encoding/jsonschema: initial support for format keyword

This CL just adds the most basic support for the format keyword, which is just to ignore it. It does add TODO entries for all the known formats with their respective versions, so it will be easy to add support for them in future CLs. Signed-off-by: Roger Peppe <[email protected]> Change-Id: I524547997f4e612a0ed501743192e916fd399900 Dispatch-Trailer: {"type":"trybot","CL":1201125,"patchset":2,"ref":"refs/changes/25/1201125/2","targetBranch":"master"}
cue-lang · Sep 12, 2024 · c0ce4d0 · c0ce4d0
1 parent 6002948
commit c0ce4d0
Show file tree

Hide file tree

Showing 85 changed files with 2,959 additions and 9,352 deletions.
diff --git a/encoding/jsonschema/constraints.go b/encoding/jsonschema/constraints.go
@@ -97,7 +97,7 @@ var constraints = []*constraint{
 	p2("exclusiveMaximum", constraintExclusiveMaximum, allVersions|openAPI),
 	p2("exclusiveMinimum", constraintExclusiveMinimum, allVersions|openAPI),
 	p1("externalDocs", constraintTODO, openAPI),
-	p1("format", constraintTODO, allVersions|openAPI),
+	p1("format", constraintFormat, allVersions|openAPI),
 	p1("id", constraintID, vto(VersionDraft4)),
 	p1("if", constraintIf, vfrom(VersionDraft7)),
 	p2("items", constraintItems, allVersions|openAPI),

diff --git a/encoding/jsonschema/constraints_string.go b/encoding/jsonschema/constraints_string.go
@@ -16,6 +16,7 @@ package jsonschema
 
 import (
 	"regexp"
+	"sync"
 
 	"cuelang.org/go/cue"
 	"cuelang.org/go/cue/ast"
@@ -61,3 +62,69 @@ func constraintPattern(key string, n cue.Value, s *state) {
 	}
 	s.add(n, stringType, &ast.UnaryExpr{Op: token.MAT, X: s.string(n)})
 }
+
+type formatFuncInfo struct {
+	versions versionSet
+	f        func(s *state)
+}
+
+var formatFuncs = sync.OnceValue(func() map[string]formatFuncInfo {
+	return map[string]formatFuncInfo{
+		"binary":                {openAPI, formatTODO},
+		"byte":                  {openAPI, formatTODO},
+		"data":                  {openAPI, formatTODO},
+		"date":                  {vfrom(VersionDraft7), formatTODO},
+		"date-time ":            {allVersions | openAPI, formatTODO},
+		"double":                {openAPI, formatTODO},
+		"duration":              {vfrom(VersionDraft2019_09), formatTODO},
+		"email":                 {allVersions | openAPI, formatTODO},
+		"float":                 {openAPI, formatTODO},
+		"hostname":              {allVersions | openAPI, formatTODO},
+		"idn-email":             {vfrom(VersionDraft7), formatTODO},
+		"idn-hostname":          {vfrom(VersionDraft7), formatTODO},
+		"int32":                 {openAPI, formatTODO},
+		"int64":                 {openAPI, formatTODO},
+		"ipv4":                  {allVersions | openAPI, formatTODO},
+		"ipv6":                  {allVersions | openAPI, formatTODO},
+		"iri":                   {vfrom(VersionDraft7), formatTODO},
+		"iri-reference":         {vfrom(VersionDraft7), formatTODO},
+		"json-pointer":          {vfrom(VersionDraft6), formatTODO},
+		"password":              {openAPI, formatTODO},
+		"regex":                 {vfrom(VersionDraft7), formatTODO},
+		"relative-json-pointer": {vfrom(VersionDraft7), formatTODO},
+		"time":                  {vfrom(VersionDraft7), formatTODO},
+		"uri":                   {allVersions | openAPI, formatTODO},
+		"uri-reference":         {vfrom(VersionDraft6), formatTODO},
+		"uri-template":          {vfrom(VersionDraft6), formatTODO},
+		"uuid":                  {vfrom(VersionDraft2019_09), formatTODO},
+	}
+})
+
+// 2019 changed "hostname" "idn-hostname"
+
+func constraintFormat(key string, n cue.Value, s *state) {
+	formatStr, ok := s.strValue(n)
+	if !ok {
+		return
+	}
+	finfo, ok := formatFuncs()[formatStr]
+	if !ok {
+		// TODO StrictKeywords isn't exactly right here, but in general
+		// we want unknown formats to be ignored even when StrictFeatures
+		// is enabled, and StrictKeywords is closest to what we want.
+		// Perhaps we should have a "lint" mode?
+		if s.cfg.StrictKeywords {
+			s.errf(n, "unknown format %q", formatStr)
+		}
+		return
+	}
+	if !finfo.versions.contains(s.schemaVersion) {
+		if s.cfg.StrictKeywords {
+			s.errf(n, "format %q is not recognized in schema version %v", formatStr, s.schemaVersion)
+		}
+		return
+	}
+	finfo.f(s)
+}
+
+func formatTODO(s *state) {}
diff --git a/encoding/jsonschema/external_teststats.txt b/encoding/jsonschema/external_teststats.txt
@@ -1,10 +1,10 @@
 # Generated by teststats. DO NOT EDIT
 v2:
-	schema extract (pass / total): 1077 / 1637 = 65.8%
-	tests (pass / total): 3467 / 7175 = 48.3%
-	tests on extracted schemas (pass / total): 3467 / 3884 = 89.3%
+	schema extract (pass / total): 1223 / 1637 = 74.7%
+	tests (pass / total): 4936 / 7175 = 68.8%
+	tests on extracted schemas (pass / total): 4936 / 6059 = 81.5%
 
 v3:
-	schema extract (pass / total): 1065 / 1637 = 65.1%
-	tests (pass / total): 3418 / 7175 = 47.6%
-	tests on extracted schemas (pass / total): 3418 / 3840 = 89.0%
+	schema extract (pass / total): 1211 / 1637 = 74.0%
+	tests (pass / total): 4887 / 7175 = 68.1%
+	tests on extracted schemas (pass / total): 4887 / 6015 = 81.2%