annotations_parser.py

import os
import re


def parse_annotated_text(text):
    # Regular expression pattern to parse YEDDA format
    yedda_pattern = re.compile(r'(\[\@(.*?)\#(\w+)\*\])', re.DOTALL)
    
    # This variable will keep track of the number of characters removed
    chars_removed = 0

    # This will store the spans of the entities in the original text
    spans_in_original_text = []

    # Buffer to store content without annotations
    buffer = []

    # Index to track last end position
    last_end = 0

    # Store labels
    labels = []

    # Loop through each match
    for match in yedda_pattern.finditer(text):
        # The entire match
        full_match = match.group(0)
        # Capture group 2 (entity)
        entity = match.group(2)
        # Capture group 2 (label)
        label = match.group(3)
        # Start position of the match in the modified string
        start = match.start()
        # End position of the match in the modified string
        end = match.end()

        labels.append(label)

        # Append the text before the annotation to the buffer
        buffer.append(text[last_end:start])

        # Remove trailing spaces after entity
        entity = entity.rstrip()
        buffer.append(entity)

        # Calculate the start and end spans in the original text
        original_start = start - chars_removed
        original_end = original_start + len(entity)

        assert original_end > original_start, text
        
        # Store the spans
        spans_in_original_text.append((original_start, original_end))
        
        # update the chars_removed counter
        chars_removed += len(full_match) - len(entity)

        # Update last_end
        last_end = end

    # Append remaining content after the last match
    buffer.append(text[last_end:])

    # Join buffer parts to get content without annotations
    content_without_annotations = "".join(buffer)

    return {
        'text': content_without_annotations,
        'spans': spans_in_original_text,
        'labels': labels
    }


def preprocess_text(text: str):
    # remove extra spaces
    text = text.strip()
    text = re.sub(r'\n+', '\n', text)
    text = re.sub(r' +', ' ', text)
    text = text.replace(' \n', '\n')
    text = text.replace('\n ', '\n')
    return text


def load_yedda_annotations(directory):

    # List to store all the annotations from all files
    all_annotations = []
    
    # Iterate through each file in the given directory
    for filename in os.listdir(directory):
        # Check if the file has the '.ann' extension
        if filename.endswith(".ann"):
            # Construct the full file path
            file_path = os.path.join(directory, filename)
            
            # Open and read the file
            with open(file_path, 'r', encoding='utf-8') as file:
                content = file.read()
            
            # Preprocess text
            content = preprocess_text(content)
            
            parsed = parse_annotated_text(content)
            file_annotations = {
                'file': filename, 
                'annotated_text': content, 
                'text': parsed['text'],
                'spans': parsed['spans'],
                'labels': parsed['labels'],
            }
            all_annotations.append(file_annotations)
    
    return all_annotations


def convert_to_ann(annotatations):
    text = annotatations['text']
    buffer = []
    i = 0
    for (j_start, j_end), label in zip(annotatations['spans'], annotatations['labels']):

        buffer += text[i:j_start]
        buffer += [f'[@{text[j_start:j_end]}#{label}*]']
        i = j_end

    buffer += [text[i:]]

    return ''.join(buffer)


if __name__ == '__main__':
    
    directory_path = 'annotations'  # The directory containing .ann files
    annotations = load_yedda_annotations(directory_path)

    counter = 0
    for file_annotation in annotations:
        counter += len(file_annotation['labels'])
        print('File:', file_annotation['file'])
        print('Text[:100]:', repr(file_annotation['text'][:100]))
        print('Number of labels:', len(file_annotation['labels']))
        assert len(file_annotation['labels']) == len(file_annotation['spans'])
        print('Average labeled sentence length:', sum(end-start for start,end in file_annotation['spans']) / len(file_annotation['spans']))
        print('--------------------------------')
    
    print('Total number of files:', len(annotations))
    print('Total label count:', counter)