config: support insteadOf for remotes' URLs (#79)

4 files changed, 239 insertions, 2 deletions

diff --git a/config/config.go b/config/config.go
index 8a99e8d..1f737b5 100644
--- a/config/config.go
+++ b/config/config.go
@@ -105,6 +105,9 @@ type Config struct {
 	// Branches list of branches, the key is the branch name and should
 	// equal Branch.Name
 	Branches map[string]*Branch
+	// URLs list of url rewrite rules, if repo url starts with URL.InsteadOf value, it will be replaced with the
+	// key instead.
+	URLs map[string]*URL
 	// Raw contains the raw information of a config file. The main goal is
 	// preserve the parsed information from the original format, to avoid
 	// dropping unsupported fields.
@@ -117,6 +120,7 @@ func NewConfig() *Config {
 		Remotes:    make(map[string]*RemoteConfig),
 		Submodules: make(map[string]*Submodule),
 		Branches:   make(map[string]*Branch),
+		URLs:       make(map[string]*URL),
 		Raw:        format.New(),
 	}
 
@@ -231,6 +235,7 @@ const (
 	authorSection    = "author"
 	committerSection = "committer"
 	initSection      = "init"
+	urlSection       = "url"
 	fetchKey         = "fetch"
 	urlKey           = "url"
 	bareKey          = "bare"
@@ -270,6 +275,10 @@ func (c *Config) Unmarshal(b []byte) error {
 		return err
 	}
 
+	if err := c.unmarshalURLs(); err != nil {
+		return err
+	}
+
 	return c.unmarshalRemotes()
 }
 
@@ -323,6 +332,25 @@ func (c *Config) unmarshalRemotes() error {
 		c.Remotes[r.Name] = r
 	}
 
+	// Apply insteadOf url rules
+	for _, r := range c.Remotes {
+		r.applyURLRules(c.URLs)
+	}
+
+	return nil
+}
+
+func (c *Config) unmarshalURLs() error {
+	s := c.Raw.Section(urlSection)
+	for _, sub := range s.Subsections {
+		r := &URL{}
+		if err := r.unmarshal(sub); err != nil {
+			return err
+		}
+
+		c.URLs[r.Name] = r
+	}
+
 	return nil
 }
 
@@ -367,6 +395,7 @@ func (c *Config) Marshal() ([]byte, error) {
 	c.marshalRemotes()
 	c.marshalSubmodules()
 	c.marshalBranches()
+	c.marshalURLs()
 	c.marshalInit()
 
 	buf := bytes.NewBuffer(nil)
@@ -491,6 +520,20 @@ func (c *Config) marshalBranches() {
 	s.Subsections = newSubsections
 }
 
+func (c *Config) marshalURLs() {
+	s := c.Raw.Section(urlSection)
+	s.Subsections = make(format.Subsections, len(c.URLs))
+
+	var i int
+	for _, r := range c.URLs {
+		section := r.marshal()
+		// the submodule section at config is a subset of the .gitmodule file
+		// we should remove the non-valid options for the config file.
+		s.Subsections[i] = section
+		i++
+	}
+}
+
 func (c *Config) marshalInit() {
 	s := c.Raw.Section(initSection)
 	if c.Init.DefaultBranch != "" {
@@ -505,6 +548,12 @@ type RemoteConfig struct {
 	// URLs the URLs of a remote repository. It must be non-empty. Fetch will
 	// always use the first URL, while push will use all of them.
 	URLs []string
+
+	// insteadOfRulesApplied have urls been modified
+	insteadOfRulesApplied bool
+	// originalURLs are the urls before applying insteadOf rules
+	originalURLs []string
+
 	// Fetch the default set of "refspec" for fetch operation
 	Fetch []RefSpec
 
@@ -565,7 +614,12 @@ func (c *RemoteConfig) marshal() *format.Subsection {
 	if len(c.URLs) == 0 {
 		c.raw.RemoveOption(urlKey)
 	} else {
-		c.raw.SetOption(urlKey, c.URLs...)
+		urls := c.URLs
+		if c.insteadOfRulesApplied {
+			urls = c.originalURLs
+		}
+
+		c.raw.SetOption(urlKey, urls...)
 	}
 
 	if len(c.Fetch) == 0 {
@@ -585,3 +639,20 @@ func (c *RemoteConfig) marshal() *format.Subsection {
 func (c *RemoteConfig) IsFirstURLLocal() bool {
 	return url.IsLocalEndpoint(c.URLs[0])
 }
+
+func (c *RemoteConfig) applyURLRules(urlRules map[string]*URL) {
+	// save original urls
+	originalURLs := make([]string, len(c.URLs))
+	copy(originalURLs, c.URLs)
+
+	for i, url := range c.URLs {
+		if matchingURLRule := findLongestInsteadOfMatch(url, urlRules); matchingURLRule != nil {
+			c.URLs[i] = matchingURLRule.ApplyInsteadOf(c.URLs[i])
+			c.insteadOfRulesApplied = true
+		}
+	}
+
+	if c.insteadOfRulesApplied {
+		c.originalURLs = originalURLs
+	}
+}
diff --git a/config/config_test.go b/config/config_test.go
index 8108a94..b4a9ac3 100644
--- a/config/config_test.go
+++ b/config/config_test.go
@@ -38,6 +38,8 @@ func (s *ConfigSuite) TestUnmarshal(c *C) {
 		url = git@github.com:src-d/go-git.git
 		fetch = +refs/heads/*:refs/remotes/origin/*
 		fetch = +refs/pull/*:refs/remotes/origin/pull/*
+[remote "insteadOf"]
+		url = https://github.com/kostyay/go-git.git
 [remote "win-local"]
 		url = X:\\Git\\
 [submodule "qux"]
@@ -49,6 +51,8 @@ func (s *ConfigSuite) TestUnmarshal(c *C) {
 		merge = refs/heads/master
 [init]
 		defaultBranch = main
+[url "ssh://git@github.com/"]
+	insteadOf = https://github.com/
 `)
 
 	cfg := NewConfig()
@@ -65,7 +69,7 @@ func (s *ConfigSuite) TestUnmarshal(c *C) {
 	c.Assert(cfg.Committer.Name, Equals, "Richard Roe")
 	c.Assert(cfg.Committer.Email, Equals, "richard@example.com")
 	c.Assert(cfg.Pack.Window, Equals, uint(20))
-	c.Assert(cfg.Remotes, HasLen, 3)
+	c.Assert(cfg.Remotes, HasLen, 4)
 	c.Assert(cfg.Remotes["origin"].Name, Equals, "origin")
 	c.Assert(cfg.Remotes["origin"].URLs, DeepEquals, []string{"git@github.com:mcuadros/go-git.git"})
 	c.Assert(cfg.Remotes["origin"].Fetch, DeepEquals, []RefSpec{"+refs/heads/*:refs/remotes/origin/*"})
@@ -74,6 +78,7 @@ func (s *ConfigSuite) TestUnmarshal(c *C) {
 	c.Assert(cfg.Remotes["alt"].Fetch, DeepEquals, []RefSpec{"+refs/heads/*:refs/remotes/origin/*", "+refs/pull/*:refs/remotes/origin/pull/*"})
 	c.Assert(cfg.Remotes["win-local"].Name, Equals, "win-local")
 	c.Assert(cfg.Remotes["win-local"].URLs, DeepEquals, []string{"X:\\Git\\"})
+	c.Assert(cfg.Remotes["insteadOf"].URLs, DeepEquals, []string{"ssh://git@github.com/kostyay/go-git.git"})
 	c.Assert(cfg.Submodules, HasLen, 1)
 	c.Assert(cfg.Submodules["qux"].Name, Equals, "qux")
 	c.Assert(cfg.Submodules["qux"].URL, Equals, "https://github.com/foo/qux.git")
@@ -94,6 +99,8 @@ func (s *ConfigSuite) TestMarshal(c *C) {
 	url = git@github.com:src-d/go-git.git
 	fetch = +refs/heads/*:refs/remotes/origin/*
 	fetch = +refs/pull/*:refs/remotes/origin/pull/*
+[remote "insteadOf"]
+	url = https://github.com/kostyay/go-git.git
 [remote "origin"]
 	url = git@github.com:mcuadros/go-git.git
 [remote "win-local"]
@@ -103,6 +110,8 @@ func (s *ConfigSuite) TestMarshal(c *C) {
 [branch "master"]
 	remote = origin
 	merge = refs/heads/master
+[url "ssh://git@github.com/"]
+	insteadOf = https://github.com/
 [init]
 	defaultBranch = main
 `)
@@ -128,6 +137,11 @@ func (s *ConfigSuite) TestMarshal(c *C) {
 		URLs: []string{"X:\\Git\\"},
 	}
 
+	cfg.Remotes["insteadOf"] = &RemoteConfig{
+		Name: "insteadOf",
+		URLs: []string{"https://github.com/kostyay/go-git.git"},
+	}
+
 	cfg.Submodules["qux"] = &Submodule{
 		Name: "qux",
 		URL:  "https://github.com/foo/qux.git",
@@ -139,6 +153,11 @@ func (s *ConfigSuite) TestMarshal(c *C) {
 		Merge:  "refs/heads/master",
 	}
 
+	cfg.URLs["ssh://git@github.com/"] = &URL{
+		Name: "ssh://git@github.com/",
+		InsteadOf: "https://github.com/",
+	}
+
 	b, err := cfg.Marshal()
 	c.Assert(err, IsNil)
 
@@ -161,6 +180,8 @@ func (s *ConfigSuite) TestUnmarshalMarshal(c *C) {
 	email = richard@example.co
 [pack]
 	window = 20
+[remote "insteadOf"]
+	url = https://github.com/kostyay/go-git.git
 [remote "origin"]
 	url = git@github.com:mcuadros/go-git.git
 	fetch = +refs/heads/*:refs/remotes/origin/*
@@ -170,6 +191,8 @@ func (s *ConfigSuite) TestUnmarshalMarshal(c *C) {
 [branch "master"]
 	remote = origin
 	merge = refs/heads/master
+[url "ssh://git@github.com/"]
+	insteadOf = https://github.com/
 `)
 
 	cfg := NewConfig()
diff --git a/config/url.go b/config/url.go
new file mode 100644
index 0000000..114d6b2
--- /dev/null
+++ b/config/url.go
@@ -0,0 +1,81 @@
+package config
+
+import (
+	"errors"
+	"strings"
+
+	format "github.com/go-git/go-git/v5/plumbing/format/config"
+)
+
+var (
+	errURLEmptyInsteadOf = errors.New("url config: empty insteadOf")
+)
+
+// Url defines Url rewrite rules
+type URL struct {
+	// Name new base url
+	Name string
+	// Any URL that starts with this value will be rewritten to start, instead, with <base>.
+	// When more than one insteadOf strings match a given URL, the longest match is used.
+	InsteadOf string
+
+	// raw representation of the subsection, filled by marshal or unmarshal are
+	// called.
+	raw *format.Subsection
+}
+
+// Validate validates fields of branch
+func (b *URL) Validate() error {
+	if b.InsteadOf == "" {
+		return errURLEmptyInsteadOf
+	}
+
+	return nil
+}
+
+const (
+	insteadOfKey = "insteadOf"
+)
+
+func (u *URL) unmarshal(s *format.Subsection) error {
+	u.raw = s
+
+	u.Name = s.Name
+	u.InsteadOf = u.raw.Option(insteadOfKey)
+	return nil
+}
+
+func (u *URL) marshal() *format.Subsection {
+	if u.raw == nil {
+		u.raw = &format.Subsection{}
+	}
+
+	u.raw.Name = u.Name
+	u.raw.SetOption(insteadOfKey, u.InsteadOf)
+
+	return u.raw
+}
+
+func findLongestInsteadOfMatch(remoteURL string, urls map[string]*URL) *URL {
+	var longestMatch *URL
+	for _, u := range urls {
+		if !strings.HasPrefix(remoteURL, u.InsteadOf) {
+			continue
+		}
+
+		// according to spec if there is more than one match, take the logest
+		if longestMatch == nil || len(longestMatch.InsteadOf) < len(u.InsteadOf) {
+			longestMatch = u
+		}
+	}
+
+	return longestMatch
+}
+
+func (u *URL) ApplyInsteadOf(url string) string {
+	if !strings.HasPrefix(url, u.InsteadOf) {
+		return url
+	}
+
+	return u.Name + url[len(u.InsteadOf):]
+}
diff --git a/config/url_test.go b/config/url_test.go
new file mode 100644
index 0000000..5afc9f3
--- /dev/null
+++ b/config/url_test.go
@@ -0,0 +1,62 @@
+package config
+
+import (
+	. "gopkg.in/check.v1"
+)
+
+type URLSuite struct{}
+
+var _ = Suite(&URLSuite{})
+
+func (b *URLSuite) TestValidateInsteadOf(c *C) {
+	goodURL := URL{
+		Name:      "ssh://github.com",
+		InsteadOf: "http://github.com",
+	}
+	badURL := URL{}
+	c.Assert(goodURL.Validate(), IsNil)
+	c.Assert(badURL.Validate(), NotNil)
+}
+
+func (b *URLSuite) TestMarshal(c *C) {
+	expected := []byte(`[core]
+	bare = false
+[url "ssh://git@github.com/"]
+	insteadOf = https://github.com/
+`)
+
+	cfg := NewConfig()
+	cfg.URLs["ssh://git@github.com/"] = &URL{
+		Name:      "ssh://git@github.com/",
+		InsteadOf: "https://github.com/",
+	}
+
+	actual, err := cfg.Marshal()
+	c.Assert(err, IsNil)
+	c.Assert(string(actual), Equals, string(expected))
+}
+
+func (b *URLSuite) TestUnmarshal(c *C) {
+	input := []byte(`[core]
+	bare = false
+[url "ssh://git@github.com/"]
+	insteadOf = https://github.com/
+`)
+
+	cfg := NewConfig()
+	err := cfg.Unmarshal(input)
+	c.Assert(err, IsNil)
+	url := cfg.URLs["ssh://git@github.com/"]
+	c.Assert(url.Name, Equals, "ssh://git@github.com/")
+	c.Assert(url.InsteadOf, Equals, "https://github.com/")
+}
+
+func (b *URLSuite) TestApplyInsteadOf(c *C) {
+	urlRule := URL{
+		Name:      "ssh://github.com",
+		InsteadOf: "http://github.com",
+	}
+
+	c.Assert(urlRule.ApplyInsteadOf("http://google.com"), Equals, "http://google.com")
+	c.Assert(urlRule.ApplyInsteadOf("http://github.com/myrepo"), Equals, "ssh://github.com/myrepo")
+}