fix(legacy): fix PACKAGE FORMAT CHANGE warnings in legacy cache export

Reduce PACKAGE FORMAT CHANGE warnings from 6,149 to ~3,200 by aligning the legacy export field values with what the Node build-classifier extracts from filenames. classify.go: - Split solaris/illumos/sunos into three separate OS patterns (Node triplet.js treats them as distinct values; lumping all to OSSunOS caused 1,483 drops) - Add mips64r6/mips64r6el arch patterns before mips64 to prevent prefix match - Add mips64le/mips64el distinct patterns before mips64 baseline - Fix amd64[_-]?v2/v3/v4 regex to match underscore form (e.g. pathman amd64_v2) buildmeta.go: - Add ArchMIPS64R6, ArchMIPS64R6EL, ArchMIPS64LE, ArchMIPSLE constants legacy.go legacyFieldBackport: - Remove x86_64_v2/v3/v4 → x86_64 translation (classifier knows these values) - Remove mips64r6/mips64r6el → mips64 translation (same reason) - Add mipsle → mipsel translation (tpm['mipsle']={arch:'mipsel'}) - Add mips64le → mips64el translation (tpm['mips64le']={arch:'mips64el'}) legacy.go legacyARMArchFromFilename: - Check "armv7" before "gnueabihf" so armv7-unknown-linux-gnueabihf → armv7 - Add armv6hf → armhf (shellcheck naming, tpm['armv6hf']=ARMHF) - Add arm-5 → armel (Gitea naming: patternToTerms converts arm-5 → armv5 → armel) - Add arm-7 → armv7 (Gitea naming: patternToTerms converts arm-7 → armv7) - Add armv5 → armel (tpm['armv5']=T.ARMEL)
2026-04-06 18:36:50 +00:00 · 2026-03-11 17:27:37 -06:00
parent 899900695c
commit 4f09649d30
4 changed files with 178 additions and 35 deletions
--- a/internal/buildmeta/buildmeta.go
+++ b/internal/buildmeta/buildmeta.go
@@ -50,10 +50,12 @@ const (
 	ArchRISCV64 Arch = "riscv64"
 	ArchS390X   Arch = "s390x"
 	ArchLoong64 Arch = "loong64"
-	ArchMIPS64LE Arch = "mips64le"
-	ArchMIPS64   Arch = "mips64"
-	ArchMIPSLE   Arch = "mipsle"
-	ArchMIPS     Arch = "mips"
+	ArchMIPS64LE   Arch = "mips64le"
+	ArchMIPS64     Arch = "mips64"
+	ArchMIPS64R6EL Arch = "mips64r6el"
+	ArchMIPS64R6   Arch = "mips64r6"
+	ArchMIPSLE     Arch = "mipsle"
+	ArchMIPS       Arch = "mips"

 	// Universal (fat) binary architectures for macOS.
 	ArchUniversal1 Arch = "universal1" // PPC + x86 (Rosetta 1 era)
--- a/internal/classify/classify.go
+++ b/internal/classify/classify.go
@@ -77,7 +77,11 @@ var osPatterns = []struct {
 	{buildmeta.OSOpenBSD, regexp.MustCompile(`(?i)` + b + `openbsd` + bEnd)},
 	{buildmeta.OSNetBSD, regexp.MustCompile(`(?i)` + b + `netbsd` + bEnd)},
 	{buildmeta.OSDragonFly, regexp.MustCompile(`(?i)` + b + `dragonfly(?:bsd)?` + bEnd)},
-	{buildmeta.OSSunOS, regexp.MustCompile(`(?i)` + b + `(?:sunos|solaris|illumos)` + bEnd)},
+	// solaris, illumos, and sunos are distinct OS values in the Node build-classifier.
+	// Keep them separate so the legacy cache matches what the classifier extracts.
+	{buildmeta.OSSolaris, regexp.MustCompile(`(?i)` + b + `solaris` + bEnd)},
+	{buildmeta.OSIllumos, regexp.MustCompile(`(?i)` + b + `illumos` + bEnd)},
+	{buildmeta.OSSunOS, regexp.MustCompile(`(?i)` + b + `sunos` + bEnd)},
 	{buildmeta.OSAIX, regexp.MustCompile(`(?i)` + b + `aix` + bEnd)},
 	{buildmeta.OSAndroid, regexp.MustCompile(`(?i)` + b + `android` + bEnd)},
 	{buildmeta.OSPlan9, regexp.MustCompile(`(?i)` + b + `plan9` + bEnd)},
@@ -102,9 +106,9 @@ var archPatterns = []struct {
 	// Universal/fat binaries before specific arches.
 	{buildmeta.ArchUniversal2, regexp.MustCompile(`(?i)` + b + `(?:universal2?|fat)` + bEnd)},
 	// amd64 micro-levels before baseline — "amd64v3" must not fall through to amd64.
-	{buildmeta.ArchAMD64v4, regexp.MustCompile(`(?i)(?:x86[_-]64[_-]v4|amd64v4|v4-amd64)`)},
-	{buildmeta.ArchAMD64v3, regexp.MustCompile(`(?i)(?:x86[_-]64[_-]v3|amd64v3|v3-amd64)`)},
-	{buildmeta.ArchAMD64v2, regexp.MustCompile(`(?i)(?:x86[_-]64[_-]v2|amd64v2|v2-amd64)`)},
+	{buildmeta.ArchAMD64v4, regexp.MustCompile(`(?i)(?:x86[_-]64[_-]v4|amd64[_-]?v4|v4-amd64)`)},
+	{buildmeta.ArchAMD64v3, regexp.MustCompile(`(?i)(?:x86[_-]64[_-]v3|amd64[_-]?v3|v3-amd64)`)},
+	{buildmeta.ArchAMD64v2, regexp.MustCompile(`(?i)(?:x86[_-]64[_-]v2|amd64[_-]?v2|v2-amd64)`)},
 	// amd64 baseline before x86 — "x86_64" must not match as x86.
 	{buildmeta.ArchAMD64, regexp.MustCompile(`(?i)(?:x86[_-]64|amd64|x64|64-?bit)`)},
 	// arm64 before armv7/armv6 — "aarch64" must not match as arm.
@@ -119,6 +123,9 @@ var archPatterns = []struct {
 	{buildmeta.ArchRISCV64, regexp.MustCompile(`(?i)riscv64`)},
 	{buildmeta.ArchS390X, regexp.MustCompile(`(?i)s390x`)},
 	{buildmeta.ArchLoong64, regexp.MustCompile(`(?i)loong(?:arch)?64`)},
+	// mips64r6 before mips64 — "mips64r6" contains "mips64" as a prefix.
+	{buildmeta.ArchMIPS64R6EL, regexp.MustCompile(`(?i)mips64r6e(?:l|le)`)},
+	{buildmeta.ArchMIPS64R6, regexp.MustCompile(`(?i)mips64r6`)},
 	{buildmeta.ArchMIPS64LE, regexp.MustCompile(`(?i)mips64(?:el|le)`)},
 	{buildmeta.ArchMIPS64, regexp.MustCompile(`(?i)mips64`)},
 	{buildmeta.ArchMIPSLE, regexp.MustCompile(`(?i)mips(?:el|le)`)},
--- a/internal/storage/legacy.go
+++ b/internal/storage/legacy.go
@@ -82,15 +82,19 @@ func (a Asset) toLegacy() LegacyAsset {
 //   - universal2/universal1 → x86_64: classifier maps "universal" in filename
 //     to x86_64. The darwin WATERFALL falls back aarch64→x86_64, so arm64
 //     users still receive these builds.
-//   - x86_64_v2 → x86_64: classifier doesn't recognize micro-arch level suffixes.
-//   - mips64r6/mips64r6el → mips64: MIPS Release 6 variants map to the base arch.
-//   - ARM (filename-based): gnueabihf/armhf→armhf, armel→armel, armv5→armel,
-//     armv7a→armv7a. Go normalizes these; Node classifier preserves the
-//     original Debian/Rust naming. See legacyARMArchFromFilename.
+//   - mipsle → mipsel: classifier normalizes both spellings to "mipsel".
+//   - mips64le → mips64el: classifier normalizes both spellings to "mips64el".
+//   - ARM (filename-based): explicit armvN takes priority over ABI tags;
+//     gnueabihf/armhf→armhf, armel→armel, armv5→armel, armv7a→armv7a.
+//     Go normalizes these; Node classifier preserves the original naming.
+//     See legacyARMArchFromFilename.
 //
 // Note: solaris/illumos/sunos are kept as-is. The build-classifier (triplet.js)
 // recognizes all three as distinct values, and the live cache uses them directly.
 //
+// Note: x86_64_v2/v3/v4 and mips64r6/mips64r6el are kept as-is. The
+// build-classifier knows these exact values and expects them to match.
+//
 // Package-specific rules replicate per-package overrides in production's releases.js:
 //   - ffmpeg: Windows .gz → .exe  (prod releases.js: rel.ext = 'exe')
 func legacyFieldBackport(pkg string, a Asset) Asset {
@@ -99,14 +103,12 @@ func legacyFieldBackport(pkg string, a Asset) Asset {
 		a.Arch = "x86_64"
 	}

-	// x86_64 micro-arch levels: classifier doesn't know these suffixes.
-	if a.Arch == "x86_64_v2" || a.Arch == "x86_64_v3" || a.Arch == "x86_64_v4" {
-		a.Arch = "x86_64"
+	// MIPS spelling normalization: classifier maps both spellings to the "el" form.
+	if a.Arch == "mipsle" {
+		a.Arch = "mipsel"
 	}
-
-	// MIPS Release 6 variants: map to the base mips64 arch.
-	if a.Arch == "mips64r6" || a.Arch == "mips64r6el" {
-		a.Arch = "mips64"
+	if a.Arch == "mips64le" {
+		a.Arch = "mips64el"
 	}

 	// ARM arch: the Node classifier re-parses filenames and expects the cache
@@ -136,18 +138,47 @@ func legacyFieldBackport(pkg string, a Asset) Asset {
 // Go canonical arch value already matches what the classifier would extract.
 //
 // The Node classifier's extraction rules differ from Go's normalization:
-//   - gnueabihf (Rust triplet) / armhf (Debian) → "armhf" (not "armv6" or "armv7")
+//   - armv7a (explicit) → "armv7a" (not "armv7")
+//   - armv7 (explicit, e.g. "armv7-unknown-linux-gnueabihf") → "armv7"
+//     The explicit version number takes priority over the ABI suffix.
+//   - arm-5 / arm-7 (Gitea naming: "linux-arm-5", "linux-arm-7") → "armel" / "armv7"
+//     patternToTerms converts "arm-5" → "armv5" and "arm-7" → "armv7".
+//   - armv6hf (shellcheck naming) → "armhf" (tpm['armv6hf'] = ARMHF)
+//   - gnueabihf (Rust triplet, no explicit armvN) → "armhf"
+//   - armhf (Debian armhf) → "armhf"
 //   - armel (Debian soft-float ABI) → "armel" (not "armv6")
-//   - armv5 → "armel" (Node tiered map: armv5 falls back to armel)
-//   - armv7a → "armv7a" (not "armv7")
+//   - armv5 (explicit) → "armel" (Node tiered map: armv5 falls back to armel)
 func legacyARMArchFromFilename(filename string) string {
 	lower := strings.ToLower(filename)
-	if strings.Contains(lower, "gnueabihf") || strings.Contains(lower, "armhf") {
-		return "armhf"
-	}
+	// armv7a before armv7 — "armv7a" contains "armv7" as a prefix.
 	if strings.Contains(lower, "armv7a") {
 		return "armv7a"
 	}
+	// Explicit armv7 in filename: takes priority over ABI suffix (gnueabihf).
+	// e.g. "armv7-unknown-linux-gnueabihf" → classifier extracts "armv7".
+	if strings.Contains(lower, "armv7") {
+		return "armv7"
+	}
+	// armv6hf (shellcheck naming): tpm['armv6hf'] = ARMHF → "armhf".
+	if strings.Contains(lower, "armv6hf") {
+		return "armhf"
+	}
+	// Gitea arm-N naming: "linux-arm-5" → patternToTerms → "armv5" → armel.
+	if strings.Contains(lower, "arm-5") {
+		return "armel"
+	}
+	// Gitea arm-N naming: "linux-arm-7" → patternToTerms → "armv7" → armv7.
+	if strings.Contains(lower, "arm-7") {
+		return "armv7"
+	}
+	// Rust gnueabihf triplet (no explicit armvN): classifier → "armhf".
+	if strings.Contains(lower, "gnueabihf") {
+		return "armhf"
+	}
+	// Debian armhf (hard-float ABI): classifier → "armhf".
+	if strings.Contains(lower, "armhf") {
+		return "armhf"
+	}
 	if strings.Contains(lower, "armel") {
 		return "armel"
 	}
--- a/internal/storage/legacy_test.go
+++ b/internal/storage/legacy_test.go
@@ -231,8 +231,9 @@ func TestExportLegacyTranslations(t *testing.T) {
 		}
 	})

-	t.Run("x86_64_v2_translated_to_x86_64", func(t *testing.T) {
-		// Micro-arch level suffixes (v2/v3/v4) are not recognized by the classifier.
+	t.Run("x86_64_v2_kept_as_is", func(t *testing.T) {
+		// Micro-arch level suffixes (v2/v3/v4): classifier has tpm entries for
+		// amd64_v2 → x86_64_v2, so the cache must match with the canonical value.
 		pd := storage.PackageData{
 			Assets: []storage.Asset{
 				{Filename: "tool-linux-x86_64_v2.tar.gz", OS: "linux", Arch: "x86_64_v2", Format: ".tar.gz"},
@@ -242,12 +243,13 @@ func TestExportLegacyTranslations(t *testing.T) {
 		if len(lc.Releases) != 1 {
 			t.Fatalf("releases = %d, want 1", len(lc.Releases))
 		}
-		if lc.Releases[0].Arch != "x86_64" {
-			t.Errorf("arch = %q, want x86_64", lc.Releases[0].Arch)
+		if lc.Releases[0].Arch != "x86_64_v2" {
+			t.Errorf("arch = %q, want x86_64_v2 (kept as-is, classifier knows this value)", lc.Releases[0].Arch)
 		}
 	})

-	t.Run("mips64r6_translated_to_mips64", func(t *testing.T) {
+	t.Run("mips64r6_kept_as_is", func(t *testing.T) {
+		// mips64r6/mips64r6el: classifier has tpm entries for these exact values.
 		pd := storage.PackageData{
 			Assets: []storage.Asset{
 				{Filename: "tool-linux-mips64r6.tar.gz", OS: "linux", Arch: "mips64r6", Format: ".tar.gz"},
@@ -258,10 +260,43 @@ func TestExportLegacyTranslations(t *testing.T) {
 		if len(lc.Releases) != 2 {
 			t.Fatalf("releases = %d, want 2", len(lc.Releases))
 		}
-		for _, r := range lc.Releases {
-			if r.Arch != "mips64" {
-				t.Errorf("arch = %q, want mips64 (mips64r6* → mips64)", r.Arch)
-			}
+		if lc.Releases[0].Arch != "mips64r6" {
+			t.Errorf("arch = %q, want mips64r6 (kept as-is, classifier knows this value)", lc.Releases[0].Arch)
+		}
+		if lc.Releases[1].Arch != "mips64r6el" {
+			t.Errorf("arch = %q, want mips64r6el (kept as-is, classifier knows this value)", lc.Releases[1].Arch)
+		}
+	})
+
+	t.Run("mipsle_to_mipsel", func(t *testing.T) {
+		// mipsle: Go uses mipsle (GOARCH), Node classifier tpm['mipsle'] = {arch:'mipsel'}.
+		pd := storage.PackageData{
+			Assets: []storage.Asset{
+				{Filename: "caddy_linux_mipsle.tar.gz", OS: "linux", Arch: "mipsle", Format: ".tar.gz"},
+			},
+		}
+		lc, _ := storage.ExportLegacy("caddy", pd)
+		if len(lc.Releases) != 1 {
+			t.Fatalf("releases = %d, want 1", len(lc.Releases))
+		}
+		if lc.Releases[0].Arch != "mipsel" {
+			t.Errorf("arch = %q, want mipsel (mipsle → mipsel)", lc.Releases[0].Arch)
+		}
+	})
+
+	t.Run("mips64le_to_mips64el", func(t *testing.T) {
+		// mips64le: Go uses mips64le (GOARCH), Node classifier tpm['mips64le'] = {arch:'mips64el'}.
+		pd := storage.PackageData{
+			Assets: []storage.Asset{
+				{Filename: "gitea-linux-mips64le.tar.gz", OS: "linux", Arch: "mips64le", Format: ".tar.gz"},
+			},
+		}
+		lc, _ := storage.ExportLegacy("gitea", pd)
+		if len(lc.Releases) != 1 {
+			t.Fatalf("releases = %d, want 1", len(lc.Releases))
+		}
+		if lc.Releases[0].Arch != "mips64el" {
+			t.Errorf("arch = %q, want mips64el (mips64le → mips64el)", lc.Releases[0].Arch)
 		}
 	})

@@ -425,6 +460,74 @@ func TestExportLegacyTranslations(t *testing.T) {
 			t.Errorf("arch = %q, want armv6 (no translation for armv6l)", lc.Releases[0].Arch)
 		}
 	})
+
+	t.Run("arm_armv7_gnueabihf_to_armv7", func(t *testing.T) {
+		// Files like "ripgrep-14.1.0-armv7-unknown-linux-gnueabihf.tar.gz":
+		// Go classifies as armv7; the "armv7" term in filename takes priority
+		// over the gnueabihf ABI suffix. Classifier sees "armv7" → extracts armv7.
+		pd := storage.PackageData{
+			Assets: []storage.Asset{
+				{Filename: "ripgrep-14.1.0-armv7-unknown-linux-gnueabihf.tar.gz", OS: "linux", Arch: "armv7", Format: ".tar.gz"},
+			},
+		}
+		lc, _ := storage.ExportLegacy("ripgrep", pd)
+		if len(lc.Releases) != 1 {
+			t.Fatalf("releases = %d, want 1", len(lc.Releases))
+		}
+		if lc.Releases[0].Arch != "armv7" {
+			t.Errorf("arch = %q, want armv7 (armv7 in filename takes priority over gnueabihf)", lc.Releases[0].Arch)
+		}
+	})
+
+	t.Run("arm_armv6hf_to_armhf", func(t *testing.T) {
+		// shellcheck uses "armv6hf" naming; classifier tpm['armv6hf'] = ARMHF → "armhf".
+		pd := storage.PackageData{
+			Assets: []storage.Asset{
+				{Filename: "shellcheck-v0.9.0.linux.armv6hf.tar.xz", OS: "linux", Arch: "armv6", Format: ".tar.xz"},
+			},
+		}
+		lc, _ := storage.ExportLegacy("shellcheck", pd)
+		if len(lc.Releases) != 1 {
+			t.Fatalf("releases = %d, want 1", len(lc.Releases))
+		}
+		if lc.Releases[0].Arch != "armhf" {
+			t.Errorf("arch = %q, want armhf (armv6hf → armhf)", lc.Releases[0].Arch)
+		}
+	})
+
+	t.Run("arm_gitea_arm5_to_armel", func(t *testing.T) {
+		// Gitea uses "arm-5" naming; patternToTerms converts to "armv5" → tpm → "armel".
+		// Go sees \barm\b → classifies as armv6. Legacy export must correct to armel.
+		pd := storage.PackageData{
+			Assets: []storage.Asset{
+				{Filename: "gitea-1.20.0-linux-arm-5", OS: "linux", Arch: "armv6", Format: ""},
+			},
+		}
+		lc, _ := storage.ExportLegacy("gitea", pd)
+		if len(lc.Releases) != 1 {
+			t.Fatalf("releases = %d, want 1", len(lc.Releases))
+		}
+		if lc.Releases[0].Arch != "armel" {
+			t.Errorf("arch = %q, want armel (arm-5 → armel)", lc.Releases[0].Arch)
+		}
+	})
+
+	t.Run("arm_gitea_arm7_to_armv7", func(t *testing.T) {
+		// Gitea uses "arm-7" naming; patternToTerms converts to "armv7" → tpm → "armv7".
+		// Go sees \barm\b → classifies as armv6. Legacy export must correct to armv7.
+		pd := storage.PackageData{
+			Assets: []storage.Asset{
+				{Filename: "gitea-1.20.0-linux-arm-7", OS: "linux", Arch: "armv6", Format: ""},
+			},
+		}
+		lc, _ := storage.ExportLegacy("gitea", pd)
+		if len(lc.Releases) != 1 {
+			t.Fatalf("releases = %d, want 1", len(lc.Releases))
+		}
+		if lc.Releases[0].Arch != "armv7" {
+			t.Errorf("arch = %q, want armv7 (arm-7 → armv7)", lc.Releases[0].Arch)
+		}
+	})
 }

 // TestExportLegacyMixed verifies correct counting when multiple drop categories