show --tree output into a dependency graph structure.
+     *
+     * @param {string} treeOutput
+     * @param {Map<string, string>} versionMap - canonical name -> resolved version
+     * @param {{directMarkers: Map<string, string>, transitiveMarkers: Map<string, Map<string, string>>}} markerData
+     * @returns {{directDeps: string[], graph: Map<string, {name: string, version: string, children: string[]}>}}
+     */
+    _parsePoetryTree(treeOutput: string, versionMap: Map<string, string>, markerData: {
+        directMarkers: Map<string, string>;
+        transitiveMarkers: Map<string, Map<string, string>>;
+    }): {
+        directDeps: string[];
+        graph: Map<string, {
+            name: string;
+            version: string;
+            children: string[];
+        }>;
+    };
+}
+import Base_pyproject from './base_pyproject.js';