test_tex_math.py

import pytest
from lxml import etree
from anthology.texmath import TexMath

texmath = TexMath()

test_cases_unicode = (
    (
        '<tex-math>K</tex-math>-Embeddings: Learning Conceptual Embeddings for Words using Context',
        'K-Embeddings: Learning Conceptual Embeddings for Words using Context',
    ),
    (
        '<tex-math>^{\\mathcal{E}}</tex-math>: a Vectorial Resource for Computing Conceptual Similarity',
        'ℰ: a Vectorial Resource for Computing Conceptual Similarity',
    ),
    (
        '<tex-math>\\sharp</tex-math>: An Enhancement Approach to Reference-based Evaluation Metrics for Open-domain Dialog Generation',
        '♯: An Enhancement Approach to Reference-based Evaluation Metrics for Open-domain Dialog Generation',
    ),
    ('<tex-math>k</tex-math>-arization of Synchronous ', 'k-arization of Synchronous '),
    ('<tex-math>N</tex-math>-Gram Translation', 'N-Gram Translation'),
    ('<tex-math>k</tex-math>-best ', 'k-best '),
    (
        '<tex-math>n</tex-math>-grams – Investigating Abstraction and Domain Dependence',
        'n-grams – Investigating Abstraction and Domain Dependence',
    ),
    (
        '<tex-math>N</tex-math>-gram Fragment Sequence Based Unsupervised Domain-Specific Document Readability',
        'N-gram Fragment Sequence Based Unsupervised Domain-Specific Document Readability',
    ),
    ('<tex-math>n</tex-math>-grams', 'n-grams'),
    ('<tex-math>O(M(n^2))</tex-math> Time', 'O(M(n2)) Time'),
    ('<tex-math>L_0</tex-math>-norm', 'L0-norm'),
    ('<tex-math>n</tex-math>-gram embedding', 'n-gram embedding'),
    ('<tex-math>N</tex-math>-best List Re-ranking', 'N-best List Re-ranking'),
    ('<tex-math>O(n^6)</tex-math> ', 'O(n6) '),
    (
        '<tex-math>^{\\circ}</tex-math>: A Referring Expression Recognition Dataset in 360',
        '∘: A Referring Expression Recognition Dataset in 360',
    ),
    ('<tex-math>^{\\circ}</tex-math> Images', '∘ Images'),
    ('<tex-math>k</tex-math>', 'k'),
    (
        '<tex-math>k</tex-math>-Fold Ensemble for Out-Of-Distribution Detection',
        'k-Fold Ensemble for Out-Of-Distribution Detection',
    ),
    ('<tex-math>Q^{2}</tex-math>: ', 'Q2: '),
    ('<tex-math>l_{0}</tex-math>-norm-based Alignment', 'l0-norm-based Alignment'),
    ('<tex-math>\\tau</tex-math> Maximization', '𝜏 Maximization'),
    (
        '<tex-math>\\ell_1</tex-math>-Norm Symmetric Nonnegative Matrix Factorization',
        'ℓ1-Norm Symmetric Nonnegative Matrix Factorization',
    ),
    (
        '<tex-math>{\\mathcal{P}^2}</tex-math>: A Plan-and-Pretrain Approach for Knowledge Graph-to-Text Generation',
        '𝒫2: A Plan-and-Pretrain Approach for Knowledge Graph-to-Text Generation',
    ),
    (
        '<tex-math>^2</tex-math> Learning: Actively reducing redundancies in Active Learning methods for Sequence Tagging and Machine Translation',
        '2 Learning: Actively reducing redundancies in Active Learning methods for Sequence Tagging and Machine Translation',
    ),
    ('<tex-math>\\ell_{1}</tex-math> Norm Optimisation', 'ℓ1 Norm Optimisation'),
    (
        '<tex-math>S^3</tex-math> - Statistical Sandhi Splitting',
        'S3 - Statistical Sandhi Splitting',
    ),
    ('<tex-math>\\epsilon</tex-math>-extension Hidden ', '𝜖-extension Hidden '),
    (
        '<tex-math>\\varepsilon</tex-math>-Skip Discriminating-Reverse Parsing on Graph-Structured Stack',
        '𝜀-Skip Discriminating-Reverse Parsing on Graph-Structured Stack',
    ),
    (
        '<tex-math>F^2</tex-math> - New Technique for Recognition of User Emotional States in Spoken Dialogue Systems',
        'F2 - New Technique for Recognition of User Emotional States in Spoken Dialogue Systems',
    ),
    (
        '<tex-math>0(n^6)</tex-math> Recognition Algorithm for Mildly Context-Sensitive Languages',
        '0(n6) Recognition Algorithm for Mildly Context-Sensitive Languages',
    ),
    ('<tex-math>\\lambda</tex-math>-', '𝜆-'),
    ('<tex-math>\\leftrightarrow</tex-math> ', '↔ '),
    ('<tex-math>\\Phi</tex-math>', '𝛷'),
)

test_cases_html = (
    (
        '<tex-math>K</tex-math>-Embeddings: Learning Conceptual Embeddings for Words using Context',
        '<span class="tex-math">K</span>-Embeddings: Learning Conceptual Embeddings for Words using Context',
    ),
    (
        '<tex-math>^{\\mathcal{E}}</tex-math>: a Vectorial Resource for Computing Conceptual Similarity',
        '<span class="tex-math"><sup>ℰ</sup></span>: a Vectorial Resource for Computing Conceptual Similarity',
    ),
    (
        '<tex-math>\\sharp</tex-math>: An Enhancement Approach to Reference-based Evaluation Metrics for Open-domain Dialog Generation',
        '<span class="tex-math">♯</span>: An Enhancement Approach to Reference-based Evaluation Metrics for Open-domain Dialog Generation',
    ),
    (
        '<tex-math>k</tex-math>-arization of Synchronous ',
        '<span class="tex-math">k</span>-arization of Synchronous ',
    ),
    (
        '<tex-math>N</tex-math>-Gram Translation',
        '<span class="tex-math">N</span>-Gram Translation',
    ),
    ('<tex-math>k</tex-math>-best ', '<span class="tex-math">k</span>-best '),
    (
        '<tex-math>n</tex-math>-grams – Investigating Abstraction and Domain Dependence',
        '<span class="tex-math">n</span>-grams – Investigating Abstraction and Domain Dependence',
    ),
    (
        '<tex-math>N</tex-math>-gram Fragment Sequence Based Unsupervised Domain-Specific Document Readability',
        '<span class="tex-math">N</span>-gram Fragment Sequence Based Unsupervised Domain-Specific Document Readability',
    ),
    ('<tex-math>n</tex-math>-grams', '<span class="tex-math">n</span>-grams'),
    (
        '<tex-math>O(M(n^2))</tex-math> Time',
        '<span class="tex-math">O(M(n<sup>2</sup>))</span> Time',
    ),
    ('<tex-math>L_0</tex-math>-norm', '<span class="tex-math">L<sub>0</sub></span>-norm'),
    (
        '<tex-math>n</tex-math>-gram embedding',
        '<span class="tex-math">n</span>-gram embedding',
    ),
    (
        '<tex-math>N</tex-math>-best List Re-ranking',
        '<span class="tex-math">N</span>-best List Re-ranking',
    ),
    ('<tex-math>O(n^6)</tex-math> ', '<span class="tex-math">O(n<sup>6</sup>)</span> '),
    (
        '<tex-math>^{\\circ}</tex-math>: A Referring Expression Recognition Dataset in 360',
        '<span class="tex-math"><sup>∘</sup></span>: A Referring Expression Recognition Dataset in 360',
    ),
    (
        '<tex-math>^{\\circ}</tex-math> Images',
        '<span class="tex-math"><sup>∘</sup></span> Images',
    ),
    ('<tex-math>k</tex-math>', '<span class="tex-math">k</span>'),
    (
        '<tex-math>k</tex-math>-Fold Ensemble for Out-Of-Distribution Detection',
        '<span class="tex-math">k</span>-Fold Ensemble for Out-Of-Distribution Detection',
    ),
    ('<tex-math>Q^{2}</tex-math>: ', '<span class="tex-math">Q<sup>2</sup></span>: '),
    (
        '<tex-math>l_{0}</tex-math>-norm-based Alignment',
        '<span class="tex-math">l<sub>0</sub></span>-norm-based Alignment',
    ),
    (
        '<tex-math>\\tau</tex-math> Maximization',
        '<span class="tex-math">𝜏</span> Maximization',
    ),
    (
        '<tex-math>\\ell_1</tex-math>-Norm Symmetric Nonnegative Matrix Factorization',
        '<span class="tex-math">ℓ<sub>1</sub></span>-Norm Symmetric Nonnegative Matrix Factorization',
    ),
    (
        '<tex-math>{\\mathcal{P}^2}</tex-math>: A Plan-and-Pretrain Approach for Knowledge Graph-to-Text Generation',
        '<span class="tex-math">𝒫<sup>2</sup></span>: A Plan-and-Pretrain Approach for Knowledge Graph-to-Text Generation',
    ),
    (
        '<tex-math>^2</tex-math> Learning: Actively reducing redundancies in Active Learning methods for Sequence Tagging and Machine Translation',
        '<span class="tex-math"><sup>2</sup></span> Learning: Actively reducing redundancies in Active Learning methods for Sequence Tagging and Machine Translation',
    ),
    (
        '<tex-math>\\ell_{1}</tex-math> Norm Optimisation',
        '<span class="tex-math">ℓ<sub>1</sub></span> Norm Optimisation',
    ),
    (
        '<tex-math>S^3</tex-math> - Statistical Sandhi Splitting',
        '<span class="tex-math">S<sup>3</sup></span> - Statistical Sandhi Splitting',
    ),
    (
        '<tex-math>\\epsilon</tex-math>-extension Hidden ',
        '<span class="tex-math">𝜖</span>-extension Hidden ',
    ),
    (
        '<tex-math>\\varepsilon</tex-math>-Skip Discriminating-Reverse Parsing on Graph-Structured Stack',
        '<span class="tex-math">𝜀</span>-Skip Discriminating-Reverse Parsing on Graph-Structured Stack',
    ),
    (
        '<tex-math>F^2</tex-math> - New Technique for Recognition of User Emotional States in Spoken Dialogue Systems',
        '<span class="tex-math">F<sup>2</sup></span> - New Technique for Recognition of User Emotional States in Spoken Dialogue Systems',
    ),
    (
        '<tex-math>0(n^6)</tex-math> Recognition Algorithm for Mildly Context-Sensitive Languages',
        '<span class="tex-math">0(n<sup>6</sup>)</span> Recognition Algorithm for Mildly Context-Sensitive Languages',
    ),
    ('<tex-math>\\lambda</tex-math>-', '<span class="tex-math">𝜆</span>-'),
    ('<tex-math>\\leftrightarrow</tex-math> ', '<span class="tex-math">↔</span> '),
    ('<tex-math>\\Phi</tex-math>', '<span class="tex-math">𝛷</span>'),
    # Manually collected tests
    (
        '<tex-math>0(n^{\\tilde{\\rho}+1})</tex-math>',
        '<span class="tex-math">0(n<sup> ̃𝜌+1</sup>)</span>',
    ),
    (
        '<tex-math>\\{mt, src\\} \\rightarrow pe</tex-math>',
        '<span class="tex-math">{mt, src} → pe</span>',
    ),
    (
        '<tex-math>p(\\boldsymbol{y}|\\textrm{do}(\\boldsymbol{x}))</tex-math>',
        '<span class="tex-math">p(<strong>y</strong>|<span class="font-weight-normal">do</span>(<strong>x</strong>))</span>',
    ),
    ('<tex-math>{\\sim}3\\%</tex-math>', '<span class="tex-math">∼3%</span>'),
    (
        '<tex-math>O(\\log_2 n)</tex-math>',
        '<span class="tex-math">O(<span class="tex-math-function">log</span><sub>2</sub> n)</span>',
    ),
    (
        '<tex-math>\\mathbf{^2}</tex-math>',
        '<span class="tex-math"><strong><sup>2</sup></strong></span>',
    ),
    (
        '<tex-math>RoBERTa_{large}</tex-math>',
        '<span class="tex-math">RoBERTa<sub>large</sub></span>',
    ),
    (
        '<tex-math>RoBERTa_{\\rm large}</tex-math>',
        '<span class="tex-math">RoBERTa<sub> large</sub></span>',
    ),
    (
        '<tex-math>RoBERTa_{\\bf large}</tex-math>',
        '<span class="tex-math">RoBERTa<sub> large</sub></span>',
    ),
    ('<tex-math>\\ell_1</tex-math>', '<span class="tex-math">ℓ<sub>1</sub></span>'),
    (
        '<tex-math>n \\log_2 \\frac{m}{n} + o(m)</tex-math>',
        '<span class="tex-math">n <span class="tex-math-function">log</span><sub>2</sub> <sup>m</sup>⁄<sub>n</sub> + o(m)</span>',
    ),
    (
        '<tex-math>\textrm{Pr}(f_1^J/e^I_1)</tex-math>',
        '<span class="tex-math">\textrmPr(f<sub>1</sub><sup>J</sup>/e<sup>I</sup><sub>1</sub>)</span>',
    ),
    (
        '<tex-math>\\# \\$ \\% \\&amp; \\_ \\{ \\} \\| \\:</tex-math>',
        '<span class="tex-math"># $ % &amp; _ { } ‖  </span>',
    ),
    (
        '<tex-math>asd\\$asd</tex-math>',
        '<span class="tex-math">asd$asd</span>',
    ),
    (
        '<tex-math>2\\_3</tex-math>',
        '<span class="tex-math">2_3</span>',
    ),
    (
        '<tex-math>2_3</tex-math>',
        '<span class="tex-math">2<sub>3</sub></span>',
    ),
    (
        '<tex-math>foo_{\\textsc{bar}}</tex-math>',
        '<span class="tex-math">foo<sub><span style="font-variant: small-caps;">bar</span></sub></span>',
    ),
    (
        '<tex-math>foo^{\\texttt{bar}}</tex-math>',
        '<span class="tex-math">foo<sup><span class="text-monospace">bar</span></sup></span>',
    ),
)


@pytest.mark.parametrize('inp, out', test_cases_unicode)
def test_unicode(inp, out):
    element = etree.fromstring(f"<span>{inp}</span>")
    math_element = element.find(".//tex-math")
    actual_out = texmath.to_unicode(math_element)
    if math_element.tail:
        actual_out += math_element.tail
    assert actual_out == out


@pytest.mark.parametrize('inp, out', test_cases_html)
def test_html(inp, out):
    element = etree.fromstring(f"<span>{inp}</span>")
    math_element = element.find(".//tex-math")
    result = texmath.to_html(math_element)
    actual_out = etree.tostring(result, encoding="unicode")
    assert actual_out == out