mathml2asciimath 0.0.9 → 0.0.13
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/.github/workflows/rake.yml +32 -0
- data/.gitignore +1 -0
- data/.hound.yml +3 -1
- data/.rubocop.yml +7 -7
- data/Gemfile +1 -2
- data/README.adoc +6 -6
- data/Rakefile +1 -1
- data/bin/m2a.rb +5 -7
- data/bin/rspec +1 -1
- data/lib/mathml2asciimath/m2a.rb +208 -210
- data/lib/mathml2asciimath/version.rb +1 -2
- data/mathml2asciimath.gemspec +7 -7
- data/spec/mathml_spec.rb +329 -327
- data/spec/spec_helper.rb +6 -7
- metadata +37 -41
- data/.github/workflows/macos.yml +0 -31
- data/.github/workflows/ubuntu.yml +0 -43
- data/.github/workflows/windows.yml +0 -34
- data/.rubocop.ribose.yml +0 -65
- data/.rubocop.tb.yml +0 -650
- data/Gemfile.lock +0 -106
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: a11fa790b22436b9f9d06a40d90264a67e2bef3fd08f02538daa01a16a4818f8
|
|
4
|
+
data.tar.gz: c017afbd1e64bb121517f6a508193d37d2157326b4aca00953d9764ca4920731
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 6b1371dffc46a011741a9d0544860327e2ce5ef275b7f0461bf00bc50d84d443af8c10760e2db75f32d7a1a3c6f57edfa986ea57df66065fc96ed5964c4fc545
|
|
7
|
+
data.tar.gz: 37ad642dd7ef784027a713433abdccf1d273d469dc3b24fcb1d74e0ef05e27f8c92eb4911605ca9d91fe14ba55c20e8366a056ba94eb54276b0ad6c4f36d4f84
|
|
@@ -0,0 +1,32 @@
|
|
|
1
|
+
# Auto-generated by Cimas: Do not edit it manually!
|
|
2
|
+
# See https://github.com/metanorma/cimas
|
|
3
|
+
name: rake
|
|
4
|
+
|
|
5
|
+
on:
|
|
6
|
+
push:
|
|
7
|
+
branches: [ master, main ]
|
|
8
|
+
tags: [ v* ]
|
|
9
|
+
pull_request:
|
|
10
|
+
|
|
11
|
+
jobs:
|
|
12
|
+
rake:
|
|
13
|
+
name: Test on Ruby ${{ matrix.ruby }} ${{ matrix.os }}
|
|
14
|
+
runs-on: ${{ matrix.os }}
|
|
15
|
+
continue-on-error: ${{ matrix.experimental }}
|
|
16
|
+
strategy:
|
|
17
|
+
fail-fast: false
|
|
18
|
+
matrix:
|
|
19
|
+
ruby: [ '3.0', '2.7', '2.6', '2.5' ]
|
|
20
|
+
os: [ ubuntu-latest, windows-latest, macos-latest ]
|
|
21
|
+
experimental: [ false ]
|
|
22
|
+
steps:
|
|
23
|
+
- uses: actions/checkout@v2
|
|
24
|
+
with:
|
|
25
|
+
submodules: true
|
|
26
|
+
|
|
27
|
+
- uses: ruby/setup-ruby@v1
|
|
28
|
+
with:
|
|
29
|
+
ruby-version: ${{ matrix.ruby }}
|
|
30
|
+
bundler-cache: true
|
|
31
|
+
|
|
32
|
+
- run: bundle exec rake
|
data/.gitignore
ADDED
|
@@ -0,0 +1 @@
|
|
|
1
|
+
.rubocop-https--*
|
data/.hound.yml
CHANGED
data/.rubocop.yml
CHANGED
|
@@ -1,10 +1,10 @@
|
|
|
1
|
-
#
|
|
2
|
-
# https://github.com/
|
|
3
|
-
# All project-specific additions and overrides should be specified in this file.
|
|
4
|
-
|
|
1
|
+
# Auto-generated by Cimas: Do not edit it manually!
|
|
2
|
+
# See https://github.com/metanorma/cimas
|
|
5
3
|
inherit_from:
|
|
6
4
|
- https://raw.githubusercontent.com/riboseinc/oss-guides/master/ci/rubocop.yml
|
|
5
|
+
|
|
6
|
+
# local repo-specific modifications
|
|
7
|
+
# ...
|
|
8
|
+
|
|
7
9
|
AllCops:
|
|
8
|
-
TargetRubyVersion: 2.
|
|
9
|
-
Rails:
|
|
10
|
-
Enabled: true
|
|
10
|
+
TargetRubyVersion: 2.5
|
data/Gemfile
CHANGED
data/README.adoc
CHANGED
|
@@ -1,12 +1,12 @@
|
|
|
1
1
|
= mathml2asciimath
|
|
2
2
|
|
|
3
3
|
image:https://img.shields.io/gem/v/mathml2asciimath.svg["Gem Version", link="https://rubygems.org/gems/mathml2asciimath"]
|
|
4
|
-
image:https://github.com/
|
|
5
|
-
image:https://github.com/
|
|
6
|
-
image:https://github.com/
|
|
7
|
-
image:https://codeclimate.com/github/
|
|
8
|
-
image:https://img.shields.io/github/issues-pr-raw/
|
|
9
|
-
image:https://img.shields.io/github/commits-since/
|
|
4
|
+
image:https://github.com/plurimath/mathml2asciimath/workflows/ubuntu/badge.svg["Ubuntu Build Status", link="https://github.com/plurimath/mathml2asciimath/actions?query=workflow%3Aubuntu"]
|
|
5
|
+
image:https://github.com/plurimath/mathml2asciimath/workflows/macos/badge.svg["OSX Build Status", link="https://github.com/plurimath/mathml2asciimath/actions?query=workflow%3Amacos"]
|
|
6
|
+
image:https://github.com/plurimath/mathml2asciimath/workflows/windows/badge.svg["Windows Build Status", link="https://github.com/plurimath/mathml2asciimath/actions?query=workflow%3Awindows"]
|
|
7
|
+
image:https://codeclimate.com/github/plurimath/mathml2asciimath/badges/gpa.svg["Code Climate", link="https://codeclimate.com/github/plurimath/mathml2asciimath"]
|
|
8
|
+
image:https://img.shields.io/github/issues-pr-raw/plurimath/mathml2asciimath.svg["Pull Requests", link="https://github.com/plurimath/mathml2asciimath/pulls"]
|
|
9
|
+
image:https://img.shields.io/github/commits-since/plurimath/mathml2asciimath/latest.svg["Commits since latest",link="https://github.com/plurimath/mathml2asciimath/releases"]
|
|
10
10
|
|
|
11
11
|
Ruby gem to convert MathML into AsciiMath
|
|
12
12
|
|
data/Rakefile
CHANGED
data/bin/m2a.rb
CHANGED
|
@@ -5,13 +5,11 @@ while a = gets
|
|
|
5
5
|
matches = a.split(%r{(<math.*?</math>)})
|
|
6
6
|
out = ""
|
|
7
7
|
matches.each do |x|
|
|
8
|
-
if /<math/.match x
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
|
|
8
|
+
out += if /<math/.match? x
|
|
9
|
+
MathML2AsciiMath.m2a(x)
|
|
10
|
+
else
|
|
11
|
+
x
|
|
12
|
+
end
|
|
13
13
|
end
|
|
14
14
|
print out
|
|
15
15
|
end
|
|
16
|
-
|
|
17
|
-
|
data/bin/rspec
CHANGED
data/lib/mathml2asciimath/m2a.rb
CHANGED
|
@@ -3,161 +3,172 @@ require "htmlentities"
|
|
|
3
3
|
require "pp"
|
|
4
4
|
|
|
5
5
|
module MathML2AsciiMath
|
|
6
|
+
def self.m2a(xml)
|
|
7
|
+
normalized = xml
|
|
6
8
|
|
|
7
|
-
|
|
8
|
-
docxml = Nokogiri::XML(
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
9
|
+
# &:noblanks skips non-significant whitespaces in MathML
|
|
10
|
+
docxml = Nokogiri::XML.parse(normalized, &:noblanks)
|
|
11
|
+
|
|
12
|
+
# Get rid of things like
|
|
13
|
+
# <mtext> </mtext>
|
|
14
|
+
parse(docxml.root).gsub(/[[:blank:]]/, " ").unicode_normalize.squeeze(" ")
|
|
15
|
+
end
|
|
16
|
+
|
|
17
|
+
def self.encodechars(xml)
|
|
18
|
+
xml.gsub(/\u03b1/, "alpha")
|
|
19
|
+
.gsub(/\u03b2/, "beta")
|
|
20
|
+
.gsub(/\u03b3/, "gamma")
|
|
21
|
+
.gsub(/\u0393/, "Gamma")
|
|
22
|
+
.gsub(/\u03b4/, "delta")
|
|
23
|
+
.gsub(/\u0394/, "Delta")
|
|
24
|
+
.gsub(/\u2206/, "Delta")
|
|
25
|
+
.gsub(/\u03b5/, "epsilon")
|
|
26
|
+
.gsub(/\u025b/, "varepsilon")
|
|
27
|
+
.gsub(/\u03b6/, "zeta")
|
|
28
|
+
.gsub(/\u03b7/, "eta")
|
|
29
|
+
.gsub(/\u03b8/, "theta")
|
|
30
|
+
.gsub(/\u0398/, "Theta")
|
|
31
|
+
.gsub(/\u03d1/, "vartheta")
|
|
32
|
+
.gsub(/\u03b9/, "iota")
|
|
33
|
+
.gsub(/\u03ba/, "kappa")
|
|
34
|
+
.gsub(/\u03bb/, "lambda")
|
|
35
|
+
.gsub(/\u039b/, "Lambda")
|
|
36
|
+
.gsub(/\u03bc/, "mu")
|
|
37
|
+
.gsub(/\u03bd/, "nu")
|
|
38
|
+
.gsub(/\u03be/, "xi")
|
|
39
|
+
.gsub(/\u039e/, "Xi")
|
|
40
|
+
.gsub(/\u03c0/, "pi")
|
|
41
|
+
.gsub(/\u03a0/, "Pi")
|
|
42
|
+
.gsub(/\u03c1/, "rho")
|
|
43
|
+
.gsub(/\u03c2/, "beta")
|
|
44
|
+
.gsub(/\u03c3/, "sigma")
|
|
45
|
+
.gsub(/\u03a3/, "Sigma")
|
|
46
|
+
.gsub(/\u03c4/, "tau")
|
|
47
|
+
.gsub(/\u03c5/, "upsilon")
|
|
48
|
+
.gsub(/\u03c6/, "phi")
|
|
49
|
+
.gsub(/\u03a6/, "Phi")
|
|
50
|
+
.gsub(/\u03d5/, "varphi")
|
|
51
|
+
.gsub(/\u03c7/, "chi")
|
|
52
|
+
.gsub(/\u03c8/, "psi")
|
|
53
|
+
.gsub(/\u03a8/, "Psi")
|
|
54
|
+
.gsub(/\u03c9/, "omega")
|
|
55
|
+
.gsub(/\u03a9/, "omega")
|
|
56
|
+
.gsub(/\u22c5/, "*")
|
|
57
|
+
.gsub(/\u2219/, "*")
|
|
58
|
+
.gsub(/\u00b7/, "*")
|
|
59
|
+
.gsub(/\u2217/, "**")
|
|
60
|
+
.gsub(/\u22c6/, "***")
|
|
61
|
+
.gsub(/\//, "//")
|
|
62
|
+
.gsub(/\\/, "\\\\")
|
|
63
|
+
.gsub(/\u00d7/, "xx")
|
|
64
|
+
.gsub(/\u22c9/, "|><")
|
|
65
|
+
.gsub(/\u22ca/, "><|")
|
|
66
|
+
.gsub(/\u22c8/, "|><|")
|
|
67
|
+
.gsub(/\u00f7/, "-:")
|
|
68
|
+
.gsub(/\u2218/, "@")
|
|
69
|
+
.gsub(/\u2295/, "o+")
|
|
70
|
+
.gsub(/\u2a01/, "o+")
|
|
71
|
+
.gsub(/\u2297/, "ox")
|
|
72
|
+
.gsub(/\u2299/, "o.")
|
|
73
|
+
.gsub(/\u2211/, "sum")
|
|
74
|
+
.gsub(/\u220f/, "prod")
|
|
75
|
+
.gsub(/\u2227/, "^^")
|
|
76
|
+
.gsub(/\u22c0/, "^^^")
|
|
77
|
+
.gsub(/\u2228/, "vv")
|
|
78
|
+
.gsub(/\u22c1/, "vvv")
|
|
79
|
+
.gsub(/\u2229/, "nn")
|
|
80
|
+
.gsub(/\u22c2/, "nnn")
|
|
81
|
+
.gsub(/\u222a/, "uu")
|
|
82
|
+
.gsub(/\u22c3/, "uuu")
|
|
83
|
+
.gsub(/\u2260/, "!=")
|
|
84
|
+
.gsub(/\u2264/, "<=")
|
|
85
|
+
.gsub(/\u2265/, ">=")
|
|
86
|
+
.gsub(/\u227a/, "-<")
|
|
87
|
+
.gsub(/\u227b/, ">-")
|
|
88
|
+
.gsub(/\u2aaf/, "-<=")
|
|
89
|
+
.gsub(/\u2ab0/, ">-=")
|
|
90
|
+
.gsub(/\u2208/, "in")
|
|
91
|
+
.gsub(/\u2209/, "!in")
|
|
92
|
+
.gsub(/\u2282/, "sub")
|
|
93
|
+
.gsub(/\u2283/, "sup")
|
|
94
|
+
.gsub(/\u2286/, "sube")
|
|
95
|
+
.gsub(/\u2287/, "supe")
|
|
96
|
+
.gsub(/\u2261/, "-=")
|
|
97
|
+
.gsub(/\u2245/, "~=")
|
|
98
|
+
.gsub(/\u2248/, "~~")
|
|
99
|
+
.gsub(/\u221d/, "prop")
|
|
100
|
+
.gsub(/\u00ac/, "not")
|
|
101
|
+
.gsub(/\u21d2/, "=>")
|
|
102
|
+
.gsub(/\u21d4/, "<=>")
|
|
103
|
+
.gsub(/\u2200/, "AA")
|
|
104
|
+
.gsub(/\u2203/, "EE")
|
|
105
|
+
.gsub(/\u22a5/, "_|_")
|
|
106
|
+
.gsub(/\u22a4/, "TT")
|
|
107
|
+
.gsub(/\u22a2/, "|--")
|
|
108
|
+
.gsub(/\u22a8/, "|==")
|
|
109
|
+
.gsub(/\u22a8/, "|==")
|
|
110
|
+
.gsub(/\u2329/, "(:")
|
|
111
|
+
.gsub(/\u232a/, ":)")
|
|
112
|
+
.gsub(/\u2329/, "<<")
|
|
113
|
+
.gsub(/\u27e8/, "<<")
|
|
114
|
+
.gsub(/\u232a/, ">>")
|
|
115
|
+
.gsub(/\u27e9/, ">>")
|
|
116
|
+
.gsub(/\u222b/, "int")
|
|
117
|
+
.gsub(/\u222e/, "oint")
|
|
118
|
+
.gsub(/\u2202/, "del")
|
|
119
|
+
.gsub(/\u2207/, "grad")
|
|
120
|
+
.gsub(/\u00b1/, "+-")
|
|
121
|
+
.gsub(/\u2205/, "O/")
|
|
122
|
+
.gsub(/\u221e/, "oo")
|
|
123
|
+
.gsub(/\u2135/, "aleph")
|
|
124
|
+
.gsub(/\u2234/, ":.")
|
|
125
|
+
.gsub(/\u2235/, ":'")
|
|
126
|
+
.gsub(/\u2220/, "/_")
|
|
127
|
+
.gsub(/\u25b3/, "/_\\")
|
|
128
|
+
.gsub(/\u2032/, "'")
|
|
129
|
+
.gsub(/~/, "tilde")
|
|
130
|
+
.gsub(/\u00a0\u00a0\u00a0\u00a0/, "qquad")
|
|
131
|
+
.gsub(/\u00a0\u00a0/, "quad")
|
|
132
|
+
.gsub(/\u00a0/, "\\ ")
|
|
133
|
+
.gsub(/\u2322/, "frown")
|
|
134
|
+
.gsub(/\u00a0/, "quad")
|
|
135
|
+
.gsub(/\u22ef/, "cdots")
|
|
136
|
+
.gsub(/\u22ee/, "vdots")
|
|
137
|
+
.gsub(/\u22f1/, "ddots")
|
|
138
|
+
.gsub(/\u22c4/, "diamond")
|
|
139
|
+
.gsub(/\u25a1/, "square")
|
|
140
|
+
.gsub(/\u230a/, "|__")
|
|
141
|
+
.gsub(/\u230b/, "__|")
|
|
142
|
+
.gsub(/\u2308/, "|~")
|
|
143
|
+
.gsub(/\u2309/, "~|")
|
|
144
|
+
.gsub(/\u2102/, "CC")
|
|
145
|
+
.gsub(/\u2115/, "NN")
|
|
146
|
+
.gsub(/\u211a/, "QQ")
|
|
147
|
+
.gsub(/\u211d/, "RR")
|
|
148
|
+
.gsub(/\u2124/, "ZZ")
|
|
149
|
+
.gsub(/\u2191/, "uarr")
|
|
150
|
+
.gsub(/\u2193/, "darr")
|
|
151
|
+
.gsub(/\u2190/, "larr")
|
|
152
|
+
.gsub(/\u2194/, "harr")
|
|
153
|
+
.gsub(/\u21d2/, "rArr")
|
|
154
|
+
.gsub(/\u21d0/, "lArr")
|
|
155
|
+
.gsub(/\u21d4/, "hArr")
|
|
156
|
+
.gsub(/\u2192/, "->")
|
|
157
|
+
.gsub(/\u21a3/, ">->")
|
|
158
|
+
.gsub(/\u21a0/, "->>")
|
|
159
|
+
.gsub(/\u2916/, ">->>")
|
|
160
|
+
.gsub(/\u21a6/, "|->")
|
|
161
|
+
.gsub(/\u2026/, "...")
|
|
162
|
+
.gsub(/\u2212/, "-")
|
|
163
|
+
.gsub(/\u2061/, "") # function application
|
|
164
|
+
.gsub(/\u2751/, "square")
|
|
165
|
+
.gsub(/[\u2028\u2029]/, " ") # normalize thin spaces like \u2009, \u2008
|
|
12
166
|
end
|
|
13
167
|
|
|
14
|
-
def self.
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
gsub(/\u0393/, "Gamma").
|
|
19
|
-
gsub(/\u03b4/, "delta").
|
|
20
|
-
gsub(/\u0394/, "Delta").
|
|
21
|
-
gsub(/\u2206/, "Delta").
|
|
22
|
-
gsub(/\u03b5/, "epsilon").
|
|
23
|
-
gsub(/\u025b/, "varepsilon").
|
|
24
|
-
gsub(/\u03b6/, "zeta").
|
|
25
|
-
gsub(/\u03b7/, "eta").
|
|
26
|
-
gsub(/\u03b8/, "theta").
|
|
27
|
-
gsub(/\u0398/, "Theta").
|
|
28
|
-
gsub(/\u03d1/, "vartheta").
|
|
29
|
-
gsub(/\u03b9/, "iota").
|
|
30
|
-
gsub(/\u03ba/, "kappa").
|
|
31
|
-
gsub(/\u03bb/, "lambda").
|
|
32
|
-
gsub(/\u039b/, "Lambda").
|
|
33
|
-
gsub(/\u03bc/, "mu").
|
|
34
|
-
gsub(/\u03bd/, "nu").
|
|
35
|
-
gsub(/\u03be/, "xi").
|
|
36
|
-
gsub(/\u039e/, "Xi").
|
|
37
|
-
gsub(/\u03c0/, "pi").
|
|
38
|
-
gsub(/\u03a0/, "Pi").
|
|
39
|
-
gsub(/\u03c1/, "rho").
|
|
40
|
-
gsub(/\u03c2/, "beta").
|
|
41
|
-
gsub(/\u03c3/, "sigma").
|
|
42
|
-
gsub(/\u03a3/, "Sigma").
|
|
43
|
-
gsub(/\u03c4/, "tau").
|
|
44
|
-
gsub(/\u03c5/, "upsilon").
|
|
45
|
-
gsub(/\u03c6/, "phi").
|
|
46
|
-
gsub(/\u03a6/, "Phi").
|
|
47
|
-
gsub(/\u03d5/, "varphi").
|
|
48
|
-
gsub(/\u03c7/, "chi").
|
|
49
|
-
gsub(/\u03c8/, "psi").
|
|
50
|
-
gsub(/\u03a8/, "Psi").
|
|
51
|
-
gsub(/\u03c9/, "omega").
|
|
52
|
-
gsub(/\u03a9/, "omega").
|
|
53
|
-
gsub(/\u22c5/, "*").
|
|
54
|
-
gsub(/\u2219/, "*").
|
|
55
|
-
gsub(/\u00b7/, "*").
|
|
56
|
-
gsub(/\u2217/, "**").
|
|
57
|
-
gsub(/\u22c6/, "***").
|
|
58
|
-
gsub(/\//, "//").
|
|
59
|
-
gsub(/\\/, "\\\\").
|
|
60
|
-
gsub(/\u00d7/, "xx").
|
|
61
|
-
gsub(/\u22c9/, "|><").
|
|
62
|
-
gsub(/\u22ca/, "><|").
|
|
63
|
-
gsub(/\u22c8/, "|><|").
|
|
64
|
-
gsub(/\u00f7/, "-:").
|
|
65
|
-
gsub(/\u2218/, "@").
|
|
66
|
-
gsub(/\u2295/, "o+").
|
|
67
|
-
gsub(/\u2a01/, "o+").
|
|
68
|
-
gsub(/\u2297/, "ox").
|
|
69
|
-
gsub(/\u2299/, "o.").
|
|
70
|
-
gsub(/\u2211/, "sum").
|
|
71
|
-
gsub(/\u220f/, "prod").
|
|
72
|
-
gsub(/\u2227/, "^^").
|
|
73
|
-
gsub(/\u22c0/, "^^^").
|
|
74
|
-
gsub(/\u2228/, "vv").
|
|
75
|
-
gsub(/\u22c1/, "vvv").
|
|
76
|
-
gsub(/\u2229/, "nn").
|
|
77
|
-
gsub(/\u22c2/, "nnn").
|
|
78
|
-
gsub(/\u222a/, "uu").
|
|
79
|
-
gsub(/\u22c3/, "uuu").
|
|
80
|
-
gsub(/\u2260/, "!=").
|
|
81
|
-
gsub(/\u2264/, "<=").
|
|
82
|
-
gsub(/\u2265/, ">=").
|
|
83
|
-
gsub(/\u227a/, "-<").
|
|
84
|
-
gsub(/\u227b/, ">-").
|
|
85
|
-
gsub(/\u2aaf/, "-<=").
|
|
86
|
-
gsub(/\u2ab0/, ">-=").
|
|
87
|
-
gsub(/\u2208/, "in").
|
|
88
|
-
gsub(/\u2209/, "!in").
|
|
89
|
-
gsub(/\u2282/, "sub").
|
|
90
|
-
gsub(/\u2283/, "sup").
|
|
91
|
-
gsub(/\u2286/, "sube").
|
|
92
|
-
gsub(/\u2287/, "supe").
|
|
93
|
-
gsub(/\u2261/, "-=").
|
|
94
|
-
gsub(/\u2245/, "~=").
|
|
95
|
-
gsub(/\u2248/, "~~").
|
|
96
|
-
gsub(/\u221d/, "prop").
|
|
97
|
-
gsub(/\u00ac/, "not").
|
|
98
|
-
gsub(/\u21d2/, "=>").
|
|
99
|
-
gsub(/\u21d4/, "<=>").
|
|
100
|
-
gsub(/\u2200/, "AA").
|
|
101
|
-
gsub(/\u2203/, "EE").
|
|
102
|
-
gsub(/\u22a5/, "_|_").
|
|
103
|
-
gsub(/\u22a4/, "TT").
|
|
104
|
-
gsub(/\u22a2/, "|--").
|
|
105
|
-
gsub(/\u22a8/, "|==").
|
|
106
|
-
gsub(/\u22a8/, "|==").
|
|
107
|
-
gsub(/\u2329/, "(:").
|
|
108
|
-
gsub(/\u232a/, ":)").
|
|
109
|
-
gsub(/\u2329/, "<<").
|
|
110
|
-
gsub(/\u27e8/, "<<").
|
|
111
|
-
gsub(/\u232a/, ">>").
|
|
112
|
-
gsub(/\u27e9/, ">>").
|
|
113
|
-
gsub(/\u222e/, "oint").
|
|
114
|
-
gsub(/\u2202/, "del").
|
|
115
|
-
gsub(/\u2207/, "grad").
|
|
116
|
-
gsub(/\u00b1/, "+-").
|
|
117
|
-
gsub(/\u2205/, "O/").
|
|
118
|
-
gsub(/\u221e/, "oo").
|
|
119
|
-
gsub(/\u2135/, "aleph").
|
|
120
|
-
gsub(/\u2234/, ":.").
|
|
121
|
-
gsub(/\u2235/, ":'").
|
|
122
|
-
gsub(/\u2220/, "/_").
|
|
123
|
-
gsub(/\u25b3/, "/_\\").
|
|
124
|
-
gsub(/\u2032/, "'").
|
|
125
|
-
gsub(/~/, "tilde").
|
|
126
|
-
gsub(/\u00a0\u00a0\u00a0\u00a0/, "qquad").
|
|
127
|
-
gsub(/\u00a0\u00a0/, "quad").
|
|
128
|
-
gsub(/\u00a0/, "\\ ").
|
|
129
|
-
gsub(/\u2322/, "frown").
|
|
130
|
-
gsub(/\u00a0/, "quad").
|
|
131
|
-
gsub(/\u22ef/, "cdots").
|
|
132
|
-
gsub(/\u22ee/, "vdots").
|
|
133
|
-
gsub(/\u22f1/, "ddots").
|
|
134
|
-
gsub(/\u22c4/, "diamond").
|
|
135
|
-
gsub(/\u25a1/, "square").
|
|
136
|
-
gsub(/\u230a/, "|__").
|
|
137
|
-
gsub(/\u230b/, "__|").
|
|
138
|
-
gsub(/\u2308/, "|~").
|
|
139
|
-
gsub(/\u2309/, "~|").
|
|
140
|
-
gsub(/\u2102/, "CC").
|
|
141
|
-
gsub(/\u2115/, "NN").
|
|
142
|
-
gsub(/\u211a/, "QQ").
|
|
143
|
-
gsub(/\u211d/, "RR").
|
|
144
|
-
gsub(/\u2124/, "ZZ").
|
|
145
|
-
gsub(/\u2191/, "uarr").
|
|
146
|
-
gsub(/\u2193/, "darr").
|
|
147
|
-
gsub(/\u2190/, "larr").
|
|
148
|
-
gsub(/\u2194/, "harr").
|
|
149
|
-
gsub(/\u21d2/, "rArr").
|
|
150
|
-
gsub(/\u21d0/, "lArr").
|
|
151
|
-
gsub(/\u21d4/, "hArr").
|
|
152
|
-
gsub(/\u2192/, "->").
|
|
153
|
-
gsub(/\u21a3/, ">->").
|
|
154
|
-
gsub(/\u21a0/, "->>").
|
|
155
|
-
gsub(/\u2916/, ">->>").
|
|
156
|
-
gsub(/\u21a6/, "|->").
|
|
157
|
-
gsub(/\u2026/, "...").
|
|
158
|
-
gsub(/\u2212/, "-").
|
|
159
|
-
gsub(/\u2061/, ""). # function application
|
|
160
|
-
gsub(/\u2751/, "square")
|
|
168
|
+
def self.join_parsed_children(children, delimiter = " ")
|
|
169
|
+
children.map do |n|
|
|
170
|
+
parse(n).strip
|
|
171
|
+
end.join(delimiter)
|
|
161
172
|
end
|
|
162
173
|
|
|
163
174
|
def self.parse(node)
|
|
@@ -168,55 +179,47 @@ module MathML2AsciiMath
|
|
|
168
179
|
|
|
169
180
|
case node.name.sub(/^[^:]*:/, "")
|
|
170
181
|
when "math"
|
|
171
|
-
|
|
172
|
-
node.elements.each { |n| outarr << parse(n).strip }
|
|
173
|
-
return outarr.join(" ")
|
|
182
|
+
join_parsed_children(node.elements)
|
|
174
183
|
|
|
175
184
|
when "annotation"
|
|
176
|
-
|
|
185
|
+
""
|
|
177
186
|
|
|
178
187
|
when "semantics"
|
|
179
|
-
|
|
180
|
-
node.elements.each { |n| outarr << parse(n).strip }
|
|
181
|
-
return outarr.join(" ")
|
|
188
|
+
join_parsed_children(node.elements)
|
|
182
189
|
|
|
183
190
|
when "mrow"
|
|
184
|
-
|
|
185
|
-
|
|
186
|
-
|
|
187
|
-
if %w{mfrac msub munder munderover}.include? node.parent.name.sub(/^[^:]*:/, "")
|
|
191
|
+
out = join_parsed_children(node.elements)
|
|
192
|
+
if %w[mfrac msub munder munderover]
|
|
193
|
+
.include? node.parent.name.sub(/^[^:]*:/, "")
|
|
188
194
|
out = "(#{out})"
|
|
189
195
|
end
|
|
190
|
-
|
|
196
|
+
out
|
|
191
197
|
|
|
192
198
|
when "mfenced"
|
|
193
|
-
|
|
194
|
-
|
|
195
|
-
|
|
199
|
+
sym_open = node["open"] || "("
|
|
200
|
+
sym_close = node["close"] || ")"
|
|
201
|
+
|
|
196
202
|
separator = "," # TODO currently ignore the supplied separators
|
|
197
|
-
node.elements
|
|
198
|
-
out
|
|
199
|
-
return "#{open}#{out}#{close}"
|
|
203
|
+
out = join_parsed_children(node.elements, separator)
|
|
204
|
+
"#{sym_open}#{out}#{sym_close}"
|
|
200
205
|
|
|
201
206
|
when "msqrt"
|
|
202
|
-
|
|
203
|
-
node.elements.each { |n| outarr << parse(n).strip }
|
|
204
|
-
return "sqrt(#{outarr.join(" ")})"
|
|
207
|
+
"sqrt(#{join_parsed_children(node.elements)})"
|
|
205
208
|
|
|
206
209
|
when "mfrac"
|
|
207
|
-
|
|
210
|
+
"(#{parse(node.elements[0])})/(#{parse(node.elements[1])})"
|
|
208
211
|
|
|
209
212
|
when "msup"
|
|
210
213
|
sup = parse(node.elements[1])
|
|
211
214
|
sup = "(#{sup})" unless sup.length == 1
|
|
212
215
|
op = parse(node.elements[0]).gsub(/ $/, "")
|
|
213
|
-
|
|
216
|
+
"#{op}^#{sup}"
|
|
214
217
|
|
|
215
218
|
when "msub"
|
|
216
219
|
sub = parse(node.elements[1])
|
|
217
220
|
sub = "(#{sub})" unless sub.length == 1
|
|
218
221
|
op = parse(node.elements[0]).gsub(/ $/, "")
|
|
219
|
-
|
|
222
|
+
"#{op}_#{sub}"
|
|
220
223
|
|
|
221
224
|
when "munderover", "msubsup"
|
|
222
225
|
sub = parse(node.elements[1])
|
|
@@ -224,27 +227,29 @@ module MathML2AsciiMath
|
|
|
224
227
|
sup = parse(node.elements[2])
|
|
225
228
|
sup = "(#{sup})" unless sup.length == 1
|
|
226
229
|
op = parse(node.elements[0]).gsub(/ $/, "")
|
|
227
|
-
|
|
230
|
+
"#{op}_#{sub}^#{sup}"
|
|
228
231
|
|
|
229
232
|
when "munder"
|
|
230
|
-
elem1 = parse(node.elements[1]).
|
|
233
|
+
elem1 = parse(node.elements[1]).strip
|
|
231
234
|
accent = case elem1
|
|
232
235
|
when "\u0332" then "ul"
|
|
233
236
|
when "\u23df" then "ubrace"
|
|
234
237
|
else
|
|
235
238
|
"underset"
|
|
236
239
|
end
|
|
240
|
+
|
|
237
241
|
if accent == "underset"
|
|
238
|
-
|
|
242
|
+
"underset(#{elem1})(#{parse(node.elements[0])})"
|
|
239
243
|
else
|
|
240
|
-
|
|
244
|
+
"#{accent} #{parse(node.elements[0])}"
|
|
241
245
|
end
|
|
246
|
+
|
|
242
247
|
when "mover"
|
|
243
|
-
elem1 = parse(node.elements[1]).
|
|
248
|
+
elem1 = parse(node.elements[1]).strip
|
|
244
249
|
accent = case elem1
|
|
245
250
|
when "\u005e" then "hat"
|
|
246
251
|
when "\u00af" then "bar"
|
|
247
|
-
#when "\u2192" then "vec"
|
|
252
|
+
# when "\u2192" then "vec"
|
|
248
253
|
when "->" then "vec"
|
|
249
254
|
when "." then "dot"
|
|
250
255
|
when ".." then "ddot"
|
|
@@ -252,53 +257,46 @@ module MathML2AsciiMath
|
|
|
252
257
|
else
|
|
253
258
|
"overset"
|
|
254
259
|
end
|
|
260
|
+
|
|
255
261
|
if accent == "overset"
|
|
256
|
-
|
|
262
|
+
"overset(#{elem1})(#{parse(node.elements[0])})"
|
|
257
263
|
else
|
|
258
|
-
|
|
264
|
+
"#{accent} #{parse(node.elements[0])}"
|
|
259
265
|
end
|
|
260
266
|
|
|
261
267
|
when "mtable"
|
|
262
|
-
|
|
263
|
-
node.elements.each { |n| outarr << parse(n).strip }
|
|
264
|
-
return "[#{outarr.join(",")}]"
|
|
268
|
+
"[#{join_parsed_children(node.elements, ',')}]"
|
|
265
269
|
|
|
266
270
|
when "mtr"
|
|
267
|
-
|
|
268
|
-
node.elements.each { |n| outarr << parse(n).strip }
|
|
269
|
-
return "[#{outarr.join(",")}]"
|
|
271
|
+
"[#{join_parsed_children(node.elements, ',')}]"
|
|
270
272
|
|
|
271
273
|
when "mtd"
|
|
272
|
-
|
|
273
|
-
node.elements.each { |n| outarr << parse(n).strip }
|
|
274
|
-
return "#{outarr.join(",")}"
|
|
274
|
+
join_parsed_children(node.elements, ",")
|
|
275
275
|
|
|
276
276
|
when "mn", "mtext"
|
|
277
|
-
|
|
278
|
-
node.children.each { |n| outarr << parse(n).strip }
|
|
279
|
-
return "#{outarr.join("")}"
|
|
277
|
+
join_parsed_children(node.children, "")
|
|
280
278
|
|
|
281
279
|
when "mi"
|
|
282
|
-
#
|
|
283
|
-
|
|
284
|
-
|
|
285
|
-
|
|
286
|
-
|
|
287
|
-
|
|
280
|
+
# FIXME: What does this comment have to do with Word?
|
|
281
|
+
# mi is not meant to have space around it,
|
|
282
|
+
# but Word is conflating operators and operands
|
|
283
|
+
join_parsed_children(node.children)
|
|
284
|
+
|
|
285
|
+
# FIXME: Why do we need to add extra spaces?
|
|
286
|
+
# out = " #{out} " if /[^a-zA-Z0-9',]|[a-z][a-z]/.match out
|
|
288
287
|
|
|
289
288
|
when "mo"
|
|
290
|
-
|
|
291
|
-
node.children.each { |n| outarr << parse(n).strip }
|
|
292
|
-
out = outarr.join(" ")
|
|
289
|
+
out = join_parsed_children(node.children)
|
|
293
290
|
out = " #{out} " unless node["fence"]
|
|
294
|
-
|
|
291
|
+
out
|
|
295
292
|
|
|
296
293
|
when "mstyle"
|
|
297
|
-
|
|
298
|
-
|
|
299
|
-
out = outarr.join(" ")
|
|
294
|
+
join_parsed_children(node.children)
|
|
295
|
+
|
|
300
296
|
else
|
|
301
|
-
|
|
297
|
+
"<math xmlns=\"http://www.w3.org/1998/Math/MathML\">" +
|
|
298
|
+
node.to_xml +
|
|
299
|
+
"</math>"
|
|
302
300
|
|
|
303
301
|
end
|
|
304
302
|
end
|