qedit/pkg/buffer/rope.go

package buffer

import (
	"io"
	"unicode/utf8"

	ropes "github.com/zyedidia/rope"
)

type RopeBuffer struct {
	rope    *ropes.Node
	anchors []*Cursor
}

func NewRopeBuffer(contents []byte) *RopeBuffer {
	return &RopeBuffer{
		ropes.New(contents),
		nil,
	}
}

// LineColToPos returns the index of the byte at line, col. If line is less than
// zero, or more than the number of available lines, the function will panic. If
// col is less than zero, the function will panic. If col is greater than the
// length of the line, the position of the last byte of the line is returned,
// instead.
func (b *RopeBuffer) LineColToPos(line, col int) int {
	pos := b.getLineStartPos(line)

	// Have to do this algorithm for safety. If this function was declared to panic
	// or index out of bounds memory, if col > the given line length, it would be
	// more efficient and simpler. But unfortunately, I believe it is necessary.
	if col > 0 {
		_, r := b.rope.SplitAt(pos)
		l, _ := r.SplitAt(b.rope.Len() - pos)

		l.EachLeaf(func(n *ropes.Node) bool {
			data := n.Value() // Reference; not a copy.
			for _, r := range string(data) {
				if col == 0 || r == '\n' {
					return true // Found the position of the column
				}
				pos++
				col--
			}
			return false // Have not gotten to the appropriate position, yet
		})
	}

	return pos
}

// Line returns a slice of the data at the given line, including the ending line-
// delimiter. line starts from zero. Data returned may or may not be a copy: do not
// write it.
func (b *RopeBuffer) Line(line int) []byte {
	pos := b.getLineStartPos(line)
	bytes := 0

	_, r := b.rope.SplitAt(pos)
	l, _ := r.SplitAt(b.rope.Len() - pos)

	var isCRLF bool // true if the last byte was '\r'
	l.EachLeaf(func(n *ropes.Node) bool {
		data := n.Value() // Reference; not a copy.
		for i, r := range string(data) {
			if r == '\r' {
				isCRLF = true
			} else if r == '\n' {
				bytes += i // Add bytes before i
				if isCRLF {
					bytes += 2 // Add the CRLF bytes
				} else {
					bytes += 1 // Add LF byte
				}
				return true // Read (past-tense) the whole line
			} else {
				isCRLF = false
			}
		}
		bytes += len(data)
		return false // Have not read the whole line, yet
	})

	return b.rope.Slice(pos, pos+bytes) // NOTE: may be faster to do it ourselves
}

// Returns a slice of the buffer from startLine, startCol, to endLine, endCol,
// inclusive bounds. The returned value may or may not be a copy of the data,
// so do not write to it.
func (b *RopeBuffer) Slice(startLine, startCol, endLine, endCol int) []byte {
	endPos := b.LineColToPos(endLine, endCol)
	if length := b.rope.Len(); endPos >= length {
		endPos = length - 1
	}
	return b.rope.Slice(b.LineColToPos(startLine, startCol), endPos+1)
}

// Bytes returns all of the bytes in the buffer. This function is very likely
// to copy all of the data in the buffer. Use sparingly. Try using other methods,
// where possible.
func (b *RopeBuffer) Bytes() []byte {
	return b.rope.Value()
}

// Insert copies a byte slice (inserting it) into the position at line, col.
func (b *RopeBuffer) Insert(line, col int, value []byte) {
	b.rope.Insert(b.LineColToPos(line, col), value)
	b.shiftAnchors(line, col, utf8.RuneCount(value))
}

// Remove deletes any characters between startLine, startCol, and endLine,
// endCol, inclusive bounds.
func (b *RopeBuffer) Remove(startLine, startCol, endLine, endCol int) {
	start := b.LineColToPos(startLine, startCol)
	end := b.LineColToPos(endLine, endCol) + 1

	if len := b.rope.Len(); end >= len {
		end = len
		if start > end {
			return
		}
	}

	b.rope.Remove(start, end)
	// Shift anchors within the range
	b.shiftAnchorsRemovedRange(start, startLine, startCol, endLine, endCol)
	// Shift anchors after the range
	b.shiftAnchors(endLine, endCol+1, start-end-1)
}

// Returns the number of occurrences of 'sequence' in the buffer, within the range
// of start line and col, to end line and col. End is exclusive.
func (b *RopeBuffer) Count(startLine, startCol, endLine, endCol int, sequence []byte) int {
	startPos := b.LineColToPos(startLine, startCol)
	endPos := b.LineColToPos(endLine, endCol)
	return b.rope.Count(startPos, endPos, sequence)
}

// Len returns the number of bytes in the buffer.
func (b *RopeBuffer) Len() int {
	return b.rope.Len()
}

// Lines returns the number of lines in the buffer. If the buffer is empty,
// 1 is returned, because there is always at least one line. This function
// basically counts the number of newline ('\n') characters in a buffer.
func (b *RopeBuffer) Lines() int {
	rope := b.rope
	return rope.Count(0, rope.Len(), []byte{'\n'}) + 1
}

// getLineStartPos returns the first byte index of the given line (starting from zero).
// The returned index can be equal to the length of the buffer, not pointing to any byte,
// which means the byte is on the last, and empty, line of the buffer. If line is greater
// than or equal to the number of lines in the buffer, a panic is issued.
func (b *RopeBuffer) getLineStartPos(line int) int {
	var pos int

	if line > 0 {
		b.rope.IndexAllFunc(0, b.rope.Len(), []byte{'\n'}, func(idx int) bool {
			line--
			pos = idx + 1    // idx+1 = start of line after delimiter
			return line <= 0 // If pos is now the start of the line we're searching for
		})
	}

	if line > 0 { // If there aren't enough lines to reach line...
		panic("getLineStartPos: not enough lines in buffer to reach position")
	}

	return pos
}

// RunesInLineWithDelim returns the number of runes in the given line. That is, the
// number of Utf-8 codepoints in the line, not bytes. Includes the line delimiter
// in the count. If that line delimiter is CRLF ('\r\n'), then it adds two.
func (b *RopeBuffer) RunesInLineWithDelim(line int) int {
	linePos := b.getLineStartPos(line)

	ropeLen := b.rope.Len()

	if linePos >= ropeLen {
		return 0
	}

	var count int

	_, r := b.rope.SplitAt(linePos)
	l, _ := r.SplitAt(ropeLen - linePos)

	l.EachLeaf(func(n *ropes.Node) bool {
		data := n.Value() // Reference; not a copy.
		for _, r := range string(data) {
			count++ // Before: we count the line delimiter
			if r == '\n' {
				return true // Read (past-tense) the whole line
			}
		}
		return false // Have not read the whole line, yet
	})

	return count
}

// RunesInLine returns the number of runes in the given line. That is, the
// number of Utf-8 codepoints in the line, not bytes. Excludes line delimiters.
func (b *RopeBuffer) RunesInLine(line int) int {
	linePos := b.getLineStartPos(line)

	ropeLen := b.rope.Len()

	if linePos >= ropeLen {
		return 0
	}

	var count int

	_, r := b.rope.SplitAt(linePos)
	l, _ := r.SplitAt(ropeLen - linePos)

	var isCRLF bool
	l.EachLeaf(func(n *ropes.Node) bool {
		data := n.Value() // Reference; not a copy.
		for _, r := range string(data) {
			if r == '\r' {
				isCRLF = true
			} else if r == '\n' {
				return true // Read (past-tense) the whole line
			} else {
				if isCRLF {
					isCRLF = false
					count++ // Add the '\r' we previously thought was part of the delim.
				}
			}
			count++
		}
		return false // Have not read the whole line, yet
	})

	return count
}

// ClampLineCol is a utility function to clamp any provided line and col to
// only possible values within the buffer, pointing to runes. It first clamps
// the line, then clamps the column. The column is clamped between zero and
// the last rune before the line delimiter.
func (b *RopeBuffer) ClampLineCol(line, col int) (int, int) {
	if line < 0 {
		line = 0
	} else if lines := b.Lines() - 1; line > lines {
		line = lines
	}

	if col < 0 {
		col = 0
	} else if runes := b.RunesInLine(line); col > runes {
		col = runes
	}

	return line, col
}

// PosToLineCol converts a byte offset (position) of the buffer's bytes, into
// a line and column. Unless you are working with the Bytes() function, this
// is unlikely to be useful to you. Position will be clamped.
func (b *RopeBuffer) PosToLineCol(pos int) (int, int) {
	var line, col int
	var wasAtNewline bool

	if pos <= 0 {
		return line, col
	}

	b.rope.EachLeaf(func(n *ropes.Node) bool {
		data := n.Value()
		var i int
		for i < len(data) {
			if wasAtNewline { // Start of line
				if data[i] != '\n' { // If the start of this line does not happen to be a delim...
					wasAtNewline = false // Say we weren't previously at a delimiter
				}
				line, col = line+1, 0
			} else if data[i] == '\n' { // End of line
				wasAtNewline = true
				col++
			} else {
				col++
			}

			_, size := utf8.DecodeRune(data[i:])
			i += size
			pos -= size

			if pos < 0 {
				return true
			}
		}
		return false
	})

	return line, col
}

func (b *RopeBuffer) WriteTo(w io.Writer) (int64, error) {
	return b.rope.WriteTo(w)
}

// Currently meant for the Remove function: imagine if the removed region passes through a Cursor position.
// We want to shift the cursor to the start of the region, based upon where the cursor position is.
func (b *RopeBuffer) shiftAnchorsRemovedRange(startPos, startLine, startCol, endLine, endCol int) {
	for _, v := range b.anchors {
		cursorLine, cursorCol := v.GetLineCol()
		if cursorLine >= startLine && cursorLine <= endLine {
			// If the anchor is not within the start or end columns
			if (cursorLine == startLine && cursorCol < startCol) || (cursorLine == endLine && cursorCol > endCol) {
				continue
			}
			cursorPos := b.LineColToPos(cursorLine, cursorCol)
			v.line, v.col = b.PosToLineCol(cursorPos + (startPos - cursorPos))
		}
	}
}

func (b *RopeBuffer) shiftAnchors(insertLine, insertCol, runeCount int) {
	for _, v := range b.anchors {
		cursorLine, cursorCol := v.GetLineCol()
		if insertLine < cursorLine || (insertLine == cursorLine && insertCol <= cursorCol) {
			v.line, v.col = b.PosToLineCol(b.LineColToPos(cursorLine, cursorCol) + runeCount)
		}
	}
}

// RegisterCursor adds the Cursor to a slice which the Buffer uses to update
// each Cursor based on changes that occur in the Buffer. Various functions are
// called on the Cursor depending upon where the edits occurred and how it should
// modify the Cursor's position. Unregister a Cursor before deleting it from
// memory, or forgetting it, with UnregisterPosition.
func (b *RopeBuffer) RegisterCursor(cursor *Cursor) {
	if cursor == nil {
		return
	}
	b.anchors = append(b.anchors, cursor)
}

// UnregisterCursor will remove the cursor from the list of watched Cursors.
// It is mandatory that a Cursor be unregistered before being freed from memory,
// or otherwise being forgotten.
func (b *RopeBuffer) UnregisterCursor(cursor *Cursor) {
	for i, v := range b.anchors {
		if cursor == v {
			// Delete item at i without preserving order
			b.anchors[i] = b.anchors[len(b.anchors)-1]
			b.anchors[len(b.anchors)-1] = nil
			b.anchors = b.anchors[:len(b.anchors)-1]
		}
	}
}