1 | # CSV -- module for generating/parsing CSV data.
|
---|
2 | # Copyright (C) 2000-2004 NAKAMURA, Hiroshi <[email protected]>.
|
---|
3 |
|
---|
4 | # $Id: csv.rb 11708 2007-02-12 23:01:19Z shyouhei $
|
---|
5 |
|
---|
6 | # This program is copyrighted free software by NAKAMURA, Hiroshi. You can
|
---|
7 | # redistribute it and/or modify it under the same terms of Ruby's license;
|
---|
8 | # either the dual license version in 2003, or any later version.
|
---|
9 |
|
---|
10 |
|
---|
11 | class CSV
|
---|
12 | class IllegalFormatError < RuntimeError; end
|
---|
13 |
|
---|
14 | # deprecated
|
---|
15 | class Cell < String
|
---|
16 | def initialize(data = "", is_null = false)
|
---|
17 | super(is_null ? "" : data)
|
---|
18 | end
|
---|
19 |
|
---|
20 | def data
|
---|
21 | to_s
|
---|
22 | end
|
---|
23 | end
|
---|
24 |
|
---|
25 | # deprecated
|
---|
26 | class Row < Array
|
---|
27 | end
|
---|
28 |
|
---|
29 | # Open a CSV formatted file for reading or writing.
|
---|
30 | #
|
---|
31 | # For reading.
|
---|
32 | #
|
---|
33 | # EXAMPLE 1
|
---|
34 | # CSV.open('csvfile.csv', 'r') do |row|
|
---|
35 | # p row
|
---|
36 | # end
|
---|
37 | #
|
---|
38 | # EXAMPLE 2
|
---|
39 | # reader = CSV.open('csvfile.csv', 'r')
|
---|
40 | # row1 = reader.shift
|
---|
41 | # row2 = reader.shift
|
---|
42 | # if row2.empty?
|
---|
43 | # p 'row2 not find.'
|
---|
44 | # end
|
---|
45 | # reader.close
|
---|
46 | #
|
---|
47 | # ARGS
|
---|
48 | # filename: filename to parse.
|
---|
49 | # col_sep: Column separator. ?, by default. If you want to separate
|
---|
50 | # fields with semicolon, give ?; here.
|
---|
51 | # row_sep: Row separator. nil by default. nil means "\r\n or \n". If you
|
---|
52 | # want to separate records with \r, give ?\r here.
|
---|
53 | #
|
---|
54 | # RETURNS
|
---|
55 | # reader instance. To get parse result, see CSV::Reader#each.
|
---|
56 | #
|
---|
57 | #
|
---|
58 | # For writing.
|
---|
59 | #
|
---|
60 | # EXAMPLE 1
|
---|
61 | # CSV.open('csvfile.csv', 'w') do |writer|
|
---|
62 | # writer << ['r1c1', 'r1c2']
|
---|
63 | # writer << ['r2c1', 'r2c2']
|
---|
64 | # writer << [nil, nil]
|
---|
65 | # end
|
---|
66 | #
|
---|
67 | # EXAMPLE 2
|
---|
68 | # writer = CSV.open('csvfile.csv', 'w')
|
---|
69 | # writer << ['r1c1', 'r1c2'] << ['r2c1', 'r2c2'] << [nil, nil]
|
---|
70 | # writer.close
|
---|
71 | #
|
---|
72 | # ARGS
|
---|
73 | # filename: filename to generate.
|
---|
74 | # col_sep: Column separator. ?, by default. If you want to separate
|
---|
75 | # fields with semicolon, give ?; here.
|
---|
76 | # row_sep: Row separator. nil by default. nil means "\r\n or \n". If you
|
---|
77 | # want to separate records with \r, give ?\r here.
|
---|
78 | #
|
---|
79 | # RETURNS
|
---|
80 | # writer instance. See CSV::Writer#<< and CSV::Writer#add_row to know how
|
---|
81 | # to generate CSV string.
|
---|
82 | #
|
---|
83 | def CSV.open(path, mode, fs = nil, rs = nil, &block)
|
---|
84 | if mode == 'r' or mode == 'rb'
|
---|
85 | open_reader(path, mode, fs, rs, &block)
|
---|
86 | elsif mode == 'w' or mode == 'wb'
|
---|
87 | open_writer(path, mode, fs, rs, &block)
|
---|
88 | else
|
---|
89 | raise ArgumentError.new("'mode' must be 'r', 'rb', 'w', or 'wb'")
|
---|
90 | end
|
---|
91 | end
|
---|
92 |
|
---|
93 | def CSV.foreach(path, rs = nil, &block)
|
---|
94 | open_reader(path, 'r', ',', rs, &block)
|
---|
95 | end
|
---|
96 |
|
---|
97 | def CSV.read(path, length = nil, offset = nil)
|
---|
98 | CSV.parse(IO.read(path, length, offset))
|
---|
99 | end
|
---|
100 |
|
---|
101 | def CSV.readlines(path, rs = nil)
|
---|
102 | reader = open_reader(path, 'r', ',', rs)
|
---|
103 | begin
|
---|
104 | reader.collect { |row| row }
|
---|
105 | ensure
|
---|
106 | reader.close
|
---|
107 | end
|
---|
108 | end
|
---|
109 |
|
---|
110 | def CSV.generate(path, fs = nil, rs = nil, &block)
|
---|
111 | open_writer(path, 'w', fs, rs, &block)
|
---|
112 | end
|
---|
113 |
|
---|
114 | # Parse lines from given string or stream. Return rows as an Array of Arrays.
|
---|
115 | def CSV.parse(str_or_readable, fs = nil, rs = nil, &block)
|
---|
116 | if File.exist?(str_or_readable)
|
---|
117 | STDERR.puts("CSV.parse(filename) is deprecated." +
|
---|
118 | " Use CSV.open(filename, 'r') instead.")
|
---|
119 | return open_reader(str_or_readable, 'r', fs, rs, &block)
|
---|
120 | end
|
---|
121 | if block
|
---|
122 | CSV::Reader.parse(str_or_readable, fs, rs) do |row|
|
---|
123 | yield(row)
|
---|
124 | end
|
---|
125 | nil
|
---|
126 | else
|
---|
127 | CSV::Reader.create(str_or_readable, fs, rs).collect { |row| row }
|
---|
128 | end
|
---|
129 | end
|
---|
130 |
|
---|
131 | # Parse a line from given string. Bear in mind it parses ONE LINE. Rest of
|
---|
132 | # the string is ignored for example "a,b\r\nc,d" => ['a', 'b'] and the
|
---|
133 | # second line 'c,d' is ignored.
|
---|
134 | #
|
---|
135 | # If you don't know whether a target string to parse is exactly 1 line or
|
---|
136 | # not, use CSV.parse_row instead of this method.
|
---|
137 | def CSV.parse_line(src, fs = nil, rs = nil)
|
---|
138 | fs ||= ','
|
---|
139 | if fs.is_a?(Fixnum)
|
---|
140 | fs = fs.chr
|
---|
141 | end
|
---|
142 | if !rs.nil? and rs.is_a?(Fixnum)
|
---|
143 | rs = rs.chr
|
---|
144 | end
|
---|
145 | idx = 0
|
---|
146 | res_type = :DT_COLSEP
|
---|
147 | row = []
|
---|
148 | begin
|
---|
149 | while res_type == :DT_COLSEP
|
---|
150 | res_type, idx, cell = parse_body(src, idx, fs, rs)
|
---|
151 | row << cell
|
---|
152 | end
|
---|
153 | rescue IllegalFormatError
|
---|
154 | return []
|
---|
155 | end
|
---|
156 | row
|
---|
157 | end
|
---|
158 |
|
---|
159 | # Create a line from cells. each cell is stringified by to_s.
|
---|
160 | def CSV.generate_line(row, fs = nil, rs = nil)
|
---|
161 | if row.size == 0
|
---|
162 | return ''
|
---|
163 | end
|
---|
164 | fs ||= ','
|
---|
165 | if fs.is_a?(Fixnum)
|
---|
166 | fs = fs.chr
|
---|
167 | end
|
---|
168 | if !rs.nil? and rs.is_a?(Fixnum)
|
---|
169 | rs = rs.chr
|
---|
170 | end
|
---|
171 | res_type = :DT_COLSEP
|
---|
172 | result_str = ''
|
---|
173 | idx = 0
|
---|
174 | while true
|
---|
175 | generate_body(row[idx], result_str, fs, rs)
|
---|
176 | idx += 1
|
---|
177 | if (idx == row.size)
|
---|
178 | break
|
---|
179 | end
|
---|
180 | generate_separator(:DT_COLSEP, result_str, fs, rs)
|
---|
181 | end
|
---|
182 | result_str
|
---|
183 | end
|
---|
184 |
|
---|
185 | # Parse a line from string. Consider using CSV.parse_line instead.
|
---|
186 | # To parse lines in CSV string, see EXAMPLE below.
|
---|
187 | #
|
---|
188 | # EXAMPLE
|
---|
189 | # src = "a,b\r\nc,d\r\ne,f"
|
---|
190 | # idx = 0
|
---|
191 | # begin
|
---|
192 | # parsed = []
|
---|
193 | # parsed_cells, idx = CSV.parse_row(src, idx, parsed)
|
---|
194 | # puts "Parsed #{ parsed_cells } cells."
|
---|
195 | # p parsed
|
---|
196 | # end while parsed_cells > 0
|
---|
197 | #
|
---|
198 | # ARGS
|
---|
199 | # src: a CSV data to be parsed. Must respond '[](idx)'.
|
---|
200 | # src[](idx) must return a char. (Not a string such as 'a', but 97).
|
---|
201 | # src[](idx_out_of_bounds) must return nil. A String satisfies this
|
---|
202 | # requirement.
|
---|
203 | # idx: index of parsing location of 'src'. 0 origin.
|
---|
204 | # out_dev: buffer for parsed cells. Must respond '<<(aString)'.
|
---|
205 | # col_sep: Column separator. ?, by default. If you want to separate
|
---|
206 | # fields with semicolon, give ?; here.
|
---|
207 | # row_sep: Row separator. nil by default. nil means "\r\n or \n". If you
|
---|
208 | # want to separate records with \r, give ?\r here.
|
---|
209 | #
|
---|
210 | # RETURNS
|
---|
211 | # parsed_cells: num of parsed cells.
|
---|
212 | # idx: index of next parsing location of 'src'.
|
---|
213 | #
|
---|
214 | def CSV.parse_row(src, idx, out_dev, fs = nil, rs = nil)
|
---|
215 | fs ||= ','
|
---|
216 | if fs.is_a?(Fixnum)
|
---|
217 | fs = fs.chr
|
---|
218 | end
|
---|
219 | if !rs.nil? and rs.is_a?(Fixnum)
|
---|
220 | rs = rs.chr
|
---|
221 | end
|
---|
222 | idx_backup = idx
|
---|
223 | parsed_cells = 0
|
---|
224 | res_type = :DT_COLSEP
|
---|
225 | begin
|
---|
226 | while res_type != :DT_ROWSEP
|
---|
227 | res_type, idx, cell = parse_body(src, idx, fs, rs)
|
---|
228 | if res_type == :DT_EOS
|
---|
229 | if idx == idx_backup #((parsed_cells == 0) and cell.nil?)
|
---|
230 | return 0, 0
|
---|
231 | end
|
---|
232 | res_type = :DT_ROWSEP
|
---|
233 | end
|
---|
234 | parsed_cells += 1
|
---|
235 | out_dev << cell
|
---|
236 | end
|
---|
237 | rescue IllegalFormatError
|
---|
238 | return 0, 0
|
---|
239 | end
|
---|
240 | return parsed_cells, idx
|
---|
241 | end
|
---|
242 |
|
---|
243 | # Convert a line from cells data to string. Consider using CSV.generate_line
|
---|
244 | # instead. To generate multi-row CSV string, see EXAMPLE below.
|
---|
245 | #
|
---|
246 | # EXAMPLE
|
---|
247 | # row1 = ['a', 'b']
|
---|
248 | # row2 = ['c', 'd']
|
---|
249 | # row3 = ['e', 'f']
|
---|
250 | # src = [row1, row2, row3]
|
---|
251 | # buf = ''
|
---|
252 | # src.each do |row|
|
---|
253 | # parsed_cells = CSV.generate_row(row, 2, buf)
|
---|
254 | # puts "Created #{ parsed_cells } cells."
|
---|
255 | # end
|
---|
256 | # p buf
|
---|
257 | #
|
---|
258 | # ARGS
|
---|
259 | # src: an Array of String to be converted to CSV string. Must respond to
|
---|
260 | # 'size' and '[](idx)'. src[idx] must return String.
|
---|
261 | # cells: num of cells in a line.
|
---|
262 | # out_dev: buffer for generated CSV string. Must respond to '<<(string)'.
|
---|
263 | # col_sep: Column separator. ?, by default. If you want to separate
|
---|
264 | # fields with semicolon, give ?; here.
|
---|
265 | # row_sep: Row separator. nil by default. nil means "\r\n or \n". If you
|
---|
266 | # want to separate records with \r, give ?\r here.
|
---|
267 | #
|
---|
268 | # RETURNS
|
---|
269 | # parsed_cells: num of converted cells.
|
---|
270 | #
|
---|
271 | def CSV.generate_row(src, cells, out_dev, fs = nil, rs = nil)
|
---|
272 | fs ||= ','
|
---|
273 | if fs.is_a?(Fixnum)
|
---|
274 | fs = fs.chr
|
---|
275 | end
|
---|
276 | if !rs.nil? and rs.is_a?(Fixnum)
|
---|
277 | rs = rs.chr
|
---|
278 | end
|
---|
279 | src_size = src.size
|
---|
280 | if (src_size == 0)
|
---|
281 | if cells == 0
|
---|
282 | generate_separator(:DT_ROWSEP, out_dev, fs, rs)
|
---|
283 | end
|
---|
284 | return 0
|
---|
285 | end
|
---|
286 | res_type = :DT_COLSEP
|
---|
287 | parsed_cells = 0
|
---|
288 | generate_body(src[parsed_cells], out_dev, fs, rs)
|
---|
289 | parsed_cells += 1
|
---|
290 | while ((parsed_cells < cells) and (parsed_cells != src_size))
|
---|
291 | generate_separator(:DT_COLSEP, out_dev, fs, rs)
|
---|
292 | generate_body(src[parsed_cells], out_dev, fs, rs)
|
---|
293 | parsed_cells += 1
|
---|
294 | end
|
---|
295 | if (parsed_cells == cells)
|
---|
296 | generate_separator(:DT_ROWSEP, out_dev, fs, rs)
|
---|
297 | else
|
---|
298 | generate_separator(:DT_COLSEP, out_dev, fs, rs)
|
---|
299 | end
|
---|
300 | parsed_cells
|
---|
301 | end
|
---|
302 |
|
---|
303 | # Private class methods.
|
---|
304 | class << self
|
---|
305 | private
|
---|
306 |
|
---|
307 | def open_reader(path, mode, fs, rs, &block)
|
---|
308 | file = File.open(path, mode)
|
---|
309 | if block
|
---|
310 | begin
|
---|
311 | CSV::Reader.parse(file, fs, rs) do |row|
|
---|
312 | yield(row)
|
---|
313 | end
|
---|
314 | ensure
|
---|
315 | file.close
|
---|
316 | end
|
---|
317 | nil
|
---|
318 | else
|
---|
319 | reader = CSV::Reader.create(file, fs, rs)
|
---|
320 | reader.close_on_terminate
|
---|
321 | reader
|
---|
322 | end
|
---|
323 | end
|
---|
324 |
|
---|
325 | def open_writer(path, mode, fs, rs, &block)
|
---|
326 | file = File.open(path, mode)
|
---|
327 | if block
|
---|
328 | begin
|
---|
329 | CSV::Writer.generate(file, fs, rs) do |writer|
|
---|
330 | yield(writer)
|
---|
331 | end
|
---|
332 | ensure
|
---|
333 | file.close
|
---|
334 | end
|
---|
335 | nil
|
---|
336 | else
|
---|
337 | writer = CSV::Writer.create(file, fs, rs)
|
---|
338 | writer.close_on_terminate
|
---|
339 | writer
|
---|
340 | end
|
---|
341 | end
|
---|
342 |
|
---|
343 | def parse_body(src, idx, fs, rs)
|
---|
344 | fs_str = fs
|
---|
345 | fs_size = fs_str.size
|
---|
346 | rs_str = rs || "\n"
|
---|
347 | rs_size = rs_str.size
|
---|
348 | fs_idx = rs_idx = 0
|
---|
349 | cell = Cell.new
|
---|
350 | state = :ST_START
|
---|
351 | quoted = cr = false
|
---|
352 | c = nil
|
---|
353 | last_idx = idx
|
---|
354 | while c = src[idx]
|
---|
355 | unless quoted
|
---|
356 | fschar = (c == fs_str[fs_idx])
|
---|
357 | rschar = (c == rs_str[rs_idx])
|
---|
358 | # simple 1 char backtrack
|
---|
359 | if !fschar and c == fs_str[0]
|
---|
360 | fs_idx = 0
|
---|
361 | fschar = true
|
---|
362 | if state == :ST_START
|
---|
363 | state = :ST_DATA
|
---|
364 | elsif state == :ST_QUOTE
|
---|
365 | raise IllegalFormatError
|
---|
366 | end
|
---|
367 | end
|
---|
368 | if !rschar and c == rs_str[0]
|
---|
369 | rs_idx = 0
|
---|
370 | rschar = true
|
---|
371 | if state == :ST_START
|
---|
372 | state = :ST_DATA
|
---|
373 | elsif state == :ST_QUOTE
|
---|
374 | raise IllegalFormatError
|
---|
375 | end
|
---|
376 | end
|
---|
377 | end
|
---|
378 | if c == ?"
|
---|
379 | fs_idx = rs_idx = 0
|
---|
380 | if cr
|
---|
381 | raise IllegalFormatError
|
---|
382 | end
|
---|
383 | cell << src[last_idx, (idx - last_idx)]
|
---|
384 | last_idx = idx
|
---|
385 | if state == :ST_DATA
|
---|
386 | if quoted
|
---|
387 | last_idx += 1
|
---|
388 | quoted = false
|
---|
389 | state = :ST_QUOTE
|
---|
390 | else
|
---|
391 | raise IllegalFormatError
|
---|
392 | end
|
---|
393 | elsif state == :ST_QUOTE
|
---|
394 | cell << c.chr
|
---|
395 | last_idx += 1
|
---|
396 | quoted = true
|
---|
397 | state = :ST_DATA
|
---|
398 | else # :ST_START
|
---|
399 | quoted = true
|
---|
400 | last_idx += 1
|
---|
401 | state = :ST_DATA
|
---|
402 | end
|
---|
403 | elsif fschar or rschar
|
---|
404 | if fschar
|
---|
405 | fs_idx += 1
|
---|
406 | end
|
---|
407 | if rschar
|
---|
408 | rs_idx += 1
|
---|
409 | end
|
---|
410 | sep = nil
|
---|
411 | if fs_idx == fs_size
|
---|
412 | if state == :ST_START and rs_idx > 0 and fs_idx < rs_idx
|
---|
413 | state = :ST_DATA
|
---|
414 | end
|
---|
415 | cell << src[last_idx, (idx - last_idx - (fs_size - 1))]
|
---|
416 | last_idx = idx
|
---|
417 | fs_idx = rs_idx = 0
|
---|
418 | if cr
|
---|
419 | raise IllegalFormatError
|
---|
420 | end
|
---|
421 | sep = :DT_COLSEP
|
---|
422 | elsif rs_idx == rs_size
|
---|
423 | if state == :ST_START and fs_idx > 0 and rs_idx < fs_idx
|
---|
424 | state = :ST_DATA
|
---|
425 | end
|
---|
426 | if !(rs.nil? and cr)
|
---|
427 | cell << src[last_idx, (idx - last_idx - (rs_size - 1))]
|
---|
428 | last_idx = idx
|
---|
429 | end
|
---|
430 | fs_idx = rs_idx = 0
|
---|
431 | sep = :DT_ROWSEP
|
---|
432 | end
|
---|
433 | if sep
|
---|
434 | if state == :ST_DATA
|
---|
435 | return sep, idx + 1, cell;
|
---|
436 | elsif state == :ST_QUOTE
|
---|
437 | return sep, idx + 1, cell;
|
---|
438 | else # :ST_START
|
---|
439 | return sep, idx + 1, nil
|
---|
440 | end
|
---|
441 | end
|
---|
442 | elsif rs.nil? and c == ?\r
|
---|
443 | # special \r treatment for backward compatibility
|
---|
444 | fs_idx = rs_idx = 0
|
---|
445 | if cr
|
---|
446 | raise IllegalFormatError
|
---|
447 | end
|
---|
448 | cell << src[last_idx, (idx - last_idx)]
|
---|
449 | last_idx = idx
|
---|
450 | if quoted
|
---|
451 | state = :ST_DATA
|
---|
452 | else
|
---|
453 | cr = true
|
---|
454 | end
|
---|
455 | else
|
---|
456 | fs_idx = rs_idx = 0
|
---|
457 | if state == :ST_DATA or state == :ST_START
|
---|
458 | if cr
|
---|
459 | raise IllegalFormatError
|
---|
460 | end
|
---|
461 | state = :ST_DATA
|
---|
462 | else # :ST_QUOTE
|
---|
463 | raise IllegalFormatError
|
---|
464 | end
|
---|
465 | end
|
---|
466 | idx += 1
|
---|
467 | end
|
---|
468 | if state == :ST_START
|
---|
469 | if fs_idx > 0 or rs_idx > 0
|
---|
470 | state = :ST_DATA
|
---|
471 | else
|
---|
472 | return :DT_EOS, idx, nil
|
---|
473 | end
|
---|
474 | elsif quoted
|
---|
475 | raise IllegalFormatError
|
---|
476 | elsif cr
|
---|
477 | raise IllegalFormatError
|
---|
478 | end
|
---|
479 | cell << src[last_idx, (idx - last_idx)]
|
---|
480 | last_idx = idx
|
---|
481 | return :DT_EOS, idx, cell
|
---|
482 | end
|
---|
483 |
|
---|
484 | def generate_body(cell, out_dev, fs, rs)
|
---|
485 | if cell.nil?
|
---|
486 | # empty
|
---|
487 | else
|
---|
488 | cell = cell.to_s
|
---|
489 | row_data = cell.dup
|
---|
490 | if (row_data.gsub!('"', '""') or
|
---|
491 | row_data.index(fs) or
|
---|
492 | (rs and row_data.index(rs)) or
|
---|
493 | (/[\r\n]/ =~ row_data) or
|
---|
494 | (cell.empty?))
|
---|
495 | out_dev << '"' << row_data << '"'
|
---|
496 | else
|
---|
497 | out_dev << row_data
|
---|
498 | end
|
---|
499 | end
|
---|
500 | end
|
---|
501 |
|
---|
502 | def generate_separator(type, out_dev, fs, rs)
|
---|
503 | case type
|
---|
504 | when :DT_COLSEP
|
---|
505 | out_dev << fs
|
---|
506 | when :DT_ROWSEP
|
---|
507 | out_dev << (rs || "\n")
|
---|
508 | end
|
---|
509 | end
|
---|
510 | end
|
---|
511 |
|
---|
512 |
|
---|
513 | # CSV formatted string/stream reader.
|
---|
514 | #
|
---|
515 | # EXAMPLE
|
---|
516 | # read CSV lines untill the first column is 'stop'.
|
---|
517 | #
|
---|
518 | # CSV::Reader.parse(File.open('bigdata', 'rb')) do |row|
|
---|
519 | # p row
|
---|
520 | # break if !row[0].is_null && row[0].data == 'stop'
|
---|
521 | # end
|
---|
522 | #
|
---|
523 | class Reader
|
---|
524 | include Enumerable
|
---|
525 |
|
---|
526 | # Parse CSV data and get lines. Given block is called for each parsed row.
|
---|
527 | # Block value is always nil. Rows are not cached for performance reason.
|
---|
528 | def Reader.parse(str_or_readable, fs = ',', rs = nil, &block)
|
---|
529 | reader = Reader.create(str_or_readable, fs, rs)
|
---|
530 | if block
|
---|
531 | reader.each do |row|
|
---|
532 | yield(row)
|
---|
533 | end
|
---|
534 | reader.close
|
---|
535 | nil
|
---|
536 | else
|
---|
537 | reader
|
---|
538 | end
|
---|
539 | end
|
---|
540 |
|
---|
541 | # Returns reader instance.
|
---|
542 | def Reader.create(str_or_readable, fs = ',', rs = nil)
|
---|
543 | case str_or_readable
|
---|
544 | when IO
|
---|
545 | IOReader.new(str_or_readable, fs, rs)
|
---|
546 | when String
|
---|
547 | StringReader.new(str_or_readable, fs, rs)
|
---|
548 | else
|
---|
549 | IOReader.new(str_or_readable, fs, rs)
|
---|
550 | end
|
---|
551 | end
|
---|
552 |
|
---|
553 | def each
|
---|
554 | while true
|
---|
555 | row = []
|
---|
556 | parsed_cells = get_row(row)
|
---|
557 | if parsed_cells == 0
|
---|
558 | break
|
---|
559 | end
|
---|
560 | yield(row)
|
---|
561 | end
|
---|
562 | nil
|
---|
563 | end
|
---|
564 |
|
---|
565 | def shift
|
---|
566 | row = []
|
---|
567 | parsed_cells = get_row(row)
|
---|
568 | row
|
---|
569 | end
|
---|
570 |
|
---|
571 | def close
|
---|
572 | terminate
|
---|
573 | end
|
---|
574 |
|
---|
575 | private
|
---|
576 |
|
---|
577 | def initialize(dev)
|
---|
578 | raise RuntimeError.new('Do not instanciate this class directly.')
|
---|
579 | end
|
---|
580 |
|
---|
581 | def get_row(row)
|
---|
582 | raise NotImplementedError.new('Method get_row must be defined in a derived class.')
|
---|
583 | end
|
---|
584 |
|
---|
585 | def terminate
|
---|
586 | # Define if needed.
|
---|
587 | end
|
---|
588 | end
|
---|
589 |
|
---|
590 |
|
---|
591 | class StringReader < Reader
|
---|
592 | def initialize(string, fs = ',', rs = nil)
|
---|
593 | @fs = fs
|
---|
594 | @rs = rs
|
---|
595 | @dev = string
|
---|
596 | @idx = 0
|
---|
597 | if @dev[0, 3] == "\xef\xbb\xbf"
|
---|
598 | @idx += 3
|
---|
599 | end
|
---|
600 | end
|
---|
601 |
|
---|
602 | private
|
---|
603 |
|
---|
604 | def get_row(row)
|
---|
605 | parsed_cells, next_idx = CSV.parse_row(@dev, @idx, row, @fs, @rs)
|
---|
606 | if parsed_cells == 0 and next_idx == 0 and @idx != @dev.size
|
---|
607 | raise IllegalFormatError.new
|
---|
608 | end
|
---|
609 | @idx = next_idx
|
---|
610 | parsed_cells
|
---|
611 | end
|
---|
612 | end
|
---|
613 |
|
---|
614 |
|
---|
615 | class IOReader < Reader
|
---|
616 | def initialize(io, fs = ',', rs = nil)
|
---|
617 | @io = io
|
---|
618 | @fs = fs
|
---|
619 | @rs = rs
|
---|
620 | @dev = CSV::IOBuf.new(@io)
|
---|
621 | @idx = 0
|
---|
622 | if @dev[0] == 0xef and @dev[1] == 0xbb and @dev[2] == 0xbf
|
---|
623 | @idx += 3
|
---|
624 | end
|
---|
625 | @close_on_terminate = false
|
---|
626 | end
|
---|
627 |
|
---|
628 | # Tell this reader to close the IO when terminated (Triggered by invoking
|
---|
629 | # CSV::IOReader#close).
|
---|
630 | def close_on_terminate
|
---|
631 | @close_on_terminate = true
|
---|
632 | end
|
---|
633 |
|
---|
634 | private
|
---|
635 |
|
---|
636 | def get_row(row)
|
---|
637 | parsed_cells, next_idx = CSV.parse_row(@dev, @idx, row, @fs, @rs)
|
---|
638 | if parsed_cells == 0 and next_idx == 0 and [email protected]_eos?
|
---|
639 | raise IllegalFormatError.new
|
---|
640 | end
|
---|
641 | dropped = @dev.drop(next_idx)
|
---|
642 | @idx = next_idx - dropped
|
---|
643 | parsed_cells
|
---|
644 | end
|
---|
645 |
|
---|
646 | def terminate
|
---|
647 | if @close_on_terminate
|
---|
648 | @io.close
|
---|
649 | end
|
---|
650 |
|
---|
651 | if @dev
|
---|
652 | @dev.close
|
---|
653 | end
|
---|
654 | end
|
---|
655 | end
|
---|
656 |
|
---|
657 |
|
---|
658 | # CSV formatted string/stream writer.
|
---|
659 | #
|
---|
660 | # EXAMPLE
|
---|
661 | # Write rows to 'csvout' file.
|
---|
662 | #
|
---|
663 | # outfile = File.open('csvout', 'wb')
|
---|
664 | # CSV::Writer.generate(outfile) do |csv|
|
---|
665 | # csv << ['c1', nil, '', '"', "\r\n", 'c2']
|
---|
666 | # ...
|
---|
667 | # end
|
---|
668 | #
|
---|
669 | # outfile.close
|
---|
670 | #
|
---|
671 | class Writer
|
---|
672 | # Given block is called with the writer instance. str_or_writable must
|
---|
673 | # handle '<<(string)'.
|
---|
674 | def Writer.generate(str_or_writable, fs = ',', rs = nil, &block)
|
---|
675 | writer = Writer.create(str_or_writable, fs, rs)
|
---|
676 | if block
|
---|
677 | yield(writer)
|
---|
678 | writer.close
|
---|
679 | nil
|
---|
680 | else
|
---|
681 | writer
|
---|
682 | end
|
---|
683 | end
|
---|
684 |
|
---|
685 | # str_or_writable must handle '<<(string)'.
|
---|
686 | def Writer.create(str_or_writable, fs = ',', rs = nil)
|
---|
687 | BasicWriter.new(str_or_writable, fs, rs)
|
---|
688 | end
|
---|
689 |
|
---|
690 | # dump CSV stream to the device. argument must be an Array of String.
|
---|
691 | def <<(row)
|
---|
692 | CSV.generate_row(row, row.size, @dev, @fs, @rs)
|
---|
693 | self
|
---|
694 | end
|
---|
695 | alias add_row <<
|
---|
696 |
|
---|
697 | def close
|
---|
698 | terminate
|
---|
699 | end
|
---|
700 |
|
---|
701 | private
|
---|
702 |
|
---|
703 | def initialize(dev)
|
---|
704 | raise RuntimeError.new('Do not instanciate this class directly.')
|
---|
705 | end
|
---|
706 |
|
---|
707 | def terminate
|
---|
708 | # Define if needed.
|
---|
709 | end
|
---|
710 | end
|
---|
711 |
|
---|
712 |
|
---|
713 | class BasicWriter < Writer
|
---|
714 | def initialize(str_or_writable, fs = ',', rs = nil)
|
---|
715 | @fs = fs
|
---|
716 | @rs = rs
|
---|
717 | @dev = str_or_writable
|
---|
718 | @close_on_terminate = false
|
---|
719 | end
|
---|
720 |
|
---|
721 | # Tell this writer to close the IO when terminated (Triggered by invoking
|
---|
722 | # CSV::BasicWriter#close).
|
---|
723 | def close_on_terminate
|
---|
724 | @close_on_terminate = true
|
---|
725 | end
|
---|
726 |
|
---|
727 | private
|
---|
728 |
|
---|
729 | def terminate
|
---|
730 | if @close_on_terminate
|
---|
731 | @dev.close
|
---|
732 | end
|
---|
733 | end
|
---|
734 | end
|
---|
735 |
|
---|
736 | private
|
---|
737 |
|
---|
738 | # Buffered stream.
|
---|
739 | #
|
---|
740 | # EXAMPLE 1 -- an IO.
|
---|
741 | # class MyBuf < StreamBuf
|
---|
742 | # # Do initialize myself before a super class. Super class might call my
|
---|
743 | # # method 'read'. (Could be awful for C++ user. :-)
|
---|
744 | # def initialize(s)
|
---|
745 | # @s = s
|
---|
746 | # super()
|
---|
747 | # end
|
---|
748 | #
|
---|
749 | # # define my own 'read' method.
|
---|
750 | # # CAUTION: Returning nil means EnfOfStream.
|
---|
751 | # def read(size)
|
---|
752 | # @s.read(size)
|
---|
753 | # end
|
---|
754 | #
|
---|
755 | # # release buffers. in Ruby which has GC, you do not have to call this...
|
---|
756 | # def terminate
|
---|
757 | # @s = nil
|
---|
758 | # super()
|
---|
759 | # end
|
---|
760 | # end
|
---|
761 | #
|
---|
762 | # buf = MyBuf.new(STDIN)
|
---|
763 | # my_str = ''
|
---|
764 | # p buf[0, 0] # => '' (null string)
|
---|
765 | # p buf[0] # => 97 (char code of 'a')
|
---|
766 | # p buf[0, 1] # => 'a'
|
---|
767 | # my_str = buf[0, 5]
|
---|
768 | # p my_str # => 'abcde' (5 chars)
|
---|
769 | # p buf[0, 6] # => "abcde\n" (6 chars)
|
---|
770 | # p buf[0, 7] # => "abcde\n" (6 chars)
|
---|
771 | # p buf.drop(3) # => 3 (dropped chars)
|
---|
772 | # p buf.get(0, 2) # => 'de' (2 chars)
|
---|
773 | # p buf.is_eos? # => false (is not EOS here)
|
---|
774 | # p buf.drop(5) # => 3 (dropped chars)
|
---|
775 | # p buf.is_eos? # => true (is EOS here)
|
---|
776 | # p buf[0] # => nil (is EOS here)
|
---|
777 | #
|
---|
778 | # EXAMPLE 2 -- String.
|
---|
779 | # This is a conceptual example. No pros with this.
|
---|
780 | #
|
---|
781 | # class StrBuf < StreamBuf
|
---|
782 | # def initialize(s)
|
---|
783 | # @str = s
|
---|
784 | # @idx = 0
|
---|
785 | # super()
|
---|
786 | # end
|
---|
787 | #
|
---|
788 | # def read(size)
|
---|
789 | # str = @str[@idx, size]
|
---|
790 | # @idx += str.size
|
---|
791 | # str
|
---|
792 | # end
|
---|
793 | # end
|
---|
794 | #
|
---|
795 | class StreamBuf
|
---|
796 | # get a char or a partial string from the stream.
|
---|
797 | # idx: index of a string to specify a start point of a string to get.
|
---|
798 | # unlike String instance, idx < 0 returns nil.
|
---|
799 | # n: size of a string to get.
|
---|
800 | # returns char at idx if n == nil.
|
---|
801 | # returns a partial string, from idx to (idx + n) if n != nil. at EOF,
|
---|
802 | # the string size could not equal to arg n.
|
---|
803 | def [](idx, n = nil)
|
---|
804 | if idx < 0
|
---|
805 | return nil
|
---|
806 | end
|
---|
807 | if (idx_is_eos?(idx))
|
---|
808 | if n and (@offset + idx == buf_size(@cur_buf))
|
---|
809 | # Like a String, 'abc'[4, 1] returns nil and
|
---|
810 | # 'abc'[3, 1] returns '' not nil.
|
---|
811 | return ''
|
---|
812 | else
|
---|
813 | return nil
|
---|
814 | end
|
---|
815 | end
|
---|
816 | my_buf = @cur_buf
|
---|
817 | my_offset = @offset
|
---|
818 | next_idx = idx
|
---|
819 | while (my_offset + next_idx >= buf_size(my_buf))
|
---|
820 | if (my_buf == @buf_tail_idx)
|
---|
821 | unless add_buf
|
---|
822 | break
|
---|
823 | end
|
---|
824 | end
|
---|
825 | next_idx = my_offset + next_idx - buf_size(my_buf)
|
---|
826 | my_buf += 1
|
---|
827 | my_offset = 0
|
---|
828 | end
|
---|
829 | loc = my_offset + next_idx
|
---|
830 | if !n
|
---|
831 | return @buf_list[my_buf][loc] # Fixnum of char code.
|
---|
832 | elsif (loc + n - 1 < buf_size(my_buf))
|
---|
833 | return @buf_list[my_buf][loc, n] # String.
|
---|
834 | else # should do loop insted of (tail) recursive call...
|
---|
835 | res = @buf_list[my_buf][loc, BufSize]
|
---|
836 | size_added = buf_size(my_buf) - loc
|
---|
837 | if size_added > 0
|
---|
838 | idx += size_added
|
---|
839 | n -= size_added
|
---|
840 | ret = self[idx, n]
|
---|
841 | if ret
|
---|
842 | res << ret
|
---|
843 | end
|
---|
844 | end
|
---|
845 | return res
|
---|
846 | end
|
---|
847 | end
|
---|
848 | alias get []
|
---|
849 |
|
---|
850 | # drop a string from the stream.
|
---|
851 | # returns dropped size. at EOF, dropped size might not equals to arg n.
|
---|
852 | # Once you drop the head of the stream, access to the dropped part via []
|
---|
853 | # or get returns nil.
|
---|
854 | def drop(n)
|
---|
855 | if is_eos?
|
---|
856 | return 0
|
---|
857 | end
|
---|
858 | size_dropped = 0
|
---|
859 | while (n > 0)
|
---|
860 | if !@is_eos or (@cur_buf != @buf_tail_idx)
|
---|
861 | if (@offset + n < buf_size(@cur_buf))
|
---|
862 | size_dropped += n
|
---|
863 | @offset += n
|
---|
864 | n = 0
|
---|
865 | else
|
---|
866 | size = buf_size(@cur_buf) - @offset
|
---|
867 | size_dropped += size
|
---|
868 | n -= size
|
---|
869 | @offset = 0
|
---|
870 | unless rel_buf
|
---|
871 | unless add_buf
|
---|
872 | break
|
---|
873 | end
|
---|
874 | @cur_buf = @buf_tail_idx
|
---|
875 | end
|
---|
876 | end
|
---|
877 | end
|
---|
878 | end
|
---|
879 | size_dropped
|
---|
880 | end
|
---|
881 |
|
---|
882 | def is_eos?
|
---|
883 | return idx_is_eos?(0)
|
---|
884 | end
|
---|
885 |
|
---|
886 | # WARN: Do not instantiate this class directly. Define your own class
|
---|
887 | # which derives this class and define 'read' instance method.
|
---|
888 | def initialize
|
---|
889 | @buf_list = []
|
---|
890 | @cur_buf = @buf_tail_idx = -1
|
---|
891 | @offset = 0
|
---|
892 | @is_eos = false
|
---|
893 | add_buf
|
---|
894 | @cur_buf = @buf_tail_idx
|
---|
895 | end
|
---|
896 |
|
---|
897 | protected
|
---|
898 |
|
---|
899 | def terminate
|
---|
900 | while (rel_buf); end
|
---|
901 | end
|
---|
902 |
|
---|
903 | # protected method 'read' must be defined in derived classes.
|
---|
904 | # CAUTION: Returning a string which size is not equal to 'size' means
|
---|
905 | # EnfOfStream. When it is not at EOS, you must block the callee, try to
|
---|
906 | # read and return the sized string.
|
---|
907 | def read(size) # raise EOFError
|
---|
908 | raise NotImplementedError.new('Method read must be defined in a derived class.')
|
---|
909 | end
|
---|
910 |
|
---|
911 | private
|
---|
912 |
|
---|
913 | def buf_size(idx)
|
---|
914 | @buf_list[idx].size
|
---|
915 | end
|
---|
916 |
|
---|
917 | def add_buf
|
---|
918 | if @is_eos
|
---|
919 | return false
|
---|
920 | end
|
---|
921 | begin
|
---|
922 | str_read = read(BufSize)
|
---|
923 | rescue EOFError
|
---|
924 | str_read = nil
|
---|
925 | rescue
|
---|
926 | terminate
|
---|
927 | raise
|
---|
928 | end
|
---|
929 | if str_read.nil?
|
---|
930 | @is_eos = true
|
---|
931 | @buf_list.push('')
|
---|
932 | @buf_tail_idx += 1
|
---|
933 | false
|
---|
934 | else
|
---|
935 | @buf_list.push(str_read)
|
---|
936 | @buf_tail_idx += 1
|
---|
937 | true
|
---|
938 | end
|
---|
939 | end
|
---|
940 |
|
---|
941 | def rel_buf
|
---|
942 | if (@cur_buf < 0)
|
---|
943 | return false
|
---|
944 | end
|
---|
945 | @buf_list[@cur_buf] = nil
|
---|
946 | if (@cur_buf == @buf_tail_idx)
|
---|
947 | @cur_buf = -1
|
---|
948 | return false
|
---|
949 | else
|
---|
950 | @cur_buf += 1
|
---|
951 | return true
|
---|
952 | end
|
---|
953 | end
|
---|
954 |
|
---|
955 | def idx_is_eos?(idx)
|
---|
956 | (@is_eos and ((@cur_buf < 0) or (@cur_buf == @buf_tail_idx)))
|
---|
957 | end
|
---|
958 |
|
---|
959 | BufSize = 1024 * 8
|
---|
960 | end
|
---|
961 |
|
---|
962 | # Buffered IO.
|
---|
963 | #
|
---|
964 | # EXAMPLE
|
---|
965 | # # File 'bigdata' could be a giga-byte size one!
|
---|
966 | # buf = CSV::IOBuf.new(File.open('bigdata', 'rb'))
|
---|
967 | # CSV::Reader.new(buf).each do |row|
|
---|
968 | # p row
|
---|
969 | # break if row[0].data == 'admin'
|
---|
970 | # end
|
---|
971 | #
|
---|
972 | class IOBuf < StreamBuf
|
---|
973 | def initialize(s)
|
---|
974 | @s = s
|
---|
975 | super()
|
---|
976 | end
|
---|
977 |
|
---|
978 | def close
|
---|
979 | terminate
|
---|
980 | end
|
---|
981 |
|
---|
982 | private
|
---|
983 |
|
---|
984 | def read(size)
|
---|
985 | @s.read(size)
|
---|
986 | end
|
---|
987 |
|
---|
988 | def terminate
|
---|
989 | super()
|
---|
990 | end
|
---|
991 | end
|
---|
992 | end
|
---|