Updates to add flexibility to lookasides and tagless handling:

- Added template "macros" to allow for complex --cdn-url specifications ( {{.Name}}, {{.Branch}}, etc.) - Added --cdn <distro> option with pre-set URL patterns to simplify use - Kept default fallback behavior and search patterns of --cdn-url if templates are not used - Removed "--altlookaside" option, as the above features make it redundant - Automatic tagless: If a proper git version/imports tag isn't found, a "tagless-mode" import will be attempted automatically at run-time - taglessmode command line option kept in case user still wants to force a tagless import -Skip Grube
2024-12-04 02:16:26 +00:00 · 2023-02-22 23:23:47 -05:00 · 2023-02-22 23:23:47 -05:00 · 3237794071
commit 3237794071
parent 06ffd1e507
5 changed files with 194 additions and 54 deletions
--- a/README.md
+++ b/README.md
@ -1,14 +1,14 @@
 # srpmproc
 Upstream package importer with auto patching. Reference implementation for OpenPatch

-# Usage
+## Usage
 ```
 Usage:
  srpmproc [flags]
  srpmproc [command]

 Available Commands:
-  fetch
+  fetch       
  help        Help about any command

 Flags:
@ -16,7 +16,8 @@ Flags:
      --basic-username string           Basic auth username
      --branch-prefix string            Branch prefix (replaces import-branch-prefix) (default "r")
      --branch-suffix string            Branch suffix to use for imported branches
-      --cdn-url string                  CDN URL to download blobs from (default "https://git.centos.org/sources")
+      --cdn string                      CDN URL shortcuts for well-known distros, auto-assigns --cdn-url.  Valid values:  rocky8, rocky, fedora, centos, centos-stream.  Setting this overrides --cdn-url
+      --cdn-url string                  CDN URL to download blobs from. Simple URL follows default rocky/centos patterns. Can be customized using macros (see docs) (default "https://git.centos.org/sources")
      --git-committer-email string      Email of committer (default "rockyautomation@rockylinux.org")
      --git-committer-name string       Name of committer (default "rockyautomation")
  -h, --help                            help for srpmproc
@ -28,6 +29,8 @@ Flags:
      --no-dup-mode                     If enabled, skips already imported tags
      --no-storage-download             If enabled, blobs are always downloaded from upstream
      --no-storage-upload               If enabled, blobs are not uploaded to blob storage
+      --package-release string          Package release to fetch
+      --package-version string          Package version to fetch
      --rpm-prefix string               Where to retrieve SRPM content. Only used when source-rpm is not a local file (default "https://git.centos.org/rpms")
      --single-tag string               If set, only this tag is imported
      --source-rpm string               Location of RPM to process
@ -35,9 +38,42 @@ Flags:
      --ssh-user string                 SSH User (default "git")
      --storage-addr string             Bucket to use as blob storage
      --strict-branch-mode              If enabled, only branches with the calculated name are imported and not prefix only
+      --taglessmode                     Tagless mode:  If set, pull the latest commit from the branch and determine version numbers from spec file.  This is auto-tried if tags aren't found.
      --tmpfs-mode string               If set, packages are imported to path and patched but not pushed
      --upstream-prefix string          Upstream git repository prefix
      --version int                     Upstream version

 Use "srpmproc [command] --help" for more information about a command.
 ```
+
+<br />
+
+## Examples:
+
+1. Import the kernel package from git.centos.org/rpms/, to local folder /opt/gitroot/rpms/kernel.git/ .  Download the lookaside source tarballs from the default CentOS file server location to local folder `/opt/fake_s3/` .  We want to grab branch "c8" (import prefix plus RHEL version), and it will be committed as branch "r8" (branch prefix plus RHEL version).  This assumes that `/opt/fake_s3` exists, and `/opt/gitroot/rpms/kernel.git` exists and is a git repository of some kind (even an empty one).
+
+```
+srpmproc --branch-prefix "r"  --import-branch-prefix "c"  --rpm-prefix "https://git.centos.org/rpms" --version 8 --storage-addr file:///opt/fake_s3  --upstream-prefix file:///opt/gitroot   --cdn centos --strict-branch-mode --source-rpm kernel
+```
+
+<br />
+
+## CDN and --cdn-url
+The --cdn-url option allows for Go-style templates to craft complex URL patterns.  These templates are: `{{.Name}}` (package name), `{{.Hash}}` (hash of lookaside file), `{{.Hashtype}}` (hash type of file, like "sha256" or "sha512"), `{{.Branch}}` (the branch we are importing), and `{{.Filename}}` (the lookaside file's name as it appears in SOURCES/).  You can add these values as part of --cdn-url to craft your lookaside pattern.
+
+
+For example, if I wanted my lookaside downloads to come from CentOS 9 Stream, I would use as part of my command:
+```
+--cdn-url "https://sources.stream.centos.org/sources/rpms/{{.Name}}/{{.Filename}}/{{.Hashtype}}/{{.Hash}}/{{.Filename}}"
+```
+
+
+**Default Behavior:**  If these templates are not used, the default behavior of `--cdn-url` is to fall back on the traditional RHEL import pattern:  `<CDN_URL>/<NAME>/<BRANCH>/<HASH>` .  If that fails, a further fallback is attempted, the simple: `<CDN_URL>/<HASH>`.  These cover the common Rocky Linux and RHEL/CentOS imports if the base lookaside URL is the only thing given.  If no `--cdn-url` is specified, it defaults to "https://git.centos.org/sources" (for RHEL imports into Rocky Linux)
+
+
+**CDN Shorthand:** For convenience, some lookaside patterns for popular distros are provided via the `--cdn` option.  You can specify this without needing to use the longer `--cdn-url`.  For example, when importing from CentOS 9 Stream, you could use `--cdn centos-stream`
+
+
+
+
+
--- a/cmd/srpmproc/main.go
+++ b/cmd/srpmproc/main.go
@ -58,7 +58,7 @@ var (
 	packageVersion       string
 	packageRelease       string
 	taglessMode          bool
-	altLookAside         bool
+	cdn                  string
 )

 var root = &cobra.Command{
@ -96,7 +96,7 @@ func mn(_ *cobra.Command, _ []string) {
 		PackageVersion:       packageVersion,
 		PackageRelease:       packageRelease,
 		TaglessMode:          taglessMode,
-		AltLookAside:         altLookAside,
+		Cdn:                  cdn,
 	})
 	if err != nil {
 		log.Fatal(err)
@ -131,7 +131,7 @@ func main() {
 	root.Flags().StringVar(&rpmPrefix, "rpm-prefix", "https://git.centos.org/rpms", "Where to retrieve SRPM content. Only used when source-rpm is not a local file")
 	root.Flags().StringVar(&importBranchPrefix, "import-branch-prefix", "c", "Import branch prefix")
 	root.Flags().StringVar(&branchPrefix, "branch-prefix", "r", "Branch prefix (replaces import-branch-prefix)")
-	root.Flags().StringVar(&cdnUrl, "cdn-url", "https://git.centos.org/sources", "CDN URL to download blobs from")
+	root.Flags().StringVar(&cdnUrl, "cdn-url", "https://git.centos.org/sources", "CDN URL to download blobs from. Simple URL follows default rocky/centos patterns. Can be customized using macros (see docs)")
 	root.Flags().StringVar(&singleTag, "single-tag", "", "If set, only this tag is imported")
 	root.Flags().BoolVar(&noDupMode, "no-dup-mode", false, "If enabled, skips already imported tags")
 	root.Flags().BoolVar(&moduleMode, "module-mode", false, "If enabled, imports a module instead of a package")
@ -146,8 +146,8 @@ func main() {
 	root.Flags().StringVar(&basicPassword, "basic-password", "", "Basic auth password")
 	root.Flags().StringVar(&packageVersion, "package-version", "", "Package version to fetch")
 	root.Flags().StringVar(&packageRelease, "package-release", "", "Package release to fetch")
-	root.Flags().BoolVar(&taglessMode, "taglessmode", false, "Tagless mode:  If set, pull the latest commit from a branch, and determine version info from spec file (aka upstream versions aren't tagged)")
-	root.Flags().BoolVar(&altLookAside, "altlookaside", false, "If set, uses the new CentOS Stream lookaside pattern (https://<SITE_PREFIX>/<RPM_NAME>/<FILE_NAME>/<SHA_VERSION>/<SHA_SUM>/<FILE_NAME>)")
+	root.Flags().BoolVar(&taglessMode, "taglessmode", false, "Tagless mode:  If set, pull the latest commit from the branch and determine version numbers from spec file.  This is auto-tried if tags aren't found.")
+	root.Flags().StringVar(&cdn, "cdn", "", "CDN URL shortcuts for well-known distros, auto-assigns --cdn-url.  Valid values:  rocky8, rocky, fedora, centos, centos-stream.  Setting this overrides --cdn-url")

 	if err := root.Execute(); err != nil {
 		log.Fatal(err)
--- a/pkg/data/process.go
+++ b/pkg/data/process.go
@ -59,5 +59,5 @@ type ProcessData struct {
 	PackageVersion       string
 	PackageRelease       string
 	TaglessMode          bool
-	AltLookAside         bool
+	Cdn                  string
 }
--- a/pkg/modes/git.go
+++ b/pkg/modes/git.go
@ -21,12 +21,14 @@
 package modes

 import (
+	"bytes"
 	"fmt"
 	"io/ioutil"
 	"net/http"
 	"path/filepath"
 	"sort"
 	"strings"
+	"text/template"
 	"time"

 	"github.com/go-git/go-git/v5/plumbing/transport"
@ -341,44 +343,69 @@ func (g *GitMode) WriteSource(pd *data.ProcessData, md *data.ModeData) error {
 			} else {

 				url := ""
-				// Alternate lookaside logic:  if enabled, we pull from a new URL pattern
-				if !pd.AltLookAside {
-					url = fmt.Sprintf("%s/%s/%s/%s", pd.CdnUrl, md.Name, branchName, hash)
-				} else {
-					// We first need the hash algorithm based on length of hash:
-					hashType := "sha512"
-					switch len(hash) {
-					case 128:
-						hashType = "sha512"
-					case 64:
-						hashType = "sha256"
-					case 40:
-						hashType = "sha1"
-					case 32:
-						hashType = "md5"
+
+				// We need to figure out the hashtype for templating purposes:
+				hashType := "sha512"
+				switch len(hash) {
+				case 128:
+					hashType = "sha512"
+				case 64:
+					hashType = "sha256"
+				case 40:
+					hashType = "sha1"
+				case 32:
+					hashType = "md5"
+				}
+
+				// need the name of the file without "SOURCES/":
+				fileName := strings.Split(path, "/")[1]
+
+				// Feed our template info to ProcessUrl and transform to the real values: ( {{.Name}}, {{.Branch}}, {{.Hash}}, {{.Hashtype}}, {{.Filename}} )
+				url, err = ProcessUrl(pd.CdnUrl, md.Name, branchName, hash, hashType, fileName)
+				if err != nil {
+					return fmt.Errorf("Could not process CDN URL template(s) in string ( {{ .Variable }} )")
+				}
+
+				var req *http.Request
+				var resp *http.Response
+
+				// Download the --cdn-url given, but *only* if it contains template strings ( {{.Name}} , {{.Hash}} , etc. )
+				// Otherwise we need to fall back to the traditional cdn-url patterns
+				if strings.Contains(pd.CdnUrl, "{{") && strings.Contains(pd.CdnUrl, "}}") {
+					pd.Log.Printf("downloading %s", url)
+
+					req, err := http.NewRequest("GET", url, nil)
+					if err != nil {
+						return fmt.Errorf("could not create new http request: %v", err)
 					}
+					req.Header.Set("Accept-Encoding", "*")

-					// need the name of the file without "SOURCES/":
-					fileName := strings.Split(path, "/")[1]
-
-					// Alt. lookaside url is of the form: <cdn> / <name> / <filename> / <hashtype> / <hash> / <filename>
-					url = fmt.Sprintf("%s/%s/%s/%s/%s/%s", pd.CdnUrl, md.Name, fileName, hashType, hash, fileName)
+					resp, err = client.Do(req)
+					if err != nil {
+						return fmt.Errorf("could not download dist-git file: %v", err)
+					}
 				}

-				pd.Log.Printf("downloading %s", url)
-
-				req, err := http.NewRequest("GET", url, nil)
-				if err != nil {
-					return fmt.Errorf("could not create new http request: %v", err)
+				// Default cdn-url:  If we don't have a templated download string, try the default <SITE>/<PKG>/<BRANCH>/<HASH> pattern:
+				if resp == nil || resp.StatusCode != http.StatusOK {
+					url = fmt.Sprintf("%s/%s/%s/%s", pd.CdnUrl, md.Name, branchName, hash)
+					pd.Log.Printf("Attempting default URL: %s", url)
+					req, err = http.NewRequest("GET", url, nil)
+					if err != nil {
+						return fmt.Errorf("could not create new http request: %v", err)
+					}
+					req.Header.Set("Accept-Encoding", "*")
+					resp, err = client.Do(req)
+					if err != nil {
+						return fmt.Errorf("could not download dist-git file: %v", err)
+					}
 				}
-				req.Header.Set("Accept-Encoding", "*")

-				resp, err := client.Do(req)
-				if err != nil {
-					return fmt.Errorf("could not download dist-git file: %v", err)
-				}
-				if resp.StatusCode != http.StatusOK {
+				// If the default URL fails, we have one more pattern to try.  The simple <SITE>/<HASH> pattern
+				// If this one fails, we are truly lost, and have to bail out w/ an error:
+				if resp == nil || resp.StatusCode != http.StatusOK {
 					url = fmt.Sprintf("%s/%s", pd.CdnUrl, hash)
+					pd.Log.Printf("Attempting 2nd fallback URL: %s", url)
 					req, err = http.NewRequest("GET", url, nil)
 					if err != nil {
 						return fmt.Errorf("could not create new http request: %v", err)
@ -458,3 +485,34 @@ func (g *GitMode) ImportName(pd *data.ProcessData, md *data.ModeData) string {

 	return strings.Replace(strings.TrimPrefix(md.TagBranch, "refs/heads/"), "%", "_", -1)
 }
+
+// Given a cdnUrl string as input, return same string, but with substituted
+// template values ( {{.Name}} , {{.Hash}}, {{.Filename}}, etc. )
+func ProcessUrl(cdnUrl string, name string, branch string, hash string, hashtype string, filename string) (string, error) {
+
+	// These 5 {{ .Value }} items are possible in our templated string:
+	type Lookaside struct {
+		Name     string
+		Branch   string
+		Hash     string
+		Hashtype string
+		Filename string
+	}
+
+	tmpUrl := Lookaside{name, branch, hash, hashtype, filename}
+
+	tmpl, err := template.New("").Parse(cdnUrl)
+	if err != nil {
+		panic(err)
+	}
+
+	var result bytes.Buffer
+	err = tmpl.Execute(&result, tmpUrl)
+
+	if err != nil {
+		panic(err)
+	}
+
+	return result.String(), nil
+
+}
--- a/pkg/srpmproc/process.go
+++ b/pkg/srpmproc/process.go
@ -101,8 +101,13 @@ type ProcessDataRequest struct {
 	PackageVersion string
 	PackageRelease string

-	TaglessMode  bool
-	AltLookAside bool
+	TaglessMode bool
+	Cdn         string
+}
+
+type LookasidePath struct {
+	Distro string
+	Url    string
 }

 func gitlabify(str string) string {
@ -113,7 +118,29 @@ func gitlabify(str string) string {
 	return strings.Replace(str, "+", "plus", -1)
 }

+// List of distros and their lookaside patterns
+// If we find one of these passed as --cdn (ex: "--cdn fedora"), then we override, and assign this URL to be our --cdn-url
+func StaticLookasides() []LookasidePath {
+
+	centos := LookasidePath{Distro: "centos", Url: "https://git.centos.org/sources/{{.Name}}/{{.Branch}}/{{.Hash}}"}
+	centosStream := LookasidePath{Distro: "centos-stream", Url: "https://sources.stream.centos.org/sources/rpms/{{.Name}}/{{.Filename}}/{{.Hashtype}}/{{.Hash}}/{{.Filename}}"}
+	rocky8 := LookasidePath{Distro: "rocky8", Url: "https://rocky-linux-sources-staging.a1.rockylinux.org/{{.Hash}}"}
+	rocky := LookasidePath{Distro: "rocky", Url: "https://sources.build.resf.org/{{.Hash}}"}
+	fedora := LookasidePath{Distro: "fedora", Url: "https://src.fedoraproject.org/repo/pkgs/{{.Name}}/{{.Filename}}/{{.Hashtype}}/{{.Hash}}/{{.Filename}}"}
+
+	return []LookasidePath{centos, centosStream, rocky8, rocky, fedora}
+
+}
+
 func NewProcessData(req *ProcessDataRequest) (*data.ProcessData, error) {
+
+	// Build the logger to use for the data import
+	var writer io.Writer = os.Stdout
+	if req.LogWriter != nil {
+		writer = req.LogWriter
+	}
+	logger := log.New(writer, "", log.LstdFlags)
+
 	// Set defaults
 	if req.ModulePrefix == "" {
 		req.ModulePrefix = ModulePrefixCentOS
@ -139,13 +166,29 @@ func NewProcessData(req *ProcessDataRequest) (*data.ProcessData, error) {
 	if req.BranchPrefix == "" {
 		req.BranchPrefix = "r"
 	}
-	if req.CdnUrl == "" && !req.AltLookAside {
+	if req.CdnUrl == "" {
 		req.CdnUrl = "https://git.centos.org/sources"
 	}
-	// If altlookaside is enabled, and the CdnUrl hasn't been changed, then automatically set it to the default
-	// CentOS Stream (the new pattern very much won't work with the old git.centos.org/sources site)
-	if (req.CdnUrl == "https://git.centos.org/sources" || req.CdnUrl == "") && req.AltLookAside {
-		req.CdnUrl = "https://sources.stream.centos.org/sources/rpms"
+
+	// If a Cdn distro is defined, loop through StaticLookasides() array of structs,
+	// see if we have a match to --cdn (matching values are things like fedora, centos, rocky8, etc.)
+	// If we match, then we want to short-circuit the CdnUrl to the assigned distro's one
+	if req.Cdn != "" {
+
+		var foundDistro = false
+
+		for _, distro := range StaticLookasides() {
+			if distro.Distro == strings.ToLower(req.Cdn) {
+				foundDistro = true
+				req.CdnUrl = distro.Url
+				logger.Printf("Discovered --cdn distro: %s .  Using override CDN URL Pattern: %s", distro.Distro, req.CdnUrl)
+				break
+			}
+		}
+
+		if foundDistro == false {
+			return nil, fmt.Errorf("Error, distro name given as --cdn argument is not valid.")
+		}
 	}

 	// Validate required
@ -214,12 +257,6 @@ func NewProcessData(req *ProcessDataRequest) (*data.ProcessData, error) {
 		reqFsCreator = req.FsCreator
 	}

-	var writer io.Writer = os.Stdout
-	if req.LogWriter != nil {
-		writer = req.LogWriter
-	}
-	logger := log.New(writer, "", log.LstdFlags)
-
 	if req.TmpFsMode != "" {
 		logger.Printf("using tmpfs dir: %s", req.TmpFsMode)
 		fsCreator = func(branch string) (billy.Filesystem, error) {
@ -276,7 +313,7 @@ func NewProcessData(req *ProcessDataRequest) (*data.ProcessData, error) {
 		PackageVersion:       req.PackageVersion,
 		PackageRelease:       req.PackageRelease,
 		TaglessMode:          req.TaglessMode,
-		AltLookAside:         req.AltLookAside,
+		Cdn:                  req.Cdn,
 	}, nil
 }

@ -334,6 +371,7 @@ func ProcessRPM(pd *data.ProcessData) (*srpmprocpb.ProcessResponse, error) {
 		list, err := remote.List(&git.ListOptions{
 			Auth: pd.Authenticator,
 		})
+
 		if err != nil {
 			log.Println("ignoring no-dup-mode")
 		} else {
@ -367,6 +405,14 @@ func ProcessRPM(pd *data.ProcessData) (*srpmprocpb.ProcessResponse, error) {
 		}
 	}

+	// If we have no valid branches to consider, then we'll automatically switch to attempt a tagless import:
+	if len(md.Branches) <= 0 {
+		log.Println("No valid tags (refs/tags/imports/*) found in repository!  Switching to perform a tagless import.")
+		pd.TaglessMode = true
+		result, err := processRPMTagless(pd)
+		return result, err
+	}
+
 	for _, branch := range md.Branches {
 		md.Repo = &sourceRepo
 		md.Worktree = &sourceWorktree