roo-andyw8 2.0.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +7 -0
- data/.simplecov +4 -0
- data/.travis.yml +13 -0
- data/CHANGELOG +438 -0
- data/Gemfile +24 -0
- data/Guardfile +24 -0
- data/LICENSE +22 -0
- data/README.md +121 -0
- data/Rakefile +23 -0
- data/examples/roo_soap_client.rb +50 -0
- data/examples/roo_soap_server.rb +26 -0
- data/examples/write_me.rb +31 -0
- data/lib/roo.rb +28 -0
- data/lib/roo/base.rb +717 -0
- data/lib/roo/csv.rb +110 -0
- data/lib/roo/excelx.rb +540 -0
- data/lib/roo/excelx/comments.rb +23 -0
- data/lib/roo/excelx/extractor.rb +20 -0
- data/lib/roo/excelx/relationships.rb +26 -0
- data/lib/roo/excelx/shared_strings.rb +40 -0
- data/lib/roo/excelx/sheet_doc.rb +175 -0
- data/lib/roo/excelx/styles.rb +62 -0
- data/lib/roo/excelx/workbook.rb +59 -0
- data/lib/roo/font.rb +17 -0
- data/lib/roo/libre_office.rb +5 -0
- data/lib/roo/link.rb +15 -0
- data/lib/roo/open_office.rb +652 -0
- data/lib/roo/spreadsheet.rb +31 -0
- data/lib/roo/utils.rb +81 -0
- data/lib/roo/version.rb +3 -0
- data/roo.gemspec +27 -0
- data/scripts/txt2html +67 -0
- data/spec/fixtures/vcr_cassettes/google_drive.yml +165 -0
- data/spec/fixtures/vcr_cassettes/google_drive_access_token.yml +73 -0
- data/spec/fixtures/vcr_cassettes/google_drive_set.yml +857 -0
- data/spec/lib/roo/base_spec.rb +4 -0
- data/spec/lib/roo/csv_spec.rb +48 -0
- data/spec/lib/roo/excelx/format_spec.rb +51 -0
- data/spec/lib/roo/excelx_spec.rb +363 -0
- data/spec/lib/roo/libreoffice_spec.rb +13 -0
- data/spec/lib/roo/openoffice_spec.rb +15 -0
- data/spec/lib/roo/spreadsheet_spec.rb +88 -0
- data/spec/lib/roo/utils_spec.rb +105 -0
- data/spec/spec_helper.rb +9 -0
- data/test/all_ss.rb +11 -0
- data/test/files/1900_base.xlsx +0 -0
- data/test/files/1904_base.xlsx +0 -0
- data/test/files/Bibelbund.csv +3741 -0
- data/test/files/Bibelbund.ods +0 -0
- data/test/files/Bibelbund.xlsx +0 -0
- data/test/files/Bibelbund1.ods +0 -0
- data/test/files/Pfand_from_windows_phone.xlsx +0 -0
- data/test/files/advanced_header.ods +0 -0
- data/test/files/bbu.ods +0 -0
- data/test/files/bbu.xlsx +0 -0
- data/test/files/bode-v1.ods.zip +0 -0
- data/test/files/bode-v1.xls.zip +0 -0
- data/test/files/boolean.csv +2 -0
- data/test/files/boolean.ods +0 -0
- data/test/files/boolean.xlsx +0 -0
- data/test/files/borders.ods +0 -0
- data/test/files/borders.xlsx +0 -0
- data/test/files/bug-numbered-sheet-names.xlsx +0 -0
- data/test/files/comments.ods +0 -0
- data/test/files/comments.xlsx +0 -0
- data/test/files/csvtypes.csv +1 -0
- data/test/files/datetime.ods +0 -0
- data/test/files/datetime.xlsx +0 -0
- data/test/files/dreimalvier.ods +0 -0
- data/test/files/emptysheets.ods +0 -0
- data/test/files/emptysheets.xlsx +0 -0
- data/test/files/encrypted-letmein.ods +0 -0
- data/test/files/file_item_error.xlsx +0 -0
- data/test/files/formula.ods +0 -0
- data/test/files/formula.xlsx +0 -0
- data/test/files/formula_string_error.xlsx +0 -0
- data/test/files/html-escape.ods +0 -0
- data/test/files/link.csv +1 -0
- data/test/files/link.xlsx +0 -0
- data/test/files/matrix.ods +0 -0
- data/test/files/named_cells.ods +0 -0
- data/test/files/named_cells.xlsx +0 -0
- data/test/files/no_spreadsheet_file.txt +1 -0
- data/test/files/numbers-export.xlsx +0 -0
- data/test/files/numbers1.csv +18 -0
- data/test/files/numbers1.ods +0 -0
- data/test/files/numbers1.xlsx +0 -0
- data/test/files/numbers1withnull.xlsx +0 -0
- data/test/files/numeric-link.xlsx +0 -0
- data/test/files/only_one_sheet.ods +0 -0
- data/test/files/only_one_sheet.xlsx +0 -0
- data/test/files/paragraph.ods +0 -0
- data/test/files/paragraph.xlsx +0 -0
- data/test/files/ric.ods +0 -0
- data/test/files/sheet1.xml +109 -0
- data/test/files/simple_spreadsheet.ods +0 -0
- data/test/files/simple_spreadsheet.xlsx +0 -0
- data/test/files/simple_spreadsheet_from_italo.ods +0 -0
- data/test/files/so_datetime.csv +8 -0
- data/test/files/style.ods +0 -0
- data/test/files/style.xlsx +0 -0
- data/test/files/time-test.csv +2 -0
- data/test/files/time-test.ods +0 -0
- data/test/files/time-test.xlsx +0 -0
- data/test/files/type_excel.ods +0 -0
- data/test/files/type_excel.xlsx +0 -0
- data/test/files/type_excelx.ods +0 -0
- data/test/files/type_openoffice.xlsx +0 -0
- data/test/files/whitespace.ods +0 -0
- data/test/files/whitespace.xlsx +0 -0
- data/test/test_generic_spreadsheet.rb +211 -0
- data/test/test_helper.rb +58 -0
- data/test/test_roo.rb +1977 -0
- metadata +318 -0
@@ -0,0 +1,652 @@
|
|
1
|
+
require 'date'
|
2
|
+
require 'nokogiri'
|
3
|
+
require 'cgi'
|
4
|
+
require 'zip/filesystem'
|
5
|
+
require 'roo/font'
|
6
|
+
|
7
|
+
class Roo::OpenOffice < Roo::Base
|
8
|
+
# initialization and opening of a spreadsheet file
|
9
|
+
# values for packed: :zip
|
10
|
+
def initialize(filename, options={})
|
11
|
+
packed = options[:packed]
|
12
|
+
file_warning = options[:file_warning] || :error
|
13
|
+
|
14
|
+
file_type_check(filename,'.ods','an Roo::OpenOffice', file_warning, packed)
|
15
|
+
@tmpdir = make_tmpdir(filename.split('/').last, options[:tmpdir_root])
|
16
|
+
@filename = local_filename(filename, @tmpdir, packed)
|
17
|
+
#TODO: @cells_read[:default] = false
|
18
|
+
Zip::File.open(@filename) do |zip_file|
|
19
|
+
if content_entry = zip_file.glob("content.xml").first
|
20
|
+
roo_content_xml_path = File.join(@tmpdir, 'roo_content.xml')
|
21
|
+
content_entry.extract(roo_content_xml_path)
|
22
|
+
decrypt_if_necessary(
|
23
|
+
zip_file,
|
24
|
+
content_entry,
|
25
|
+
roo_content_xml_path,
|
26
|
+
options
|
27
|
+
)
|
28
|
+
else
|
29
|
+
raise ArgumentError, 'file missing required content.xml'
|
30
|
+
end
|
31
|
+
end
|
32
|
+
super(filename, options)
|
33
|
+
@formula = Hash.new
|
34
|
+
@style = Hash.new
|
35
|
+
@style_defaults = Hash.new { |h,k| h[k] = [] }
|
36
|
+
@style_definitions = Hash.new
|
37
|
+
@comment = Hash.new
|
38
|
+
@comments_read = Hash.new
|
39
|
+
end
|
40
|
+
|
41
|
+
# If the ODS file has an encryption-data element, then try to decrypt.
|
42
|
+
# If successful, the temporary content.xml will be overwritten with
|
43
|
+
# decrypted contents.
|
44
|
+
def decrypt_if_necessary(
|
45
|
+
zip_file,
|
46
|
+
content_entry,
|
47
|
+
roo_content_xml_path, options
|
48
|
+
)
|
49
|
+
# Check if content.xml is encrypted by extracting manifest.xml
|
50
|
+
# and searching for a manifest:encryption-data element
|
51
|
+
|
52
|
+
if manifest_entry = zip_file.glob("META-INF/manifest.xml").first
|
53
|
+
roo_manifest_xml_path = File.join(@tmpdir, "roo_manifest.xml")
|
54
|
+
manifest_entry.extract(roo_manifest_xml_path)
|
55
|
+
manifest = ::Roo::Utils.load_xml(roo_manifest_xml_path)
|
56
|
+
|
57
|
+
# XPath search for manifest:encryption-data only for the content.xml
|
58
|
+
# file
|
59
|
+
|
60
|
+
encryption_data = manifest.xpath(
|
61
|
+
"//manifest:file-entry[@manifest:full-path='content.xml']"\
|
62
|
+
"/manifest:encryption-data"
|
63
|
+
).first
|
64
|
+
|
65
|
+
# If XPath returns a node, then we know content.xml is encrypted
|
66
|
+
|
67
|
+
if !encryption_data.nil?
|
68
|
+
|
69
|
+
# Since we know it's encrypted, we check for the password option
|
70
|
+
# and if it doesn't exist, raise an argument error
|
71
|
+
|
72
|
+
password = options[:password]
|
73
|
+
if !password.nil?
|
74
|
+
perform_decryption(
|
75
|
+
encryption_data,
|
76
|
+
password,
|
77
|
+
content_entry,
|
78
|
+
roo_content_xml_path
|
79
|
+
)
|
80
|
+
else
|
81
|
+
raise ArgumentError,
|
82
|
+
'file is encrypted but password was not supplied'
|
83
|
+
end
|
84
|
+
end
|
85
|
+
else
|
86
|
+
raise ArgumentError, 'file missing required META-INF/manifest.xml'
|
87
|
+
end
|
88
|
+
end
|
89
|
+
|
90
|
+
# Process the ODS encryption manifest and perform the decryption
|
91
|
+
def perform_decryption(
|
92
|
+
encryption_data,
|
93
|
+
password,
|
94
|
+
content_entry,
|
95
|
+
roo_content_xml_path
|
96
|
+
)
|
97
|
+
# Extract various expected attributes from the manifest that
|
98
|
+
# describe the encryption
|
99
|
+
|
100
|
+
algorithm_node = encryption_data.xpath("manifest:algorithm").first
|
101
|
+
key_derivation_node =
|
102
|
+
encryption_data.xpath("manifest:key-derivation").first
|
103
|
+
start_key_generation_node =
|
104
|
+
encryption_data.xpath("manifest:start-key-generation").first
|
105
|
+
|
106
|
+
# If we have all the expected elements, then we can perform
|
107
|
+
# the decryption.
|
108
|
+
|
109
|
+
if !algorithm_node.nil? && !key_derivation_node.nil? &&
|
110
|
+
!start_key_generation_node.nil?
|
111
|
+
|
112
|
+
# The algorithm is a URI describing the algorithm used
|
113
|
+
algorithm = algorithm_node['manifest:algorithm-name']
|
114
|
+
|
115
|
+
# The initialization vector is base-64 encoded
|
116
|
+
iv = Base64.decode64(
|
117
|
+
algorithm_node['manifest:initialisation-vector']
|
118
|
+
)
|
119
|
+
key_derivation_name =
|
120
|
+
key_derivation_node['manifest:key-derivation-name']
|
121
|
+
key_size = key_derivation_node['manifest:key-size'].to_i
|
122
|
+
iteration_count =
|
123
|
+
key_derivation_node['manifest:iteration-count'].to_i
|
124
|
+
salt = Base64.decode64(key_derivation_node['manifest:salt'])
|
125
|
+
|
126
|
+
# The key is hashed with an algorithm represented by this URI
|
127
|
+
key_generation_name =
|
128
|
+
start_key_generation_node[
|
129
|
+
'manifest:start-key-generation-name'
|
130
|
+
]
|
131
|
+
key_generation_size =
|
132
|
+
start_key_generation_node['manifest:key-size'].to_i
|
133
|
+
|
134
|
+
hashed_password = password
|
135
|
+
key = nil
|
136
|
+
|
137
|
+
if key_generation_name.eql?(
|
138
|
+
"http://www.w3.org/2000/09/xmldsig#sha256"
|
139
|
+
)
|
140
|
+
hashed_password = Digest::SHA256.digest(password)
|
141
|
+
else
|
142
|
+
raise ArgumentError, 'Unknown key generation algorithm ' +
|
143
|
+
key_generation_name
|
144
|
+
end
|
145
|
+
|
146
|
+
cipher = find_cipher(
|
147
|
+
algorithm,
|
148
|
+
key_derivation_name,
|
149
|
+
hashed_password,
|
150
|
+
salt,
|
151
|
+
iteration_count,
|
152
|
+
iv
|
153
|
+
)
|
154
|
+
|
155
|
+
begin
|
156
|
+
decrypted = decrypt(content_entry, cipher)
|
157
|
+
|
158
|
+
# Finally, inflate the decrypted stream and overwrite
|
159
|
+
# content.xml
|
160
|
+
IO.binwrite(
|
161
|
+
roo_content_xml_path,
|
162
|
+
Zlib::Inflate.new(-Zlib::MAX_WBITS).inflate(decrypted)
|
163
|
+
)
|
164
|
+
rescue StandardError => error
|
165
|
+
raise ArgumentError,
|
166
|
+
'Invalid password or other data error: ' + error.to_s
|
167
|
+
end
|
168
|
+
else
|
169
|
+
raise ArgumentError,
|
170
|
+
'manifest.xml missing encryption-data elements'
|
171
|
+
end
|
172
|
+
end
|
173
|
+
|
174
|
+
# Create a cipher based on an ODS algorithm URI from manifest.xml
|
175
|
+
def find_cipher(
|
176
|
+
algorithm,
|
177
|
+
key_derivation_name,
|
178
|
+
hashed_password,
|
179
|
+
salt,
|
180
|
+
iteration_count,
|
181
|
+
iv
|
182
|
+
)
|
183
|
+
cipher = nil
|
184
|
+
if algorithm.eql? "http://www.w3.org/2001/04/xmlenc#aes256-cbc"
|
185
|
+
cipher = OpenSSL::Cipher.new('AES-256-CBC')
|
186
|
+
cipher.decrypt
|
187
|
+
cipher.padding = 0
|
188
|
+
cipher.key = find_cipher_key(
|
189
|
+
cipher,
|
190
|
+
key_derivation_name,
|
191
|
+
hashed_password,
|
192
|
+
salt,
|
193
|
+
iteration_count
|
194
|
+
)
|
195
|
+
cipher.iv = iv
|
196
|
+
else
|
197
|
+
raise ArgumentError, 'Unknown algorithm ' + algorithm
|
198
|
+
end
|
199
|
+
cipher
|
200
|
+
end
|
201
|
+
|
202
|
+
# Create a cipher key based on an ODS algorithm string from manifest.xml
|
203
|
+
def find_cipher_key(
|
204
|
+
cipher,
|
205
|
+
key_derivation_name,
|
206
|
+
hashed_password,
|
207
|
+
salt,
|
208
|
+
iteration_count
|
209
|
+
)
|
210
|
+
if key_derivation_name.eql? "PBKDF2"
|
211
|
+
key = OpenSSL::PKCS5.pbkdf2_hmac_sha1(
|
212
|
+
hashed_password,
|
213
|
+
salt,
|
214
|
+
iteration_count,
|
215
|
+
cipher.key_len
|
216
|
+
)
|
217
|
+
else
|
218
|
+
raise ArgumentError, 'Unknown key derivation name ' +
|
219
|
+
key_derivation_name
|
220
|
+
end
|
221
|
+
key
|
222
|
+
end
|
223
|
+
|
224
|
+
# Block decrypt raw bytes from the zip file based on the cipher
|
225
|
+
def decrypt(content_entry, cipher)
|
226
|
+
# Zip::Entry.extract writes a 0-length file when trying
|
227
|
+
# to extract an encrypted stream, so we read the
|
228
|
+
# raw bytes based on the offset and lengths
|
229
|
+
decrypted = ""
|
230
|
+
File.open(@filename, "rb") do |zipfile|
|
231
|
+
zipfile.seek(
|
232
|
+
content_entry.local_header_offset +
|
233
|
+
content_entry.calculate_local_header_size
|
234
|
+
)
|
235
|
+
total_to_read = content_entry.compressed_size
|
236
|
+
block_size = 4096
|
237
|
+
if block_size > total_to_read
|
238
|
+
block_size = total_to_read
|
239
|
+
end
|
240
|
+
while buffer = zipfile.read(block_size)
|
241
|
+
decrypted += cipher.update(buffer)
|
242
|
+
total_to_read -= buffer.length
|
243
|
+
if total_to_read == 0
|
244
|
+
break
|
245
|
+
end
|
246
|
+
if block_size > total_to_read
|
247
|
+
block_size = total_to_read
|
248
|
+
end
|
249
|
+
end
|
250
|
+
end
|
251
|
+
decrypted + cipher.final
|
252
|
+
end
|
253
|
+
|
254
|
+
def method_missing(m,*args)
|
255
|
+
read_labels
|
256
|
+
# is method name a label name
|
257
|
+
if @label.has_key?(m.to_s)
|
258
|
+
row,col = label(m.to_s)
|
259
|
+
cell(row,col)
|
260
|
+
else
|
261
|
+
# call super for methods like #a1
|
262
|
+
super
|
263
|
+
end
|
264
|
+
end
|
265
|
+
|
266
|
+
# Returns the content of a spreadsheet-cell.
|
267
|
+
# (1,1) is the upper left corner.
|
268
|
+
# (1,1), (1,'A'), ('A',1), ('a',1) all refers to the
|
269
|
+
# cell at the first line and first row.
|
270
|
+
def cell(row, col, sheet=nil)
|
271
|
+
sheet ||= default_sheet
|
272
|
+
read_cells(sheet)
|
273
|
+
row,col = normalize(row,col)
|
274
|
+
if celltype(row,col,sheet) == :date
|
275
|
+
yyyy,mm,dd = @cell[sheet][[row,col]].to_s.split('-')
|
276
|
+
return Date.new(yyyy.to_i,mm.to_i,dd.to_i)
|
277
|
+
end
|
278
|
+
@cell[sheet][[row,col]]
|
279
|
+
end
|
280
|
+
|
281
|
+
# Returns the formula at (row,col).
|
282
|
+
# Returns nil if there is no formula.
|
283
|
+
# The method #formula? checks if there is a formula.
|
284
|
+
def formula(row,col,sheet=nil)
|
285
|
+
sheet ||= default_sheet
|
286
|
+
read_cells(sheet)
|
287
|
+
row,col = normalize(row,col)
|
288
|
+
@formula[sheet][[row,col]]
|
289
|
+
end
|
290
|
+
alias_method :formula?, :formula
|
291
|
+
|
292
|
+
# returns each formula in the selected sheet as an array of elements
|
293
|
+
# [row, col, formula]
|
294
|
+
def formulas(sheet=nil)
|
295
|
+
sheet ||= default_sheet
|
296
|
+
read_cells(sheet)
|
297
|
+
if @formula[sheet]
|
298
|
+
@formula[sheet].each.collect do |elem|
|
299
|
+
[elem[0][0], elem[0][1], elem[1]]
|
300
|
+
end
|
301
|
+
else
|
302
|
+
[]
|
303
|
+
end
|
304
|
+
end
|
305
|
+
|
306
|
+
# Given a cell, return the cell's style
|
307
|
+
def font(row, col, sheet=nil)
|
308
|
+
sheet ||= default_sheet
|
309
|
+
read_cells(sheet)
|
310
|
+
row,col = normalize(row,col)
|
311
|
+
style_name = @style[sheet][[row,col]] || @style_defaults[sheet][col - 1] || 'Default'
|
312
|
+
@style_definitions[style_name]
|
313
|
+
end
|
314
|
+
|
315
|
+
# returns the type of a cell:
|
316
|
+
# * :float
|
317
|
+
# * :string
|
318
|
+
# * :date
|
319
|
+
# * :percentage
|
320
|
+
# * :formula
|
321
|
+
# * :time
|
322
|
+
# * :datetime
|
323
|
+
def celltype(row,col,sheet=nil)
|
324
|
+
sheet ||= default_sheet
|
325
|
+
read_cells(sheet)
|
326
|
+
row,col = normalize(row,col)
|
327
|
+
if @formula[sheet][[row,col]]
|
328
|
+
return :formula
|
329
|
+
else
|
330
|
+
@cell_type[sheet][[row,col]]
|
331
|
+
end
|
332
|
+
end
|
333
|
+
|
334
|
+
def sheets
|
335
|
+
doc.xpath("//*[local-name()='table']").map do |sheet|
|
336
|
+
sheet.attributes["name"].value
|
337
|
+
end
|
338
|
+
end
|
339
|
+
|
340
|
+
# version of the Roo::OpenOffice document
|
341
|
+
# at 2007 this is always "1.0"
|
342
|
+
def officeversion
|
343
|
+
oo_version
|
344
|
+
@officeversion
|
345
|
+
end
|
346
|
+
|
347
|
+
# shows the internal representation of all cells
|
348
|
+
# mainly for debugging purposes
|
349
|
+
def to_s(sheet=nil)
|
350
|
+
sheet ||= default_sheet
|
351
|
+
read_cells(sheet)
|
352
|
+
@cell[sheet].inspect
|
353
|
+
end
|
354
|
+
|
355
|
+
# returns the row,col values of the labelled cell
|
356
|
+
# (nil,nil) if label is not defined
|
357
|
+
def label(labelname)
|
358
|
+
read_labels
|
359
|
+
unless @label.size > 0
|
360
|
+
return nil,nil,nil
|
361
|
+
end
|
362
|
+
if @label.has_key? labelname
|
363
|
+
return @label[labelname][1].to_i,
|
364
|
+
::Roo::Utils.letter_to_number(@label[labelname][2]),
|
365
|
+
@label[labelname][0]
|
366
|
+
else
|
367
|
+
return nil,nil,nil
|
368
|
+
end
|
369
|
+
end
|
370
|
+
|
371
|
+
# Returns an array which all labels. Each element is an array with
|
372
|
+
# [labelname, [row,col,sheetname]]
|
373
|
+
def labels(sheet=nil)
|
374
|
+
read_labels
|
375
|
+
@label.map do |label|
|
376
|
+
[ label[0], # name
|
377
|
+
[ label[1][1].to_i, # row
|
378
|
+
::Roo::Utils.letter_to_number(label[1][2]), # column
|
379
|
+
label[1][0], # sheet
|
380
|
+
] ]
|
381
|
+
end
|
382
|
+
end
|
383
|
+
|
384
|
+
# returns the comment at (row/col)
|
385
|
+
# nil if there is no comment
|
386
|
+
def comment(row,col,sheet=nil)
|
387
|
+
sheet ||= default_sheet
|
388
|
+
read_cells(sheet)
|
389
|
+
row,col = normalize(row,col)
|
390
|
+
return nil unless @comment[sheet]
|
391
|
+
@comment[sheet][[row,col]]
|
392
|
+
end
|
393
|
+
|
394
|
+
# returns each comment in the selected sheet as an array of elements
|
395
|
+
# [row, col, comment]
|
396
|
+
def comments(sheet=nil)
|
397
|
+
sheet ||= default_sheet
|
398
|
+
read_comments(sheet) unless @comments_read[sheet]
|
399
|
+
if @comment[sheet]
|
400
|
+
@comment[sheet].each.collect do |elem|
|
401
|
+
[elem[0][0],elem[0][1],elem[1]]
|
402
|
+
end
|
403
|
+
else
|
404
|
+
[]
|
405
|
+
end
|
406
|
+
end
|
407
|
+
|
408
|
+
private
|
409
|
+
|
410
|
+
def doc
|
411
|
+
@doc ||= ::Roo::Utils.load_xml(File.join(@tmpdir, "roo_content.xml"))
|
412
|
+
end
|
413
|
+
|
414
|
+
# read the version of the OO-Version
|
415
|
+
def oo_version
|
416
|
+
doc.xpath("//*[local-name()='document-content']").each do |office|
|
417
|
+
@officeversion = attr(office,'version')
|
418
|
+
end
|
419
|
+
end
|
420
|
+
|
421
|
+
# helper function to set the internal representation of cells
|
422
|
+
def set_cell_values(sheet,x,y,i,v,value_type,formula,table_cell,str_v,style_name)
|
423
|
+
key = [y,x+i]
|
424
|
+
@cell_type[sheet] = {} unless @cell_type[sheet]
|
425
|
+
@cell_type[sheet][key] = Roo::OpenOffice.oo_type_2_roo_type(value_type)
|
426
|
+
@formula[sheet] = {} unless @formula[sheet]
|
427
|
+
if formula
|
428
|
+
['of:', 'oooc:'].each do |prefix|
|
429
|
+
if formula[0,prefix.length] == prefix
|
430
|
+
formula = formula[prefix.length..-1]
|
431
|
+
end
|
432
|
+
end
|
433
|
+
@formula[sheet][key] = formula
|
434
|
+
end
|
435
|
+
@cell[sheet] = {} unless @cell[sheet]
|
436
|
+
@style[sheet] = {} unless @style[sheet]
|
437
|
+
@style[sheet][key] = style_name
|
438
|
+
case @cell_type[sheet][key]
|
439
|
+
when :float
|
440
|
+
@cell[sheet][key] = v.to_f
|
441
|
+
when :string
|
442
|
+
@cell[sheet][key] = str_v
|
443
|
+
when :date
|
444
|
+
#TODO: if table_cell.attributes['date-value'].size != "XXXX-XX-XX".size
|
445
|
+
if attr(table_cell,'date-value').size != "XXXX-XX-XX".size
|
446
|
+
#-- dann ist noch eine Uhrzeit vorhanden
|
447
|
+
#-- "1961-11-21T12:17:18"
|
448
|
+
@cell[sheet][key] = DateTime.parse(attr(table_cell,'date-value').to_s)
|
449
|
+
@cell_type[sheet][key] = :datetime
|
450
|
+
else
|
451
|
+
@cell[sheet][key] = table_cell.attributes['date-value']
|
452
|
+
end
|
453
|
+
when :percentage
|
454
|
+
@cell[sheet][key] = v.to_f
|
455
|
+
when :time
|
456
|
+
hms = v.split(':')
|
457
|
+
@cell[sheet][key] = hms[0].to_i*3600 + hms[1].to_i*60 + hms[2].to_i
|
458
|
+
else
|
459
|
+
@cell[sheet][key] = v
|
460
|
+
end
|
461
|
+
end
|
462
|
+
|
463
|
+
# read all cells in the selected sheet
|
464
|
+
#--
|
465
|
+
# the following construct means '4 blanks'
|
466
|
+
# some content <text:s text:c="3"/>
|
467
|
+
#++
|
468
|
+
def read_cells(sheet = default_sheet)
|
469
|
+
validate_sheet!(sheet)
|
470
|
+
return if @cells_read[sheet]
|
471
|
+
|
472
|
+
sheet_found = false
|
473
|
+
doc.xpath("//*[local-name()='table']").each do |ws|
|
474
|
+
if sheet == attr(ws,'name')
|
475
|
+
sheet_found = true
|
476
|
+
col = 1
|
477
|
+
row = 1
|
478
|
+
ws.children.each do |table_element|
|
479
|
+
case table_element.name
|
480
|
+
when 'table-column'
|
481
|
+
@style_defaults[sheet] << table_element.attributes['default-cell-style-name']
|
482
|
+
when 'table-row'
|
483
|
+
if table_element.attributes['number-rows-repeated']
|
484
|
+
skip_row = attr(table_element,'number-rows-repeated').to_s.to_i
|
485
|
+
row = row + skip_row - 1
|
486
|
+
end
|
487
|
+
table_element.children.each do |cell|
|
488
|
+
skip_col = attr(cell, 'number-columns-repeated')
|
489
|
+
formula = attr(cell,'formula')
|
490
|
+
value_type = attr(cell,'value-type')
|
491
|
+
v = attr(cell,'value')
|
492
|
+
style_name = attr(cell,'style-name')
|
493
|
+
case value_type
|
494
|
+
when 'string'
|
495
|
+
str_v = ''
|
496
|
+
# insert \n if there is more than one paragraph
|
497
|
+
para_count = 0
|
498
|
+
cell.children.each do |str|
|
499
|
+
# begin comments
|
500
|
+
=begin
|
501
|
+
- <table:table-cell office:value-type="string">
|
502
|
+
- <office:annotation office:display="true" draw:style-name="gr1" draw:text-style-name="P1" svg:width="1.1413in" svg:height="0.3902in" svg:x="2.0142in" svg:y="0in" draw:caption-point-x="-0.2402in" draw:caption-point-y="0.5661in">
|
503
|
+
<dc:date>2011-09-20T00:00:00</dc:date>
|
504
|
+
<text:p text:style-name="P1">Kommentar fuer B4</text:p>
|
505
|
+
</office:annotation>
|
506
|
+
<text:p>B4 (mit Kommentar)</text:p>
|
507
|
+
</table:table-cell>
|
508
|
+
=end
|
509
|
+
if str.name == 'annotation'
|
510
|
+
str.children.each do |annotation|
|
511
|
+
if annotation.name == 'p'
|
512
|
+
# @comment ist ein Hash mit Sheet als Key (wie bei @cell)
|
513
|
+
# innerhalb eines Elements besteht ein Eintrag aus einem
|
514
|
+
# weiteren Hash mit Key [row,col] und dem eigentlichen
|
515
|
+
# Kommentartext als Inhalt
|
516
|
+
@comment[sheet] = Hash.new unless @comment[sheet]
|
517
|
+
key = [row,col]
|
518
|
+
@comment[sheet][key] = annotation.text
|
519
|
+
end
|
520
|
+
end
|
521
|
+
end
|
522
|
+
# end comments
|
523
|
+
if str.name == 'p'
|
524
|
+
v = str.content
|
525
|
+
str_v += "\n" if para_count > 0
|
526
|
+
para_count += 1
|
527
|
+
if str.children.size > 1
|
528
|
+
str_v += children_to_string(str.children)
|
529
|
+
else
|
530
|
+
str.children.each do |child|
|
531
|
+
str_v += child.content #.text
|
532
|
+
end
|
533
|
+
end
|
534
|
+
str_v.gsub!(/'/,"'") # special case not supported by unescapeHTML
|
535
|
+
str_v = CGI.unescapeHTML(str_v)
|
536
|
+
end # == 'p'
|
537
|
+
end
|
538
|
+
when 'time'
|
539
|
+
cell.children.each do |str|
|
540
|
+
if str.name == 'p'
|
541
|
+
v = str.content
|
542
|
+
end
|
543
|
+
end
|
544
|
+
when '', nil, 'date', 'percentage', 'float'
|
545
|
+
#
|
546
|
+
when 'boolean'
|
547
|
+
v = attr(cell,'boolean-value').to_s
|
548
|
+
else
|
549
|
+
# raise "unknown type #{value_type}"
|
550
|
+
end
|
551
|
+
if skip_col
|
552
|
+
if v != nil or cell.attributes['date-value']
|
553
|
+
0.upto(skip_col.to_i-1) do |i|
|
554
|
+
set_cell_values(sheet,col,row,i,v,value_type,formula,cell,str_v,style_name)
|
555
|
+
end
|
556
|
+
end
|
557
|
+
col += (skip_col.to_i - 1)
|
558
|
+
end # if skip
|
559
|
+
set_cell_values(sheet,col,row,0,v,value_type,formula,cell,str_v,style_name)
|
560
|
+
col += 1
|
561
|
+
end
|
562
|
+
row += 1
|
563
|
+
col = 1
|
564
|
+
end
|
565
|
+
end
|
566
|
+
end
|
567
|
+
end
|
568
|
+
doc.xpath("//*[local-name()='automatic-styles']").each do |style|
|
569
|
+
read_styles(style)
|
570
|
+
end
|
571
|
+
if !sheet_found
|
572
|
+
raise RangeError
|
573
|
+
end
|
574
|
+
@cells_read[sheet] = true
|
575
|
+
@comments_read[sheet] = true
|
576
|
+
end
|
577
|
+
|
578
|
+
# Only calls read_cells because Roo::Base calls read_comments
|
579
|
+
# whereas the reading of comments is done in read_cells for Roo::OpenOffice-objects
|
580
|
+
def read_comments(sheet=nil)
|
581
|
+
read_cells(sheet)
|
582
|
+
end
|
583
|
+
|
584
|
+
def read_labels
|
585
|
+
@label ||= Hash[doc.xpath("//table:named-range").map do |ne|
|
586
|
+
#-
|
587
|
+
# $Sheet1.$C$5
|
588
|
+
#+
|
589
|
+
name = attr(ne,'name').to_s
|
590
|
+
sheetname,coords = attr(ne,'cell-range-address').to_s.split('.$')
|
591
|
+
col, row = coords.split('$')
|
592
|
+
sheetname = sheetname[1..-1] if sheetname[0,1] == '$'
|
593
|
+
[name, [sheetname,row,col]]
|
594
|
+
end]
|
595
|
+
end
|
596
|
+
|
597
|
+
def read_styles(style_elements)
|
598
|
+
@style_definitions['Default'] = Roo::Font.new
|
599
|
+
style_elements.each do |style|
|
600
|
+
next unless style.name == 'style'
|
601
|
+
style_name = attr(style,'name')
|
602
|
+
style.each do |properties|
|
603
|
+
font = Roo::OpenOffice::Font.new
|
604
|
+
font.bold = attr(properties,'font-weight')
|
605
|
+
font.italic = attr(properties,'font-style')
|
606
|
+
font.underline = attr(properties,'text-underline-style')
|
607
|
+
@style_definitions[style_name] = font
|
608
|
+
end
|
609
|
+
end
|
610
|
+
end
|
611
|
+
|
612
|
+
A_ROO_TYPE = {
|
613
|
+
"float" => :float,
|
614
|
+
"string" => :string,
|
615
|
+
"date" => :date,
|
616
|
+
"percentage" => :percentage,
|
617
|
+
"time" => :time,
|
618
|
+
}
|
619
|
+
|
620
|
+
def self.oo_type_2_roo_type(ootype)
|
621
|
+
return A_ROO_TYPE[ootype]
|
622
|
+
end
|
623
|
+
|
624
|
+
# helper method to convert compressed spaces and other elements within
|
625
|
+
# an text into a string
|
626
|
+
def children_to_string(children)
|
627
|
+
result = ''
|
628
|
+
children.each {|child|
|
629
|
+
if child.text?
|
630
|
+
result = result + child.content
|
631
|
+
else
|
632
|
+
if child.name == 's'
|
633
|
+
compressed_spaces = child.attributes['c'].to_s.to_i
|
634
|
+
# no explicit number means a count of 1:
|
635
|
+
if compressed_spaces == 0
|
636
|
+
compressed_spaces = 1
|
637
|
+
end
|
638
|
+
result = result + " "*compressed_spaces
|
639
|
+
else
|
640
|
+
result = result + child.content
|
641
|
+
end
|
642
|
+
end
|
643
|
+
}
|
644
|
+
result
|
645
|
+
end
|
646
|
+
|
647
|
+
def attr(node, attr_name)
|
648
|
+
if node.attributes[attr_name]
|
649
|
+
node.attributes[attr_name].value
|
650
|
+
end
|
651
|
+
end
|
652
|
+
end
|