bibtex-ruby 1.3.5 → 1.3.6
Sign up to get free protection for your applications and to get access to all the features.
Potentially problematic release.
This version of bibtex-ruby might be problematic. Click here for more details.
- data/Gemfile.lock +4 -2
- data/History.txt +5 -0
- data/Manifest +4 -0
- data/README.md +9 -12
- data/bibtex-ruby.gemspec +1 -0
- data/features/issues/braced_strings.feature +48 -0
- data/features/issues/trailing_comma.feature +21 -0
- data/features/step_definitions/bibtex_steps.rb +9 -1
- data/features/strings.feature +1 -0
- data/lib/bibtex/bibliography.rb +25 -22
- data/lib/bibtex/bibtex.y +11 -23
- data/lib/bibtex/elements.rb +11 -13
- data/lib/bibtex/entry.rb +4 -2
- data/lib/bibtex/lexer.rb +142 -163
- data/lib/bibtex/names.rb +10 -10
- data/lib/bibtex/parser.output +135 -132
- data/lib/bibtex/parser.rb +69 -81
- data/lib/bibtex/utilities.rb +1 -1
- data/lib/bibtex/version.rb +1 -1
- data/test/bibtex/test_lexer.rb +11 -0
- data/test/fixtures/benchmark.bib +1 -1
- data/test/profile.dot +129 -74
- data/test/profile.rb +4 -2
- metadata +18 -3
data/lib/bibtex/entry.rb
CHANGED
@@ -227,6 +227,8 @@ module BibTeX
|
|
227
227
|
def added_to_bibliography(bibliography)
|
228
228
|
super
|
229
229
|
bibliography.entries[key] = self
|
230
|
+
parse_names if bibliography.options[:parse_names]
|
231
|
+
parse_months if bibliography.options[:parse_months]
|
230
232
|
self
|
231
233
|
end
|
232
234
|
|
@@ -262,10 +264,10 @@ module BibTeX
|
|
262
264
|
# Parses all name values of the entry. Tries to replace and join the
|
263
265
|
# value prior to parsing.
|
264
266
|
def parse_names
|
267
|
+
strings = bibliography ? bibliography.strings.values : []
|
265
268
|
NAME_FIELDS.each do |key|
|
266
269
|
if name = @fields[key]
|
267
|
-
name.replace(
|
268
|
-
name.join
|
270
|
+
name.replace(strings).join
|
269
271
|
name = name.to_name
|
270
272
|
@fields[key] = name
|
271
273
|
end
|
data/lib/bibtex/lexer.rb
CHANGED
@@ -17,7 +17,7 @@
|
|
17
17
|
#++
|
18
18
|
|
19
19
|
require 'strscan'
|
20
|
-
|
20
|
+
require 'forwardable'
|
21
21
|
|
22
22
|
module BibTeX
|
23
23
|
|
@@ -25,12 +25,19 @@ module BibTeX
|
|
25
25
|
# The BibTeX::Lexer handles the lexical analysis of BibTeX bibliographies.
|
26
26
|
#
|
27
27
|
class Lexer
|
28
|
+
extend Forwardable
|
29
|
+
include Enumerable
|
28
30
|
|
29
|
-
attr_reader :
|
31
|
+
attr_reader :options, :stack, :mode, :scanner
|
32
|
+
attr_writer :mode
|
33
|
+
|
34
|
+
def_delegator :@scanner, :string, :data
|
35
|
+
def_delegators :@stack, :each
|
30
36
|
|
31
37
|
DEFAULTS = { :include => [:errors], :strict => true }.freeze
|
32
38
|
|
33
|
-
|
39
|
+
MODE = Hash.new(:meta).merge(:bibtex => :bibtex, :entry => :bibtex, :string => :bibtex, :preamble => :bibtex, :comment => :bibtex, :meta => :meta, :literal => :literal, :content => :content).freeze
|
40
|
+
|
34
41
|
#
|
35
42
|
# Creates a new instance. Possible options and their respective
|
36
43
|
# default values are:
|
@@ -45,53 +52,38 @@ module BibTeX
|
|
45
52
|
#
|
46
53
|
def initialize(options = {})
|
47
54
|
@options = DEFAULTS.merge(options)
|
48
|
-
|
55
|
+
reset
|
49
56
|
end
|
50
57
|
|
58
|
+
def reset
|
59
|
+
@stack, @brace_level, @mode, @active_object = [], 0, :meta, nil
|
60
|
+
@scanner.reset if @scanner
|
61
|
+
|
62
|
+
# cache options for speed
|
63
|
+
@include_meta_content = @options[:include].include?(:meta_content)
|
64
|
+
@include_errors = @options[:include].include?(:errors)
|
65
|
+
|
66
|
+
self
|
67
|
+
end
|
68
|
+
|
51
69
|
# Sets the source for the lexical analysis and resets the internal state.
|
52
|
-
def data=(
|
53
|
-
@
|
54
|
-
|
55
|
-
@mode = :meta
|
56
|
-
@active_object = nil
|
57
|
-
@data = StringScanner.new(string)
|
58
|
-
|
59
|
-
# @line_breaks = []
|
60
|
-
# @line_breaks << @data.pos until @data.scan_until(/\n|$/).empty?
|
61
|
-
# @data.reset
|
70
|
+
def data=(data)
|
71
|
+
@scanner = StringScanner.new(data)
|
72
|
+
reset
|
62
73
|
end
|
63
74
|
|
64
|
-
|
65
|
-
|
66
|
-
0 # (@line_breaks.find_index { |n| n >= index } || 0) + 1
|
67
|
-
end
|
68
|
-
|
75
|
+
def symbols; @stack.map(&:first); end
|
76
|
+
|
69
77
|
# Returns the next token from the parse stack.
|
70
|
-
def next_token
|
71
|
-
@stack.shift
|
72
|
-
end
|
78
|
+
def next_token; @stack.shift; end
|
73
79
|
|
74
|
-
def mode=(mode)
|
75
|
-
# Log.debug("Lexer: switching to #{mode} mode...")
|
76
|
-
|
77
|
-
@active_object = case
|
78
|
-
when [:comment,:string,:preamble,:entry].include?(mode) then mode
|
79
|
-
when mode == :meta then nil
|
80
|
-
else @active_object
|
81
|
-
end
|
82
|
-
|
83
|
-
@mode = mode
|
84
|
-
end
|
85
|
-
|
86
80
|
# Returns true if the lexer is currenty parsing a BibTeX object.
|
87
81
|
def bibtex_mode?
|
88
|
-
[:bibtex
|
82
|
+
MODE[@mode] == :bibtex
|
89
83
|
end
|
90
84
|
|
91
|
-
|
92
|
-
define_method
|
93
|
-
mode == m.to_sym
|
94
|
-
end
|
85
|
+
[:meta, :literal, :content].each do |m|
|
86
|
+
define_method("#{m}_mode?") { @mode == m }
|
95
87
|
end
|
96
88
|
|
97
89
|
# Returns true if the lexer is currently parsing the given object type.
|
@@ -102,155 +94,142 @@ module BibTeX
|
|
102
94
|
# Returns true if the lexer is currently in strict mode.
|
103
95
|
def strict?; !!(@options[:strict]); end
|
104
96
|
|
105
|
-
# Pushes a value onto the parse stack.
|
97
|
+
# Pushes a value onto the parse stack. Returns the Lexer.
|
106
98
|
def push(value)
|
107
|
-
case
|
108
|
-
|
109
|
-
|
110
|
-
|
111
|
-
|
112
|
-
|
113
|
-
leave_object
|
114
|
-
when value[0] == :META_CONTENT
|
115
|
-
if @options[:include].include?(:meta_content)
|
116
|
-
value[1] = [value[1], line_number_at(@data.pos)]
|
117
|
-
@stack.push(value)
|
99
|
+
case value[0]
|
100
|
+
when :CONTENT, :STRING_LITERAL
|
101
|
+
if !@stack.empty? && value[0] == @stack[-1][0]
|
102
|
+
@stack[-1][1] << value[1]
|
103
|
+
else
|
104
|
+
@stack.push(value)
|
118
105
|
end
|
106
|
+
when :ERROR
|
107
|
+
@stack.push(value) if @include_errors
|
108
|
+
leave_object
|
109
|
+
when :META_CONTENT
|
110
|
+
@stack.push(value) if @include_meta_content
|
119
111
|
else
|
120
|
-
value[1] = [value[1], line_number_at(@data.pos)]
|
121
112
|
@stack.push(value)
|
122
113
|
end
|
114
|
+
|
123
115
|
self
|
124
116
|
end
|
125
117
|
|
126
118
|
# Start the lexical analysis.
|
127
|
-
def analyse(
|
128
|
-
raise(ArgumentError, 'Lexer: failed to start analysis: no source given!')
|
129
|
-
|
130
|
-
|
131
|
-
self.data =
|
132
|
-
@data.reset
|
119
|
+
def analyse(string = nil)
|
120
|
+
raise(ArgumentError, 'Lexer: failed to start analysis: no source given!') unless
|
121
|
+
string || @scanner
|
122
|
+
|
123
|
+
self.data = string || @scanner.string
|
133
124
|
|
134
|
-
|
135
|
-
|
136
|
-
when bibtex_mode?
|
137
|
-
parse_bibtex
|
138
|
-
when meta_mode?
|
139
|
-
parse_meta
|
140
|
-
when content_mode?
|
141
|
-
parse_content
|
142
|
-
when literal_mode?
|
143
|
-
parse_literal
|
144
|
-
end
|
125
|
+
until @scanner.eos?
|
126
|
+
send("parse_#{MODE[@mode]}")
|
145
127
|
end
|
146
128
|
|
147
|
-
|
148
|
-
push [false, '$end']
|
129
|
+
push([false, '$end'])
|
149
130
|
end
|
150
131
|
|
132
|
+
private
|
133
|
+
|
151
134
|
def parse_bibtex
|
152
135
|
case
|
153
|
-
when @
|
154
|
-
when @
|
136
|
+
when @scanner.scan(/[\t\r\n\s]+/o)
|
137
|
+
when @scanner.scan(/\{/o)
|
155
138
|
@brace_level += 1
|
156
|
-
push
|
157
|
-
if
|
158
|
-
|
159
|
-
end
|
160
|
-
when @data.scan(/\}/o)
|
161
|
-
return error_unbalanced_braces if @brace_level < 1
|
139
|
+
push([:LBRACE,'{'])
|
140
|
+
@mode = :content if @brace_level > 1 || @brace_level == 1 && active?(:comment)
|
141
|
+
when @scanner.scan(/\}/o)
|
162
142
|
@brace_level -= 1
|
163
|
-
push
|
164
|
-
leave_object if @brace_level == 0
|
165
|
-
|
166
|
-
|
167
|
-
|
168
|
-
|
169
|
-
|
170
|
-
|
171
|
-
|
172
|
-
|
173
|
-
|
174
|
-
|
175
|
-
|
176
|
-
|
177
|
-
|
178
|
-
|
179
|
-
enter_object
|
180
|
-
when @data.scan(/./o)
|
181
|
-
error_unexpected_token
|
143
|
+
push([:RBRACE,'}'])
|
144
|
+
return leave_object if @brace_level == 0
|
145
|
+
return error_unbalanced_braces if @brace_level < 0
|
146
|
+
when @scanner.scan( /=/o)
|
147
|
+
push([:EQ,'='])
|
148
|
+
when @scanner.scan(/,/o)
|
149
|
+
push([:COMMA,','])
|
150
|
+
when @scanner.scan(/\d+/o)
|
151
|
+
push([:NUMBER,@scanner.matched])
|
152
|
+
when @scanner.scan(/[a-z\d\/:_!$\.%&*-]+/io)
|
153
|
+
push([:NAME,@scanner.matched])
|
154
|
+
when @scanner.scan(/"/o)
|
155
|
+
@mode = :literal
|
156
|
+
when @scanner.scan(/#/o)
|
157
|
+
push([:SHARP,'#'])
|
158
|
+
when @scanner.scan(/@/o)
|
182
159
|
enter_object
|
160
|
+
when @scanner.scan(/./o)
|
161
|
+
error_unexpected_token
|
183
162
|
end
|
184
163
|
end
|
185
164
|
|
186
165
|
def parse_meta
|
187
|
-
match = @
|
188
|
-
|
189
|
-
push
|
166
|
+
match = @scanner.scan_until(strict? ? /@[\t ]*/o : /(^|\n)[\t ]*@[\t ]*/o)
|
167
|
+
if @scanner.matched
|
168
|
+
push([:META_CONTENT,match.chop])
|
190
169
|
enter_object
|
191
170
|
else
|
192
|
-
push
|
193
|
-
@
|
171
|
+
push([:META_CONTENT,@scanner.rest])
|
172
|
+
@scanner.terminate
|
194
173
|
end
|
195
174
|
end
|
196
175
|
|
197
176
|
def parse_content
|
198
|
-
match = @
|
199
|
-
case @
|
177
|
+
match = @scanner.scan_until(/\{|\}/o)
|
178
|
+
case @scanner.matched
|
200
179
|
when '{'
|
201
180
|
@brace_level += 1
|
202
|
-
push
|
181
|
+
push([:CONTENT,match])
|
203
182
|
when '}'
|
204
183
|
@brace_level -= 1
|
205
184
|
case
|
206
|
-
when @brace_level < 0
|
207
|
-
push [:CONTENT,match.chop]
|
208
|
-
error_unbalanced_braces
|
209
185
|
when @brace_level == 0
|
210
|
-
push
|
211
|
-
push
|
186
|
+
push([:CONTENT,match.chop])
|
187
|
+
push([:RBRACE,'}'])
|
212
188
|
leave_object
|
213
|
-
when @brace_level == 1 && active?(:
|
214
|
-
push
|
215
|
-
push
|
216
|
-
|
189
|
+
when @brace_level == 1 && !active?(:comment)
|
190
|
+
push([:CONTENT,match.chop])
|
191
|
+
push([:RBRACE,'}'])
|
192
|
+
@mode = :bibtex
|
193
|
+
when @brace_level < 0
|
194
|
+
push([:CONTENT,match.chop])
|
195
|
+
error_unbalanced_braces
|
217
196
|
else
|
218
|
-
push
|
197
|
+
push([:CONTENT,match])
|
219
198
|
end
|
220
199
|
else
|
221
|
-
push
|
222
|
-
@
|
200
|
+
push([:CONTENT,@scanner.rest])
|
201
|
+
@scanner.terminate
|
223
202
|
error_unterminated_content
|
224
203
|
end
|
225
204
|
end
|
226
205
|
|
227
206
|
def parse_literal
|
228
|
-
match = @
|
229
|
-
case @
|
207
|
+
match = @scanner.scan_until(/[\{\}"\n]/o)
|
208
|
+
case @scanner.matched
|
230
209
|
when '{'
|
231
210
|
@brace_level += 1
|
232
|
-
push
|
211
|
+
push([:STRING_LITERAL,match])
|
233
212
|
when '}'
|
234
213
|
@brace_level -= 1
|
235
214
|
if @brace_level < 1
|
236
|
-
push
|
215
|
+
push([:STRING_LITERAL,match.chop])
|
237
216
|
error_unbalanced_braces
|
238
217
|
else
|
239
|
-
push
|
218
|
+
push([:STRING_LITERAL,match])
|
240
219
|
end
|
241
220
|
when '"'
|
242
221
|
if @brace_level == 1
|
243
|
-
push
|
244
|
-
|
222
|
+
push([:STRING_LITERAL,match.chop])
|
223
|
+
@mode = :bibtex
|
245
224
|
else
|
246
|
-
push
|
225
|
+
push([:STRING_LITERAL,match])
|
247
226
|
end
|
248
227
|
when "\n"
|
249
|
-
push
|
228
|
+
push([:STRING_LITERAL,match.chop])
|
250
229
|
error_unterminated_string
|
251
230
|
else
|
252
|
-
push
|
253
|
-
@
|
231
|
+
push([:STRING_LITERAL,@scanner.rest])
|
232
|
+
@scanner.terminate
|
254
233
|
error_unterminated_string
|
255
234
|
end
|
256
235
|
end
|
@@ -258,63 +237,63 @@ module BibTeX
|
|
258
237
|
# Called when the lexer encounters a new BibTeX object.
|
259
238
|
def enter_object
|
260
239
|
@brace_level = 0
|
261
|
-
self.mode = :bibtex
|
262
240
|
push [:AT,'@']
|
263
241
|
|
264
242
|
case
|
265
|
-
when @
|
266
|
-
|
267
|
-
push [:STRING, @
|
268
|
-
when @
|
269
|
-
|
270
|
-
push [:PREAMBLE,
|
271
|
-
when @
|
272
|
-
|
273
|
-
push [:COMMENT,
|
274
|
-
when @
|
275
|
-
|
276
|
-
push [:NAME, @
|
243
|
+
when @scanner.scan(/string/io)
|
244
|
+
@mode = @active_object = :string
|
245
|
+
push [:STRING, @scanner.matched]
|
246
|
+
when @scanner.scan(/preamble/io)
|
247
|
+
@mode = @active_object = :preamble
|
248
|
+
push [:PREAMBLE, @scanner.matched]
|
249
|
+
when @scanner.scan(/comment/io)
|
250
|
+
@mode = @active_object = :comment
|
251
|
+
push [:COMMENT, @scanner.matched]
|
252
|
+
when @scanner.scan(/[a-z\d:_!\.$%&*-]+/io)
|
253
|
+
@mode = @active_object = :entry
|
254
|
+
push [:NAME, @scanner.matched]
|
255
|
+
else
|
256
|
+
error_unexpected_object
|
277
257
|
end
|
278
258
|
end
|
279
259
|
|
280
260
|
# Called when parser leaves a BibTeX object.
|
281
261
|
def leave_object
|
282
|
-
|
283
|
-
@brace_level = 0
|
262
|
+
@mode, @active_object, @brace_level = :meta, nil, 0
|
284
263
|
end
|
285
264
|
|
286
|
-
|
287
265
|
def error_unbalanced_braces
|
288
|
-
|
289
|
-
|
290
|
-
backtrace [:E_UNBALANCED_BRACES, [self.data.matched,n]]
|
266
|
+
Log.warn("Lexer: unbalanced braces at #{@scanner.pos}; brace level #{@brace_level}; mode #{@mode.inspect}.")
|
267
|
+
backtrace [:E_UNBALANCED, @scanner.matched]
|
291
268
|
end
|
292
269
|
|
293
270
|
def error_unterminated_string
|
294
|
-
|
295
|
-
|
296
|
-
backtrace [:E_UNTERMINATED_STRING, [@data.matched,n]]
|
271
|
+
Log.warn("Lexer: unterminated string at #{@scanner.pos}; brace level #{@brace_level}; mode #{@mode.inspect}.")
|
272
|
+
backtrace [:E_UNTERMINATED_STRING, @scanner.matched]
|
297
273
|
end
|
298
274
|
|
299
275
|
def error_unterminated_content
|
300
|
-
|
301
|
-
|
302
|
-
backtrace [:E_UNTERMINATED_CONTENT, [@data.matched,n]]
|
276
|
+
Log.warn("Lexer: unterminated content at #{@scanner.pos}; brace level #{@brace_level}; mode #{@mode.inspect}.")
|
277
|
+
backtrace [:E_UNTERMINATED_CONTENT, @scanner.matched]
|
303
278
|
end
|
304
279
|
|
305
280
|
def error_unexpected_token
|
306
|
-
|
307
|
-
|
308
|
-
backtrace [:E_UNEXPECTED_TOKEN, [@data.matched,n]]
|
281
|
+
Log.warn("Lexer: unexpected token `#{@scanner.matched}' at #{@scanner.pos}; brace level #{@brace_level}; mode #{@mode.inspect}.")
|
282
|
+
backtrace [:E_UNEXPECTED_TOKEN, @scanner.matched]
|
309
283
|
end
|
310
|
-
|
311
|
-
def
|
312
|
-
|
313
|
-
|
314
|
-
trace << error
|
315
|
-
push [:ERROR,trace]
|
284
|
+
|
285
|
+
def error_unexpected_object
|
286
|
+
Log.warn("Lexer: unexpected object at #{@scanner.pos}; brace level #{@brace_level}; mode #{@mode.inspect}.")
|
287
|
+
backtrace [:E_UNEXPECTED_OBJECT, '@']
|
316
288
|
end
|
317
289
|
|
290
|
+
def backtrace(error)
|
291
|
+
bt = []
|
292
|
+
bt.unshift(@stack.pop) until @stack.empty? || (!bt.empty? && [:AT,:META_CONTENT].include?(bt[0][0]))
|
293
|
+
bt << error
|
294
|
+
push [:ERROR,bt]
|
295
|
+
end
|
296
|
+
|
318
297
|
end
|
319
298
|
|
320
299
|
end
|