// Copyright 2018 The Fuchsia Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

// This generates a language definition file because writing it by hand is too hard.

import { OnigRegExp } from "oniguruma";

// Format of language definition JSON
const tmSchema =
  "https://raw.githubusercontent.com/martinring/tmlanguage/master/tmlanguage.json";
type TmCaptures = { [k: string]: { name: string } };
type TmIncludePattern = { include: string };
type TmMatchPattern = {
  name?: string;
  match: string;
  captures?: TmCaptures;
};
type TmBlockPattern = {
  name?: string;
  begin: string;
  beginCaptures?: TmCaptures;
  end: string;
  endCaptures?: TmCaptures;
  patterns: TmPattern[];
};
type TmPattern = TmIncludePattern | TmMatchPattern | TmBlockPattern;
type TmLanguage = {
  $schema: string;
  name: string;
  scopeName: string;
  patterns: TmPattern[];
  repository: {
    [key: string]: { patterns: TmPattern[] };
  };
};

class NamedPattern {
  readonly re: string;
  readonly names: string[];
  constructor(re: string, names?: string[]) {
    this.re = re;
    this.names = names || [];
    // Hack to only validate if there are no unresolved recursive groups
    const recursive_re = /\\g<(.*?)>/g;
    let match;
    while ((match = recursive_re.exec(re)) !== null) {
      if (!re.includes(`(?<${match[1]}>`)) {
        // Group definition not found, can't validate yet.
        return;
      }
    }
    this.validate();
  }

  validate() {
    const num_groups = new OnigRegExp(this.re + "|").searchSync("")!.length - 1;
    if (num_groups !== this.names.length) {
      throw new Error(
        `Found ${num_groups} but expected ${this.names.length} groups in ${this.re}`
      );
    }
  }

  toString() {
    return this.re;
  }

  assert(s: string) {
    const re = new OnigRegExp(this.re);
    const m = re.searchSync(s);
    if (!m) {
      throw Error(
        `${JSON.stringify(s)} did not match pattern ${JSON.stringify(this.re)}`
      );
    }
    const c = m[0];
    if (c.index !== 0 || c.start !== 0 || c.length !== s.length) {
      throw Error(
        `${JSON.stringify(s)} did not fully match pattern ${JSON.stringify(
          this.re
        )}, only matched ${JSON.stringify(m[0])}`
      );
    }
  }
}

function _captures(pattern: Pattern): TmCaptures {
  const captures: { [k: string]: { name: string } } = {};
  const names = _names(pattern);
  for (let i = 0; i < names.length; i++) {
    captures[`${i + 1}`] = { name: names[i] };
  }
  return captures;
}

function include(name: string): TmIncludePattern {
  return { include: `#${name}` };
}

function match(name: string, pat: Pattern): TmMatchPattern {
  return {
    name: `${name}.fidl`,
    match: pat.toString(),
    captures: _captures(pat),
  };
}

function anonMatch(pattern: Pattern | TmPattern): TmPattern {
  if (pattern instanceof String) {
    return {
      match: pattern.toString(),
    };
  }
  if (pattern instanceof NamedPattern) {
    return {
      match: pattern.toString(),
      captures: _captures(pattern),
    };
  }
  return pattern;
}

function block(args: {
  name: string;
  begin: Pattern;
  end: Pattern;
  patterns: Array<Pattern | TmPattern>;
}): TmBlockPattern {
  const tmPatterns: TmPattern[] = args.patterns.map(anonMatch);
  return {
    name: `${args.name}.fidl`,
    begin: args.begin.toString(),
    beginCaptures: _captures(args.begin),
    end: args.end.toString(),
    endCaptures: _captures(args.end),
    patterns: tmPatterns,
  };
}

type Pattern = NamedPattern | String;

/** Return names of the groups in the pattern. */
function _names(pat: Pattern | Pattern[]): string[] {
  if (pat instanceof Array) {
    return pat.map(_names).reduce((prev, names) => [...(prev || []), ...names]);
  }
  if (pat instanceof NamedPattern) {
    return pat.names;
  } else {
    return [];
  }
}

function _join(
  pats: Pattern[],
  prefix: string,
  separator: string,
  suffix: string
): NamedPattern {
  const re = pats.map((p) => p.toString()).join(separator);
  return new NamedPattern(prefix + re + suffix, _names(pats));
}

/**
 * Create a new pattern consisting of adjacent input patterns optionally separated by whitespace.
 * @param pats Patterns
 */
function seq(...pats: Pattern[]): NamedPattern {
  return _join(pats, "", "\\s*", "");
}

function one_of(...pats: Pattern[]): NamedPattern {
  return _join(pats, "(?:", "|", ")");
}

function one_of_words(...words: string[]) {
  return word(one_of(...words));
}

function comma_separated(pat: Pattern): NamedPattern {
  return new NamedPattern(`(?:${pat})(?:,\\s*${pat})*`, [
    ..._names(pat),
    ..._names(pat),
  ]);
}

function zero_or_more(pat: Pattern): NamedPattern {
  return new NamedPattern(`(?:${pat}\\s*)*`, _names(pat));
}

function optional(pat: Pattern): NamedPattern {
  return new NamedPattern(`(?:${pat})?`, _names(pat));
}

function word(word: Pattern): NamedPattern {
  return new NamedPattern(`\\b${word}\\b`, _names(word));
}

function named(pat: Pattern, name: string) {
  return new NamedPattern(`(${pat})`, [name, ..._names(pat)]);
}

const NUMERIC_LITERAL = named(
  "-?\\b(?:(?:0(?:x|X)[0-9a-fA-F]*)|(?:0(?:b|B)[01]*)|(?:(?:[0-9]+\\.?[0-9]*)|(?:\\.[0-9]+))(?:(?:e|E)(?:\\+|-)?[0-9]+)?)\\b",
  "constant.numeric"
);
const BOOLEAN_LITERAL = named(
  one_of_words("true", "false"),
  "constant.language"
);
const STRING_LITERAL = named('"(?:[^\\"]|\\.)*"', "string.quoted.double");
const LITERAL = one_of(NUMERIC_LITERAL, BOOLEAN_LITERAL, STRING_LITERAL);

const IDENTIFIER = "\\b[a-zA-Z_][0-9a-zA-Z_]*\\b";
const COMPOUND_IDENTIFIER = new NamedPattern(
  `${IDENTIFIER}(?:\\.${IDENTIFIER})*`
);

const CONSTANT = one_of(LITERAL, COMPOUND_IDENTIFIER);
const NUMERIC_CONSTANT = one_of(NUMERIC_LITERAL, COMPOUND_IDENTIFIER);

function angle_brackets(contents: Pattern) {
  const scope = "punctuation.bracket.angle";
  return seq(named("<", scope), contents, named(">", scope));
}

function keyword(keyword: string, name: string = "keyword.control") {
  return word(named(keyword, name));
}

function separator(sep: string) {
  return named(sep, "punctuation.separator");
}

const primitive_types = [
  "bool",
  "float32",
  "float64",
  "int8",
  "int16",
  "int32",
  "int64",
  "uint8",
  "uint16",
  "uint32",
  "uint64",
];

const PRIMITIVE_TYPE = named(
  one_of_words(...primitive_types),
  "storage.type.basic"
);
const EOL = new NamedPattern("(;)", ["punctuation.terminator"]);
const LIBRARY_NAME = named(COMPOUND_IDENTIFIER, "entity.name.type");
const LOCAL_TYPE = named(IDENTIFIER, "entity.name.type");
const LAYOUT_REFERENCE = named(COMPOUND_IDENTIFIER, "entity.name.type");
const VARIABLE = named(IDENTIFIER, "variable");

const ATTRIBUTE_TAG = named(seq(
  "@",
  IDENTIFIER
), "entity.other.attribute-name");

const MODIFIERS = named(zero_or_more(
  one_of(
    word("strict"),
    word("flexible"),
    word("resource"),
  )
), "storage.type.modifier");

const ORDINAL = seq(NUMERIC_LITERAL, separator(":"));

const LAYOUT_KIND = one_of(
  keyword("union"),
  keyword("struct"),
  keyword("table"),
  keyword("enum"),
  keyword("bits")
);

const SUBTYPE = seq(separator(":"), LAYOUT_REFERENCE);

// currently the only valid placement for a type layout parameter is the
// first parameter, so take advantage of this fact to simplify this rule
const TYPE_PARAMETERS = angle_brackets(seq(LAYOUT_REFERENCE, optional(
  seq(
    separator(","),
    NUMERIC_CONSTANT,
  )
)));

const TYPE_CONSTRAINT = named(CONSTANT, "storage.type.constraint");

const TYPE_CONSTRAINTS = seq(
  ":",
  one_of(TYPE_CONSTRAINT, angle_brackets(comma_separated(TYPE_CONSTRAINT)))
);

// Checks

const INLINE_LAYOUT_PREFIX = seq(
  MODIFIERS,
  LAYOUT_KIND,
  optional(SUBTYPE),
  "{"
);

comma_separated("[a-zA-Z]+").assert("foo, bar");

COMPOUND_IDENTIFIER.assert("foo");
COMPOUND_IDENTIFIER.assert("foo_bar");
COMPOUND_IDENTIFIER.assert("foo.bar.baz");

NUMERIC_LITERAL.assert("-42");
NUMERIC_LITERAL.assert("0x12345deadbeef");
NUMERIC_LITERAL.assert("0b101110");

NUMERIC_CONSTANT.assert("0");
NUMERIC_CONSTANT.assert("ABC");

MODIFIERS.assert("strict");
MODIFIERS.assert("flexible");
MODIFIERS.assert("resource");
MODIFIERS.assert("strict resource");
MODIFIERS.assert("flexible resource");
MODIFIERS.assert("resource strict");
MODIFIERS.assert("resource flexible");

TYPE_PARAMETERS.assert("<Foo, 3>");
TYPE_PARAMETERS.assert("<Foo>");

TYPE_CONSTRAINTS.assert(":optional");
TYPE_CONSTRAINTS.assert(":zx.rights.DUPLICATE");
TYPE_CONSTRAINTS.assert(":MAX_SIZE");
TYPE_CONSTRAINTS.assert(":<MAX_SIZE, optional>");
TYPE_CONSTRAINTS.assert(":<CHANNEL, zx.rights.DUPLICATE, optional>");

INLINE_LAYOUT_PREFIX.assert("resource flexible struct {");
INLINE_LAYOUT_PREFIX.assert("strict enum : int32 {");

ATTRIBUTE_TAG.assert("@foo");

const WITH_COMMENTS_AND_ATTRIBUTES = [
  include("comments"),
  include("attributes"),
];

// TODO: support handle rights
const tmLanguage: TmLanguage = {
  $schema: tmSchema,
  name: "FIDL",
  scopeName: "source.fidl",
  patterns: [
    ...WITH_COMMENTS_AND_ATTRIBUTES,

    // Library declaration
    match("meta.library", seq(keyword("library"), LIBRARY_NAME, EOL)),

    // Variations of using
    match("meta.library", seq(keyword("using"), LIBRARY_NAME, EOL)),
    match(
      "meta.library",
      seq(keyword("using"), LIBRARY_NAME, keyword("as"), LOCAL_TYPE, EOL)
    ),

    // Aliases: an aliased type can only be a layout reference, and cannot be re-parameterized
    match(
      "meta.type-alias",
      seq(
        keyword("alias"),
        named(IDENTIFIER, "variable.alias"),
        separator("="),
        LAYOUT_REFERENCE,
        optional(TYPE_CONSTRAINTS)
      )
    ),

    // Const declaration
    block({
      name: "meta.const",
      begin: seq(
        keyword("const"),
        named(IDENTIFIER, "variable.constant"),
        named(one_of(COMPOUND_IDENTIFIER, PRIMITIVE_TYPE), "storage.type"),
        separator("="),
      ),
      end: EOL,
      patterns: [include("const-value")],
    }),

    // Protocols
    block({
      name: "meta.protocol-block",
      begin: seq(keyword("protocol"), LOCAL_TYPE, "{"),
      end: "}",
      patterns: [
        ...WITH_COMMENTS_AND_ATTRIBUTES,
        seq(keyword("compose"), named(COMPOUND_IDENTIFIER, "entity.name.type")),
        include("method"),
      ],
    }),

    // Type declaration
    block({
      name: "meta.type",
      begin: seq(
        keyword("type"),
        LOCAL_TYPE,
        separator("="),
      ),
      end: EOL,
      patterns: [include("type-constructor")],
    }),
  ],
  repository: {
    comments: {
      patterns: [
        match("invalid.illegal.stray-comment-end", "\\*/.*\\n"),
        match("comment.line.documentation", "///.*\\n"),
        match("comment.line.double-slash", "//.*\\n"),
      ],
    },
    "attribute-with-args": {
      patterns: [
        block({
          name: "meta.attribute.with-args",
          begin: seq(ATTRIBUTE_TAG, "\\("),
          end: "\\)",
          patterns: [
            seq(optional(separator(",")), IDENTIFIER, separator("=")),
            include("const-value"),
          ],
        }),
      ],
    },
    "attribute-no-args": {
      patterns: [
        match("meta.attribute.no-args", ATTRIBUTE_TAG),
      ],
    },
    attributes: {
      patterns: [
        include("attribute-with-args"),
        include("attribute-no-args"),
      ],
    },
    method: {
      patterns: [
        block({
          name: "meta.method",
          begin: seq(named(IDENTIFIER, "entity.name.function")),
          // TODO: support anonymous layout errors
          end: seq(optional(seq(keyword("error"), LAYOUT_REFERENCE)), EOL),
          patterns: [include("method-argument"), separator("->")],
        }),
      ],
    },
    "const-value": {
      patterns: [
        match("meta.separator", separator("\\|")),
        match("storage.type.operand", CONSTANT),
      ],
    },
    "numeric-const-value": {
      patterns: [
        match("meta.separator", separator("\\|")),
        match("storage.type.operand", NUMERIC_CONSTANT),
      ],
    },
    "default-value": {
      patterns: [
        match("meta.separator", separator("=")),
        include("const-value"),
      ],
    },
    "method-argument": {
      patterns: [
        block({
          name: "meta.method.arguments",
          begin: "\\(",
          end: "\\)",
          patterns: [include("type-constructor")],
        }),
      ],
    },
    "type-constructor": {
      patterns: [
        include("type-constructor-inline"),
        include("type-constructor-inline-arg"),
        include("type-constructor-reference"),
      ],
    },
    // a type constructor that contains an inline layout, e.g. `foo struct { ... };
    "type-constructor-inline": {
      patterns: [
        block({
          name: "meta.type-constructor-inline",
          begin: seq(INLINE_LAYOUT_PREFIX),
          end: seq(
            "}",
            // no user-defined layouts currently accept parameters, so don't
            // worry about syntax highlighting for them yet; just check for
            // constraints
            optional(TYPE_CONSTRAINTS),
          ),
          patterns: [...WITH_COMMENTS_AND_ATTRIBUTES, include("layout-member")],
        }),
      ],
    },
    // a type constructor with a layout parameter that is an inline layout, e.g.
    // `foo bar<struct {...}, ...>:<...>`;
    "type-constructor-inline-arg": {
      patterns: [
        // currently the only valid placement for a type layout parameter is the
        // first parameter, so take advantage of this fact to simplify this rule
        block({
          name: "meta.type-constructor-inline-arg",
          begin: seq(
            LAYOUT_REFERENCE,
            "<",
            INLINE_LAYOUT_PREFIX
          ),
          end: seq(
            // the rest of the layout parameters (i.e. array size)
            optional(seq(",", CONSTANT)),
            ">",
            optional(TYPE_CONSTRAINTS),
          ),
          patterns: [...WITH_COMMENTS_AND_ATTRIBUTES, include("layout-member")],
        }),
      ],
    },
    // a type constructor that references another type, e.g. `foo bar<...>:<...>;`
    "type-constructor-reference": {
      patterns: [
        match(
          "meta.type-constructor-reference",
          seq(
            LAYOUT_REFERENCE,
            optional(TYPE_PARAMETERS),
            optional(TYPE_CONSTRAINTS),
          )
        ),
      ],
    },
    "layout-member": {
      // the ordering of the patterns below is important, since we want to attempt value before we
      // attempt a default struct member
      patterns: [
        include("layout-member-reserved"),
        include("layout-member-value"),
        include("layout-member-struct"),
        include("layout-member-ordinaled"),
      ],
    },
    // the reserved member, e.g. `1: reserved;`, only applies to unions and tables
    "layout-member-reserved": {
      patterns: [
        match(
          "meta.layout.reserved-member",
          seq(ORDINAL, keyword("reserved"), EOL)
        ),
      ],
    },
    // an ordinaled layout member, like `1: a bool;`, only applies to unions and tables
    "layout-member-ordinaled": {
      patterns: [
        block({
          name: "meta.layout.ordinaled-member",
          begin: seq(ORDINAL, VARIABLE),
          end: EOL,
          patterns: [include("type-constructor")],
        }),
      ],
    },
    // a struct member; may have an optional default value, e.g. `foo uint8 = 3;`
    "layout-member-struct": {
      patterns: [
        block({
          name: "meta.layout.struct-struct-member",
          begin: VARIABLE,
          end: EOL,
          patterns: [
            include("type-constructor"),
            include("default-value"),
          ],
        }),
      ],
    },
    // a layout member that specifies a value, e.g. `foo = 1;`, only applies to bits and enums
    "layout-member-value": {
      patterns: [
        block({
          name: "meta.layout.value-member",
          begin: seq(
            VARIABLE,
            separator("="),
          ),
          end: EOL,
          patterns: [include("numeric-const-value")],
        }),
      ],
    },
  },
};

console.log(JSON.stringify(tmLanguage, null, "    "));