Replace core SBOM-creation API with builder pattern #1383

wagoodman · 2022-12-02T17:30:18Z

Adds a top-level replacement for the syft API. The idea is to allow for encapsulation of more kinds of cataloging without the need to share a data interface. This allows for file-base cataloging and package-based cataloging to share the same approach to selection and configuration.

The existing cataloging functions have been removed, which is why this is a breaking change. I initially attempted to keep both schemes in place, however, the configuration management became ultimately confusing.

This PR adds high-level configuration:

syft/cataloging/*.go: cross-cutting configuration that could affect all catalogers, the artifacts they produce, or add downstream artifacts based on these descriptions. This is a set of configurations NOT capabilities (e.g. behavior, such as catalogers themselves).
syft/cataloging/pkgcataloging/*.go: wires up configurations for all package catalogers
syft/cataloging/filecataloging/*.go: wires up all configurations for file catalogers

This PR removes the existing configurations:

syft/pkg/cataloger/config.go

From a high-level, the CreateSBOMConfig is the entrypoint to all cataloging. The configuration itself describes what should be done. Ultimately all capabilities (file cataloging, pkg cataloging, linux distro identification, and cross-cutting relationship additions) are expressed as "tasks". Tasks act like a facade, similar to the command pattern, and encapsulates pre-configured behavior that ultimately writes to an SBOM. The notion of "tasks" has not been exported to the public API.

Secondarily, this PR makes the following adjustments:

migrates all cross-cutting relationship functions to an internal relationship package
leverages the recent UI enhancements to show catalogers in a tree on the CLI
migrates some application configurations in a breaking fashion:
- exclude-binary-overlap-by-ownership has been moved to package.exclude-binary-overlap-by-ownership
- default-image-pull-source has been moved to source.image.default-pull-source

Minimal example of using the new API:

src := source.New(...)
sbom, err := syft.CreateSBOM(ctx, src, nil)

or

src := source.New(...)
cfg := syft.DefaultCreateSBOMConfig()
sbom, err := syft.CreateSBOM(ctx, src, cfg)

Leveraging a little more of the API:

cfg := syft.DefaultCreateSBOMConfig().
	WithTool("my-tool", "v1.0.2").
	WithParallelism(5).
	WithCatalogerSelection("+sbom-cataloger", "-rpm").
	WithRelationshipsConfig(
		cataloging.RelationshipsConfig{
			FileOwnership:        false,
			FileOwnershipOverlap: false,
			ExcludeBinaryPackagesWithFileOwnershipOverlap: false,
		},
	).
	WithSearchConfig(
		cataloging.SearchConfig{
			Scope: source.SquashedScope,
		},
	).
	WithDataGenerationConfig(
		cataloging.DataGenerationConfig{
			GenerateCPEs:          false,
		},
	).
	WithFilesConfig(
		filecataloging.Config{
			Selection: file.OwnedFilesSelection,
			Hashers: []crypto.Hash{
				crypto.SHA256,
				crypto.SHA1,
			},
		},
	).
	WithPackagesConfig(
		pkgcataloging.Config{
			Golang: golang.CatalogerConfig{
				SearchLocalModCacheLicenses: true,
				SearchRemoteLicenses:        true,
			},
			LinuxKernel: kernel.LinuxKernelCatalogerConfig{
				CatalogModules: true,
			},
			Python: python.CatalogerConfig{
				GuessUnpinnedRequirements: true,
			},
			Java: java.CatalogerConfig{
				ArchiveSearchConfig: cataloging.ArchiveSearchConfig{
					IncludeIndexedArchives:   true,
					IncludeUnindexedArchives: false,
				},
				UseNetwork: true,
			},
		},
	)

sbom, err := syft.CreateSBOM(ctx, src, cfg)

Today when the cataloging process is run, the application configuration is captured to show the exact input. I've changed this some to instead capture an API-level construct instead of a construct that is in the cmd package. Here is an example of the syft-json descriptor section:

{
  "name": "syft",
  "version": "v0.99.0",
  "configuration": {
    "catalogers": {
      "requested": {
        "default": [
          "binary"
        ],
        "selection": []
      },
      "used": [
        "binary-cataloger",
        "cargo-auditable-binary-cataloger",
        "dotnet-portable-executable-cataloger",
        "go-module-binary-cataloger"
      ]
    },
    "data-generation": {
      "generate-cpes": true
    },
    "extra": null,
    "files": {
      "hashers": [
        "sha-1",
        "sha-256"
      ],
      "selection": "owned-files"
    },
    "packages": {
      "golang": {...},
      "java": {...},
      "javascript": {... },
      "linux-kernel": {... },
      "python": {...}
    },
    "relationships": {
      "exclude-binary-packages-with-file-ownership-overlap": true,
      "file-ownership": true,
      "file-ownership-overlap": true
    },
    "search": {
      "scope": "squashed"
    }
  }
}

click to see all options in an example

{
  "name": "syft",
  "version": "[not provided]",
  "configuration": {
    "catalogers": {
      "requested": {
        "default": [
          "binary"
        ],
        "selection": []
      },
      "used": [
        "binary-cataloger",
        "cargo-auditable-binary-cataloger",
        "dotnet-portable-executable-cataloger",
        "go-module-binary-cataloger"
      ]
    },
    "data-generation": {
      "generate-cpes": true
    },
    "extra": null,
    "files": {
      "hashers": [
        "sha-1",
        "sha-256"
      ],
      "selection": "owned-files"
    },
    "packages": {
      "golang": {
        "local-mod-cache-dir": "/Users/wagoodman/.local/share/rtx/installs/go/1.21.1/packages/pkg/mod",
        "proxies": [
          "https://proxy.golang.org",
          "direct"
        ],
        "search-local-mod-cache-licenses": false,
        "search-remote-licenses": false
      },
      "java-archive": {
        "include-indexed-archives": true,
        "include-unindexed-archives": false,
        "maven-base-url": "https://repo1.maven.org/maven2",
        "max-parent-recursive-depth": 5,
        "use-network": false
      },
      "javascript": {
        "npm-base-url": "https://registry.npmjs.org",
        "search-remote-licenses": false
      },
      "linux-kernel": {
        "catalog-modules": true
      },
      "python": {
        "guess-unpinned-requirements": false
      }
    },
    "relationships": {
      "exclude-binary-packages-with-file-ownership-overlap": true,
      "file-ownership": true,
      "file-ownership-overlap": true
    },
    "search": {
      "scope": "squashed"
    }
  }
}

This beats the current approach of using the catalogers: https://gist.github.com/wagoodman/57ed59a6d57600c23913071b8470175b

PRs broken off of this one

Follow up PRs

Add cataloger list command #2366

Partially implements #558

Fixes #2136
Closes #1731
Closes #1039
Closes #477

github-actions · 2022-12-16T18:19:32Z

Benchmark Test Results

Benchmark results from the latest changes vs base branch

name                                                       old time/op    new time/op    delta
ImagePackageCatalogers/alpmdb-cataloger-2                    11.4ms ± 1%    13.8ms ± 4%  +20.83%  (p=0.016 n=4+5)
ImagePackageCatalogers/ruby-gemspec-cataloger-2              1.31ms ± 1%    1.68ms ± 7%  +28.60%  (p=0.016 n=4+5)
ImagePackageCatalogers/python-package-cataloger-2            3.34ms ± 1%    4.00ms ± 4%  +19.64%  (p=0.008 n=5+5)
ImagePackageCatalogers/php-composer-installed-cataloger-2    1.08ms ± 1%    1.31ms ± 3%  +21.59%  (p=0.008 n=5+5)
ImagePackageCatalogers/javascript-package-cataloger-2         768µs ± 2%     932µs ± 3%  +21.35%  (p=0.008 n=5+5)
ImagePackageCatalogers/dpkgdb-cataloger-2                     886µs ± 1%    1116µs ± 5%  +25.94%  (p=0.008 n=5+5)
ImagePackageCatalogers/rpm-db-cataloger-2                    1.30ms ± 1%    1.62ms ± 5%  +25.00%  (p=0.008 n=5+5)
ImagePackageCatalogers/java-cataloger-2                      14.8ms ± 1%    17.4ms ± 5%  +17.32%  (p=0.008 n=5+5)
ImagePackageCatalogers/apkdb-cataloger-2                      891µs ± 2%    1084µs ± 2%  +21.64%  (p=0.008 n=5+5)
ImagePackageCatalogers/go-module-binary-cataloger-2          6.41µs ± 1%    7.41µs ± 4%  +15.58%  (p=0.008 n=5+5)
ImagePackageCatalogers/dotnet-deps-cataloger-2               1.37ms ± 2%    1.70ms ± 4%  +23.99%  (p=0.008 n=5+5)
ImagePackageCatalogers/portage-cataloger-2                    715µs ± 1%     882µs ± 3%  +23.45%  (p=0.008 n=5+5)
ImagePackageCatalogers/sbom-cataloger-2                      4.46ms ± 0%    5.30ms ± 1%  +18.79%  (p=0.008 n=5+5)
ImagePackageCatalogers/binary-cataloger-2                    3.90ms ± 1%    4.72ms ± 2%  +21.03%  (p=0.008 n=5+5)

name                                                       old alloc/op   new alloc/op   delta
ImagePackageCatalogers/alpmdb-cataloger-2                    5.26MB ± 0%    5.27MB ± 0%     ~     (p=0.095 n=5+5)
ImagePackageCatalogers/ruby-gemspec-cataloger-2               205kB ± 0%     205kB ± 0%   -0.08%  (p=0.032 n=5+5)
ImagePackageCatalogers/python-package-cataloger-2             963kB ± 0%     961kB ± 0%   -0.12%  (p=0.008 n=5+5)
ImagePackageCatalogers/php-composer-installed-cataloger-2     218kB ± 0%     217kB ± 0%   -0.11%  (p=0.032 n=5+5)
ImagePackageCatalogers/javascript-package-cataloger-2         159kB ± 0%     159kB ± 0%     ~     (p=0.444 n=5+5)
ImagePackageCatalogers/dpkgdb-cataloger-2                     200kB ± 0%     199kB ± 0%   -0.18%  (p=0.008 n=5+5)
ImagePackageCatalogers/rpm-db-cataloger-2                     303kB ± 0%     302kB ± 0%   -0.20%  (p=0.008 n=5+5)
ImagePackageCatalogers/java-cataloger-2                      3.49MB ± 0%    3.49MB ± 0%     ~     (p=0.548 n=5+5)
ImagePackageCatalogers/apkdb-cataloger-2                      182kB ± 0%     182kB ± 0%   -0.06%  (p=0.032 n=5+5)
ImagePackageCatalogers/go-module-binary-cataloger-2          1.12kB ± 0%    1.12kB ± 0%     ~     (all equal)
ImagePackageCatalogers/dotnet-deps-cataloger-2                374kB ± 0%     375kB ± 0%   +0.15%  (p=0.008 n=5+5)
ImagePackageCatalogers/portage-cataloger-2                    139kB ± 0%     138kB ± 0%   -0.07%  (p=0.032 n=5+5)
ImagePackageCatalogers/sbom-cataloger-2                       722kB ± 0%     722kB ± 0%   +0.02%  (p=0.008 n=5+5)
ImagePackageCatalogers/binary-cataloger-2                     656kB ± 0%     656kB ± 0%   +0.04%  (p=0.008 n=5+5)

name                                                       old allocs/op  new allocs/op  delta
ImagePackageCatalogers/alpmdb-cataloger-2                     85.7k ± 0%     85.7k ± 0%     ~     (p=0.683 n=5+5)
ImagePackageCatalogers/ruby-gemspec-cataloger-2               4.25k ± 0%     4.25k ± 0%     ~     (p=0.444 n=5+5)
ImagePackageCatalogers/python-package-cataloger-2             16.5k ± 0%     16.5k ± 0%   -0.05%  (p=0.008 n=5+5)
ImagePackageCatalogers/php-composer-installed-cataloger-2     5.50k ± 0%     5.50k ± 0%     ~     (p=0.556 n=4+5)
ImagePackageCatalogers/javascript-package-cataloger-2         3.33k ± 0%     3.33k ± 0%     ~     (all equal)
ImagePackageCatalogers/dpkgdb-cataloger-2                     4.47k ± 0%     4.47k ± 0%     ~     (all equal)
ImagePackageCatalogers/rpm-db-cataloger-2                     8.12k ± 0%     8.12k ± 0%     ~     (all equal)
ImagePackageCatalogers/java-cataloger-2                       57.5k ± 0%     57.5k ± 0%     ~     (p=0.111 n=5+5)
ImagePackageCatalogers/apkdb-cataloger-2                      5.23k ± 0%     5.23k ± 0%     ~     (p=0.444 n=5+5)
ImagePackageCatalogers/go-module-binary-cataloger-2            38.0 ± 0%      38.0 ± 0%     ~     (all equal)
ImagePackageCatalogers/dotnet-deps-cataloger-2                7.12k ± 0%     7.12k ± 0%     ~     (all equal)
ImagePackageCatalogers/portage-cataloger-2                    3.58k ± 0%     3.58k ± 0%     ~     (p=1.000 n=5+4)
ImagePackageCatalogers/sbom-cataloger-2                       24.4k ± 0%     24.4k ± 0%     ~     (all equal)
ImagePackageCatalogers/binary-cataloger-2                     22.2k ± 0%     22.2k ± 0%     ~     (all equal)

syft/file/selection.go

syft/create_sbom_config.go

Signed-off-by: Alex Goodman <[email protected]>

syft/create_sbom.go

syft/create_sbom_config.go

wagoodman · 2024-01-10T21:43:01Z

internal/task/file_tasks.go

+			}
+		})
+
+		result, err := digestsCataloger.Catalog(resolver, coordinates...)


from @willmurphyscode , blocking: we need to explicitly pass all coordinates, since there is no guarantee to have any results from a owned-files indication

I'm going to fix the functional problem in this PR, but to address the signature and generator issue I really should break that into a separate PR that I follow up with after this PR

I think this will have some play into the solution here #2487

internal/sbomsync/builder.go

Signed-off-by: Alex Goodman <[email protected]>

internal/task/file_tasks.go

Signed-off-by: Alex Goodman <[email protected]>

willmurphyscode

Thanks for all the careful thought in making the API and configs easier to use for the future.

* remove existing cataloging API Signed-off-by: Alex Goodman <[email protected]> * add file cataloging config Signed-off-by: Alex Goodman <[email protected]> * add package cataloging config Signed-off-by: Alex Goodman <[email protected]> * add configs for cross-cutting concerns Signed-off-by: Alex Goodman <[email protected]> * rename CLI option configs to not require import aliases later Signed-off-by: Alex Goodman <[email protected]> * update all nested structs for the Catalog struct Signed-off-by: Alex Goodman <[email protected]> * update Catalog cli options - add new cataloger selection options (selection and default) - remove the excludeBinaryOverlapByOwnership - deprecate "catalogers" flag - add new javascript configuration Signed-off-by: Alex Goodman <[email protected]> * migrate relationship capabilities to separate internal package Signed-off-by: Alex Goodman <[email protected]> * refactor golang cataloger to use configuration options when creating packages Signed-off-by: Alex Goodman <[email protected]> * create internal object to facilitate reading from and writing to an SBOM Signed-off-by: Alex Goodman <[email protected]> * create a command-like object (task) to facilitate partial SBOM creation Signed-off-by: Alex Goodman <[email protected]> * add cataloger selection capability - be able to parse string expressions into a set of resolved actions against sets - be able to use expressions to select/add/remove tasks to/from the final set of tasks to run Signed-off-by: Alex Goodman <[email protected]> * add package, file, and environment related tasks Signed-off-by: Alex Goodman <[email protected]> * update existing file catalogers to use nested UI elements Signed-off-by: Alex Goodman <[email protected]> * add CreateSBOMConfig that drives the SBOM creation process Signed-off-by: Alex Goodman <[email protected]> * capture SBOM creation info as a struct Signed-off-by: Alex Goodman <[email protected]> * add CreateSBOM() function Signed-off-by: Alex Goodman <[email protected]> * fix tests Signed-off-by: Alex Goodman <[email protected]> * update docs with SBOM selection help + breaking changes Signed-off-by: Alex Goodman <[email protected]> * fix multiple override default inputs Signed-off-by: Alex Goodman <[email protected]> * fix deprecation flag printing to stdout Signed-off-by: Alex Goodman <[email protected]> * refactor cataloger selection description to separate object Signed-off-by: Alex Goodman <[email protected]> * address review comments Signed-off-by: Alex Goodman <[email protected]> * keep expression errors and show specific suggestions only Signed-off-by: Alex Goodman <[email protected]> * address additional review feedback Signed-off-by: Alex Goodman <[email protected]> * address more review comments Signed-off-by: Alex Goodman <[email protected]> * addressed additional PR review feedback Signed-off-by: Alex Goodman <[email protected]> * fix file selection references Signed-off-by: Alex Goodman <[email protected]> * remove guess language data generation option Signed-off-by: Alex Goodman <[email protected]> * add tests for coordinatesForSelection Signed-off-by: Alex Goodman <[email protected]> * rename relationship attributes Signed-off-by: Alex Goodman <[email protected]> * add descriptions to relationships config fields Signed-off-by: Alex Goodman <[email protected]> * improve documentation around configuration options Signed-off-by: Alex Goodman <[email protected]> * add explicit errors around legacy config entries Signed-off-by: Alex Goodman <[email protected]> --------- Signed-off-by: Alex Goodman <[email protected]>

wagoodman mentioned this pull request Dec 16, 2022

Top-level API should be more composable #558

Closed

5 tasks

wagoodman force-pushed the refactor-cataloging-api branch from 57a9877 to 64f7af5 Compare December 16, 2022 18:12

wagoodman force-pushed the refactor-cataloging-api branch from 64f7af5 to 99f9d1e Compare December 16, 2022 18:24

This comment was marked as outdated.

Sign in to view

spiffcs mentioned this pull request Dec 19, 2022

bug: spdx checksum empty array; allow syft to generate SHA1 for spdx-tag-value documents #1404

Merged

This comment was marked as outdated.

Sign in to view

wagoodman mentioned this pull request Jan 11, 2023

Continue - Option to enable specific language or ecosystem cataloger #888

Closed

This was referenced Feb 16, 2023

Cataloger filtering cross matches wrong catalogers #1573

Closed

Fix cataloger selection to be more specific #1582

Merged

wagoodman self-assigned this Apr 26, 2023

This was referenced May 19, 2023

Minor cataloger improvements for the CLI #1831

Closed

Why are different catalogers disabled (by default) in different scenarios? #1776

Closed

wagoodman mentioned this pull request Oct 19, 2023

Files are not output in CycloneDX formats #1710

Open

wagoodman added the breaking-change Change is not backwards compatible label Nov 14, 2023

wagoodman changed the title ~~Add SBOM builder configuration~~ Replace core SBOM-creation API with builder pattern Nov 14, 2023

github-actions bot removed the breaking-change Change is not backwards compatible label Nov 16, 2023

wagoodman force-pushed the refactor-cataloging-api branch from d36be02 to efb97f4 Compare November 16, 2023 20:48

This was referenced Nov 20, 2023

Improve cataloger selection capabilities #1039

Closed

When specify java-cataloger, java-pom-cataloger will also be selected #2136

Closed

wagoodman force-pushed the refactor-cataloging-api branch 5 times, most recently from 624ff9f to f8aaae5 Compare November 27, 2023 20:16