ohm-js/src/pexprs-toArgumentNameList.js

An object-oriented language for parsing and pattern matching

import {abstract, copyWithoutDuplicates} from './common.js';
import * as pexprs from './pexprs-main.js';

// --------------------------------------------------------------------
// Private stuff
// --------------------------------------------------------------------

function isRestrictedJSIdentifier(str) {
  return /^[a-zA-Z_$][0-9a-zA-Z_$]*$/.test(str);
}

function resolveDuplicatedNames(argumentNameList) {
  // `count` is used to record the number of times each argument name occurs in the list,
  // this is useful for checking duplicated argument name. It maps argument names to ints.
  const count = Object.create(null);
  argumentNameList.forEach(argName => {
    count[argName] = (count[argName] || 0) + 1;
  });

  // Append subscripts ('_1', '_2', ...) to duplicate argument names.
  Object.keys(count).forEach(dupArgName => {
    if (count[dupArgName] <= 1) {
      return;
    }

    // This name shows up more than once, so add subscripts.
    let subscript = 1;
    argumentNameList.forEach((argName, idx) => {
      if (argName === dupArgName) {
        argumentNameList[idx] = argName + '_' + subscript++;
      }
    });
  });
}

// --------------------------------------------------------------------
// Operations
// --------------------------------------------------------------------

/*
  Returns a list of strings that will be used as the default argument names for its receiver
  (a pexpr) in a semantic action. This is used exclusively by the Semantics Editor.

  `firstArgIndex` is the 1-based index of the first argument name that will be generated for this
  pexpr. It enables us to name arguments positionally, e.g., if the second argument is a
  non-alphanumeric terminal like "+", it will be named '$2'.

  `noDupCheck` is true if the caller of `toArgumentNameList` is not a top level caller. It enables
  us to avoid nested duplication subscripts appending, e.g., '_1_1', '_1_2', by only checking
  duplicates at the top level.

  Here is a more elaborate example that illustrates how this method works:
  `(a "+" b).toArgumentNameList(1)` evaluates to `['a', '$2', 'b']` with the following recursive
  calls:

    (a).toArgumentNameList(1) -> ['a'],
    ("+").toArgumentNameList(2) -> ['$2'],
    (b).toArgumentNameList(3) -> ['b']

  Notes:
  * This method must only be called on well-formed expressions, e.g., the receiver must
    not have any Alt sub-expressions with inconsistent arities.
  * e.getArity() === e.toArgumentNameList(1).length
*/
// function(firstArgIndex, noDupCheck) { ... }
pexprs.PExpr.prototype.toArgumentNameList = abstract('toArgumentNameList');

pexprs.any.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  return ['any'];
};

pexprs.end.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  return ['end'];
};

pexprs.Terminal.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  if (typeof this.obj === 'string' && /^[_a-zA-Z0-9]+$/.test(this.obj)) {
    // If this terminal is a valid suffix for a JS identifier, just prepend it with '_'
    return ['_' + this.obj];
  } else {
    // Otherwise, name it positionally.
    return ['$' + firstArgIndex];
  }
};

pexprs.Range.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  let argName = this.from + '_to_' + this.to;
  // If the `argName` is not valid then try to prepend a `_`.
  if (!isRestrictedJSIdentifier(argName)) {
    argName = '_' + argName;
  }
  // If the `argName` still not valid after prepending a `_`, then name it positionally.
  if (!isRestrictedJSIdentifier(argName)) {
    argName = '$' + firstArgIndex;
  }
  return [argName];
};

pexprs.Alt.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  // `termArgNameLists` is an array of arrays where each row is the
  // argument name list that corresponds to a term in this alternation.
  const termArgNameLists = this.terms.map(term =>
    term.toArgumentNameList(firstArgIndex, true),
  );

  const argumentNameList = [];
  const numArgs = termArgNameLists[0].length;
  for (let colIdx = 0; colIdx < numArgs; colIdx++) {
    const col = [];
    for (let rowIdx = 0; rowIdx < this.terms.length; rowIdx++) {
      col.push(termArgNameLists[rowIdx][colIdx]);
    }
    const uniqueNames = copyWithoutDuplicates(col);
    argumentNameList.push(uniqueNames.join('_or_'));
  }

  if (!noDupCheck) {
    resolveDuplicatedNames(argumentNameList);
  }
  return argumentNameList;
};

pexprs.Seq.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  // Generate the argument name list, without worrying about duplicates.
  let argumentNameList = [];
  this.factors.forEach(factor => {
    const factorArgumentNameList = factor.toArgumentNameList(firstArgIndex, true);
    argumentNameList = argumentNameList.concat(factorArgumentNameList);

    // Shift the firstArgIndex to take this factor's argument names into account.
    firstArgIndex += factorArgumentNameList.length;
  });
  if (!noDupCheck) {
    resolveDuplicatedNames(argumentNameList);
  }
  return argumentNameList;
};

pexprs.Iter.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  const argumentNameList = this.expr
      .toArgumentNameList(firstArgIndex, noDupCheck)
      .map(exprArgumentString =>
      exprArgumentString[exprArgumentString.length - 1] === 's' ?
        exprArgumentString + 'es' :
        exprArgumentString + 's',
      );
  if (!noDupCheck) {
    resolveDuplicatedNames(argumentNameList);
  }
  return argumentNameList;
};

pexprs.Opt.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  return this.expr.toArgumentNameList(firstArgIndex, noDupCheck).map(argName => {
    return 'opt' + argName[0].toUpperCase() + argName.slice(1);
  });
};

pexprs.Not.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  return [];
};

pexprs.Lookahead.prototype.toArgumentNameList = pexprs.Lex.prototype.toArgumentNameList =
  function(firstArgIndex, noDupCheck) {
    return this.expr.toArgumentNameList(firstArgIndex, noDupCheck);
  };

pexprs.Apply.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  return [this.ruleName];
};

pexprs.UnicodeChar.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  return ['$' + firstArgIndex];
};

pexprs.Param.prototype.toArgumentNameList = function(firstArgIndex, noDupCheck) {
  return ['param' + this.index];
};

// "Value pexprs" (Value, Str, Arr, Obj) are going away soon, so we don't worry about them here.