refactor: clean duplicate files with aws chroot builder

pull/7459/head
Marin Salinas 7 years ago committed by Megan Marsh
parent 488535118d
commit ed2d2b9058

@ -11,6 +11,7 @@ import (
"net/http"
"runtime"
awschroot "github.com/hashicorp/packer/builder/amazon/chroot"
osccommon "github.com/hashicorp/packer/builder/osc/common"
"github.com/hashicorp/packer/common"
"github.com/hashicorp/packer/helper/config"
@ -237,7 +238,7 @@ func (b *Builder) Run(ui packer.Ui, hook packer.Hook) (packer.Artifact, error) {
}
steps = append(steps,
&StepFlock{},
&awschroot.StepFlock{},
&StepPrepareDevice{},
&StepCreateVolume{
RootVolumeType: b.config.RootVolumeType,
@ -246,21 +247,21 @@ func (b *Builder) Run(ui packer.Ui, hook packer.Hook) (packer.Artifact, error) {
Ctx: b.config.ctx,
},
&StepLinkVolume{},
&StepEarlyUnflock{},
&StepPreMountCommands{
&awschroot.StepEarlyUnflock{},
&awschroot.StepPreMountCommands{
Commands: b.config.PreMountCommands,
},
&StepMountDevice{
MountOptions: b.config.MountOptions,
MountPartition: b.config.MountPartition,
},
&StepPostMountCommands{
&awschroot.StepPostMountCommands{
Commands: b.config.PostMountCommands,
},
&StepMountExtra{},
&StepCopyFiles{},
&StepChrootProvision{},
&StepEarlyCleanup{},
&awschroot.StepMountExtra{},
&awschroot.StepCopyFiles{},
&awschroot.StepChrootProvision{},
&awschroot.StepEarlyCleanup{},
&StepSnapshot{},
&osccommon.StepDeregisterOMI{
AccessConfig: &b.config.AccessConfig,

@ -1,10 +0,0 @@
package chroot
import (
"github.com/hashicorp/packer/helper/multistep"
)
// Cleanup is an interface that some steps implement for early cleanup.
type Cleanup interface {
CleanupFunc(multistep.StateBag) error
}

@ -51,7 +51,7 @@ func devicePrefix() (string, error) {
defer f.Close()
dirs, err := f.Readdirnames(-1)
if dirs != nil && len(dirs) > 0 {
if len(dirs) > 0 {
for _, dir := range dirs {
dirBase := filepath.Base(dir)
for _, prefix := range available {

@ -1,16 +0,0 @@
// +build windows
package chroot
import (
"errors"
"os"
)
func lockFile(*os.File) error {
return errors.New("not supported on Windows")
}
func unlockFile(f *os.File) error {
return nil
}

@ -1,27 +0,0 @@
// +build !windows
package chroot
import (
"os"
"golang.org/x/sys/unix"
)
// See: http://linux.die.net/include/sys/file.h
const LOCK_EX = 2
const LOCK_NB = 4
const LOCK_UN = 8
func lockFile(f *os.File) error {
err := unix.Flock(int(f.Fd()), LOCK_EX)
if err != nil {
return err
}
return nil
}
func unlockFile(f *os.File) error {
return unix.Flock(int(f.Fd()), LOCK_UN)
}

@ -1,39 +0,0 @@
package chroot
import (
"fmt"
sl "github.com/hashicorp/packer/common/shell-local"
"github.com/hashicorp/packer/packer"
"github.com/hashicorp/packer/template/interpolate"
)
func RunLocalCommands(commands []string, wrappedCommand CommandWrapper, ctx interpolate.Context, ui packer.Ui) error {
for _, rawCmd := range commands {
intCmd, err := interpolate.Render(rawCmd, &ctx)
if err != nil {
return fmt.Errorf("Error interpolating: %s", err)
}
command, err := wrappedCommand(intCmd)
if err != nil {
return fmt.Errorf("Error wrapping command: %s", err)
}
ui.Say(fmt.Sprintf("Executing command: %s", command))
comm := &sl.Communicator{
ExecuteCommand: []string{"sh", "-c", command},
}
cmd := &packer.RemoteCmd{Command: command}
if err := cmd.StartWithUi(comm, ui); err != nil {
return fmt.Errorf("Error executing command: %s", err)
}
if cmd.ExitStatus != 0 {
return fmt.Errorf(
"Received non-zero exit code %d from command: %s",
cmd.ExitStatus,
command)
}
}
return nil
}

@ -1,37 +0,0 @@
package chroot
import (
"context"
"log"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
// StepChrootProvision provisions the instance within a chroot.
type StepChrootProvision struct {
}
func (s *StepChrootProvision) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
hook := state.Get("hook").(packer.Hook)
mountPath := state.Get("mount_path").(string)
ui := state.Get("ui").(packer.Ui)
wrappedCommand := state.Get("wrappedCommand").(CommandWrapper)
// Create our communicator
comm := &Communicator{
Chroot: mountPath,
CmdWrapper: wrappedCommand,
}
// Provision
log.Println("Running the provision hook")
if err := hook.Run(packer.HookProvision, ui, comm, nil); err != nil {
state.Put("error", err)
return multistep.ActionHalt
}
return multistep.ActionContinue
}
func (s *StepChrootProvision) Cleanup(state multistep.StateBag) {}

@ -1,91 +0,0 @@
package chroot
import (
"bytes"
"context"
"fmt"
"log"
"path/filepath"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
// StepCopyFiles copies some files from the host into the chroot environment.
//
// Produces:
// copy_files_cleanup CleanupFunc - A function to clean up the copied files
// early.
type StepCopyFiles struct {
files []string
}
func (s *StepCopyFiles) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
config := state.Get("config").(*Config)
mountPath := state.Get("mount_path").(string)
ui := state.Get("ui").(packer.Ui)
wrappedCommand := state.Get("wrappedCommand").(CommandWrapper)
stderr := new(bytes.Buffer)
s.files = make([]string, 0, len(config.CopyFiles))
if len(config.CopyFiles) > 0 {
ui.Say("Copying files from host to chroot...")
for _, path := range config.CopyFiles {
ui.Message(path)
chrootPath := filepath.Join(mountPath, path)
log.Printf("Copying '%s' to '%s'", path, chrootPath)
cmdText, err := wrappedCommand(fmt.Sprintf("cp --remove-destination %s %s", path, chrootPath))
if err != nil {
err := fmt.Errorf("Error building copy command: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
stderr.Reset()
cmd := ShellCommand(cmdText)
cmd.Stderr = stderr
if err := cmd.Run(); err != nil {
err := fmt.Errorf(
"Error copying file: %s\nnStderr: %s", err, stderr.String())
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
s.files = append(s.files, chrootPath)
}
}
state.Put("copy_files_cleanup", s)
return multistep.ActionContinue
}
func (s *StepCopyFiles) Cleanup(state multistep.StateBag) {
ui := state.Get("ui").(packer.Ui)
if err := s.CleanupFunc(state); err != nil {
ui.Error(err.Error())
}
}
func (s *StepCopyFiles) CleanupFunc(state multistep.StateBag) error {
wrappedCommand := state.Get("wrappedCommand").(CommandWrapper)
if s.files != nil {
for _, file := range s.files {
log.Printf("Removing: %s", file)
localCmdText, err := wrappedCommand(fmt.Sprintf("rm -f %s", file))
if err != nil {
return err
}
localCmd := ShellCommand(localCmdText)
if err := localCmd.Run(); err != nil {
return err
}
}
}
s.files = nil
return nil
}

@ -1,39 +0,0 @@
package chroot
import (
"context"
"fmt"
"log"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
// StepEarlyCleanup performs some of the cleanup steps early in order to
// prepare for snapshotting and creating an AMI.
type StepEarlyCleanup struct{}
func (s *StepEarlyCleanup) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
ui := state.Get("ui").(packer.Ui)
cleanupKeys := []string{
"copy_files_cleanup",
"mount_extra_cleanup",
"mount_device_cleanup",
"attach_cleanup",
}
for _, key := range cleanupKeys {
c := state.Get(key).(Cleanup)
log.Printf("Running cleanup func: %s", key)
if err := c.CleanupFunc(state); err != nil {
err := fmt.Errorf("Error cleaning up: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
}
return multistep.ActionContinue
}
func (s *StepEarlyCleanup) Cleanup(state multistep.StateBag) {}

@ -1,30 +0,0 @@
package chroot
import (
"context"
"fmt"
"log"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
// StepEarlyUnflock unlocks the flock.
type StepEarlyUnflock struct{}
func (s *StepEarlyUnflock) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
cleanup := state.Get("flock_cleanup").(Cleanup)
ui := state.Get("ui").(packer.Ui)
log.Println("Unlocking file lock...")
if err := cleanup.CleanupFunc(state); err != nil {
err := fmt.Errorf("Error unlocking file lock: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
return multistep.ActionContinue
}
func (s *StepEarlyUnflock) Cleanup(state multistep.StateBag) {}

@ -1,74 +0,0 @@
package chroot
import (
"context"
"fmt"
"log"
"os"
"path/filepath"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
// StepFlock provisions the vm within a chroot.
//
// Produces:
// flock_cleanup Cleanup - To perform early cleanup
type StepFlock struct {
fh *os.File
}
func (s *StepFlock) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
ui := state.Get("ui").(packer.Ui)
lockfile := "/var/lock/packer-chroot/lock"
if err := os.MkdirAll(filepath.Dir(lockfile), 0755); err != nil {
err := fmt.Errorf("Error creating lock: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
log.Printf("Obtaining lock: %s", lockfile)
f, err := os.Create(lockfile)
if err != nil {
err := fmt.Errorf("Error creating lock: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
// LOCK!
if err := lockFile(f); err != nil {
err := fmt.Errorf("Error obtaining lock: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
// Set the file handle, we can't close it because we need to hold
// the lock.
s.fh = f
state.Put("flock_cleanup", s)
return multistep.ActionContinue
}
func (s *StepFlock) Cleanup(state multistep.StateBag) {
s.CleanupFunc(state)
}
func (s *StepFlock) CleanupFunc(state multistep.StateBag) error {
if s.fh == nil {
return nil
}
log.Printf("Unlocking: %s", s.fh.Name())
if err := unlockFile(s.fh); err != nil {
return err
}
s.fh = nil
return nil
}

@ -1,137 +0,0 @@
package chroot
import (
"bytes"
"context"
"fmt"
"os"
"os/exec"
"syscall"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
// StepMountExtra mounts the attached device.
//
// Produces:
// mount_extra_cleanup CleanupFunc - To perform early cleanup
type StepMountExtra struct {
mounts []string
}
func (s *StepMountExtra) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
config := state.Get("config").(*Config)
mountPath := state.Get("mount_path").(string)
ui := state.Get("ui").(packer.Ui)
wrappedCommand := state.Get("wrappedCommand").(CommandWrapper)
s.mounts = make([]string, 0, len(config.ChrootMounts))
ui.Say("Mounting additional paths within the chroot...")
for _, mountInfo := range config.ChrootMounts {
innerPath := mountPath + mountInfo[2]
if err := os.MkdirAll(innerPath, 0755); err != nil {
err := fmt.Errorf("Error creating mount directory: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
flags := "-t " + mountInfo[0]
if mountInfo[0] == "bind" {
flags = "--bind"
}
ui.Message(fmt.Sprintf("Mounting: %s", mountInfo[2]))
stderr := new(bytes.Buffer)
mountCommand, err := wrappedCommand(fmt.Sprintf(
"mount %s %s %s",
flags,
mountInfo[1],
innerPath))
if err != nil {
err := fmt.Errorf("Error creating mount command: %s", err)
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
cmd := ShellCommand(mountCommand)
cmd.Stderr = stderr
if err := cmd.Run(); err != nil {
err := fmt.Errorf(
"Error mounting: %s\nStderr: %s", err, stderr.String())
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
s.mounts = append(s.mounts, innerPath)
}
state.Put("mount_extra_cleanup", s)
return multistep.ActionContinue
}
func (s *StepMountExtra) Cleanup(state multistep.StateBag) {
ui := state.Get("ui").(packer.Ui)
if err := s.CleanupFunc(state); err != nil {
ui.Error(err.Error())
return
}
}
func (s *StepMountExtra) CleanupFunc(state multistep.StateBag) error {
if s.mounts == nil {
return nil
}
wrappedCommand := state.Get("wrappedCommand").(CommandWrapper)
for len(s.mounts) > 0 {
var path string
lastIndex := len(s.mounts) - 1
path, s.mounts = s.mounts[lastIndex], s.mounts[:lastIndex]
grepCommand, err := wrappedCommand(fmt.Sprintf("grep %s /proc/mounts", path))
if err != nil {
return fmt.Errorf("Error creating grep command: %s", err)
}
// Before attempting to unmount,
// check to see if path is already unmounted
stderr := new(bytes.Buffer)
cmd := ShellCommand(grepCommand)
cmd.Stderr = stderr
if err := cmd.Run(); err != nil {
if exitError, ok := err.(*exec.ExitError); ok {
if status, ok := exitError.Sys().(syscall.WaitStatus); ok {
exitStatus := status.ExitStatus()
if exitStatus == 1 {
// path has already been unmounted
// just skip this path
continue
}
}
}
}
unmountCommand, err := wrappedCommand(fmt.Sprintf("umount %s", path))
if err != nil {
return fmt.Errorf("Error creating unmount command: %s", err)
}
stderr = new(bytes.Buffer)
cmd = ShellCommand(unmountCommand)
cmd.Stderr = stderr
if err := cmd.Run(); err != nil {
return fmt.Errorf(
"Error unmounting device: %s\nStderr: %s", err, stderr.String())
}
}
s.mounts = nil
return nil
}

@ -1,47 +0,0 @@
package chroot
import (
"context"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
type postMountCommandsData struct {
Device string
MountPath string
}
// StepPostMountCommands allows running arbitrary commands after mounting the
// device, but prior to the bind mount and copy steps.
type StepPostMountCommands struct {
Commands []string
}
func (s *StepPostMountCommands) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
config := state.Get("config").(*Config)
device := state.Get("device").(string)
mountPath := state.Get("mount_path").(string)
ui := state.Get("ui").(packer.Ui)
wrappedCommand := state.Get("wrappedCommand").(CommandWrapper)
if len(s.Commands) == 0 {
return multistep.ActionContinue
}
ctx := config.ctx
ctx.Data = &postMountCommandsData{
Device: device,
MountPath: mountPath,
}
ui.Say("Running post-mount commands...")
if err := RunLocalCommands(s.Commands, wrappedCommand, ctx, ui); err != nil {
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
return multistep.ActionContinue
}
func (s *StepPostMountCommands) Cleanup(state multistep.StateBag) {}

@ -1,41 +0,0 @@
package chroot
import (
"context"
"github.com/hashicorp/packer/helper/multistep"
"github.com/hashicorp/packer/packer"
)
type preMountCommandsData struct {
Device string
}
// StepPreMountCommands sets up the a new block device when building from scratch
type StepPreMountCommands struct {
Commands []string
}
func (s *StepPreMountCommands) Run(_ context.Context, state multistep.StateBag) multistep.StepAction {
config := state.Get("config").(*Config)
device := state.Get("device").(string)
ui := state.Get("ui").(packer.Ui)
wrappedCommand := state.Get("wrappedCommand").(CommandWrapper)
if len(s.Commands) == 0 {
return multistep.ActionContinue
}
ctx := config.ctx
ctx.Data = &preMountCommandsData{Device: device}
ui.Say("Running device setup commands...")
if err := RunLocalCommands(s.Commands, wrappedCommand, ctx, ui); err != nil {
state.Put("error", err)
ui.Error(err.Error())
return multistep.ActionHalt
}
return multistep.ActionContinue
}
func (s *StepPreMountCommands) Cleanup(state multistep.StateBag) {}

@ -1,4 +1,5 @@
// Code generated by pigeon; DO NOT EDIT.
package bootcommand
import (
@ -7,7 +8,9 @@ import (
"fmt"
"io"
"io/ioutil"
"math"
"os"
"sort"
"strconv"
"strings"
"time"
@ -788,18 +791,85 @@ var (
// errNoRule is returned when the grammar to parse has no rule.
errNoRule = errors.New("grammar has no rule")
// errInvalidEntrypoint is returned when the specified entrypoint rule
// does not exit.
errInvalidEntrypoint = errors.New("invalid entrypoint")
// errInvalidEncoding is returned when the source is not properly
// utf8-encoded.
errInvalidEncoding = errors.New("invalid encoding")
// errNoMatch is returned if no match could be found.
errNoMatch = errors.New("no match found")
// errMaxExprCnt is used to signal that the maximum number of
// expressions have been parsed.
errMaxExprCnt = errors.New("max number of expresssions parsed")
)
// Option is a function that can set an option on the parser. It returns
// the previous setting as an Option.
type Option func(*parser) Option
// MaxExpressions creates an Option to stop parsing after the provided
// number of expressions have been parsed, if the value is 0 then the parser will
// parse for as many steps as needed (possibly an infinite number).
//
// The default for maxExprCnt is 0.
func MaxExpressions(maxExprCnt uint64) Option {
return func(p *parser) Option {
oldMaxExprCnt := p.maxExprCnt
p.maxExprCnt = maxExprCnt
return MaxExpressions(oldMaxExprCnt)
}
}
// Entrypoint creates an Option to set the rule name to use as entrypoint.
// The rule name must have been specified in the -alternate-entrypoints
// if generating the parser with the -optimize-grammar flag, otherwise
// it may have been optimized out. Passing an empty string sets the
// entrypoint to the first rule in the grammar.
//
// The default is to start parsing at the first rule in the grammar.
func Entrypoint(ruleName string) Option {
return func(p *parser) Option {
oldEntrypoint := p.entrypoint
p.entrypoint = ruleName
if ruleName == "" {
p.entrypoint = g.rules[0].name
}
return Entrypoint(oldEntrypoint)
}
}
// Statistics adds a user provided Stats struct to the parser to allow
// the user to process the results after the parsing has finished.
// Also the key for the "no match" counter is set.
//
// Example usage:
//
// input := "input"
// stats := Stats{}
// _, err := Parse("input-file", []byte(input), Statistics(&stats, "no match"))
// if err != nil {
// log.Panicln(err)
// }
// b, err := json.MarshalIndent(stats.ChoiceAltCnt, "", " ")
// if err != nil {
// log.Panicln(err)
// }
// fmt.Println(string(b))
//
func Statistics(stats *Stats, choiceNoMatch string) Option {
return func(p *parser) Option {
oldStats := p.Stats
p.Stats = stats
oldChoiceNoMatch := p.choiceNoMatch
p.choiceNoMatch = choiceNoMatch
if p.Stats.ChoiceAltCnt == nil {
p.Stats.ChoiceAltCnt = make(map[string]map[string]int)
}
return Statistics(oldStats, oldChoiceNoMatch)
}
}
// Debug creates an Option to set the debug flag to b. When set to true,
// debugging information is printed to stdout while parsing.
//
@ -826,6 +896,20 @@ func Memoize(b bool) Option {
}
}
// AllowInvalidUTF8 creates an Option to allow invalid UTF-8 bytes.
// Every invalid UTF-8 byte is treated as a utf8.RuneError (U+FFFD)
// by character class matchers and is matched by the any matcher.
// The returned matched value, c.text and c.offset are NOT affected.
//
// The default is false.
func AllowInvalidUTF8(b bool) Option {
return func(p *parser) Option {
old := p.allowInvalidUTF8
p.allowInvalidUTF8 = b
return AllowInvalidUTF8(old)
}
}
// Recover creates an Option to set the recover flag to b. When set to
// true, this causes the parser to recover from panics and convert it
// to an error. Setting it to false can be useful while debugging to
@ -840,13 +924,37 @@ func Recover(b bool) Option {
}
}
// GlobalStore creates an Option to set a key to a certain value in
// the globalStore.
func GlobalStore(key string, value interface{}) Option {
return func(p *parser) Option {
old := p.cur.globalStore[key]
p.cur.globalStore[key] = value
return GlobalStore(key, old)
}
}
// InitState creates an Option to set a key to a certain value in
// the global "state" store.
func InitState(key string, value interface{}) Option {
return func(p *parser) Option {
old := p.cur.state[key]
p.cur.state[key] = value
return InitState(key, old)
}
}
// ParseFile parses the file identified by filename.
func ParseFile(filename string, opts ...Option) (interface{}, error) {
func ParseFile(filename string, opts ...Option) (i interface{}, err error) {
f, err := os.Open(filename)
if err != nil {
return nil, err
}
defer f.Close()
defer func() {
if closeErr := f.Close(); closeErr != nil {
err = closeErr
}
}()
return ParseReader(filename, f, opts...)
}
@ -887,8 +995,22 @@ type savepoint struct {
type current struct {
pos position // start position of the match
text []byte // raw text of the match
// state is a store for arbitrary key,value pairs that the user wants to be
// tied to the backtracking of the parser.
// This is always rolled back if a parsing rule fails.
state storeDict
// globalStore is a general store for the user to store arbitrary key-value
// pairs that they need to manage and that they do not want tied to the
// backtracking of the parser. This is only modified by the user and never
// rolled back by the parser. It is always up to the user to keep this in a
// consistent state.
globalStore storeDict
}
type storeDict map[string]interface{}
// the AST types...
type grammar struct {
@ -914,11 +1036,23 @@ type actionExpr struct {
run func(*parser) (interface{}, error)
}
type recoveryExpr struct {
pos position
expr interface{}
recoverExpr interface{}
failureLabel []string
}
type seqExpr struct {
pos position
exprs []interface{}
}
type throwExpr struct {
pos position
label string
}
type labeledExpr struct {
pos position
label string
@ -941,6 +1075,11 @@ type ruleRefExpr struct {
name string
}
type stateCodeExpr struct {
pos position
run func(*parser) error
}
type andCodeExpr struct {
pos position
run func(*parser) (bool, error)
@ -958,13 +1097,14 @@ type litMatcher struct {
}
type charClassMatcher struct {
pos position
val string
chars []rune
ranges []rune
classes []*unicode.RangeTable
ignoreCase bool
inverted bool
pos position
val string
basicLatinChars [128]bool
chars []rune
ranges []rune
classes []*unicode.RangeTable
ignoreCase bool
inverted bool
}
type anyMatcher position
@ -1018,9 +1158,10 @@ func (e errList) Error() string {
// parserError wraps an error with a prefix indicating the rule in which
// the error occurred. The original error is stored in the Inner field.
type parserError struct {
Inner error
pos position
prefix string
Inner error
pos position
prefix string
expected []string
}
// Error returns the error message.
@ -1030,14 +1171,32 @@ func (p *parserError) Error() string {
// newParser creates a parser with the specified input source and options.
func newParser(filename string, b []byte, opts ...Option) *parser {
stats := Stats{
ChoiceAltCnt: make(map[string]map[string]int),
}
p := &parser{
filename: filename,
errs: new(errList),
data: b,
pt: savepoint{position: position{line: 1}},
recover: true,
cur: current{
state: make(storeDict),
globalStore: make(storeDict),
},
maxFailPos: position{col: 1, line: 1},
maxFailExpected: make([]string, 0, 20),
Stats: &stats,
// start rule is rule [0] unless an alternate entrypoint is specified
entrypoint: g.rules[0].name,
}
p.setOptions(opts)
if p.maxExprCnt == 0 {
p.maxExprCnt = math.MaxUint64
}
return p
}
@ -1054,6 +1213,30 @@ type resultTuple struct {
end savepoint
}
const choiceNoMatch = -1
// Stats stores some statistics, gathered during parsing
type Stats struct {
// ExprCnt counts the number of expressions processed during parsing
// This value is compared to the maximum number of expressions allowed
// (set by the MaxExpressions option).
ExprCnt uint64
// ChoiceAltCnt is used to count for each ordered choice expression,
// which alternative is used how may times.
// These numbers allow to optimize the order of the ordered choice expression
// to increase the performance of the parser
//
// The outer key of ChoiceAltCnt is composed of the name of the rule as well
// as the line and the column of the ordered choice.
// The inner key of ChoiceAltCnt is the number (one-based) of the matching alternative.
// For each alternative the number of matches are counted. If an ordered choice does not
// match, a special counter is incremented. The name of this counter is set with
// the parser option Statistics.
// For an alternative to be included in ChoiceAltCnt, it has to match at least once.
ChoiceAltCnt map[string]map[string]int
}
type parser struct {
filename string
pt savepoint
@ -1062,9 +1245,9 @@ type parser struct {
data []byte
errs *errList
depth int
recover bool
debug bool
depth int
memoize bool
// memoization table for the packrat algorithm:
@ -1078,8 +1261,23 @@ type parser struct {
// rule stack, allows identification of the current rule in errors
rstack []*rule
// stats
exprCnt int
// parse fail
maxFailPos position
maxFailExpected []string
maxFailInvertExpected bool
// max number of expressions to be parsed
maxExprCnt uint64
// entrypoint for the parser
entrypoint string
allowInvalidUTF8 bool
*Stats
choiceNoMatch string
// recovery expression stack, keeps track of the currently available recovery expression, these are traversed in reverse
recoveryStack []map[string]interface{}
}
// push a variable set on the vstack.
@ -1114,6 +1312,31 @@ func (p *parser) popV() {
p.vstack = p.vstack[:len(p.vstack)-1]
}
// push a recovery expression with its labels to the recoveryStack
func (p *parser) pushRecovery(labels []string, expr interface{}) {
if cap(p.recoveryStack) == len(p.recoveryStack) {
// create new empty slot in the stack
p.recoveryStack = append(p.recoveryStack, nil)
} else {
// slice to 1 more
p.recoveryStack = p.recoveryStack[:len(p.recoveryStack)+1]
}
m := make(map[string]interface{}, len(labels))
for _, fl := range labels {
m[fl] = expr
}
p.recoveryStack[len(p.recoveryStack)-1] = m
}
// pop a recovery expression from the recoveryStack
func (p *parser) popRecovery() {
// GC that map
p.recoveryStack[len(p.recoveryStack)-1] = nil
p.recoveryStack = p.recoveryStack[:len(p.recoveryStack)-1]
}
func (p *parser) print(prefix, s string) string {
if !p.debug {
return s
@ -1135,10 +1358,10 @@ func (p *parser) out(s string) string {
}
func (p *parser) addErr(err error) {
p.addErrAt(err, p.pt.position)
p.addErrAt(err, p.pt.position, []string{})
}
func (p *parser) addErrAt(err error, pos position) {
func (p *parser) addErrAt(err error, pos position, expected []string) {
var buf bytes.Buffer
if p.filename != "" {
buf.WriteString(p.filename)
@ -1158,10 +1381,29 @@ func (p *parser) addErrAt(err error, pos position) {
buf.WriteString("rule " + rule.name)
}
}
pe := &parserError{Inner: err, pos: pos, prefix: buf.String()}
pe := &parserError{Inner: err, pos: pos, prefix: buf.String(), expected: expected}
p.errs.add(pe)
}
func (p *parser) failAt(fail bool, pos position, want string) {
// process fail if parsing fails and not inverted or parsing succeeds and invert is set
if fail == p.maxFailInvertExpected {
if pos.offset < p.maxFailPos.offset {
return
}
if pos.offset > p.maxFailPos.offset {
p.maxFailPos = pos
p.maxFailExpected = p.maxFailExpected[:0]
}
if p.maxFailInvertExpected {
want = "!" + want
}
p.maxFailExpected = append(p.maxFailExpected, want)
}
}
// read advances the parser to the next rune.
func (p *parser) read() {
p.pt.offset += p.pt.w
@ -1174,8 +1416,8 @@ func (p *parser) read() {
p.pt.col = 0
}
if rn == utf8.RuneError {
if n == 1 {
if rn == utf8.RuneError && n == 1 { // see utf8.DecodeRune
if !p.allowInvalidUTF8 {
p.addErr(errInvalidEncoding)
}
}
@ -1192,6 +1434,43 @@ func (p *parser) restore(pt savepoint) {
p.pt = pt
}
// Cloner is implemented by any value that has a Clone method, which returns a
// copy of the value. This is mainly used for types which are not passed by
// value (e.g map, slice, chan) or structs that contain such types.
//
// This is used in conjunction with the global state feature to create proper
// copies of the state to allow the parser to properly restore the state in
// the case of backtracking.
type Cloner interface {
Clone() interface{}
}
// clone and return parser current state.
func (p *parser) cloneState() storeDict {
if p.debug {
defer p.out(p.in("cloneState"))
}
state := make(storeDict, len(p.cur.state))
for k, v := range p.cur.state {
if c, ok := v.(Cloner); ok {
state[k] = c.Clone()
} else {
state[k] = v
}
}
return state
}
// restore parser current state to the state storeDict.
// every restoreState should applied only one time for every cloned state
func (p *parser) restoreState(state storeDict) {
if p.debug {
defer p.out(p.in("restoreState"))
}
p.cur.state = state
}
// get the slice of bytes from the savepoint start to the current position.
func (p *parser) sliceFrom(start savepoint) []byte {
return p.data[start.position.offset:p.pt.position.offset]
@ -1257,19 +1536,54 @@ func (p *parser) parse(g *grammar) (val interface{}, err error) {
}()
}
// start rule is rule [0]
startRule, ok := p.rules[p.entrypoint]
if !ok {
p.addErr(errInvalidEntrypoint)
return nil, p.errs.err()
}
p.read() // advance to first rune
val, ok := p.parseRule(g.rules[0])
val, ok = p.parseRule(startRule)
if !ok {
if len(*p.errs) == 0 {
// make sure this doesn't go out silently
p.addErr(errNoMatch)
// If parsing fails, but no errors have been recorded, the expected values
// for the farthest parser position are returned as error.
maxFailExpectedMap := make(map[string]struct{}, len(p.maxFailExpected))
for _, v := range p.maxFailExpected {
maxFailExpectedMap[v] = struct{}{}
}
expected := make([]string, 0, len(maxFailExpectedMap))
eof := false
if _, ok := maxFailExpectedMap["!."]; ok {
delete(maxFailExpectedMap, "!.")
eof = true
}
for k := range maxFailExpectedMap {
expected = append(expected, k)
}
sort.Strings(expected)
if eof {
expected = append(expected, "EOF")
}
p.addErrAt(errors.New("no match found, expected: "+listJoin(expected, ", ", "or")), p.maxFailPos, expected)
}
return nil, p.errs.err()
}
return val, p.errs.err()
}
func listJoin(list []string, sep string, lastSep string) string {
switch len(list) {
case 0:
return ""
case 1:
return list[0]
default:
return fmt.Sprintf("%s %s %s", strings.Join(list[:len(list)-1], sep), lastSep, list[len(list)-1])
}
}
func (p *parser) parseRule(rule *rule) (interface{}, bool) {
if p.debug {
defer p.out(p.in("parseRule " + rule.name))
@ -1301,7 +1615,6 @@ func (p *parser) parseRule(rule *rule) (interface{}, bool) {
func (p *parser) parseExpr(expr interface{}) (interface{}, bool) {
var pt savepoint
var ok bool
if p.memoize {
res, ok := p.getMemoized(expr)
@ -1312,8 +1625,13 @@ func (p *parser) parseExpr(expr interface{}) (interface{}, bool) {
pt = p.pt
}
p.exprCnt++
p.ExprCnt++
if p.ExprCnt > p.maxExprCnt {
panic(errMaxExprCnt)
}
var val interface{}
var ok bool
switch expr := expr.(type) {
case *actionExpr:
val, ok = p.parseActionExpr(expr)
@ -1337,10 +1655,16 @@ func (p *parser) parseExpr(expr interface{}) (interface{}, bool) {
val, ok = p.parseNotExpr(expr)
case *oneOrMoreExpr:
val, ok = p.parseOneOrMoreExpr(expr)
case *recoveryExpr:
val, ok = p.parseRecoveryExpr(expr)
case *ruleRefExpr:
val, ok = p.parseRuleRefExpr(expr)
case *seqExpr:
val, ok = p.parseSeqExpr(expr)
case *stateCodeExpr:
val, ok = p.parseStateCodeExpr(expr)
case *throwExpr:
val, ok = p.parseThrowExpr(expr)
case *zeroOrMoreExpr:
val, ok = p.parseZeroOrMoreExpr(expr)
case *zeroOrOneExpr:
@ -1364,10 +1688,13 @@ func (p *parser) parseActionExpr(act *actionExpr) (interface{}, bool) {
if ok {
p.cur.pos = start.position
p.cur.text = p.sliceFrom(start)
state := p.cloneState()
actVal, err := act.run(p)
if err != nil {
p.addErrAt(err, start.position)
p.addErrAt(err, start.position, []string{})
}
p.restoreState(state)
val = actVal
}
if ok && p.debug {
@ -1381,10 +1708,14 @@ func (p *parser) parseAndCodeExpr(and *andCodeExpr) (interface{}, bool) {
defer p.out(p.in("parseAndCodeExpr"))
}
state := p.cloneState()
ok, err := and.run(p)
if err != nil {
p.addErr(err)
}
p.restoreState(state)
return nil, ok
}
@ -1394,10 +1725,13 @@ func (p *parser) parseAndExpr(and *andExpr) (interface{}, bool) {
}
pt := p.pt
state := p.cloneState()
p.pushV()
_, ok := p.parseExpr(and.expr)
p.popV()
p.restoreState(state)
p.restore(pt)
return nil, ok
}
@ -1406,12 +1740,15 @@ func (p *parser) parseAnyMatcher(any *anyMatcher) (interface{}, bool) {
defer p.out(p.in("parseAnyMatcher"))
}
if p.pt.rn != utf8.RuneError {
start := p.pt
p.read()
return p.sliceFrom(start), true
if p.pt.rn == utf8.RuneError && p.pt.w == 0 {
// EOF - see utf8.DecodeRune
p.failAt(false, p.pt.position, ".")
return nil, false
}
return nil, false
start := p.pt
p.read()
p.failAt(true, start.position, ".")
return p.sliceFrom(start), true
}
func (p *parser) parseCharClassMatcher(chr *charClassMatcher) (interface{}, bool) {
@ -1420,11 +1757,14 @@ func (p *parser) parseCharClassMatcher(chr *charClassMatcher) (interface{}, bool
}
cur := p.pt.rn
start := p.pt
// can't match EOF
if cur == utf8.RuneError {
if cur == utf8.RuneError && p.pt.w == 0 { // see utf8.DecodeRune
p.failAt(false, start.position, chr.val)
return nil, false
}
start := p.pt
if chr.ignoreCase {
cur = unicode.ToLower(cur)
}
@ -1433,9 +1773,11 @@ func (p *parser) parseCharClassMatcher(chr *charClassMatcher) (interface{}, bool
for _, rn := range chr.chars {
if rn == cur {
if chr.inverted {
p.failAt(false, start.position, chr.val)
return nil, false
}
p.read()
p.failAt(true, start.position, chr.val)
return p.sliceFrom(start), true
}
}
@ -1444,9 +1786,11 @@ func (p *parser) parseCharClassMatcher(chr *charClassMatcher) (interface{}, bool
for i := 0; i < len(chr.ranges); i += 2 {
if cur >= chr.ranges[i] && cur <= chr.ranges[i+1] {
if chr.inverted {
p.failAt(false, start.position, chr.val)
return nil, false
}
p.read()
p.failAt(true, start.position, chr.val)
return p.sliceFrom(start), true
}
}
@ -1455,33 +1799,60 @@ func (p *parser) parseCharClassMatcher(chr *charClassMatcher) (interface{}, bool
for _, cl := range chr.classes {
if unicode.Is(cl, cur) {
if chr.inverted {
p.failAt(false, start.position, chr.val)
return nil, false
}
p.read()
p.failAt(true, start.position, chr.val)
return p.sliceFrom(start), true
}
}
if chr.inverted {
p.read()
p.failAt(true, start.position, chr.val)
return p.sliceFrom(start), true
}
p.failAt(false, start.position, chr.val)
return nil, false
}
func (p *parser) incChoiceAltCnt(ch *choiceExpr, altI int) {
choiceIdent := fmt.Sprintf("%s %d:%d", p.rstack[len(p.rstack)-1].name, ch.pos.line, ch.pos.col)
m := p.ChoiceAltCnt[choiceIdent]
if m == nil {
m = make(map[string]int)
p.ChoiceAltCnt[choiceIdent] = m
}
// We increment altI by 1, so the keys do not start at 0
alt := strconv.Itoa(altI + 1)
if altI == choiceNoMatch {
alt = p.choiceNoMatch
}
m[alt]++
}
func (p *parser) parseChoiceExpr(ch *choiceExpr) (interface{}, bool) {
if p.debug {
defer p.out(p.in("parseChoiceExpr"))
}
for _, alt := range ch.alternatives {
for altI, alt := range ch.alternatives {
// dummy assignment to prevent compile error if optimized
_ = altI
state := p.cloneState()
p.pushV()
val, ok := p.parseExpr(alt)
p.popV()
if ok {
p.incChoiceAltCnt(ch, altI)
return val, ok
}
p.restoreState(state)
}
p.incChoiceAltCnt(ch, choiceNoMatch)
return nil, false
}
@ -1505,6 +1876,11 @@ func (p *parser) parseLitMatcher(lit *litMatcher) (interface{}, bool) {
defer p.out(p.in("parseLitMatcher"))
}
ignoreCase := ""
if lit.ignoreCase {
ignoreCase = "i"
}
val := fmt.Sprintf("%q%s", lit.val, ignoreCase)
start := p.pt
for _, want := range lit.val {
cur := p.pt.rn
@ -1512,11 +1888,13 @@ func (p *parser) parseLitMatcher(lit *litMatcher) (interface{}, bool) {
cur = unicode.ToLower(cur)
}
if cur != want {
p.failAt(false, start.position, val)
p.restore(start)
return nil, false
}
p.read()
}
p.failAt(true, start.position, val)
return p.sliceFrom(start), true
}
@ -1525,10 +1903,14 @@ func (p *parser) parseNotCodeExpr(not *notCodeExpr) (interface{}, bool) {
defer p.out(p.in("parseNotCodeExpr"))
}
state := p.cloneState()
ok, err := not.run(p)
if err != nil {
p.addErr(err)
}
p.restoreState(state)
return nil, !ok
}
@ -1538,10 +1920,15 @@ func (p *parser) parseNotExpr(not *notExpr) (interface{}, bool) {
}
pt := p.pt
state := p.cloneState()
p.pushV()
p.maxFailInvertExpected = !p.maxFailInvertExpected
_, ok := p.parseExpr(not.expr)
p.maxFailInvertExpected = !p.maxFailInvertExpected
p.popV()
p.restoreState(state)
p.restore(pt)
return nil, !ok
}
@ -1567,6 +1954,18 @@ func (p *parser) parseOneOrMoreExpr(expr *oneOrMoreExpr) (interface{}, bool) {
}
}
func (p *parser) parseRecoveryExpr(recover *recoveryExpr) (interface{}, bool) {
if p.debug {
defer p.out(p.in("parseRecoveryExpr (" + strings.Join(recover.failureLabel, ",") + ")"))
}
p.pushRecovery(recover.failureLabel, recover.recoverExpr)
val, ok := p.parseExpr(recover.expr)
p.popRecovery()
return val, ok
}
func (p *parser) parseRuleRefExpr(ref *ruleRefExpr) (interface{}, bool) {
if p.debug {
defer p.out(p.in("parseRuleRefExpr " + ref.name))
@ -1589,12 +1988,14 @@ func (p *parser) parseSeqExpr(seq *seqExpr) (interface{}, bool) {
defer p.out(p.in("parseSeqExpr"))
}
var vals []interface{}
vals := make([]interface{}, 0, len(seq.exprs))
pt := p.pt
state := p.cloneState()
for _, expr := range seq.exprs {
val, ok := p.parseExpr(expr)
if !ok {
p.restoreState(state)
p.restore(pt)
return nil, false
}
@ -1603,6 +2004,34 @@ func (p *parser) parseSeqExpr(seq *seqExpr) (interface{}, bool) {
return vals, true
}
func (p *parser) parseStateCodeExpr(state *stateCodeExpr) (interface{}, bool) {
if p.debug {
defer p.out(p.in("parseStateCodeExpr"))
}
err := state.run(p)
if err != nil {
p.addErr(err)
}
return nil, true
}
func (p *parser) parseThrowExpr(expr *throwExpr) (interface{}, bool) {
if p.debug {
defer p.out(p.in("parseThrowExpr"))
}
for i := len(p.recoveryStack) - 1; i >= 0; i-- {
if recoverExpr, ok := p.recoveryStack[i][expr.label]; ok {
if val, ok := p.parseExpr(recoverExpr); ok {
return val, ok
}
}
}
return nil, false
}
func (p *parser) parseZeroOrMoreExpr(expr *zeroOrMoreExpr) (interface{}, bool) {
if p.debug {
defer p.out(p.in("parseZeroOrMoreExpr"))
@ -1632,18 +2061,3 @@ func (p *parser) parseZeroOrOneExpr(expr *zeroOrOneExpr) (interface{}, bool) {
// whether it matched or not, consider it a match
return val, true
}
func rangeTable(class string) *unicode.RangeTable {
if rt, ok := unicode.Categories[class]; ok {
return rt
}
if rt, ok := unicode.Properties[class]; ok {
return rt
}
if rt, ok := unicode.Scripts[class]; ok {
return rt
}
// cannot happen
panic(fmt.Sprintf("invalid Unicode class: %s", class))
}

Loading…
Cancel
Save