def test_r_group_table(self):
# Define a simple table structure
table = [Sentence('R1 R2'), Sentence('1a CH3 C'), Sentence('1b Me Br')]
output = r_group.resolve_r_group_grid(table)
r_groups_list = r_group.separate_duplicate_r_groups(output)
# Test the first r_group pair
var1, value1, labels1 = r_groups_list[0][0].convert_to_tuple()
var2, value2, labels2 = r_groups_list[0][1].convert_to_tuple()
self.assertEqual(var1.text, 'R1')
self.assertEqual(value1.text, 'CH3')
self.assertEqual(labels1[0].text, '1a')
self.assertEqual(var2.text, 'R2')
self.assertEqual(value2.text, 'C')
self.assertEqual(labels2[0].text, '1a')
# Test the second r_group pair
var1, value1, labels1 = r_groups_list[1][0].convert_to_tuple()
var2, value2, labels2 = r_groups_list[1][1].convert_to_tuple()
self.assertEqual(var1.text, 'R1')
self.assertEqual(value1.text, 'Me')
self.assertEqual(labels1[0].text, '1b')
self.assertEqual(var2.text, 'R2')
self.assertEqual(value2.text, 'Br')
self.assertEqual(labels2[0].text, '1b')
def test_r_group_assignment(self):
"""
Test assignment of multiple lines
"""
sentences = [Sentence('R1 = R2 = H'), Sentence('R1 = R2 = Ac')]
out = []
for sentence in sentences:
r_groups = r_group.detect_r_group_from_sentence(sentence, indicator='=')
r_groups = r_group.standardize_values(r_groups)
# Resolving positional labels where possible for 'or' cases
r_groups = r_group.filter_repeated_labels(r_groups)
# Separate duplicate variables into separate lists
r_groups_list = r_group.separate_duplicate_r_groups(r_groups)
out.append(r_groups_list)
self.assertEqual(out[0][0][0].var.text, 'R1')
self.assertEqual(out[0][0][0].value.text, 'R2')
self.assertEqual(out[0][0][1].var.text, 'R2')
self.assertEqual(out[0][0][1].value.text, '[H]')
self.assertEqual(out[1][0][0].var.text, 'R1')
self.assertEqual(out[1][0][0].value.text, 'R2')
self.assertEqual(out[1][0][1].var.text, 'R2')
self.assertEqual(out[1][0][1].value.text, 'Ac')
def test_r_group_simple_table(self):
# Define a simple table structure
table = [Sentence('R'), Sentence('1a CH3'), Sentence('1b Me')]
output = r_group.resolve_r_group_grid(table)
var, value, labels = output[0].convert_to_tuple()
var2, value2, labels2 = output[1].convert_to_tuple()
# tuple_output = [ (var.text, value.text, labels.text) for var, value, labels in output[0].convert_to_tuple()]
self.assertEqual(var.text, 'R')
self.assertEqual(value.text, 'CH3')
self.assertEqual(labels[0].text, '1a')
self.assertEqual(var2.text, 'R')
self.assertEqual(value2.text, 'Me')
self.assertEqual(labels2[0].text, '1b')
def do_parse(self, input, expected):
s = Sentence(input)
log.debug(s)
log.debug(s.tagged_tokens)
result = next(mp_phrase.scan(s.tagged_tokens))[0]
log.debug(etree.tostring(result, pretty_print=True, encoding='unicode'))
self.assertEqual(expected, etree.tostring(result, encoding='unicode'))
def read_label(fig, label, whitelist=LABEL_WHITELIST):
""" Reads a label paragraph objects using ocr
:param numpy.ndarray img: Input unprocessedimage
:param Label label: Label object containing appropriate bounding box
:rtype List[List[str]]
"""
size = 5
img = convert_greyscale(fig.img)
cropped_img = crop(img, label.left, label.right, label.top, label.bottom)
padded_img = pad(cropped_img, size, mode='constant', constant_values=(1, 1))
text = get_text(padded_img, x_offset=label.left, y_offset=label.top, psm=PSM.SINGLE_BLOCK, whitelist=whitelist)
if not text:
label.text = []
return label, 0
raw_sentences = get_sentences(text)
if len(raw_sentences) is not 0:
# Tag each sentence
tagged_sentences = [Sentence(sentence, word_tokenizer=ChemSchematicResolverTokeniser(),
parsers=[LabelParser()]) for sentence in raw_sentences]
else:
tagged_sentences = []
label.text = tagged_sentences
# Calculating average confidence for the block
confidences = [t.confidence for t in text]
avg_conf = np.mean(confidences)
log.info('Confidence in OCR: %s' % avg_conf)
return label, avg_conf
def do_parse(self, input, expected):
s = Sentence(input)
results = []
for result, *_ in mc_phrase.scan(s.tagged_tokens):
results.append(etree.tostring(result, encoding='unicode'))
log.debug(etree.tostring(result, pretty_print=True, encoding='unicode'))
self.assertEqual(expected, results)
def test_duplicate_r_group_vars_in_one_sentence(self):
sent = Sentence('A R1=H R2=NH B R1=H R2=C')
# sent = Sentence(text=[Token('A', 0, 1), Token('R1', 2, 3), Token('=', 4, 5), Token('H', 6, 7),
# Token('R2', 8, 9), Token('=', 10, 11), Token('NH', 12, 13),
# Token('B', 14, 15), Token('R1', 16, 17), Token('=', 18, 19), Token('H', 20, 21),
# Token('R2', 21, 22), Token('=', 23, 24), Token('H', 25, 26)],
# start=0,
# end=26,
# sentence_tokenizer=ChemSentenceTokenizer(),
# word_tokenizer=ChemWordTokenizer(),
# lexicon=ChemLexicon(),
# abbreviation_detector=ChemAbbreviationDetector(),
# pos_tagger=ChemCrfPosTagger(), # ChemPerceptronTagger()
# ner_tagger=CemTagger()
# )
var_value_pairs = r_group.detect_r_group_from_sentence(sent)
r_groups = r_group.get_label_candidates(sent, var_value_pairs)
r_groups = r_group.standardize_values(r_groups)
# Resolving positional labels where possible for 'or' cases
r_groups = r_group.filter_repeated_labels(r_groups)
# Separate duplicate variables into separate lists
r_groups_list = r_group.separate_duplicate_r_groups(r_groups)
output = []
for r_groups in r_groups_list:
output.append(r_group.convert_r_groups_to_tuples(r_groups))
def do_parse(self, input, expected):
s = Sentence(input)
log.debug(s)
log.debug(s.tagged_tokens)
result = [c.serialize() for c in TemParser().parse(s.tagged_tokens)]
if len(result) == 0:
result = ['']
self.assertEqual(expected, result[0])
def do_parse(self, input, expected):
s = Sentence(input)
log.debug(s)
log.debug(s.tagged_tokens)
result = next(ir.scan(s.tagged_tokens))[0]
log.debug(etree.tostring(result, pretty_print=True, encoding='unicode'))
self.assertEqual(expected, etree.tostring(result, encoding='unicode'))
for c in IrParser().parse(s.tagged_tokens):
print(c.serialize())
def do_parse(self, input, expected):
s = Sentence(input)
log.debug(s)
log.debug(s.tagged_tokens)
results = []
for i, r in enumerate(chemical_label_phrase.scan(s.tagged_tokens)):
log.debug(etree.tostring(r[0], pretty_print=True, encoding='unicode'))
results.append(etree.tostring(r[0], encoding='unicode'))
self.assertEqual(expected, results)
def test_mp1(self):
# Declaration
s = Sentence(
'Colorless solid (81% yield, 74.8 mg, 0.22 mmol); mp 77.2–77.5 °C.'
)
expected = '<mp_phrase><mp><value>77.2–77.5</value><units>°C</units></mp></mp_phrase>'
# Testing
result = next(mp_phrase.scan(s.tagged_tokens))[0]
#Assertion
self.assertEqual(expected, etree.tostring(result, encoding='unicode'))
def test_label_parsing(self):
test_sentence = Sentence('3', parsers=[LabelParser()])
self.assertEqual(test_sentence.records.serialize(), [{
'labels': ['3']
}])