Exemplo n.º 1
0
func (p *Pattern) anchoredForwardSearch(it *colltab.Iter) (start, end int) {
	if it.Next() {
		if end := p.searchOnce(it); end != -1 {
			return 0, end
		}
	}
	return -1, -1
}
Exemplo n.º 2
0
// CompileString compiles and returns a pattern that can be used for faster
// searching.
func (m *Matcher) CompileString(s string) *Pattern {
	p := &Pattern{m: m}
	iter := newcolltab.Iter{Weighter: m.w}
	for iter.SetInputString(s); iter.Next(); {
	}
	p.ce = iter.Elems
	p.deleteEmptyElements()
	return p
}
Exemplo n.º 3
0
// Compile compiles and returns a pattern that can be used for faster searching.
func (m *Matcher) Compile(b []byte) *Pattern {
	p := &Pattern{m: m}
	iter := colltab.Iter{Weighter: m.w}
	for iter.SetInput(b); iter.Next(); {
	}
	p.ce = iter.Elems
	p.deleteEmptyElements()
	return p
}
Exemplo n.º 4
0
func (p *Pattern) forwardSearch(it *colltab.Iter) (start, end int) {
	for start := 0; it.Next(); it.Reset(start) {
		nextStart := it.End()
		if end := p.searchOnce(it); end != -1 {
			return start, end
		}
		start = nextStart
	}
	return -1, -1
}
Exemplo n.º 5
0
// searchOnce tries to match the pattern s.p at the text position i. s.buf needs
// to be filled with collation elements of the first segment, where n is the
// number of source bytes consumed for this segment. It will return the end
// position of the match or -1.
func (p *Pattern) searchOnce(it *colltab.Iter) (end int) {
	var pLevel [4]int

	m := p.m
	for {
		k := 0
		for ; k < it.N; k++ {
			if v := it.Elems[k].Primary(); v > 0 {
				if w, ok := p.next(&pLevel[0], colltab.Elem.Primary); !ok || v != w {
					return -1
				}
			}

			if !m.ignoreDiacritics {
				if v := it.Elems[k].Secondary(); v > 0 {
					if w, ok := p.next(&pLevel[1], colltab.Elem.Secondary); !ok || v != w {
						return -1
					}
				}
			} else if it.Elems[k].Primary() == 0 {
				// We ignore tertiary values of collation elements of the
				// secondary level.
				continue
			}

			// TODO: distinguish between case and width. This will be easier to
			// implement after we moved to the new collation implementation.
			if !m.ignoreWidth && !m.ignoreCase {
				if v := it.Elems[k].Tertiary(); v > 0 {
					if w, ok := p.next(&pLevel[2], tertiary); !ok || int(v) != w {
						return -1
					}
				}
			}
			// TODO: check quaternary weight
		}
		it.Discard() // Remove the current segment from the buffer.

		// Check for completion.
		switch {
		// If any of these cases match, we are not at the end.
		case pLevel[0] < len(p.ce):
		case !m.ignoreDiacritics && pLevel[1] < len(p.ce):
		case !(m.ignoreWidth || m.ignoreCase) && pLevel[2] < len(p.ce):
		default:
			// At this point, both the segment and pattern has matched fully.
			// However, the segment may still be have trailing modifiers.
			// This can be verified by another call to next.
			end = it.End()
			if it.Next() && it.Elems[0].Primary() == 0 {
				if !m.ignoreDiacritics {
					return -1
				}
				end = it.End()
			}
			return end
		}

		// Fill the buffer with the next batch of collation elements.
		if !it.Next() {
			return -1
		}
	}
}