@krassowski/jupyterlab-lsp/lib/extractors/regexp.js

Language Server Protocol integration for JupyterLab

import { position_at_offset } from '../positioning';
export function getIndexOfCaptureGroup(expression, matched_string, value_of_captured_group) {
    // TODO: use https://github.com/tc39/proposal-regexp-match-indices once supported in >95% of browsers
    //  (probably around 2025)
    // get index of the part that is being extracted to foreign document
    let captured_groups = expression.exec(matched_string);
    if (captured_groups == null) {
        console.warn(`No capture group found for ${expression} in ${matched_string}`);
        return -1;
    }
    let offset_in_match = 0;
    // first element is a full match
    let full_matched = captured_groups[0];
    for (let group of captured_groups.slice(1)) {
        if (typeof group === 'undefined') {
            continue;
        }
        if (group === value_of_captured_group) {
            offset_in_match += full_matched.indexOf(group);
            break;
        }
        let group_end_offset = full_matched.indexOf(group) + group.length;
        full_matched = full_matched.slice(group_end_offset);
        offset_in_match += group_end_offset;
    }
    return offset_in_match;
}
export class RegExpForeignCodeExtractor {
    constructor(options) {
        this.language = options.language;
        this.options = options;
        this.global_expression = new RegExp(options.pattern, 'g');
        this.test_expression = new RegExp(options.pattern, 'g');
        this.expression = new RegExp(options.pattern);
        this.standalone = this.options.is_standalone;
        this.file_extension = this.options.file_extension;
    }
    has_foreign_code(code) {
        let result = this.test_expression.test(code);
        this.test_expression.lastIndex = 0;
        return result;
    }
    extract_foreign_code(code) {
        let lines = code.split('\n');
        let extracts = new Array();
        let started_from = this.global_expression.lastIndex;
        let match = this.global_expression.exec(code);
        let host_code_fragment;
        let chosen_replacer;
        let is_new_api_replacer = false;
        if (typeof this.options.foreign_replacer !== 'undefined') {
            chosen_replacer = this.options.foreign_replacer;
            is_new_api_replacer = true;
        }
        else if (typeof this.options.foreign_capture_groups !== 'undefined') {
            chosen_replacer = '$' + this.options.foreign_capture_groups.join('$');
            is_new_api_replacer = true;
        }
        else if (this.options.extract_to_foreign) {
            chosen_replacer = this.options.extract_to_foreign;
        }
        else {
            console.warn(`Foreign replacer not defined for extractor: {this.expression} - this is deprecated; use 'foreign_replacer' to define it`);
            return [];
        }
        while (match != null) {
            let matched_string = match[0];
            let position_shift = null;
            let foreign_code_fragment = matched_string.replace(this.expression, 
            // eslint-disable-next-line @typescript-eslint/ban-ts-comment
            // @ts-ignore
            chosen_replacer);
            let prefix = '';
            if (typeof this.options.extract_arguments !== 'undefined') {
                prefix = matched_string.replace(this.expression, 
                // eslint-disable-next-line @typescript-eslint/ban-ts-comment
                // @ts-ignore
                this.options.extract_arguments);
                position_shift = position_at_offset(prefix.length, prefix.split('\n'));
            }
            // NOTE:
            // match.index + matched_string.length === this.sticky_expression.lastIndex
            let end_index = this.global_expression.lastIndex;
            if (this.options.keep_in_host || this.options.keep_in_host == null) {
                host_code_fragment = code.substring(started_from, end_index);
            }
            else {
                if (started_from === match.index) {
                    host_code_fragment = '';
                }
                else {
                    host_code_fragment = code.substring(started_from, match.index) + '\n';
                }
            }
            let foreign_code_group_value = foreign_code_fragment;
            if (is_new_api_replacer) {
                foreign_code_group_value = matched_string.replace(this.expression, '$' + Math.min(...this.options.foreign_capture_groups));
            }
            const foreign_group_index_in_match = getIndexOfCaptureGroup(this.expression, matched_string, foreign_code_group_value);
            let start_offset = match.index + foreign_group_index_in_match;
            let start = position_at_offset(start_offset, lines);
            let end = position_at_offset(start_offset + foreign_code_fragment.length, lines);
            extracts.push({
                host_code: host_code_fragment,
                foreign_code: prefix + foreign_code_fragment,
                range: { start, end },
                virtual_shift: position_shift
            });
            started_from = this.global_expression.lastIndex;
            match = this.global_expression.exec(code);
        }
        if (started_from !== code.length) {
            let final_host_code_fragment = code.substring(started_from, code.length);
            extracts.push({
                host_code: final_host_code_fragment,
                foreign_code: null,
                range: null,
                virtual_shift: null
            });
        }
        return extracts;
    }
}
//# sourceMappingURL=regexp.js.map