buildkit/frontend/dockerfile/parser/parser.go

348 lines
9.7 KiB
Go
Raw Normal View History

// Package parser implements a parser and parse tree dumper for Dockerfiles.
package parser
import (
"bufio"
"bytes"
"fmt"
"io"
"regexp"
"strconv"
"strings"
"unicode"
"github.com/moby/buildkit/frontend/dockerfile/command"
"github.com/pkg/errors"
)
// Node is a structure used to represent a parse tree.
//
// In the node there are three fields, Value, Next, and Children. Value is the
// current token's string value. Next is always the next non-child token, and
// children contains all the children. Here's an example:
//
// (value next (child child-next child-next-next) next-next)
//
// This data structure is frankly pretty lousy for handling complex languages,
// but lucky for us the Dockerfile isn't very complicated. This structure
// works a little more effectively than a "proper" parse tree for our needs.
//
type Node struct {
Value string // actual content
Next *Node // the next item in the current sexp
Children []*Node // the children of this sexp
Attributes map[string]bool // special attributes for this node
Original string // original line used before parsing
Flags []string // only top Node should have this set
StartLine int // the line in the original dockerfile where the node begins
EndLine int // the line in the original dockerfile where the node ends
}
// Dump dumps the AST defined by `node` as a list of sexps.
// Returns a string suitable for printing.
func (node *Node) Dump() string {
str := ""
str += node.Value
if len(node.Flags) > 0 {
str += fmt.Sprintf(" %q", node.Flags)
}
for _, n := range node.Children {
str += "(" + n.Dump() + ")\n"
}
for n := node.Next; n != nil; n = n.Next {
if len(n.Children) > 0 {
str += " " + n.Dump()
} else {
str += " " + strconv.Quote(n.Value)
}
}
return strings.TrimSpace(str)
}
func (node *Node) lines(start, end int) {
node.StartLine = start
node.EndLine = end
}
// AddChild adds a new child node, and updates line information
func (node *Node) AddChild(child *Node, startLine, endLine int) {
child.lines(startLine, endLine)
if node.StartLine < 0 {
node.StartLine = startLine
}
node.EndLine = endLine
node.Children = append(node.Children, child)
}
var (
dispatch map[string]func(string, *directives) (*Node, map[string]bool, error)
reWhitespace = regexp.MustCompile(`[\t\v\f\r ]+`)
reDirectives = regexp.MustCompile(`^#\s*([a-zA-Z][a-zA-Z0-9]*)\s*=\s*(.+?)\s*$`)
reComment = regexp.MustCompile(`^#.*$`)
)
// DefaultEscapeToken is the default escape token
const DefaultEscapeToken = '\\'
var validDirectives = map[string]struct{}{
"escape": {},
"syntax": {},
}
// directive is the structure used during a build run to hold the state of
// parsing directives.
type directives struct {
escapeToken rune // Current escape token
lineContinuationRegex *regexp.Regexp // Current line continuation regex
done bool // Whether we are done looking for directives
seen map[string]struct{} // Whether the escape directive has been seen
}
// setEscapeToken sets the default token for escaping characters in a Dockerfile.
func (d *directives) setEscapeToken(s string) error {
if s != "`" && s != "\\" {
return errors.Errorf("invalid escape token '%s' does not match ` or \\", s)
}
d.escapeToken = rune(s[0])
d.lineContinuationRegex = regexp.MustCompile(`\` + s + `[ \t]*$`)
return nil
}
// possibleParserDirective looks for parser directives, eg '# escapeToken=<char>'.
// Parser directives must precede any builder instruction or other comments,
// and cannot be repeated.
func (d *directives) possibleParserDirective(line string) error {
if d.done {
return nil
}
match := reDirectives.FindStringSubmatch(line)
if len(match) == 0 {
d.done = true
return nil
}
k := strings.ToLower(match[1])
_, ok := validDirectives[k]
if !ok {
d.done = true
return nil
}
if _, ok := d.seen[k]; ok {
return errors.Errorf("only one %s parser directive can be used", k)
}
d.seen[k] = struct{}{}
if k == "escape" {
return d.setEscapeToken(match[2])
}
return nil
}
// newDefaultDirectives returns a new directives structure with the default escapeToken token
func newDefaultDirectives() *directives {
d := &directives{
seen: map[string]struct{}{},
}
d.setEscapeToken(string(DefaultEscapeToken))
return d
}
func init() {
// Dispatch Table. see line_parsers.go for the parse functions.
// The command is parsed and mapped to the line parser. The line parser
// receives the arguments but not the command, and returns an AST after
// reformulating the arguments according to the rules in the parser
// functions. Errors are propagated up by Parse() and the resulting AST can
// be incorporated directly into the existing AST as a next.
dispatch = map[string]func(string, *directives) (*Node, map[string]bool, error){
Add support for user-defined healthchecks This PR adds support for user-defined health-check probes for Docker containers. It adds a `HEALTHCHECK` instruction to the Dockerfile syntax plus some corresponding "docker run" options. It can be used with a restart policy to automatically restart a container if the check fails. The `HEALTHCHECK` instruction has two forms: * `HEALTHCHECK [OPTIONS] CMD command` (check container health by running a command inside the container) * `HEALTHCHECK NONE` (disable any healthcheck inherited from the base image) The `HEALTHCHECK` instruction tells Docker how to test a container to check that it is still working. This can detect cases such as a web server that is stuck in an infinite loop and unable to handle new connections, even though the server process is still running. When a container has a healthcheck specified, it has a _health status_ in addition to its normal status. This status is initially `starting`. Whenever a health check passes, it becomes `healthy` (whatever state it was previously in). After a certain number of consecutive failures, it becomes `unhealthy`. The options that can appear before `CMD` are: * `--interval=DURATION` (default: `30s`) * `--timeout=DURATION` (default: `30s`) * `--retries=N` (default: `1`) The health check will first run **interval** seconds after the container is started, and then again **interval** seconds after each previous check completes. If a single run of the check takes longer than **timeout** seconds then the check is considered to have failed. It takes **retries** consecutive failures of the health check for the container to be considered `unhealthy`. There can only be one `HEALTHCHECK` instruction in a Dockerfile. If you list more than one then only the last `HEALTHCHECK` will take effect. The command after the `CMD` keyword can be either a shell command (e.g. `HEALTHCHECK CMD /bin/check-running`) or an _exec_ array (as with other Dockerfile commands; see e.g. `ENTRYPOINT` for details). The command's exit status indicates the health status of the container. The possible values are: - 0: success - the container is healthy and ready for use - 1: unhealthy - the container is not working correctly - 2: starting - the container is not ready for use yet, but is working correctly If the probe returns 2 ("starting") when the container has already moved out of the "starting" state then it is treated as "unhealthy" instead. For example, to check every five minutes or so that a web-server is able to serve the site's main page within three seconds: HEALTHCHECK --interval=5m --timeout=3s \ CMD curl -f http://localhost/ || exit 1 To help debug failing probes, any output text (UTF-8 encoded) that the command writes on stdout or stderr will be stored in the health status and can be queried with `docker inspect`. Such output should be kept short (only the first 4096 bytes are stored currently). When the health status of a container changes, a `health_status` event is generated with the new status. The health status is also displayed in the `docker ps` output. Signed-off-by: Thomas Leonard <thomas.leonard@docker.com> Signed-off-by: Sebastiaan van Stijn <github@gone.nl> rewritten from github.com/moby/moby b6c7becbfe1d76b1250f6d8e991e645e13808a9c
2016-04-18 09:48:13 +00:00
command.Add: parseMaybeJSONToList,
command.Arg: parseNameOrNameVal,
Add support for user-defined healthchecks This PR adds support for user-defined health-check probes for Docker containers. It adds a `HEALTHCHECK` instruction to the Dockerfile syntax plus some corresponding "docker run" options. It can be used with a restart policy to automatically restart a container if the check fails. The `HEALTHCHECK` instruction has two forms: * `HEALTHCHECK [OPTIONS] CMD command` (check container health by running a command inside the container) * `HEALTHCHECK NONE` (disable any healthcheck inherited from the base image) The `HEALTHCHECK` instruction tells Docker how to test a container to check that it is still working. This can detect cases such as a web server that is stuck in an infinite loop and unable to handle new connections, even though the server process is still running. When a container has a healthcheck specified, it has a _health status_ in addition to its normal status. This status is initially `starting`. Whenever a health check passes, it becomes `healthy` (whatever state it was previously in). After a certain number of consecutive failures, it becomes `unhealthy`. The options that can appear before `CMD` are: * `--interval=DURATION` (default: `30s`) * `--timeout=DURATION` (default: `30s`) * `--retries=N` (default: `1`) The health check will first run **interval** seconds after the container is started, and then again **interval** seconds after each previous check completes. If a single run of the check takes longer than **timeout** seconds then the check is considered to have failed. It takes **retries** consecutive failures of the health check for the container to be considered `unhealthy`. There can only be one `HEALTHCHECK` instruction in a Dockerfile. If you list more than one then only the last `HEALTHCHECK` will take effect. The command after the `CMD` keyword can be either a shell command (e.g. `HEALTHCHECK CMD /bin/check-running`) or an _exec_ array (as with other Dockerfile commands; see e.g. `ENTRYPOINT` for details). The command's exit status indicates the health status of the container. The possible values are: - 0: success - the container is healthy and ready for use - 1: unhealthy - the container is not working correctly - 2: starting - the container is not ready for use yet, but is working correctly If the probe returns 2 ("starting") when the container has already moved out of the "starting" state then it is treated as "unhealthy" instead. For example, to check every five minutes or so that a web-server is able to serve the site's main page within three seconds: HEALTHCHECK --interval=5m --timeout=3s \ CMD curl -f http://localhost/ || exit 1 To help debug failing probes, any output text (UTF-8 encoded) that the command writes on stdout or stderr will be stored in the health status and can be queried with `docker inspect`. Such output should be kept short (only the first 4096 bytes are stored currently). When the health status of a container changes, a `health_status` event is generated with the new status. The health status is also displayed in the `docker ps` output. Signed-off-by: Thomas Leonard <thomas.leonard@docker.com> Signed-off-by: Sebastiaan van Stijn <github@gone.nl> rewritten from github.com/moby/moby b6c7becbfe1d76b1250f6d8e991e645e13808a9c
2016-04-18 09:48:13 +00:00
command.Cmd: parseMaybeJSON,
command.Copy: parseMaybeJSONToList,
Add support for user-defined healthchecks This PR adds support for user-defined health-check probes for Docker containers. It adds a `HEALTHCHECK` instruction to the Dockerfile syntax plus some corresponding "docker run" options. It can be used with a restart policy to automatically restart a container if the check fails. The `HEALTHCHECK` instruction has two forms: * `HEALTHCHECK [OPTIONS] CMD command` (check container health by running a command inside the container) * `HEALTHCHECK NONE` (disable any healthcheck inherited from the base image) The `HEALTHCHECK` instruction tells Docker how to test a container to check that it is still working. This can detect cases such as a web server that is stuck in an infinite loop and unable to handle new connections, even though the server process is still running. When a container has a healthcheck specified, it has a _health status_ in addition to its normal status. This status is initially `starting`. Whenever a health check passes, it becomes `healthy` (whatever state it was previously in). After a certain number of consecutive failures, it becomes `unhealthy`. The options that can appear before `CMD` are: * `--interval=DURATION` (default: `30s`) * `--timeout=DURATION` (default: `30s`) * `--retries=N` (default: `1`) The health check will first run **interval** seconds after the container is started, and then again **interval** seconds after each previous check completes. If a single run of the check takes longer than **timeout** seconds then the check is considered to have failed. It takes **retries** consecutive failures of the health check for the container to be considered `unhealthy`. There can only be one `HEALTHCHECK` instruction in a Dockerfile. If you list more than one then only the last `HEALTHCHECK` will take effect. The command after the `CMD` keyword can be either a shell command (e.g. `HEALTHCHECK CMD /bin/check-running`) or an _exec_ array (as with other Dockerfile commands; see e.g. `ENTRYPOINT` for details). The command's exit status indicates the health status of the container. The possible values are: - 0: success - the container is healthy and ready for use - 1: unhealthy - the container is not working correctly - 2: starting - the container is not ready for use yet, but is working correctly If the probe returns 2 ("starting") when the container has already moved out of the "starting" state then it is treated as "unhealthy" instead. For example, to check every five minutes or so that a web-server is able to serve the site's main page within three seconds: HEALTHCHECK --interval=5m --timeout=3s \ CMD curl -f http://localhost/ || exit 1 To help debug failing probes, any output text (UTF-8 encoded) that the command writes on stdout or stderr will be stored in the health status and can be queried with `docker inspect`. Such output should be kept short (only the first 4096 bytes are stored currently). When the health status of a container changes, a `health_status` event is generated with the new status. The health status is also displayed in the `docker ps` output. Signed-off-by: Thomas Leonard <thomas.leonard@docker.com> Signed-off-by: Sebastiaan van Stijn <github@gone.nl> rewritten from github.com/moby/moby b6c7becbfe1d76b1250f6d8e991e645e13808a9c
2016-04-18 09:48:13 +00:00
command.Entrypoint: parseMaybeJSON,
command.Env: parseEnv,
Add support for user-defined healthchecks This PR adds support for user-defined health-check probes for Docker containers. It adds a `HEALTHCHECK` instruction to the Dockerfile syntax plus some corresponding "docker run" options. It can be used with a restart policy to automatically restart a container if the check fails. The `HEALTHCHECK` instruction has two forms: * `HEALTHCHECK [OPTIONS] CMD command` (check container health by running a command inside the container) * `HEALTHCHECK NONE` (disable any healthcheck inherited from the base image) The `HEALTHCHECK` instruction tells Docker how to test a container to check that it is still working. This can detect cases such as a web server that is stuck in an infinite loop and unable to handle new connections, even though the server process is still running. When a container has a healthcheck specified, it has a _health status_ in addition to its normal status. This status is initially `starting`. Whenever a health check passes, it becomes `healthy` (whatever state it was previously in). After a certain number of consecutive failures, it becomes `unhealthy`. The options that can appear before `CMD` are: * `--interval=DURATION` (default: `30s`) * `--timeout=DURATION` (default: `30s`) * `--retries=N` (default: `1`) The health check will first run **interval** seconds after the container is started, and then again **interval** seconds after each previous check completes. If a single run of the check takes longer than **timeout** seconds then the check is considered to have failed. It takes **retries** consecutive failures of the health check for the container to be considered `unhealthy`. There can only be one `HEALTHCHECK` instruction in a Dockerfile. If you list more than one then only the last `HEALTHCHECK` will take effect. The command after the `CMD` keyword can be either a shell command (e.g. `HEALTHCHECK CMD /bin/check-running`) or an _exec_ array (as with other Dockerfile commands; see e.g. `ENTRYPOINT` for details). The command's exit status indicates the health status of the container. The possible values are: - 0: success - the container is healthy and ready for use - 1: unhealthy - the container is not working correctly - 2: starting - the container is not ready for use yet, but is working correctly If the probe returns 2 ("starting") when the container has already moved out of the "starting" state then it is treated as "unhealthy" instead. For example, to check every five minutes or so that a web-server is able to serve the site's main page within three seconds: HEALTHCHECK --interval=5m --timeout=3s \ CMD curl -f http://localhost/ || exit 1 To help debug failing probes, any output text (UTF-8 encoded) that the command writes on stdout or stderr will be stored in the health status and can be queried with `docker inspect`. Such output should be kept short (only the first 4096 bytes are stored currently). When the health status of a container changes, a `health_status` event is generated with the new status. The health status is also displayed in the `docker ps` output. Signed-off-by: Thomas Leonard <thomas.leonard@docker.com> Signed-off-by: Sebastiaan van Stijn <github@gone.nl> rewritten from github.com/moby/moby b6c7becbfe1d76b1250f6d8e991e645e13808a9c
2016-04-18 09:48:13 +00:00
command.Expose: parseStringsWhitespaceDelimited,
command.From: parseStringsWhitespaceDelimited,
Add support for user-defined healthchecks This PR adds support for user-defined health-check probes for Docker containers. It adds a `HEALTHCHECK` instruction to the Dockerfile syntax plus some corresponding "docker run" options. It can be used with a restart policy to automatically restart a container if the check fails. The `HEALTHCHECK` instruction has two forms: * `HEALTHCHECK [OPTIONS] CMD command` (check container health by running a command inside the container) * `HEALTHCHECK NONE` (disable any healthcheck inherited from the base image) The `HEALTHCHECK` instruction tells Docker how to test a container to check that it is still working. This can detect cases such as a web server that is stuck in an infinite loop and unable to handle new connections, even though the server process is still running. When a container has a healthcheck specified, it has a _health status_ in addition to its normal status. This status is initially `starting`. Whenever a health check passes, it becomes `healthy` (whatever state it was previously in). After a certain number of consecutive failures, it becomes `unhealthy`. The options that can appear before `CMD` are: * `--interval=DURATION` (default: `30s`) * `--timeout=DURATION` (default: `30s`) * `--retries=N` (default: `1`) The health check will first run **interval** seconds after the container is started, and then again **interval** seconds after each previous check completes. If a single run of the check takes longer than **timeout** seconds then the check is considered to have failed. It takes **retries** consecutive failures of the health check for the container to be considered `unhealthy`. There can only be one `HEALTHCHECK` instruction in a Dockerfile. If you list more than one then only the last `HEALTHCHECK` will take effect. The command after the `CMD` keyword can be either a shell command (e.g. `HEALTHCHECK CMD /bin/check-running`) or an _exec_ array (as with other Dockerfile commands; see e.g. `ENTRYPOINT` for details). The command's exit status indicates the health status of the container. The possible values are: - 0: success - the container is healthy and ready for use - 1: unhealthy - the container is not working correctly - 2: starting - the container is not ready for use yet, but is working correctly If the probe returns 2 ("starting") when the container has already moved out of the "starting" state then it is treated as "unhealthy" instead. For example, to check every five minutes or so that a web-server is able to serve the site's main page within three seconds: HEALTHCHECK --interval=5m --timeout=3s \ CMD curl -f http://localhost/ || exit 1 To help debug failing probes, any output text (UTF-8 encoded) that the command writes on stdout or stderr will be stored in the health status and can be queried with `docker inspect`. Such output should be kept short (only the first 4096 bytes are stored currently). When the health status of a container changes, a `health_status` event is generated with the new status. The health status is also displayed in the `docker ps` output. Signed-off-by: Thomas Leonard <thomas.leonard@docker.com> Signed-off-by: Sebastiaan van Stijn <github@gone.nl> rewritten from github.com/moby/moby b6c7becbfe1d76b1250f6d8e991e645e13808a9c
2016-04-18 09:48:13 +00:00
command.Healthcheck: parseHealthConfig,
command.Label: parseLabel,
command.Maintainer: parseString,
command.Onbuild: parseSubCommand,
command.Run: parseMaybeJSON,
command.Shell: parseMaybeJSON,
command.StopSignal: parseString,
command.User: parseString,
command.Volume: parseMaybeJSONToList,
command.Workdir: parseString,
}
}
// newNodeFromLine splits the line into parts, and dispatches to a function
// based on the command and command arguments. A Node is created from the
// result of the dispatch.
func newNodeFromLine(line string, d *directives) (*Node, error) {
cmd, flags, args, err := splitCommand(line)
if err != nil {
return nil, err
}
fn := dispatch[cmd]
// Ignore invalid Dockerfile instructions
if fn == nil {
fn = parseIgnore
}
next, attrs, err := fn(args, d)
if err != nil {
return nil, err
}
return &Node{
Value: cmd,
Original: line,
Flags: flags,
Next: next,
Attributes: attrs,
}, nil
}
// Result is the result of parsing a Dockerfile
type Result struct {
AST *Node
EscapeToken rune
Warnings []string
}
// PrintWarnings to the writer
func (r *Result) PrintWarnings(out io.Writer) {
if len(r.Warnings) == 0 {
return
}
fmt.Fprintf(out, strings.Join(r.Warnings, "\n")+"\n")
}
// Parse reads lines from a Reader, parses the lines into an AST and returns
// the AST and escape token
func Parse(rwc io.Reader) (*Result, error) {
d := newDefaultDirectives()
currentLine := 0
root := &Node{StartLine: -1}
scanner := bufio.NewScanner(rwc)
warnings := []string{}
var err error
for scanner.Scan() {
bytesRead := scanner.Bytes()
if currentLine == 0 {
// First line, strip the byte-order-marker if present
bytesRead = bytes.TrimPrefix(bytesRead, utf8bom)
}
bytesRead, err = processLine(d, bytesRead, true)
if err != nil {
return nil, withLocation(err, currentLine, 0)
}
currentLine++
startLine := currentLine
line, isEndOfLine := trimContinuationCharacter(string(bytesRead), d)
if isEndOfLine && line == "" {
continue
}
var hasEmptyContinuationLine bool
for !isEndOfLine && scanner.Scan() {
bytesRead, err := processLine(d, scanner.Bytes(), false)
if err != nil {
return nil, withLocation(err, currentLine, 0)
}
currentLine++
if isComment(scanner.Bytes()) {
// original line was a comment (processLine strips comments)
continue
}
if isEmptyContinuationLine(bytesRead) {
hasEmptyContinuationLine = true
continue
}
continuationLine := string(bytesRead)
continuationLine, isEndOfLine = trimContinuationCharacter(continuationLine, d)
line += continuationLine
}
if hasEmptyContinuationLine {
warnings = append(warnings, "[WARNING]: Empty continuation line found in:\n "+line)
}
child, err := newNodeFromLine(line, d)
if err != nil {
return nil, withLocation(err, startLine, currentLine)
}
root.AddChild(child, startLine, currentLine)
}
if len(warnings) > 0 {
warnings = append(warnings, "[WARNING]: Empty continuation lines will become errors in a future release.")
}
if root.StartLine < 0 {
return nil, withLocation(errors.New("file with no instructions"), currentLine, 0)
}
return &Result{
AST: root,
Warnings: warnings,
EscapeToken: d.escapeToken,
}, withLocation(handleScannerError(scanner.Err()), currentLine, 0)
}
func trimComments(src []byte) []byte {
return reComment.ReplaceAll(src, []byte{})
}
func trimWhitespace(src []byte) []byte {
return bytes.TrimLeftFunc(src, unicode.IsSpace)
}
func isComment(line []byte) bool {
return reComment.Match(trimWhitespace(line))
}
func isEmptyContinuationLine(line []byte) bool {
return len(trimWhitespace(line)) == 0
}
var utf8bom = []byte{0xEF, 0xBB, 0xBF}
func trimContinuationCharacter(line string, d *directives) (string, bool) {
if d.lineContinuationRegex.MatchString(line) {
line = d.lineContinuationRegex.ReplaceAllString(line, "")
return line, false
}
return line, true
}
// TODO: remove stripLeftWhitespace after deprecation period. It seems silly
// to preserve whitespace on continuation lines. Why is that done?
func processLine(d *directives, token []byte, stripLeftWhitespace bool) ([]byte, error) {
if stripLeftWhitespace {
token = trimWhitespace(token)
}
return trimComments(token), d.possibleParserDirective(string(token))
}
func handleScannerError(err error) error {
switch err {
case bufio.ErrTooLong:
return errors.Errorf("dockerfile line greater than max allowed size of %d", bufio.MaxScanTokenSize-1)
default:
return err
}
}