/// <summary>
/// Extracts a block of text delimited by double quotes. It is
/// assumed the parser is positioned at the first quote. The
/// quotes are not included in the returned string. On return,
/// the parser is positioned at the closing quote or at the end of
/// the text if the closing quote was not found.
/// </summary>
/// <param name="parser">TextParser object</param>
/// <returns>The extracted text</returns>
private string ExtractQuote(TextParser parser)
{
// Extract contents of quote
parser.MoveAhead();
int start = parser.Position;
while (!parser.EndOfText && parser.Peek() != '"')
{
parser.MoveAhead();
}
return(parser.Extract(start, parser.Position));
}
/// <summary>
/// Extracts a block of text delimited by the specified open and close
/// characters. It is assumed the parser is positioned at an
/// occurrence of the open character. The open and closing characters
/// are not included in the returned string. On return, the parser is
/// positioned at the closing character or at the end of the text if
/// the closing character was not found.
/// </summary>
/// <param name="parser">TextParser object</param>
/// <param name="openChar">Start-of-block delimiter</param>
/// <param name="closeChar">End-of-block delimiter</param>
/// <returns>The extracted text</returns>
private string ExtractBlock(TextParser parser, char openChar, char closeChar)
{
// Track delimiter depth
int depth = 1;
// Extract characters between delimiters
parser.MoveAhead();
int start = parser.Position;
while (!parser.EndOfText)
{
if (parser.Peek() == openChar)
{
// Increase block depth
depth++;
}
else if (parser.Peek() == closeChar)
{
// Decrease block depth
depth--;
// Test for end of block
if (depth == 0)
{
break;
}
}
else if (parser.Peek() == '"')
{
// Don't count delimiters within quoted text
ExtractQuote(parser);
}
// Move to next character
parser.MoveAhead();
}
return(parser.Extract(start, parser.Position));
}
/// <summary>
/// Parses a query segment and converts it to an expression
/// tree.
/// </summary>
/// <param name="query">Query segment to convert</param>
/// <param name="defaultConjunction">Implicit conjunction type</param>
/// <returns>Root node of expression tree</returns>
private INode ParseNode(string query, ConjunctionTypes defaultConjunction)
{
TermForms termForm = TermForms.Inflectional;
bool termExclude = false;
ConjunctionTypes conjunction = defaultConjunction;
bool resetState = true;
INode root = null;
INode node;
string term;
TextParser parser = new TextParser(query);
while (!parser.EndOfText)
{
if (resetState)
{
// Reset modifiers
termForm = TermForms.Inflectional;
termExclude = false;
conjunction = defaultConjunction;
resetState = false;
}
parser.MovePastWhitespace();
if (!parser.EndOfText &&
!Punctuation.Contains(parser.Peek()))
{
// Extract query term
int start = parser.Position;
parser.MoveAhead();
while (!parser.EndOfText &&
!Punctuation.Contains(parser.Peek()) &&
!Char.IsWhiteSpace(parser.Peek()))
{
parser.MoveAhead();
}
// Allow trailing wildcard
if (parser.Peek() == '*')
{
parser.MoveAhead();
termForm = TermForms.Literal;
}
// Interpret token
term = parser.Extract(start, parser.Position);
if (String.Compare(term, "AND", true) == 0)
{
conjunction = ConjunctionTypes.And;
}
else if (String.Compare(term, "OR", true) == 0)
{
conjunction = ConjunctionTypes.Or;
}
else if (String.Compare(term, "NEAR", true) == 0)
{
conjunction = ConjunctionTypes.Near;
}
else if (String.Compare(term, "NOT", true) == 0)
{
termExclude = true;
}
else
{
root = AddNode(root, term, termForm, termExclude, conjunction);
resetState = true;
}
continue;
}
else if (parser.Peek() == '"')
{
// Match next term exactly
termForm = TermForms.Literal;
// Extract quoted term
term = ExtractQuote(parser);
root = AddNode(root, term.Trim(), termForm, termExclude, conjunction);
resetState = true;
}
else if (parser.Peek() == '(')
{
// Parse parentheses block
term = ExtractBlock(parser, '(', ')');
node = ParseNode(term, defaultConjunction);
root = AddNode(root, node, conjunction, true);
resetState = true;
}
else if (parser.Peek() == '<')
{
// Parse angle brackets block
term = ExtractBlock(parser, '<', '>');
node = ParseNode(term, ConjunctionTypes.Near);
root = AddNode(root, node, conjunction);
resetState = true;
}
else if (parser.Peek() == '-')
{
// Match when next term is not present
termExclude = true;
}
else if (parser.Peek() == '+')
{
// Match next term exactly
termForm = TermForms.Literal;
}
else if (parser.Peek() == '~')
{
// Match synonyms of next term
termForm = TermForms.Thesaurus;
}
// Advance to next character
parser.MoveAhead();
}
return(root);
}
/// <summary>
/// Parses a query segment and converts it to an expression
/// tree.
/// </summary>
/// <param name="query">Query segment to convert</param>
/// <param name="defaultConjunction">Implicit conjunction type</param>
/// <returns>Root node of expression tree</returns>
private INode ParseNode(string query, ConjunctionTypes defaultConjunction)
{
TermForms termForm = TermForms.Inflectional;
bool termExclude = false;
ConjunctionTypes conjunction = defaultConjunction;
bool resetState = true;
INode root = null;
INode node;
string term;
TextParser parser = new TextParser(query);
while (!parser.EndOfText)
{
if (resetState)
{
// Reset modifiers
termForm = TermForms.Inflectional;
termExclude = false;
conjunction = defaultConjunction;
resetState = false;
}
parser.MovePastWhitespace();
if (!parser.EndOfText &&
!Punctuation.Contains(parser.Peek()))
{
// Extract query term
int start = parser.Position;
parser.MoveAhead();
while (!parser.EndOfText &&
!Punctuation.Contains(parser.Peek()) &&
!Char.IsWhiteSpace(parser.Peek()))
parser.MoveAhead();
// Allow trailing wildcard
if (parser.Peek() == '*')
{
parser.MoveAhead();
termForm = TermForms.Literal;
}
// Interpret token
term = parser.Extract(start, parser.Position);
if (String.Compare(term, "AND", true) == 0)
conjunction = ConjunctionTypes.And;
else if (String.Compare(term, "OR", true) == 0)
conjunction = ConjunctionTypes.Or;
else if (String.Compare(term, "NEAR", true) == 0)
conjunction = ConjunctionTypes.Near;
else if (String.Compare(term, "NOT", true) == 0)
termExclude = true;
else
{
root = AddNode(root, term, termForm, termExclude, conjunction);
resetState = true;
}
continue;
}
else if (parser.Peek() == '"')
{
// Match next term exactly
termForm = TermForms.Literal;
// Extract quoted term
term = ExtractQuote(parser);
root = AddNode(root, term.Trim(), termForm, termExclude, conjunction);
resetState = true;
}
else if (parser.Peek() == '(')
{
// Parse parentheses block
term = ExtractBlock(parser, '(', ')');
node = ParseNode(term, defaultConjunction);
root = AddNode(root, node, conjunction, true);
resetState = true;
}
else if (parser.Peek() == '<')
{
// Parse angle brackets block
term = ExtractBlock(parser, '<', '>');
node = ParseNode(term, ConjunctionTypes.Near);
root = AddNode(root, node, conjunction);
resetState = true;
}
else if (parser.Peek() == '-')
{
// Match when next term is not present
termExclude = true;
}
else if (parser.Peek() == '+')
{
// Match next term exactly
termForm = TermForms.Literal;
}
else if (parser.Peek() == '~')
{
// Match synonyms of next term
termForm = TermForms.Thesaurus;
}
// Advance to next character
parser.MoveAhead();
}
return root;
}
/// <summary>
/// Extracts a block of text delimited by double quotes. It is
/// assumed the parser is positioned at the first quote. The
/// quotes are not included in the returned string. On return,
/// the parser is positioned at the closing quote or at the end of
/// the text if the closing quote was not found.
/// </summary>
/// <param name="parser">TextParser object</param>
/// <returns>The extracted text</returns>
private string ExtractQuote(TextParser parser)
{
// Extract contents of quote
parser.MoveAhead();
int start = parser.Position;
while (!parser.EndOfText && parser.Peek() != '"')
parser.MoveAhead();
return parser.Extract(start, parser.Position);
}
/// <summary>
/// Extracts a block of text delimited by the specified open and close
/// characters. It is assumed the parser is positioned at an
/// occurrence of the open character. The open and closing characters
/// are not included in the returned string. On return, the parser is
/// positioned at the closing character or at the end of the text if
/// the closing character was not found.
/// </summary>
/// <param name="parser">TextParser object</param>
/// <param name="openChar">Start-of-block delimiter</param>
/// <param name="closeChar">End-of-block delimiter</param>
/// <returns>The extracted text</returns>
private string ExtractBlock(TextParser parser, char openChar, char closeChar)
{
// Track delimiter depth
int depth = 1;
// Extract characters between delimiters
parser.MoveAhead();
int start = parser.Position;
while (!parser.EndOfText)
{
if (parser.Peek() == openChar)
{
// Increase block depth
depth++;
}
else if (parser.Peek() == closeChar)
{
// Decrease block depth
depth--;
// Test for end of block
if (depth == 0)
break;
}
else if (parser.Peek() == '"')
{
// Don't count delimiters within quoted text
ExtractQuote(parser);
}
// Move to next character
parser.MoveAhead();
}
return parser.Extract(start, parser.Position);
}