feat: faster parser (#3)

* fix(load): support nested objects

* feat: initial golang parser

Moves most of load.libsonnet to a Golang app for hopefully much faster
parsing (fingers crossed)

* feat(fast): nested objects, subpackages

* feat: incorporate fast mode into main binary

Moves the work done in the fast directory into the main executable,
removing the Jsonnet based parser in favor of the much faster Golang one.
This commit is contained in:
sh0rez 2020-05-10 23:09:12 +02:00 committed by GitHub
parent 75f8e7373c
commit 4e4d4a7170
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
7 changed files with 442 additions and 57 deletions

234
fast/main.go Normal file
View file

@ -0,0 +1,234 @@
package main
import (
"encoding/json"
"fmt"
"io/ioutil"
"log"
"path/filepath"
"strings"
"time"
"github.com/google/go-jsonnet"
"github.com/sh0rez/docsonnet/pkg/docsonnet"
"github.com/sh0rez/docsonnet/pkg/render"
)
func main() {
data, err := eval()
if err != nil {
log.Fatalln(err)
}
// fmt.Println(string(data))
var d DS
if err := json.Unmarshal(data, &d); err != nil {
log.Fatalln(err)
}
pkg := load(d)
fmt.Println("render")
res := render.Render(pkg)
for k, v := range res {
fmt.Println(k)
if err := ioutil.WriteFile(filepath.Join("docs", k), []byte(v), 0644); err != nil {
log.Fatalln(err)
}
}
}
// load docsonnet
//
// Data assumptions:
// - only map[string]interface{} and docsonnet fields
// - docsonnet fields (#...) coming first
func load(d DS) docsonnet.Package {
start := time.Now()
pkg := d.Package()
fmt.Println("load", pkg.Name)
pkg.API = make(docsonnet.Fields)
pkg.Sub = make(map[string]docsonnet.Package)
for k, v := range d {
if k == "#" {
continue
}
f := v.(map[string]interface{})
// docsonnet field
name := strings.TrimPrefix(k, "#")
if strings.HasPrefix(k, "#") {
pkg.API[name] = loadField(name, f, d)
continue
}
// non-docsonnet
// subpackage?
if _, ok := f["#"]; ok {
p := load(DS(f))
pkg.Sub[p.Name] = p
continue
}
// non-annotated nested?
// try to load, but skip when already loaded as annotated above
if nested, ok := loadNested(name, f); ok && !fieldsHas(pkg.API, name) {
pkg.API[name] = *nested
continue
}
}
fmt.Println("done load", pkg.Name, time.Since(start))
return pkg
}
func fieldsHas(f docsonnet.Fields, key string) bool {
_, b := f[key]
return b
}
func loadNested(name string, msi map[string]interface{}) (*docsonnet.Field, bool) {
out := docsonnet.Object{
Name: name,
Fields: make(docsonnet.Fields),
}
ok := false
for k, v := range msi {
f := v.(map[string]interface{})
n := strings.TrimPrefix(k, "#")
if !strings.HasPrefix(k, "#") {
if l, ok := loadNested(k, f); ok {
out.Fields[n] = *l
}
continue
}
ok = true
l := loadField(n, f, msi)
out.Fields[n] = l
}
if !ok {
return nil, false
}
return &docsonnet.Field{Object: &out}, true
}
func loadField(name string, field map[string]interface{}, parent map[string]interface{}) docsonnet.Field {
if ifn, ok := field["function"]; ok {
return loadFn(name, ifn.(map[string]interface{}))
}
if iobj, ok := field["object"]; ok {
return loadObj(name, iobj.(map[string]interface{}), parent)
}
panic("docsonnet field lacking {function | object}")
}
func loadFn(name string, msi map[string]interface{}) docsonnet.Field {
fn := docsonnet.Function{
Name: name,
Help: msi["help"].(string),
}
if args, ok := msi["args"]; ok {
fn.Args = loadArgs(args.([]interface{}))
}
return docsonnet.Field{Function: &fn}
}
func loadArgs(is []interface{}) []docsonnet.Argument {
args := make([]docsonnet.Argument, len(is))
for i := range is {
arg := is[i].(map[string]interface{})
args[i] = docsonnet.Argument{
Name: arg["name"].(string),
Type: docsonnet.Type(arg["type"].(string)),
Default: arg["default"],
}
}
return args
}
func fieldNames(msi map[string]interface{}) []string {
out := make([]string, 0, len(msi))
for k := range msi {
out = append(out, k)
}
return out
}
func loadObj(name string, msi map[string]interface{}, parent map[string]interface{}) docsonnet.Field {
obj := docsonnet.Object{
Name: name,
Help: msi["help"].(string),
Fields: make(docsonnet.Fields),
}
// look for children in same key without #
var iChilds interface{}
var ok bool
if iChilds, ok = parent[name]; !ok {
fmt.Println("aborting, no", name, strings.Join(fieldNames(parent), ", "))
return docsonnet.Field{Object: &obj}
}
childs := iChilds.(map[string]interface{})
for k, v := range childs {
name := strings.TrimPrefix(k, "#")
f := v.(map[string]interface{})
if !strings.HasPrefix(k, "#") {
if l, ok := loadNested(k, f); ok {
obj.Fields[name] = *l
}
continue
}
obj.Fields[name] = loadField(name, f, childs)
}
return docsonnet.Field{Object: &obj}
}
type DS map[string]interface{}
func (d DS) Package() docsonnet.Package {
hash, ok := d["#"]
if !ok {
log.Fatalln("Package declaration missing")
}
pkg := hash.(map[string]interface{})
return docsonnet.Package{
Help: pkg["help"].(string),
Name: pkg["name"].(string),
Import: pkg["import"].(string),
}
}
func eval() ([]byte, error) {
fmt.Println("eval start")
start := time.Now()
vm := jsonnet.MakeVM()
vm.Importer(&jsonnet.FileImporter{JPaths: []string{".."}})
data, err := ioutil.ReadFile("fast.libsonnet")
if err != nil {
return nil, err
}
out, err := vm.EvaluateSnippet("fast.libsonnet", string(data))
if err != nil {
return nil, err
}
fmt.Println("eval:", time.Since(start))
return []byte(out), nil
}

2
go.sum
View file

@ -2,8 +2,6 @@ github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSs
github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
github.com/fatih/color v1.7.0/go.mod h1:Zm6kSWBoL9eyXnKyktHP6abPY2pDugNf5KwzbycvMj4=
github.com/go-clix/cli v0.1.1 h1:T9N0AdMbmpFM9cLw42TcLL5sQ3YgyxTyHUhBK0GW1LI=
github.com/go-clix/cli v0.1.1/go.mod h1:dYJevXraB9mXZFhz5clyQestG0qGcmT5rRC/P9etoRQ=
github.com/go-clix/cli v0.1.2-0.20200502172020-b8f4629e879a h1:nh+UOawbjKgiUAJAgi8JHctNebEu6mjwDXsv8Xdln8w=
github.com/go-clix/cli v0.1.2-0.20200502172020-b8f4629e879a/go.mod h1:dYJevXraB9mXZFhz5clyQestG0qGcmT5rRC/P9etoRQ=
github.com/gobuffalo/here v0.6.0 h1:hYrd0a6gDmWxBM4TnrGw8mQg24iSVoIkHEk7FodQcBI=

View file

@ -1,59 +1,24 @@
local lib = {
// reshape converts the Jsonnet structure to the one used by docsonnet:
// - put fields into an `api` key
// - put subpackages into `sub` key
reshape(pkg)::
local aux(old, key) =
if key == '#' then
old
else if std.objectHas(pkg[key], '#') then
old { sub+: { [key]: $.package(pkg[key]) } }
else
old { api+: { [key]: pkg[key] } };
std.foldl(aux, std.objectFields(pkg), {})
+ pkg['#'],
// fillObjects creates docsonnet objects from Jsonnet ones,
// also filling those that have been specified explicitely
fillObjects(api)::
scan(obj)::
local aux(old, key) =
if std.startsWith(key, '#') then
old { [key]: api[key] }
else if std.isObject(api[key]) && std.length(std.objectFields(api[key])) > 0 then
old { ['#' + key]+: { object+: {
fields: api[key],
} } }
true
else if std.isObject(obj[key]) then
old || $.scan(obj[key])
else old;
std.foldl(aux, std.objectFieldsAll(obj), false),
load(pkg)::
local aux(old, key) =
if !std.isObject(pkg[key]) then
old
else if std.startsWith(key, '#') then
old { [key]: pkg[key] }
else if self.scan(pkg[key]) then
old { [key]: $.load(pkg[key]) }
else old;
std.foldl(aux, std.objectFields(api), {}),
// clean removes all hashes from field names
clean(api):: {
[std.lstripChars(key, '#')]:
if std.isObject(api[key]) then $.clean(api[key])
else api[key]
for key in std.objectFields(api)
},
cleanNonObj(api):: {
[key]:
if std.startsWith(key, "#") then api[key]
else if std.isObject(api[key]) then $.cleanNonObj(api[key])
else api[key]
for key in std.objectFieldsAll(api)
if std.isObject(api[key])
},
// package loads docsonnet from a Jsonnet package
package(pkg)::
local cleaned = self.cleanNonObj(pkg);
local reshaped = self.reshape(cleaned);
local filled =
if std.objectHas(reshaped, 'api')
then reshaped { api: $.fillObjects(reshaped.api) }
else reshaped;
self.clean(filled),
std.foldl(aux, std.objectFieldsAll(pkg), {}),
};
lib.package(std.extVar("main"))
lib.load(std.extVar('main'))

183
pkg/docsonnet/fast.go Normal file
View file

@ -0,0 +1,183 @@
package docsonnet
import (
"fmt"
"log"
"strings"
"time"
)
// load docsonnet
//
// Data assumptions:
// - only map[string]interface{} and fields
// - fields (#...) coming first
func fastLoad(d DS) Package {
start := time.Now()
pkg := d.Package()
fmt.Println("load", pkg.Name)
pkg.API = make(Fields)
pkg.Sub = make(map[string]Package)
for k, v := range d {
if k == "#" {
continue
}
f := v.(map[string]interface{})
// field
name := strings.TrimPrefix(k, "#")
if strings.HasPrefix(k, "#") {
pkg.API[name] = loadField(name, f, d)
continue
}
// non-docsonnet
// subpackage?
if _, ok := f["#"]; ok {
p := fastLoad(DS(f))
pkg.Sub[p.Name] = p
continue
}
// non-annotated nested?
// try to load, but skip when already loaded as annotated above
if nested, ok := loadNested(name, f); ok && !fieldsHas(pkg.API, name) {
pkg.API[name] = *nested
continue
}
}
fmt.Println("done load", pkg.Name, time.Since(start))
return pkg
}
func fieldsHas(f Fields, key string) bool {
_, b := f[key]
return b
}
func loadNested(name string, msi map[string]interface{}) (*Field, bool) {
out := Object{
Name: name,
Fields: make(Fields),
}
ok := false
for k, v := range msi {
f := v.(map[string]interface{})
n := strings.TrimPrefix(k, "#")
if !strings.HasPrefix(k, "#") {
if l, ok := loadNested(k, f); ok {
out.Fields[n] = *l
}
continue
}
ok = true
l := loadField(n, f, msi)
out.Fields[n] = l
}
if !ok {
return nil, false
}
return &Field{Object: &out}, true
}
func loadField(name string, field map[string]interface{}, parent map[string]interface{}) Field {
if ifn, ok := field["function"]; ok {
return loadFn(name, ifn.(map[string]interface{}))
}
if iobj, ok := field["object"]; ok {
return loadObj(name, iobj.(map[string]interface{}), parent)
}
panic("field lacking {function | object}")
}
func loadFn(name string, msi map[string]interface{}) Field {
fn := Function{
Name: name,
Help: msi["help"].(string),
}
if args, ok := msi["args"]; ok {
fn.Args = loadArgs(args.([]interface{}))
}
return Field{Function: &fn}
}
func loadArgs(is []interface{}) []Argument {
args := make([]Argument, len(is))
for i := range is {
arg := is[i].(map[string]interface{})
args[i] = Argument{
Name: arg["name"].(string),
Type: Type(arg["type"].(string)),
Default: arg["default"],
}
}
return args
}
func fieldNames(msi map[string]interface{}) []string {
out := make([]string, 0, len(msi))
for k := range msi {
out = append(out, k)
}
return out
}
func loadObj(name string, msi map[string]interface{}, parent map[string]interface{}) Field {
obj := Object{
Name: name,
Help: msi["help"].(string),
Fields: make(Fields),
}
// look for children in same key without #
var iChilds interface{}
var ok bool
if iChilds, ok = parent[name]; !ok {
fmt.Println("aborting, no", name, strings.Join(fieldNames(parent), ", "))
return Field{Object: &obj}
}
childs := iChilds.(map[string]interface{})
for k, v := range childs {
name := strings.TrimPrefix(k, "#")
f := v.(map[string]interface{})
if !strings.HasPrefix(k, "#") {
if l, ok := loadNested(k, f); ok {
obj.Fields[name] = *l
}
continue
}
obj.Fields[name] = loadField(name, f, childs)
}
return Field{Object: &obj}
}
type DS map[string]interface{}
func (d DS) Package() Package {
hash, ok := d["#"]
if !ok {
log.Fatalln("Package declaration missing")
}
pkg := hash.(map[string]interface{})
return Package{
Help: pkg["help"].(string),
Name: pkg["name"].(string),
Import: pkg["import"].(string),
}
}

View file

@ -32,18 +32,22 @@ func Load(filename string) (*Package, error) {
// invoke load.libsonnet
vm.ExtCode("main", fmt.Sprintf(`(import "%s")`, filename))
log.Println("evaluating Jsonnet")
data, err := vm.EvaluateSnippet("load.libsonnet", string(load))
if err != nil {
return nil, err
}
log.Println("parsing result")
// parse the result
var d Package
var d DS
if err := json.Unmarshal([]byte(data), &d); err != nil {
log.Fatalln(err)
}
return &d, nil
p := fastLoad(d)
return &p, nil
}
// importer wraps jsonnet.FileImporter, to statically provide load.libsonnet,

View file

@ -15,6 +15,7 @@ func Render(pkg docsonnet.Package) map[string]string {
}
func render(pkg docsonnet.Package, parents []string, root bool) map[string]string {
fmt.Println("render", pkg.Name)
link := "/" + strings.Join(append(parents, pkg.Name), "/")
if root {
link = "/"

File diff suppressed because one or more lines are too long