ipynbdiff 0.3.2 → 0.3.6
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +57 -2
- data/ipynbdiff.gemspec +1 -1
- data/lib/ipynbdiff.rb +28 -13
- data/lib/output_transformer.rb +61 -0
- data/lib/transformer.rb +60 -61
- metadata +3 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 10f77cf53513157724c6e1d8c8abc4659e3254dd77c3c14ebb2aaabd3155639c
|
4
|
+
data.tar.gz: 10815dc6a9cb76c73b4d6c9f712bdb28be9217a2b40efbcf076342e9f5d34811
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 06e895b990e7099d094b6e78c8a54fc6e5c618b28a9ae266cdb47a127189d67df7a7468f5a96fb2da21529f30911dfca761d7e4b0fc5452fc351e1878c697811
|
7
|
+
data.tar.gz: 54ee8fb62e05b130b304989787ee1032ffe41a2dc68dcdfde5bf7480588bdcd3a4f9a20c3b63eeb7cf21dd0ed89527b34d39a7ec15170f7cddda6114ea8fac0a
|
data/README.md
CHANGED
@@ -1,3 +1,58 @@
|
|
1
|
-
#
|
1
|
+
# IpynbDiff: Better diff for Jupyter Notebooks
|
2
2
|
|
3
|
-
This is a
|
3
|
+
This is a simple diff tool that cleans up jupyter notebooks, transforming each [notebook](example/1/from.ipynb)
|
4
|
+
into a [readable markdown file](example/1/from_html.md), keeping the output of cells, and running the
|
5
|
+
diff after. Markdowns are generated using an opinionated Jupyter to Markdown conversion. This means
|
6
|
+
that the entire file is readable on the diff.
|
7
|
+
|
8
|
+
The result are diffs that are much easier to read:
|
9
|
+
|
10
|
+
| Diff | IpynbDiff - HTML | IpynbDiff - Percent |
|
11
|
+
| ------ | ------ | ------ |
|
12
|
+
| [Here](example/diff.txt) | [Here](example/ipynbdiff.txt) | [Here](example/ipynbdiff_percent.txt) |
|
13
|
+
| ![](example/img/diff.png) | ![](example/img/ipynbdiff_html.png) | ![](example/img/ipynbdiff_percent.png) |
|
14
|
+
|
15
|
+
|
16
|
+
This started as a port of This is a port of [ipynbdiff](https://gitlab.com/gitlab-org/incubation-engineering/mlops/ipynbdiff),
|
17
|
+
but now has extended functionality although not working as git driver.
|
18
|
+
|
19
|
+
## Usage
|
20
|
+
|
21
|
+
### Generating diffs
|
22
|
+
|
23
|
+
```ruby
|
24
|
+
IpynbDiff.diff(from_path, to_path, options)
|
25
|
+
```
|
26
|
+
|
27
|
+
Options:
|
28
|
+
|
29
|
+
```ruby
|
30
|
+
@default_transform_options = {
|
31
|
+
preprocess_input: TRUE, # Whether the input should be transformed
|
32
|
+
write_output_to: nil, # Pass a path to save the output to a file
|
33
|
+
format: :text, # These are the formats Diffy accepts https://github.com/samg/diffy
|
34
|
+
sources_are_files: FALSE, # Weather to use the from/to as string or path to a file
|
35
|
+
raise_if_invalid_notebook: FALSE, # Raises an error if the notebooks are invalid, otherwise returns nil
|
36
|
+
transform_options: @default_transform_options, # See below for transform options
|
37
|
+
diff_opts: {
|
38
|
+
include_diff_info: FALSE # These are passed to Diffy https://github.com/samg/diffy
|
39
|
+
}
|
40
|
+
}
|
41
|
+
```
|
42
|
+
|
43
|
+
### Transforming the notebooks
|
44
|
+
|
45
|
+
It might be necessary to have the transformed files in addition to the diff.
|
46
|
+
|
47
|
+
```ruby
|
48
|
+
IpynbDiff.transform(notebook, options)
|
49
|
+
```
|
50
|
+
|
51
|
+
Options:
|
52
|
+
|
53
|
+
```ruby
|
54
|
+
@default_transform_options = {
|
55
|
+
include_metadata: FALSE, # Whether to include or not the notebook metadata (kernel, language, etc)
|
56
|
+
cell_decorator: :html # :html is useful to add styling with css, :percent is better for text format
|
57
|
+
}
|
58
|
+
```
|
data/ipynbdiff.gemspec
CHANGED
@@ -2,7 +2,7 @@
|
|
2
2
|
|
3
3
|
Gem::Specification.new do |s|
|
4
4
|
s.name = 'ipynbdiff'
|
5
|
-
s.version = '0.3.
|
5
|
+
s.version = '0.3.6'
|
6
6
|
s.summary = 'Human Readable diffs for Jupyter Notebooks'
|
7
7
|
s.description = 'Better diff for Jupyter Notebooks by first preprocessing them and removing clutter'
|
8
8
|
s.authors = ['Eduardo Bonet']
|
data/lib/ipynbdiff.rb
CHANGED
@@ -2,42 +2,57 @@
|
|
2
2
|
|
3
3
|
# Human Readable Jupyter Diffs
|
4
4
|
module IpynbDiff
|
5
|
-
require 'transformer
|
5
|
+
require 'transformer'
|
6
6
|
require 'diffy'
|
7
7
|
|
8
|
-
@
|
8
|
+
@default_transform_options = {
|
9
|
+
include_metadata: FALSE,
|
10
|
+
cell_decorator: :html
|
11
|
+
}
|
12
|
+
|
13
|
+
@default_diff_options = {
|
9
14
|
preprocess_input: TRUE,
|
10
15
|
write_output_to: nil,
|
11
16
|
format: :text,
|
12
17
|
sources_are_files: FALSE,
|
13
|
-
|
18
|
+
raise_if_invalid_notebook: FALSE,
|
19
|
+
transform_options: @default_transform_options,
|
14
20
|
diff_opts: {
|
15
21
|
include_diff_info: FALSE
|
16
22
|
}
|
17
23
|
}.freeze
|
18
24
|
|
19
25
|
def self.prepare_input(to_prepare, options)
|
20
|
-
|
21
|
-
prepared = File.read(prepared) if options[:sources_are_files]
|
22
|
-
prepared = Transformer.transform(prepared, include_metadata: options[:include_metadata]) if options[:preprocess_input]
|
26
|
+
return '' unless to_prepare
|
23
27
|
|
24
|
-
|
28
|
+
prep = to_prepare
|
29
|
+
prep = File.read(prep) if options[:sources_are_files]
|
30
|
+
prep = transform(prep, raise_errors: TRUE, options: options[:transform_options]) if options[:preprocess_input]
|
31
|
+
prep
|
25
32
|
end
|
26
33
|
|
27
34
|
def self.diff(
|
28
35
|
from_notebook,
|
29
36
|
to_notebook,
|
30
|
-
options =
|
37
|
+
options = @default_diff_options
|
31
38
|
)
|
32
|
-
options = @
|
39
|
+
options = @default_diff_options.merge(options)
|
33
40
|
|
34
|
-
from =
|
35
|
-
to =
|
41
|
+
from = prepare_input(from_notebook, options)
|
42
|
+
to = prepare_input(to_notebook, options)
|
36
43
|
|
37
44
|
d = Diffy::Diff.new(from, to, **options[:diff_opts]).to_s(options[:format])
|
38
|
-
|
39
45
|
File.write(options[:write_output_to], d) if options[:write_output_to]
|
40
|
-
|
41
46
|
d
|
47
|
+
rescue InvalidNotebookError
|
48
|
+
raise if options[:raise_if_invalid_notebook]
|
49
|
+
end
|
50
|
+
|
51
|
+
def self.transform(notebook, raise_errors: FALSE, options: @default_transform_options)
|
52
|
+
options = @default_transform_options.merge(options)
|
53
|
+
|
54
|
+
Transformer.new(**options).transform(notebook)
|
55
|
+
rescue InvalidNotebookError
|
56
|
+
raise if raise_errors
|
42
57
|
end
|
43
58
|
end
|
@@ -0,0 +1,61 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module IpynbDiff
|
4
|
+
|
5
|
+
# Transforms Jupyter output data into markdown
|
6
|
+
class OutputTransformer
|
7
|
+
|
8
|
+
ORDERED_KEYS = {
|
9
|
+
'execute_result' => %w[image/png image/svg+xml image/jpeg text/markdown text/latex text/plain],
|
10
|
+
'display_data' => %w[image/png image/svg+xml image/jpeg text/markdown text/latex]
|
11
|
+
}.freeze
|
12
|
+
|
13
|
+
def transform(output)
|
14
|
+
case (output_type = output['output_type'])
|
15
|
+
when 'error'
|
16
|
+
transform_error(output['traceback'])
|
17
|
+
when 'execute_result', 'display_data'
|
18
|
+
transform_non_error(ORDERED_KEYS[output_type], output['data'])
|
19
|
+
end
|
20
|
+
end
|
21
|
+
|
22
|
+
def transform_error(traceback)
|
23
|
+
traceback.map do |t|
|
24
|
+
t.split("\n").map do |line|
|
25
|
+
line.gsub(/\[[0-9][0-9;]*m/, '').sub("\u001B", ' ').gsub(/\u001B/, '').rstrip << "\n"
|
26
|
+
end
|
27
|
+
end
|
28
|
+
end
|
29
|
+
|
30
|
+
def transform_non_error(accepted_keys, elements)
|
31
|
+
accepted_keys.map do |key|
|
32
|
+
transform_element(key, elements[key]) if elements.key?(key)
|
33
|
+
end.flatten
|
34
|
+
end
|
35
|
+
|
36
|
+
def transform_element(output_type, output_element)
|
37
|
+
case output_type
|
38
|
+
when 'image/png', 'image/jpeg'
|
39
|
+
transform_image(output_type, output_element)
|
40
|
+
when 'image/svg+xml'
|
41
|
+
transform_svg(output_element)
|
42
|
+
when 'text/markdown', 'text/latex', 'text/plain'
|
43
|
+
transform_text(output_element)
|
44
|
+
end
|
45
|
+
end
|
46
|
+
|
47
|
+
def transform_image(image_type, image_content)
|
48
|
+
[" ![](data:#{image_type};base64,#{image_content.gsub("\n", '')})", "\n"]
|
49
|
+
end
|
50
|
+
|
51
|
+
def transform_svg(image_content)
|
52
|
+
single_line = image_content.map(&:strip).join('').gsub(/\s+/, ' ')
|
53
|
+
|
54
|
+
[" ![](data:image/svg+xml;utf8,#{single_line})", "\n"]
|
55
|
+
end
|
56
|
+
|
57
|
+
def transform_text(text_content)
|
58
|
+
text_content.map { |line| " #{line}" }.append("\n")
|
59
|
+
end
|
60
|
+
end
|
61
|
+
end
|
data/lib/transformer.rb
CHANGED
@@ -1,93 +1,92 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
2
|
|
3
3
|
module IpynbDiff
|
4
|
+
class InvalidNotebookError < StandardError
|
5
|
+
end
|
6
|
+
|
4
7
|
# Returns a markdown version of the Jupyter Notebook
|
5
8
|
class Transformer
|
6
9
|
require 'json'
|
7
10
|
require 'yaml'
|
11
|
+
require 'output_transformer'
|
8
12
|
|
9
|
-
|
10
|
-
|
11
|
-
|
12
|
-
|
13
|
-
|
13
|
+
@cell_decorator = :html
|
14
|
+
@include_metadata = TRUE
|
15
|
+
|
16
|
+
|
17
|
+
def initialize(include_metadata: TRUE, cell_decorator: :html)
|
18
|
+
@include_metadata = include_metadata
|
19
|
+
@cell_decorator = cell_decorator
|
20
|
+
@output_transformer = OutputTransformer.new
|
14
21
|
end
|
15
22
|
|
16
|
-
def
|
17
|
-
|
23
|
+
def validate_notebook(notebook)
|
24
|
+
notebook_json = JSON.parse(notebook)
|
25
|
+
|
26
|
+
return notebook_json if notebook_json.key?('cells') && notebook_json.key?('metadata')
|
27
|
+
|
28
|
+
raise InvalidNotebookError
|
29
|
+
rescue JSON::ParserError
|
30
|
+
raise InvalidNotebookError
|
18
31
|
end
|
19
32
|
|
20
|
-
def
|
21
|
-
|
33
|
+
def transform(notebook)
|
34
|
+
notebook_json = validate_notebook(notebook)
|
35
|
+
transformed_blocks = notebook_json['cells'].map do |cell|
|
36
|
+
decorate_cell(transform_cell(cell, notebook_json), cell)
|
37
|
+
end
|
22
38
|
|
23
|
-
|
24
|
-
|
25
|
-
%(``` #{notebook['metadata']['kernelspec']['language']}\n),
|
26
|
-
*cell['source'],
|
27
|
-
"\n```\n",
|
28
|
-
*cell['outputs'].map { |output| transform_output(output) },
|
29
|
-
"\n</div>\n"
|
30
|
-
].join('')
|
39
|
+
transformed_blocks.prepend(transform_metadata(notebook_json)) if @include_metadata
|
40
|
+
transformed_blocks.join("\n")
|
31
41
|
end
|
32
42
|
|
33
|
-
def
|
34
|
-
|
35
|
-
|
36
|
-
|
37
|
-
|
38
|
-
|
43
|
+
def decorate_cell(rows, cell)
|
44
|
+
tags = cell['metadata']&.fetch('tags', [])
|
45
|
+
type = cell['cell_type'] || 'raw'
|
46
|
+
|
47
|
+
case @cell_decorator
|
48
|
+
when :html
|
49
|
+
rows.prepend(%(<div class="cell #{type}" data-id="#{cell['id']}" data-tags="#{tags&.join(' ')}">\n\n))
|
50
|
+
.append("\n</div>\n")
|
51
|
+
when :percent
|
52
|
+
rows.prepend(%(%% Cell type:#{type} id:#{cell['id']} tags:#{tags&.join(',')}\n\n))
|
53
|
+
else
|
54
|
+
rows
|
55
|
+
end.join('')
|
39
56
|
end
|
40
57
|
|
41
|
-
def
|
42
|
-
[
|
43
|
-
%(\n<div class="output execute_result">\n\n),
|
44
|
-
*output['data']['text/plain'].map { |line| " #{line}" },
|
45
|
-
"\n\n</div>\n"
|
46
|
-
].join('')
|
58
|
+
def transform_cell(cell, notebook)
|
59
|
+
cell['cell_type'] == 'code' ? transform_code_cell(cell, notebook) : transform_text_cell(cell)
|
47
60
|
end
|
48
61
|
|
49
|
-
def
|
50
|
-
if
|
51
|
-
[
|
52
|
-
|
53
|
-
|
54
|
-
"\n\n</div>\n"
|
55
|
-
].join('')
|
62
|
+
def decorate_output(output_rows, output)
|
63
|
+
if @cell_decorator == :html
|
64
|
+
output_rows.prepend(%(\n<div class="output #{output['output_type']}">\n\n)).append("\n</div>\n")
|
65
|
+
else
|
66
|
+
output_rows.prepend(%(\n%%%% Output: #{output['output_type']}\n\n))
|
56
67
|
end
|
57
68
|
end
|
58
69
|
|
59
|
-
def
|
70
|
+
def transform_code_cell(cell, notebook)
|
60
71
|
[
|
61
|
-
%(
|
62
|
-
|
63
|
-
"\n
|
64
|
-
|
72
|
+
%(``` #{notebook['metadata']['kernelspec']['language']}\n),
|
73
|
+
*cell['source'],
|
74
|
+
"\n```\n",
|
75
|
+
*cell['outputs'].map { |output| transform_output(output) }
|
76
|
+
]
|
65
77
|
end
|
66
78
|
|
67
|
-
def
|
68
|
-
|
69
|
-
when 'execute_result'
|
70
|
-
transform_execute_result(output)
|
71
|
-
when 'display_data'
|
72
|
-
transform_image_result(output)
|
73
|
-
when 'error'
|
74
|
-
transform_error_result(output)
|
75
|
-
end
|
76
|
-
end
|
79
|
+
def transform_output(output)
|
80
|
+
transformed = @output_transformer.transform(output)
|
77
81
|
|
78
|
-
|
79
|
-
|
80
|
-
id = cell['id']
|
81
|
-
cell_type = cell['cell_type']
|
82
|
+
decorate_output(transformed, output).join('') if transformed
|
83
|
+
end
|
82
84
|
|
83
|
-
|
84
|
-
|
85
|
-
*cell['source'],
|
86
|
-
"\n\n</div>\n"
|
87
|
-
].join('')
|
85
|
+
def transform_text_cell(cell)
|
86
|
+
cell['source'].append("\n")
|
88
87
|
end
|
89
88
|
|
90
|
-
def
|
89
|
+
def transform_metadata(notebook_json)
|
91
90
|
{
|
92
91
|
'jupyter' => {
|
93
92
|
'kernelspec' => notebook_json['metadata']['kernelspec'],
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: ipynbdiff
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.3.
|
4
|
+
version: 0.3.6
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Eduardo Bonet
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2021-10-
|
11
|
+
date: 2021-10-21 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: diffy
|
@@ -121,6 +121,7 @@ files:
|
|
121
121
|
- README.md
|
122
122
|
- ipynbdiff.gemspec
|
123
123
|
- lib/ipynbdiff.rb
|
124
|
+
- lib/output_transformer.rb
|
124
125
|
- lib/transformer.rb
|
125
126
|
homepage: https://gitlab.com/gitlab-org/incubation-engineering/mlops/rb-ipynbdiff
|
126
127
|
licenses:
|