prettier 0.12.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/CHANGELOG.md +368 -0
- data/CODE_OF_CONDUCT.md +76 -0
- data/CONTRIBUTING.md +143 -0
- data/LICENSE +21 -0
- data/README.md +151 -0
- data/bin/console +7 -0
- data/exe/rbprettier +8 -0
- data/lib/prettier.rb +18 -0
- data/lib/prettier/rake/task.rb +57 -0
- data/node_modules/prettier/bin-prettier.js +44620 -0
- data/node_modules/prettier/index.js +42565 -0
- data/node_modules/prettier/third-party.js +5326 -0
- data/package.json +39 -0
- data/src/builders.js +9 -0
- data/src/escapePattern.js +45 -0
- data/src/nodes.js +594 -0
- data/src/nodes/alias.js +32 -0
- data/src/nodes/arrays.js +162 -0
- data/src/nodes/blocks.js +66 -0
- data/src/nodes/calls.js +28 -0
- data/src/nodes/case.js +57 -0
- data/src/nodes/commands.js +70 -0
- data/src/nodes/conditionals.js +154 -0
- data/src/nodes/hashes.js +134 -0
- data/src/nodes/hooks.js +15 -0
- data/src/nodes/lambdas.js +59 -0
- data/src/nodes/loops.js +46 -0
- data/src/nodes/methods.js +42 -0
- data/src/nodes/params.js +75 -0
- data/src/nodes/regexp.js +18 -0
- data/src/nodes/rescue.js +77 -0
- data/src/nodes/strings.js +143 -0
- data/src/parse.js +16 -0
- data/src/print.js +23 -0
- data/src/ripper.rb +542 -0
- data/src/ruby.js +127 -0
- data/src/toProc.js +82 -0
- data/src/utils.js +150 -0
- metadata +109 -0
@@ -0,0 +1,143 @@
|
|
1
|
+
const {
|
2
|
+
concat,
|
3
|
+
group,
|
4
|
+
hardline,
|
5
|
+
indent,
|
6
|
+
join,
|
7
|
+
literalline,
|
8
|
+
softline
|
9
|
+
} = require("../builders");
|
10
|
+
const { concatBody, empty, makeList, surround } = require("../utils");
|
11
|
+
const escapePattern = require("../escapePattern");
|
12
|
+
|
13
|
+
// If there is some part of this string that matches an escape sequence or that
|
14
|
+
// contains the interpolation pattern ("#{"), then we are locked into whichever
|
15
|
+
// quote the user chose. (If they chose single quotes, then double quoting
|
16
|
+
// would activate the escape sequence, and if they chose double quotes, then
|
17
|
+
// single quotes would deactivate it.)
|
18
|
+
const isQuoteLocked = string =>
|
19
|
+
string.body.some(
|
20
|
+
part =>
|
21
|
+
part.type === "@tstring_content" &&
|
22
|
+
(escapePattern.test(part.body) || part.body.includes("#{"))
|
23
|
+
);
|
24
|
+
|
25
|
+
// A string is considered to be able to use single quotes if it contains only
|
26
|
+
// plain string content and that content does not contain a single quote.
|
27
|
+
const isSingleQuotable = string =>
|
28
|
+
string.body.every(
|
29
|
+
part => part.type === "@tstring_content" && !part.body.includes("'")
|
30
|
+
);
|
31
|
+
|
32
|
+
const getStringQuote = (string, preferSingleQuotes) => {
|
33
|
+
if (isQuoteLocked(string)) {
|
34
|
+
return string.quote;
|
35
|
+
}
|
36
|
+
|
37
|
+
return preferSingleQuotes && isSingleQuotable(string) ? "'" : '"';
|
38
|
+
};
|
39
|
+
|
40
|
+
const quotePattern = new RegExp("\\\\([\\s\\S])|(['\"])", "g");
|
41
|
+
|
42
|
+
const makeString = (content, enclosingQuote) => {
|
43
|
+
const otherQuote = enclosingQuote === '"' ? "'" : enclosingQuote;
|
44
|
+
|
45
|
+
// Escape and unescape single and double quotes as needed to be able to
|
46
|
+
// enclose `content` with `enclosingQuote`.
|
47
|
+
return content.replace(quotePattern, (match, escaped, quote) => {
|
48
|
+
if (escaped === otherQuote) {
|
49
|
+
return escaped;
|
50
|
+
}
|
51
|
+
|
52
|
+
if (quote === enclosingQuote) {
|
53
|
+
return `\\${quote}`;
|
54
|
+
}
|
55
|
+
|
56
|
+
if (quote) {
|
57
|
+
return quote;
|
58
|
+
}
|
59
|
+
|
60
|
+
return `\\${escaped}`;
|
61
|
+
});
|
62
|
+
};
|
63
|
+
|
64
|
+
module.exports = {
|
65
|
+
"@CHAR": (path, { preferSingleQuotes }, _print) => {
|
66
|
+
const { body } = path.getValue();
|
67
|
+
|
68
|
+
if (body.length !== 2) {
|
69
|
+
return body;
|
70
|
+
}
|
71
|
+
|
72
|
+
const quote = preferSingleQuotes ? "'" : '"';
|
73
|
+
return body.length === 2 ? concat([quote, body.slice(1), quote]) : body;
|
74
|
+
},
|
75
|
+
heredoc: (path, opts, print) => {
|
76
|
+
const { beging, ending } = path.getValue();
|
77
|
+
|
78
|
+
return concat([
|
79
|
+
beging,
|
80
|
+
concat([literalline].concat(path.map(print, "body"))),
|
81
|
+
ending
|
82
|
+
]);
|
83
|
+
},
|
84
|
+
string: makeList,
|
85
|
+
string_concat: (path, opts, print) =>
|
86
|
+
group(
|
87
|
+
concat([
|
88
|
+
path.call(print, "body", 0),
|
89
|
+
" \\",
|
90
|
+
indent(concat([hardline, path.call(print, "body", 1)]))
|
91
|
+
])
|
92
|
+
),
|
93
|
+
string_dvar: surround("#{", "}"),
|
94
|
+
string_embexpr: (path, opts, print) =>
|
95
|
+
group(
|
96
|
+
concat([
|
97
|
+
"#{",
|
98
|
+
indent(concat([softline, path.call(print, "body", 0)])),
|
99
|
+
concat([softline, "}"])
|
100
|
+
])
|
101
|
+
),
|
102
|
+
string_literal: (path, { preferSingleQuotes }, print) => {
|
103
|
+
const string = path.getValue().body[0];
|
104
|
+
|
105
|
+
// If this string is actually a heredoc, bail out and return to the print
|
106
|
+
// function for heredocs
|
107
|
+
if (string.type === "heredoc") {
|
108
|
+
return path.call(print, "body", 0);
|
109
|
+
}
|
110
|
+
|
111
|
+
// If the string is empty, it will not have any parts, so just print out the
|
112
|
+
// quotes corresponding to the config
|
113
|
+
if (string.body.length === 0) {
|
114
|
+
return preferSingleQuotes ? "''" : '""';
|
115
|
+
}
|
116
|
+
|
117
|
+
const quote = getStringQuote(string, preferSingleQuotes);
|
118
|
+
const parts = [];
|
119
|
+
|
120
|
+
string.body.forEach((part, index) => {
|
121
|
+
if (part.type === "@tstring_content") {
|
122
|
+
// In this case, the part of the string is just regular string content
|
123
|
+
parts.push(makeString(part.body, quote));
|
124
|
+
} else {
|
125
|
+
// In this case, the part of the string is an embedded expression
|
126
|
+
parts.push(path.call(print, "body", 0, "body", index));
|
127
|
+
}
|
128
|
+
});
|
129
|
+
|
130
|
+
return concat([quote].concat(parts).concat([quote]));
|
131
|
+
},
|
132
|
+
word_add: concatBody,
|
133
|
+
word_new: empty,
|
134
|
+
xstring: makeList,
|
135
|
+
xstring_literal: (path, opts, print) =>
|
136
|
+
group(
|
137
|
+
concat([
|
138
|
+
"`",
|
139
|
+
indent(concat([softline, join(softline, path.call(print, "body", 0))])),
|
140
|
+
concat([softline, "`"])
|
141
|
+
])
|
142
|
+
)
|
143
|
+
};
|
data/src/parse.js
ADDED
@@ -0,0 +1,16 @@
|
|
1
|
+
const { spawnSync } = require("child_process");
|
2
|
+
const path = require("path");
|
3
|
+
|
4
|
+
module.exports = (text, _parsers, _opts) => {
|
5
|
+
const child = spawnSync("ruby", [path.join(__dirname, "./ripper.rb")], {
|
6
|
+
input: text
|
7
|
+
});
|
8
|
+
|
9
|
+
const error = child.stderr.toString();
|
10
|
+
if (error) {
|
11
|
+
throw new Error(error);
|
12
|
+
}
|
13
|
+
|
14
|
+
const response = child.stdout.toString();
|
15
|
+
return JSON.parse(response);
|
16
|
+
};
|
data/src/print.js
ADDED
@@ -0,0 +1,23 @@
|
|
1
|
+
const { printComments } = require("./utils");
|
2
|
+
const nodes = require("./nodes");
|
3
|
+
|
4
|
+
module.exports = (path, opts, print) => {
|
5
|
+
const { type, body, comments, start } = path.getValue();
|
6
|
+
|
7
|
+
if (type in nodes) {
|
8
|
+
const printed = nodes[type](path, opts, print);
|
9
|
+
|
10
|
+
if (comments) {
|
11
|
+
return printComments(printed, start, comments);
|
12
|
+
}
|
13
|
+
return printed;
|
14
|
+
}
|
15
|
+
|
16
|
+
if (type[0] === "@") {
|
17
|
+
return body;
|
18
|
+
}
|
19
|
+
|
20
|
+
throw new Error(
|
21
|
+
`Unsupported node encountered: ${type}\n${JSON.stringify(body, null, 2)}`
|
22
|
+
);
|
23
|
+
};
|
data/src/ripper.rb
ADDED
@@ -0,0 +1,542 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
|
3
|
+
REQUIRED_VERSION = Gem::Version.new('2.5')
|
4
|
+
if Gem::Version.new(RUBY_VERSION) < REQUIRED_VERSION
|
5
|
+
raise "Ruby version #{RUBY_VERSION} not supported. " \
|
6
|
+
"Please upgrade to #{REQUIRED_VERSION} or above."
|
7
|
+
end
|
8
|
+
|
9
|
+
require 'json' unless defined?(JSON)
|
10
|
+
require 'ripper'
|
11
|
+
|
12
|
+
module Layer
|
13
|
+
# Some nodes are lists that come back from the parser. They always start with
|
14
|
+
# a *_new node (or in the case of string, *_content) and each additional node
|
15
|
+
# in the list is a *_add node. This layer takes those nodes and turns them
|
16
|
+
# into one node with an array body.
|
17
|
+
module Lists
|
18
|
+
events = %i[
|
19
|
+
args
|
20
|
+
mlhs
|
21
|
+
mrhs
|
22
|
+
qsymbols
|
23
|
+
qwords
|
24
|
+
regexp
|
25
|
+
stmts
|
26
|
+
string
|
27
|
+
symbols
|
28
|
+
words
|
29
|
+
xstring
|
30
|
+
]
|
31
|
+
|
32
|
+
private
|
33
|
+
|
34
|
+
events.each do |event|
|
35
|
+
suffix = event == :string ? 'content' : 'new'
|
36
|
+
|
37
|
+
define_method(:"on_#{event}_#{suffix}") do
|
38
|
+
{ type: event, body: [], start: lineno, end: lineno }
|
39
|
+
end
|
40
|
+
|
41
|
+
define_method(:"on_#{event}_add") do |parts, part|
|
42
|
+
parts.tap do |node|
|
43
|
+
node[:body] << part
|
44
|
+
node[:end] = lineno
|
45
|
+
end
|
46
|
+
end
|
47
|
+
end
|
48
|
+
end
|
49
|
+
|
50
|
+
# For most nodes, it's enough to look at the child nodes to determine the
|
51
|
+
# start of the parent node. However, for some nodes it's necessary to keep
|
52
|
+
# track of the keywords as they come in from the lexer and to modify the start
|
53
|
+
# node once we have it.
|
54
|
+
module StartLine
|
55
|
+
events = %i[begin else elsif ensure rescue until while]
|
56
|
+
|
57
|
+
def initialize(*args)
|
58
|
+
super(*args)
|
59
|
+
@keywords = []
|
60
|
+
end
|
61
|
+
|
62
|
+
def self.prepended(base)
|
63
|
+
base.attr_reader :keywords
|
64
|
+
end
|
65
|
+
|
66
|
+
private
|
67
|
+
|
68
|
+
def find_start(body)
|
69
|
+
keywords[keywords.rindex { |keyword| keyword[:body] == body }][:start]
|
70
|
+
end
|
71
|
+
|
72
|
+
events.each do |event|
|
73
|
+
keyword = event.to_s
|
74
|
+
|
75
|
+
define_method(:"on_#{event}") do |*body|
|
76
|
+
super(*body).tap { |sexp| sexp.merge!(start: find_start(keyword)) }
|
77
|
+
end
|
78
|
+
end
|
79
|
+
|
80
|
+
def on_kw(body)
|
81
|
+
super(body).tap { |sexp| keywords << sexp }
|
82
|
+
end
|
83
|
+
|
84
|
+
def on_program(*body)
|
85
|
+
super(*body).tap { |sexp| sexp.merge!(start: 1) }
|
86
|
+
end
|
87
|
+
end
|
88
|
+
|
89
|
+
# Nodes that are always on their own line occur when the lexer is in the
|
90
|
+
# EXPR_BEG node. Those comments are tracked within the @block_comments
|
91
|
+
# instance variable. Then for each node that could contain them, we attach
|
92
|
+
# them after the node has been built.
|
93
|
+
module BlockComments
|
94
|
+
events = {
|
95
|
+
begin: [0, :body, 0],
|
96
|
+
bodystmt: [0],
|
97
|
+
class: [2, :body, 0],
|
98
|
+
def: [2, :body, 0],
|
99
|
+
defs: [4, :body, 0],
|
100
|
+
else: [0],
|
101
|
+
elsif: [1],
|
102
|
+
ensure: [0],
|
103
|
+
if: [1],
|
104
|
+
program: [0],
|
105
|
+
rescue: [2],
|
106
|
+
sclass: [1, :body, 0],
|
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
|
118
|
+
|
119
|
+
def self.prepended(base)
|
120
|
+
base.attr_reader :block_comments, :current_embdoc
|
121
|
+
end
|
122
|
+
|
123
|
+
private
|
124
|
+
|
125
|
+
def attach_comments(sexp, stmts)
|
126
|
+
range = sexp[:start]..sexp[:end]
|
127
|
+
comments =
|
128
|
+
block_comments.group_by { |comment| range.include?(comment[:start]) }
|
129
|
+
|
130
|
+
if comments[true]
|
131
|
+
stmts[:body] =
|
132
|
+
(stmts[:body] + comments[true]).sort_by { |node| node[:start] }
|
133
|
+
|
134
|
+
@block_comments = comments.fetch(false) { [] }
|
135
|
+
end
|
136
|
+
end
|
137
|
+
|
138
|
+
events.each do |event, path|
|
139
|
+
define_method(:"on_#{event}") do |*body|
|
140
|
+
super(*body).tap { |sexp| attach_comments(sexp, body.dig(*path)) }
|
141
|
+
end
|
142
|
+
end
|
143
|
+
|
144
|
+
def on_comment(body)
|
145
|
+
super(body).tap do |sexp|
|
146
|
+
block_comments << sexp if RipperJS.lex_state_name(state) == 'EXPR_BEG'
|
147
|
+
end
|
148
|
+
end
|
149
|
+
|
150
|
+
def on_embdoc_beg(comment)
|
151
|
+
@current_embdoc = {
|
152
|
+
type: :embdoc, body: comment, start: lineno, end: lineno
|
153
|
+
}
|
154
|
+
end
|
155
|
+
|
156
|
+
def on_embdoc(comment)
|
157
|
+
@current_embdoc[:body] << comment
|
158
|
+
end
|
159
|
+
|
160
|
+
def on_embdoc_end(comment)
|
161
|
+
@current_embdoc[:body] << comment.chomp
|
162
|
+
@block_comments << @current_embdoc
|
163
|
+
@current_embdoc = nil
|
164
|
+
end
|
165
|
+
|
166
|
+
def on_method_add_block(*body)
|
167
|
+
super(*body).tap do |sexp|
|
168
|
+
stmts = body[1][:body][1]
|
169
|
+
stmts = stmts[:type] == :stmts ? stmts : body[1][:body][1][:body][0]
|
170
|
+
|
171
|
+
attach_comments(sexp, stmts)
|
172
|
+
end
|
173
|
+
end
|
174
|
+
end
|
175
|
+
|
176
|
+
# Tracking heredocs in somewhat interesting. Straight-line heredocs are
|
177
|
+
# reported as strings, whereas squiggly-line heredocs are reported as
|
178
|
+
# heredocs.
|
179
|
+
module Heredocs
|
180
|
+
def initialize(*args)
|
181
|
+
super(*args)
|
182
|
+
@heredoc_stack = []
|
183
|
+
end
|
184
|
+
|
185
|
+
def self.prepended(base)
|
186
|
+
base.attr_reader :heredoc_stack
|
187
|
+
end
|
188
|
+
|
189
|
+
private
|
190
|
+
|
191
|
+
def on_embexpr_beg(body)
|
192
|
+
super(body).tap { |sexp| heredoc_stack << sexp }
|
193
|
+
end
|
194
|
+
|
195
|
+
def on_embexpr_end(body)
|
196
|
+
super(body).tap { heredoc_stack.pop }
|
197
|
+
end
|
198
|
+
|
199
|
+
def on_heredoc_beg(beging)
|
200
|
+
heredoc = { type: :heredoc, beging: beging, start: lineno, end: lineno }
|
201
|
+
heredoc_stack << heredoc
|
202
|
+
end
|
203
|
+
|
204
|
+
def on_heredoc_end(ending)
|
205
|
+
heredoc_stack[-1].merge!(ending: ending.chomp, end: lineno)
|
206
|
+
end
|
207
|
+
|
208
|
+
def on_heredoc_dedent(string, _width)
|
209
|
+
heredoc = heredoc_stack.pop
|
210
|
+
string.merge!(heredoc.slice(:type, :beging, :ending, :start, :end))
|
211
|
+
end
|
212
|
+
|
213
|
+
def on_string_literal(string)
|
214
|
+
heredoc = heredoc_stack[-1]
|
215
|
+
|
216
|
+
if heredoc && string[:type] != :heredoc && heredoc[:type] == :heredoc
|
217
|
+
heredoc_stack.pop
|
218
|
+
string.merge!(heredoc.slice(:type, :beging, :ending, :start, :end))
|
219
|
+
else
|
220
|
+
super
|
221
|
+
end
|
222
|
+
end
|
223
|
+
end
|
224
|
+
|
225
|
+
# These are the event types that contain _actual_ string content. If there is
|
226
|
+
# an encoding magic comment at the top of the file, ripper will actually
|
227
|
+
# change into that encoding for the storage of the string. This will break
|
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]
|
232
|
+
|
233
|
+
events.each do |event|
|
234
|
+
define_method(:"on_#{event}") do |body|
|
235
|
+
super(body.force_encoding('UTF-8'))
|
236
|
+
end
|
237
|
+
end
|
238
|
+
end
|
239
|
+
|
240
|
+
# This layer keeps track of inline comments as they come in. Ripper itself
|
241
|
+
# doesn't attach comments to the AST, so we need to do it manually. In this
|
242
|
+
# case, inline comments are defined as any comments wherein the lexer state is
|
243
|
+
# not equal to EXPR_BEG (tracked in the BlockComments layer).
|
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
|
262
|
+
|
263
|
+
def self.prepended(base)
|
264
|
+
base.attr_reader :inline_comments, :last_sexp
|
265
|
+
end
|
266
|
+
|
267
|
+
private
|
268
|
+
|
269
|
+
events.each do |event, path|
|
270
|
+
define_method(:"on_#{event}") do |*body|
|
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
|
276
|
+
end
|
277
|
+
end
|
278
|
+
|
279
|
+
SPECIAL_LITERALS = %i[qsymbols qwords symbols words].freeze
|
280
|
+
|
281
|
+
# Special array literals are handled in different ways and so their comments
|
282
|
+
# need to be passed up to their parent array node.
|
283
|
+
def on_array(*body)
|
284
|
+
@last_sexp =
|
285
|
+
super(*body).tap do |sexp|
|
286
|
+
next unless SPECIAL_LITERALS.include?(body.dig(0, :type))
|
287
|
+
|
288
|
+
comments = sexp.dig(:body, 0).delete(:comments)
|
289
|
+
sexp.merge!(comments: comments) if comments
|
290
|
+
end
|
291
|
+
end
|
292
|
+
|
293
|
+
# Handling this specially because we want to pull the comments out of both
|
294
|
+
# child nodes.
|
295
|
+
def on_assoc_new(*body)
|
296
|
+
@last_sexp =
|
297
|
+
super(*body).tap do |sexp|
|
298
|
+
comments =
|
299
|
+
(sexp.dig(:body, 0).delete(:comments) || []) +
|
300
|
+
(sexp.dig(:body, 1).delete(:comments) || [])
|
301
|
+
|
302
|
+
sexp.merge!(comments: comments) if comments.any?
|
303
|
+
end
|
304
|
+
end
|
305
|
+
|
306
|
+
# Most scanner events don't stand on their own a s-expressions, but the CHAR
|
307
|
+
# scanner event is effectively just a string, so we need to track it as a
|
308
|
+
# s-expression.
|
309
|
+
def on_CHAR(body)
|
310
|
+
@last_sexp = super(body)
|
311
|
+
end
|
312
|
+
|
313
|
+
# We need to know exactly where the comment is, switching off the current
|
314
|
+
# lexer state. In Ruby 2.7.0-dev, that's defined as:
|
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
|
349
|
+
|
350
|
+
defined_events = private_instance_methods(false).grep(/\Aon_/) { $'.to_sym }
|
351
|
+
|
352
|
+
(Ripper::PARSER_EVENTS - defined_events).each do |event|
|
353
|
+
define_method(:"on_#{event}") do |*body|
|
354
|
+
super(*body).tap do |sexp|
|
355
|
+
@last_sexp = sexp
|
356
|
+
next if inline_comments.empty?
|
357
|
+
|
358
|
+
sexp[:comments] = inline_comments.reverse
|
359
|
+
@inline_comments = []
|
360
|
+
end
|
361
|
+
end
|
362
|
+
end
|
363
|
+
end
|
364
|
+
|
365
|
+
# Handles __END__ syntax, which allows individual scripts to keep content
|
366
|
+
# after the main ruby code that can be read through DATA.
|
367
|
+
module Ending
|
368
|
+
def initialize(source, *args)
|
369
|
+
super(source, *args)
|
370
|
+
@source = source
|
371
|
+
@ending = nil
|
372
|
+
end
|
373
|
+
|
374
|
+
def self.prepended(base)
|
375
|
+
base.attr_reader :source, :ending
|
376
|
+
end
|
377
|
+
|
378
|
+
private
|
379
|
+
|
380
|
+
def on___end__(body)
|
381
|
+
@ending = super(source.split("\n")[lineno..-1].join("\n"))
|
382
|
+
end
|
383
|
+
|
384
|
+
def on_program(*body)
|
385
|
+
super(*body).tap { |sexp| sexp[:body][0][:body] << ending if ending }
|
386
|
+
end
|
387
|
+
end
|
388
|
+
|
389
|
+
# Adds the used quote type onto string nodes.
|
390
|
+
module Strings
|
391
|
+
private
|
392
|
+
|
393
|
+
def on_tstring_end(quote)
|
394
|
+
last_sexp.merge!(quote: quote)
|
395
|
+
end
|
396
|
+
|
397
|
+
def on_label_end(quote)
|
398
|
+
last_sexp.merge!(quote: quote[0]) # quote is ": or ':
|
399
|
+
end
|
400
|
+
end
|
401
|
+
|
402
|
+
# Normally access controls are reported as vcall nodes. This module creates a
|
403
|
+
# new node type to explicitly track those nodes instead.
|
404
|
+
module AccessControls
|
405
|
+
def initialize(source, *args)
|
406
|
+
super(source, *args)
|
407
|
+
@lines = source.split("\n")
|
408
|
+
end
|
409
|
+
|
410
|
+
def self.prepended(base)
|
411
|
+
base.attr_reader :lines
|
412
|
+
end
|
413
|
+
|
414
|
+
private
|
415
|
+
|
416
|
+
def on_vcall(ident)
|
417
|
+
super(ident).tap do |sexp|
|
418
|
+
if !%w[private protected public].include?(ident[:body]) ||
|
419
|
+
ident[:body] != lines[lineno - 1].strip
|
420
|
+
next
|
421
|
+
end
|
422
|
+
|
423
|
+
sexp.merge!(type: :access_ctrl)
|
424
|
+
end
|
425
|
+
end
|
426
|
+
end
|
427
|
+
end
|
428
|
+
|
429
|
+
class RipperJS < Ripper
|
430
|
+
private
|
431
|
+
|
432
|
+
SCANNER_EVENTS.each do |event|
|
433
|
+
define_method(:"on_#{event}") do |body|
|
434
|
+
{ type: :"@#{event}", body: body, start: lineno, end: lineno }
|
435
|
+
end
|
436
|
+
end
|
437
|
+
|
438
|
+
PARSER_EVENTS.each do |event|
|
439
|
+
define_method(:"on_#{event}") do |*body|
|
440
|
+
min = body.map { |part| part.is_a?(Hash) ? part[:start] : lineno }.min
|
441
|
+
{ type: event, body: body, start: min || lineno, end: lineno }
|
442
|
+
end
|
443
|
+
end
|
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
|
454
|
+
|
455
|
+
# When the only statement inside of a `def` node is a `begin` node, then you
|
456
|
+
# can safely replace the body of the `def` with the body of the `begin`. For
|
457
|
+
# example:
|
458
|
+
#
|
459
|
+
# def foo
|
460
|
+
# begin
|
461
|
+
# try_something
|
462
|
+
# rescue SomeError => error
|
463
|
+
# handle_error(error)
|
464
|
+
# end
|
465
|
+
# end
|
466
|
+
#
|
467
|
+
# can get transformed into:
|
468
|
+
#
|
469
|
+
# def foo
|
470
|
+
# try_something
|
471
|
+
# rescue SomeError => error
|
472
|
+
# handle_error(error)
|
473
|
+
# end
|
474
|
+
#
|
475
|
+
# This module handles this by hoisting up the `bodystmt` node from the inner
|
476
|
+
# `begin` up to the `def`.
|
477
|
+
prepend(
|
478
|
+
Module.new do
|
479
|
+
private
|
480
|
+
|
481
|
+
def on_def(ident, params, bodystmt)
|
482
|
+
def_bodystmt = bodystmt
|
483
|
+
stmts, *other_parts = bodystmt[:body]
|
484
|
+
|
485
|
+
if !other_parts.any? && stmts[:body].length == 1 &&
|
486
|
+
stmts.dig(:body, 0, :type) == :begin
|
487
|
+
def_bodystmt = stmts.dig(:body, 0, :body, 0)
|
488
|
+
end
|
489
|
+
|
490
|
+
super(ident, params, def_bodystmt)
|
491
|
+
end
|
492
|
+
end
|
493
|
+
)
|
494
|
+
|
495
|
+
# By default, Ripper parses the expression `lambda { foo }` as a
|
496
|
+
# `method_add_block` node, so we can't turn it back into `-> { foo }`. This
|
497
|
+
# module overrides that behavior and reports it back as a `lambda` node
|
498
|
+
# instead.
|
499
|
+
prepend(
|
500
|
+
Module.new do
|
501
|
+
private
|
502
|
+
|
503
|
+
def on_method_add_block(invocation, block)
|
504
|
+
# It's possible to hit a `method_add_block` node without going through
|
505
|
+
# `method_add_arg` node, ex: `super {}`. In that case we're definitely
|
506
|
+
# not going to transform into a lambda.
|
507
|
+
return super if invocation[:type] != :method_add_arg
|
508
|
+
|
509
|
+
fcall, args = invocation[:body]
|
510
|
+
|
511
|
+
# If there are arguments to the `lambda`, that means `lambda` has been
|
512
|
+
# overridden as a function so we cannot transform it into a `lambda`
|
513
|
+
# node.
|
514
|
+
if fcall[:type] != :fcall || args[:type] != :args || args[:body].any?
|
515
|
+
return super
|
516
|
+
end
|
517
|
+
|
518
|
+
ident = fcall.dig(:body, 0)
|
519
|
+
return super if ident[:type] != :@ident || ident[:body] != 'lambda'
|
520
|
+
|
521
|
+
super.tap do |sexp|
|
522
|
+
params, stmts = block[:body]
|
523
|
+
params ||= { type: :params, body: [] }
|
524
|
+
|
525
|
+
sexp.merge!(type: :lambda, body: [params, stmts])
|
526
|
+
end
|
527
|
+
end
|
528
|
+
end
|
529
|
+
)
|
530
|
+
end
|
531
|
+
|
532
|
+
if $0 == __FILE__
|
533
|
+
builder = RipperJS.new($stdin.read)
|
534
|
+
response = builder.parse
|
535
|
+
|
536
|
+
if !response && builder.error?
|
537
|
+
STDERR.puts 'Invalid ruby'
|
538
|
+
exit 1
|
539
|
+
end
|
540
|
+
|
541
|
+
puts JSON.fast_generate(response)
|
542
|
+
end
|