hmcgowan-roo 1.2.4
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- data/History.txt +225 -0
- data/README.txt +43 -0
- data/lib/roo/excel.rb +455 -0
- data/lib/roo/excelx.rb +654 -0
- data/lib/roo/generic_spreadsheet.rb +636 -0
- data/lib/roo/google.rb +411 -0
- data/lib/roo/openoffice.rb +508 -0
- data/lib/roo/roo_rails_helper.rb +81 -0
- data/lib/roo/version.rb +9 -0
- data/lib/roo.rb +11 -0
- data/test/Bibelbund.csv +3741 -0
- data/test/Bibelbund.ods +0 -0
- data/test/Bibelbund.xls +0 -0
- data/test/Bibelbund.xlsx +0 -0
- data/test/Bibelbund1.ods +0 -0
- data/test/bbu.ods +0 -0
- data/test/bbu.xls +0 -0
- data/test/bbu.xlsx +0 -0
- data/test/bode-v1.ods.zip +0 -0
- data/test/bode-v1.xls.zip +0 -0
- data/test/boolean.ods +0 -0
- data/test/boolean.xls +0 -0
- data/test/boolean.xlsx +0 -0
- data/test/borders.ods +0 -0
- data/test/borders.xls +0 -0
- data/test/borders.xlsx +0 -0
- data/test/bug-row-column-fixnum-float.xls +0 -0
- data/test/datetime.ods +0 -0
- data/test/datetime.xls +0 -0
- data/test/datetime.xlsx +0 -0
- data/test/emptysheets.ods +0 -0
- data/test/emptysheets.xls +0 -0
- data/test/false_encoding.xls +0 -0
- data/test/formula.ods +0 -0
- data/test/formula.xls +0 -0
- data/test/formula.xlsx +0 -0
- data/test/html-escape.ods +0 -0
- data/test/no_spreadsheet_file.txt +1 -0
- data/test/numbers1.csv +18 -0
- data/test/numbers1.ods +0 -0
- data/test/numbers1.xls +0 -0
- data/test/numbers1.xlsx +0 -0
- data/test/numbers1_excel.csv +18 -0
- data/test/only_one_sheet.ods +0 -0
- data/test/only_one_sheet.xls +0 -0
- data/test/only_one_sheet.xlsx +0 -0
- data/test/ric.ods +0 -0
- data/test/simple_spreadsheet.ods +0 -0
- data/test/simple_spreadsheet.xls +0 -0
- data/test/simple_spreadsheet.xlsx +0 -0
- data/test/simple_spreadsheet_from_italo.ods +0 -0
- data/test/simple_spreadsheet_from_italo.xls +0 -0
- data/test/style.ods +0 -0
- data/test/style.xls +0 -0
- data/test/style.xlsx +0 -0
- data/test/test_helper.rb +19 -0
- data/test/test_roo.rb +4946 -0
- data/test/time-test.csv +2 -0
- data/test/time-test.ods +0 -0
- data/test/time-test.xls +0 -0
- data/test/time-test.xlsx +0 -0
- metadata +225 -0
data/lib/roo/excelx.rb
ADDED
@@ -0,0 +1,654 @@
|
|
1
|
+
|
2
|
+
require 'rubygems'
|
3
|
+
require 'rexml/document'
|
4
|
+
require 'fileutils'
|
5
|
+
require 'zip/zipfilesystem'
|
6
|
+
require 'date'
|
7
|
+
|
8
|
+
class String
|
9
|
+
def end_with?(str)
|
10
|
+
self[-str.length,str.length] == str
|
11
|
+
end
|
12
|
+
end
|
13
|
+
|
14
|
+
class Excelx < GenericSpreadsheet
|
15
|
+
FORMATS = {
|
16
|
+
'General' => :float,
|
17
|
+
'0' => :float,
|
18
|
+
'0.00' => :float,
|
19
|
+
'#,##0' => :float,
|
20
|
+
'#,##0.00' => :float,
|
21
|
+
'0%' => :percentage,
|
22
|
+
'0.00%' => :percentage,
|
23
|
+
'0.00E+00' => :float,
|
24
|
+
'# ?/?' => :float, #??? TODO:
|
25
|
+
'# ??/??' => :float, #??? TODO:
|
26
|
+
'mm-dd-yy' => :date,
|
27
|
+
'd-mmm-yy' => :date,
|
28
|
+
'd-mmm' => :date,
|
29
|
+
'mmm-yy' => :date,
|
30
|
+
'h:mm AM/PM' => :date,
|
31
|
+
'h:mm:ss AM/PM' => :date,
|
32
|
+
'h:mm' => :time,
|
33
|
+
'h:mm:ss' => :time,
|
34
|
+
'm/d/yy h:mm' => :date,
|
35
|
+
'#,##0 ;(#,##0)' => :float,
|
36
|
+
'#,##0 ;[Red](#,##0)' => :float,
|
37
|
+
'#,##0.00;(#,##0.00)' => :float,
|
38
|
+
'#,##0.00;[Red](#,##0.00)' => :float,
|
39
|
+
'mm:ss' => :time,
|
40
|
+
'[h]:mm:ss' => :time,
|
41
|
+
'mmss.0' => :time,
|
42
|
+
'##0.0E+0' => :float,
|
43
|
+
'@' => :float,
|
44
|
+
#-- zusaetzliche Formate, die nicht standardmaessig definiert sind:
|
45
|
+
"yyyy\\-mm\\-dd" => :date,
|
46
|
+
'dd/mm/yy' => :date,
|
47
|
+
'hh:mm:ss' => :time,
|
48
|
+
"dd/mm/yy\\ hh:mm" => :datetime,
|
49
|
+
}
|
50
|
+
STANDARD_FORMATS = {
|
51
|
+
0 => 'General',
|
52
|
+
1 => '0',
|
53
|
+
2 => '0.00',
|
54
|
+
3 => '#,##0',
|
55
|
+
4 => '#,##0.00',
|
56
|
+
9 => '0%',
|
57
|
+
10 => '0.00%',
|
58
|
+
11 => '0.00E+00',
|
59
|
+
12 => '# ?/?',
|
60
|
+
13 => '# ??/??',
|
61
|
+
14 => 'mm-dd-yy',
|
62
|
+
15 => 'd-mmm-yy',
|
63
|
+
16 => 'd-mmm',
|
64
|
+
17 => 'mmm-yy',
|
65
|
+
18 => 'h:mm AM/PM',
|
66
|
+
19 => 'h:mm:ss AM/PM',
|
67
|
+
20 => 'h:mm',
|
68
|
+
21 => 'h:mm:ss',
|
69
|
+
22 => 'm/d/yy h:mm',
|
70
|
+
37 => '#,##0 ;(#,##0)',
|
71
|
+
38 => '#,##0 ;[Red](#,##0)',
|
72
|
+
39 => '#,##0.00;(#,##0.00)',
|
73
|
+
40 => '#,##0.00;[Red](#,##0.00)',
|
74
|
+
45 => 'mm:ss',
|
75
|
+
46 => '[h]:mm:ss',
|
76
|
+
47 => 'mmss.0',
|
77
|
+
48 => '##0.0E+0',
|
78
|
+
49 => '@',
|
79
|
+
}
|
80
|
+
@@nr = 0
|
81
|
+
|
82
|
+
# initialization and opening of a spreadsheet file
|
83
|
+
# values for packed: :zip
|
84
|
+
def initialize(filename, packed=nil, file_warning = :error) #, create = false)
|
85
|
+
super()
|
86
|
+
@file_warning = file_warning
|
87
|
+
@tmpdir = "oo_"+$$.to_s
|
88
|
+
@tmpdir = File.join(ENV['ROO_TMP'], @tmpdir) if ENV['ROO_TMP']
|
89
|
+
unless File.exists?(@tmpdir)
|
90
|
+
FileUtils::mkdir(@tmpdir)
|
91
|
+
end
|
92
|
+
filename = open_from_uri(filename) if filename[0,7] == "http://"
|
93
|
+
filename = unzip(filename) if packed and packed == :zip
|
94
|
+
begin
|
95
|
+
file_type_check(filename,'.xlsx','an Excel-xlsx')
|
96
|
+
@cells_read = Hash.new
|
97
|
+
@filename = filename
|
98
|
+
unless File.file?(@filename)
|
99
|
+
raise IOError, "file #{@filename} does not exist"
|
100
|
+
end
|
101
|
+
@@nr += 1
|
102
|
+
@file_nr = @@nr
|
103
|
+
extract_content(@filename)
|
104
|
+
file = File.new(File.join(@tmpdir, @file_nr.to_s+"_roo_workbook.xml"))
|
105
|
+
@workbook_doc = REXML::Document.new file
|
106
|
+
file.close
|
107
|
+
@shared_table = []
|
108
|
+
if File.exist?(File.join(@tmpdir, @file_nr.to_s+'_roo_sharedStrings.xml'))
|
109
|
+
file = File.new(File.join(@tmpdir, @file_nr.to_s+'_roo_sharedStrings.xml'))
|
110
|
+
@sharedstring_doc = REXML::Document.new file
|
111
|
+
file.close
|
112
|
+
read_shared_strings(@sharedstring_doc)
|
113
|
+
end
|
114
|
+
@styles_table = []
|
115
|
+
@style_definitions = Array.new { |h,k| h[k] = {} }
|
116
|
+
if File.exist?(File.join(@tmpdir, @file_nr.to_s+'_roo_styles.xml'))
|
117
|
+
file = File.new(File.join(@tmpdir, @file_nr.to_s+'_roo_styles.xml'))
|
118
|
+
@styles_doc = REXML::Document.new file
|
119
|
+
file.close
|
120
|
+
read_styles(@styles_doc)
|
121
|
+
end
|
122
|
+
@sheet_doc = []
|
123
|
+
@sheet_files.each_with_index do |item, i|
|
124
|
+
file = File.new(item)
|
125
|
+
@sheet_doc[i] = REXML::Document.new file
|
126
|
+
file.close
|
127
|
+
end
|
128
|
+
ensure
|
129
|
+
#if ENV["roo_local"] != "thomas-p"
|
130
|
+
FileUtils::rm_r(@tmpdir)
|
131
|
+
#end
|
132
|
+
end
|
133
|
+
@default_sheet = nil
|
134
|
+
# no need to set default_sheet if there is only one sheet in the document
|
135
|
+
if self.sheets.size == 1
|
136
|
+
@default_sheet = self.sheets.first
|
137
|
+
end
|
138
|
+
@cell = Hash.new
|
139
|
+
@cell_type = Hash.new
|
140
|
+
@formula = Hash.new
|
141
|
+
@first_row = Hash.new
|
142
|
+
@last_row = Hash.new
|
143
|
+
@first_column = Hash.new
|
144
|
+
@last_column = Hash.new
|
145
|
+
@header_line = 1
|
146
|
+
@excelx_type = Hash.new
|
147
|
+
@excelx_value = Hash.new
|
148
|
+
@s_attribute = Hash.new # TODO: ggf. wieder entfernen nur lokal benoetigt
|
149
|
+
end
|
150
|
+
|
151
|
+
# Returns the content of a spreadsheet-cell.
|
152
|
+
# (1,1) is the upper left corner.
|
153
|
+
# (1,1), (1,'A'), ('A',1), ('a',1) all refers to the
|
154
|
+
# cell at the first line and first row.
|
155
|
+
def cell(row, col, sheet=nil)
|
156
|
+
sheet = @default_sheet unless sheet
|
157
|
+
read_cells(sheet) unless @cells_read[sheet]
|
158
|
+
row,col = normalize(row,col)
|
159
|
+
if celltype(row,col,sheet) == :date
|
160
|
+
yyyy,mm,dd = @cell[sheet][[row,col]].split('-')
|
161
|
+
return Date.new(yyyy.to_i,mm.to_i,dd.to_i)
|
162
|
+
elsif celltype(row,col,sheet) == :datetime
|
163
|
+
date_part,time_part = @cell[sheet][[row,col]].split(' ')
|
164
|
+
yyyy,mm,dd = date_part.split('-')
|
165
|
+
hh,mi,ss = time_part.split(':')
|
166
|
+
return DateTime.civil(yyyy.to_i,mm.to_i,dd.to_i,hh.to_i,mi.to_i,ss.to_i)
|
167
|
+
|
168
|
+
end
|
169
|
+
@cell[sheet][[row,col]]
|
170
|
+
end
|
171
|
+
|
172
|
+
# Returns the formula at (row,col).
|
173
|
+
# Returns nil if there is no formula.
|
174
|
+
# The method #formula? checks if there is a formula.
|
175
|
+
def formula(row,col,sheet=nil)
|
176
|
+
sheet = @default_sheet unless sheet
|
177
|
+
read_cells(sheet) unless @cells_read[sheet]
|
178
|
+
row,col = normalize(row,col)
|
179
|
+
if @formula[sheet][[row,col]] == nil
|
180
|
+
return nil
|
181
|
+
else
|
182
|
+
return @formula[sheet][[row,col]]
|
183
|
+
end
|
184
|
+
end
|
185
|
+
|
186
|
+
# true, if there is a formula
|
187
|
+
def formula?(row,col,sheet=nil)
|
188
|
+
sheet = @default_sheet unless sheet
|
189
|
+
read_cells(sheet) unless @cells_read[sheet]
|
190
|
+
row,col = normalize(row,col)
|
191
|
+
formula(row,col) != nil
|
192
|
+
end
|
193
|
+
|
194
|
+
class Font
|
195
|
+
attr_accessor :bold, :italic, :underline
|
196
|
+
|
197
|
+
def bold?
|
198
|
+
@bold == true
|
199
|
+
end
|
200
|
+
|
201
|
+
def italic?
|
202
|
+
@italic == true
|
203
|
+
end
|
204
|
+
|
205
|
+
def underline?
|
206
|
+
@underline == true
|
207
|
+
end
|
208
|
+
end
|
209
|
+
|
210
|
+
# Given a cell, return the cell's style
|
211
|
+
def font(row, col, sheet=nil)
|
212
|
+
sheet = @default_sheet unless sheet
|
213
|
+
read_cells(sheet) unless @cells_read[sheet]
|
214
|
+
row,col = normalize(row,col)
|
215
|
+
s_attribute = @s_attribute[sheet][[row,col]]
|
216
|
+
s_attribute ||= 0
|
217
|
+
s_attribute = s_attribute.to_i
|
218
|
+
@style_definitions[s_attribute]
|
219
|
+
end
|
220
|
+
|
221
|
+
# set a cell to a certain value
|
222
|
+
# (this will not be saved back to the spreadsheet file!)
|
223
|
+
def set(row,col,value,sheet=nil) #:nodoc:
|
224
|
+
sheet = @default_sheet unless sheet
|
225
|
+
read_cells(sheet) unless @cells_read[sheet]
|
226
|
+
row,col = normalize(row,col)
|
227
|
+
set_value(row,col,value,sheet)
|
228
|
+
if value.class == Fixnum
|
229
|
+
set_type(row,col,:float,sheet)
|
230
|
+
elsif value.class == String
|
231
|
+
set_type(row,col,:string,sheet)
|
232
|
+
elsif value.class == Float
|
233
|
+
set_type(row,col,:string,sheet)
|
234
|
+
else
|
235
|
+
raise ArgumentError, "Type for "+value.to_s+" not set"
|
236
|
+
end
|
237
|
+
end
|
238
|
+
|
239
|
+
# returns the type of a cell:
|
240
|
+
# * :float
|
241
|
+
# * :string,
|
242
|
+
# * :date
|
243
|
+
# * :percentage
|
244
|
+
# * :formula
|
245
|
+
# * :time
|
246
|
+
# * :datetime
|
247
|
+
def celltype(row,col,sheet=nil)
|
248
|
+
sheet = @default_sheet unless sheet
|
249
|
+
read_cells(sheet) unless @cells_read[sheet]
|
250
|
+
row,col = normalize(row,col)
|
251
|
+
if @formula[sheet][[row,col]]
|
252
|
+
return :formula
|
253
|
+
else
|
254
|
+
@cell_type[sheet][[row,col]]
|
255
|
+
end
|
256
|
+
end
|
257
|
+
|
258
|
+
# returns the internal type of an excel cell
|
259
|
+
# * :numeric_or_formula
|
260
|
+
# * :string
|
261
|
+
# Note: this is only available within the Excelx class
|
262
|
+
def excelx_type(row,col,sheet=nil)
|
263
|
+
sheet = @default_sheet unless sheet
|
264
|
+
read_cells(sheet) unless @cells_read[sheet]
|
265
|
+
row,col = normalize(row,col)
|
266
|
+
return @excelx_type[sheet][[row,col]]
|
267
|
+
end
|
268
|
+
|
269
|
+
# returns the internal value of an excelx cell
|
270
|
+
# Note: this is only available within the Excelx class
|
271
|
+
def excelx_value(row,col,sheet=nil)
|
272
|
+
sheet = @default_sheet unless sheet
|
273
|
+
read_cells(sheet) unless @cells_read[sheet]
|
274
|
+
row,col = normalize(row,col)
|
275
|
+
return @excelx_value[sheet][[row,col]]
|
276
|
+
end
|
277
|
+
|
278
|
+
# returns the internal format of an excel cell
|
279
|
+
def excelx_format(row,col,sheet=nil)
|
280
|
+
sheet = @default_sheet unless sheet
|
281
|
+
read_cells(sheet) unless @cells_read[sheet]
|
282
|
+
row,col = normalize(row,col)
|
283
|
+
s = @s_attribute[sheet][[row,col]]
|
284
|
+
result = attribute2format(s)
|
285
|
+
result
|
286
|
+
end
|
287
|
+
|
288
|
+
# returns an array of sheet names in the spreadsheet
|
289
|
+
def sheets
|
290
|
+
return_sheets = []
|
291
|
+
@workbook_doc.each_element do |workbook|
|
292
|
+
workbook.each_element do |el|
|
293
|
+
if el.name == "sheets"
|
294
|
+
el.each_element do |sheet|
|
295
|
+
return_sheets << sheet.attributes['name']
|
296
|
+
end
|
297
|
+
end
|
298
|
+
end
|
299
|
+
end
|
300
|
+
return_sheets
|
301
|
+
end
|
302
|
+
|
303
|
+
# shows the internal representation of all cells
|
304
|
+
# for debugging purposes
|
305
|
+
def to_s(sheet=nil)
|
306
|
+
sheet = @default_sheet unless sheet
|
307
|
+
read_cells(sheet) unless @cells_read[sheet]
|
308
|
+
@cell[sheet].inspect
|
309
|
+
end
|
310
|
+
|
311
|
+
# returns each formula in the selected sheet as an array of elements
|
312
|
+
# [row, col, formula]
|
313
|
+
def formulas(sheet=nil)
|
314
|
+
theformulas = Array.new
|
315
|
+
sheet = @default_sheet unless sheet
|
316
|
+
read_cells(sheet) unless @cells_read[sheet]
|
317
|
+
first_row(sheet).upto(last_row(sheet)) {|row|
|
318
|
+
first_column(sheet).upto(last_column(sheet)) {|col|
|
319
|
+
if formula?(row,col,sheet)
|
320
|
+
f = [row, col, formula(row,col,sheet)]
|
321
|
+
theformulas << f
|
322
|
+
end
|
323
|
+
}
|
324
|
+
}
|
325
|
+
theformulas
|
326
|
+
end
|
327
|
+
|
328
|
+
private
|
329
|
+
|
330
|
+
# helper function to set the internal representation of cells
|
331
|
+
def set_cell_values(sheet,x,y,i,v,vt,formula,tr,str_v,
|
332
|
+
excelx_type=nil,
|
333
|
+
excelx_value=nil,
|
334
|
+
s_attribute=nil)
|
335
|
+
key = [y,x+i]
|
336
|
+
@cell_type[sheet] = {} unless @cell_type[sheet]
|
337
|
+
@cell_type[sheet][key] = vt
|
338
|
+
@formula[sheet] = {} unless @formula[sheet]
|
339
|
+
@formula[sheet][key] = formula if formula
|
340
|
+
@cell[sheet] = {} unless @cell[sheet]
|
341
|
+
case @cell_type[sheet][key]
|
342
|
+
when :float
|
343
|
+
@cell[sheet][key] = v.to_f
|
344
|
+
when :string
|
345
|
+
@cell[sheet][key] = str_v
|
346
|
+
when :date
|
347
|
+
@cell[sheet][key] = (Date.new(1899,12,30)+v.to_i).strftime("%Y-%m-%d")
|
348
|
+
when :datetime
|
349
|
+
@cell[sheet][key] = (DateTime.new(1899,12,30)+v.to_f).strftime("%Y-%m-%d %H:%M:%S")
|
350
|
+
when :percentage
|
351
|
+
@cell[sheet][key] = v.to_f
|
352
|
+
when :time
|
353
|
+
@cell[sheet][key] = v.to_f*(24*60*60)
|
354
|
+
else
|
355
|
+
@cell[sheet][key] = v
|
356
|
+
end
|
357
|
+
@excelx_type[sheet] = {} unless @excelx_type[sheet]
|
358
|
+
@excelx_type[sheet][key] = excelx_type
|
359
|
+
@excelx_value[sheet] = {} unless @excelx_value[sheet]
|
360
|
+
@excelx_value[sheet][key] = excelx_value
|
361
|
+
@s_attribute[sheet] = {} unless @s_attribute[sheet]
|
362
|
+
@s_attribute[sheet][key] = s_attribute
|
363
|
+
end
|
364
|
+
|
365
|
+
# splits a coordinate like "AA12" into the parts "AA" (String) and 12 (Fixnum)
|
366
|
+
def split_coord(s)
|
367
|
+
letter = ""
|
368
|
+
number = 0
|
369
|
+
i = 0
|
370
|
+
while i<s.length and "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz".include?(s[i,1])
|
371
|
+
letter += s[i,1]
|
372
|
+
i+=1
|
373
|
+
end
|
374
|
+
while i<s.length and "0123456789".include?(s[i,1])
|
375
|
+
number = number*10 + s[i,1].to_i
|
376
|
+
i+=1
|
377
|
+
end
|
378
|
+
if letter=="" or number==0
|
379
|
+
raise ArgumentError
|
380
|
+
end
|
381
|
+
return letter,number
|
382
|
+
end
|
383
|
+
|
384
|
+
def split_coordinate(str)
|
385
|
+
letter,number = split_coord(str)
|
386
|
+
x = GenericSpreadsheet.letter_to_number(letter)
|
387
|
+
y = number
|
388
|
+
return x,y
|
389
|
+
end
|
390
|
+
|
391
|
+
# read all cells in the selected sheet
|
392
|
+
def format2type(format)
|
393
|
+
if FORMATS.has_key? format
|
394
|
+
FORMATS[format]
|
395
|
+
else
|
396
|
+
:float
|
397
|
+
end
|
398
|
+
end
|
399
|
+
|
400
|
+
# read all cells in the selected sheet
|
401
|
+
def read_cells(sheet=nil)
|
402
|
+
sheet = @default_sheet unless sheet
|
403
|
+
sheet_found = false
|
404
|
+
raise ArgumentError, "Error: sheet '#{sheet||'nil'}' not valid" if @default_sheet == nil and sheet==nil
|
405
|
+
raise RangeError unless self.sheets.include? sheet
|
406
|
+
n = self.sheets.index(sheet)
|
407
|
+
@sheet_doc[n].each_element do |worksheet|
|
408
|
+
worksheet.each_element do |elem|
|
409
|
+
if elem.name == 'sheetData'
|
410
|
+
elem.each_element do |sheetdata|
|
411
|
+
if sheetdata.name == 'row'
|
412
|
+
sheetdata.each_element do |row|
|
413
|
+
if row.name == 'c'
|
414
|
+
s_attribute = row.attributes['s']
|
415
|
+
if row.attributes['t'] == 's'
|
416
|
+
tmp_type = :shared
|
417
|
+
elsif row.attributes['t'] == 'b'
|
418
|
+
tmp_type = :boolean
|
419
|
+
else
|
420
|
+
format = attribute2format(s_attribute)
|
421
|
+
tmp_type = format2type(format)
|
422
|
+
end
|
423
|
+
formula = nil
|
424
|
+
row.each_element do |cell|
|
425
|
+
# puts "cell.name: #{cell.name}" if cell.text.include? "22606.5120"
|
426
|
+
# puts "cell.text: #{cell.text}" if cell.text.include? "22606.5120"
|
427
|
+
if cell.name == 'f'
|
428
|
+
formula = cell.text
|
429
|
+
end
|
430
|
+
if cell.name == 'v'
|
431
|
+
#puts "tmp_type: #{tmp_type}" if cell.text.include? "22606.5120"
|
432
|
+
#puts cell.name
|
433
|
+
if tmp_type == :time or tmp_type == :datetime #2008-07-26
|
434
|
+
#p cell.text
|
435
|
+
# p cell.text.to_f if cell.text.include? "22606.5120"
|
436
|
+
if cell.text.to_f >= 1.0 # 2008-07-26
|
437
|
+
# puts ">= 1.0" if cell.text.include? "22606.5120"
|
438
|
+
# puts "cell.text.to_f: #{cell.text.to_f}" if cell.text.include? "22606.5120"
|
439
|
+
#puts "cell.text.to_f.floor: #{cell.text.to_f.floor}" if cell.text.include? "22606.5120"
|
440
|
+
if (cell.text.to_f - cell.text.to_f.floor).abs > 0.000001 #TODO:
|
441
|
+
# puts "abs ist groesser" if cell.text.include? "22606.5120"
|
442
|
+
# @cell[sheet][key] = DateTime.parse(tr.attributes['date-value'])
|
443
|
+
tmp_type = :datetime
|
444
|
+
|
445
|
+
else
|
446
|
+
#puts ":date"
|
447
|
+
tmp_type = :date # 2008-07-26
|
448
|
+
end
|
449
|
+
else
|
450
|
+
#puts "<1.0"
|
451
|
+
end # 2008-07-26
|
452
|
+
end # 2008-07-26
|
453
|
+
excelx_type = [:numeric_or_formula,format]
|
454
|
+
excelx_value = cell.text
|
455
|
+
if tmp_type == :shared
|
456
|
+
vt = :string
|
457
|
+
str_v = @shared_table[cell.text.to_i]
|
458
|
+
excelx_type = :string
|
459
|
+
elsif tmp_type == :boolean
|
460
|
+
vt = :boolean
|
461
|
+
cell.text.to_i == 1 ? v = 'TRUE' : v = 'FALSE'
|
462
|
+
elsif tmp_type == :date
|
463
|
+
vt = :date
|
464
|
+
v = cell.text
|
465
|
+
elsif tmp_type == :time
|
466
|
+
vt = :time
|
467
|
+
v = cell.text
|
468
|
+
elsif tmp_type == :datetime
|
469
|
+
vt = :datetime
|
470
|
+
v = cell.text
|
471
|
+
elsif tmp_type == :formula
|
472
|
+
vt = :formula
|
473
|
+
v = cell.text.to_f #TODO: !!!!
|
474
|
+
else
|
475
|
+
vt = :float
|
476
|
+
v = cell.text
|
477
|
+
end
|
478
|
+
#puts "vt: #{vt}" if cell.text.include? "22606.5120"
|
479
|
+
x,y = split_coordinate(row.attributes['r'])
|
480
|
+
tr=nil #TODO: ???s
|
481
|
+
set_cell_values(sheet,x,y,0,v,vt,formula,tr,str_v,excelx_type,excelx_value,s_attribute)
|
482
|
+
end
|
483
|
+
end
|
484
|
+
end
|
485
|
+
end
|
486
|
+
end
|
487
|
+
end
|
488
|
+
end
|
489
|
+
end
|
490
|
+
end
|
491
|
+
sheet_found = true #TODO:
|
492
|
+
if !sheet_found
|
493
|
+
raise RangeError
|
494
|
+
end
|
495
|
+
@cells_read[sheet] = true
|
496
|
+
end
|
497
|
+
|
498
|
+
# Checks if the default_sheet exists. If not an RangeError exception is
|
499
|
+
# raised
|
500
|
+
def check_default_sheet
|
501
|
+
sheet_found = false
|
502
|
+
raise ArgumentError, "Error: default_sheet not set" if @default_sheet == nil
|
503
|
+
@workbook_doc.each_element do |workbook|
|
504
|
+
workbook.each_element do |el|
|
505
|
+
if el.name == "sheets"
|
506
|
+
el.each_element do |sheet|
|
507
|
+
if @default_sheet == sheet.attributes['name']
|
508
|
+
sheet_found = true
|
509
|
+
end
|
510
|
+
end
|
511
|
+
end
|
512
|
+
end
|
513
|
+
end
|
514
|
+
if ! sheet_found
|
515
|
+
raise RangeError, "sheet '#{@default_sheet}' not found"
|
516
|
+
end
|
517
|
+
end
|
518
|
+
|
519
|
+
# extracts all needed files from the zip file
|
520
|
+
def process_zipfile(zipfilename, zip, path='')
|
521
|
+
@sheet_files = []
|
522
|
+
Zip::ZipFile.open(zipfilename) {|zf|
|
523
|
+
zf.entries.each {|entry|
|
524
|
+
#entry.extract
|
525
|
+
if entry.to_s.end_with?('workbook.xml')
|
526
|
+
open(@tmpdir+'/'+@file_nr.to_s+'_roo_workbook.xml','wb') {|f|
|
527
|
+
f << zip.read(entry)
|
528
|
+
}
|
529
|
+
end
|
530
|
+
if entry.to_s.end_with?('sharedStrings.xml')
|
531
|
+
open(@tmpdir+'/'+@file_nr.to_s+'_roo_sharedStrings.xml','wb') {|f|
|
532
|
+
f << zip.read(entry)
|
533
|
+
}
|
534
|
+
end
|
535
|
+
if entry.to_s.end_with?('styles.xml')
|
536
|
+
open(@tmpdir+'/'+@file_nr.to_s+'_roo_styles.xml','wb') {|f|
|
537
|
+
f << zip.read(entry)
|
538
|
+
}
|
539
|
+
end
|
540
|
+
if entry.to_s =~ /sheet([0-9]+).xml$/
|
541
|
+
nr = $1
|
542
|
+
open(@tmpdir+'/'+@file_nr.to_s+"_roo_sheet#{nr}",'wb') {|f|
|
543
|
+
f << zip.read(entry)
|
544
|
+
}
|
545
|
+
@sheet_files[nr.to_i-1] = @tmpdir+'/'+@file_nr.to_s+"_roo_sheet#{nr}"
|
546
|
+
end
|
547
|
+
}
|
548
|
+
}
|
549
|
+
return
|
550
|
+
end
|
551
|
+
|
552
|
+
# extract files from the zip file
|
553
|
+
def extract_content(zipfilename)
|
554
|
+
Zip::ZipFile.open(@filename) do |zip|
|
555
|
+
process_zipfile(zipfilename,zip)
|
556
|
+
end
|
557
|
+
end
|
558
|
+
|
559
|
+
# sets the value of a cell
|
560
|
+
def set_value(row,col,value,sheet=nil)
|
561
|
+
sheet = @default_value unless sheet
|
562
|
+
@cell[sheet][[row,col]] = value
|
563
|
+
end
|
564
|
+
|
565
|
+
# sets the type of a cell
|
566
|
+
def set_type(row,col,type,sheet=nil)
|
567
|
+
sheet = @default_value unless sheet
|
568
|
+
@cell_type[sheet][[row,col]] = type
|
569
|
+
end
|
570
|
+
|
571
|
+
# read the shared strings xml document
|
572
|
+
def read_shared_strings(doc)
|
573
|
+
doc.each_element do |sst|
|
574
|
+
if sst.name == 'sst'
|
575
|
+
sst.each_element do |si|
|
576
|
+
if si.name == 'si'
|
577
|
+
si.each_element do |elem|
|
578
|
+
if elem.name == 't'
|
579
|
+
@shared_table << elem.text
|
580
|
+
end
|
581
|
+
end
|
582
|
+
end
|
583
|
+
end
|
584
|
+
end
|
585
|
+
end
|
586
|
+
end
|
587
|
+
|
588
|
+
# read the styles elements of an excelx document
|
589
|
+
def read_styles(doc)
|
590
|
+
@numFmts = []
|
591
|
+
@cellXfs = []
|
592
|
+
fonts = []
|
593
|
+
doc.each_element do |e1|
|
594
|
+
if e1.name == "styleSheet"
|
595
|
+
e1.each_element do |e2|
|
596
|
+
if e2.name == "numFmts"
|
597
|
+
e2.each_element do |e3|
|
598
|
+
if e3.name == 'numFmt'
|
599
|
+
numFmtId = e3.attributes['numFmtId']
|
600
|
+
formatCode = e3.attributes['formatCode']
|
601
|
+
@numFmts << [numFmtId, formatCode]
|
602
|
+
end
|
603
|
+
end
|
604
|
+
elsif e2.name == "fonts"
|
605
|
+
e2.each_element do |e3|
|
606
|
+
if e3.name == 'font'
|
607
|
+
font = Excelx::Font.new
|
608
|
+
e3.each do |e4|
|
609
|
+
case e4.name
|
610
|
+
when 'b'
|
611
|
+
font.bold = true
|
612
|
+
when 'i'
|
613
|
+
font.italic = true
|
614
|
+
when 'u'
|
615
|
+
font.underline = true
|
616
|
+
end
|
617
|
+
end
|
618
|
+
fonts << font
|
619
|
+
end
|
620
|
+
end
|
621
|
+
elsif e2.name == "cellXfs"
|
622
|
+
e2.each_element do |e3|
|
623
|
+
if e3.name == 'xf'
|
624
|
+
numFmtId = e3.attributes['numFmtId']
|
625
|
+
@cellXfs << [numFmtId]
|
626
|
+
fontId = e3.attributes['fontId'].to_i
|
627
|
+
@style_definitions << fonts[fontId]
|
628
|
+
end
|
629
|
+
end
|
630
|
+
end
|
631
|
+
end
|
632
|
+
end
|
633
|
+
end
|
634
|
+
end
|
635
|
+
|
636
|
+
# convert internal excelx attribute to a format
|
637
|
+
def attribute2format(s)
|
638
|
+
result = nil
|
639
|
+
@numFmts.each {|nf|
|
640
|
+
if nf.first == @cellXfs[s.to_i].first
|
641
|
+
result = nf[1]
|
642
|
+
break
|
643
|
+
end
|
644
|
+
}
|
645
|
+
unless result
|
646
|
+
id = @cellXfs[s.to_i].first.to_i
|
647
|
+
if STANDARD_FORMATS.has_key? id
|
648
|
+
result = STANDARD_FORMATS[id]
|
649
|
+
end
|
650
|
+
end
|
651
|
+
result
|
652
|
+
end
|
653
|
+
|
654
|
+
end # class
|