html-to-markdown 2.5.6 → 2.6.0
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.
- checksums.yaml +4 -4
- data/.rubocop.yml +29 -0
- data/Gemfile +15 -0
- data/Gemfile.lock +80 -0
- data/README.md +7 -4
- data/Rakefile +24 -0
- data/{extconf.rb → ext/html-to-markdown-rb/extconf.rb} +1 -0
- data/html-to-markdown-rb.gemspec +59 -0
- data/lib/html_to_markdown/version.rb +1 -1
- metadata +13 -83
- data/Cargo.toml +0 -28
- data/src/lib.rs +0 -432
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 71fe440f4811019fdf9c1d88b72b092c97d8223dab27a83f6b23e2675d98317a
|
|
4
|
+
data.tar.gz: 1bbe81a26da2c50e16e5207e48bda25e309afde2f3e22600889b85855624de2c
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 49c83e6230f9df3a96088e7f0e4eeb885da4b9601ac7bad407675a1393d228eb4a9357ee5a036e328dd13d1b28aa3d74a50e216030acd6a359296ca5a16888d3
|
|
7
|
+
data.tar.gz: ee85846cd5a92e271719632edba3c2415e2263803a4bd74948e6bae8acd93176c6a4daec380d5423ce4670472219a5ed06bb61f047b47a53bf4778e85b9ed459
|
data/.rubocop.yml
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
plugins:
|
|
2
|
+
- rubocop-rspec
|
|
3
|
+
|
|
4
|
+
AllCops:
|
|
5
|
+
NewCops: enable
|
|
6
|
+
TargetRubyVersion: 3.2
|
|
7
|
+
Exclude:
|
|
8
|
+
- "tmp/**/*"
|
|
9
|
+
- "vendor/**/*"
|
|
10
|
+
|
|
11
|
+
Style/Documentation:
|
|
12
|
+
Enabled: false
|
|
13
|
+
|
|
14
|
+
Metrics/BlockLength:
|
|
15
|
+
Exclude:
|
|
16
|
+
- "spec/**/*"
|
|
17
|
+
- "*.gemspec"
|
|
18
|
+
|
|
19
|
+
Metrics/MethodLength:
|
|
20
|
+
Max: 15
|
|
21
|
+
|
|
22
|
+
RSpec/MultipleExpectations:
|
|
23
|
+
Enabled: false
|
|
24
|
+
|
|
25
|
+
RSpec/ExampleLength:
|
|
26
|
+
Enabled: false
|
|
27
|
+
|
|
28
|
+
RSpec/SpecFilePathFormat:
|
|
29
|
+
Enabled: false
|
data/Gemfile
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
source 'https://rubygems.org'
|
|
4
|
+
|
|
5
|
+
ruby '>= 3.2'
|
|
6
|
+
|
|
7
|
+
gemspec
|
|
8
|
+
|
|
9
|
+
group :development, :test do
|
|
10
|
+
gem 'rake-compiler'
|
|
11
|
+
gem 'rb_sys' # provides build tooling when developing locally
|
|
12
|
+
gem 'rspec'
|
|
13
|
+
gem 'rubocop', require: false
|
|
14
|
+
gem 'rubocop-rspec', require: false
|
|
15
|
+
end
|
data/Gemfile.lock
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
1
|
+
PATH
|
|
2
|
+
remote: .
|
|
3
|
+
specs:
|
|
4
|
+
html-to-markdown (2.6.0)
|
|
5
|
+
rb_sys (>= 0.9, < 1.0)
|
|
6
|
+
|
|
7
|
+
GEM
|
|
8
|
+
remote: https://rubygems.org/
|
|
9
|
+
specs:
|
|
10
|
+
ast (2.4.3)
|
|
11
|
+
diff-lcs (1.6.2)
|
|
12
|
+
json (2.15.2)
|
|
13
|
+
language_server-protocol (3.17.0.5)
|
|
14
|
+
lint_roller (1.1.0)
|
|
15
|
+
parallel (1.27.0)
|
|
16
|
+
parser (3.3.10.0)
|
|
17
|
+
ast (~> 2.4.1)
|
|
18
|
+
racc
|
|
19
|
+
prism (1.6.0)
|
|
20
|
+
racc (1.8.1)
|
|
21
|
+
rainbow (3.1.1)
|
|
22
|
+
rake (13.3.1)
|
|
23
|
+
rake-compiler (1.3.0)
|
|
24
|
+
rake
|
|
25
|
+
rake-compiler-dock (1.9.1)
|
|
26
|
+
rb_sys (0.9.117)
|
|
27
|
+
rake-compiler-dock (= 1.9.1)
|
|
28
|
+
regexp_parser (2.11.3)
|
|
29
|
+
rspec (3.13.2)
|
|
30
|
+
rspec-core (~> 3.13.0)
|
|
31
|
+
rspec-expectations (~> 3.13.0)
|
|
32
|
+
rspec-mocks (~> 3.13.0)
|
|
33
|
+
rspec-core (3.13.6)
|
|
34
|
+
rspec-support (~> 3.13.0)
|
|
35
|
+
rspec-expectations (3.13.5)
|
|
36
|
+
diff-lcs (>= 1.2.0, < 2.0)
|
|
37
|
+
rspec-support (~> 3.13.0)
|
|
38
|
+
rspec-mocks (3.13.7)
|
|
39
|
+
diff-lcs (>= 1.2.0, < 2.0)
|
|
40
|
+
rspec-support (~> 3.13.0)
|
|
41
|
+
rspec-support (3.13.6)
|
|
42
|
+
rubocop (1.81.7)
|
|
43
|
+
json (~> 2.3)
|
|
44
|
+
language_server-protocol (~> 3.17.0.2)
|
|
45
|
+
lint_roller (~> 1.1.0)
|
|
46
|
+
parallel (~> 1.10)
|
|
47
|
+
parser (>= 3.3.0.2)
|
|
48
|
+
rainbow (>= 2.2.2, < 4.0)
|
|
49
|
+
regexp_parser (>= 2.9.3, < 3.0)
|
|
50
|
+
rubocop-ast (>= 1.47.1, < 2.0)
|
|
51
|
+
ruby-progressbar (~> 1.7)
|
|
52
|
+
unicode-display_width (>= 2.4.0, < 4.0)
|
|
53
|
+
rubocop-ast (1.47.1)
|
|
54
|
+
parser (>= 3.3.7.2)
|
|
55
|
+
prism (~> 1.4)
|
|
56
|
+
rubocop-rspec (3.7.0)
|
|
57
|
+
lint_roller (~> 1.1)
|
|
58
|
+
rubocop (~> 1.72, >= 1.72.1)
|
|
59
|
+
ruby-progressbar (1.13.0)
|
|
60
|
+
unicode-display_width (3.2.0)
|
|
61
|
+
unicode-emoji (~> 4.1)
|
|
62
|
+
unicode-emoji (4.1.0)
|
|
63
|
+
|
|
64
|
+
PLATFORMS
|
|
65
|
+
arm64-darwin-24
|
|
66
|
+
ruby
|
|
67
|
+
|
|
68
|
+
DEPENDENCIES
|
|
69
|
+
html-to-markdown!
|
|
70
|
+
rake-compiler
|
|
71
|
+
rb_sys
|
|
72
|
+
rspec
|
|
73
|
+
rubocop
|
|
74
|
+
rubocop-rspec
|
|
75
|
+
|
|
76
|
+
RUBY VERSION
|
|
77
|
+
ruby 3.2.9p248
|
|
78
|
+
|
|
79
|
+
BUNDLED WITH
|
|
80
|
+
2.5.12
|
data/README.md
CHANGED
|
@@ -3,9 +3,12 @@
|
|
|
3
3
|
Blazing-fast HTML → Markdown conversion for Ruby, powered by the same Rust engine used by our Python, Node.js, and WebAssembly packages. Ship identical Markdown across every runtime while enjoying native extension performance.
|
|
4
4
|
|
|
5
5
|
[](https://crates.io/crates/html-to-markdown-rs)
|
|
6
|
-
[](https://www.npmjs.com/package/html-to-markdown-node)
|
|
7
|
+
[](https://www.npmjs.com/package/html-to-markdown-wasm)
|
|
8
|
+
[](https://www.npmjs.com/package/html-to-markdown)
|
|
9
|
+
[](https://pypi.org/project/html-to-markdown/)
|
|
10
|
+
[](https://packagist.org/packages/goldziher/html-to-markdown)
|
|
11
|
+
[](https://rubygems.org/gems/html-to-markdown)
|
|
9
12
|
[](https://github.com/Goldziher/html-to-markdown/blob/main/LICENSE)
|
|
10
13
|
|
|
11
14
|
## Features
|
|
@@ -163,7 +166,7 @@ You can also call the CLI binary directly for scripting:
|
|
|
163
166
|
|
|
164
167
|
```ruby
|
|
165
168
|
HtmlToMarkdown::CLIProxy.call(['--version'])
|
|
166
|
-
# => "html-to-markdown 2.5.
|
|
169
|
+
# => "html-to-markdown 2.5.7"
|
|
167
170
|
```
|
|
168
171
|
|
|
169
172
|
Rebuild the CLI locally if you see `CLI binary not built` during tests:
|
data/Rakefile
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
require 'bundler/gem_tasks'
|
|
4
|
+
require 'rb_sys/extensiontask'
|
|
5
|
+
require 'rspec/core/rake_task'
|
|
6
|
+
|
|
7
|
+
GEMSPEC = Gem::Specification.load(File.expand_path('html-to-markdown-rb.gemspec', __dir__))
|
|
8
|
+
|
|
9
|
+
RbSys::ExtensionTask.new('html-to-markdown-rb', GEMSPEC) do |ext|
|
|
10
|
+
ext.lib_dir = 'lib'
|
|
11
|
+
ext.ext_dir = 'ext/html-to-markdown-rb'
|
|
12
|
+
ext.cross_compile = true
|
|
13
|
+
ext.cross_platform = %w[
|
|
14
|
+
x86_64-linux
|
|
15
|
+
x86_64-darwin
|
|
16
|
+
arm64-darwin
|
|
17
|
+
x64-mingw32
|
|
18
|
+
]
|
|
19
|
+
end
|
|
20
|
+
|
|
21
|
+
RSpec::Core::RakeTask.new(:spec)
|
|
22
|
+
|
|
23
|
+
task spec: :compile
|
|
24
|
+
task default: :spec
|
|
@@ -0,0 +1,59 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
require_relative 'lib/html_to_markdown/version'
|
|
4
|
+
|
|
5
|
+
repo_root = File.expand_path('../..', __dir__)
|
|
6
|
+
crate_prefix = 'packages/ruby/'
|
|
7
|
+
git_cmd = %(git -C "#{repo_root}" ls-files -z #{crate_prefix})
|
|
8
|
+
git_files =
|
|
9
|
+
`#{git_cmd}`.split("\x0")
|
|
10
|
+
.select { |path| path.start_with?(crate_prefix) }
|
|
11
|
+
.map { |path| path.delete_prefix(crate_prefix) }
|
|
12
|
+
fallback_files = Dir.chdir(__dir__) do
|
|
13
|
+
Dir.glob(
|
|
14
|
+
%w[
|
|
15
|
+
README.md
|
|
16
|
+
ext/html-to-markdown-rb/extconf.rb
|
|
17
|
+
exe/*
|
|
18
|
+
lib/**/*.rb
|
|
19
|
+
lib/bin/*
|
|
20
|
+
src/**/*.rs
|
|
21
|
+
spec/**/*.rb
|
|
22
|
+
]
|
|
23
|
+
)
|
|
24
|
+
end
|
|
25
|
+
files = git_files.empty? ? fallback_files : git_files
|
|
26
|
+
|
|
27
|
+
Gem::Specification.new do |spec|
|
|
28
|
+
spec.name = 'html-to-markdown'
|
|
29
|
+
spec.version = HtmlToMarkdown::VERSION
|
|
30
|
+
spec.authors = ["Na'aman Hirschfeld"]
|
|
31
|
+
spec.email = ['nhirschfeld@gmail.com']
|
|
32
|
+
|
|
33
|
+
spec.summary = 'Blazing-fast HTML to Markdown conversion for Ruby, powered by Rust.'
|
|
34
|
+
spec.description = <<~DESC.strip
|
|
35
|
+
html-to-markdown is a native Ruby extension built on the shared Rust engine that powers the html-to-markdown project.
|
|
36
|
+
It delivers identical HTML-to-Markdown output across languages, exposes inline image extraction, and ships with a CLI for automation workflows.
|
|
37
|
+
DESC
|
|
38
|
+
spec.homepage = 'https://github.com/Goldziher/html-to-markdown'
|
|
39
|
+
spec.license = 'MIT'
|
|
40
|
+
|
|
41
|
+
spec.required_ruby_version = Gem::Requirement.new('>= 3.2')
|
|
42
|
+
|
|
43
|
+
spec.bindir = 'exe'
|
|
44
|
+
spec.executables = ['html-to-markdown']
|
|
45
|
+
spec.require_paths = ['lib']
|
|
46
|
+
|
|
47
|
+
spec.files = files
|
|
48
|
+
spec.extra_rdoc_files = ['README.md']
|
|
49
|
+
|
|
50
|
+
spec.extensions = ['ext/html-to-markdown-rb/extconf.rb']
|
|
51
|
+
|
|
52
|
+
spec.add_dependency 'rb_sys', '>= 0.9', '< 1.0'
|
|
53
|
+
spec.metadata['rubygems_mfa_required'] = 'true'
|
|
54
|
+
spec.metadata['homepage_uri'] = 'https://github.com/Goldziher/html-to-markdown'
|
|
55
|
+
spec.metadata['source_code_uri'] = 'https://github.com/Goldziher/html-to-markdown'
|
|
56
|
+
spec.metadata['bug_tracker_uri'] = 'https://github.com/Goldziher/html-to-markdown/issues'
|
|
57
|
+
spec.metadata['changelog_uri'] = 'https://github.com/Goldziher/html-to-markdown/releases'
|
|
58
|
+
spec.metadata['documentation_uri'] = 'https://github.com/Goldziher/html-to-markdown/blob/main/packages/ruby/README.md'
|
|
59
|
+
end
|
metadata
CHANGED
|
@@ -1,14 +1,14 @@
|
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
|
2
2
|
name: html-to-markdown
|
|
3
3
|
version: !ruby/object:Gem::Version
|
|
4
|
-
version: 2.
|
|
4
|
+
version: 2.6.0
|
|
5
5
|
platform: ruby
|
|
6
6
|
authors:
|
|
7
7
|
- Na'aman Hirschfeld
|
|
8
8
|
autorequire:
|
|
9
9
|
bindir: exe
|
|
10
10
|
cert_chain: []
|
|
11
|
-
date: 2025-
|
|
11
|
+
date: 2025-11-07 00:00:00.000000000 Z
|
|
12
12
|
dependencies:
|
|
13
13
|
- !ruby/object:Gem::Dependency
|
|
14
14
|
name: rb_sys
|
|
@@ -30,95 +30,26 @@ dependencies:
|
|
|
30
30
|
- - "<"
|
|
31
31
|
- !ruby/object:Gem::Version
|
|
32
32
|
version: '1.0'
|
|
33
|
-
description:
|
|
34
|
-
Ruby
|
|
35
|
-
|
|
36
|
-
performance.\n\n[](https://crates.io/crates/html-to-markdown-rs)\n[](https://www.npmjs.com/package/html-to-markdown-node)\n[](https://pypi.org/project/html-to-markdown/)\n[](https://rubygems.org/gems/html-to-markdown)\n[](https://github.com/Goldziher/html-to-markdown/blob/main/LICENSE)\n\n##
|
|
41
|
-
Features\n\n- ⚡ **Rust-fast**: Ruby bindings around a highly optimised Rust core
|
|
42
|
-
(60‑80× faster than BeautifulSoup-based converters).\n- \U0001F501 **Identical output**:
|
|
43
|
-
Shares logic with the Python wheels, npm bindings, WASM package, and CLI — consistent
|
|
44
|
-
Markdown everywhere.\n- ⚙️ **Rich configuration**: Control heading styles, list
|
|
45
|
-
indentation, whitespace handling, HTML preprocessing, and more.\n- \U0001F5BC️ **Inline
|
|
46
|
-
image extraction**: Pull out embedded images (PNG/JPEG/SVG/data URIs) alongside
|
|
47
|
-
Markdown.\n- \U0001F9F0 **Bundled CLI proxy**: Call the Rust CLI straight from Ruby
|
|
48
|
-
or shell scripts.\n- \U0001F6E0️ **First-class Rails support**: Works with `Gem.win_platform?`
|
|
49
|
-
builds, supports Trusted Publishing, and compiles on install if no native gem matches.\n\n##
|
|
50
|
-
Documentation & Support\n\n- [GitHub repository](https://github.com/Goldziher/html-to-markdown)\n-
|
|
51
|
-
[Issue tracker](https://github.com/Goldziher/html-to-markdown/issues)\n- [Changelog](https://github.com/Goldziher/html-to-markdown/blob/main/CHANGELOG.md)\n-
|
|
52
|
-
[Live demo (WASM)](https://goldziher.github.io/html-to-markdown/)\n\n## Installation\n\n```bash\nbundle
|
|
53
|
-
add html-to-markdown\n# or\ngem install html-to-markdown\n```\n\nAdd the gem to
|
|
54
|
-
your project and Bundler will compile the native Rust extension on first install.\n\n###
|
|
55
|
-
Requirements\n\n- Ruby **3.2+** (Magnus relies on the fiber scheduler APIs added
|
|
56
|
-
in 3.2)\n- Rust toolchain **1.85+** with Cargo available on your `$PATH`\n- Ruby
|
|
57
|
-
development headers (`ruby-dev`, `ruby-devel`, or the platform equivalent)\n\n**Windows**:
|
|
58
|
-
install [RubyInstaller with MSYS2](https://rubyinstaller.org/) (UCRT64). Run once:\n\n```powershell\nridk
|
|
59
|
-
exec pacman -S --needed --noconfirm base-devel mingw-w64-ucrt-x86_64-toolchain\n```\n\nThis
|
|
60
|
-
provides the standard headers (including `strings.h`) required for the bindgen step.\n\n##
|
|
61
|
-
Performance Snapshot\n\nApple M4 • Real Wikipedia documents • `HtmlToMarkdown.convert`
|
|
62
|
-
(Ruby)\n\n| Document | Size | Latency | Throughput | Docs/sec |\n| -------------------
|
|
63
|
-
| ----- | ------- | ---------- | -------- |\n| Lists (Timeline) | 129KB | 0.69ms
|
|
64
|
-
\ | 187 MB/s | 1,450 |\n| Tables (Countries) | 360KB | 2.19ms | 164 MB/s
|
|
65
|
-
\ | 456 |\n| Mixed (Python wiki) | 656KB | 4.88ms | 134 MB/s | 205 |\n\n>
|
|
66
|
-
Same core, same benchmarks: the Ruby extension stays within single-digit % of the
|
|
67
|
-
Rust CLI and mirrors the Python/Node numbers.\n\n## Quick Start\n\n```ruby\nrequire
|
|
68
|
-
'html_to_markdown'\n\nhtml = <<~HTML\n <h1>Welcome</h1>\n <p>This is <strong>Rust-fast</strong>
|
|
69
|
-
conversion!</p>\n <ul>\n <li>Native extension</li>\n <li>Identical output
|
|
70
|
-
across languages</li>\n </ul>\nHTML\n\nmarkdown = HtmlToMarkdown.convert(html)\nputs
|
|
71
|
-
markdown\n# # Welcome\n#\n# This is **Rust-fast** conversion!\n#\n# - Native extension\n#
|
|
72
|
-
- Identical output across languages\n```\n\n## API\n\n### Conversion Options\n\nPass
|
|
73
|
-
a Ruby hash (string or symbol keys) to tweak rendering. Every option maps one-for-one
|
|
74
|
-
with the Rust/Python/Node APIs.\n\n```ruby\nrequire 'html_to_markdown'\n\nmarkdown
|
|
75
|
-
= HtmlToMarkdown.convert(\n '<pre><code class=\"language-ruby\">puts \"hi\"</code></pre>',\n
|
|
76
|
-
\ heading_style: :atx,\n code_block_style: :fenced,\n bullets: '*+-',\n list_indent_type:
|
|
77
|
-
:spaces,\n list_indent_width: 2,\n whitespace_mode: :normalized,\n highlight_style:
|
|
78
|
-
:double_equal\n)\n\nputs markdown\n```\n\n### HTML Preprocessing\n\nClean up scraped
|
|
79
|
-
HTML (navigation, forms, malformed markup) before conversion:\n\n```ruby\nrequire
|
|
80
|
-
'html_to_markdown'\n\nmarkdown = HtmlToMarkdown.convert(\n html,\n preprocessing:
|
|
81
|
-
{\n enabled: true,\n preset: :aggressive, # :minimal, :standard, :aggressive\n
|
|
82
|
-
\ remove_navigation: true,\n remove_forms: true\n }\n)\n```\n\n### Inline
|
|
83
|
-
Images\n\nExtract inline binary data (data URIs, SVG) together with the converted
|
|
84
|
-
Markdown.\n\n```ruby\nrequire 'html_to_markdown'\n\nresult = HtmlToMarkdown.convert_with_inline_images(\n
|
|
85
|
-
\ '<img src=\"data:image/png;base64,iVBORw0...\" alt=\"Pixel\">',\n image_config:
|
|
86
|
-
{\n max_decoded_size_bytes: 1 * 1024 * 1024,\n infer_dimensions: true,\n filename_prefix:
|
|
87
|
-
'img_',\n capture_svg: true\n }\n)\n\nputs result.markdown\nresult.inline_images.each
|
|
88
|
-
do |img|\n puts \"#{img.filename} -> #{img.format} (#{img.data.bytesize} bytes)\"\nend\n```\n\n##
|
|
89
|
-
CLI\n\nThe gem bundles a small proxy for the Rust CLI binary. Use it when you need
|
|
90
|
-
parity with the standalone `html-to-markdown` executable.\n\n```ruby\nrequire 'html_to_markdown/cli'\n\nHtmlToMarkdown::CLI.run(%w[--heading-style
|
|
91
|
-
atx input.html], stdout: $stdout)\n# => writes converted Markdown to STDOUT\n```\n\nYou
|
|
92
|
-
can also call the CLI binary directly for scripting:\n\n```ruby\nHtmlToMarkdown::CLIProxy.call(['--version'])\n#
|
|
93
|
-
=> \"html-to-markdown 2.5.6\"\n```\n\nRebuild the CLI locally if you see `CLI binary
|
|
94
|
-
not built` during tests:\n\n```bash\nbundle exec rake compile # builds
|
|
95
|
-
the extension\nbundle exec ruby scripts/prepare_ruby_gem.rb # copies the CLI into
|
|
96
|
-
lib/bin/\n```\n\n## Error Handling\n\nConversion errors raise `HtmlToMarkdown::Error`
|
|
97
|
-
(wrapping the Rust error context). CLI invocations use specialised subclasses:\n\n-
|
|
98
|
-
`HtmlToMarkdown::CLIProxy::MissingBinaryError`\n- `HtmlToMarkdown::CLIProxy::CLIExecutionError`\n\nRescue
|
|
99
|
-
them to provide clearer feedback in your application.\n\n## Consistent Across Languages\n\nThe
|
|
100
|
-
Ruby gem shares the exact Rust core with:\n\n- [Python wheels](https://pypi.org/project/html-to-markdown/)\n-
|
|
101
|
-
[Node.js / Bun bindings](https://www.npmjs.com/package/html-to-markdown-node)\n-
|
|
102
|
-
[WebAssembly package](https://www.npmjs.com/package/html-to-markdown-wasm)\n- The
|
|
103
|
-
Rust crate and CLI\n\nUse whichever runtime fits your stack while keeping formatting
|
|
104
|
-
behaviour identical.\n\n## Development\n\n```bash\nbundle exec rake compile #
|
|
105
|
-
build the native extension\nbundle exec rspec # run test suite\n```\n\nThe
|
|
106
|
-
extension uses [Magnus](https://github.com/matsadler/magnus) plus `rb-sys` for bindgen.
|
|
107
|
-
When editing the Rust code under `src/`, rerun `rake compile`.\n\n## License\n\nMIT
|
|
108
|
-
© Na'aman Hirschfeld\n"
|
|
33
|
+
description: |-
|
|
34
|
+
html-to-markdown is a native Ruby extension built on the shared Rust engine that powers the html-to-markdown project.
|
|
35
|
+
It delivers identical HTML-to-Markdown output across languages, exposes inline image extraction, and ships with a CLI for automation workflows.
|
|
109
36
|
email:
|
|
110
37
|
- nhirschfeld@gmail.com
|
|
111
38
|
executables:
|
|
112
39
|
- html-to-markdown
|
|
113
40
|
extensions:
|
|
114
|
-
- extconf.rb
|
|
41
|
+
- ext/html-to-markdown-rb/extconf.rb
|
|
115
42
|
extra_rdoc_files:
|
|
116
43
|
- README.md
|
|
117
44
|
files:
|
|
118
|
-
-
|
|
45
|
+
- ".rubocop.yml"
|
|
46
|
+
- Gemfile
|
|
47
|
+
- Gemfile.lock
|
|
119
48
|
- README.md
|
|
49
|
+
- Rakefile
|
|
120
50
|
- exe/html-to-markdown
|
|
121
|
-
- extconf.rb
|
|
51
|
+
- ext/html-to-markdown-rb/extconf.rb
|
|
52
|
+
- html-to-markdown-rb.gemspec
|
|
122
53
|
- lib/html_to_markdown.rb
|
|
123
54
|
- lib/html_to_markdown/cli.rb
|
|
124
55
|
- lib/html_to_markdown/cli_proxy.rb
|
|
@@ -126,7 +57,6 @@ files:
|
|
|
126
57
|
- spec/cli_proxy_spec.rb
|
|
127
58
|
- spec/convert_spec.rb
|
|
128
59
|
- spec/spec_helper.rb
|
|
129
|
-
- src/lib.rs
|
|
130
60
|
homepage: https://github.com/Goldziher/html-to-markdown
|
|
131
61
|
licenses:
|
|
132
62
|
- MIT
|
|
@@ -136,7 +66,7 @@ metadata:
|
|
|
136
66
|
source_code_uri: https://github.com/Goldziher/html-to-markdown
|
|
137
67
|
bug_tracker_uri: https://github.com/Goldziher/html-to-markdown/issues
|
|
138
68
|
changelog_uri: https://github.com/Goldziher/html-to-markdown/releases
|
|
139
|
-
documentation_uri: https://github.com/Goldziher/html-to-markdown/blob/main/
|
|
69
|
+
documentation_uri: https://github.com/Goldziher/html-to-markdown/blob/main/packages/ruby/README.md
|
|
140
70
|
post_install_message:
|
|
141
71
|
rdoc_options: []
|
|
142
72
|
require_paths:
|
data/Cargo.toml
DELETED
|
@@ -1,28 +0,0 @@
|
|
|
1
|
-
[package]
|
|
2
|
-
name = "html-to-markdown-rb"
|
|
3
|
-
version = "2.5.6"
|
|
4
|
-
edition = "2021"
|
|
5
|
-
authors = ["Na'aman Hirschfeld <nhirschfeld@gmail.com>"]
|
|
6
|
-
license = "MIT"
|
|
7
|
-
repository = "https://github.com/Goldziher/html-to-markdown"
|
|
8
|
-
homepage = "https://github.com/Goldziher/html-to-markdown"
|
|
9
|
-
documentation = "https://docs.rs/html-to-markdown-rs"
|
|
10
|
-
readme = "README.md"
|
|
11
|
-
rust-version = "1.80"
|
|
12
|
-
description = "Ruby bindings (Magnus) for html-to-markdown - high-performance HTML to Markdown converter"
|
|
13
|
-
keywords = ["html", "markdown", "ruby", "magnus", "bindings"]
|
|
14
|
-
categories = ["api-bindings"]
|
|
15
|
-
|
|
16
|
-
[lib]
|
|
17
|
-
name = "html_to_markdown_rb"
|
|
18
|
-
crate-type = ["cdylib", "rlib"]
|
|
19
|
-
|
|
20
|
-
[features]
|
|
21
|
-
default = []
|
|
22
|
-
|
|
23
|
-
[dependencies]
|
|
24
|
-
html-to-markdown-rs = { version = "2.5.6", features = ["inline-images"] }
|
|
25
|
-
magnus = { git = "https://github.com/matsadler/magnus", rev = "f6db11769efb517427bf7f121f9c32e18b059b38", features = ["rb-sys"] }
|
|
26
|
-
|
|
27
|
-
[dev-dependencies]
|
|
28
|
-
pretty_assertions = "1.4"
|
data/src/lib.rs
DELETED
|
@@ -1,432 +0,0 @@
|
|
|
1
|
-
use html_to_markdown_rs::{
|
|
2
|
-
convert as convert_inner, convert_with_inline_images as convert_with_inline_images_inner, error::ConversionError,
|
|
3
|
-
CodeBlockStyle, ConversionOptions, HeadingStyle, HighlightStyle, HtmlExtraction, InlineImage, InlineImageConfig,
|
|
4
|
-
InlineImageFormat, InlineImageSource, InlineImageWarning, ListIndentType, NewlineStyle, PreprocessingOptions,
|
|
5
|
-
PreprocessingPreset, WhitespaceMode,
|
|
6
|
-
};
|
|
7
|
-
use magnus::prelude::*;
|
|
8
|
-
use magnus::{function, scan_args::scan_args, Error, RArray, RHash, Ruby, Symbol, TryConvert, Value};
|
|
9
|
-
|
|
10
|
-
const DEFAULT_INLINE_IMAGE_LIMIT: u64 = 5 * 1024 * 1024;
|
|
11
|
-
|
|
12
|
-
fn conversion_error(err: ConversionError) -> Error {
|
|
13
|
-
match err {
|
|
14
|
-
ConversionError::ConfigError(msg) => arg_error(msg),
|
|
15
|
-
other => runtime_error(other.to_string()),
|
|
16
|
-
}
|
|
17
|
-
}
|
|
18
|
-
|
|
19
|
-
fn arg_error(message: impl Into<String>) -> Error {
|
|
20
|
-
let ruby = Ruby::get().expect("Ruby not initialised");
|
|
21
|
-
Error::new(ruby.exception_arg_error(), message.into())
|
|
22
|
-
}
|
|
23
|
-
|
|
24
|
-
fn runtime_error(message: impl Into<String>) -> Error {
|
|
25
|
-
let ruby = Ruby::get().expect("Ruby not initialised");
|
|
26
|
-
Error::new(ruby.exception_runtime_error(), message.into())
|
|
27
|
-
}
|
|
28
|
-
|
|
29
|
-
fn symbol_to_string(value: Value) -> Result<String, Error> {
|
|
30
|
-
if let Some(symbol) = Symbol::from_value(value) {
|
|
31
|
-
Ok(symbol.name()?.to_string())
|
|
32
|
-
} else {
|
|
33
|
-
String::try_convert(value)
|
|
34
|
-
}
|
|
35
|
-
}
|
|
36
|
-
|
|
37
|
-
fn get_kw(ruby: &Ruby, hash: RHash, name: &str) -> Option<Value> {
|
|
38
|
-
let sym = ruby.intern(name);
|
|
39
|
-
hash.get(sym).or_else(|| hash.get(name))
|
|
40
|
-
}
|
|
41
|
-
|
|
42
|
-
fn parse_heading_style(value: Value) -> Result<HeadingStyle, Error> {
|
|
43
|
-
match symbol_to_string(value)?.as_str() {
|
|
44
|
-
"underlined" => Ok(HeadingStyle::Underlined),
|
|
45
|
-
"atx" => Ok(HeadingStyle::Atx),
|
|
46
|
-
"atx_closed" => Ok(HeadingStyle::AtxClosed),
|
|
47
|
-
other => Err(arg_error(format!("invalid heading_style: {other}"))),
|
|
48
|
-
}
|
|
49
|
-
}
|
|
50
|
-
|
|
51
|
-
fn parse_list_indent_type(value: Value) -> Result<ListIndentType, Error> {
|
|
52
|
-
match symbol_to_string(value)?.as_str() {
|
|
53
|
-
"spaces" => Ok(ListIndentType::Spaces),
|
|
54
|
-
"tabs" => Ok(ListIndentType::Tabs),
|
|
55
|
-
other => Err(arg_error(format!("invalid list_indent_type: {other}"))),
|
|
56
|
-
}
|
|
57
|
-
}
|
|
58
|
-
|
|
59
|
-
fn parse_highlight_style(value: Value) -> Result<HighlightStyle, Error> {
|
|
60
|
-
match symbol_to_string(value)?.as_str() {
|
|
61
|
-
"double_equal" => Ok(HighlightStyle::DoubleEqual),
|
|
62
|
-
"html" => Ok(HighlightStyle::Html),
|
|
63
|
-
"bold" => Ok(HighlightStyle::Bold),
|
|
64
|
-
"none" => Ok(HighlightStyle::None),
|
|
65
|
-
other => Err(arg_error(format!("invalid highlight_style: {other}"))),
|
|
66
|
-
}
|
|
67
|
-
}
|
|
68
|
-
|
|
69
|
-
fn parse_whitespace_mode(value: Value) -> Result<WhitespaceMode, Error> {
|
|
70
|
-
match symbol_to_string(value)?.as_str() {
|
|
71
|
-
"normalized" => Ok(WhitespaceMode::Normalized),
|
|
72
|
-
"strict" => Ok(WhitespaceMode::Strict),
|
|
73
|
-
other => Err(arg_error(format!("invalid whitespace_mode: {other}"))),
|
|
74
|
-
}
|
|
75
|
-
}
|
|
76
|
-
|
|
77
|
-
fn parse_newline_style(value: Value) -> Result<NewlineStyle, Error> {
|
|
78
|
-
match symbol_to_string(value)?.as_str() {
|
|
79
|
-
"spaces" => Ok(NewlineStyle::Spaces),
|
|
80
|
-
"backslash" => Ok(NewlineStyle::Backslash),
|
|
81
|
-
other => Err(arg_error(format!("invalid newline_style: {other}"))),
|
|
82
|
-
}
|
|
83
|
-
}
|
|
84
|
-
|
|
85
|
-
fn parse_code_block_style(value: Value) -> Result<CodeBlockStyle, Error> {
|
|
86
|
-
match symbol_to_string(value)?.as_str() {
|
|
87
|
-
"indented" => Ok(CodeBlockStyle::Indented),
|
|
88
|
-
"backticks" => Ok(CodeBlockStyle::Backticks),
|
|
89
|
-
"tildes" => Ok(CodeBlockStyle::Tildes),
|
|
90
|
-
other => Err(arg_error(format!("invalid code_block_style: {other}"))),
|
|
91
|
-
}
|
|
92
|
-
}
|
|
93
|
-
|
|
94
|
-
fn parse_preset(value: Value) -> Result<PreprocessingPreset, Error> {
|
|
95
|
-
match symbol_to_string(value)?.as_str() {
|
|
96
|
-
"minimal" => Ok(PreprocessingPreset::Minimal),
|
|
97
|
-
"standard" => Ok(PreprocessingPreset::Standard),
|
|
98
|
-
"aggressive" => Ok(PreprocessingPreset::Aggressive),
|
|
99
|
-
other => Err(arg_error(format!("invalid preprocessing preset: {other}"))),
|
|
100
|
-
}
|
|
101
|
-
}
|
|
102
|
-
|
|
103
|
-
fn parse_vec_of_strings(value: Value) -> Result<Vec<String>, Error> {
|
|
104
|
-
let array = RArray::from_value(value).ok_or_else(|| arg_error("expected an Array of strings"))?;
|
|
105
|
-
|
|
106
|
-
array.to_vec::<String>()
|
|
107
|
-
}
|
|
108
|
-
|
|
109
|
-
fn parse_preprocessing_options(ruby: &Ruby, value: Value) -> Result<PreprocessingOptions, Error> {
|
|
110
|
-
let hash = RHash::from_value(value).ok_or_else(|| arg_error("expected preprocessing to be a Hash"))?;
|
|
111
|
-
|
|
112
|
-
let mut opts = PreprocessingOptions::default();
|
|
113
|
-
|
|
114
|
-
if let Some(enabled) = get_kw(ruby, hash, "enabled") {
|
|
115
|
-
opts.enabled = bool::try_convert(enabled)?;
|
|
116
|
-
}
|
|
117
|
-
|
|
118
|
-
if let Some(preset) = get_kw(ruby, hash, "preset") {
|
|
119
|
-
opts.preset = parse_preset(preset)?;
|
|
120
|
-
}
|
|
121
|
-
|
|
122
|
-
if let Some(remove_navigation) = get_kw(ruby, hash, "remove_navigation") {
|
|
123
|
-
opts.remove_navigation = bool::try_convert(remove_navigation)?;
|
|
124
|
-
}
|
|
125
|
-
|
|
126
|
-
if let Some(remove_forms) = get_kw(ruby, hash, "remove_forms") {
|
|
127
|
-
opts.remove_forms = bool::try_convert(remove_forms)?;
|
|
128
|
-
}
|
|
129
|
-
|
|
130
|
-
Ok(opts)
|
|
131
|
-
}
|
|
132
|
-
|
|
133
|
-
fn build_conversion_options(ruby: &Ruby, options: Option<Value>) -> Result<ConversionOptions, Error> {
|
|
134
|
-
let mut opts = ConversionOptions::default();
|
|
135
|
-
|
|
136
|
-
let Some(options) = options else {
|
|
137
|
-
return Ok(opts);
|
|
138
|
-
};
|
|
139
|
-
|
|
140
|
-
if options.is_nil() {
|
|
141
|
-
return Ok(opts);
|
|
142
|
-
}
|
|
143
|
-
|
|
144
|
-
let hash = RHash::from_value(options).ok_or_else(|| arg_error("options must be provided as a Hash"))?;
|
|
145
|
-
|
|
146
|
-
if let Some(heading_style) = get_kw(ruby, hash, "heading_style") {
|
|
147
|
-
opts.heading_style = parse_heading_style(heading_style)?;
|
|
148
|
-
}
|
|
149
|
-
|
|
150
|
-
if let Some(list_indent_type) = get_kw(ruby, hash, "list_indent_type") {
|
|
151
|
-
opts.list_indent_type = parse_list_indent_type(list_indent_type)?;
|
|
152
|
-
}
|
|
153
|
-
|
|
154
|
-
if let Some(list_indent_width) = get_kw(ruby, hash, "list_indent_width") {
|
|
155
|
-
opts.list_indent_width = usize::try_convert(list_indent_width)?;
|
|
156
|
-
}
|
|
157
|
-
|
|
158
|
-
if let Some(bullets) = get_kw(ruby, hash, "bullets") {
|
|
159
|
-
opts.bullets = String::try_convert(bullets)?;
|
|
160
|
-
}
|
|
161
|
-
|
|
162
|
-
if let Some(strong_em_symbol) = get_kw(ruby, hash, "strong_em_symbol") {
|
|
163
|
-
let value = String::try_convert(strong_em_symbol)?;
|
|
164
|
-
let mut chars = value.chars();
|
|
165
|
-
let ch = chars
|
|
166
|
-
.next()
|
|
167
|
-
.ok_or_else(|| arg_error("strong_em_symbol must not be empty"))?;
|
|
168
|
-
if chars.next().is_some() {
|
|
169
|
-
return Err(arg_error("strong_em_symbol must be a single character"));
|
|
170
|
-
}
|
|
171
|
-
opts.strong_em_symbol = ch;
|
|
172
|
-
}
|
|
173
|
-
|
|
174
|
-
if let Some(escape_asterisks) = get_kw(ruby, hash, "escape_asterisks") {
|
|
175
|
-
opts.escape_asterisks = bool::try_convert(escape_asterisks)?;
|
|
176
|
-
}
|
|
177
|
-
|
|
178
|
-
if let Some(escape_underscores) = get_kw(ruby, hash, "escape_underscores") {
|
|
179
|
-
opts.escape_underscores = bool::try_convert(escape_underscores)?;
|
|
180
|
-
}
|
|
181
|
-
|
|
182
|
-
if let Some(escape_misc) = get_kw(ruby, hash, "escape_misc") {
|
|
183
|
-
opts.escape_misc = bool::try_convert(escape_misc)?;
|
|
184
|
-
}
|
|
185
|
-
|
|
186
|
-
if let Some(escape_ascii) = get_kw(ruby, hash, "escape_ascii") {
|
|
187
|
-
opts.escape_ascii = bool::try_convert(escape_ascii)?;
|
|
188
|
-
}
|
|
189
|
-
|
|
190
|
-
if let Some(code_language) = get_kw(ruby, hash, "code_language") {
|
|
191
|
-
opts.code_language = String::try_convert(code_language)?;
|
|
192
|
-
}
|
|
193
|
-
|
|
194
|
-
if let Some(autolinks) = get_kw(ruby, hash, "autolinks") {
|
|
195
|
-
opts.autolinks = bool::try_convert(autolinks)?;
|
|
196
|
-
}
|
|
197
|
-
|
|
198
|
-
if let Some(default_title) = get_kw(ruby, hash, "default_title") {
|
|
199
|
-
opts.default_title = bool::try_convert(default_title)?;
|
|
200
|
-
}
|
|
201
|
-
|
|
202
|
-
if let Some(br_in_tables) = get_kw(ruby, hash, "br_in_tables") {
|
|
203
|
-
opts.br_in_tables = bool::try_convert(br_in_tables)?;
|
|
204
|
-
}
|
|
205
|
-
|
|
206
|
-
if let Some(hocr_spatial_tables) = get_kw(ruby, hash, "hocr_spatial_tables") {
|
|
207
|
-
opts.hocr_spatial_tables = bool::try_convert(hocr_spatial_tables)?;
|
|
208
|
-
}
|
|
209
|
-
|
|
210
|
-
if let Some(highlight_style) = get_kw(ruby, hash, "highlight_style") {
|
|
211
|
-
opts.highlight_style = parse_highlight_style(highlight_style)?;
|
|
212
|
-
}
|
|
213
|
-
|
|
214
|
-
if let Some(extract_metadata) = get_kw(ruby, hash, "extract_metadata") {
|
|
215
|
-
opts.extract_metadata = bool::try_convert(extract_metadata)?;
|
|
216
|
-
}
|
|
217
|
-
|
|
218
|
-
if let Some(whitespace_mode) = get_kw(ruby, hash, "whitespace_mode") {
|
|
219
|
-
opts.whitespace_mode = parse_whitespace_mode(whitespace_mode)?;
|
|
220
|
-
}
|
|
221
|
-
|
|
222
|
-
if let Some(strip_newlines) = get_kw(ruby, hash, "strip_newlines") {
|
|
223
|
-
opts.strip_newlines = bool::try_convert(strip_newlines)?;
|
|
224
|
-
}
|
|
225
|
-
|
|
226
|
-
if let Some(wrap) = get_kw(ruby, hash, "wrap") {
|
|
227
|
-
opts.wrap = bool::try_convert(wrap)?;
|
|
228
|
-
}
|
|
229
|
-
|
|
230
|
-
if let Some(wrap_width) = get_kw(ruby, hash, "wrap_width") {
|
|
231
|
-
opts.wrap_width = usize::try_convert(wrap_width)?;
|
|
232
|
-
}
|
|
233
|
-
|
|
234
|
-
if let Some(convert_as_inline) = get_kw(ruby, hash, "convert_as_inline") {
|
|
235
|
-
opts.convert_as_inline = bool::try_convert(convert_as_inline)?;
|
|
236
|
-
}
|
|
237
|
-
|
|
238
|
-
if let Some(sub_symbol) = get_kw(ruby, hash, "sub_symbol") {
|
|
239
|
-
opts.sub_symbol = String::try_convert(sub_symbol)?;
|
|
240
|
-
}
|
|
241
|
-
|
|
242
|
-
if let Some(sup_symbol) = get_kw(ruby, hash, "sup_symbol") {
|
|
243
|
-
opts.sup_symbol = String::try_convert(sup_symbol)?;
|
|
244
|
-
}
|
|
245
|
-
|
|
246
|
-
if let Some(newline_style) = get_kw(ruby, hash, "newline_style") {
|
|
247
|
-
opts.newline_style = parse_newline_style(newline_style)?;
|
|
248
|
-
}
|
|
249
|
-
|
|
250
|
-
if let Some(code_block_style) = get_kw(ruby, hash, "code_block_style") {
|
|
251
|
-
opts.code_block_style = parse_code_block_style(code_block_style)?;
|
|
252
|
-
}
|
|
253
|
-
|
|
254
|
-
if let Some(keep_inline_images_in) = get_kw(ruby, hash, "keep_inline_images_in") {
|
|
255
|
-
opts.keep_inline_images_in = parse_vec_of_strings(keep_inline_images_in)?;
|
|
256
|
-
}
|
|
257
|
-
|
|
258
|
-
if let Some(preprocessing) = get_kw(ruby, hash, "preprocessing") {
|
|
259
|
-
opts.preprocessing = parse_preprocessing_options(ruby, preprocessing)?;
|
|
260
|
-
}
|
|
261
|
-
|
|
262
|
-
if let Some(encoding) = get_kw(ruby, hash, "encoding") {
|
|
263
|
-
opts.encoding = String::try_convert(encoding)?;
|
|
264
|
-
}
|
|
265
|
-
|
|
266
|
-
if let Some(debug) = get_kw(ruby, hash, "debug") {
|
|
267
|
-
opts.debug = bool::try_convert(debug)?;
|
|
268
|
-
}
|
|
269
|
-
|
|
270
|
-
if let Some(strip_tags) = get_kw(ruby, hash, "strip_tags") {
|
|
271
|
-
opts.strip_tags = parse_vec_of_strings(strip_tags)?;
|
|
272
|
-
}
|
|
273
|
-
|
|
274
|
-
if let Some(preserve_tags) = get_kw(ruby, hash, "preserve_tags") {
|
|
275
|
-
opts.preserve_tags = parse_vec_of_strings(preserve_tags)?;
|
|
276
|
-
}
|
|
277
|
-
|
|
278
|
-
Ok(opts)
|
|
279
|
-
}
|
|
280
|
-
|
|
281
|
-
fn build_inline_image_config(ruby: &Ruby, config: Option<Value>) -> Result<InlineImageConfig, Error> {
|
|
282
|
-
let mut cfg = InlineImageConfig::new(DEFAULT_INLINE_IMAGE_LIMIT);
|
|
283
|
-
|
|
284
|
-
let Some(config) = config else {
|
|
285
|
-
return Ok(cfg);
|
|
286
|
-
};
|
|
287
|
-
|
|
288
|
-
if config.is_nil() {
|
|
289
|
-
return Ok(cfg);
|
|
290
|
-
}
|
|
291
|
-
|
|
292
|
-
let hash = RHash::from_value(config).ok_or_else(|| arg_error("inline image config must be provided as a Hash"))?;
|
|
293
|
-
|
|
294
|
-
if let Some(limit) = get_kw(ruby, hash, "max_decoded_size_bytes") {
|
|
295
|
-
cfg.max_decoded_size_bytes = u64::try_convert(limit)?;
|
|
296
|
-
}
|
|
297
|
-
|
|
298
|
-
if let Some(prefix) = get_kw(ruby, hash, "filename_prefix") {
|
|
299
|
-
cfg.filename_prefix = if prefix.is_nil() {
|
|
300
|
-
None
|
|
301
|
-
} else {
|
|
302
|
-
Some(String::try_convert(prefix)?)
|
|
303
|
-
};
|
|
304
|
-
}
|
|
305
|
-
|
|
306
|
-
if let Some(capture_svg) = get_kw(ruby, hash, "capture_svg") {
|
|
307
|
-
cfg.capture_svg = bool::try_convert(capture_svg)?;
|
|
308
|
-
}
|
|
309
|
-
|
|
310
|
-
if let Some(infer_dimensions) = get_kw(ruby, hash, "infer_dimensions") {
|
|
311
|
-
cfg.infer_dimensions = bool::try_convert(infer_dimensions)?;
|
|
312
|
-
}
|
|
313
|
-
|
|
314
|
-
Ok(cfg)
|
|
315
|
-
}
|
|
316
|
-
|
|
317
|
-
fn inline_image_to_value(ruby: &Ruby, image: InlineImage) -> Result<Value, Error> {
|
|
318
|
-
let InlineImage {
|
|
319
|
-
data,
|
|
320
|
-
format,
|
|
321
|
-
filename,
|
|
322
|
-
description,
|
|
323
|
-
dimensions,
|
|
324
|
-
source,
|
|
325
|
-
attributes,
|
|
326
|
-
} = image;
|
|
327
|
-
|
|
328
|
-
let hash = ruby.hash_new();
|
|
329
|
-
let data_value = ruby.str_from_slice(&data);
|
|
330
|
-
hash.aset(ruby.intern("data"), data_value)?;
|
|
331
|
-
|
|
332
|
-
let format_value = match format {
|
|
333
|
-
InlineImageFormat::Png => "png".to_string(),
|
|
334
|
-
InlineImageFormat::Jpeg => "jpeg".to_string(),
|
|
335
|
-
InlineImageFormat::Gif => "gif".to_string(),
|
|
336
|
-
InlineImageFormat::Bmp => "bmp".to_string(),
|
|
337
|
-
InlineImageFormat::Webp => "webp".to_string(),
|
|
338
|
-
InlineImageFormat::Svg => "svg".to_string(),
|
|
339
|
-
InlineImageFormat::Other(other) => other,
|
|
340
|
-
};
|
|
341
|
-
hash.aset(ruby.intern("format"), format_value)?;
|
|
342
|
-
|
|
343
|
-
match filename {
|
|
344
|
-
Some(name) => hash.aset(ruby.intern("filename"), name)?,
|
|
345
|
-
None => hash.aset(ruby.intern("filename"), ruby.qnil())?,
|
|
346
|
-
}
|
|
347
|
-
|
|
348
|
-
match description {
|
|
349
|
-
Some(desc) => hash.aset(ruby.intern("description"), desc)?,
|
|
350
|
-
None => hash.aset(ruby.intern("description"), ruby.qnil())?,
|
|
351
|
-
}
|
|
352
|
-
|
|
353
|
-
if let Some((width, height)) = dimensions {
|
|
354
|
-
let dims = ruby.ary_new();
|
|
355
|
-
dims.push(width as i64)?;
|
|
356
|
-
dims.push(height as i64)?;
|
|
357
|
-
hash.aset(ruby.intern("dimensions"), dims)?;
|
|
358
|
-
} else {
|
|
359
|
-
hash.aset(ruby.intern("dimensions"), ruby.qnil())?;
|
|
360
|
-
}
|
|
361
|
-
|
|
362
|
-
let source_value = match source {
|
|
363
|
-
InlineImageSource::ImgDataUri => "img_data_uri",
|
|
364
|
-
InlineImageSource::SvgElement => "svg_element",
|
|
365
|
-
};
|
|
366
|
-
hash.aset(ruby.intern("source"), source_value)?;
|
|
367
|
-
|
|
368
|
-
let attrs = ruby.hash_new();
|
|
369
|
-
for (key, value) in attributes {
|
|
370
|
-
attrs.aset(key, value)?;
|
|
371
|
-
}
|
|
372
|
-
hash.aset(ruby.intern("attributes"), attrs)?;
|
|
373
|
-
|
|
374
|
-
Ok(hash.as_value())
|
|
375
|
-
}
|
|
376
|
-
|
|
377
|
-
fn warning_to_value(ruby: &Ruby, warning: InlineImageWarning) -> Result<Value, Error> {
|
|
378
|
-
let hash = ruby.hash_new();
|
|
379
|
-
hash.aset(ruby.intern("index"), warning.index as i64)?;
|
|
380
|
-
hash.aset(ruby.intern("message"), warning.message)?;
|
|
381
|
-
Ok(hash.as_value())
|
|
382
|
-
}
|
|
383
|
-
|
|
384
|
-
fn extraction_to_value(ruby: &Ruby, extraction: HtmlExtraction) -> Result<Value, Error> {
|
|
385
|
-
let hash = ruby.hash_new();
|
|
386
|
-
hash.aset(ruby.intern("markdown"), extraction.markdown)?;
|
|
387
|
-
|
|
388
|
-
let inline_images = ruby.ary_new();
|
|
389
|
-
for image in extraction.inline_images {
|
|
390
|
-
inline_images.push(inline_image_to_value(ruby, image)?)?;
|
|
391
|
-
}
|
|
392
|
-
hash.aset(ruby.intern("inline_images"), inline_images)?;
|
|
393
|
-
|
|
394
|
-
let warnings = ruby.ary_new();
|
|
395
|
-
for warning in extraction.warnings {
|
|
396
|
-
warnings.push(warning_to_value(ruby, warning)?)?;
|
|
397
|
-
}
|
|
398
|
-
hash.aset(ruby.intern("warnings"), warnings)?;
|
|
399
|
-
|
|
400
|
-
Ok(hash.as_value())
|
|
401
|
-
}
|
|
402
|
-
|
|
403
|
-
fn convert_fn(ruby: &Ruby, args: &[Value]) -> Result<String, Error> {
|
|
404
|
-
let parsed = scan_args::<(String,), (Option<Value>,), (), (), (), ()>(args)?;
|
|
405
|
-
let html = parsed.required.0;
|
|
406
|
-
let options = build_conversion_options(ruby, parsed.optional.0)?;
|
|
407
|
-
|
|
408
|
-
convert_inner(&html, Some(options)).map_err(conversion_error)
|
|
409
|
-
}
|
|
410
|
-
|
|
411
|
-
fn convert_with_inline_images_fn(ruby: &Ruby, args: &[Value]) -> Result<Value, Error> {
|
|
412
|
-
let parsed = scan_args::<(String,), (Option<Value>, Option<Value>), (), (), (), ()>(args)?;
|
|
413
|
-
let html = parsed.required.0;
|
|
414
|
-
let options = build_conversion_options(ruby, parsed.optional.0)?;
|
|
415
|
-
let config = build_inline_image_config(ruby, parsed.optional.1)?;
|
|
416
|
-
|
|
417
|
-
let extraction = convert_with_inline_images_inner(&html, Some(options), config).map_err(conversion_error)?;
|
|
418
|
-
|
|
419
|
-
extraction_to_value(ruby, extraction)
|
|
420
|
-
}
|
|
421
|
-
|
|
422
|
-
#[magnus::init]
|
|
423
|
-
fn init(ruby: &Ruby) -> Result<(), Error> {
|
|
424
|
-
let module = ruby.define_module("HtmlToMarkdown")?;
|
|
425
|
-
module.define_singleton_method("convert", function!(convert_fn, -1))?;
|
|
426
|
-
module.define_singleton_method(
|
|
427
|
-
"convert_with_inline_images",
|
|
428
|
-
function!(convert_with_inline_images_fn, -1),
|
|
429
|
-
)?;
|
|
430
|
-
|
|
431
|
-
Ok(())
|
|
432
|
-
}
|