/// <summary>
/// Removes all the token sequences from a specified set. Only
/// sequences already in this set will be removed.
/// </summary>
/// <param name="set">The set to remove from</param>
public void RemoveAll(LookAheadSet set)
{
foreach (var elt in set.elements)
{
this.Remove(elt);
}
}
/// <summary>
/// Returns a look-ahead set with all conflicts between
/// alternatives in a production pattern.
/// </summary>
/// <param name="pattern">The production pattern</param>
/// <param name="maxLength">The maximum token sequence length</param>
/// <returns>A look-ahead set with the conflicts found</returns>
/// <exception cref="ParserCreationException">
/// If an inherent ambiguity was found among the look-ahead sets
/// </exception>
private LookAheadSet FindConflicts(
ProductionPattern pattern,
int maxLength)
{
LookAheadSet result = new LookAheadSet(maxLength);
LookAheadSet set1;
LookAheadSet set2;
for (int i = 0; i < pattern.Count; i++)
{
set1 = pattern[i].LookAhead;
for (int j = 0; j < i; j++)
{
set2 = pattern[j].LookAhead;
result.AddAll(set1.CreateIntersection(set2));
}
}
if (result.IsRepetitive)
{
this.ThrowAmbiguityException(pattern.Name, null, result);
}
return(result);
}
/// <summary>
/// Throws a parse exception that matches the specified look-ahead
/// set. This method will take into account any initial matching
/// tokens in the look-ahead set.
/// </summary>
/// <param name="set">The look-ahead set to match</param>
/// <exception cref="ParseException">Always thrown by this method</exception>
private void ThrowParseException(LookAheadSet set)
{
Token token;
IList <string> list = new List <string>();
int[] initials;
// Read tokens until mismatch
while (set.IsNext(this, 1))
{
set = set.CreateNextSet(this.NextToken().Id);
}
// Find next token descriptions
initials = set.InitialTokens;
for (int i = 0; i < initials.Length; i++)
{
list.Add(this.GetTokenDescription(initials[i]));
}
// Create exception
token = this.NextToken();
throw new ParseException(
ParseException.ErrorType.UnexpectedToken,
token.ToShortString(),
list,
token.StartLine,
token.StartColumn);
}
/// <summary>
/// Adds all the token sequences from a specified set. Only
/// sequences not already in this set will be added.
/// </summary>
/// <param name="set">The set to add from</param>
public void AddAll(LookAheadSet set)
{
foreach (var elt in set.elements)
{
this.Add(elt);
}
}
/**
* Creates a new look-ahead set filter. The filter will contain
* all sequences from this set, possibly left trimmed by each one
* of the sequences in the specified set.
*
* @param set the look-ahead set to trim with
*
* @return a new look-ahead set filter
*/
public LookAheadSet CreateFilter(LookAheadSet set)
{
LookAheadSet result = new LookAheadSet(maxLength);
Sequence first;
Sequence second;
// Handle special cases
if (Size() <= 0 || set.Size() <= 0)
{
return(this);
}
// Create combinations
for (int i = 0; i < elements.Count; i++)
{
first = (Sequence)elements[i];
for (int j = 0; j < set.elements.Count; j++)
{
second = (Sequence)set.elements[j];
if (first.StartsWith(second))
{
result.Add(first.Subsequence(second.Length()));
}
}
}
return(result);
}
/// <summary>
/// Initializes a new instance of the <see cref="ProductionPatternElement"/> class.
/// If the maximum value if zero (0) or negative, it will be set to <c>int.MaxValue</c>.
/// </summary>
/// <param name="isToken">The token flag</param>
/// <param name="id">The node identity</param>
/// <param name="min">The minimum number of occurrences</param>
/// <param name="max">
/// The maximum number of occurrences, or a negative value for infinite
/// </param>
public ProductionPatternElement(
bool isToken,
int id,
int min,
int max)
{
this.token = isToken;
this.id = id;
if (min < 0)
{
min = 0;
}
this.min = min;
if (max <= 0)
{
max = int.MaxValue;
}
else if (max < min)
{
max = min;
}
this.max = max;
this.lookAhead = null;
}
/**
* Finds the look-ahead set for a production pattern element. The
* maximum look-ahead length must be specified. This method does
* NOT take the element repeat into consideration when creating
* the look-ahead set. It is also possible to specify a look-ahead
* set filter, which will make sure that unnecessary token
* sequences will be avoided.
*
* @param elem the production pattern element
* @param length the maximum look-ahead length
* @param dummy a parameter to distinguish the method
* @param stack the call stack used for loop detection
* @param filter the look-ahead set filter
*
* @return the look-ahead set for the pattern element
*
* @throws ParserCreationException if an infinite loop was found
* in the grammar
*/
private LookAheadSet FindLookAhead(ProductionPatternElement elem,
int length,
int dummy,
CallStack stack,
LookAheadSet filter)
{
LookAheadSet result;
ProductionPattern pattern;
if (elem.IsToken())
{
result = new LookAheadSet(length);
result.Add(elem.Id);
}
else
{
pattern = GetPattern(elem.Id);
result = FindLookAhead(pattern, length, stack, filter);
if (stack.Contains(pattern.Name))
{
result = result.CreateRepetitive();
}
}
return(result);
}
/**
* Finds the look-ahead set for a production pattern. The maximum
* look-ahead length must be specified. It is also possible to
* specify a look-ahead set filter, which will make sure that
* unnecessary token sequences will be avoided.
*
* @param pattern the production pattern
* @param length the maximum look-ahead length
* @param stack the call stack used for loop detection
* @param filter the look-ahead set filter
*
* @return the look-ahead set for the production pattern
*
* @throws ParserCreationException if an infinite loop was found
* in the grammar
*/
private LookAheadSet FindLookAhead(ProductionPattern pattern,
int length,
CallStack stack,
LookAheadSet filter)
{
LookAheadSet result;
LookAheadSet temp;
// Check for infinite loop
if (stack.Contains(pattern.Name, length))
{
throw new ParserCreationException(
ParserCreationException.ErrorType.INFINITE_LOOP,
pattern.Name,
(String)null);
}
// Find pattern look-ahead
stack.Push(pattern.Name, length);
result = new LookAheadSet(length);
for (int i = 0; i < pattern.Count; i++)
{
temp = FindLookAhead(pattern[i],
length,
0,
stack,
filter);
result.AddAll(temp);
}
stack.Pop();
return(result);
}
/**
* Removes all the token sequences from a specified set. Only
* sequences already in this set will be removed.
*
* @param set the set to remove from
*/
public void RemoveAll(LookAheadSet set)
{
for (int i = 0; i < set.elements.Count; i++)
{
Remove((Sequence)set.elements[i]);
}
}
/**
* Adds all the token sequences from a specified set. Only
* sequences not already in this set will be added.
*
* @param set the set to add from
*/
public void AddAll(LookAheadSet set)
{
for (int i = 0; i < set.elements.Count; i++)
{
Add((Sequence)set.elements[i]);
}
}
/**
* Throws a parse exception that matches the specified look-ahead
* set. This method will take into account any initial matching
* tokens in the look-ahead set.
*
* @param set the look-ahead set to match
*
* @throws ParseException always thrown by this method
*/
private void ThrowParseException(LookAheadSet set)
{
Token token;
ArrayList list = new ArrayList();
int[] initials;
// Read tokens until mismatch
while (set.IsNext(this, 1))
{
set = set.CreateNextSet(NextToken().Id);
}
// Find next token descriptions
initials = set.GetInitialTokens();
for (int i = 0; i < initials.Length; i++)
{
list.Add(GetTokenDescription(initials[i]));
}
// Create exception
token = NextToken();
throw new ParseException(ParseException.ErrorType.UNEXPECTED_TOKEN,
token.ToShortString(),
list,
token.StartLine,
token.StartColumn);
}
/// <summary>
/// Finds the look-ahead set for a production pattern element. The
/// maximum look-ahead length must be specified. This method does
/// NOT take the element repeat into consideration when creating
/// the look-ahead set. It is also possible to specify a look-ahead
/// set filter, which will make sure that unnecessary token
/// sequences will be avoided.
/// </summary>
/// <param name="elem">The production pattern element</param>
/// <param name="length">The maximum look ahead length</param>
/// <param name="dummy">A parameter to distinguish the method</param>
/// <param name="stack">The call stack used for loop detection</param>
/// <param name="filter">The look ahead set filter</param>
/// <returns>The look-ahead set for the pattern element</returns>
/// <exception cref="ParserCreationException">
/// If an infinite loop was found in the grammar
/// </exception>
private LookAheadSet FindLookAhead(
ProductionPatternElement elem,
int length,
#pragma warning disable IDE0060 // Remove unused parameter
int dummy,
#pragma warning restore IDE0060 // Remove unused parameter
CallStack stack,
LookAheadSet filter)
{
LookAheadSet result;
ProductionPattern pattern;
if (elem.IsToken)
{
result = new LookAheadSet(length);
result.Add(elem.Id);
}
else
{
pattern = this.GetPattern(elem.Id);
result = this.FindLookAhead(pattern, length, stack, filter);
if (stack.Contains(pattern.Name))
{
result = result.CreateRepetitive();
}
}
return(result);
}
/**
* Creates a new production pattern.
*
* @param id the production pattern id
* @param name the production pattern name
*/
public ProductionPattern(int id, string name) {
this.id = id;
this.name = name;
this.synthetic = false;
this.alternatives = new ArrayList();
this.defaultAlt = -1;
this.lookAhead = null;
}
/// <summary>
/// Initializes a new instance of the <see cref="ProductionPattern"/> class.
/// </summary>
/// <param name="id">The production pattern id</param>
/// <param name="name">The production pattern name</param>
public ProductionPattern(int id, string name)
{
this.id = id;
this.name = name;
this.synthetic = false;
this.alternatives = new List <ProductionPatternAlternative>();
this.defaultAlt = -1;
this.lookAhead = null;
}
/**
* Creates a new production pattern.
*
* @param id the production pattern id
* @param name the production pattern name
*/
public ProductionPattern(int id, string name)
{
id = id;
name = name;
synthetic = false;
alternatives = new ArrayList();
defaultAlt = -1;
lookAhead = null;
}
/**
* Checks if some token sequence is present in both this set
* and a specified one.
*
* @param set the look-ahead set to compare with
*
* @return true if the look-ahead sets intersect, or
* false otherwise
*/
public bool Intersects(LookAheadSet set)
{
for (int i = 0; i < elements.Count; i++)
{
if (set.Contains((Sequence)elements[i]))
{
return(true);
}
}
return(false);
}
/**
* Checks if another look-ahead set has an overlapping token
* sequence. An overlapping token sequence is a token sequence
* that is identical to another sequence, but for the length.
* I.e. one of the two sequences may be longer than the other.
*
* @param set the look-ahead set to check
*
* @return true if there is some token sequence that overlaps, or
* false otherwise
*/
public bool IsOverlap(LookAheadSet set)
{
for (int i = 0; i < elements.Count; i++)
{
if (set.IsOverlap((Sequence)elements[i]))
{
return(true);
}
}
return(false);
}
/// <summary>
/// Checks if another look-ahead set has an overlapping token
/// sequence. An overlapping token sequence is a token sequence
/// that is identical to another sequence, but for the length.
/// I.e. one of the two sequences may be longer than the other.
/// </summary>
/// <param name="set">The look-ahead set to check</param>
/// <returns>
/// True if there is some token sequence that overlaps, or false otherwise
/// </returns>
public bool IsOverlap(LookAheadSet set)
{
foreach (var seq in this.elements)
{
if (set.IsOverlap(seq))
{
return(true);
}
}
return(false);
}
/// <summary>
/// Checks if some token sequence is present in both this set
/// and a specified one.
/// </summary>
/// <param name="set">The look-ahead set to compare with</param>
/// <returns>
/// True if the look-ahead sets intersect, or false otherwise
/// </returns>
public bool Intersects(LookAheadSet set)
{
foreach (var elem in this.elements)
{
if (set.Contains(elem))
{
return(true);
}
}
return(false);
}
/// <summary>
/// Finds the look-ahead set for a production pattern element. The
/// maximum look-ahead length must be specified. This method takes
/// the element repeats into consideration when creating the
/// look-ahead set, but does NOT include an empty sequence even if
/// the minimum count is zero (0). It is also possible to specify a
/// look-ahead set filter, which will make sure that unnecessary
/// token sequences will be avoided.
/// </summary>
/// <param name="elem">The production pattern element</param>
/// <param name="length">The maximum look-ahead length</param>
/// <param name="stack">The call stack used for loop detection</param>
/// <param name="filter">The look-ahead set filter</param>
/// <returns>The look-ahead set for the pattern element</returns>
/// <exception cref="ParserCreationException">
/// If an infinite loop was found in the grammar
/// </exception>
private LookAheadSet FindLookAhead(
ProductionPatternElement elem,
int length,
CallStack stack,
LookAheadSet filter)
{
LookAheadSet result;
LookAheadSet first;
LookAheadSet follow;
int max;
// Find initial element look-ahead
first = this.FindLookAhead(elem, length, 0, stack, filter);
result = new LookAheadSet(length);
result.AddAll(first);
if (filter == null || !filter.IsOverlap(result))
{
return(result);
}
// Handle element repetitions
if (elem.MaxCount == int.MaxValue)
{
first = first.CreateRepetitive();
}
max = elem.MaxCount;
if (length < max)
{
max = length;
}
for (int i = 1; i < max; i++)
{
first = first.CreateOverlaps(filter);
if (first.Size <= 0 || first.MinLength >= length)
{
break;
}
follow = this.FindLookAhead(
elem,
length,
0,
stack,
filter.CreateFilter(first));
first = first.CreateCombination(follow);
result.AddAll(first);
}
return(result);
}
/**
* Checks if the next tokens match a production pattern
* alternative. The pattern alternative look-ahead set will be
* used if existing, otherwise this method returns false.
*
* @param alt the pattern alternative to check
*
* @return true if the next tokens match, or
* false otherwise
*/
private bool IsNext(ProductionPatternAlternative alt)
{
LookAheadSet set = alt.LookAhead;
if (set == null)
{
return(false);
}
else
{
return(set.IsNext(this));
}
}
/**
* Checks if the next tokens match a production pattern. The
* pattern look-ahead set will be used if existing, otherwise
* this method returns false.
*
* @param pattern the pattern to check
*
* @return true if the next tokens match, or
* false otherwise
*/
private bool IsNext(ProductionPattern pattern)
{
LookAheadSet set = pattern.LookAhead;
if (set == null)
{
return(false);
}
else
{
return(set.IsNext(this));
}
}
/**
* Returns a look-ahead set with all conflicts between two
* look-ahead sets.
*
* @param pattern the pattern name being analyzed
* @param location the pattern location
* @param set1 the first look-ahead set
* @param set2 the second look-ahead set
*
* @return a look-ahead set with the conflicts found
*
* @throws ParserCreationException if an inherent ambiguity was
* found among the look-ahead sets
*/
private LookAheadSet FindConflicts(string pattern,
string location,
LookAheadSet set1,
LookAheadSet set2)
{
LookAheadSet result;
result = set1.CreateIntersection(set2);
if (result.IsRepetitive())
{
ThrowAmbiguityException(pattern, location, result);
}
return(result);
}
/// <summary>
/// Creates a new look-ahead set with overlaps from another. All
/// token sequences in this set that overlaps with the other set
/// will be added to the new look-ahead set.
/// </summary>
/// <param name="set">The look-ahead set to check with</param>
/// <returns>A new look-ahead set containing the overlaps</returns>
public LookAheadSet CreateOverlaps(LookAheadSet set)
{
LookAheadSet result = new LookAheadSet(this.maxLength);
foreach (var seq in this.elements)
{
if (set.IsOverlap(seq))
{
result.Add(seq);
}
}
return(result);
}
/**
* Creates a new look-ahead set with overlaps from another. All
* token sequences in this set that overlaps with the other set
* will be added to the new look-ahead set.
*
* @param set the look-ahead set to check with
*
* @return a new look-ahead set containing the overlaps
*/
public LookAheadSet CreateOverlaps(LookAheadSet set)
{
LookAheadSet result = new LookAheadSet(maxLength);
Sequence seq;
for (int i = 0; i < elements.Count; i++)
{
seq = (Sequence)elements[i];
if (set.IsOverlap(seq))
{
result.Add(seq);
}
}
return(result);
}
/// <summary>
/// Creates a new look-ahead set that is the result of reading
/// the specified token. The new look-ahead set will contain
/// the rest of all the token sequences that started with the
/// specified token.
/// </summary>
/// <param name="token">The token to read</param>
/// <returns>
/// A new look-ahead set containing the remaining tokens
/// </returns>
public LookAheadSet CreateNextSet(int token)
{
LookAheadSet result = new LookAheadSet(this.maxLength - 1);
object value;
foreach (var seq in this.elements)
{
value = seq.GetToken(0);
if (value != null && token == (int)value)
{
result.Add(seq.Subsequence(1));
}
}
return(result);
}
/**
* Checks if the next tokens match a production pattern
* element. If the element has a look-ahead set it will be
* used, otherwise the look-ahead set of the referenced
* production or token will be used.
*
* @param elem the pattern element to check
*
* @return true if the next tokens match, or
* false otherwise
*/
private bool IsNext(ProductionPatternElement elem)
{
LookAheadSet set = elem.LookAhead;
if (set != null)
{
return(set.IsNext(this));
}
else if (elem.IsToken())
{
return(elem.IsMatch(PeekToken(0)));
}
else
{
return(IsNext(GetPattern(elem.Id)));
}
}
/**
* Creates a new look-ahead set that is the result of reading
* the specified token. The new look-ahead set will contain
* the rest of all the token sequences that started with the
* specified token.
*
* @param token the token to read
*
* @return a new look-ahead set containing the remaining tokens
*/
public LookAheadSet CreateNextSet(int token)
{
LookAheadSet result = new LookAheadSet(maxLength - 1);
Sequence seq;
object value;
for (int i = 0; i < elements.Count; i++)
{
seq = (Sequence)elements[i];
value = seq.GetToken(0);
if (value != null && token == (int)value)
{
result.Add(seq.Subsequence(1));
}
}
return(result);
}
/// <summary>
/// Finds the look-ahead set for a production pattern alternative.
/// The pattern position and maximum look-ahead length must be
/// specified. It is also possible to specify a look-ahead set
/// filter, which will make sure that unnecessary token sequences
/// will be avoided.
/// </summary>
/// <param name="alt">The production pattern alternative</param>
/// <param name="length">The maximum look-ahead length</param>
/// <param name="pos">The pattern element position</param>
/// <param name="stack">The call stack used for loop detection</param>
/// <param name="filter">The look-ahead set filter</param>
/// <returns>
/// The look-ahead set for the pattern alternative
/// </returns>
/// <exception cref="ParserCreationException">
/// If an infinite loop was found in the grammar
/// </exception>
private LookAheadSet FindLookAhead(
ProductionPatternAlternative alt,
int length,
int pos,
CallStack stack,
LookAheadSet filter)
{
LookAheadSet first;
LookAheadSet follow;
LookAheadSet overlaps;
// Check trivial cases
if (length <= 0 || pos >= alt.Count)
{
return(new LookAheadSet(0));
}
// Find look-ahead for this element
first = this.FindLookAhead(alt[pos], length, stack, filter);
if (alt[pos].MinCount == 0)
{
first.AddEmpty();
}
// Find remaining look-ahead
if (filter == null)
{
length -= first.MinLength;
if (length > 0)
{
follow = this.FindLookAhead(alt, length, pos + 1, stack, null);
first = first.CreateCombination(follow);
}
}
else if (filter.IsOverlap(first))
{
overlaps = first.CreateOverlaps(filter);
length -= overlaps.MinLength;
filter = filter.CreateFilter(overlaps);
follow = this.FindLookAhead(alt, length, pos + 1, stack, filter);
first.RemoveAll(overlaps);
first.AddAll(overlaps.CreateCombination(follow));
}
return(first);
}
/// <summary>
/// Throws a parser creation exception for an ambiguity. The
/// specified look-ahead set contains the token conflicts to be
/// reported.
/// </summary>
/// <param name="pattern">The production pattern name</param>
/// <param name="location">The production pattern location, or null</param>
/// <param name="set">The look-ahead set with conflicts</param>
/// <exception cref="ParserCreationException">
/// Always thrown by this method
/// </exception>
private void ThrowAmbiguityException(
string pattern,
string location,
LookAheadSet set)
{
int[] initials;
// Find next token descriptions
initials = set.InitialTokens;
IList <string> list = initials.Select(i => this.GetTokenDescription(i)).ToList();
// Create exception
throw new ParserCreationException(
ParserCreationException.ErrorType.InherentAmbiguity,
pattern,
location,
list);
}
/// <summary>
/// Creates a new identical look-ahead set, except for the
/// repeat flag being set in each token sequence.
/// </summary>
/// <returns>A new repetitive look-ahead set</returns>
public LookAheadSet CreateRepetitive()
{
LookAheadSet result = new LookAheadSet(this.maxLength);
foreach (var seq in this.elements)
{
if (seq.IsRepetitive)
{
result.Add(seq);
}
else
{
result.Add(new Sequence(true, seq));
}
}
return(result);
}
/**
* Calculates the look-ahead needed for the specified production
* pattern. This method attempts to resolve any conflicts and
* stores the results in the pattern look-ahead object.
*
* @param pattern the production pattern
*
* @throws ParserCreationException if the look-ahead set couldn't
* be determined due to inherent ambiguities
*/
private void CalculateLookAhead(ProductionPattern pattern) {
ProductionPatternAlternative alt;
LookAheadSet result;
LookAheadSet[] alternatives;
LookAheadSet conflicts;
LookAheadSet previous = new LookAheadSet(0);
int length = 1;
int i;
CallStack stack = new CallStack();
// Calculate simple look-ahead
stack.Push(pattern.Name, 1);
result = new LookAheadSet(1);
alternatives = new LookAheadSet[pattern.Count];
for (i = 0; i < pattern.Count; i++) {
alt = pattern[i];
alternatives[i] = FindLookAhead(alt, 1, 0, stack, null);
alt.LookAhead = alternatives[i];
result.AddAll(alternatives[i]);
}
if (pattern.LookAhead == null) {
pattern.LookAhead = result;
}
conflicts = FindConflicts(pattern, 1);
// Resolve conflicts
while (conflicts.Size() > 0) {
length++;
stack.Clear();
stack.Push(pattern.Name, length);
conflicts.AddAll(previous);
for (i = 0; i < pattern.Count; i++) {
alt = pattern[i];
if (alternatives[i].Intersects(conflicts)) {
alternatives[i] = FindLookAhead(alt,
length,
0,
stack,
conflicts);
alt.LookAhead = alternatives[i];
}
if (alternatives[i].Intersects(conflicts)) {
if (pattern.DefaultAlternative == null) {
pattern.DefaultAlternative = alt;
} else if (pattern.DefaultAlternative != alt) {
result = alternatives[i].CreateIntersection(conflicts);
ThrowAmbiguityException(pattern.Name,
null,
result);
}
}
}
previous = conflicts;
conflicts = FindConflicts(pattern, length);
}
// Resolve conflicts inside rules
for (i = 0; i < pattern.Count; i++) {
CalculateLookAhead(pattern[i], 0);
}
}
/**
* Finds the look-ahead set for a production pattern element. The
* maximum look-ahead length must be specified. This method takes
* the element repeats into consideration when creating the
* look-ahead set, but does NOT include an empty sequence even if
* the minimum count is zero (0). It is also possible to specify a
* look-ahead set filter, which will make sure that unnecessary
* token sequences will be avoided.
*
* @param elem the production pattern element
* @param length the maximum look-ahead length
* @param stack the call stack used for loop detection
* @param filter the look-ahead set filter
*
* @return the look-ahead set for the pattern element
*
* @throws ParserCreationException if an infinite loop was found
* in the grammar
*/
private LookAheadSet FindLookAhead(ProductionPatternElement elem,
int length,
CallStack stack,
LookAheadSet filter) {
LookAheadSet result;
LookAheadSet first;
LookAheadSet follow;
int max;
// Find initial element look-ahead
first = FindLookAhead(elem, length, 0, stack, filter);
result = new LookAheadSet(length);
result.AddAll(first);
if (filter == null || !filter.IsOverlap(result)) {
return result;
}
// Handle element repetitions
if (elem.MaxCount == Int32.MaxValue) {
first = first.CreateRepetitive();
}
max = elem.MaxCount;
if (length < max) {
max = length;
}
for (int i = 1; i < max; i++) {
first = first.CreateOverlaps(filter);
if (first.Size() <= 0 || first.GetMinLength() >= length) {
break;
}
follow = FindLookAhead(elem,
length,
0,
stack,
filter.CreateFilter(first));
first = first.CreateCombination(follow);
result.AddAll(first);
}
return result;
}
/**
* Finds the look-ahead set for a production pattern. The maximum
* look-ahead length must be specified. It is also possible to
* specify a look-ahead set filter, which will make sure that
* unnecessary token sequences will be avoided.
*
* @param pattern the production pattern
* @param length the maximum look-ahead length
* @param stack the call stack used for loop detection
* @param filter the look-ahead set filter
*
* @return the look-ahead set for the production pattern
*
* @throws ParserCreationException if an infinite loop was found
* in the grammar
*/
private LookAheadSet FindLookAhead(ProductionPattern pattern,
int length,
CallStack stack,
LookAheadSet filter) {
LookAheadSet result;
LookAheadSet temp;
// Check for infinite loop
if (stack.Contains(pattern.Name, length)) {
throw new ParserCreationException(
ParserCreationException.ErrorType.INFINITE_LOOP,
pattern.Name,
(String) null);
}
// Find pattern look-ahead
stack.Push(pattern.Name, length);
result = new LookAheadSet(length);
for (int i = 0; i < pattern.Count; i++) {
temp = FindLookAhead(pattern[i],
length,
0,
stack,
filter);
result.AddAll(temp);
}
stack.Pop();
return result;
}
/**
* Calculates the look-aheads needed for the specified pattern
* alternative. This method attempts to resolve any conflicts in
* optional elements by recalculating look-aheads for referenced
* productions.
*
* @param alt the production pattern alternative
* @param pos the pattern element position
*
* @throws ParserCreationException if the look-ahead set couldn't
* be determined due to inherent ambiguities
*/
private void CalculateLookAhead(ProductionPatternAlternative alt,
int pos) {
ProductionPattern pattern;
ProductionPatternElement elem;
LookAheadSet first;
LookAheadSet follow;
LookAheadSet conflicts;
LookAheadSet previous = new LookAheadSet(0);
String location;
int length = 1;
// Check trivial cases
if (pos >= alt.Count) {
return;
}
// Check for non-optional element
pattern = alt.Pattern;
elem = alt[pos];
if (elem.MinCount == elem.MaxCount) {
CalculateLookAhead(alt, pos + 1);
return;
}
// Calculate simple look-aheads
first = FindLookAhead(elem, 1, new CallStack(), null);
follow = FindLookAhead(alt, 1, pos + 1, new CallStack(), null);
// Resolve conflicts
location = "at position " + (pos + 1);
conflicts = FindConflicts(pattern.Name,
location,
first,
follow);
while (conflicts.Size() > 0) {
length++;
conflicts.AddAll(previous);
first = FindLookAhead(elem,
length,
new CallStack(),
conflicts);
follow = FindLookAhead(alt,
length,
pos + 1,
new CallStack(),
conflicts);
first = first.CreateCombination(follow);
elem.LookAhead = first;
if (first.Intersects(conflicts)) {
first = first.CreateIntersection(conflicts);
ThrowAmbiguityException(pattern.Name, location, first);
}
previous = conflicts;
conflicts = FindConflicts(pattern.Name,
location,
first,
follow);
}
// Check remaining elements
CalculateLookAhead(alt, pos + 1);
}
/**
* Adds all the token sequences from a specified set. Only
* sequences not already in this set will be added.
*
* @param set the set to add from
*/
public void AddAll(LookAheadSet set) {
for (int i = 0; i < set.elements.Count; i++) {
Add((Sequence) set.elements[i]);
}
}
/**
* Checks if another look-ahead set has an overlapping token
* sequence. An overlapping token sequence is a token sequence
* that is identical to another sequence, but for the length.
* I.e. one of the two sequences may be longer than the other.
*
* @param set the look-ahead set to check
*
* @return true if there is some token sequence that overlaps, or
* false otherwise
*/
public bool IsOverlap(LookAheadSet set) {
for (int i = 0; i < elements.Count; i++) {
if (set.IsOverlap((Sequence) elements[i])) {
return true;
}
}
return false;
}
/**
* Creates a new look-ahead set that is the result of reading
* the specified token. The new look-ahead set will contain
* the rest of all the token sequences that started with the
* specified token.
*
* @param token the token to read
*
* @return a new look-ahead set containing the remaining tokens
*/
public LookAheadSet CreateNextSet(int token) {
LookAheadSet result = new LookAheadSet(maxLength - 1);
Sequence seq;
object value;
for (int i = 0; i < elements.Count; i++) {
seq = (Sequence) elements[i];
value = seq.GetToken(0);
if (value != null && token == (int) value) {
result.Add(seq.Subsequence(1));
}
}
return result;
}
/**
* Creates a new element. If the maximum value if zero (0) or
* negative, it will be set to Int32.MaxValue.
*
* @param isToken the token flag
* @param id the node identity
* @param min the minimum number of occurancies
* @param max the maximum number of occurancies, or
* negative for infinite
*/
public ProductionPatternElement(bool isToken,
int id,
int min,
int max) {
this.token = isToken;
this.id = id;
if (min < 0) {
min = 0;
}
this.min = min;
if (max <= 0) {
max = Int32.MaxValue;
} else if (max < min) {
max = min;
}
this.max = max;
this.lookAhead = null;
}
/**
* Creates a duplicate look-ahead set, possibly with a
* different maximum length.
*
* @param maxLength the maximum token sequence length
* @param set the look-ahead set to copy
*/
public LookAheadSet(int maxLength, LookAheadSet set)
: this(maxLength) {
AddAll(set);
}
/**
* Creates a new look-ahead set filter. The filter will contain
* all sequences from this set, possibly left trimmed by each one
* of the sequences in the specified set.
*
* @param set the look-ahead set to trim with
*
* @return a new look-ahead set filter
*/
public LookAheadSet CreateFilter(LookAheadSet set) {
LookAheadSet result = new LookAheadSet(maxLength);
Sequence first;
Sequence second;
// Handle special cases
if (this.Size() <= 0 || set.Size() <= 0) {
return this;
}
// Create combinations
for (int i = 0; i < elements.Count; i++) {
first = (Sequence) elements[i];
for (int j = 0; j < set.elements.Count; j++) {
second = (Sequence) set.elements[j];
if (first.StartsWith(second)) {
result.Add(first.Subsequence(second.Length()));
}
}
}
return result;
}
/**
* Creates a new look-ahead set that is the combination of
* this set with another set. The combination is created by
* creating new token sequences that consist of appending all
* elements from the specified set onto all elements in this
* set. This is sometimes referred to as the cartesian
* product.
*
* @param set the set to combine with
*
* @return a new look-ahead set containing the combination
*/
public LookAheadSet CreateCombination(LookAheadSet set) {
LookAheadSet result = new LookAheadSet(maxLength);
Sequence first;
Sequence second;
// Handle special cases
if (this.Size() <= 0) {
return set;
} else if (set.Size() <= 0) {
return this;
}
// Create combinations
for (int i = 0; i < elements.Count; i++) {
first = (Sequence) elements[i];
if (first.Length() >= maxLength) {
result.Add(first);
} else if (first.Length() <= 0) {
result.AddAll(set);
} else {
for (int j = 0; j < set.elements.Count; j++) {
second = (Sequence) set.elements[j];
result.Add(first.Concat(maxLength, second));
}
}
}
return result;
}
/**
* Checks if some token sequence is present in both this set
* and a specified one.
*
* @param set the look-ahead set to compare with
*
* @return true if the look-ahead sets intersect, or
* false otherwise
*/
public bool Intersects(LookAheadSet set) {
for (int i = 0; i < elements.Count; i++) {
if (set.Contains((Sequence) elements[i])) {
return true;
}
}
return false;
}
/**
* Finds the look-ahead set for a production pattern element. The
* maximum look-ahead length must be specified. This method does
* NOT take the element repeat into consideration when creating
* the look-ahead set. It is also possible to specify a look-ahead
* set filter, which will make sure that unnecessary token
* sequences will be avoided.
*
* @param elem the production pattern element
* @param length the maximum look-ahead length
* @param dummy a parameter to distinguish the method
* @param stack the call stack used for loop detection
* @param filter the look-ahead set filter
*
* @return the look-ahead set for the pattern element
*
* @throws ParserCreationException if an infinite loop was found
* in the grammar
*/
private LookAheadSet FindLookAhead(ProductionPatternElement elem,
int length,
int dummy,
CallStack stack,
LookAheadSet filter) {
LookAheadSet result;
ProductionPattern pattern;
if (elem.IsToken()) {
result = new LookAheadSet(length);
result.Add(elem.Id);
} else {
pattern = GetPattern(elem.Id);
result = FindLookAhead(pattern, length, stack, filter);
if (stack.Contains(pattern.Name)) {
result = result.CreateRepetitive();
}
}
return result;
}
/**
* Removes all the token sequences from a specified set. Only
* sequences already in this set will be removed.
*
* @param set the set to remove from
*/
public void RemoveAll(LookAheadSet set) {
for (int i = 0; i < set.elements.Count; i++) {
Remove((Sequence) set.elements[i]);
}
}
/**
* Returns a look-ahead set with all conflics between
* alternatives in a production pattern.
*
* @param pattern the production pattern
* @param maxLength the maximum token sequence length
*
* @return a look-ahead set with the conflicts found
*
* @throws ParserCreationException if an inherent ambiguity was
* found among the look-ahead sets
*/
private LookAheadSet FindConflicts(ProductionPattern pattern,
int maxLength) {
LookAheadSet result = new LookAheadSet(maxLength);
LookAheadSet set1;
LookAheadSet set2;
for (int i = 0; i < pattern.Count; i++) {
set1 = pattern[i].LookAhead;
for (int j = 0; j < i; j++) {
set2 = pattern[j].LookAhead;
result.AddAll(set1.CreateIntersection(set2));
}
}
if (result.IsRepetitive()) {
ThrowAmbiguityException(pattern.Name, null, result);
}
return result;
}
/**
* Creates a new look-ahead set that is the intersection of
* this set with another set. The token sequences in the net
* set will only have the repeat flag set if it was set in
* both the identical token sequences.
*
* @param set the set to intersect with
*
* @return a new look-ahead set containing the intersection
*/
public LookAheadSet CreateIntersection(LookAheadSet set) {
LookAheadSet result = new LookAheadSet(maxLength);
Sequence seq1;
Sequence seq2;
for (int i = 0; i < elements.Count; i++) {
seq1 = (Sequence) elements[i];
seq2 = set.FindSequence(seq1);
if (seq2 != null && seq1.IsRepetitive()) {
result.Add(seq2);
} else if (seq2 != null) {
result.Add(seq1);
}
}
return result;
}
/**
* Returns a look-ahead set with all conflicts between two
* look-ahead sets.
*
* @param pattern the pattern name being analyzed
* @param location the pattern location
* @param set1 the first look-ahead set
* @param set2 the second look-ahead set
*
* @return a look-ahead set with the conflicts found
*
* @throws ParserCreationException if an inherent ambiguity was
* found among the look-ahead sets
*/
private LookAheadSet FindConflicts(string pattern,
string location,
LookAheadSet set1,
LookAheadSet set2) {
LookAheadSet result;
result = set1.CreateIntersection(set2);
if (result.IsRepetitive()) {
ThrowAmbiguityException(pattern, location, result);
}
return result;
}
/**
* Creates a new look-ahead set with overlaps from another. All
* token sequences in this set that overlaps with the other set
* will be added to the new look-ahead set.
*
* @param set the look-ahead set to check with
*
* @return a new look-ahead set containing the overlaps
*/
public LookAheadSet CreateOverlaps(LookAheadSet set) {
LookAheadSet result = new LookAheadSet(maxLength);
Sequence seq;
for (int i = 0; i < elements.Count; i++) {
seq = (Sequence) elements[i];
if (set.IsOverlap(seq)) {
result.Add(seq);
}
}
return result;
}
/**
* Returns the union of all alternative look-ahead sets in a
* production pattern.
*
* @param pattern the production pattern
*
* @return a unified look-ahead set
*/
private LookAheadSet FindUnion(ProductionPattern pattern) {
LookAheadSet result;
int length = 0;
int i;
for (i = 0; i < pattern.Count; i++) {
result = pattern[i].LookAhead;
if (result.GetMaxLength() > length) {
length = result.GetMaxLength();
}
}
result = new LookAheadSet(length);
for (i = 0; i < pattern.Count; i++) {
result.AddAll(pattern[i].LookAhead);
}
return result;
}
/**
* Creates a new identical look-ahead set, except for the
* repeat flag being set in each token sequence.
*
* @return a new repetitive look-ahead set
*/
public LookAheadSet CreateRepetitive() {
LookAheadSet result = new LookAheadSet(maxLength);
Sequence seq;
for (int i = 0; i < elements.Count; i++) {
seq = (Sequence) elements[i];
if (seq.IsRepetitive()) {
result.Add(seq);
} else {
result.Add(new Sequence(true, seq));
}
}
return result;
}
/**
* Throws a parse exception that matches the specified look-ahead
* set. This method will take into account any initial matching
* tokens in the look-ahead set.
*
* @param set the look-ahead set to match
*
* @throws ParseException always thrown by this method
*/
private void ThrowParseException(LookAheadSet set) {
Token token;
ArrayList list = new ArrayList();
int[] initials;
// Read tokens until mismatch
while (set.IsNext(this, 1)) {
set = set.CreateNextSet(NextToken().Id);
}
// Find next token descriptions
initials = set.GetInitialTokens();
for (int i = 0; i < initials.Length; i++) {
list.Add(GetTokenDescription(initials[i]));
}
// Create exception
token = NextToken();
throw new ParseException(ParseException.ErrorType.UNEXPECTED_TOKEN,
token.ToShortString(),
list,
token.StartLine,
token.StartColumn);
}
/**
* Finds the look-ahead set for a production pattern alternative.
* The pattern position and maximum look-ahead length must be
* specified. It is also possible to specify a look-ahead set
* filter, which will make sure that unnecessary token sequences
* will be avoided.
*
* @param alt the production pattern alternative
* @param length the maximum look-ahead length
* @param pos the pattern element position
* @param stack the call stack used for loop detection
* @param filter the look-ahead set filter
*
* @return the look-ahead set for the pattern alternative
*
* @throws ParserCreationException if an infinite loop was found
* in the grammar
*/
private LookAheadSet FindLookAhead(ProductionPatternAlternative alt,
int length,
int pos,
CallStack stack,
LookAheadSet filter) {
LookAheadSet first;
LookAheadSet follow;
LookAheadSet overlaps;
// Check trivial cases
if (length <= 0 || pos >= alt.Count) {
return new LookAheadSet(0);
}
// Find look-ahead for this element
first = FindLookAhead(alt[pos], length, stack, filter);
if (alt[pos].MinCount == 0) {
first.AddEmpty();
}
// Find remaining look-ahead
if (filter == null) {
length -= first.GetMinLength();
if (length > 0) {
follow = FindLookAhead(alt, length, pos + 1, stack, null);
first = first.CreateCombination(follow);
}
} else if (filter.IsOverlap(first)) {
overlaps = first.CreateOverlaps(filter);
length -= overlaps.GetMinLength();
filter = filter.CreateFilter(overlaps);
follow = FindLookAhead(alt, length, pos + 1, stack, filter);
first.RemoveAll(overlaps);
first.AddAll(overlaps.CreateCombination(follow));
}
return first;
}
/**
* Throws a parser creation exception for an ambiguity. The
* specified look-ahead set contains the token conflicts to be
* reported.
*
* @param pattern the production pattern name
* @param location the production pattern location, or null
* @param set the look-ahead set with conflicts
*
* @throws ParserCreationException always thrown by this method
*/
private void ThrowAmbiguityException(string pattern,
string location,
LookAheadSet set) {
ArrayList list = new ArrayList();
int[] initials;
// Find next token descriptions
initials = set.GetInitialTokens();
for (int i = 0; i < initials.Length; i++) {
list.Add(GetTokenDescription(initials[i]));
}
// Create exception
throw new ParserCreationException(
ParserCreationException.ErrorType.INHERENT_AMBIGUITY,
pattern,
location,
list);
}