gered/tree-sitter-dockerfile
1
0
Fork 0
Code Issues Pull requests Actions Packages Projects Releases Wiki Activity

Add support for single-quoted strings and distinguish JSON strings from

Browse source 
the double-quoted strings that support variable expansion.
This commit is contained in:
Martin Jambon 2023-06-22 19:44:07 -07:00
parent fd30df68c4
commit 98fdcad2a0
7 changed files with 242 additions and 57 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
corpus/cmd.txt
View file

@ -41,7 +41,7 @@ cmd []
(source_file (source_file
(cmd_instruction (cmd_instruction
(string_array))) (json_string_array)))
================== ==================
Run with shell array Run with shell array
@ -52,8 +52,8 @@ cmd ["echo", "test"]
--- ---
(source_file (source_file
(cmd_instruction (cmd_instruction
(string_array (json_string_array
(double_quoted_string) (json_string)
(double_quoted_string)))) (json_string))))

11
corpus/entrypoint.txt
View file

@ -41,7 +41,7 @@ ENTRYPOINT []
(source_file (source_file
(entrypoint_instruction (entrypoint_instruction
(string_array))) (json_string_array)))
================== ==================
Run with shell array Run with shell array
@ -52,8 +52,7 @@ ENTRYPOINT ["echo", "test"]
--- ---
(source_file (source_file
(entrypoint_instruction (entrypoint_instruction
(string_array (json_string_array
(double_quoted_string) (json_string)
(double_quoted_string)))) (json_string))))

10
corpus/run.txt
View file

@ -41,7 +41,7 @@ run []
(source_file (source_file
(run_instruction (run_instruction
(string_array))) (json_string_array)))
================== ==================
Run with shell array Run with shell array
@ -52,10 +52,10 @@ run ["echo", "test"]
--- ---
(source_file (source_file
(run_instruction (run_instruction
(string_array (json_string_array
(double_quoted_string) (json_string)
(double_quoted_string)))) (json_string))))
================== ==================
Run with options Run with options

6
corpus/shell_instruction.txt
View file

@ -8,7 +8,7 @@ SHELL ["powershell","-command"]
(source_file (source_file
(shell_instruction (shell_instruction
(string_array (json_string_array
(double_quoted_string) (json_string)
(double_quoted_string)))) (json_string))))

146
corpus/strings.txt Normal file
View file

@ -0,0 +1,146 @@
==========================
JSON string arrays
==========================
CMD ["a", "bcd"]
CMD ["\"\\"]
CMD ["\b\f\n\r\t\u0ABC\u12Cd"]
---
(source_file
(cmd_instruction
(json_string_array
(json_string)
(json_string)))
(cmd_instruction
(json_string_array
(json_string)))
(cmd_instruction
(json_string_array
(json_string))))
==========================
Double-quoted strings
==========================
ENV KEY "value"
ENV KEY "conti\
nue"
ENV KEY "quotes: \" '"
ENV KEY "backslash: \\"
ENV KEY "not escape sequences: \a \n"
ENV KEY "expansions: $FOO ${BAR}"
---
(source_file
(env_instruction
(env_pair
(unquoted_string)
(double_quoted_string)))
(env_instruction
(env_pair
(unquoted_string)
(double_quoted_string
(line_continuation))))
(env_instruction
(env_pair
(unquoted_string)
(double_quoted_string
(double_quoted_escape_sequence))))
(env_instruction
(env_pair
(unquoted_string)
(double_quoted_string
(double_quoted_escape_sequence))))
(env_instruction
(env_pair
(unquoted_string)
(double_quoted_string
(double_quoted_escape_sequence)
(double_quoted_escape_sequence))))
(env_instruction
(env_pair
(unquoted_string)
(double_quoted_string
(expansion
(variable))
(expansion
(variable))))))
==========================
Single-quoted strings
==========================
ENV KEY 'value'
ENV KEY 'conti\
nue'
ENV KEY 'quotes: \' \''
ENV KEY 'backslash: \\'
ENV KEY 'not escape sequences: \a \n'
---
(source_file
(env_instruction
(env_pair
(unquoted_string)
(single_quoted_string)))
(env_instruction
(env_pair
(unquoted_string)
(single_quoted_string
(line_continuation))))
(env_instruction
(env_pair
(unquoted_string)
(single_quoted_string
(single_quoted_escape_sequence)
(single_quoted_escape_sequence))))
(env_instruction
(env_pair
(unquoted_string)
(single_quoted_string
(single_quoted_escape_sequence))))
(env_instruction
(env_pair
(unquoted_string)
(single_quoted_string
(single_quoted_escape_sequence)
(single_quoted_escape_sequence)))))
==========================
Unquoted strings
==========================
ENV KEY value
ENV KEY two\ words
ENV KEY conti\
nue
ENV KEY $FOO${BAR}baz
---
(source_file
(env_instruction
(env_pair
(unquoted_string)
(unquoted_string)))
(env_instruction
(env_pair
(unquoted_string)
(unquoted_string)))
(env_instruction
(env_pair
(unquoted_string)
(unquoted_string
(line_continuation))))
(env_instruction
(env_pair
(unquoted_string)
(unquoted_string
(expansion
(variable))
(expansion
(variable))))))

10
corpus/volume.txt
View file

@ -49,8 +49,8 @@ VOLUME ["/test/myvol"]
(source_file (source_file
(volume_instruction (volume_instruction
(string_array (json_string_array
(double_quoted_string)))) (json_string))))
================== ==================
Multiline array Multiline array
@ -63,7 +63,7 @@ VOLUME ["/test/myvol", \
(source_file (source_file
(volume_instruction (volume_instruction
(string_array (json_string_array
(double_quoted_string) (json_string)
(line_continuation) (line_continuation)
(double_quoted_string)))) (json_string))))

106
grammar.js
View file

@ -46,13 +46,13 @@ module.exports = grammar({
$.mount_param $.mount_param
) )
), ),
choice($.string_array, $.shell_command) choice($.json_string_array, $.shell_command)
), ),
cmd_instruction: ($) => cmd_instruction: ($) =>
seq( seq(
alias(/[cC][mM][dD]/, "CMD"), alias(/[cC][mM][dD]/, "CMD"),
choice($.string_array, $.shell_command) choice($.json_string_array, $.shell_command)
), ),
label_instruction: ($) => label_instruction: ($) =>
@ -93,14 +93,14 @@ module.exports = grammar({
entrypoint_instruction: ($) => entrypoint_instruction: ($) =>
seq( seq(
alias(/[eE][nN][tT][rR][yY][pP][oO][iI][nN][tT]/, "ENTRYPOINT"), alias(/[eE][nN][tT][rR][yY][pP][oO][iI][nN][tT]/, "ENTRYPOINT"),
choice($.string_array, $.shell_command) choice($.json_string_array, $.shell_command)
), ),
volume_instruction: ($) => volume_instruction: ($) =>
seq( seq(
alias(/[vV][oO][lL][uU][mM][eE]/, "VOLUME"), alias(/[vV][oO][lL][uU][mM][eE]/, "VOLUME"),
choice( choice(
$.string_array, $.json_string_array,
seq($.path, repeat(seq($._non_newline_whitespace, $.path))) seq($.path, repeat(seq($._non_newline_whitespace, $.path)))
) )
), ),
@ -144,7 +144,12 @@ module.exports = grammar({
optional( optional(
seq( seq(
token.immediate("="), token.immediate("="),
field("default", choice($.double_quoted_string, $.unquoted_string)) field("default",
choice(
$.double_quoted_string,
$.single_quoted_string,
$.unquoted_string
))
) )
) )
), ),
@ -171,7 +176,7 @@ module.exports = grammar({
), ),
shell_instruction: ($) => shell_instruction: ($) =>
seq(alias(/[sS][hH][eE][lL][lL]/, "SHELL"), $.string_array), seq(alias(/[sS][hH][eE][lL][lL]/, "SHELL"), $.json_string_array),
maintainer_instruction: ($) => maintainer_instruction: ($) =>
seq( seq(
@ -222,7 +227,12 @@ module.exports = grammar({
field("name", $._env_key), field("name", $._env_key),
token.immediate("="), token.immediate("="),
optional( optional(
field("value", choice($.double_quoted_string, $.unquoted_string)) field("value",
choice(
$.double_quoted_string,
$.single_quoted_string,
$.unquoted_string
))
) )
), ),
@ -230,7 +240,12 @@ module.exports = grammar({
seq( seq(
field("name", $._env_key), field("name", $._env_key),
token.immediate(/\s+/), token.immediate(/\s+/),
field("value", choice($.double_quoted_string, $.unquoted_string)) field("value",
choice(
$.double_quoted_string,
$.single_quoted_string,
$.unquoted_string
))
), ),
_env_key: ($) => _env_key: ($) =>
@ -242,7 +257,12 @@ module.exports = grammar({
seq( seq(
field("key", alias(/[-a-zA-Z0-9\._]+/, $.unquoted_string)), field("key", alias(/[-a-zA-Z0-9\._]+/, $.unquoted_string)),
token.immediate("="), token.immediate("="),
field("value", choice($.double_quoted_string, $.unquoted_string)) field("value",
choice(
$.double_quoted_string,
$.single_quoted_string,
$.unquoted_string
))
), ),
image_spec: ($) => image_spec: ($) =>
@ -311,15 +331,6 @@ module.exports = grammar({
repeat(choice(token.immediate(/[-a-zA-Z0-9_]+/), $._immediate_expansion)) repeat(choice(token.immediate(/[-a-zA-Z0-9_]+/), $._immediate_expansion))
), ),
string_array: ($) =>
seq(
"[",
optional(
seq($.double_quoted_string, repeat(seq(",", $.double_quoted_string)))
),
"]"
),
shell_command: ($) => shell_command: ($) =>
seq( seq(
repeat($._comment_line), repeat($._comment_line),
@ -359,41 +370,70 @@ module.exports = grammar({
_anon_comment: ($) => seq("#", /.*/), _anon_comment: ($) => seq("#", /.*/),
json_string_array: ($) =>
seq(
"[",
optional(
seq($.json_string, repeat(seq(",", $.json_string)))
),
"]"
),
// Note that JSON strings are different from the other double-quoted
// strings. They don't support $-expansions.
// Convenient reference: https://www.json.org/
json_string: ($) => /"(?:[^"\\]|\\(?:["\\/bfnrt]|u[0-9A-Fa-f]{4}))*"/,
double_quoted_string: ($) => double_quoted_string: ($) =>
seq( seq(
'"', '"',
repeat( repeat(
choice( choice(
token.immediate(/[^"\n\\\$]+/), token.immediate(/[^"\n\\\$]+/),
$.escape_sequence, $.double_quoted_escape_sequence,
$._immediate_expansion $._immediate_expansion
) )
), ),
'"' '"'
), ),
// same as double_quoted_string but without $-expansions:
single_quoted_string: ($) =>
seq(
"'",
repeat(
choice(
token.immediate(/[^'\n\\]+/),
$.single_quoted_escape_sequence
)
),
"'"
),
unquoted_string: ($) => unquoted_string: ($) =>
repeat1( repeat1(
choice( choice(
token.immediate(/[^\s\n\"\\\$]+/), token.immediate(/[^\s\n\"'\\\$]+/),
token.immediate("\\ "), token.immediate("\\ "),
$._immediate_expansion $._immediate_expansion
) )
), ),
escape_sequence: ($) => double_quoted_escape_sequence: ($) => token.immediate(
token.immediate( choice(
seq( "\\\\",
"\\", "\\\"",
choice( "\\"
/[^xuU]/, )
/\d{2,3}/, ),
/x[0-9a-fA-F]{2,}/,
/u[0-9a-fA-F]{4}/, single_quoted_escape_sequence: ($) => token.immediate(
/U[0-9a-fA-F]{8}/ choice(
) "\\\\",
) "\\'",
), "\\"
)
),
_non_newline_whitespace: ($) => /[\t ]+/, _non_newline_whitespace: ($) => /[\t ]+/,