@wgslx/wgslx/src/sequence.ts

Extended WebGPU shading language tools

import {TextMatcher} from './patterns';
import {preprocess} from './preprocess';

/** Global match regex for all blankspace. */
const BLANKSPACE_GLOBAL_REGEX =
  /[\u0020\u0009\u000a\u000b\u000c\u000d\u0085\u200e\u200f\u2028\u2029]+/g;

/** Line break regex. */
const LINE_BREAK_REGEX =
  /(?:\u000d\u000a?|[\u000a\u000b\u000c\u0085\u2028\u2029])/;

export interface Segment {
  /** The text contained in the segment. */
  text: string;

  /** The file that the segment belongs to. */
  file: string;

  /** The line number of the segment. */
  line: number;

  /** The beginning column number of the text. */
  column: number;
}

/** Cursor pointing to a specific location in a code sequence. */
export interface Cursor {
  readonly segment: number;
  readonly offset: number;
}

export function Cursor(segment: number, start = 0): Cursor {
  return {segment, offset: start};
}

export function compareCursor(a: Cursor, b: Cursor): number {
  if (a.segment === b.segment) {
    return a.offset - b.offset;
  }

  return a.segment - b.segment;
}

export function cursorGreaterThan(candidate: Cursor, current: Cursor): boolean {
  if (candidate.segment > current.segment) {
    return true;
  }

  if (candidate.segment === current.segment) {
    return candidate.offset > current.offset;
  }

  return false;
}

export function cursorGreaterOrEqualThan(
  candidate: Cursor,
  current: Cursor
): boolean {
  if (candidate.segment >= current.segment) {
    return true;
  }

  if (candidate.segment === current.segment) {
    return candidate.offset >= current.offset;
  }

  return false;
}

export interface SourceCursor {
  readonly line: number;
  readonly column: number;
}

/** Match in a code sequence containing the match and the advanced cursor. */
export interface TextMatch {
  /** The text of the match. */
  readonly text: string;

  /** The segment number of the match. */
  readonly segment: number;

  /** The starting position of the match inside the sequence. */
  readonly offset: number;

  /** The ending position of the match inside the sequence. */
  readonly end: number;

  /** The next cursor. */
  readonly cursor: Cursor;
}

/**
 * Code sequence for parsing. The segments are separated by whitespace and comments are removed.
 * Matching tokens never crosses whitespace delimiters in WGSL and this will allow us to inject
 * code from multiple files.
 */
export class Sequence {
  /** List of segments that make up the sequence. */
  readonly segments: Segment[];

  stringify(cursor: Cursor): string {
    if (cursor.segment >= this.segments.length) {
      return 'eof';
    }

    const segment = this.segments[cursor.segment];
    return `${segment.line}:${segment.column + cursor.offset}:${segment.file}`;
  }

  toSourceCursor(cursor: Cursor): SourceCursor {
    // End of file.
    if (cursor.segment === this.segments.length && cursor.offset === 0) {
      return {
        line: this.segments[this.segments.length - 1].line + 1,
        column: this.segments[this.segments.length - 1].column + 1,
      };
    }

    if (cursor.segment >= this.segments.length) {
      return {line: -1, column: -1};
    }

    const segment = this.segments[cursor.segment];
    return {line: segment.line + 1, column: segment.column + cursor.offset + 1};
  }

  /** Tries to match the text matcher at the given position. */
  match(cursor: Cursor, matcher: TextMatcher): TextMatch | null {
    if (cursor.segment >= this.segments.length) {
      return null;
    }

    const segment = this.segments[cursor.segment];
    const match = matcher(segment.text, cursor.offset);

    // Match not found
    if (!match) {
      return null;
    }

    let next: Cursor;

    if (cursor.offset + match.length === segment.text.length) {
      // Match found and takes the rest of the segment.
      next = {
        segment: cursor.segment + 1,
        offset: 0,
      };
    } else {
      // Match was found and there is more in this segment.
      next = {
        segment: cursor.segment,
        offset: cursor.offset + match.length,
      };
    }

    return {
      text: match,
      segment: cursor.segment,
      offset: cursor.offset,
      end: cursor.offset + match.length,

      cursor: next,
    };
  }

  /** Creates a sequence from the specified file text and file name. */
  static from(text: string, file: string): Sequence {
    text = preprocess(text);
    // Split by new lines.
    const lines = text.split(LINE_BREAK_REGEX);
    const segments: Segment[] = [];

    for (let l = 0; l < lines.length; l++) {
      const line = lines[l];
      const whitespaceMatches = line.matchAll(BLANKSPACE_GLOBAL_REGEX);
      let i = 0;

      for (const match of whitespaceMatches) {
        // Starts with whitespace, should only be at
        if (match.index === i) {
          i += match[0].length;
          continue;
        }

        // Add text sequence
        const segment = {
          text: line.substring(i, match.index),
          file,
          line: l,
          column: i,
        };

        if (segment.text.match(BLANKSPACE_GLOBAL_REGEX)) {
          throw new Error('Whitespace in text segment.');
        }

        segments.push(segment);
        i = match.index + match[0].length;
      }

      // Does not end in whitespace, add the remaining characters.
      if (i !== line.length) {
        segments.push({
          text: line.substring(i),
          file,
          line: l,
          column: i,
        });
      }
    }

    return new Sequence(segments);
  }

  protected constructor(segments: Segment[]) {
    this.segments = segments;
  }
}