From 0055be63c9e12fa7ac30878c418a839267f630b2 Mon Sep 17 00:00:00 2001
From: Tim Vergenz <vergenzt@gmail.com>
Date: Fri, 11 Oct 2024 13:12:27 -0400
Subject: [PATCH] feat: add |||- chomped text block syntax

Resolves google/jsonnet#289

Companion PR to google/jsonnet#1175
---
 internal/formatter/fix_indentation.go           |  2 +-
 internal/formatter/unparser.go                  |  9 ++++++++-
 internal/parser/lexer.go                        | 17 +++++++++++++++--
 testdata/block_string_chomped.golden            |  1 +
 testdata/block_string_chomped.jsonnet           |  5 +++++
 testdata/block_string_chomped.linter.golden     |  0
 testdata/block_string_chomped_concatted.golden  |  1 +
 testdata/block_string_chomped_concatted.jsonnet |  5 +++++
 ...block_string_chomped_concatted.linter.golden |  0
 9 files changed, 36 insertions(+), 4 deletions(-)
 create mode 100644 testdata/block_string_chomped.golden
 create mode 100644 testdata/block_string_chomped.jsonnet
 create mode 100644 testdata/block_string_chomped.linter.golden
 create mode 100644 testdata/block_string_chomped_concatted.golden
 create mode 100644 testdata/block_string_chomped_concatted.jsonnet
 create mode 100644 testdata/block_string_chomped_concatted.linter.golden

diff --git a/internal/formatter/fix_indentation.go b/internal/formatter/fix_indentation.go
index 90229d197..103f95b50 100644
--- a/internal/formatter/fix_indentation.go
+++ b/internal/formatter/fix_indentation.go
@@ -685,7 +685,7 @@ func (c *FixIndentation) Visit(expr ast.Node, currIndent indent, crowded bool) {
 			node.BlockIndent = strings.Repeat(" ", currIndent.base+c.Options.Indent)
 			node.BlockTermIndent = strings.Repeat(" ", currIndent.base)
 			c.column = currIndent.base // blockTermIndent
-			c.column += 3              // "|||"
+			c.column += 3              // always "|||" (never "|||-" because we're only accounting for block end)
 		case ast.VerbatimStringSingle:
 			c.column += 3 // Include @, start and end quotes
 			for _, r := range node.Value {
diff --git a/internal/formatter/unparser.go b/internal/formatter/unparser.go
index d3e92cdfb..d9d3a1ba4 100644
--- a/internal/formatter/unparser.go
+++ b/internal/formatter/unparser.go
@@ -467,7 +467,11 @@ func (u *unparser) unparse(expr ast.Node, crowded bool) {
 			u.write(node.Value)
 			u.write("'")
 		case ast.StringBlock:
-			u.write("|||\n")
+			u.write("|||")
+			if node.Value[len(node.Value)-1] != '\n' {
+				u.write("-")
+			}
+			u.write("\n")
 			if node.Value[0] != '\n' {
 				u.write(node.BlockIndent)
 			}
@@ -481,6 +485,9 @@ func (u *unparser) unparse(expr ast.Node, crowded bool) {
 					u.write(node.BlockIndent)
 				}
 			}
+			if node.Value[len(node.Value)-1] != '\n' {
+				u.write("\n")
+			}
 			u.write(node.BlockTermIndent)
 			u.write("|||")
 		case ast.VerbatimStringDouble:
diff --git a/internal/parser/lexer.go b/internal/parser/lexer.go
index 374343625..0b3437263 100644
--- a/internal/parser/lexer.go
+++ b/internal/parser/lexer.go
@@ -719,6 +719,13 @@ func (l *lexer) lexSymbol() error {
 	if r == '|' && strings.HasPrefix(l.input[l.pos.byteNo:], "||") {
 		commentStartLoc := l.tokenStartLoc
 		l.acceptN(2) // Skip "||"
+
+		var chompTrailingNl bool = false
+		if l.peek() == '-' {
+			chompTrailingNl = true
+			l.next()
+		}
+
 		var cb bytes.Buffer
 
 		// Skip whitespace
@@ -775,7 +782,13 @@ func (l *lexer) lexSymbol() error {
 					return l.makeStaticErrorPoint("Text block not terminated with |||", commentStartLoc)
 				}
 				l.acceptN(3) // Skip '|||'
-				l.emitFullToken(tokenStringBlock, cb.String(),
+
+				var str string = cb.String()
+				if chompTrailingNl {
+					str = str[:len(str)-1]
+				}
+
+				l.emitFullToken(tokenStringBlock, str,
 					stringBlockIndent, stringBlockTermIndent)
 				l.resetTokenStart()
 				return nil
@@ -793,7 +806,7 @@ func (l *lexer) lexSymbol() error {
 		if r == '/' && strings.HasPrefix(l.input[l.pos.byteNo:], "*") {
 			break
 		}
-		// Not allowed ||| in operators
+		// Not allowed ||| in operators (accounts for |||-)
 		if r == '|' && strings.HasPrefix(l.input[l.pos.byteNo:], "||") {
 			break
 		}
diff --git a/testdata/block_string_chomped.golden b/testdata/block_string_chomped.golden
new file mode 100644
index 000000000..1d0a12680
--- /dev/null
+++ b/testdata/block_string_chomped.golden
@@ -0,0 +1 @@
+"foo\nbar\nbaz"
diff --git a/testdata/block_string_chomped.jsonnet b/testdata/block_string_chomped.jsonnet
new file mode 100644
index 000000000..559901f51
--- /dev/null
+++ b/testdata/block_string_chomped.jsonnet
@@ -0,0 +1,5 @@
+|||-
+  foo
+  bar
+  baz
+|||
diff --git a/testdata/block_string_chomped.linter.golden b/testdata/block_string_chomped.linter.golden
new file mode 100644
index 000000000..e69de29bb
diff --git a/testdata/block_string_chomped_concatted.golden b/testdata/block_string_chomped_concatted.golden
new file mode 100644
index 000000000..47dc09463
--- /dev/null
+++ b/testdata/block_string_chomped_concatted.golden
@@ -0,0 +1 @@
+"foo bar baz all one line"
diff --git a/testdata/block_string_chomped_concatted.jsonnet b/testdata/block_string_chomped_concatted.jsonnet
new file mode 100644
index 000000000..db4b69103
--- /dev/null
+++ b/testdata/block_string_chomped_concatted.jsonnet
@@ -0,0 +1,5 @@
+|||-
+  foo bar baz 
+||| + |||-
+  all one line
+|||
diff --git a/testdata/block_string_chomped_concatted.linter.golden b/testdata/block_string_chomped_concatted.linter.golden
new file mode 100644
index 000000000..e69de29bb