all: fix parsing literal paragraph

A line that start with one or more space will be considered
as literal paragraph.
Once literal line found it will end until an empty line.

1 files changed, 59 insertions, 12 deletions

diff --git a/document_parser.go b/document_parser.go
index 8de91cc..18c0817 100644
--- a/document_parser.go
+++ b/document_parser.go
@@ -11,6 +11,13 @@ import (
 	"git.sr.ht/~shulhan/pakakeh.go/lib/ascii"
 )
 
+// From the [Document.parse] its all start from parseBlock.
+//
+//  -> parseBlock()
+//     -> line()
+//        -> whatKindOfLine()
+//     The parsing branched here based on docp.kind.
+
 const debugLevel = 0
 
 type documentParser struct {
@@ -133,6 +140,48 @@ func (docp *documentParser) consumeLinesUntil(el *element, term int, terms []int
 	return line
 }
 
+// consumeLiteralParagraph consumes lines that start with spaces and returns
+// the next line that does not have it.
+func (docp *documentParser) consumeLiteralParagraph(el *element,
+	origSpaces []byte, terms ...int,
+) (line []byte) {
+	logp := `consumeLiteralParagraph`
+	lenOrigSpaces := len(origSpaces)
+	var spaces []byte
+	var ok bool
+	for {
+		spaces, line, ok = docp.line(logp)
+		if !ok {
+			break
+		}
+		if bytes.HasPrefix(spaces, origSpaces) {
+			el.Write(spaces[lenOrigSpaces:])
+			el.Write(line)
+			el.WriteByte('\n')
+			continue
+		}
+		if len(spaces) != 0 {
+			// If start with spaces, remove single space only and
+			// append the rest.
+			el.Write(spaces[1:])
+			el.Write(line)
+			el.WriteByte('\n')
+			continue
+		}
+		for _, kind := range terms {
+			if kind == docp.kind {
+				goto out
+			}
+		}
+		// Keep consume lines until we found terminating conditions.
+		el.Write(line)
+		el.WriteByte('\n')
+	}
+out:
+	el.raw = applySubstitutions(docp.doc, el.raw)
+	return line
+}
+
 // hasPreamble will return true if the contents contains preamble, indicated
 // by the first section that found after current line.
 func (docp *documentParser) hasPreamble() bool {
@@ -285,12 +334,13 @@ func (docp *documentParser) parseBlock(parent *element, term int) {
 		el   = &element{}
 
 		line   []byte
+		spaces []byte
 		isTerm bool
 		ok     bool
 	)
 	for !isTerm {
 		if len(line) == 0 {
-			_, line, ok = docp.line(logp)
+			spaces, line, ok = docp.line(logp)
 			if !ok {
 				return
 			}
@@ -514,17 +564,14 @@ func (docp *documentParser) parseBlock(parent *element, term int) {
 				el.addRole(classNameLiteralBlock)
 				el.Write(line)
 				el.WriteByte('\n')
-				line = docp.consumeLinesUntil(
-					el,
-					lineKindEmpty,
-					[]int{
-						term,
-						elKindBlockListing,
-						elKindBlockListingNamed,
-						elKindBlockLiteral,
-						elKindBlockLiteralNamed,
-					})
-				el.raw = applySubstitutions(docp.doc, el.raw)
+				line = docp.consumeLiteralParagraph(el,
+					spaces, lineKindEmpty,
+					term,
+					elKindBlockListing,
+					elKindBlockListingNamed,
+					elKindBlockLiteral,
+					elKindBlockLiteralNamed,
+				)
 			}
 			parent.addChild(el)
 			el = &element{}