provenant/parsers/

misc.rs

// SPDX-FileCopyrightText: nexB Inc. and others
// ScanCode is a trademark of nexB Inc.
// SPDX-FileCopyrightText: Provenant contributors
// SPDX-License-Identifier: Apache-2.0
// Derived from ScanCode Toolkit (Apache-2.0); modified. See NOTICE.

//! File type recognizers for various package archives and binary formats.
//!
//! This module contains simple file-type recognizers that identify packages by
//! their file extensions or path patterns. These recognizers do NOT parse file
//! contents - they only tag files with the appropriate package_type and datasource_id.
//!
//! # Implementation Notes
//!
//! - Most recognizers use the `file_recognizer!` macro to reduce boilerplate and
//!   return minimal PackageData with only package_type and datasource_id set
//! - These correspond to Python's misc.py NonAssemblableDatafileHandler classes
//! - The JVM-archive recognizers (`.jar`/`.war`/`.aar`) are an exception: they
//!   perform bounded ZIP introspection of `META-INF/MANIFEST.MF` and
//!   `META-INF/maven/**/pom.properties` to recover real package metadata
//! - Some recognizers use magic byte detection for disambiguation (Squashfs, NSIS, InstallShield)

use std::path::Path;

use super::PackageParser;
use super::maven::{JvmArchiveKind, extract_jvm_archive};
use crate::models::{DatasourceId, PackageData, PackageType};
use crate::utils::magic;

/// Helper macro to define file-type recognizers with minimal boilerplate.
///
/// Each recognizer matches specific file patterns and returns a minimal
/// PackageData structure with only package_type and datasource_id populated.
///
/// # Arguments
///
/// * `$name` - Struct name for the recognizer
/// * `$pkg_type` - Package type string (e.g., "jar", "war", "meteor")
/// * `$datasource` - Datasource ID string (e.g., "java_jar", "meteor_package")
/// * `$match_fn` - Closure that takes a &Path and returns bool for matching
macro_rules! file_recognizer {
    ($name:ident, $pkg_type:expr, $datasource:expr, $match_fn:expr) => {
        pub struct $name;

        impl PackageParser for $name {
            const PACKAGE_TYPE: PackageType = $pkg_type;

            fn is_match(path: &Path) -> bool {
                ($match_fn)(path)
            }

            fn extract_packages(path: &Path) -> Vec<PackageData> {
                let _ = path;
                vec![PackageData {
                    package_type: Some($pkg_type),
                    datasource_id: Some($datasource),
                    ..Default::default()
                }]
            }
        }
    };
}

// Java Archives

/// `.jar` archive introspector.
///
/// Reads `META-INF/MANIFEST.MF` and `META-INF/maven/**/pom.properties` from the
/// archive via the shared bounded-ZIP reader and falls back to the bare
/// recognizer row when no metadata can be recovered.
pub struct JavaJarRecognizer;

impl PackageParser for JavaJarRecognizer {
    const PACKAGE_TYPE: PackageType = PackageType::Jar;

    fn is_match(path: &Path) -> bool {
        path.extension().and_then(|e| e.to_str()) == Some("jar")
    }

    fn extract_packages(path: &Path) -> Vec<PackageData> {
        extract_jvm_archive(path, JvmArchiveKind::Jar)
    }
}

/// `.war` web-application archive introspector. See [`JavaJarRecognizer`].
pub struct JavaWarRecognizer;

impl PackageParser for JavaWarRecognizer {
    const PACKAGE_TYPE: PackageType = PackageType::War;

    fn is_match(path: &Path) -> bool {
        path.extension().and_then(|e| e.to_str()) == Some("war")
    }

    fn extract_packages(path: &Path) -> Vec<PackageData> {
        extract_jvm_archive(path, JvmArchiveKind::War)
    }
}

file_recognizer!(
    JavaWarWebXmlRecognizer,
    PackageType::War,
    DatasourceId::JavaWarWebXml,
    |path: &Path| path
        .to_str()
        .is_some_and(|p| p.ends_with("/WEB-INF/web.xml") || p.ends_with("WEB-INF/web.xml"))
);

file_recognizer!(
    JavaEarRecognizer,
    PackageType::Ear,
    DatasourceId::JavaEarArchive,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("ear")
);

file_recognizer!(
    JavaEarAppXmlRecognizer,
    PackageType::Ear,
    DatasourceId::JavaEarApplicationXml,
    |path: &Path| path.to_str().is_some_and(
        |p| p.ends_with("/META-INF/application.xml") || p.ends_with("META-INF/application.xml")
    )
);

// Apache Axis2

file_recognizer!(
    Axis2ModuleXmlRecognizer,
    PackageType::Axis2,
    DatasourceId::Axis2ModuleXml,
    |path: &Path| {
        path.to_str().is_some_and(|p| {
            let lower = p.to_lowercase();
            lower.ends_with("/meta-inf/module.xml") || lower.ends_with("meta-inf/module.xml")
        })
    }
);

file_recognizer!(
    Axis2MarRecognizer,
    PackageType::Axis2,
    DatasourceId::Axis2Mar,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("mar")
);

// JBoss

file_recognizer!(
    JBossSarRecognizer,
    PackageType::JbossService,
    DatasourceId::JbossSar,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("sar")
);

file_recognizer!(
    JBossServiceXmlRecognizer,
    PackageType::JbossService,
    DatasourceId::JbossServiceXml,
    |path: &Path| {
        path.to_str().is_some_and(|p| {
            let lower = p.to_lowercase();
            lower.ends_with("/meta-inf/jboss-service.xml")
                || lower.ends_with("meta-inf/jboss-service.xml")
        })
    }
);

// Meteor

file_recognizer!(
    MeteorPackageRecognizer,
    PackageType::Meteor,
    DatasourceId::MeteorPackage,
    |path: &Path| path.to_str().is_some_and(|p| p.ends_with("/package.js"))
);

// Mobile Apps

/// `.aar` Android library archive introspector. See [`JavaJarRecognizer`].
pub struct AndroidLibraryRecognizer;

impl PackageParser for AndroidLibraryRecognizer {
    const PACKAGE_TYPE: PackageType = PackageType::AndroidLib;

    fn is_match(path: &Path) -> bool {
        path.extension().and_then(|e| e.to_str()) == Some("aar")
    }

    fn extract_packages(path: &Path) -> Vec<PackageData> {
        extract_jvm_archive(path, JvmArchiveKind::Aar)
    }
}

file_recognizer!(
    MozillaXpiRecognizer,
    PackageType::Mozilla,
    DatasourceId::MozillaXpi,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("xpi")
);

file_recognizer!(
    ChromeCrxRecognizer,
    PackageType::Chrome,
    DatasourceId::ChromeCrx,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("crx")
);

file_recognizer!(
    IosIpaRecognizer,
    PackageType::Ios,
    DatasourceId::IosIpa,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("ipa")
);

// Archives

file_recognizer!(
    CabArchiveRecognizer,
    PackageType::Cab,
    DatasourceId::MicrosoftCabinet,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("cab")
);

file_recognizer!(
    SharArchiveRecognizer,
    PackageType::Shar,
    DatasourceId::SharShellArchive,
    |path: &Path| path.extension().and_then(|e| e.to_str()) == Some("shar")
);

// Disk Images

file_recognizer!(
    AppleDmgRecognizer,
    PackageType::Dmg,
    DatasourceId::AppleDmg,
    |path: &Path| {
        path.extension()
            .and_then(|e| e.to_str())
            .is_some_and(|ext| ext == "dmg" || ext == "sparseimage")
    }
);

file_recognizer!(
    IsoImageRecognizer,
    PackageType::Iso,
    DatasourceId::IsoDiskImage,
    |path: &Path| {
        path.extension()
            .and_then(|e| e.to_str())
            .is_some_and(|ext| ext == "iso" || ext == "udf" || ext == "img")
    }
);

// Installers and Binary Formats (require magic byte detection)

file_recognizer!(
    SquashfsRecognizer,
    PackageType::Squashfs,
    DatasourceId::SquashfsDiskImage,
    |path: &Path| magic::is_squashfs(path)
);

file_recognizer!(
    NsisRecognizer,
    PackageType::Nsis,
    DatasourceId::NsisInstaller,
    |path: &Path| {
        path.extension()
            .and_then(|e| e.to_str())
            .is_some_and(|ext| ext == "exe")
            && magic::is_nsis_installer(path)
    }
);

file_recognizer!(
    InstallShieldRecognizer,
    PackageType::Installshield,
    DatasourceId::InstallshieldInstaller,
    |path: &Path| {
        path.extension()
            .and_then(|e| e.to_str())
            .is_some_and(|ext| ext == "exe")
            && magic::is_zip(path)
    }
);

pub(crate) static RECOGNIZER_METADATA: &[super::metadata::ParserMetadata] = &[
    super::metadata::ParserMetadata {
        description: "Misc file type recognizers (JAR, WAR, EAR, AAR, iOS, Chrome, Mozilla, installers, disk images, etc.)",
        file_patterns: &[
            "**/*.jar",
            "**/*.war",
            "**/WEB-INF/web.xml",
            "**/*.ear",
            "**/META-INF/application.xml",
            "**/meta-inf/module.xml",
            "**/*.mar",
            "**/*.sar",
            "**/meta-inf/jboss-service.xml",
            "**/package.js",
            "**/*.aar",
            "**/*.xpi",
            "**/*.crx",
            "**/*.ipa",
            "**/*.cab",
            "**/*.shar",
            "**/*.dmg",
            "**/*.sparseimage",
            "**/*.iso",
            "**/*.udf",
            "**/*.img",
            "**/*.exe",
        ],
        package_type: "",
        primary_language: "",
        documentation_url: None,
    },
];