cmd: relocate search.MatchPattern to cmd/internal/pkgpattern

Relocate cmd/go's search.MatchPattern helper routine to a new package
in cmd/internal from its current location, as to allow it to be used
in other tools that accept package pattern command line flags. No
change in functionality along the way.

Updates #51430.

Change-Id: I726e974ccd66a055bb5a94497b36b8d68d47cad1
Reviewed-on: https://go-review.googlesource.com/c/go/+/432757
Reviewed-by: Bryan Mills <bcmills@google.com>
This commit is contained in:
Than McIntosh 2022-09-21 15:54:33 -04:00
parent fb6c210dc3
commit 1e4989c336
9 changed files with 209 additions and 132 deletions

View file

@ -41,6 +41,7 @@ import (
"cmd/go/internal/str"
"cmd/go/internal/trace"
"cmd/go/internal/vcs"
"cmd/internal/pkgpattern"
"cmd/internal/sys"
"golang.org/x/mod/modfile"
@ -3201,7 +3202,7 @@ func PackagesAndErrorsOutsideModule(ctx context.Context, opts PackageOpts, args
matchers := make([]func(string) bool, len(patterns))
for i, p := range patterns {
if strings.Contains(p, "...") {
matchers[i] = search.MatchPattern(p)
matchers[i] = pkgpattern.MatchPattern(p)
}
}
return pkgs, nil

View file

@ -9,6 +9,7 @@ import (
"strings"
"cmd/go/internal/search"
"cmd/internal/pkgpattern"
)
// MatchPackage(pattern, cwd)(p) reports whether package p matches pattern in the working directory cwd.
@ -29,7 +30,7 @@ func MatchPackage(pattern, cwd string) func(*Package) bool {
if pattern == "" {
return func(p *Package) bool { return p.Dir == dir }
}
matchPath := search.MatchPattern(pattern)
matchPath := pkgpattern.MatchPattern(pattern)
return func(p *Package) bool {
// Compute relative path to dir and see if it matches the pattern.
rel, err := filepath.Rel(dir, p.Dir)
@ -50,7 +51,7 @@ func MatchPackage(pattern, cwd string) func(*Package) bool {
case pattern == "cmd":
return func(p *Package) bool { return p.Standard && strings.HasPrefix(p.ImportPath, "cmd/") }
default:
matchPath := search.MatchPattern(pattern)
matchPath := pkgpattern.MatchPattern(pattern)
return func(p *Package) bool { return matchPath(p.ImportPath) }
}
}

View file

@ -15,6 +15,7 @@ import (
"cmd/go/internal/modload"
"cmd/go/internal/search"
"cmd/go/internal/str"
"cmd/internal/pkgpattern"
"golang.org/x/mod/module"
)
@ -165,8 +166,8 @@ func newQuery(raw string) (*query, error) {
version: version,
}
if strings.Contains(q.pattern, "...") {
q.matchWildcard = search.MatchPattern(q.pattern)
q.canMatchWildcardInModule = search.TreeCanMatchPattern(q.pattern)
q.matchWildcard = pkgpattern.MatchPattern(q.pattern)
q.canMatchWildcardInModule = pkgpattern.TreeCanMatchPattern(q.pattern)
}
if err := q.validate(); err != nil {
return q, err

View file

@ -20,6 +20,7 @@ import (
"cmd/go/internal/modfetch/codehost"
"cmd/go/internal/modinfo"
"cmd/go/internal/search"
"cmd/internal/pkgpattern"
"golang.org/x/mod/module"
)
@ -225,7 +226,7 @@ func listModules(ctx context.Context, rs *Requirements, args []string, mode List
if arg == "all" {
match = func(string) bool { return true }
} else if strings.Contains(arg, "...") {
match = search.MatchPattern(arg)
match = pkgpattern.MatchPattern(arg)
} else {
var v string
if mg == nil {

View file

@ -25,6 +25,7 @@ import (
"cmd/go/internal/search"
"cmd/go/internal/str"
"cmd/go/internal/trace"
"cmd/internal/pkgpattern"
"golang.org/x/mod/module"
"golang.org/x/mod/semver"
@ -624,7 +625,7 @@ func QueryPattern(ctx context.Context, pattern, query string, current func(strin
}
var match func(mod module.Version, roots []string, isLocal bool) *search.Match
matchPattern := search.MatchPattern(pattern)
matchPattern := pkgpattern.MatchPattern(pattern)
if i := strings.Index(pattern, "..."); i >= 0 {
base = pathpkg.Dir(pattern[:i+3])

View file

@ -24,6 +24,7 @@ import (
"cmd/go/internal/par"
"cmd/go/internal/search"
"cmd/go/internal/trace"
"cmd/internal/pkgpattern"
"golang.org/x/mod/module"
)
@ -47,8 +48,8 @@ func matchPackages(ctx context.Context, m *search.Match, tags map[string]bool, f
isMatch := func(string) bool { return true }
treeCanMatch := func(string) bool { return true }
if !m.IsMeta() {
isMatch = search.MatchPattern(m.Pattern())
treeCanMatch = search.TreeCanMatchPattern(m.Pattern())
isMatch = pkgpattern.MatchPattern(m.Pattern())
treeCanMatch = pkgpattern.TreeCanMatchPattern(m.Pattern())
}
var mu sync.Mutex

View file

@ -8,13 +8,13 @@ import (
"cmd/go/internal/base"
"cmd/go/internal/cfg"
"cmd/go/internal/fsys"
"cmd/internal/pkgpattern"
"fmt"
"go/build"
"io/fs"
"os"
"path"
"path/filepath"
"regexp"
"strings"
)
@ -109,8 +109,8 @@ func (m *Match) MatchPackages() {
match := func(string) bool { return true }
treeCanMatch := func(string) bool { return true }
if !m.IsMeta() {
match = MatchPattern(m.pattern)
treeCanMatch = TreeCanMatchPattern(m.pattern)
match = pkgpattern.MatchPattern(m.pattern)
treeCanMatch = pkgpattern.TreeCanMatchPattern(m.pattern)
}
have := map[string]bool{
@ -233,7 +233,7 @@ func (m *Match) MatchDirs(modRoots []string) {
cleanPattern = "." + string(os.PathSeparator) + cleanPattern
}
slashPattern := filepath.ToSlash(cleanPattern)
match := MatchPattern(slashPattern)
match := pkgpattern.MatchPattern(slashPattern)
// Find directory to begin the scan.
// Could be smarter but this one optimization
@ -332,90 +332,6 @@ func (m *Match) MatchDirs(modRoots []string) {
}
}
// TreeCanMatchPattern(pattern)(name) reports whether
// name or children of name can possibly match pattern.
// Pattern is the same limited glob accepted by matchPattern.
func TreeCanMatchPattern(pattern string) func(name string) bool {
wildCard := false
if i := strings.Index(pattern, "..."); i >= 0 {
wildCard = true
pattern = pattern[:i]
}
return func(name string) bool {
return len(name) <= len(pattern) && hasPathPrefix(pattern, name) ||
wildCard && strings.HasPrefix(name, pattern)
}
}
// MatchPattern(pattern)(name) reports whether
// name matches pattern. Pattern is a limited glob
// pattern in which '...' means 'any string' and there
// is no other special syntax.
// Unfortunately, there are two special cases. Quoting "go help packages":
//
// First, /... at the end of the pattern can match an empty string,
// so that net/... matches both net and packages in its subdirectories, like net/http.
// Second, any slash-separated pattern element containing a wildcard never
// participates in a match of the "vendor" element in the path of a vendored
// package, so that ./... does not match packages in subdirectories of
// ./vendor or ./mycode/vendor, but ./vendor/... and ./mycode/vendor/... do.
// Note, however, that a directory named vendor that itself contains code
// is not a vendored package: cmd/vendor would be a command named vendor,
// and the pattern cmd/... matches it.
func MatchPattern(pattern string) func(name string) bool {
// Convert pattern to regular expression.
// The strategy for the trailing /... is to nest it in an explicit ? expression.
// The strategy for the vendor exclusion is to change the unmatchable
// vendor strings to a disallowed code point (vendorChar) and to use
// "(anything but that codepoint)*" as the implementation of the ... wildcard.
// This is a bit complicated but the obvious alternative,
// namely a hand-written search like in most shell glob matchers,
// is too easy to make accidentally exponential.
// Using package regexp guarantees linear-time matching.
const vendorChar = "\x00"
if strings.Contains(pattern, vendorChar) {
return func(name string) bool { return false }
}
re := regexp.QuoteMeta(pattern)
re = replaceVendor(re, vendorChar)
switch {
case strings.HasSuffix(re, `/`+vendorChar+`/\.\.\.`):
re = strings.TrimSuffix(re, `/`+vendorChar+`/\.\.\.`) + `(/vendor|/` + vendorChar + `/\.\.\.)`
case re == vendorChar+`/\.\.\.`:
re = `(/vendor|/` + vendorChar + `/\.\.\.)`
case strings.HasSuffix(re, `/\.\.\.`):
re = strings.TrimSuffix(re, `/\.\.\.`) + `(/\.\.\.)?`
}
re = strings.ReplaceAll(re, `\.\.\.`, `[^`+vendorChar+`]*`)
reg := regexp.MustCompile(`^` + re + `$`)
return func(name string) bool {
if strings.Contains(name, vendorChar) {
return false
}
return reg.MatchString(replaceVendor(name, vendorChar))
}
}
// replaceVendor returns the result of replacing
// non-trailing vendor path elements in x with repl.
func replaceVendor(x, repl string) string {
if !strings.Contains(x, "vendor") {
return x
}
elem := strings.Split(x, "/")
for i := 0; i < len(elem)-1; i++ {
if elem[i] == "vendor" {
elem[i] = repl
}
}
return strings.Join(elem, "/")
}
// WarnUnmatched warns about patterns that didn't match any packages.
func WarnUnmatched(matches []*Match) {
for _, m := range matches {
@ -512,22 +428,6 @@ func CleanPatterns(patterns []string) []string {
return out
}
// hasPathPrefix reports whether the path s begins with the
// elements in prefix.
func hasPathPrefix(s, prefix string) bool {
switch {
default:
return false
case len(s) == len(prefix):
return s == prefix
case len(s) > len(prefix):
if prefix != "" && prefix[len(prefix)-1] == '/' {
return strings.HasPrefix(s, prefix)
}
return s[len(prefix)] == '/' && s[:len(prefix)] == prefix
}
}
// hasFilepathPrefix reports whether the path s begins with the
// elements in prefix.
func hasFilepathPrefix(s, prefix string) bool {

View file

@ -1,8 +1,8 @@
// Copyright 2012 The Go Authors. All rights reserved.
// Copyright 2022 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package search
package pkgpattern
import (
"strings"
@ -12,19 +12,19 @@ import (
var matchPatternTests = `
pattern ...
match foo
pattern net
match net
not net/http
pattern net/http
match net/http
not net
pattern net...
match net net/http netchan
not not/http not/net/http
# Special cases. Quoting docs:
# First, /... at the end of the pattern can match an empty string,
@ -43,23 +43,23 @@ var matchPatternTests = `
pattern ./...
match ./vendor ./mycode/vendor
not ./vendor/foo ./mycode/vendor/foo
pattern ./vendor/...
match ./vendor/foo ./vendor/foo/vendor
not ./vendor/foo/vendor/bar
pattern mycode/vendor/...
match mycode/vendor mycode/vendor/foo mycode/vendor/foo/vendor
not mycode/vendor/foo/vendor/bar
pattern x/vendor/y
match x/vendor/y
not x/vendor
pattern x/vendor/y/...
match x/vendor/y x/vendor/y/z x/vendor/y/vendor x/vendor/y/z/vendor
not x/vendor/y/vendor/z
pattern .../vendor/...
match x/vendor/y x/vendor/y/z x/vendor/y/vendor x/vendor/y/z/vendor
`
@ -70,17 +70,51 @@ func TestMatchPattern(t *testing.T) {
})
}
var treeCanMatchPatternTests = `
var matchSimplePatternTests = `
pattern ...
match foo
pattern .../bar/.../baz
match foo/bar/abc/baz
pattern net
match net
not net/http
pattern net/http
match net/http
not net
pattern net...
match net net/http netchan
not not/http not/net/http
# Special cases. Quoting docs:
# First, /... at the end of the pattern can match an empty string,
# so that net/... matches both net and packages in its subdirectories, like net/http.
pattern net/...
match net net/http
not not/http not/net/http netchan
`
func TestSimpleMatchPattern(t *testing.T) {
testPatterns(t, "MatchSimplePattern", matchSimplePatternTests, func(pattern, name string) bool {
return MatchSimplePattern(pattern)(name)
})
}
var treeCanMatchPatternTests = `
pattern ...
match foo
pattern net
match net
not net/http
pattern net/http
match net net/http
pattern net...
match net netchan net/http
not not/http not/net/http
@ -88,18 +122,18 @@ var treeCanMatchPatternTests = `
pattern net/...
match net net/http
not not/http netchan
pattern abc.../def
match abcxyz
not xyzabc
pattern x/y/z/...
match x x/y x/y/z x/y/z/w
pattern x/y/z
match x x/y x/y/z
not x/y/z/w
pattern x/.../y/z
match x/a/b/c
not y/x/a/b/c

View file

@ -0,0 +1,137 @@
// Copyright 2022 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package pkgpattern
import (
"regexp"
"strings"
)
// Note: most of this code was originally part of the cmd/go/internal/search
// package; it was migrated here in order to support the use case of
// commands other than cmd/go that need to accept package pattern args.
// TreeCanMatchPattern(pattern)(name) reports whether
// name or children of name can possibly match pattern.
// Pattern is the same limited glob accepted by MatchPattern.
func TreeCanMatchPattern(pattern string) func(name string) bool {
wildCard := false
if i := strings.Index(pattern, "..."); i >= 0 {
wildCard = true
pattern = pattern[:i]
}
return func(name string) bool {
return len(name) <= len(pattern) && hasPathPrefix(pattern, name) ||
wildCard && strings.HasPrefix(name, pattern)
}
}
// MatchPattern(pattern)(name) reports whether
// name matches pattern. Pattern is a limited glob
// pattern in which '...' means 'any string' and there
// is no other special syntax.
// Unfortunately, there are two special cases. Quoting "go help packages":
//
// First, /... at the end of the pattern can match an empty string,
// so that net/... matches both net and packages in its subdirectories, like net/http.
// Second, any slash-separated pattern element containing a wildcard never
// participates in a match of the "vendor" element in the path of a vendored
// package, so that ./... does not match packages in subdirectories of
// ./vendor or ./mycode/vendor, but ./vendor/... and ./mycode/vendor/... do.
// Note, however, that a directory named vendor that itself contains code
// is not a vendored package: cmd/vendor would be a command named vendor,
// and the pattern cmd/... matches it.
func MatchPattern(pattern string) func(name string) bool {
return matchPatternInternal(pattern, true)
}
// MatchSimplePattern returns a function that can be used to check
// whether a given name matches a pattern, where pattern is a limited
// glob pattern in which '...' means 'any string', with no other
// special syntax. There is one special case for MatchPatternSimple:
// according to the rules in "go help packages": a /... at the end of
// the pattern can match an empty string, so that net/... matches both
// net and packages in its subdirectories, like net/http.
func MatchSimplePattern(pattern string) func(name string) bool {
return matchPatternInternal(pattern, false)
}
func matchPatternInternal(pattern string, vendorExclude bool) func(name string) bool {
// Convert pattern to regular expression.
// The strategy for the trailing /... is to nest it in an explicit ? expression.
// The strategy for the vendor exclusion is to change the unmatchable
// vendor strings to a disallowed code point (vendorChar) and to use
// "(anything but that codepoint)*" as the implementation of the ... wildcard.
// This is a bit complicated but the obvious alternative,
// namely a hand-written search like in most shell glob matchers,
// is too easy to make accidentally exponential.
// Using package regexp guarantees linear-time matching.
const vendorChar = "\x00"
if vendorExclude && strings.Contains(pattern, vendorChar) {
return func(name string) bool { return false }
}
re := regexp.QuoteMeta(pattern)
wild := `.*`
if vendorExclude {
wild = `[^` + vendorChar + `]*`
re = replaceVendor(re, vendorChar)
switch {
case strings.HasSuffix(re, `/`+vendorChar+`/\.\.\.`):
re = strings.TrimSuffix(re, `/`+vendorChar+`/\.\.\.`) + `(/vendor|/` + vendorChar + `/\.\.\.)`
case re == vendorChar+`/\.\.\.`:
re = `(/vendor|/` + vendorChar + `/\.\.\.)`
}
}
if strings.HasSuffix(re, `/\.\.\.`) {
re = strings.TrimSuffix(re, `/\.\.\.`) + `(/\.\.\.)?`
}
re = strings.ReplaceAll(re, `\.\.\.`, wild)
reg := regexp.MustCompile(`^` + re + `$`)
return func(name string) bool {
if vendorExclude {
if strings.Contains(name, vendorChar) {
return false
}
name = replaceVendor(name, vendorChar)
}
return reg.MatchString(name)
}
}
// hasPathPrefix reports whether the path s begins with the
// elements in prefix.
func hasPathPrefix(s, prefix string) bool {
switch {
default:
return false
case len(s) == len(prefix):
return s == prefix
case len(s) > len(prefix):
if prefix != "" && prefix[len(prefix)-1] == '/' {
return strings.HasPrefix(s, prefix)
}
return s[len(prefix)] == '/' && s[:len(prefix)] == prefix
}
}
// replaceVendor returns the result of replacing
// non-trailing vendor path elements in x with repl.
func replaceVendor(x, repl string) string {
if !strings.Contains(x, "vendor") {
return x
}
elem := strings.Split(x, "/")
for i := 0; i < len(elem)-1; i++ {
if elem[i] == "vendor" {
elem[i] = repl
}
}
return strings.Join(elem, "/")
}