Nuclex.Support/Source/Settings/ConfigurationFileStore.Parsing.cs

#region CPL License
/*
Nuclex Framework
Copyright (C) 2002-2014 Nuclex Development Labs

This library is free software; you can redistribute it and/or
modify it under the terms of the IBM Common Public License as
published by the IBM Corporation; either version 1.0 of the
License, or (at your option) any later version.

This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
IBM Common Public License for more details.

You should have received a copy of the IBM Common Public
License along with this library
*/
#endregion

using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.IO;

using Nuclex.Support.Parsing;

namespace Nuclex.Support.Settings {

  partial class ConfigurationFileStore {

    #region class ParserState

    /// <summary>Remembers the target store and current category of the parser</summary>
    private class ParserState {

      /// <summary>Store to which the parsed categories and options will be added</summary>
      public ConfigurationFileStore Store;

      /// <summary>Current category options belong to</summary>
      public Category Category;

    }

    #endregion // class ParserState

    /// <summary>Parses a configuration file from the specified text reader</summary>
    /// <param name="reader">Reader the configuration file will be parsed from</param>
    /// <returns>The configuration file parsed from the specified reader</returns>
    public static ConfigurationFileStore Parse(TextReader reader) {
      var store = new ConfigurationFileStore();
      var state = new ParserState() {
        Store = store,
        Category = store.rootCategory
      };

      for(; ; ) {
        string line = reader.ReadLine();
        if(line == null) {
          break;
        }

        parseLine(state, line);
      }

      return store;
    }

    /// <summary>Incrementally parses a line read from a configuration file</summary>
    /// <param name="state">Current parser state</param>
    /// <param name="line">Line that has been read</param>
    private static void parseLine(ParserState state, string line) {

      // If the line is empty, ignore it
      int length = line.Length;
      if(length == 0) {
        return;
      }

      // Skip all spaces at the beginning of the line
      int firstCharacterIndex = 0;
      ParserHelper.SkipSpaces(line, ref firstCharacterIndex);

      // If the line contained nothing but spaces, ignore it
      if(firstCharacterIndex == length) {
        return;
      }

      // If the line is a comment, ignore it
      if((line[firstCharacterIndex] == '#') || (line[firstCharacterIndex] == ';')) {
        return;
      }

      // Now the line is either a category definition or some attempt to set an option
      if(line[firstCharacterIndex] == '[') {
        parseCategory(state, line, firstCharacterIndex);
      } else {
        parseOption(state, line, firstCharacterIndex);
      }
      state.Category.Lines.Add(line);

    }

    /// <summary>Parses a category definition encountered on a line</summary>
    /// <param name="state">Current parser state</param>
    /// <param name="line">Line containing the category definition</param>
    /// <param name="firstCharacterIndex">Index of the definition's first character</param>
    private static void parseCategory(
      ParserState state, string line, int firstCharacterIndex
    ) {
      Debug.Assert(line[firstCharacterIndex] == '[');

      int nameStartIndex = firstCharacterIndex + 1;
      ParserHelper.SkipSpaces(line, ref nameStartIndex);

      int lastCharacterIndex = line.Length - 1;
      if(nameStartIndex >= lastCharacterIndex) {
        return; // No space left for closing brace
      }

      int nameEndIndex = line.IndexOf(']', nameStartIndex);
      if(nameEndIndex == -1) {
        return; // No closing brace in line
      }

      // Skip any whitespaces between the last character and the closing brace
      do {
        --nameEndIndex;
      } while(char.IsWhiteSpace(line, nameEndIndex));

      // Now we know that the line holds a category definition and where exactly in
      // the line the category name is located. Create the category.
      state.Category = new Category() {
        CategoryName = new StringSegment(
          line, nameStartIndex, nameEndIndex - nameStartIndex + 1
        ),
        OptionLookup = new Dictionary<string, Option>(),
        Lines = new List<string>()
      };
      state.Store.categoryLookup.Add(state.Category.CategoryName.ToString(), state.Category);
    }

    /// <summary>Parses an option definition encountered on a line</summary>
    /// <param name="state">Current parser state</param>
    /// <param name="line">Line containing the option definition</param>
    /// <param name="firstCharacterIndex">Index of the definition's first character</param>
    private static void parseOption(
      ParserState state, string line, int firstCharacterIndex
    ) {
      int assignmentIndex = line.IndexOf('=', firstCharacterIndex + 1);
      if(assignmentIndex == -1) {
        return; // No assignment took place
      }

      // Cut off any whitespaces between the option name and the assignment
      int nameEndIndex = assignmentIndex;
      do {
        --nameEndIndex;
      } while(char.IsWhiteSpace(line, nameEndIndex));

      // We have enough information to know that this is an assignment of some kind
      Option option = new Option() {
        LineIndex = state.Category.Lines.Count,
        OptionName = new StringSegment(
          line, firstCharacterIndex, nameEndIndex - firstCharacterIndex + 1
        )
      };

      parseOptionValue(option, line, assignmentIndex + 1);

      // We've got the option assignment, either with an empty or proper value
      state.Store.options.Add(option);
      state.Category.OptionLookup.Add(option.OptionName.ToString(), option);
    }

    /// <summary>Parses the value assigned to an option</summary>
    /// <param name="option">Option to which a value is being assigned</param>
    /// <param name="line">Line containing the option assignment</param>
    /// <param name="assignmentEndIndex">Index one after the assignment character</param>
    private static void parseOptionValue(Option option, string line, int assignmentEndIndex) {
      int firstCharacterIndex = assignmentEndIndex;
      ParserHelper.SkipSpaces(line, ref firstCharacterIndex);

      // Just for beauty, when the option value is empty but padded with spaces,
      // leave one space between the equals sign and the value.
      if(firstCharacterIndex > assignmentEndIndex) {
        ++assignmentEndIndex;
      }

      // If the line consists of only whitespace, create an empty value
      if(firstCharacterIndex == line.Length) {
        option.OptionValue = new StringSegment(line, assignmentEndIndex, 0);
        return;
      }

      char firstCharacter = line[firstCharacterIndex];

      // Values can be quoted to allow for comments characters appearing in them
      int lastCharacterIndex;
      if(firstCharacter == '"') {
        lastCharacterIndex = line.LastIndexOf('"');
      } else {
        lastCharacterIndex = firstCharacterIndex;
      }

      int commentStartIndex = line.IndexOf(';', lastCharacterIndex);
      if(commentStartIndex == -1) {
        commentStartIndex = line.IndexOf('#', lastCharacterIndex);
      }
      if(commentStartIndex == -1) {
        lastCharacterIndex = line.Length - 1;
      } else {
        lastCharacterIndex = commentStartIndex - 1;
      }

      while(lastCharacterIndex > firstCharacterIndex) {
        if(char.IsWhiteSpace(line, lastCharacterIndex)) {
          --lastCharacterIndex;
        } else {
          break;
        }
      }

      option.OptionValue = new StringSegment(
        line, firstCharacterIndex, lastCharacterIndex - firstCharacterIndex + 1
      );
    }

    /// <summary>Determines the best matching type for an option value</summary>
    /// <param name="value">Value for which the best matching type will be found</param>
    /// <returns>The best matching type for the specified option value</returns>
    private static Type getBestMatchingType(ref StringSegment value) {
      if(value.Count == 0) {
        return typeof(string);
      }

      // If there are at least two characters, it may be an integer with
      // a sign in front of it
      if(value.Count >= 2) {
        int index = value.Offset;
        if(ParserHelper.SkipInteger(value.Text, ref index)) {
          if(index >= value.Offset + value.Count) {
            return typeof(int);
          }
          if(value.Text[index] == '.') {
            return typeof(float);
          }
        }
      } else { // If it's just a single character, it may be a number
        if(char.IsNumber(value.Text, value.Offset)) {
          return typeof(int);
        }
      }

      // If it parses as a boolean literal, then it must be a boolean
      if(parseBooleanLiteral(ref value) != null) {
        return typeof(bool);
      }

      return typeof(string);
    }

    /// <summary>Tries to parse a boolean literal</summary>
    /// <param name="value">Value that will be parsed as a boolean literal</param>
    /// <returns>
    ///   True or false if the value was a boolean literal, null if it wasn't
    /// </returns>
    private static bool? parseBooleanLiteral(ref StringSegment value) {
      switch(value.Count) {

        // If the string spells 'no', it is considered a boolean
        case 2: {
          bool isSpellingNo =
            ((value.Text[value.Offset + 0] == 'n') || (value.Text[value.Offset + 0] == 'N')) &&
            ((value.Text[value.Offset + 1] == 'o') || (value.Text[value.Offset + 1] == 'O'));
          return isSpellingNo ? new Nullable<bool>(false) : null;
        }

        // If the string spells 'yes', it is considered a boolean
        case 3: {
          bool isSpellingYes =
            ((value.Text[value.Offset + 0] == 'y') || (value.Text[value.Offset + 0] == 'Y')) &&
            ((value.Text[value.Offset + 1] == 'e') || (value.Text[value.Offset + 1] == 'E')) &&
            ((value.Text[value.Offset + 2] == 's') || (value.Text[value.Offset + 2] == 'S'));
          return isSpellingYes ? new Nullable<bool>(true) : null;
        }

        // If the string spells 'true', it is considered a boolean
        case 4: {
          bool isSpellingTrue =
            ((value.Text[value.Offset + 0] == 't') || (value.Text[value.Offset + 0] == 'T')) &&
            ((value.Text[value.Offset + 1] == 'r') || (value.Text[value.Offset + 1] == 'R')) &&
            ((value.Text[value.Offset + 2] == 'u') || (value.Text[value.Offset + 2] == 'U')) &&
            ((value.Text[value.Offset + 3] == 'e') || (value.Text[value.Offset + 3] == 'E'));
          return isSpellingTrue ? new Nullable<bool>(true) : null;
        }

        // If the string spells 'false', it is considered a boolean
        case 5: {
          bool isSpellingFalse =
            ((value.Text[value.Offset + 0] == 'f') || (value.Text[value.Offset + 0] == 'F')) &&
            ((value.Text[value.Offset + 1] == 'a') || (value.Text[value.Offset + 1] == 'A')) &&
            ((value.Text[value.Offset + 2] == 'l') || (value.Text[value.Offset + 2] == 'L')) &&
            ((value.Text[value.Offset + 3] == 's') || (value.Text[value.Offset + 3] == 'S')) &&
            ((value.Text[value.Offset + 4] == 'e') || (value.Text[value.Offset + 4] == 'E'));
          return isSpellingFalse ? new Nullable<bool>(false) : null;
        }

        // Anything else is not considered a boolean
        default: {
          return null;
        }

      }
    }

  }

} // namespace Nuclex.Support.Configuration