prettier 0.12.2 → 0.12.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +49 -1
- data/CONTRIBUTING.md +1 -1
- data/README.md +34 -18
- data/node_modules/prettier/bin-prettier.js +22 -13
- data/node_modules/prettier/index.js +22 -13
- data/package.json +2 -2
- data/src/nodes.js +10 -576
- data/src/nodes/alias.js +1 -1
- data/src/nodes/args.js +88 -0
- data/src/nodes/arrays.js +1 -1
- data/src/nodes/assign.js +39 -0
- data/src/nodes/blocks.js +18 -3
- data/src/nodes/calls.js +37 -3
- data/src/nodes/case.js +1 -1
- data/src/nodes/commands.js +2 -1
- data/src/nodes/conditionals.js +118 -57
- data/src/nodes/constants.js +25 -0
- data/src/nodes/flow.js +64 -0
- data/src/nodes/hashes.js +3 -2
- data/src/nodes/hooks.js +1 -1
- data/src/nodes/ints.js +24 -0
- data/src/nodes/lambdas.js +1 -1
- data/src/nodes/loops.js +1 -1
- data/src/nodes/massign.js +70 -0
- data/src/nodes/methods.js +40 -2
- data/src/nodes/operators.js +44 -0
- data/src/nodes/params.js +14 -3
- data/src/nodes/regexp.js +1 -1
- data/src/nodes/rescue.js +1 -1
- data/src/nodes/scopes.js +61 -0
- data/src/nodes/statements.js +105 -0
- data/src/nodes/strings.js +9 -2
- data/src/{builders.js → prettier.js} +9 -2
- data/src/ripper.rb +412 -353
- data/src/utils.js +1 -1
- metadata +40 -3
@@ -0,0 +1,105 @@
|
|
1
|
+
const {
|
2
|
+
concat,
|
3
|
+
dedent,
|
4
|
+
group,
|
5
|
+
hardline,
|
6
|
+
indent,
|
7
|
+
join,
|
8
|
+
line,
|
9
|
+
literalline,
|
10
|
+
markAsRoot,
|
11
|
+
softline,
|
12
|
+
trim
|
13
|
+
} = require("../prettier");
|
14
|
+
|
15
|
+
module.exports = {
|
16
|
+
"@__end__": (path, _opts, _print) => {
|
17
|
+
const { body } = path.getValue();
|
18
|
+
return concat([trim, "__END__", literalline, body]);
|
19
|
+
},
|
20
|
+
bodystmt: (path, opts, print) => {
|
21
|
+
const [_statements, rescue, elseClause, ensure] = path.getValue().body;
|
22
|
+
const parts = [path.call(print, "body", 0)];
|
23
|
+
|
24
|
+
if (rescue) {
|
25
|
+
parts.push(dedent(concat([hardline, path.call(print, "body", 1)])));
|
26
|
+
}
|
27
|
+
|
28
|
+
if (elseClause) {
|
29
|
+
// Before Ruby 2.6, this piece of bodystmt was an explicit "else" node
|
30
|
+
const stmts =
|
31
|
+
elseClause.type === "else"
|
32
|
+
? path.call(print, "body", 2, "body", 0)
|
33
|
+
: path.call(print, "body", 2);
|
34
|
+
|
35
|
+
parts.push(concat([dedent(concat([hardline, "else"])), hardline, stmts]));
|
36
|
+
}
|
37
|
+
|
38
|
+
if (ensure) {
|
39
|
+
parts.push(dedent(concat([hardline, path.call(print, "body", 3)])));
|
40
|
+
}
|
41
|
+
|
42
|
+
return group(concat(parts));
|
43
|
+
},
|
44
|
+
embdoc: (path, _opts, _print) => concat([trim, path.getValue().body]),
|
45
|
+
paren: (path, opts, print) => {
|
46
|
+
if (!path.getValue().body[0]) {
|
47
|
+
return "()";
|
48
|
+
}
|
49
|
+
|
50
|
+
let content = path.call(print, "body", 0);
|
51
|
+
|
52
|
+
if (
|
53
|
+
["args", "args_add_star", "args_add_block"].includes(
|
54
|
+
path.getValue().body[0].type
|
55
|
+
)
|
56
|
+
) {
|
57
|
+
content = join(concat([",", line]), content);
|
58
|
+
}
|
59
|
+
|
60
|
+
return group(
|
61
|
+
concat([
|
62
|
+
"(",
|
63
|
+
indent(concat([softline, content])),
|
64
|
+
concat([softline, ")"])
|
65
|
+
])
|
66
|
+
);
|
67
|
+
},
|
68
|
+
program: (path, opts, print) =>
|
69
|
+
markAsRoot(
|
70
|
+
concat([join(literalline, path.map(print, "body")), literalline])
|
71
|
+
),
|
72
|
+
stmts: (path, opts, print) => {
|
73
|
+
const stmts = path.getValue().body;
|
74
|
+
const parts = [];
|
75
|
+
let lineNo = null;
|
76
|
+
|
77
|
+
stmts.forEach((stmt, index) => {
|
78
|
+
if (stmt.type === "void_stmt") {
|
79
|
+
return;
|
80
|
+
}
|
81
|
+
|
82
|
+
const printed = path.call(print, "body", index);
|
83
|
+
|
84
|
+
if (lineNo === null) {
|
85
|
+
parts.push(printed);
|
86
|
+
} else if (
|
87
|
+
stmt.start - lineNo > 1 ||
|
88
|
+
[stmt.type, stmts[index - 1].type].includes("access_ctrl")
|
89
|
+
) {
|
90
|
+
parts.push(hardline, hardline, printed);
|
91
|
+
} else if (
|
92
|
+
stmt.start !== lineNo ||
|
93
|
+
path.getParentNode().type !== "string_embexpr"
|
94
|
+
) {
|
95
|
+
parts.push(hardline, printed);
|
96
|
+
} else {
|
97
|
+
parts.push("; ", printed);
|
98
|
+
}
|
99
|
+
|
100
|
+
lineNo = stmt.end;
|
101
|
+
});
|
102
|
+
|
103
|
+
return concat(parts);
|
104
|
+
}
|
105
|
+
};
|
data/src/nodes/strings.js
CHANGED
@@ -6,8 +6,8 @@ const {
|
|
6
6
|
join,
|
7
7
|
literalline,
|
8
8
|
softline
|
9
|
-
} = require("../
|
10
|
-
const { concatBody, empty, makeList, surround } = require("../utils");
|
9
|
+
} = require("../prettier");
|
10
|
+
const { concatBody, empty, makeList, prefix, surround } = require("../utils");
|
11
11
|
const escapePattern = require("../escapePattern");
|
12
12
|
|
13
13
|
// If there is some part of this string that matches an escape sequence or that
|
@@ -72,6 +72,11 @@ module.exports = {
|
|
72
72
|
const quote = preferSingleQuotes ? "'" : '"';
|
73
73
|
return body.length === 2 ? concat([quote, body.slice(1), quote]) : body;
|
74
74
|
},
|
75
|
+
dyna_symbol: (path, opts, print) => {
|
76
|
+
const { quote } = path.getValue().body[0];
|
77
|
+
|
78
|
+
return concat([":", quote, concat(path.call(print, "body", 0)), quote]);
|
79
|
+
},
|
75
80
|
heredoc: (path, opts, print) => {
|
76
81
|
const { beging, ending } = path.getValue();
|
77
82
|
|
@@ -129,6 +134,8 @@ module.exports = {
|
|
129
134
|
|
130
135
|
return concat([quote].concat(parts).concat([quote]));
|
131
136
|
},
|
137
|
+
symbol: prefix(":"),
|
138
|
+
symbol_literal: concatBody,
|
132
139
|
word_add: concatBody,
|
133
140
|
word_new: empty,
|
134
141
|
xstring: makeList,
|
@@ -4,6 +4,13 @@
|
|
4
4
|
const source = process.env.RBPRETTIER ? "../node_modules/prettier" : "prettier";
|
5
5
|
|
6
6
|
// eslint-disable-next-line import/no-dynamic-require
|
7
|
-
const
|
7
|
+
const prettier = require(source);
|
8
8
|
|
9
|
-
|
9
|
+
// Just combine all the things into one big object so that we can import
|
10
|
+
// whatever we need from prettier without having to dive too deeply.
|
11
|
+
module.exports = Object.assign(
|
12
|
+
{},
|
13
|
+
prettier.doc.builders,
|
14
|
+
prettier.doc.utils,
|
15
|
+
prettier.util
|
16
|
+
);
|
data/src/ripper.rb
CHANGED
@@ -9,448 +9,503 @@ end
|
|
9
9
|
require 'json' unless defined?(JSON)
|
10
10
|
require 'ripper'
|
11
11
|
|
12
|
-
|
12
|
+
class RipperJS < Ripper
|
13
|
+
private
|
14
|
+
|
15
|
+
# Scanner events occur when the lexer hits a new token, like a keyword or an
|
16
|
+
# end. These nodes always contain just one argument which is a string
|
17
|
+
# representing the content. For the most part these can just be printed
|
18
|
+
# directly, which very few exceptions.
|
19
|
+
SCANNER_EVENTS.each do |event|
|
20
|
+
define_method(:"on_#{event}") do |body|
|
21
|
+
{ type: :"@#{event}", body: body, start: lineno, end: lineno }
|
22
|
+
end
|
23
|
+
end
|
24
|
+
|
25
|
+
# Parser events represent nodes in the ripper abstract syntax tree. The event
|
26
|
+
# is reported after the children of the node have already been built.
|
27
|
+
PARSER_EVENTS.each do |event|
|
28
|
+
define_method(:"on_#{event}") do |*body|
|
29
|
+
min = body.map { |part| part.is_a?(Hash) ? part[:start] : lineno }.min
|
30
|
+
{ type: event, body: body, start: min || lineno, end: lineno }
|
31
|
+
end
|
32
|
+
end
|
33
|
+
|
13
34
|
# Some nodes are lists that come back from the parser. They always start with
|
14
|
-
# a
|
15
|
-
# in the list is a
|
16
|
-
# into one node with an array body.
|
17
|
-
|
18
|
-
|
19
|
-
|
20
|
-
|
21
|
-
|
22
|
-
|
23
|
-
|
24
|
-
|
25
|
-
|
26
|
-
|
27
|
-
|
28
|
-
|
29
|
-
|
30
|
-
|
31
|
-
|
32
|
-
|
33
|
-
|
34
|
-
|
35
|
-
|
36
|
-
|
37
|
-
|
38
|
-
|
39
|
-
|
40
|
-
|
41
|
-
|
42
|
-
|
43
|
-
|
44
|
-
|
35
|
+
# a `*_new` node (or in the case of string, `*_content`) and each additional
|
36
|
+
# node in the list is a `*_add` node. This module takes those nodes and turns
|
37
|
+
# them into one node with an array body.
|
38
|
+
#
|
39
|
+
# For example, the statement `[a, b, c]` would be parsed as:
|
40
|
+
#
|
41
|
+
# [:args_add,
|
42
|
+
# [:args_add,
|
43
|
+
# [:args_add,
|
44
|
+
# [:args_new],
|
45
|
+
# [:vcall, [:@ident, "a", [1, 1]]]
|
46
|
+
# ],
|
47
|
+
# [:vcall, [:@ident, "b", [1, 4]]]
|
48
|
+
# ],
|
49
|
+
# [:vcall, [:@ident, "c", [1, 7]]]
|
50
|
+
# ]
|
51
|
+
#
|
52
|
+
# But after this module is applied that is instead parsed as:
|
53
|
+
#
|
54
|
+
# [:args,
|
55
|
+
# [
|
56
|
+
# [:vcall, [:@ident, "a", [1, 1]]],
|
57
|
+
# [:vcall, [:@ident, "b", [1, 4]]],
|
58
|
+
# [:vcall, [:@ident, "c", [1, 7]]]
|
59
|
+
# ]
|
60
|
+
# ]
|
61
|
+
#
|
62
|
+
# This makes it a lot easier to join things with commas, and ends up resulting
|
63
|
+
# in a much flatter `prettier` tree once it has been converted. Note that
|
64
|
+
# because of this module some extra node types are added (the aggregate of
|
65
|
+
# the previous `*_add` nodes) and some nodes now have arrays in places where
|
66
|
+
# they previously had single nodes.
|
67
|
+
prepend(
|
68
|
+
Module.new do
|
69
|
+
events = %i[
|
70
|
+
args
|
71
|
+
mlhs
|
72
|
+
mrhs
|
73
|
+
qsymbols
|
74
|
+
qwords
|
75
|
+
regexp
|
76
|
+
stmts
|
77
|
+
string
|
78
|
+
symbols
|
79
|
+
words
|
80
|
+
xstring
|
81
|
+
]
|
82
|
+
|
83
|
+
private
|
84
|
+
|
85
|
+
events.each do |event|
|
86
|
+
suffix = event == :string ? 'content' : 'new'
|
87
|
+
|
88
|
+
define_method(:"on_#{event}_#{suffix}") do
|
89
|
+
{ type: event, body: [], start: lineno, end: lineno }
|
90
|
+
end
|
91
|
+
|
92
|
+
define_method(:"on_#{event}_add") do |parts, part|
|
93
|
+
parts.tap do |node|
|
94
|
+
node[:body] << part
|
95
|
+
node[:end] = lineno
|
96
|
+
end
|
45
97
|
end
|
46
98
|
end
|
47
99
|
end
|
48
|
-
|
100
|
+
)
|
49
101
|
|
50
102
|
# For most nodes, it's enough to look at the child nodes to determine the
|
51
103
|
# start of the parent node. However, for some nodes it's necessary to keep
|
52
104
|
# track of the keywords as they come in from the lexer and to modify the start
|
53
|
-
# node once we have it.
|
54
|
-
|
55
|
-
|
105
|
+
# node once we have it. We need accurate start and end lines so that we can
|
106
|
+
# embed block comments into the right kind of node.
|
107
|
+
prepend(
|
108
|
+
Module.new do
|
109
|
+
events = %i[begin else elsif ensure if rescue until while]
|
56
110
|
|
57
|
-
|
58
|
-
|
59
|
-
|
60
|
-
|
111
|
+
def initialize(*args)
|
112
|
+
super(*args)
|
113
|
+
@keywords = []
|
114
|
+
end
|
61
115
|
|
62
|
-
|
63
|
-
|
64
|
-
|
116
|
+
def self.prepended(base)
|
117
|
+
base.attr_reader :keywords
|
118
|
+
end
|
65
119
|
|
66
|
-
|
120
|
+
private
|
67
121
|
|
68
|
-
|
69
|
-
|
70
|
-
|
122
|
+
def find_start(body)
|
123
|
+
keywords[keywords.rindex { |keyword| keyword[:body] == body }][:start]
|
124
|
+
end
|
71
125
|
|
72
|
-
|
73
|
-
|
126
|
+
events.each do |event|
|
127
|
+
keyword = event.to_s
|
74
128
|
|
75
|
-
|
76
|
-
|
129
|
+
define_method(:"on_#{event}") do |*body|
|
130
|
+
super(*body).tap { |sexp| sexp.merge!(start: find_start(keyword)) }
|
131
|
+
end
|
77
132
|
end
|
78
|
-
end
|
79
133
|
|
80
|
-
|
81
|
-
|
82
|
-
|
134
|
+
def on_kw(body)
|
135
|
+
super(body).tap { |sexp| keywords << sexp }
|
136
|
+
end
|
83
137
|
|
84
|
-
|
85
|
-
|
138
|
+
def on_program(*body)
|
139
|
+
super(*body).tap { |sexp| sexp.merge!(start: 1) }
|
140
|
+
end
|
86
141
|
end
|
87
|
-
|
142
|
+
)
|
88
143
|
|
89
|
-
#
|
90
|
-
#
|
91
|
-
#
|
92
|
-
#
|
93
|
-
|
94
|
-
|
95
|
-
|
96
|
-
|
97
|
-
|
98
|
-
|
99
|
-
|
100
|
-
|
101
|
-
|
102
|
-
|
103
|
-
|
104
|
-
|
105
|
-
|
106
|
-
|
107
|
-
unless: [1],
|
108
|
-
until: [1],
|
109
|
-
when: [1],
|
110
|
-
while: [1]
|
111
|
-
}
|
112
|
-
|
113
|
-
def initialize(*args)
|
114
|
-
super(*args)
|
115
|
-
@block_comments = []
|
116
|
-
@current_embdoc = nil
|
117
|
-
end
|
144
|
+
# This layer keeps track of inline comments as they come in. Ripper itself
|
145
|
+
# doesn't attach comments to the AST, so we need to do it manually. In this
|
146
|
+
# case, inline comments are defined as any comments wherein the lexer state is
|
147
|
+
# not equal to EXPR_BEG (tracked in the BlockComments layer).
|
148
|
+
prepend(
|
149
|
+
Module.new do
|
150
|
+
# Certain events needs to steal the comments from their children in order
|
151
|
+
# for them to display properly.
|
152
|
+
events = {
|
153
|
+
aref: [:body, 1],
|
154
|
+
args_add_block: [:body, 0],
|
155
|
+
break: [:body, 0],
|
156
|
+
command: [:body, 1],
|
157
|
+
command_call: [:body, 3],
|
158
|
+
regexp_literal: [:body, 0],
|
159
|
+
string_literal: [:body, 0],
|
160
|
+
symbol_literal: [:body, 0]
|
161
|
+
}
|
118
162
|
|
119
|
-
|
120
|
-
|
121
|
-
|
163
|
+
def initialize(*args)
|
164
|
+
super(*args)
|
165
|
+
@inline_comments = []
|
166
|
+
@last_sexp = nil
|
167
|
+
end
|
168
|
+
|
169
|
+
def self.prepended(base)
|
170
|
+
base.attr_reader :inline_comments, :last_sexp
|
171
|
+
end
|
172
|
+
|
173
|
+
private
|
122
174
|
|
123
|
-
|
175
|
+
events.each do |event, path|
|
176
|
+
define_method(:"on_#{event}") do |*body|
|
177
|
+
@last_sexp =
|
178
|
+
super(*body).tap do |sexp|
|
179
|
+
comments = (sexp.dig(*path) || {}).delete(:comments)
|
180
|
+
sexp.merge!(comments: comments) if comments
|
181
|
+
end
|
182
|
+
end
|
183
|
+
end
|
124
184
|
|
125
|
-
|
126
|
-
range = sexp[:start]..sexp[:end]
|
127
|
-
comments =
|
128
|
-
block_comments.group_by { |comment| range.include?(comment[:start]) }
|
185
|
+
SPECIAL_LITERALS = %i[qsymbols qwords symbols words].freeze
|
129
186
|
|
130
|
-
|
131
|
-
|
132
|
-
|
187
|
+
# Special array literals are handled in different ways and so their
|
188
|
+
# comments need to be passed up to their parent array node.
|
189
|
+
def on_array(*body)
|
190
|
+
@last_sexp =
|
191
|
+
super(*body).tap do |sexp|
|
192
|
+
next unless SPECIAL_LITERALS.include?(body.dig(0, :type))
|
133
193
|
|
134
|
-
|
194
|
+
comments = sexp.dig(:body, 0).delete(:comments)
|
195
|
+
sexp.merge!(comments: comments) if comments
|
196
|
+
end
|
135
197
|
end
|
136
|
-
end
|
137
198
|
|
138
|
-
|
139
|
-
|
140
|
-
|
199
|
+
# Handling this specially because we want to pull the comments out of both
|
200
|
+
# child nodes.
|
201
|
+
def on_assoc_new(*body)
|
202
|
+
@last_sexp =
|
203
|
+
super(*body).tap do |sexp|
|
204
|
+
comments =
|
205
|
+
(sexp.dig(:body, 0).delete(:comments) || []) +
|
206
|
+
(sexp.dig(:body, 1).delete(:comments) || [])
|
207
|
+
|
208
|
+
sexp.merge!(comments: comments) if comments.any?
|
209
|
+
end
|
141
210
|
end
|
142
|
-
end
|
143
211
|
|
144
|
-
|
145
|
-
|
146
|
-
|
212
|
+
# Most scanner events don't stand on their own as s-expressions, but the
|
213
|
+
# CHAR scanner event is effectively just a string, so we need to track it
|
214
|
+
# as a s-expression.
|
215
|
+
def on_CHAR(body)
|
216
|
+
@last_sexp = super(body)
|
147
217
|
end
|
148
|
-
end
|
149
218
|
|
150
|
-
|
151
|
-
|
152
|
-
|
153
|
-
|
154
|
-
|
219
|
+
# We need to know exactly where the comment is, switching off the current
|
220
|
+
# lexer state. In Ruby 2.7.0-dev, that's defined as:
|
221
|
+
#
|
222
|
+
# enum lex_state_bits {
|
223
|
+
# EXPR_BEG_bit, /* ignore newline, +/- is a sign. */
|
224
|
+
# EXPR_END_bit, /* newline significant, +/- is an operator. */
|
225
|
+
# EXPR_ENDARG_bit, /* ditto, and unbound braces. */
|
226
|
+
# EXPR_ENDFN_bit, /* ditto, and unbound braces. */
|
227
|
+
# EXPR_ARG_bit, /* newline significant, +/- is an operator. */
|
228
|
+
# EXPR_CMDARG_bit, /* newline significant, +/- is an operator. */
|
229
|
+
# EXPR_MID_bit, /* newline significant, +/- is an operator. */
|
230
|
+
# EXPR_FNAME_bit, /* ignore newline, no reserved words. */
|
231
|
+
# EXPR_DOT_bit, /* right after `.' or `::', no reserved words. */
|
232
|
+
# EXPR_CLASS_bit, /* immediate after `class', no here document. */
|
233
|
+
# EXPR_LABEL_bit, /* flag bit, label is allowed. */
|
234
|
+
# EXPR_LABELED_bit, /* flag bit, just after a label. */
|
235
|
+
# EXPR_FITEM_bit, /* symbol literal as FNAME. */
|
236
|
+
# EXPR_MAX_STATE
|
237
|
+
# };
|
238
|
+
def on_comment(body)
|
239
|
+
sexp = { type: :@comment, body: body.chomp, start: lineno, end: lineno }
|
240
|
+
|
241
|
+
case RipperJS.lex_state_name(state)
|
242
|
+
when 'EXPR_END', 'EXPR_ARG|EXPR_LABELED', 'EXPR_ENDFN'
|
243
|
+
last_sexp.merge!(comments: [sexp])
|
244
|
+
when 'EXPR_CMDARG', 'EXPR_END|EXPR_ENDARG', 'EXPR_ENDARG', 'EXPR_ARG',
|
245
|
+
'EXPR_FNAME|EXPR_FITEM', 'EXPR_CLASS', 'EXPR_END|EXPR_LABEL'
|
246
|
+
inline_comments << sexp
|
247
|
+
when 'EXPR_BEG|EXPR_LABEL', 'EXPR_MID'
|
248
|
+
inline_comments << sexp.merge!(break: true)
|
249
|
+
when 'EXPR_DOT'
|
250
|
+
last_sexp.merge!(comments: [sexp.merge!(break: true)])
|
251
|
+
end
|
155
252
|
|
156
|
-
|
157
|
-
|
158
|
-
end
|
253
|
+
sexp
|
254
|
+
end
|
159
255
|
|
160
|
-
|
161
|
-
@current_embdoc[:body] << comment.chomp
|
162
|
-
@block_comments << @current_embdoc
|
163
|
-
@current_embdoc = nil
|
164
|
-
end
|
256
|
+
defined = private_instance_methods(false).grep(/\Aon_/) { $'.to_sym }
|
165
257
|
|
166
|
-
|
167
|
-
|
168
|
-
|
169
|
-
|
258
|
+
(Ripper::PARSER_EVENTS - defined).each do |event|
|
259
|
+
define_method(:"on_#{event}") do |*body|
|
260
|
+
super(*body).tap do |sexp|
|
261
|
+
@last_sexp = sexp
|
262
|
+
next if inline_comments.empty?
|
170
263
|
|
171
|
-
|
264
|
+
sexp[:comments] = inline_comments.reverse
|
265
|
+
@inline_comments = []
|
266
|
+
end
|
267
|
+
end
|
172
268
|
end
|
173
269
|
end
|
174
|
-
|
270
|
+
)
|
175
271
|
|
176
|
-
#
|
177
|
-
#
|
178
|
-
#
|
179
|
-
|
180
|
-
|
181
|
-
|
182
|
-
|
183
|
-
|
272
|
+
# Nodes that are always on their own line occur when the lexer is in the
|
273
|
+
# EXPR_BEG state. Those comments are tracked within the @block_comments
|
274
|
+
# instance variable. Then for each node that could contain them, we attach
|
275
|
+
# them after the node has been built.
|
276
|
+
prepend(
|
277
|
+
Module.new do
|
278
|
+
events = {
|
279
|
+
begin: [0, :body, 0],
|
280
|
+
bodystmt: [0],
|
281
|
+
class: [2, :body, 0],
|
282
|
+
def: [2, :body, 0],
|
283
|
+
defs: [4, :body, 0],
|
284
|
+
else: [0],
|
285
|
+
elsif: [1],
|
286
|
+
ensure: [0],
|
287
|
+
if: [1],
|
288
|
+
program: [0],
|
289
|
+
rescue: [2],
|
290
|
+
sclass: [1, :body, 0],
|
291
|
+
unless: [1],
|
292
|
+
until: [1],
|
293
|
+
when: [1],
|
294
|
+
while: [1]
|
295
|
+
}
|
184
296
|
|
185
|
-
|
186
|
-
|
187
|
-
|
297
|
+
def initialize(*args)
|
298
|
+
super(*args)
|
299
|
+
@block_comments = []
|
300
|
+
@current_embdoc = nil
|
301
|
+
end
|
188
302
|
|
189
|
-
|
303
|
+
def self.prepended(base)
|
304
|
+
base.attr_reader :block_comments, :current_embdoc
|
305
|
+
end
|
190
306
|
|
191
|
-
|
192
|
-
super(body).tap { |sexp| heredoc_stack << sexp }
|
193
|
-
end
|
307
|
+
private
|
194
308
|
|
195
|
-
|
196
|
-
|
197
|
-
|
309
|
+
def attach_comments(sexp, stmts)
|
310
|
+
range = sexp[:start]..sexp[:end]
|
311
|
+
comments =
|
312
|
+
block_comments.group_by { |comment| range.include?(comment[:start]) }
|
198
313
|
|
199
|
-
|
200
|
-
|
201
|
-
|
202
|
-
end
|
314
|
+
if comments[true]
|
315
|
+
stmts[:body] =
|
316
|
+
(stmts[:body] + comments[true]).sort_by { |node| node[:start] }
|
203
317
|
|
204
|
-
|
205
|
-
|
206
|
-
|
318
|
+
@block_comments = comments.fetch(false) { [] }
|
319
|
+
end
|
320
|
+
end
|
207
321
|
|
208
|
-
|
209
|
-
|
210
|
-
|
211
|
-
|
322
|
+
events.each do |event, path|
|
323
|
+
define_method(:"on_#{event}") do |*body|
|
324
|
+
super(*body).tap { |sexp| attach_comments(sexp, body.dig(*path)) }
|
325
|
+
end
|
326
|
+
end
|
212
327
|
|
213
|
-
|
214
|
-
|
328
|
+
def on_comment(body)
|
329
|
+
super(body).tap do |sexp|
|
330
|
+
block_comments << sexp if RipperJS.lex_state_name(state) == 'EXPR_BEG'
|
331
|
+
end
|
332
|
+
end
|
215
333
|
|
216
|
-
|
217
|
-
|
218
|
-
|
219
|
-
|
220
|
-
super
|
334
|
+
def on_embdoc_beg(comment)
|
335
|
+
@current_embdoc = {
|
336
|
+
type: :embdoc, body: comment, start: lineno, end: lineno
|
337
|
+
}
|
221
338
|
end
|
222
|
-
end
|
223
|
-
end
|
224
339
|
|
225
|
-
|
226
|
-
|
227
|
-
|
228
|
-
# everything, so we need to force the encoding back into UTF-8 so that
|
229
|
-
# the JSON library won't break.
|
230
|
-
module Encoding
|
231
|
-
events = %w[comment ident tstring_content]
|
340
|
+
def on_embdoc(comment)
|
341
|
+
@current_embdoc[:body] << comment
|
342
|
+
end
|
232
343
|
|
233
|
-
|
234
|
-
|
235
|
-
|
344
|
+
def on_embdoc_end(comment)
|
345
|
+
@current_embdoc[:body] << comment.chomp
|
346
|
+
@block_comments << @current_embdoc
|
347
|
+
@current_embdoc = nil
|
236
348
|
end
|
237
|
-
end
|
238
|
-
end
|
239
349
|
|
240
|
-
|
241
|
-
|
242
|
-
|
243
|
-
|
244
|
-
module InlineComments
|
245
|
-
# Certain events needs to steal the comments from their children in order
|
246
|
-
# for them to display properly.
|
247
|
-
events = {
|
248
|
-
args_add_block: [:body, 0],
|
249
|
-
break: [:body, 0],
|
250
|
-
command: [:body, 1],
|
251
|
-
command_call: [:body, 3],
|
252
|
-
regexp_literal: [:body, 0],
|
253
|
-
string_literal: [:body, 0],
|
254
|
-
symbol_literal: [:body, 0]
|
255
|
-
}
|
256
|
-
|
257
|
-
def initialize(*args)
|
258
|
-
super(*args)
|
259
|
-
@inline_comments = []
|
260
|
-
@last_sexp = nil
|
261
|
-
end
|
350
|
+
def on_method_add_block(*body)
|
351
|
+
super(*body).tap do |sexp|
|
352
|
+
stmts = body[1][:body][1]
|
353
|
+
stmts = stmts[:type] == :stmts ? stmts : body[1][:body][1][:body][0]
|
262
354
|
|
263
|
-
|
264
|
-
|
355
|
+
attach_comments(sexp, stmts)
|
356
|
+
end
|
357
|
+
end
|
265
358
|
end
|
359
|
+
)
|
266
360
|
|
267
|
-
|
361
|
+
# Tracking heredocs in somewhat interesting. Straight-line heredocs are
|
362
|
+
# reported as strings, whereas squiggly-line heredocs are reported as
|
363
|
+
# heredocs. We track the start and matching end of the heredoc as "beging" and
|
364
|
+
# "ending" respectively.
|
365
|
+
prepend(
|
366
|
+
Module.new do
|
367
|
+
def initialize(*args)
|
368
|
+
super(*args)
|
369
|
+
@heredoc_stack = []
|
370
|
+
end
|
268
371
|
|
269
|
-
|
270
|
-
|
271
|
-
@last_sexp =
|
272
|
-
super(*body).tap do |sexp|
|
273
|
-
comments = (sexp.dig(*path) || {}).delete(:comments)
|
274
|
-
sexp.merge!(comments: comments) if comments
|
275
|
-
end
|
372
|
+
def self.prepended(base)
|
373
|
+
base.attr_reader :heredoc_stack
|
276
374
|
end
|
277
|
-
end
|
278
375
|
|
279
|
-
|
376
|
+
private
|
280
377
|
|
281
|
-
|
282
|
-
|
283
|
-
|
284
|
-
@last_sexp =
|
285
|
-
super(*body).tap do |sexp|
|
286
|
-
next unless SPECIAL_LITERALS.include?(body.dig(0, :type))
|
378
|
+
def on_embexpr_beg(body)
|
379
|
+
super(body).tap { |sexp| heredoc_stack << sexp }
|
380
|
+
end
|
287
381
|
|
288
|
-
|
289
|
-
|
290
|
-
|
291
|
-
end
|
382
|
+
def on_embexpr_end(body)
|
383
|
+
super(body).tap { heredoc_stack.pop }
|
384
|
+
end
|
292
385
|
|
293
|
-
|
294
|
-
|
295
|
-
|
296
|
-
|
297
|
-
super(*body).tap do |sexp|
|
298
|
-
comments =
|
299
|
-
(sexp.dig(:body, 0).delete(:comments) || []) +
|
300
|
-
(sexp.dig(:body, 1).delete(:comments) || [])
|
386
|
+
def on_heredoc_beg(beging)
|
387
|
+
heredoc = { type: :heredoc, beging: beging, start: lineno, end: lineno }
|
388
|
+
heredoc_stack << heredoc
|
389
|
+
end
|
301
390
|
|
302
|
-
|
303
|
-
end
|
304
|
-
|
391
|
+
def on_heredoc_end(ending)
|
392
|
+
heredoc_stack[-1].merge!(ending: ending.chomp, end: lineno)
|
393
|
+
end
|
305
394
|
|
306
|
-
|
307
|
-
|
308
|
-
|
309
|
-
|
310
|
-
@last_sexp = super(body)
|
311
|
-
end
|
395
|
+
def on_heredoc_dedent(string, _width)
|
396
|
+
heredoc = heredoc_stack.pop
|
397
|
+
string.merge!(heredoc.slice(:type, :beging, :ending, :start, :end))
|
398
|
+
end
|
312
399
|
|
313
|
-
|
314
|
-
|
315
|
-
#
|
316
|
-
# enum lex_state_bits {
|
317
|
-
# EXPR_BEG_bit, /* ignore newline, +/- is a sign. */
|
318
|
-
# EXPR_END_bit, /* newline significant, +/- is an operator. */
|
319
|
-
# EXPR_ENDARG_bit, /* ditto, and unbound braces. */
|
320
|
-
# EXPR_ENDFN_bit, /* ditto, and unbound braces. */
|
321
|
-
# EXPR_ARG_bit, /* newline significant, +/- is an operator. */
|
322
|
-
# EXPR_CMDARG_bit, /* newline significant, +/- is an operator. */
|
323
|
-
# EXPR_MID_bit, /* newline significant, +/- is an operator. */
|
324
|
-
# EXPR_FNAME_bit, /* ignore newline, no reserved words. */
|
325
|
-
# EXPR_DOT_bit, /* right after `.' or `::', no reserved words. */
|
326
|
-
# EXPR_CLASS_bit, /* immediate after `class', no here document. */
|
327
|
-
# EXPR_LABEL_bit, /* flag bit, label is allowed. */
|
328
|
-
# EXPR_LABELED_bit, /* flag bit, just after a label. */
|
329
|
-
# EXPR_FITEM_bit, /* symbol literal as FNAME. */
|
330
|
-
# EXPR_MAX_STATE
|
331
|
-
# };
|
332
|
-
def on_comment(body)
|
333
|
-
sexp = { type: :@comment, body: body.chomp, start: lineno, end: lineno }
|
334
|
-
|
335
|
-
case RipperJS.lex_state_name(state)
|
336
|
-
when 'EXPR_END', 'EXPR_ARG|EXPR_LABELED', 'EXPR_ENDFN'
|
337
|
-
last_sexp.merge!(comments: [sexp])
|
338
|
-
when 'EXPR_CMDARG', 'EXPR_END|EXPR_ENDARG', 'EXPR_ENDARG', 'EXPR_ARG',
|
339
|
-
'EXPR_FNAME|EXPR_FITEM', 'EXPR_CLASS', 'EXPR_END|EXPR_LABEL'
|
340
|
-
inline_comments << sexp
|
341
|
-
when 'EXPR_BEG|EXPR_LABEL', 'EXPR_MID'
|
342
|
-
inline_comments << sexp.merge!(break: true)
|
343
|
-
when 'EXPR_DOT'
|
344
|
-
last_sexp.merge!(comments: [sexp.merge!(break: true)])
|
345
|
-
end
|
346
|
-
|
347
|
-
sexp
|
348
|
-
end
|
400
|
+
def on_string_literal(string)
|
401
|
+
heredoc = heredoc_stack[-1]
|
349
402
|
|
350
|
-
|
403
|
+
if heredoc && string[:type] != :heredoc && heredoc[:type] == :heredoc
|
404
|
+
heredoc_stack.pop
|
405
|
+
string.merge!(heredoc.slice(:type, :beging, :ending, :start, :end))
|
406
|
+
else
|
407
|
+
super
|
408
|
+
end
|
409
|
+
end
|
410
|
+
end
|
411
|
+
)
|
351
412
|
|
352
|
-
|
353
|
-
|
354
|
-
|
355
|
-
|
356
|
-
|
413
|
+
# These are the event types that contain _actual_ string content. If there is
|
414
|
+
# an encoding magic comment at the top of the file, ripper will actually
|
415
|
+
# change into that encoding for the storage of the string. This will break
|
416
|
+
# everything, so we need to force the encoding back into UTF-8 so that
|
417
|
+
# the JSON library won't break.
|
418
|
+
prepend(
|
419
|
+
Module.new do
|
420
|
+
private
|
357
421
|
|
358
|
-
|
359
|
-
|
422
|
+
%w[comment ident tstring_content].each do |event|
|
423
|
+
define_method(:"on_#{event}") do |body|
|
424
|
+
super(body.force_encoding('UTF-8'))
|
360
425
|
end
|
361
426
|
end
|
362
427
|
end
|
363
|
-
|
428
|
+
)
|
364
429
|
|
365
430
|
# Handles __END__ syntax, which allows individual scripts to keep content
|
366
|
-
# after the main ruby code that can be read through DATA.
|
367
|
-
|
368
|
-
|
369
|
-
|
370
|
-
|
371
|
-
|
372
|
-
|
431
|
+
# after the main ruby code that can be read through DATA. Which looks like:
|
432
|
+
#
|
433
|
+
# foo.bar
|
434
|
+
#
|
435
|
+
# __END__
|
436
|
+
# some other content that isn't read by ripper normally
|
437
|
+
prepend(
|
438
|
+
Module.new do
|
439
|
+
def initialize(source, *args)
|
440
|
+
super(source, *args)
|
441
|
+
@source = source
|
442
|
+
@ending = nil
|
443
|
+
end
|
373
444
|
|
374
|
-
|
375
|
-
|
376
|
-
|
445
|
+
def self.prepended(base)
|
446
|
+
base.attr_reader :source, :ending
|
447
|
+
end
|
377
448
|
|
378
|
-
|
449
|
+
private
|
379
450
|
|
380
|
-
|
381
|
-
|
382
|
-
|
451
|
+
def on___end__(body)
|
452
|
+
@ending = super(source.split("\n")[lineno..-1].join("\n"))
|
453
|
+
end
|
383
454
|
|
384
|
-
|
385
|
-
|
455
|
+
def on_program(*body)
|
456
|
+
super(*body).tap { |sexp| sexp[:body][0][:body] << ending if ending }
|
457
|
+
end
|
386
458
|
end
|
387
|
-
|
459
|
+
)
|
388
460
|
|
389
|
-
# Adds the used quote type onto string nodes.
|
390
|
-
|
391
|
-
|
461
|
+
# Adds the used quote type onto string nodes. This is necessary because we're
|
462
|
+
# going to have to stick to whatever quote the user chose if there are escape
|
463
|
+
# sequences within the string. For example, if you have '\n' we can't switch
|
464
|
+
# to double quotes without changing what it means.
|
465
|
+
prepend(
|
466
|
+
Module.new do
|
467
|
+
private
|
392
468
|
|
393
|
-
|
394
|
-
|
395
|
-
|
469
|
+
def on_tstring_end(quote)
|
470
|
+
last_sexp.merge!(quote: quote)
|
471
|
+
end
|
396
472
|
|
397
|
-
|
398
|
-
|
473
|
+
def on_label_end(quote)
|
474
|
+
last_sexp.merge!(quote: quote[0]) # quote is ": or ':
|
475
|
+
end
|
399
476
|
end
|
400
|
-
|
477
|
+
)
|
401
478
|
|
402
479
|
# Normally access controls are reported as vcall nodes. This module creates a
|
403
|
-
# new node type to explicitly track those nodes instead
|
404
|
-
|
405
|
-
|
406
|
-
|
407
|
-
|
408
|
-
end
|
409
|
-
|
410
|
-
def self.prepended(base)
|
411
|
-
base.attr_reader :lines
|
412
|
-
end
|
413
|
-
|
414
|
-
private
|
480
|
+
# new node type to explicitly track those nodes instead, so that the printer
|
481
|
+
# can add new lines as necessary.
|
482
|
+
prepend(
|
483
|
+
Module.new do
|
484
|
+
KEYWORDS = %w[private protected public].freeze
|
415
485
|
|
416
|
-
|
417
|
-
|
418
|
-
|
419
|
-
|
420
|
-
next
|
421
|
-
end
|
486
|
+
def initialize(source, *args)
|
487
|
+
super(source, *args)
|
488
|
+
@lines = source.split("\n")
|
489
|
+
end
|
422
490
|
|
423
|
-
|
491
|
+
def self.prepended(base)
|
492
|
+
base.attr_reader :lines
|
424
493
|
end
|
425
|
-
end
|
426
|
-
end
|
427
|
-
end
|
428
494
|
|
429
|
-
|
430
|
-
private
|
495
|
+
private
|
431
496
|
|
432
|
-
|
433
|
-
|
434
|
-
|
435
|
-
|
436
|
-
|
497
|
+
def on_vcall(ident)
|
498
|
+
super(ident).tap do |sexp|
|
499
|
+
if !KEYWORDS.include?(ident[:body]) ||
|
500
|
+
ident[:body] != lines[lineno - 1].strip
|
501
|
+
next
|
502
|
+
end
|
437
503
|
|
438
|
-
|
439
|
-
|
440
|
-
|
441
|
-
{ type: event, body: body, start: min || lineno, end: lineno }
|
504
|
+
sexp.merge!(type: :access_ctrl)
|
505
|
+
end
|
506
|
+
end
|
442
507
|
end
|
443
|
-
|
444
|
-
|
445
|
-
prepend Layer::Lists
|
446
|
-
prepend Layer::StartLine
|
447
|
-
prepend Layer::InlineComments
|
448
|
-
prepend Layer::BlockComments
|
449
|
-
prepend Layer::Heredocs
|
450
|
-
prepend Layer::Encoding
|
451
|
-
prepend Layer::Ending
|
452
|
-
prepend Layer::Strings
|
453
|
-
prepend Layer::AccessControls
|
508
|
+
)
|
454
509
|
|
455
510
|
# When the only statement inside of a `def` node is a `begin` node, then you
|
456
511
|
# can safely replace the body of the `def` with the body of the `begin`. For
|
@@ -529,6 +584,10 @@ class RipperJS < Ripper
|
|
529
584
|
)
|
530
585
|
end
|
531
586
|
|
587
|
+
# If this is the main file we're executing, then most likely this is being
|
588
|
+
# executed from the parse.js spawn. In that case, read the ruby source from
|
589
|
+
# stdin and report back the AST over stdout.
|
590
|
+
|
532
591
|
if $0 == __FILE__
|
533
592
|
builder = RipperJS.new($stdin.read)
|
534
593
|
response = builder.parse
|