v2: Implement RegExp Vars Matcher (#2997)

* implement regexp var matcher

* use subtests pattern for tests

* be more consistent with naming: MatchVarRE -> MatchVarsRE, var_regexp -> vars_regexp

2 files changed, 155 insertions, 0 deletions

diff --git a/modules/caddyhttp/matchers_test.go b/modules/caddyhttp/matchers_test.go
index 06f137e..58df171 100644
--- a/modules/caddyhttp/matchers_test.go
+++ b/modules/caddyhttp/matchers_test.go
@@ -545,6 +545,92 @@ func TestHeaderREMatcher(t *testing.T) {
 	}
 }
 
+func TestVarREMatcher(t *testing.T) {
+	for i, tc := range []struct {
+		desc       string
+		match      MatchVarsRE
+		input      VarsMiddleware
+		expect     bool
+		expectRepl map[string]string
+	}{
+		{
+			desc:   "match static value within var set by the VarsMiddleware succeeds",
+			match:  MatchVarsRE{"Var1": &MatchRegexp{Pattern: "foo"}},
+			input:  VarsMiddleware{"Var1": "here is foo val"},
+			expect: true,
+		},
+		{
+			desc:   "value set by VarsMiddleware not satisfying regexp matcher fails to match",
+			match:  MatchVarsRE{"Var1": &MatchRegexp{Pattern: "$foo^"}},
+			input:  VarsMiddleware{"Var1": "foobar"},
+			expect: false,
+		},
+		{
+			desc:       "successfully matched value is captured and its placeholder is added to replacer",
+			match:      MatchVarsRE{"Var1": &MatchRegexp{Pattern: "^foo(.*)$", Name: "name"}},
+			input:      VarsMiddleware{"Var1": "foobar"},
+			expect:     true,
+			expectRepl: map[string]string{"name.1": "bar"},
+		},
+		{
+			desc:   "matching against a value of standard variables succeeds",
+			match:  MatchVarsRE{"{http.request.method}": &MatchRegexp{Pattern: "^G.[tT]$"}},
+			input:  VarsMiddleware{},
+			expect: true,
+		},
+		{
+			desc:   "matching agaist value of var set by the VarsMiddleware and referenced by its placeholder succeeds",
+			match:  MatchVarsRE{"{http.vars.Var1}": &MatchRegexp{Pattern: "[vV]ar[0-9]"}},
+			input:  VarsMiddleware{"Var1": "var1Value"},
+			expect: true,
+		},
+	} {
+		tc := tc // capture range value
+		t.Run(tc.desc, func(t *testing.T) {
+			t.Parallel()
+			// compile the regexp and validate its name
+			err := tc.match.Provision(caddy.Context{})
+			if err != nil {
+				t.Errorf("Test %d %v: Provisioning: %v", i, tc.match, err)
+				return
+			}
+			err = tc.match.Validate()
+			if err != nil {
+				t.Errorf("Test %d %v: Validating: %v", i, tc.match, err)
+				return
+			}
+
+			// set up the fake request and its Replacer
+			req := &http.Request{URL: new(url.URL), Method: http.MethodGet}
+			repl := caddy.NewReplacer()
+			ctx := context.WithValue(req.Context(), caddy.ReplacerCtxKey, repl)
+			ctx = context.WithValue(ctx, VarsCtxKey, make(map[string]interface{}))
+			req = req.WithContext(ctx)
+
+			addHTTPVarsToReplacer(repl, req, httptest.NewRecorder())
+
+			tc.input.ServeHTTP(httptest.NewRecorder(), req, emptyHandler)
+
+			actual := tc.match.Match(req)
+			if actual != tc.expect {
+				t.Errorf("Test %d [%v]: Expected %t, got %t for input '%s'",
+					i, tc.match, tc.expect, actual, tc.input)
+				return
+			}
+
+			for key, expectVal := range tc.expectRepl {
+				placeholder := fmt.Sprintf("{http.regexp.%s}", key)
+				actualVal := repl.ReplaceAll(placeholder, "<empty>")
+				if actualVal != expectVal {
+					t.Errorf("Test %d [%v]: Expected placeholder {http.regexp.%s} to be '%s' but got '%s'",
+						i, tc.match, key, expectVal, actualVal)
+					return
+				}
+			}
+		})
+	}
+}
+
 func TestResponseMatcher(t *testing.T) {
 	for i, tc := range []struct {
 		require ResponseMatcher
diff --git a/modules/caddyhttp/vars.go b/modules/caddyhttp/vars.go
index 1208d9c..9561d46 100644
--- a/modules/caddyhttp/vars.go
+++ b/modules/caddyhttp/vars.go
@@ -25,6 +25,7 @@ import (
 func init() {
 	caddy.RegisterModule(VarsMiddleware{})
 	caddy.RegisterModule(VarsMatcher{})
+	caddy.RegisterModule(MatchVarsRE{})
 }
 
 // VarsMiddleware is an HTTP middleware which sets variables
@@ -88,6 +89,74 @@ func (m VarsMatcher) Match(r *http.Request) bool {
 	return true
 }
 
+// MatchVarsRE matches the value of the context variables by a given regular expression.
+//
+// Upon a match, it adds placeholders to the request: `{http.regexp.name.capture_group}`
+// where `name` is the regular expression's name, and `capture_group` is either
+// the named or positional capture group from the expression itself. If no name
+// is given, then the placeholder omits the name: `{http.regexp.capture_group}`
+// (potentially leading to collisions).
+type MatchVarsRE map[string]*MatchRegexp
+
+// CaddyModule returns the Caddy module information.
+func (MatchVarsRE) CaddyModule() caddy.ModuleInfo {
+	return caddy.ModuleInfo{
+		ID:  "http.matchers.vars_regexp",
+		New: func() caddy.Module { return new(MatchVarsRE) },
+	}
+}
+
+// Provision compiles m's regular expressions.
+func (m MatchVarsRE) Provision(ctx caddy.Context) error {
+	for _, rm := range m {
+		err := rm.Provision(ctx)
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// Match returns true if r matches m.
+func (m MatchVarsRE) Match(r *http.Request) bool {
+	vars := r.Context().Value(VarsCtxKey).(map[string]interface{})
+	repl := r.Context().Value(caddy.ReplacerCtxKey).(*caddy.Replacer)
+	for k, rm := range m {
+		var varStr string
+		switch vv := vars[k].(type) {
+		case string:
+			varStr = vv
+		case fmt.Stringer:
+			varStr = vv.String()
+		case error:
+			varStr = vv.Error()
+		default:
+			varStr = fmt.Sprintf("%v", vv)
+		}
+		valExpanded := repl.ReplaceAll(varStr, "")
+		if match := rm.Match(valExpanded, repl); match {
+			return match
+		}
+
+		replacedVal := repl.ReplaceAll(k, "")
+		if match := rm.Match(replacedVal, repl); match {
+			return match
+		}
+	}
+	return false
+}
+
+// Validate validates m's regular expressions.
+func (m MatchVarsRE) Validate() error {
+	for _, rm := range m {
+		err := rm.Validate()
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
 // GetVar gets a value out of the context's variable table by key.
 // If the key does not exist, the return value will be nil.
 func GetVar(ctx context.Context, key string) interface{} {