hmcgowan-roo 1.3.5 → 1.3.6
Sign up to get free protection for your applications and to get access to all the features.
- data/README.markdown +8 -2
- data/lib/roo/excel.rb +45 -80
- data/lib/roo/excelx.rb +1 -5
- data/lib/roo/generic_spreadsheet.rb +2 -12
- data/lib/roo/google.rb +1 -43
- data/lib/roo/openoffice.rb +1 -5
- data/lib/roo/version.rb +1 -1
- data/lib/roo.rb +18 -0
- data/test/1900_base.xls +0 -0
- data/test/1904_base.xls +0 -0
- data/test/skipped_tests.rb +789 -0
- data/test/test_roo.rb +572 -3789
- data/test/whitespace.ods +0 -0
- data/test/whitespace.xls +0 -0
- data/test/whitespace.xlsx +0 -0
- metadata +8 -3
- data/test/numbers1_excel.csv +0 -18
data/README.markdown
CHANGED
@@ -1,6 +1,6 @@
|
|
1
1
|
# README for Roo
|
2
2
|
|
3
|
-
|
3
|
+
1.3.5 is now available on Rubyforge. You can install the official release with 'gem install roo' or refer to the installation instructions below for the latest development gem.
|
4
4
|
|
5
5
|
|
6
6
|
## Installation
|
@@ -35,7 +35,7 @@ This is now the official roo repository and I'll be making a release on RubyForg
|
|
35
35
|
s.cell(1,'A',s.sheets[0]) # same cell
|
36
36
|
|
37
37
|
# almost all methods have an optional argument 'sheet'.
|
38
|
-
# If this parameter is
|
38
|
+
# If this parameter is omitted, the default_sheet will be used.
|
39
39
|
|
40
40
|
s.info # prints infos about the spreadsheet file
|
41
41
|
|
@@ -44,6 +44,12 @@ This is now the official roo repository and I'll be making a release on RubyForg
|
|
44
44
|
s.first_column # the number of the first column
|
45
45
|
s.last_column # the number of the last column
|
46
46
|
|
47
|
+
# limited font information is available
|
48
|
+
|
49
|
+
s.font(1,1).bold?
|
50
|
+
s.font(1,1).italic?
|
51
|
+
s.font(1,1).underline?
|
52
|
+
|
47
53
|
|
48
54
|
see http://roo.rubyforge.org for a more complete tutorial
|
49
55
|
|
data/lib/roo/excel.rb
CHANGED
@@ -6,6 +6,26 @@ rescue LoadError => e
|
|
6
6
|
false
|
7
7
|
end
|
8
8
|
|
9
|
+
# The Spreadsheet library has a bug in handling Excel
|
10
|
+
# base dates so if the file is a 1904 base date then
|
11
|
+
# dates are off by a day. 1900 base dates work fine
|
12
|
+
module Spreadsheet
|
13
|
+
module Excel
|
14
|
+
class Row < Spreadsheet::Row
|
15
|
+
def _date data # :nodoc:
|
16
|
+
return data if data.is_a?(Date)
|
17
|
+
date = @worksheet.date_base + data.to_i
|
18
|
+
if LEAP_ERROR > @worksheet.date_base
|
19
|
+
date -= 1
|
20
|
+
end
|
21
|
+
date
|
22
|
+
end
|
23
|
+
public :_datetime
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
27
|
+
|
28
|
+
|
9
29
|
# ruby-spreadsheet has a font object so we're extending it
|
10
30
|
# with our own functionality but still providing full access
|
11
31
|
# to the user for other font information
|
@@ -55,11 +75,7 @@ class Excel < GenericSpreadsheet
|
|
55
75
|
raise IOError, "file #{@filename} does not exist"
|
56
76
|
end
|
57
77
|
@workbook = Spreadsheet.open(filename)
|
58
|
-
@default_sheet =
|
59
|
-
# no need to set default_sheet if there is only one sheet in the document
|
60
|
-
if self.sheets.size == 1
|
61
|
-
@default_sheet = self.sheets.first
|
62
|
-
end
|
78
|
+
@default_sheet = self.sheets.first
|
63
79
|
ensure
|
64
80
|
#if ENV["roo_local"] != "thomas-p"
|
65
81
|
FileUtils::rm_r(@tmpdir)
|
@@ -128,38 +144,6 @@ class Excel < GenericSpreadsheet
|
|
128
144
|
end
|
129
145
|
end
|
130
146
|
|
131
|
-
# returns the first non empty column
|
132
|
-
def first_column(sheet=nil)
|
133
|
-
sheet = @default_sheet unless sheet
|
134
|
-
return @first_column[sheet] if @first_column[sheet]
|
135
|
-
fr, lr, fc, lc = get_firsts_lasts(sheet)
|
136
|
-
fc
|
137
|
-
end
|
138
|
-
|
139
|
-
# returns the last non empty column
|
140
|
-
def last_column(sheet=nil)
|
141
|
-
sheet = @default_sheet unless sheet
|
142
|
-
return @last_column[sheet] if @last_column[sheet]
|
143
|
-
fr, lr, fc, lc = get_firsts_lasts(sheet)
|
144
|
-
lc
|
145
|
-
end
|
146
|
-
|
147
|
-
# returns the first non empty row
|
148
|
-
def first_row(sheet=nil)
|
149
|
-
sheet = @default_sheet unless sheet
|
150
|
-
return @first_row[sheet] if @first_row[sheet]
|
151
|
-
fr, lr, fc, lc = get_firsts_lasts(sheet)
|
152
|
-
fr
|
153
|
-
end
|
154
|
-
|
155
|
-
# returns the last non empty row
|
156
|
-
def last_row(sheet=nil)
|
157
|
-
sheet = @default_sheet unless sheet
|
158
|
-
return @last_row[sheet] if @last_row[sheet]
|
159
|
-
fr, lr, fc, lc = get_firsts_lasts(sheet)
|
160
|
-
lr
|
161
|
-
end
|
162
|
-
|
163
147
|
# returns NO formula in excel spreadsheets
|
164
148
|
def formula(row,col,sheet=nil)
|
165
149
|
raise EXCEL_NO_FORMULAS
|
@@ -192,42 +176,6 @@ class Excel < GenericSpreadsheet
|
|
192
176
|
end
|
193
177
|
|
194
178
|
private
|
195
|
-
# determine the first and last boundaries
|
196
|
-
def get_firsts_lasts(sheet=nil)
|
197
|
-
|
198
|
-
# 2008-09-14 BEGINf
|
199
|
-
fr=lr=fc=lc=nil
|
200
|
-
sheet = @default_sheet unless sheet
|
201
|
-
if ! @cells_read[sheet]
|
202
|
-
read_cells(sheet)
|
203
|
-
end
|
204
|
-
if @cell[sheet] # nur wenn ueberhaupt Zellen belegt sind
|
205
|
-
@cell[sheet].each {|cellitem|
|
206
|
-
key = cellitem.first
|
207
|
-
y,x = key
|
208
|
-
|
209
|
-
if cellitem[1].class != String or
|
210
|
-
(cellitem[1].class == String and cellitem[1] != "")
|
211
|
-
fr = y unless fr
|
212
|
-
fr = y if y < fr
|
213
|
-
|
214
|
-
lr = y unless lr
|
215
|
-
lr = y if y > lr
|
216
|
-
|
217
|
-
fc = x unless fc
|
218
|
-
fc = x if x < fc
|
219
|
-
|
220
|
-
lc = x unless lc
|
221
|
-
lc = x if x > lc
|
222
|
-
end
|
223
|
-
}
|
224
|
-
end
|
225
|
-
@first_row[sheet] = fr
|
226
|
-
@last_row[sheet] = lr
|
227
|
-
@first_column[sheet] = fc
|
228
|
-
@last_column[sheet] = lc
|
229
|
-
return fr, lr, fc, lc
|
230
|
-
end
|
231
179
|
|
232
180
|
# converts name of a sheet to index (0,1,2,..)
|
233
181
|
def sheet_no(name)
|
@@ -355,7 +303,7 @@ class Excel < GenericSpreadsheet
|
|
355
303
|
(0..row.size).each do |cell_index|
|
356
304
|
cell = row.at(cell_index)
|
357
305
|
next if cell.nil? #skip empty cells
|
358
|
-
next if cell.class == Spreadsheet::Formula
|
306
|
+
next if cell.class == Spreadsheet::Formula && cell.value.nil? # skip empty formla cells
|
359
307
|
if date_or_time?(row, cell_index)
|
360
308
|
vt, v = read_cell_date_or_time(row, cell_index)
|
361
309
|
else
|
@@ -371,12 +319,20 @@ class Excel < GenericSpreadsheet
|
|
371
319
|
end # worksheet
|
372
320
|
@cells_read[sheet] = true
|
373
321
|
end
|
374
|
-
|
322
|
+
|
323
|
+
# Get the contents of a cell, accounting for the
|
324
|
+
# way formula stores the value
|
325
|
+
def read_cell_content(row, idx)
|
326
|
+
cell = row.at(idx)
|
327
|
+
cell = cell.value if cell.class == Spreadsheet::Formula
|
328
|
+
cell
|
329
|
+
end
|
330
|
+
|
375
331
|
# Test the cell to see if it's a valid date/time.
|
376
332
|
def date_or_time?(row, idx)
|
377
333
|
format = row.format(idx)
|
378
334
|
if format.date_or_time?
|
379
|
-
cell = row
|
335
|
+
cell = read_cell_content(row, idx)
|
380
336
|
true if Float(cell) > 0 rescue false
|
381
337
|
else
|
382
338
|
false
|
@@ -387,7 +343,8 @@ class Excel < GenericSpreadsheet
|
|
387
343
|
# Read the date-time cell and convert to,
|
388
344
|
# the date-time values for Roo
|
389
345
|
def read_cell_date_or_time(row, idx)
|
390
|
-
cell = row
|
346
|
+
cell = read_cell_content(row, idx)
|
347
|
+
cell = cell.to_s.to_f
|
391
348
|
if cell < 1.0
|
392
349
|
value_type = :time
|
393
350
|
f = cell*24.0*60.0*60.0
|
@@ -399,7 +356,11 @@ class Excel < GenericSpreadsheet
|
|
399
356
|
s = secs
|
400
357
|
value = h*3600+m*60+s
|
401
358
|
else
|
402
|
-
|
359
|
+
if row.at(idx).class == Spreadsheet::Formula
|
360
|
+
datetime = row._datetime(cell)
|
361
|
+
else
|
362
|
+
datetime = row.datetime(idx)
|
363
|
+
end
|
403
364
|
if datetime.hour != 0 or
|
404
365
|
datetime.min != 0 or
|
405
366
|
datetime.sec != 0
|
@@ -407,7 +368,11 @@ class Excel < GenericSpreadsheet
|
|
407
368
|
value = datetime
|
408
369
|
else
|
409
370
|
value_type = :date
|
410
|
-
|
371
|
+
if row.at(idx).class == Spreadsheet::Formula
|
372
|
+
value = row._date(cell)
|
373
|
+
else
|
374
|
+
value = row.date(idx)
|
375
|
+
end
|
411
376
|
value = sprintf("%04d-%02d-%02d",value.year,value.month,value.day)
|
412
377
|
end
|
413
378
|
end
|
@@ -418,7 +383,7 @@ class Excel < GenericSpreadsheet
|
|
418
383
|
# Read the cell and based on the class,
|
419
384
|
# return the values for Roo
|
420
385
|
def read_cell(row, idx)
|
421
|
-
cell = row
|
386
|
+
cell = read_cell_content(row, idx)
|
422
387
|
case cell
|
423
388
|
when Float, Integer, Fixnum, Bignum
|
424
389
|
value_type = :float
|
data/lib/roo/excelx.rb
CHANGED
@@ -128,11 +128,7 @@ class Excelx < GenericSpreadsheet
|
|
128
128
|
FileUtils::rm_r(@tmpdir)
|
129
129
|
#end
|
130
130
|
end
|
131
|
-
@default_sheet =
|
132
|
-
# no need to set default_sheet if there is only one sheet in the document
|
133
|
-
if self.sheets.size == 1
|
134
|
-
@default_sheet = self.sheets.first
|
135
|
-
end
|
131
|
+
@default_sheet = self.sheets.first
|
136
132
|
@cell = Hash.new
|
137
133
|
@cell_type = Hash.new
|
138
134
|
@formula = Hash.new
|
@@ -258,18 +258,8 @@ class GenericSpreadsheet
|
|
258
258
|
sheet = @default_sheet unless sheet
|
259
259
|
read_cells(sheet) unless @cells_read[sheet]
|
260
260
|
result = []
|
261
|
-
|
262
|
-
|
263
|
-
y,x = key # _to_string(key).split(',')
|
264
|
-
x = x.to_i
|
265
|
-
y = y.to_i
|
266
|
-
if y == rownumber
|
267
|
-
tmp_arr[x] = value
|
268
|
-
end
|
269
|
-
}
|
270
|
-
result = tmp_arr[1..-1]
|
271
|
-
while result && result[-1] == nil
|
272
|
-
result = result[0..-2]
|
261
|
+
first_column(sheet).upto(last_column(sheet)) do |col|
|
262
|
+
result << cell(rownumber,col,sheet)
|
273
263
|
end
|
274
264
|
result
|
275
265
|
end
|
data/lib/roo/google.rb
CHANGED
@@ -109,10 +109,7 @@ class Google < GenericSpreadsheet
|
|
109
109
|
#-- ----------------------------------------------------------------------
|
110
110
|
#-- TODO: Behandlung von Berechtigungen hier noch einbauen ???
|
111
111
|
#-- ----------------------------------------------------------------------
|
112
|
-
|
113
|
-
if self.sheets.size == 1
|
114
|
-
@default_sheet = self.sheets.first
|
115
|
-
end
|
112
|
+
@default_sheet = self.sheets.first
|
116
113
|
end
|
117
114
|
|
118
115
|
# returns an array of sheet names in the spreadsheet
|
@@ -241,28 +238,6 @@ class Google < GenericSpreadsheet
|
|
241
238
|
theformulas
|
242
239
|
end
|
243
240
|
|
244
|
-
# returns all values in this row as an array
|
245
|
-
# row numbers are 1,2,3,... like in the spreadsheet
|
246
|
-
def row(rownumber,sheet=nil)
|
247
|
-
sheet = @default_sheet unless sheet
|
248
|
-
read_cells(sheet) unless @cells_read[sheet]
|
249
|
-
result = []
|
250
|
-
tmp_arr = []
|
251
|
-
@cell[sheet].each_pair {|key,value|
|
252
|
-
y,x = key.split(',')
|
253
|
-
x = x.to_i
|
254
|
-
y = y.to_i
|
255
|
-
if y == rownumber
|
256
|
-
tmp_arr[x] = value
|
257
|
-
end
|
258
|
-
}
|
259
|
-
result = tmp_arr[1..-1]
|
260
|
-
while result[-1] == nil
|
261
|
-
result = result[0..-2]
|
262
|
-
end
|
263
|
-
result
|
264
|
-
end
|
265
|
-
|
266
241
|
# true, if the cell is empty
|
267
242
|
def empty?(row, col, sheet=nil)
|
268
243
|
value = cell(row, col, sheet)
|
@@ -273,23 +248,6 @@ class Google < GenericSpreadsheet
|
|
273
248
|
value.empty?
|
274
249
|
end
|
275
250
|
|
276
|
-
# returns all values in this column as an array
|
277
|
-
# column numbers are 1,2,3,... like in the spreadsheet
|
278
|
-
#--
|
279
|
-
#TODO: refactoring nach GenericSpreadsheet?
|
280
|
-
def column(columnnumber, sheet=nil)
|
281
|
-
if columnnumber.class == String
|
282
|
-
columnnumber = GenericSpreadsheet.letter_to_number(columnnumber)
|
283
|
-
end
|
284
|
-
sheet = @default_sheet unless sheet
|
285
|
-
read_cells(sheet) unless @cells_read[sheet]
|
286
|
-
result = []
|
287
|
-
first_row(sheet).upto(last_row(sheet)) do |row|
|
288
|
-
result << cell(row,columnnumber,sheet)
|
289
|
-
end
|
290
|
-
result
|
291
|
-
end
|
292
|
-
|
293
251
|
# sets the cell to the content of 'value'
|
294
252
|
# a formula can be set in the form of '=SUM(...)'
|
295
253
|
def set_value(row,col,value,sheet=nil)
|
data/lib/roo/openoffice.rb
CHANGED
@@ -43,11 +43,7 @@ class Openoffice < GenericSpreadsheet
|
|
43
43
|
FileUtils::rm_r(@tmpdir)
|
44
44
|
#end
|
45
45
|
end
|
46
|
-
@default_sheet =
|
47
|
-
# no need to set default_sheet if there is only one sheet in the document
|
48
|
-
if self.sheets.size == 1
|
49
|
-
@default_sheet = self.sheets.first
|
50
|
-
end
|
46
|
+
@default_sheet = self.sheets.first
|
51
47
|
@cell = Hash.new
|
52
48
|
@cell_type = Hash.new
|
53
49
|
@formula = Hash.new
|
data/lib/roo/version.rb
CHANGED
data/lib/roo.rb
CHANGED
@@ -1,4 +1,22 @@
|
|
1
1
|
module Roo
|
2
|
+
class Spreadsheet
|
3
|
+
class << self
|
4
|
+
def open(file)
|
5
|
+
case File.extname(file)
|
6
|
+
when '.xls'
|
7
|
+
Excel.new(file)
|
8
|
+
when '.xlsx'
|
9
|
+
Excelx.new(file)
|
10
|
+
when '.ods'
|
11
|
+
Openoffice.new(file)
|
12
|
+
when ''
|
13
|
+
Google.new(file)
|
14
|
+
else
|
15
|
+
raise ArgumentError, "Don't know how to open file #{filename}"
|
16
|
+
end
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
2
20
|
end
|
3
21
|
|
4
22
|
require 'roo/version'
|
data/test/1900_base.xls
ADDED
Binary file
|
data/test/1904_base.xls
ADDED
Binary file
|