csv-utils 0.3.4 → 0.3.9
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.ruby-version +1 -1
- data/Gemfile.lock +33 -31
- data/bin/csv-explorer +12 -0
- data/bin/csv-find-error +3 -1
- data/bin/csv-validator +17 -13
- data/csv-utils.gemspec +1 -1
- data/lib/csv-utils.rb +2 -0
- data/lib/csv_utils/csv_compare.rb +83 -0
- data/lib/csv_utils/csv_iterator.rb +51 -0
- data/lib/csv_utils/csv_report.rb +10 -5
- data/lib/csv_utils/csv_sort.rb +6 -6
- data/lib/csv_utils/csv_wrapper.rb +5 -1
- metadata +7 -3
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: b5b6f30da64b899586ef7d801904b8da9b01b6053f9d975f0702edcc5a7b65e0
|
4
|
+
data.tar.gz: e7e5006c6b63a9b8472e4658c748071619d09adf0ca73a84109b43b8862f1bcd
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 7ded9a4318f44f77b6c2ccf72a844a7cec734745052c62be5e19c2dbadc8e989b3c89528058fabfd63cbcb59a1575293b9cc87a343bb4d68dbad3051ba04616d
|
7
|
+
data.tar.gz: dd8f0d7cd953eee05489423d0ef55e0b486500d2499d442d7d523636ced12858a2e9cec7ad60ce9d0126516121674d3c1e2a879ea89bb3a93cf7100709204b9d
|
data/.ruby-version
CHANGED
@@ -1 +1 @@
|
|
1
|
-
|
1
|
+
3.0.0
|
data/Gemfile.lock
CHANGED
@@ -1,50 +1,52 @@
|
|
1
1
|
GEM
|
2
2
|
remote: http://rubygems.org/
|
3
3
|
specs:
|
4
|
-
ast (2.4.
|
5
|
-
diff-lcs (1.
|
6
|
-
docile (1.3.
|
4
|
+
ast (2.4.2)
|
5
|
+
diff-lcs (1.4.4)
|
6
|
+
docile (1.3.5)
|
7
7
|
inheritance-helper (0.1.5)
|
8
|
-
parallel (1.
|
9
|
-
parser (
|
8
|
+
parallel (1.20.1)
|
9
|
+
parser (3.0.0.0)
|
10
10
|
ast (~> 2.4.1)
|
11
11
|
rainbow (3.0.0)
|
12
|
-
rake (13.0.
|
13
|
-
regexp_parser (1.
|
12
|
+
rake (13.0.3)
|
13
|
+
regexp_parser (2.1.1)
|
14
14
|
rexml (3.2.4)
|
15
|
-
rspec (3.
|
16
|
-
rspec-core (~> 3.
|
17
|
-
rspec-expectations (~> 3.
|
18
|
-
rspec-mocks (~> 3.
|
19
|
-
rspec-core (3.
|
20
|
-
rspec-support (~> 3.
|
21
|
-
rspec-expectations (3.
|
15
|
+
rspec (3.10.0)
|
16
|
+
rspec-core (~> 3.10.0)
|
17
|
+
rspec-expectations (~> 3.10.0)
|
18
|
+
rspec-mocks (~> 3.10.0)
|
19
|
+
rspec-core (3.10.1)
|
20
|
+
rspec-support (~> 3.10.0)
|
21
|
+
rspec-expectations (3.10.1)
|
22
22
|
diff-lcs (>= 1.2.0, < 2.0)
|
23
|
-
rspec-support (~> 3.
|
24
|
-
rspec-mocks (3.
|
23
|
+
rspec-support (~> 3.10.0)
|
24
|
+
rspec-mocks (3.10.2)
|
25
25
|
diff-lcs (>= 1.2.0, < 2.0)
|
26
|
-
rspec-support (~> 3.
|
27
|
-
rspec-support (3.
|
28
|
-
rubocop (
|
26
|
+
rspec-support (~> 3.10.0)
|
27
|
+
rspec-support (3.10.2)
|
28
|
+
rubocop (1.11.0)
|
29
29
|
parallel (~> 1.10)
|
30
|
-
parser (>=
|
30
|
+
parser (>= 3.0.0.0)
|
31
31
|
rainbow (>= 2.2.2, < 4.0)
|
32
|
-
regexp_parser (>= 1.
|
32
|
+
regexp_parser (>= 1.8, < 3.0)
|
33
33
|
rexml
|
34
|
-
rubocop-ast (>=
|
34
|
+
rubocop-ast (>= 1.2.0, < 2.0)
|
35
35
|
ruby-progressbar (~> 1.7)
|
36
|
-
unicode-display_width (>= 1.4.0, <
|
37
|
-
rubocop-ast (
|
38
|
-
parser (>= 2.7.
|
39
|
-
ruby-progressbar (1.
|
40
|
-
simplecov (0.
|
36
|
+
unicode-display_width (>= 1.4.0, < 3.0)
|
37
|
+
rubocop-ast (1.4.1)
|
38
|
+
parser (>= 2.7.1.5)
|
39
|
+
ruby-progressbar (1.11.0)
|
40
|
+
simplecov (0.21.2)
|
41
41
|
docile (~> 1.1)
|
42
42
|
simplecov-html (~> 0.11)
|
43
|
-
|
44
|
-
|
43
|
+
simplecov_json_formatter (~> 0.1)
|
44
|
+
simplecov-html (0.12.3)
|
45
|
+
simplecov_json_formatter (0.1.2)
|
46
|
+
unicode-display_width (2.0.0)
|
45
47
|
|
46
48
|
PLATFORMS
|
47
|
-
|
49
|
+
x86_64-darwin-20
|
48
50
|
|
49
51
|
DEPENDENCIES
|
50
52
|
inheritance-helper
|
@@ -54,4 +56,4 @@ DEPENDENCIES
|
|
54
56
|
simplecov
|
55
57
|
|
56
58
|
BUNDLED WITH
|
57
|
-
|
59
|
+
2.2.3
|
data/bin/csv-explorer
ADDED
data/bin/csv-find-error
CHANGED
@@ -3,10 +3,12 @@
|
|
3
3
|
require 'csv'
|
4
4
|
require 'shellwords'
|
5
5
|
|
6
|
+
prev_row = nil
|
6
7
|
begin
|
7
|
-
CSV.open(ARGV[0], 'rb').each { }
|
8
|
+
CSV.open(ARGV[0], 'rb').each { |row| prev_row = row }
|
8
9
|
rescue CSV::MalformedCSVError => e
|
9
10
|
puts e.class.to_s + ': ' + e.message
|
11
|
+
puts "previous row was #{prev_row}"
|
10
12
|
if e.message =~ /line (\d+)/
|
11
13
|
lineno = $1.to_i
|
12
14
|
cmd = "csv-readline #{Shellwords.escape(ARGV[0])} #{lineno}"
|
data/bin/csv-validator
CHANGED
@@ -35,20 +35,24 @@ id_column_name = ARGV[1]
|
|
35
35
|
headers = csv.shift
|
36
36
|
strip_bom!(headers[0])
|
37
37
|
|
38
|
-
|
39
|
-
|
40
|
-
|
41
|
-
|
42
|
-
exit 1
|
43
|
-
end
|
44
|
-
|
45
|
-
id_column_num = headers.index(id_column_name)
|
38
|
+
id_column_name ||= headers[0]
|
39
|
+
unless headers.include?(id_column_name)
|
40
|
+
$stderr.puts("header #{id_column_name} not found in current set of headers")
|
41
|
+
exit 1
|
46
42
|
end
|
47
43
|
|
44
|
+
id_column_num = headers.index(id_column_name)
|
45
|
+
|
48
46
|
out = nil
|
49
|
-
|
50
|
-
out
|
51
|
-
|
47
|
+
out_proc = Proc.new do |row|
|
48
|
+
out ||=
|
49
|
+
begin
|
50
|
+
out = CSV.open('utf8-correctsion.csv', 'wb')
|
51
|
+
out << [id_column_name, 'Row', 'Col', 'Header', 'Value']
|
52
|
+
out
|
53
|
+
end
|
54
|
+
|
55
|
+
out << row
|
52
56
|
end
|
53
57
|
|
54
58
|
csv_lineno = 1
|
@@ -61,12 +65,12 @@ while (row = csv.shift)
|
|
61
65
|
end
|
62
66
|
|
63
67
|
row.each_with_index do |col, idx|
|
64
|
-
next if utf8?(col)
|
68
|
+
next if col.nil? || utf8?(col)
|
65
69
|
|
66
70
|
$stderr.puts "row(#{csv_lineno}),col(#{idx + 1}) #{headers[idx]}: none UTF-8 characters found in \"#{col}\""
|
67
71
|
if (col_utf8_encoded = convert_to_utf8(col, detect_encoding(col)))
|
68
72
|
puts "row(#{csv_lineno}),col(#{idx + 1}) #{headers[idx]}: converted to UTF-8 from #{detect_encoding(col)} \"#{col_utf8_encoded}\""
|
69
|
-
|
73
|
+
out_proc.call [row[id_column_num], csv_lineno, (idx + 1), headers[idx], col_utf8_encoded]
|
70
74
|
else
|
71
75
|
$stderr.puts "row(#{csv_lineno}),col(#{idx + 1}) #{headers[idx]}: unknown character encoding"
|
72
76
|
end
|
data/csv-utils.gemspec
CHANGED
data/lib/csv-utils.rb
CHANGED
@@ -2,7 +2,9 @@ require 'csv'
|
|
2
2
|
|
3
3
|
# Collection of tools for working with CSV files.
|
4
4
|
module CSVUtils
|
5
|
+
autoload :CSVCompare, 'csv_utils/csv_compare'
|
5
6
|
autoload :CSVExtender, 'csv_utils/csv_extender'
|
7
|
+
autoload :CSVIterator, 'csv_utils/csv_iterator'
|
6
8
|
autoload :CSVOptions, 'csv_utils/csv_options'
|
7
9
|
autoload :CSVReport, 'csv_utils/csv_report'
|
8
10
|
autoload :CSVRow, 'csv_utils/csv_row'
|
@@ -0,0 +1,83 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
# CSVUtils::CSVCompare purpose is to determine which rows in the secondary_data_file need to be created, deleted or updated
|
4
|
+
# **requires both CSV files to be sorted on the same columns, CSVUtils::CSVSort can accomplish this
|
5
|
+
# In order to receive updates, update_comparison_columns must configured or use inheritance and change the update_row? method
|
6
|
+
class CSVUtils::CSVCompare
|
7
|
+
# primary_data_file is the source of truth
|
8
|
+
# compare_proc used to compare the id column(s)
|
9
|
+
# update_comparison_columns column(s) to compare for equality, ex: updated_at, timestamp, hash
|
10
|
+
# caveat: update_comparison_columns need to be in both csv files
|
11
|
+
attr_reader :primary_data_file,
|
12
|
+
:update_comparison_columns,
|
13
|
+
:compare_proc
|
14
|
+
|
15
|
+
def initialize(primary_data_file, update_comparison_columns=nil, &block)
|
16
|
+
@primary_data_file = primary_data_file
|
17
|
+
@update_comparison_columns = update_comparison_columns
|
18
|
+
@compare_proc = block
|
19
|
+
end
|
20
|
+
|
21
|
+
def compare(secondary_data_file)
|
22
|
+
src = CSV.open(primary_data_file)
|
23
|
+
src_headers = src.shift
|
24
|
+
dest = CSV.open(secondary_data_file)
|
25
|
+
dest_headers = dest.shift
|
26
|
+
|
27
|
+
read_next_src = true
|
28
|
+
read_next_dest = true
|
29
|
+
|
30
|
+
while(!src.eof? || !dest.eof?)
|
31
|
+
src_record = next_record_from_file(src_headers, src) if read_next_src
|
32
|
+
dest_record = next_record_from_file(dest_headers, dest) if read_next_dest
|
33
|
+
|
34
|
+
if ! src_record
|
35
|
+
read_next_src = false
|
36
|
+
read_next_dest = true
|
37
|
+
|
38
|
+
yield :delete, dest_record
|
39
|
+
elsif ! dest_record
|
40
|
+
read_next_src = true
|
41
|
+
read_next_dest = false
|
42
|
+
|
43
|
+
yield :create, src_record
|
44
|
+
elsif compare_proc.call(src_record, dest_record) == 0
|
45
|
+
read_next_src = true
|
46
|
+
read_next_dest = true
|
47
|
+
|
48
|
+
yield(:update, src_record) if update_row?(src_record, dest_record)
|
49
|
+
elsif compare_proc.call(src_record, dest_record) > 0
|
50
|
+
read_next_src = false
|
51
|
+
read_next_dest = true
|
52
|
+
|
53
|
+
yield :delete, dest_record
|
54
|
+
else
|
55
|
+
read_next_src = true
|
56
|
+
read_next_dest = false
|
57
|
+
|
58
|
+
yield :create, src_record
|
59
|
+
end
|
60
|
+
end
|
61
|
+
|
62
|
+
src.close
|
63
|
+
dest.close
|
64
|
+
end
|
65
|
+
|
66
|
+
private
|
67
|
+
|
68
|
+
def next_record_from_file(headers, file)
|
69
|
+
return nil if file.eof?
|
70
|
+
|
71
|
+
Hash[headers.zip(file.shift)]
|
72
|
+
end
|
73
|
+
|
74
|
+
def update_row?(src_record, dest_record)
|
75
|
+
return false unless update_comparison_columns
|
76
|
+
|
77
|
+
update_comparison_columns.each do |column_name|
|
78
|
+
return true unless src_record[column_name] == dest_record[column_name]
|
79
|
+
end
|
80
|
+
|
81
|
+
false
|
82
|
+
end
|
83
|
+
end
|
@@ -0,0 +1,51 @@
|
|
1
|
+
# Search a CSV given a series of steps
|
2
|
+
class CSVUtils::CSVIterator
|
3
|
+
include Enumerable
|
4
|
+
|
5
|
+
attr_reader :prev_row
|
6
|
+
|
7
|
+
class RowWrapper < Hash
|
8
|
+
attr_accessor :lineno
|
9
|
+
|
10
|
+
def self.create(headers, row, lineno)
|
11
|
+
row_wrapper = RowWrapper[headers.zip(row)]
|
12
|
+
row_wrapper.lineno = lineno
|
13
|
+
row_wrapper
|
14
|
+
end
|
15
|
+
|
16
|
+
def to_pretty_s
|
17
|
+
reject { |_, v| v.strip.empty? }
|
18
|
+
.each_with_index
|
19
|
+
.map { |(k, v), idx| sprintf(' %-3d %s: %s', idx+1, k, v) }
|
20
|
+
.join("\n") + "\n"
|
21
|
+
end
|
22
|
+
end
|
23
|
+
|
24
|
+
def initialize(src_csv, csv_options = {})
|
25
|
+
@src_csv = CSVUtils::CSVWrapper.new(src_csv, 'rb', csv_options)
|
26
|
+
end
|
27
|
+
|
28
|
+
def each(headers = nil)
|
29
|
+
@src_csv.rewind
|
30
|
+
|
31
|
+
lineno = 0
|
32
|
+
unless headers
|
33
|
+
headers = @src_csv.shift
|
34
|
+
strip_bom!(headers[0])
|
35
|
+
lineno += 1
|
36
|
+
end
|
37
|
+
|
38
|
+
@prev_row = nil
|
39
|
+
while (row = @src_csv.shift)
|
40
|
+
lineno += 1
|
41
|
+
yield RowWrapper.create(headers, row, lineno)
|
42
|
+
@prev_row = row
|
43
|
+
end
|
44
|
+
end
|
45
|
+
|
46
|
+
private
|
47
|
+
|
48
|
+
def strip_bom!(col)
|
49
|
+
col.sub!("\xEF\xBB\xBF".force_encoding('ASCII-8BIT'), '')
|
50
|
+
end
|
51
|
+
end
|
data/lib/csv_utils/csv_report.rb
CHANGED
@@ -9,19 +9,20 @@ module CSVUtils
|
|
9
9
|
if csv.is_a?(String)
|
10
10
|
@must_close = true
|
11
11
|
mode = csv_options.delete(:mode) || 'wb'
|
12
|
-
CSV.open(csv, mode, csv_options)
|
12
|
+
CSV.open(csv, mode, **csv_options)
|
13
13
|
else
|
14
14
|
@must_close = false
|
15
15
|
csv
|
16
16
|
end
|
17
17
|
|
18
|
-
|
18
|
+
add_headers(headers) if headers
|
19
|
+
|
20
|
+
generate(&block) if block
|
19
21
|
end
|
20
22
|
|
21
|
-
def generate
|
22
|
-
add_headers(headers) if headers
|
23
|
+
def generate
|
23
24
|
yield self
|
24
|
-
|
25
|
+
close if @must_close
|
25
26
|
end
|
26
27
|
|
27
28
|
def append(csv_row)
|
@@ -37,5 +38,9 @@ module CSVUtils
|
|
37
38
|
def add_headers(csv_row)
|
38
39
|
append(csv_row.is_a?(Array) ? csv_row : csv_row.csv_headers)
|
39
40
|
end
|
41
|
+
|
42
|
+
def close
|
43
|
+
@csv.close
|
44
|
+
end
|
40
45
|
end
|
41
46
|
end
|
data/lib/csv_utils/csv_sort.rb
CHANGED
@@ -25,9 +25,9 @@ class CSVUtils::CSVSort
|
|
25
25
|
private
|
26
26
|
|
27
27
|
def merge_sort_csv_files(src_csv_file1, src_csv_file2, dest_csv_file)
|
28
|
-
src1 = CSV.open(src_csv_file1, 'rb', csv_options)
|
29
|
-
src2 = CSV.open(src_csv_file2, 'rb', csv_options)
|
30
|
-
dest = CSV.open(dest_csv_file, 'wb', csv_options)
|
28
|
+
src1 = CSV.open(src_csv_file1, 'rb', **csv_options)
|
29
|
+
src2 = CSV.open(src_csv_file2, 'rb', **csv_options)
|
30
|
+
dest = CSV.open(dest_csv_file, 'wb', **csv_options)
|
31
31
|
|
32
32
|
if @headers
|
33
33
|
dest << @headers
|
@@ -66,7 +66,7 @@ class CSVUtils::CSVSort
|
|
66
66
|
end
|
67
67
|
|
68
68
|
def create_sorted_csv_part_files(batch_size, &block)
|
69
|
-
src = CSV.open(csv_file, 'rb', csv_options)
|
69
|
+
src = CSV.open(csv_file, 'rb', **csv_options)
|
70
70
|
|
71
71
|
@headers = src.shift if has_headers
|
72
72
|
|
@@ -74,7 +74,7 @@ class CSVUtils::CSVSort
|
|
74
74
|
create_batch_part_proc = Proc.new do
|
75
75
|
batch.sort!(&block)
|
76
76
|
@csv_part_files << "#{new_csv_file}.part.#{@csv_part_files.size}"
|
77
|
-
CSV.open(@csv_part_files.last, 'wb', csv_options) do |csv|
|
77
|
+
CSV.open(@csv_part_files.last, 'wb', **csv_options) do |csv|
|
78
78
|
csv << @headers if @headers
|
79
79
|
batch.each { |row| csv << row }
|
80
80
|
end
|
@@ -107,6 +107,6 @@ class CSVUtils::CSVSort
|
|
107
107
|
File.unlink(csv_part_file2)
|
108
108
|
end
|
109
109
|
|
110
|
-
FileUtils.mv(@csv_part_files.last, new_csv_file)
|
110
|
+
FileUtils.mv(@csv_part_files.last || @csv_file, new_csv_file)
|
111
111
|
end
|
112
112
|
end
|
@@ -20,7 +20,7 @@ class CSVUtils::CSVWrapper
|
|
20
20
|
def open(csv, mode, csv_options)
|
21
21
|
if csv.is_a?(String)
|
22
22
|
@close_when_done = true
|
23
|
-
@csv = CSV.open(csv, mode, csv_options)
|
23
|
+
@csv = CSV.open(csv, mode, **csv_options)
|
24
24
|
else
|
25
25
|
@close_when_done = false
|
26
26
|
@csv = csv
|
@@ -35,6 +35,10 @@ class CSVUtils::CSVWrapper
|
|
35
35
|
csv.shift
|
36
36
|
end
|
37
37
|
|
38
|
+
def rewind
|
39
|
+
csv.rewind
|
40
|
+
end
|
41
|
+
|
38
42
|
def close
|
39
43
|
csv.close if close_when_done?
|
40
44
|
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: csv-utils
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.3.
|
4
|
+
version: 0.3.9
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Doug Youch
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date:
|
11
|
+
date: 2021-03-12 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: inheritance-helper
|
@@ -28,6 +28,7 @@ description: Tools for debugging malformed CSV files
|
|
28
28
|
email: dougyouch@gmail.com
|
29
29
|
executables:
|
30
30
|
- csv-change-eol
|
31
|
+
- csv-explorer
|
31
32
|
- csv-find-error
|
32
33
|
- csv-readline
|
33
34
|
- csv-validator
|
@@ -42,12 +43,15 @@ files:
|
|
42
43
|
- LICENSE
|
43
44
|
- README.md
|
44
45
|
- bin/csv-change-eol
|
46
|
+
- bin/csv-explorer
|
45
47
|
- bin/csv-find-error
|
46
48
|
- bin/csv-readline
|
47
49
|
- bin/csv-validator
|
48
50
|
- csv-utils.gemspec
|
49
51
|
- lib/csv-utils.rb
|
52
|
+
- lib/csv_utils/csv_compare.rb
|
50
53
|
- lib/csv_utils/csv_extender.rb
|
54
|
+
- lib/csv_utils/csv_iterator.rb
|
51
55
|
- lib/csv_utils/csv_options.rb
|
52
56
|
- lib/csv_utils/csv_report.rb
|
53
57
|
- lib/csv_utils/csv_row.rb
|
@@ -74,7 +78,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
|
|
74
78
|
- !ruby/object:Gem::Version
|
75
79
|
version: '0'
|
76
80
|
requirements: []
|
77
|
-
rubygems_version: 3.
|
81
|
+
rubygems_version: 3.2.3
|
78
82
|
signing_key:
|
79
83
|
specification_version: 4
|
80
84
|
summary: CSV Utils
|