pkg/front_end/test/spell_checking_utils.dart

// Copyright (c) 2019, the Dart project authors.  Please see the AUTHORS file
// for details. All rights reserved. Use of this source code is governed by a
// BSD-style license that can be found in the LICENSE file.

import 'dart:io';

Set<String> _dictionary;

Set<String> spellcheckString(String s, {bool splitAsCode: false}) {
  Set<String> wrongWords;
  _dictionary ??= _loadDictionary();
  List<String> words = splitStringIntoWords(s, splitAsCode: splitAsCode);
  for (int i = 0; i < words.length; i++) {
    String word = words[i].toLowerCase();
    if (!_dictionary.contains(word)) {
      wrongWords ??= new Set<String>();
      wrongWords.add(word);
    }
  }
  return wrongWords;
}

Set<String> _loadDictionary() {
  Set<String> dictionary = new Set<String>();
  addWords(Uri uri) {
    for (String word in File.fromUri(uri)
        .readAsStringSync()
        .split("\n")
        .map((s) => s.toLowerCase())) {
      if (word.startsWith("#")) continue;
      int indexOfHash = word.indexOf(" #");
      if (indexOfHash >= 0) {
        // Strip out comment.
        word = word.substring(0, indexOfHash).trim();
      }
      if (word == "") continue;
      if (word.contains(" ")) throw "'$word' contains spaces";
      dictionary.add(word);
    }
  }

  // TODO(jensj): Split list into several:
  // * A common one with correctly spelled words.
  // * A special one for messages.yaml.
  // * A special one for source code in 'src'.
  // * A special one for source code not in 'src'.
  // and allow the caller to specify the combination of lists we want to use.
  addWords(Uri.base.resolve("pkg/front_end/test/spell_checking_list.txt"));
  return dictionary;
}

List<String> splitStringIntoWords(String s, {bool splitAsCode: false}) {
  List<String> result = new List<String>();
  // Match whitespace and the characters "-", "=", "|", "/", ",".
  String regExpStringInner = r"\s-=\|\/,";
  if (splitAsCode) {
    // If splitting as code also split by "_", ":", ".", "(", ")", "<", ">",
    // "[", "]", "{", "}", "@", "&", "#", "?". (As well as doing stuff to camel
    // casing further below).
    regExpStringInner = "${regExpStringInner}_:\\.\\(\\)<>\\[\\]\{\}@&#\\?";
  }
  // Match one or more of the characters specified above.
  String regExp = "[$regExpStringInner]+";
  if (splitAsCode) {
    // If splitting as code we also want to remove the two characters "\n".
    regExp = "([$regExpStringInner]|(\\\\n))+";
  }

  List<String> split = s.split(new RegExp(regExp));
  for (int i = 0; i < split.length; i++) {
    String word = split[i].trim();
    if (word.isEmpty) continue;
    int start = 0;
    int end = word.length;
    bool changedStart = false;
    while (start < end) {
      int unit = word.codeUnitAt(start);
      if (unit >= 65 && unit <= 90) {
        // A-Z => Good.
        break;
      } else if (unit >= 97 && unit <= 122) {
        // a-z => Good.
        break;
      } else {
        changedStart = true;
        start++;
      }
    }
    bool changedEnd = false;
    while (end > start) {
      int unit = word.codeUnitAt(end - 1);
      if (unit >= 65 && unit <= 90) {
        // A-Z => Good.
        break;
      } else if (unit >= 97 && unit <= 122) {
        // a-z => Good.
        break;
      } else {
        changedEnd = true;
        end--;
      }
    }
    if (changedEnd && word.codeUnitAt(end) == 41) {
      // Special case trimmed ')' if there's a '(' inside the string.
      for (int i = start; i < end; i++) {
        if (word.codeUnitAt(i) == 40) {
          end++;
          break;
        }
      }
    }
    if (start == end) continue;

    if (splitAsCode) {
      bool prevCapitalized = false;
      for (int i = start; i < end; i++) {
        bool thisCapitalized = false;
        int unit = word.codeUnitAt(i);
        if (unit >= 65 && unit <= 90) {
          thisCapitalized = true;
        } else if (unit >= 48 && unit <= 57) {
          // Number inside --- allow that.
          continue;
        }
        if (prevCapitalized && thisCapitalized) {
          // Sort-of-weird thing, something like "thisIsTheCNN". Carry on.

          // Except if the previous was an 'A' and that both the previous
          // (before that) and the next (if any) is not capitalized, i.e.
          // we special-case the case of 'A' as in 'AWord' being 'a word'.
          int prevUnit = word.codeUnitAt(i - 1);
          if (prevUnit == 65) {
            bool doSpecialCase = true;
            if (i + 1 < end) {
              int nextUnit = word.codeUnitAt(i + 1);
              if (nextUnit >= 65 && nextUnit <= 90) {
                // Next is capitalized too.
                doSpecialCase = false;
              }
            }
            if (i - 2 >= start) {
              int prevPrevUnit = word.codeUnitAt(i - 2);
              if (prevPrevUnit >= 65 && prevPrevUnit <= 90) {
                // Prev-prev was capitalized too.
                doSpecialCase = false;
              }
            }
            if (doSpecialCase) {
              result.add(word.substring(start, i));
              start = i;
            }
          }

          // And the case where the next one is not capitalized --- we must
          // assume that "TheCNNAlso" should be "The", "CNN", "Also".
          if (start < i && i + 1 < end) {
            int nextUnit = word.codeUnitAt(i + 1);
            if (nextUnit >= 97 && nextUnit <= 122) {
              // Next is not capitalized.
              result.add(word.substring(start, i));
              start = i;
            }
          }
        } else if (!prevCapitalized && thisCapitalized) {
          // Starting a new camel case word.
          if (i > start) {
            result.add(word.substring(start, i));
            start = i;
          }
        } else if (prevCapitalized && !thisCapitalized) {
          // This should have been handled above.
        } else if (!prevCapitalized && !thisCapitalized) {
          // Continued word.
        }
        if (i + 1 == end) {
          // End of string.
          if (i >= start) {
            result.add(word.substring(start, end));
          }
        }
        prevCapitalized = thisCapitalized;
      }
    } else {
      result.add(
          (changedStart || changedEnd) ? word.substring(start, end) : word);
    }
  }
  return result;
}