diff --git a/vendor/google.golang.org/protobuf/internal/filedesc/desc_list.go b/vendor/google.golang.org/protobuf/internal/filedesc/desc_list.go
new file mode 100644
index 0000000..aa294ff
--- /dev/null
+++ b/vendor/google.golang.org/protobuf/internal/filedesc/desc_list.go
@@ -0,0 +1,450 @@
+// Copyright 2019 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 filedesc
+
+import (
+	"fmt"
+	"math"
+	"sort"
+	"sync"
+
+	"google.golang.org/protobuf/internal/genid"
+
+	"google.golang.org/protobuf/encoding/protowire"
+	"google.golang.org/protobuf/internal/descfmt"
+	"google.golang.org/protobuf/internal/errors"
+	"google.golang.org/protobuf/internal/pragma"
+	"google.golang.org/protobuf/reflect/protoreflect"
+	pref "google.golang.org/protobuf/reflect/protoreflect"
+)
+
+type FileImports []pref.FileImport
+
+func (p *FileImports) Len() int                            { return len(*p) }
+func (p *FileImports) Get(i int) pref.FileImport           { return (*p)[i] }
+func (p *FileImports) Format(s fmt.State, r rune)          { descfmt.FormatList(s, r, p) }
+func (p *FileImports) ProtoInternal(pragma.DoNotImplement) {}
+
+type Names struct {
+	List []pref.Name
+	once sync.Once
+	has  map[pref.Name]int // protected by once
+}
+
+func (p *Names) Len() int                            { return len(p.List) }
+func (p *Names) Get(i int) pref.Name                 { return p.List[i] }
+func (p *Names) Has(s pref.Name) bool                { return p.lazyInit().has[s] > 0 }
+func (p *Names) Format(s fmt.State, r rune)          { descfmt.FormatList(s, r, p) }
+func (p *Names) ProtoInternal(pragma.DoNotImplement) {}
+func (p *Names) lazyInit() *Names {
+	p.once.Do(func() {
+		if len(p.List) > 0 {
+			p.has = make(map[pref.Name]int, len(p.List))
+			for _, s := range p.List {
+				p.has[s] = p.has[s] + 1
+			}
+		}
+	})
+	return p
+}
+
+// CheckValid reports any errors with the set of names with an error message
+// that completes the sentence: "ranges is invalid because it has ..."
+func (p *Names) CheckValid() error {
+	for s, n := range p.lazyInit().has {
+		switch {
+		case n > 1:
+			return errors.New("duplicate name: %q", s)
+		case false && !s.IsValid():
+			// NOTE: The C++ implementation does not validate the identifier.
+			// See https://github.com/protocolbuffers/protobuf/issues/6335.
+			return errors.New("invalid name: %q", s)
+		}
+	}
+	return nil
+}
+
+type EnumRanges struct {
+	List   [][2]pref.EnumNumber // start inclusive; end inclusive
+	once   sync.Once
+	sorted [][2]pref.EnumNumber // protected by once
+}
+
+func (p *EnumRanges) Len() int                     { return len(p.List) }
+func (p *EnumRanges) Get(i int) [2]pref.EnumNumber { return p.List[i] }
+func (p *EnumRanges) Has(n pref.EnumNumber) bool {
+	for ls := p.lazyInit().sorted; len(ls) > 0; {
+		i := len(ls) / 2
+		switch r := enumRange(ls[i]); {
+		case n < r.Start():
+			ls = ls[:i] // search lower
+		case n > r.End():
+			ls = ls[i+1:] // search upper
+		default:
+			return true
+		}
+	}
+	return false
+}
+func (p *EnumRanges) Format(s fmt.State, r rune)          { descfmt.FormatList(s, r, p) }
+func (p *EnumRanges) ProtoInternal(pragma.DoNotImplement) {}
+func (p *EnumRanges) lazyInit() *EnumRanges {
+	p.once.Do(func() {
+		p.sorted = append(p.sorted, p.List...)
+		sort.Slice(p.sorted, func(i, j int) bool {
+			return p.sorted[i][0] < p.sorted[j][0]
+		})
+	})
+	return p
+}
+
+// CheckValid reports any errors with the set of names with an error message
+// that completes the sentence: "ranges is invalid because it has ..."
+func (p *EnumRanges) CheckValid() error {
+	var rp enumRange
+	for i, r := range p.lazyInit().sorted {
+		r := enumRange(r)
+		switch {
+		case !(r.Start() <= r.End()):
+			return errors.New("invalid range: %v", r)
+		case !(rp.End() < r.Start()) && i > 0:
+			return errors.New("overlapping ranges: %v with %v", rp, r)
+		}
+		rp = r
+	}
+	return nil
+}
+
+type enumRange [2]protoreflect.EnumNumber
+
+func (r enumRange) Start() protoreflect.EnumNumber { return r[0] } // inclusive
+func (r enumRange) End() protoreflect.EnumNumber   { return r[1] } // inclusive
+func (r enumRange) String() string {
+	if r.Start() == r.End() {
+		return fmt.Sprintf("%d", r.Start())
+	}
+	return fmt.Sprintf("%d to %d", r.Start(), r.End())
+}
+
+type FieldRanges struct {
+	List   [][2]pref.FieldNumber // start inclusive; end exclusive
+	once   sync.Once
+	sorted [][2]pref.FieldNumber // protected by once
+}
+
+func (p *FieldRanges) Len() int                      { return len(p.List) }
+func (p *FieldRanges) Get(i int) [2]pref.FieldNumber { return p.List[i] }
+func (p *FieldRanges) Has(n pref.FieldNumber) bool {
+	for ls := p.lazyInit().sorted; len(ls) > 0; {
+		i := len(ls) / 2
+		switch r := fieldRange(ls[i]); {
+		case n < r.Start():
+			ls = ls[:i] // search lower
+		case n > r.End():
+			ls = ls[i+1:] // search upper
+		default:
+			return true
+		}
+	}
+	return false
+}
+func (p *FieldRanges) Format(s fmt.State, r rune)          { descfmt.FormatList(s, r, p) }
+func (p *FieldRanges) ProtoInternal(pragma.DoNotImplement) {}
+func (p *FieldRanges) lazyInit() *FieldRanges {
+	p.once.Do(func() {
+		p.sorted = append(p.sorted, p.List...)
+		sort.Slice(p.sorted, func(i, j int) bool {
+			return p.sorted[i][0] < p.sorted[j][0]
+		})
+	})
+	return p
+}
+
+// CheckValid reports any errors with the set of ranges with an error message
+// that completes the sentence: "ranges is invalid because it has ..."
+func (p *FieldRanges) CheckValid(isMessageSet bool) error {
+	var rp fieldRange
+	for i, r := range p.lazyInit().sorted {
+		r := fieldRange(r)
+		switch {
+		case !isValidFieldNumber(r.Start(), isMessageSet):
+			return errors.New("invalid field number: %d", r.Start())
+		case !isValidFieldNumber(r.End(), isMessageSet):
+			return errors.New("invalid field number: %d", r.End())
+		case !(r.Start() <= r.End()):
+			return errors.New("invalid range: %v", r)
+		case !(rp.End() < r.Start()) && i > 0:
+			return errors.New("overlapping ranges: %v with %v", rp, r)
+		}
+		rp = r
+	}
+	return nil
+}
+
+// isValidFieldNumber reports whether the field number is valid.
+// Unlike the FieldNumber.IsValid method, it allows ranges that cover the
+// reserved number range.
+func isValidFieldNumber(n protoreflect.FieldNumber, isMessageSet bool) bool {
+	return protowire.MinValidNumber <= n && (n <= protowire.MaxValidNumber || isMessageSet)
+}
+
+// CheckOverlap reports an error if p and q overlap.
+func (p *FieldRanges) CheckOverlap(q *FieldRanges) error {
+	rps := p.lazyInit().sorted
+	rqs := q.lazyInit().sorted
+	for pi, qi := 0, 0; pi < len(rps) && qi < len(rqs); {
+		rp := fieldRange(rps[pi])
+		rq := fieldRange(rqs[qi])
+		if !(rp.End() < rq.Start() || rq.End() < rp.Start()) {
+			return errors.New("overlapping ranges: %v with %v", rp, rq)
+		}
+		if rp.Start() < rq.Start() {
+			pi++
+		} else {
+			qi++
+		}
+	}
+	return nil
+}
+
+type fieldRange [2]protoreflect.FieldNumber
+
+func (r fieldRange) Start() protoreflect.FieldNumber { return r[0] }     // inclusive
+func (r fieldRange) End() protoreflect.FieldNumber   { return r[1] - 1 } // inclusive
+func (r fieldRange) String() string {
+	if r.Start() == r.End() {
+		return fmt.Sprintf("%d", r.Start())
+	}
+	return fmt.Sprintf("%d to %d", r.Start(), r.End())
+}
+
+type FieldNumbers struct {
+	List []pref.FieldNumber
+	once sync.Once
+	has  map[pref.FieldNumber]struct{} // protected by once
+}
+
+func (p *FieldNumbers) Len() int                   { return len(p.List) }
+func (p *FieldNumbers) Get(i int) pref.FieldNumber { return p.List[i] }
+func (p *FieldNumbers) Has(n pref.FieldNumber) bool {
+	p.once.Do(func() {
+		if len(p.List) > 0 {
+			p.has = make(map[pref.FieldNumber]struct{}, len(p.List))
+			for _, n := range p.List {
+				p.has[n] = struct{}{}
+			}
+		}
+	})
+	_, ok := p.has[n]
+	return ok
+}
+func (p *FieldNumbers) Format(s fmt.State, r rune)          { descfmt.FormatList(s, r, p) }
+func (p *FieldNumbers) ProtoInternal(pragma.DoNotImplement) {}
+
+type OneofFields struct {
+	List   []pref.FieldDescriptor
+	once   sync.Once
+	byName map[pref.Name]pref.FieldDescriptor        // protected by once
+	byJSON map[string]pref.FieldDescriptor           // protected by once
+	byText map[string]pref.FieldDescriptor           // protected by once
+	byNum  map[pref.FieldNumber]pref.FieldDescriptor // protected by once
+}
+
+func (p *OneofFields) Len() int                                         { return len(p.List) }
+func (p *OneofFields) Get(i int) pref.FieldDescriptor                   { return p.List[i] }
+func (p *OneofFields) ByName(s pref.Name) pref.FieldDescriptor          { return p.lazyInit().byName[s] }
+func (p *OneofFields) ByJSONName(s string) pref.FieldDescriptor         { return p.lazyInit().byJSON[s] }
+func (p *OneofFields) ByTextName(s string) pref.FieldDescriptor         { return p.lazyInit().byText[s] }
+func (p *OneofFields) ByNumber(n pref.FieldNumber) pref.FieldDescriptor { return p.lazyInit().byNum[n] }
+func (p *OneofFields) Format(s fmt.State, r rune)                       { descfmt.FormatList(s, r, p) }
+func (p *OneofFields) ProtoInternal(pragma.DoNotImplement)              {}
+
+func (p *OneofFields) lazyInit() *OneofFields {
+	p.once.Do(func() {
+		if len(p.List) > 0 {
+			p.byName = make(map[pref.Name]pref.FieldDescriptor, len(p.List))
+			p.byJSON = make(map[string]pref.FieldDescriptor, len(p.List))
+			p.byText = make(map[string]pref.FieldDescriptor, len(p.List))
+			p.byNum = make(map[pref.FieldNumber]pref.FieldDescriptor, len(p.List))
+			for _, f := range p.List {
+				// Field names and numbers are guaranteed to be unique.
+				p.byName[f.Name()] = f
+				p.byJSON[f.JSONName()] = f
+				p.byText[f.TextName()] = f
+				p.byNum[f.Number()] = f
+			}
+		}
+	})
+	return p
+}
+
+type SourceLocations struct {
+	// List is a list of SourceLocations.
+	// The SourceLocation.Next field does not need to be populated
+	// as it will be lazily populated upon first need.
+	List []pref.SourceLocation
+
+	// File is the parent file descriptor that these locations are relative to.
+	// If non-nil, ByDescriptor verifies that the provided descriptor
+	// is a child of this file descriptor.
+	File pref.FileDescriptor
+
+	once   sync.Once
+	byPath map[pathKey]int
+}
+
+func (p *SourceLocations) Len() int                      { return len(p.List) }
+func (p *SourceLocations) Get(i int) pref.SourceLocation { return p.lazyInit().List[i] }
+func (p *SourceLocations) byKey(k pathKey) pref.SourceLocation {
+	if i, ok := p.lazyInit().byPath[k]; ok {
+		return p.List[i]
+	}
+	return pref.SourceLocation{}
+}
+func (p *SourceLocations) ByPath(path pref.SourcePath) pref.SourceLocation {
+	return p.byKey(newPathKey(path))
+}
+func (p *SourceLocations) ByDescriptor(desc pref.Descriptor) pref.SourceLocation {
+	if p.File != nil && desc != nil && p.File != desc.ParentFile() {
+		return pref.SourceLocation{} // mismatching parent files
+	}
+	var pathArr [16]int32
+	path := pathArr[:0]
+	for {
+		switch desc.(type) {
+		case pref.FileDescriptor:
+			// Reverse the path since it was constructed in reverse.
+			for i, j := 0, len(path)-1; i < j; i, j = i+1, j-1 {
+				path[i], path[j] = path[j], path[i]
+			}
+			return p.byKey(newPathKey(path))
+		case pref.MessageDescriptor:
+			path = append(path, int32(desc.Index()))
+			desc = desc.Parent()
+			switch desc.(type) {
+			case pref.FileDescriptor:
+				path = append(path, int32(genid.FileDescriptorProto_MessageType_field_number))
+			case pref.MessageDescriptor:
+				path = append(path, int32(genid.DescriptorProto_NestedType_field_number))
+			default:
+				return pref.SourceLocation{}
+			}
+		case pref.FieldDescriptor:
+			isExtension := desc.(pref.FieldDescriptor).IsExtension()
+			path = append(path, int32(desc.Index()))
+			desc = desc.Parent()
+			if isExtension {
+				switch desc.(type) {
+				case pref.FileDescriptor:
+					path = append(path, int32(genid.FileDescriptorProto_Extension_field_number))
+				case pref.MessageDescriptor:
+					path = append(path, int32(genid.DescriptorProto_Extension_field_number))
+				default:
+					return pref.SourceLocation{}
+				}
+			} else {
+				switch desc.(type) {
+				case pref.MessageDescriptor:
+					path = append(path, int32(genid.DescriptorProto_Field_field_number))
+				default:
+					return pref.SourceLocation{}
+				}
+			}
+		case pref.OneofDescriptor:
+			path = append(path, int32(desc.Index()))
+			desc = desc.Parent()
+			switch desc.(type) {
+			case pref.MessageDescriptor:
+				path = append(path, int32(genid.DescriptorProto_OneofDecl_field_number))
+			default:
+				return pref.SourceLocation{}
+			}
+		case pref.EnumDescriptor:
+			path = append(path, int32(desc.Index()))
+			desc = desc.Parent()
+			switch desc.(type) {
+			case pref.FileDescriptor:
+				path = append(path, int32(genid.FileDescriptorProto_EnumType_field_number))
+			case pref.MessageDescriptor:
+				path = append(path, int32(genid.DescriptorProto_EnumType_field_number))
+			default:
+				return pref.SourceLocation{}
+			}
+		case pref.EnumValueDescriptor:
+			path = append(path, int32(desc.Index()))
+			desc = desc.Parent()
+			switch desc.(type) {
+			case pref.EnumDescriptor:
+				path = append(path, int32(genid.EnumDescriptorProto_Value_field_number))
+			default:
+				return pref.SourceLocation{}
+			}
+		case pref.ServiceDescriptor:
+			path = append(path, int32(desc.Index()))
+			desc = desc.Parent()
+			switch desc.(type) {
+			case pref.FileDescriptor:
+				path = append(path, int32(genid.FileDescriptorProto_Service_field_number))
+			default:
+				return pref.SourceLocation{}
+			}
+		case pref.MethodDescriptor:
+			path = append(path, int32(desc.Index()))
+			desc = desc.Parent()
+			switch desc.(type) {
+			case pref.ServiceDescriptor:
+				path = append(path, int32(genid.ServiceDescriptorProto_Method_field_number))
+			default:
+				return pref.SourceLocation{}
+			}
+		default:
+			return pref.SourceLocation{}
+		}
+	}
+}
+func (p *SourceLocations) lazyInit() *SourceLocations {
+	p.once.Do(func() {
+		if len(p.List) > 0 {
+			// Collect all the indexes for a given path.
+			pathIdxs := make(map[pathKey][]int, len(p.List))
+			for i, l := range p.List {
+				k := newPathKey(l.Path)
+				pathIdxs[k] = append(pathIdxs[k], i)
+			}
+
+			// Update the next index for all locations.
+			p.byPath = make(map[pathKey]int, len(p.List))
+			for k, idxs := range pathIdxs {
+				for i := 0; i < len(idxs)-1; i++ {
+					p.List[idxs[i]].Next = idxs[i+1]
+				}
+				p.List[idxs[len(idxs)-1]].Next = 0
+				p.byPath[k] = idxs[0] // record the first location for this path
+			}
+		}
+	})
+	return p
+}
+func (p *SourceLocations) ProtoInternal(pragma.DoNotImplement) {}
+
+// pathKey is a comparable representation of protoreflect.SourcePath.
+type pathKey struct {
+	arr [16]uint8 // first n-1 path segments; last element is the length
+	str string    // used if the path does not fit in arr
+}
+
+func newPathKey(p pref.SourcePath) (k pathKey) {
+	if len(p) < len(k.arr) {
+		for i, ps := range p {
+			if ps < 0 || math.MaxUint8 <= ps {
+				return pathKey{str: p.String()}
+			}
+			k.arr[i] = uint8(ps)
+		}
+		k.arr[len(k.arr)-1] = uint8(len(p))
+		return k
+	}
+	return pathKey{str: p.String()}
+}