ffs, i should have committed wayyy earlier

parser/blocks.go

@@ -0,0 +1,97 @@
+package parser
+
+type BlockType string
+
+const (
+	BlockData     BlockType = "data"     // content between lines "{" and "}"
+	BlockMatching BlockType = "matching" // everything outside data blocks
+)
+
+type Block struct {
+	Type       BlockType
+	Start, End int // byte offsets [Start, End)
+	src        *string
+}
+
+func (b Block) GetContent() string {
+	if b.src == nil || b.Start < 0 || b.End > len(*b.src) || b.Start > b.End {
+		return ""
+	}
+	return (*b.src)[b.Start:b.End]
+}
+
+// ExtractBlocks scans once, emitting:
+// - data blocks: inner content between a line that's exactly "{" and a line that's exactly "}"
+// - matching blocks: gaps between data blocks (excluding the brace lines themselves)
+func ExtractBlocks(src string) []Block {
+	var out []Block
+	var curlyIndex int
+
+	const CLOSING = '}'
+	const OPENING = '{'
+
+	if len(src) > 0 && src[0] == OPENING {
+		curlyIndex = 1
+		out = append(out, Block{
+			Start: 0,
+			Type:  BlockData,
+			src:   &src,
+		})
+	} else {
+		out = append(out, Block{
+			Start: 0,
+			Type:  BlockMatching,
+			src:   &src,
+		})
+	}
+
+	for i, r := range src {
+
+		var nextCurlyIndex = curlyIndex
+
+		switch r {
+		case OPENING:
+			nextCurlyIndex++
+		case CLOSING:
+			nextCurlyIndex--
+		}
+
+		var nextChar rune = ' '
+		if i+1 < len(src) {
+			nextChar = rune(src[i+1])
+		}
+
+		if curlyIndex == 0 && nextCurlyIndex == 1 {
+			out[len(out)-1].End = i
+			out = append(out, Block{
+				Start: i,
+				Type:  BlockData,
+				src:   &src,
+			})
+		} else if curlyIndex == 1 && nextCurlyIndex == 0 {
+			out[len(out)-1].End = i + 1
+			if nextChar == OPENING {
+				out = append(out, Block{
+					Start: i + 1,
+					Type:  BlockData,
+					src:   &src,
+				})
+			} else {
+				out = append(out, Block{
+					Start: i + 1,
+					Type:  BlockMatching,
+					src:   &src,
+				})
+			}
+		}
+
+		curlyIndex = nextCurlyIndex
+	}
+
+	var lastBlock = out[len(out)-1]
+	if lastBlock.End == 0 {
+		out = out[:len(out)-1]
+	}
+
+	return out
+}