@wdprlib/ast 2.0.0 → 2.2.0

package/src/expr-eval.ts

@@ -0,0 +1,586 @@
+/**
+ * Expression evaluator for Wikidot [[#expr]] and [[#ifexpr]]
+ *
+ * Supported:
+ * - Operators: +, -, *, /, % (modulo), ^ (power)
+ * - Comparison: <, >, <=, >=, =, !=
+ * - Logic: and, or, not
+ * - Functions: abs(), min(), max(), floor(), ceil(), round()
+ * - Parentheses for grouping
+ * - Negative numbers
+ *
+ * Expression limit: 256 characters (enforced by parser)
+ */
+
+/**
+ * Set of string values considered falsy by Wikidot's `#if` construct.
+ * Case-insensitive after lowercasing and trimming.
+ */
+const FALSE_VALUES = new Set(["false", "null", "", "0"]);
+
+/**
+ * Determine whether a string value is truthy for Wikidot's `#if` construct.
+ *
+ * The value is lowercased and trimmed before checking against the set of
+ * known falsy strings (`"false"`, `"null"`, `""`, `"0"`).
+ *
+ * @param value - The condition string to check.
+ * @returns `true` if the value is not in the falsy set.
+ */
+export function isTruthy(value: string): boolean {
+  return !FALSE_VALUES.has(value.toLowerCase().trim());
+}
+
+/** Maximum allowed expression length (enforced by the parser, checked here as a safety net). */
+const MAX_EXPRESSION_LENGTH = 256;
+
+/**
+ * Determine whether a number is truthy for logical operators (`and`, `or`, `not`).
+ *
+ * 0 and `NaN` are falsy; all other finite and infinite values are truthy.
+ *
+ * @param n - The number to check.
+ * @returns `true` if the number is non-zero and not `NaN`.
+ */
+function isTruthyNum(n: number): boolean {
+  return n !== 0 && !Number.isNaN(n);
+}
+
+/**
+ * Result of evaluating a mathematical expression.
+ * Either a successful numeric value or an error message string.
+ */
+export type ExprResult = { success: true; value: number } | { success: false; error: string };
+
+/**
+ * Format a numeric expression result for display.
+ *
+ * Uses JavaScript's default `String(n)` so the full precision of the
+ * computed value is preserved (e.g. `1/3` becomes `"0.3333333333333333"`,
+ * matching the `Number` → `String` conversion rather than truncating to
+ * a fixed number of decimals). Used by both the inline renderer and the
+ * opener preprocess so the same expression produces the same string
+ * regardless of where it appears in the source.
+ */
+export function formatExprValue(n: number): string {
+  return String(n);
+}
+
+/**
+ * Evaluate a mathematical expression string and return the result.
+ *
+ * The expression is tokenized, parsed with a recursive descent parser,
+ * and evaluated in a single pass. Errors produce Wikidot-compatible
+ * messages (e.g., `"division by zero"`, `"too many values in the stack"`).
+ *
+ * NaN and Infinity results are treated as division-by-zero errors.
+ *
+ * @param expr - The expression string to evaluate.
+ * @returns A success result with a numeric value, or an error result with a message.
+ */
+export function evaluateExpression(expr: string): ExprResult {
+  try {
+    // Enforce length limit
+    if (expr.length > MAX_EXPRESSION_LENGTH) {
+      return { success: false, error: "expression too long" };
+    }
+    if (expr.trim() === "") {
+      return { success: false, error: "empty expression" };
+    }
+    const tokens = tokenize(expr);
+    if (tokens.length <= 1) {
+      // Only EOF token
+      return { success: false, error: "empty expression" };
+    }
+    const parser = new ExprParser(tokens);
+    const result = parser.parse();
+    // Treat NaN and Infinity as division by zero (Wikidot-compatible)
+    if (!Number.isFinite(result)) {
+      return { success: false, error: "division by zero" };
+    }
+    return { success: true, value: result };
+  } catch (e) {
+    const msg = e instanceof Error ? e.message : "unknown error";
+    return { success: false, error: msg };
+  }
+}
+
+/** Discriminant for expression tokens. */
+type TokenKind =
+  | "NUMBER"
+  | "IDENTIFIER"
+  | "PLUS"
+  | "MINUS"
+  | "STAR"
+  | "SLASH"
+  | "PERCENT"
+  | "CARET"
+  | "LPAREN"
+  | "RPAREN"
+  | "COMMA"
+  | "LT"
+  | "GT"
+  | "LE"
+  | "GE"
+  | "EQ"
+  | "NE"
+  | "BANG"
+  | "EOF";
+
+/** A single token produced by the expression tokenizer. */
+interface ExprToken {
+  /** The type of this token. */
+  kind: TokenKind;
+  /** Numeric value for NUMBER tokens; string representation for all others. */
+  value: string | number;
+}
+
+/**
+ * Tokenize a mathematical expression string into a sequence of tokens.
+ *
+ * Handles numbers (including decimals), identifiers (function names and
+ * keywords like `and`/`or`/`not`), two-character operators (`<=`, `>=`,
+ * `!=`, `<>`), and single-character operators. Unknown characters throw
+ * an error.
+ *
+ * @param expr - The expression string to tokenize.
+ * @returns Array of tokens, always ending with an EOF token.
+ * @throws {Error} On encountering an unknown character or invalid number.
+ */
+function tokenize(expr: string): ExprToken[] {
+  const tokens: ExprToken[] = [];
+  let i = 0;
+
+  while (i < expr.length) {
+    const ch = expr[i]!;
+
+    // Skip whitespace
+    if (/\s/.test(ch)) {
+      i++;
+      continue;
+    }
+
+    // Number (including decimals)
+    if (/\d/.test(ch) || (ch === "." && /\d/.test(expr[i + 1] ?? ""))) {
+      let numStr = "";
+      let hasDot = false;
+      while (i < expr.length) {
+        const c = expr[i]!;
+        if (c === ".") {
+          if (hasDot) break; // Only one decimal point allowed
+          hasDot = true;
+        } else if (!/\d/.test(c)) {
+          break;
+        }
+        numStr += c;
+        i++;
+      }
+      const num = parseFloat(numStr);
+      if (!Number.isFinite(num)) {
+        throw new Error("Invalid number");
+      }
+      tokens.push({ kind: "NUMBER", value: num });
+      continue;
+    }
+
+    // Identifier (function names, keywords like and/or/not)
+    if (/[a-zA-Z_]/.test(ch)) {
+      let id = "";
+      while (i < expr.length) {
+        const c = expr[i]!;
+        if (!/[a-zA-Z0-9_]/.test(c)) break;
+        id += c;
+        i++;
+      }
+      tokens.push({ kind: "IDENTIFIER", value: id.toLowerCase() });
+      continue;
+    }
+
+    // Two-character operators
+    if (ch === "<" && expr[i + 1] === "=") {
+      tokens.push({ kind: "LE", value: "<=" });
+      i += 2;
+      continue;
+    }
+    if (ch === ">" && expr[i + 1] === "=") {
+      tokens.push({ kind: "GE", value: ">=" });
+      i += 2;
+      continue;
+    }
+    if (ch === "!" && expr[i + 1] === "=") {
+      tokens.push({ kind: "NE", value: "!=" });
+      i += 2;
+      continue;
+    }
+    if (ch === "<" && expr[i + 1] === ">") {
+      tokens.push({ kind: "NE", value: "<>" });
+      i += 2;
+      continue;
+    }
+
+    // Unary logical NOT (Wikidot's `!` operator). The two-character `!=`
+    // has already been handled above, so a bare `!` here is always
+    // unary not.
+    if (ch === "!") {
+      tokens.push({ kind: "BANG", value: "!" });
+      i++;
+      continue;
+    }
+
+    // Single-character operators
+    switch (ch) {
+      case "+":
+        tokens.push({ kind: "PLUS", value: "+" });
+        break;
+      case "-":
+        tokens.push({ kind: "MINUS", value: "-" });
+        break;
+      case "*":
+        tokens.push({ kind: "STAR", value: "*" });
+        break;
+      case "/":
+        tokens.push({ kind: "SLASH", value: "/" });
+        break;
+      case "%":
+        tokens.push({ kind: "PERCENT", value: "%" });
+        break;
+      case "^":
+        tokens.push({ kind: "CARET", value: "^" });
+        break;
+      case "(":
+        tokens.push({ kind: "LPAREN", value: "(" });
+        break;
+      case ")":
+        tokens.push({ kind: "RPAREN", value: ")" });
+        break;
+      case ",":
+        tokens.push({ kind: "COMMA", value: "," });
+        break;
+      case "<":
+        tokens.push({ kind: "LT", value: "<" });
+        break;
+      case ">":
+        tokens.push({ kind: "GT", value: ">" });
+        break;
+      case "=":
+        tokens.push({ kind: "EQ", value: "=" });
+        break;
+      default:
+        // Unknown character is an error
+        throw new Error(`Unknown character: ${ch}`);
+    }
+    i++;
+  }
+
+  tokens.push({ kind: "EOF", value: "" });
+  return tokens;
+}
+
+/**
+ * Recursive descent parser for expressions
+ * Precedence (low to high):
+ *   or
+ *   and
+ *   not (unary)
+ *   comparison (<, >, <=, >=, =, !=)
+ *   addition (+, -)
+ *   multiplication (*, /, %)
+ *   power (^)
+ *   unary (-, +)
+ *   primary (number, parentheses, function call)
+ */
+class ExprParser {
+  private pos = 0;
+
+  constructor(private tokens: ExprToken[]) {}
+
+  parse(): number {
+    const result = this.parseOr();
+    if (this.current().kind !== "EOF") {
+      // Wikidot-compatible error message when extra values remain
+      throw new Error("too many values in the stack");
+    }
+    return result;
+  }
+
+  private current(): ExprToken {
+    return this.tokens[this.pos] ?? { kind: "EOF", value: "" };
+  }
+
+  private advance(): ExprToken {
+    const token = this.current();
+    this.pos++;
+    return token;
+  }
+
+  private parseOr(): number {
+    let left = this.parseAnd();
+
+    while (this.current().kind === "IDENTIFIER" && this.current().value === "or") {
+      this.advance();
+      const right = this.parseAnd();
+      // Treat 0 and NaN as falsy
+      left = isTruthyNum(left) || isTruthyNum(right) ? 1 : 0;
+    }
+
+    return left;
+  }
+
+  private parseAnd(): number {
+    let left = this.parseNot();
+
+    while (this.current().kind === "IDENTIFIER" && this.current().value === "and") {
+      this.advance();
+      const right = this.parseNot();
+      // Treat 0 and NaN as falsy
+      left = isTruthyNum(left) && isTruthyNum(right) ? 1 : 0;
+    }
+
+    return left;
+  }
+
+  private parseNot(): number {
+    const cur = this.current();
+    if ((cur.kind === "IDENTIFIER" && cur.value === "not") || cur.kind === "BANG") {
+      this.advance();
+      const value = this.parseNot();
+      // Treat 0 and NaN as falsy
+      return isTruthyNum(value) ? 0 : 1;
+    }
+    return this.parseComparison();
+  }
+
+  private parseComparison(): number {
+    let left = this.parseAddition();
+
+    const kind = this.current().kind;
+    if (
+      kind === "LT" ||
+      kind === "GT" ||
+      kind === "LE" ||
+      kind === "GE" ||
+      kind === "EQ" ||
+      kind === "NE"
+    ) {
+      this.advance();
+      const right = this.parseAddition();
+
+      switch (kind) {
+        case "LT":
+          return left < right ? 1 : 0;
+        case "GT":
+          return left > right ? 1 : 0;
+        case "LE":
+          return left <= right ? 1 : 0;
+        case "GE":
+          return left >= right ? 1 : 0;
+        case "EQ":
+          return left === right ? 1 : 0;
+        case "NE":
+          return left !== right ? 1 : 0;
+      }
+    }
+
+    return left;
+  }
+
+  private parseAddition(): number {
+    let left = this.parseMultiplication();
+
+    while (true) {
+      const kind = this.current().kind;
+      if (kind === "PLUS") {
+        this.advance();
+        left = left + this.parseMultiplication();
+      } else if (kind === "MINUS") {
+        this.advance();
+        left = left - this.parseMultiplication();
+      } else {
+        break;
+      }
+    }
+
+    return left;
+  }
+
+  private parseMultiplication(): number {
+    let left = this.parsePower();
+
+    while (true) {
+      const kind = this.current().kind;
+      if (kind === "STAR") {
+        this.advance();
+        left = left * this.parsePower();
+      } else if (kind === "SLASH") {
+        this.advance();
+        left = left / this.parsePower();
+      } else if (kind === "PERCENT") {
+        this.advance();
+        left = left % this.parsePower();
+      } else {
+        break;
+      }
+    }
+
+    return left;
+  }
+
+  private parsePower(): number {
+    const left = this.parseUnary();
+
+    if (this.current().kind === "CARET") {
+      this.advance();
+      // Right-associative
+      const right = this.parsePower();
+      return Math.pow(left, right);
+    }
+
+    return left;
+  }
+
+  private parseUnary(): number {
+    const kind = this.current().kind;
+
+    if (kind === "MINUS") {
+      this.advance();
+      return -this.parseUnary();
+    }
+    if (kind === "PLUS") {
+      this.advance();
+      return +this.parseUnary();
+    }
+    // `!` here applies when the parser descends past the top-level
+    // `parseNot` (e.g. on the RHS of a comparison: `a != !(b)`).
+    if (kind === "BANG") {
+      this.advance();
+      const value = this.parseUnary();
+      return isTruthyNum(value) ? 0 : 1;
+    }
+
+    return this.parsePrimary();
+  }
+
+  private parsePrimary(): number {
+    const token = this.current();
+
+    if (token.kind === "NUMBER") {
+      this.advance();
+      return token.value as number;
+    }
+
+    if (token.kind === "LPAREN") {
+      this.advance();
+      const value = this.parseOr();
+      if (this.current().kind !== "RPAREN") {
+        throw new Error("Expected )");
+      }
+      this.advance();
+      return value;
+    }
+
+    if (token.kind === "IDENTIFIER") {
+      const name = token.value as string;
+      this.advance();
+
+      // Function call
+      if (this.current().kind === "LPAREN") {
+        return this.parseFunctionCall(name);
+      }
+
+      // Wikidot accepts `true` / `false` as boolean literals inside
+      // `[[#expr]]` / `[[#ifexpr]]` (lowercased above by the tokenizer).
+      if (name === "true") return 1;
+      if (name === "false") return 0;
+
+      throw new Error(`undefined constant "${name}"`);
+    }
+
+    throw new Error("Expected expression");
+  }
+
+  private parseFunctionCall(name: string): number {
+    if (this.current().kind !== "LPAREN") {
+      throw new Error("Expected (");
+    }
+    this.advance();
+
+    const args: number[] = [];
+
+    if (this.current().kind !== "RPAREN") {
+      args.push(this.parseOr());
+
+      while (this.current().kind === "COMMA") {
+        this.advance();
+        args.push(this.parseOr());
+      }
+    }
+
+    if (this.current().kind !== "RPAREN") {
+      throw new Error("Expected )");
+    }
+    this.advance();
+
+    return this.callFunction(name, args);
+  }
+
+  private callFunction(name: string, args: number[]): number {
+    switch (name) {
+      case "abs":
+        this.checkArgs(name, args, 1);
+        return Math.abs(args[0]!);
+      case "min":
+        this.checkArgsMin(name, args, 1);
+        return Math.min(...args);
+      case "max":
+        this.checkArgsMin(name, args, 1);
+        return Math.max(...args);
+      case "floor":
+        this.checkArgs(name, args, 1);
+        return Math.floor(args[0]!);
+      case "ceil":
+        this.checkArgs(name, args, 1);
+        return Math.ceil(args[0]!);
+      case "round":
+        this.checkArgs(name, args, 1);
+        return Math.round(args[0]!);
+      case "sqrt":
+        this.checkArgs(name, args, 1);
+        return Math.sqrt(args[0]!);
+      case "sin":
+        this.checkArgs(name, args, 1);
+        return Math.sin(args[0]!);
+      case "cos":
+        this.checkArgs(name, args, 1);
+        return Math.cos(args[0]!);
+      case "tan":
+        this.checkArgs(name, args, 1);
+        return Math.tan(args[0]!);
+      case "ln":
+        this.checkArgs(name, args, 1);
+        return Math.log(args[0]!);
+      case "log":
+        this.checkArgs(name, args, 1);
+        return Math.log10(args[0]!);
+      case "exp":
+        this.checkArgs(name, args, 1);
+        return Math.exp(args[0]!);
+      case "pow":
+        this.checkArgs(name, args, 2);
+        return Math.pow(args[0]!, args[1]!);
+      default:
+        throw new Error(`undefined function "${name}"`);
+    }
+  }
+
+  private checkArgs(name: string, args: number[], expected: number): void {
+    if (args.length !== expected) {
+      throw new Error(`${name}() expects ${expected} argument(s), got ${args.length}`);
+    }
+  }
+
+  private checkArgsMin(name: string, args: number[], min: number): void {
+    if (args.length < min) {
+      throw new Error(`${name}() expects at least ${min} argument(s), got ${args.length}`);
+    }
+  }
+}

package/src/index.ts

@@ -0,0 +1,123 @@
+/**
+ * AST type definitions for the Wikidot markup parser.
+ *
+ * This package provides the TypeScript types that describe the abstract
+ * syntax tree (AST) produced by `@wdprlib/parser` and consumed by
+ * `@wdprlib/render`. It also exports factory helpers for constructing
+ * common node types and context-dependent settings for controlling
+ * parser/renderer behaviour.
+ *
+ * @packageDocumentation
+ */
+
+export type { Position, Point } from "./position";
+export { createPoint, createPosition } from "./position";
+
+/**
+ * Identifies the source markup dialect.
+ *
+ * Currently only `"wikidot"` is supported. Included in {@link SyntaxTree}
+ * so consumers can branch on the dialect if other formats are added later.
+ *
+ * @group Core
+ */
+export type Version = "wikidot";
+
+// Element types (output AST)
+export type {
+  Element,
+  ElementName,
+  ElementData,
+  ElementOf,
+  ElementDataMap,
+  SyntaxTree,
+  ContainerType,
+  StringContainerType,
+  ContainerData,
+  AttributeMap,
+  VariableMap,
+  Alignment,
+  LinkType,
+  LinkLocation,
+  LinkLabel,
+  PageRef,
+  ImageSource,
+  FloatAlignment,
+  ListType,
+  ListItem,
+  ListData,
+  CodeBlockData,
+  TabData,
+  TableCell,
+  TableRow,
+  TableData,
+  DefinitionListItem,
+  Module,
+  CollapsibleData,
+  ClearFloat,
+  AnchorTarget,
+  HeaderType,
+  AlignType,
+  HeadingLevel,
+  Heading,
+  DateItem,
+  Embed,
+  EmbedBlockData,
+  TocEntry,
+  AnchorData,
+  LinkData,
+  ImageData,
+  TableOfContentsData,
+  FootnoteBlockData,
+  BibliographyCiteData,
+  BibliographyBlockData,
+  UserData,
+  DateData,
+  ColorData,
+  MathData,
+  MathInlineData,
+  HtmlData,
+  IframeData,
+  IncludeData,
+  IfTagsData,
+  ExprData,
+  IfCondData,
+  IfExprData,
+} from "./element";
+export {
+  text,
+  container,
+  paragraph,
+  bold,
+  italics,
+  heading,
+  lineBreak,
+  horizontalRule,
+  link,
+  list,
+  listItemElements,
+  listItemSubList,
+  isStringContainerType,
+  isHeaderType,
+  isAlignType,
+  isContainerTypeParagraphSafe,
+  isParagraphSafe,
+} from "./element";
+
+// Diagnostics
+export type { Diagnostic, DiagnosticSeverity, ParseResult } from "./diagnostic";
+
+// Constants
+export { STYLE_SLOT_PREFIX } from "./constants";
+
+// CSS value definitions
+export type { CssLengthUnit } from "./css";
+export { CSS_LENGTH_UNITS } from "./css";
+
+// Wikitext settings
+export type { WikitextMode, WikitextSettings } from "./settings";
+
+// Expression evaluator (shared by parser preprocess and render).
+export { evaluateExpression, isTruthy, formatExprValue } from "./expr-eval";
+export type { ExprResult } from "./expr-eval";
+export { createSettings, DEFAULT_SETTINGS } from "./settings";