From 678459bc35a19d4379bf82268162ddfd8b23f509 Mon Sep 17 00:00:00 2001
From: yoshikazusawa <883514+yoshikazusawa@users.noreply.github.com>
Date: Mon, 10 Jun 2024 00:23:15 +0900
Subject: [PATCH] Supports additional indicators on block scalars

ref: https://yaml-multiline.info/
---
 lib/TAP/Parser/YAMLish/Reader.pm | 76 +++++++++++++++++++++++-------
 t/yamlish.t                      | 80 ++++++++++++++++++++++++++++++--
 2 files changed, 135 insertions(+), 21 deletions(-)

diff --git a/lib/TAP/Parser/YAMLish/Reader.pm b/lib/TAP/Parser/YAMLish/Reader.pm
index fb823784..6f09de9d 100644
--- a/lib/TAP/Parser/YAMLish/Reader.pm
+++ b/lib/TAP/Parser/YAMLish/Reader.pm
@@ -137,23 +137,9 @@ sub _read_scalar {
     return {} if $string eq '{}';
     return [] if $string eq '[]';
 
-    if ( $string eq '>' || $string eq '|' ) {
-
-        my ( $line, $indent ) = $self->_peek;
-        die "Multi-line scalar content missing" unless defined $line;
-
-        my @multiline = ($line);
-
-        while (1) {
-            $self->_next;
-            my ( $next, $ind ) = $self->_peek;
-            last if $ind < $indent;
-
-            my $pad = $string eq '|' ? ( ' ' x ( $ind - $indent ) ) : '';
-            push @multiline, $pad . $next;
-        }
-
-        return join( ( $string eq '>' ? ' ' : "\n" ), @multiline ) . "\n";
+    if ( $string =~ /^([>|])([+-]?)([1-9]?)$/ ) {
+        my ( $style, $chomping, $indent_base ) = ( $1, $2, $3 );
+        return $self->_read_block_scalar( $style, $chomping, $indent_base );
     }
 
     if ( $string =~ /^ ' (.*) ' $/x ) {
@@ -175,6 +161,62 @@ sub _read_scalar {
     return $string;
 }
 
+sub _read_block_scalar {
+    my ( $self, $style, $chomping, $indent_base ) = @_;
+
+    my ( $line, $line_indent ) = $self->_peek;
+    die "Multi-line scalar content missing" unless defined $line;
+
+    $indent_base ||= $line_indent;
+
+    my $pad = ' ' x ( $line_indent - $indent_base );
+    my @multi_lines = [ $pad, $line ];
+
+    while (1) {
+        $self->_next;
+        my ( $content, $line_indent ) = $self->_peek;
+        last if $line_indent < $indent_base;
+
+        my $pad = ' ' x ( $line_indent - $indent_base );
+        push @multi_lines, [ $pad, $content ];
+    }
+
+    my $block = '';
+    my $previous_line;
+
+    for my $current_line ( @multi_lines ) {
+        my ( $pad, $content ) = @$current_line;
+        unless ( defined $previous_line ) {
+            $block .= join( '', $pad, $content );
+            $previous_line = $current_line;
+            next;
+        }
+
+        if ( $style eq '>'
+             && length $content > 0
+             && length $pad == 0
+             && length $previous_line->[0] == 0
+             && length $previous_line->[1] > 0 )
+        {
+            $block .= ' ' . $content;
+        } else {
+            $block .= join( '', "\n", $pad, $content );
+        }
+        $previous_line = $current_line;
+    }
+
+    $block .= "\n";
+
+    if ( $chomping eq '-' ) {
+        $block =~ s/\n+$//;
+    } elsif ( $chomping eq '+' ) {
+        # noop: keep newlines
+    } else {
+         $block =~ s/\n+$/\n/;
+    }
+    return $block;
+}
+
 sub _read_nested {
     my $self = shift;
 
diff --git a/t/yamlish.t b/t/yamlish.t
index 07c36838..07bdb01b 100644
--- a/t/yamlish.t
+++ b/t/yamlish.t
@@ -40,16 +40,50 @@ BEGIN {
             ],
             out => "Hello, World",
         },
-        {   name => 'Hello World 4',
+        {   name => 'Block Scalars with folded style',
             in   => [
                 '--- >',
                 '   Hello,',
-                '      World',
+                '   World',
                 '...',
             ],
             out => "Hello, World\n",
         },
-        {   name => 'Hello World Block',
+        {   name => 'Block Scalars with folded style including extra indentation',
+            in   => [
+                '--- >',
+                '   Hello,',
+                '   World',
+                '     with extra',
+                '     indentation',
+                '...',
+            ],
+            out => "Hello, World\n  with extra\n  indentation\n",
+        },
+        {   name => 'Hello World with folded style and no newline at end',
+            in   => [
+                '--- >-',
+                '   Hello,',
+                '   World',
+                '   ',
+                '...',
+            ],
+            out => "Hello, World",
+        },
+        {   name => 'Hello World with folded style and keeping newlines from end',
+            in   => [
+                '--- >+',
+                '   Hello,',
+                '   World',
+                '     ',
+                '   ',
+                '   Hello World again',
+                '   ',
+                '...',
+            ],
+            out => "Hello, World\n  \n\nHello World again\n\n",
+        },
+        {   name => 'Hello World with literal style',
             in   => [
                 '--- |',
                 '   Hello,',
@@ -58,7 +92,45 @@ BEGIN {
             ],
             out => "Hello,\n   World\n",
         },
-        {   name => 'Hello World 5',
+        {   name => 'Hello World with literal style and no newline at end',
+            in   => [
+                '--- |-',
+                '   Hello,',
+                '      World',
+                '...',
+            ],
+            out => "Hello,\n   World",
+        },
+        {   name => 'Hello World with literal style and keeping newlines from end',
+            in   => [
+                '--- |+',
+                '   Hello,',
+                '      World',
+                '   ',
+                '    ',
+                '...',
+            ],
+            out => "Hello,\n   World\n\n \n",
+        },
+        {   name => 'Hello World with literal style and no newline at end and indentation indicator',
+            in   => [
+                '--- |-3',
+                '      Hello,',
+                '    World',
+                '...',
+            ],
+            out => "   Hello,\n World",
+        },
+        {   name => 'Hello World with indentation indicator',
+            in   => [
+                '--- >1',
+                '   Hello,',
+                '  World',
+                '...',
+            ],
+            out => "  Hello,\n World\n",
+        },
+        {   name => 'Hello World with broken indentation',
             in   => [
                 '--- >',
                 '   Hello,',