caret / packages / core / src / parser.ts

import { EditorState } from "./editorState";
import { CaretNode } from "./node";
import { MissingNode } from "./nodes/missing";
import { UnparseableNode } from "./nodes/unparseable";
import {
  CaretJuxtapositionParselet,
  CaretInfixParselet,
  CaretLeafParselet,
  CaretParselet,
  CaretPostfixParselet,
  CaretPrefixParselet,
  CannotParseError,
} from "./parselet";
import { StrandPath } from "./path";
import { Strand } from "./strand";
import { Token } from "./token";

export class CaretParser {
  private _position: number = 0;
  private _editorState: EditorState;
  private _strandPath: StrandPath;
  private _parselets: CaretParselet[];
  private _parseletBindingPower: Map<CaretParselet, number> = new Map();
  // public activeFlags: Set<Symbol> = new Set();

  constructor(
    editorState: EditorState,
    strandPath: StrandPath = [],
    parselets: (CaretParselet | Set<CaretParselet>)[]
    // activeFlags: Set<Symbol> = new Set()
  ) {
    this._editorState = editorState;
    this._strandPath = strandPath;
    // this.activeFlags = activeFlags;
    this._parselets = parselets.flatMap((p) =>
      p instanceof Set ? Array.from(p) : p
    );
    for (let i = 0; i < parselets.length; i++) {
      const parseletOrSet = parselets[i];
      const bindingPower = parselets.length - i;
      if (parseletOrSet instanceof Set) {
        for (const parselet of parseletOrSet) {
          this._parseletBindingPower.set(parselet, bindingPower);
        }
      } else {
        this._parseletBindingPower.set(parseletOrSet, bindingPower);
      }
    }
  }

  public get editorState() {
    return this._editorState;
  }

  public get strand(): Strand {
    const strand = this._editorState.content.getStrand(this._strandPath);
    if (!strand) {
      throw new Error(
        `Strand not found at path: ${JSON.stringify(this._strandPath)}`
      );
    }
    return strand;
  }

  public parse() {
    return this.parseMinBP(0);
  }

  public parseMinBP(minBindingPower: number | CaretParselet): CaretNode {
    if (typeof minBindingPower !== "number") {
      minBindingPower = this._parseletBindingPower.get(minBindingPower) ?? 0;
    }

    if (this.done()) return new MissingNode();

    const parselets = this.filterParselets(["leaf", "prefix"]);
    let resultNode: CaretNode | undefined;
    for (const parselet of parselets) {
      const positionBefore = this._position;
      try {
        resultNode = parselet.parse(this);
        break;
      } catch (error) {
        if (error instanceof CannotParseError) {
          this._position = positionBefore;
          continue;
        }
        throw error;
      }
    }
    if (!resultNode) {
      const unparsedTokens = this.strand.tokens.slice(this._position);
      this._position = this.strand.tokens.length;
      return UnparseableNode.from(null, unparsedTokens);
    }

    let left: CaretNode = resultNode;

    while (!this.done()) {
      let resultNode: CaretNode | undefined;

      {
        const parselets = this.filterParselets(["infix", "postfix"]);
        for (const parselet of parselets) {
          const bindingPower = this.getBindingPower(parselet);
          if (bindingPower < minBindingPower) continue;
          if (
            bindingPower === minBindingPower &&
            parselet instanceof CaretInfixParselet &&
            parselet.associativity === "left"
          ) {
            continue;
          }

          const positionBefore = this._position;
          try {
            resultNode = parselet.parse(this, left);
            break;
          } catch (error) {
            if (error instanceof CannotParseError) {
              this._position = positionBefore;
              continue;
            }
            throw error;
          }
        }
      }

      if (resultNode) {
        left = resultNode;
        continue;
      }

      {
        const parselets = this.filterParselets(["juxtaposition"]);
        for (const parselet of parselets) {
          const bindingPower = this.getBindingPower(parselet);
          if (bindingPower < minBindingPower) continue;

          const positionBefore = this._position;
          try {
            const right = this.parseMinBP(bindingPower);
            if (
              !(left instanceof UnparseableNode) &&
              !(right instanceof UnparseableNode) &&
              parselet.canParse(left, right)
            ) {
              resultNode = parselet.parse(this, left, right);
              break;
            } else {
              this._position = positionBefore;
            }
          } catch (error) {
            if (error instanceof CannotParseError) {
              this._position = positionBefore;
              continue;
            }
            throw error;
          }
        }
      }

      if (resultNode) {
        left = resultNode;
        continue;
      }

      // No parselets matched; stop parsing here
      break;
    }

    // Sometimes ending early is okay (if we are sub-parsing), but if we're at the root
    // level and there are still tokens left, we should return an UnparseableNode
    if (this.peek() !== null && minBindingPower === 0) {
      return UnparseableNode.from(
        resultNode,
        this.strand.tokens.slice(this._position)
      );
    }

    return left;
  }

  public parseSubStrand(
    strandOrPath: StrandPath | Strand
    // activeFlags: Set<Symbol> = this.activeFlags
  ): CaretNode {
    const path =
      strandOrPath instanceof Strand
        ? this.editorState.content.findStrandPath(strandOrPath)
        : strandOrPath;

    if (!path) {
      throw new Error(
        `Strand not found in editorState: ${JSON.stringify(strandOrPath)}`
      );
    }

    let tokenAtPath: Token | null = null;

    if (path.length > 0) {
      tokenAtPath = this.editorState.content.getToken({
        strandPath: path.slice(0, -1),
        tokenIndex: path[path.length - 1].tokenIndex,
      });
    }

    const subParser = new CaretParser(
      this._editorState,
      path,
      this._parselets
      // activeFlags
    );
    return subParser.parse();
  }

  public peek(): Token | null {
    if (this._position < this.strand.tokens.length) {
      return this.strand.tokens[this._position];
    }
    return null;
  }

  public consume(): Token | null {
    if (this._position < this.strand.tokens.length) {
      return this.strand.tokens[this._position++];
    }
    return null;
  }

  public done(): boolean {
    return this._position >= this.strand.tokens.length;
  }

  private filterParselets<
    T extends "leaf" | "prefix" | "infix" | "postfix" | "juxtaposition"
  >(validTypes: T[]): ParseletType<T>[] {
    return this._parselets.filter((parselet) =>
      validTypes.some((typeName) => {
        switch (typeName) {
          case "leaf":
            return parselet instanceof CaretLeafParselet;
          case "prefix":
            return parselet instanceof CaretPrefixParselet;
          case "infix":
            return parselet instanceof CaretInfixParselet;
          case "postfix":
            return parselet instanceof CaretPostfixParselet;
          case "juxtaposition":
            return parselet instanceof CaretJuxtapositionParselet;
          default:
            typeName satisfies never;
            return false;
        }
      })
    ) as ParseletType<T>[];
  }

  private getBindingPower(parselet: CaretParselet): number {
    const bindingPower = this._parseletBindingPower.get(parselet);
    if (bindingPower === undefined) {
      throw new Error(`Parselet not registered: ${parselet.constructor.name}`);
    }
    return bindingPower;
  }
}

type ParseletType<
  T extends "leaf" | "prefix" | "infix" | "postfix" | "juxtaposition"
> =
  | (T extends "leaf" ? CaretLeafParselet : never)
  | (T extends "prefix" ? CaretPrefixParselet : never)
  | (T extends "infix" ? CaretInfixParselet : never)
  | (T extends "postfix" ? CaretPostfixParselet : never)
  | (T extends "juxtaposition" ? CaretJuxtapositionParselet : never);