refactor: Use '-N1' for reading characters from input

This prevents a lot of the truncating and mangling errors that
were previously occuring. This feature was added in Bash 4.1

Author: hyperupcall

pkg/lib/commands/parse.sh

@@ -9,7 +9,7 @@ bash_toml.do_parse() {
 	declare -i PARSER_LINE_NUMBER=1
 	declare -i PARSER_COLUMN_NUMBER=0
 
-	while IFS= read -rn 1 char; do
+	while IFS= read -rN1 char; do
 		if bash_toml.is.newline "$char"; then
 			PARSER_COLUMN_NUMBER=0
 			PARSER_LINE_NUMBER+=1
@@ -55,16 +55,14 @@ bash_toml.do_parse() {
 		MODE_ANY_BEFORE_VALUE)
 			if bash_toml.is.whitespace "$char"; then
 				:
-			elif bash_toml.is.newline "$char"; then
-				# TODO: not being fired?
-				bash_toml.parse_fail 'KEY_ABSENT' "Key name found without value 2"
-				return 1
 			elif bash_toml.is.double_quote "$char"; then
 				mode='MODE_DOUBLEQUOTE_DURING_VALUE'
 				bash_toml.init_value_string
 			elif bash_toml.is.single_quote "$char"; then
 				mode='MODE_SINGLEQUOTE_DURING_VALUE'
 				bash_toml.init_value_string
+			elif bash_toml.is.newline "$char"; then
+				bash_toml.parse_fail 'UNEXPECTED_NEWLINE' 'Expected to find value on the same line'
 			elif bash_toml.is.empty "$char"; then
 				bash_toml.parse_fail 'UNEXPECTED_EOF' 'Expected to find value on the same line'
 				return 1
@@ -120,15 +118,15 @@ bash_toml.do_parse() {
 		# directly after the `"`
 		MODE_DOUBLEQUOTE_DURING_VALUE)
 			if bash_toml.is.double_quote "$char"; then
-				mode='MODE_ANY_AFTER_VALUE'
+				mode='MODE_DEFAULT_END'
 			elif bash_toml.is.backslash "$char"; then
 				mode='MODE_DOUBLEQUOTE_DURING_ESCAPE_SEQUENCE'
+			elif bash_toml.is.newline "$char"; then
+				bash_toml.parse_fail 'UNEXPECTED_NEWLINE' "Literal newlines must not be present in double quotes"
+				return 1
 			elif bash_toml.is.control_character "$char"; then
 				# TODO: this code path won't get activated
 				:
-			elif bash_toml.is.empty "$char"; then
-				bash_toml.parse_fail 'UNEXPECTED_EOF' "Must complete the basic string with a double quote"
-				return 1
 			else
 				bash_toml.append_value_string "$char"
 			fi
@@ -224,9 +222,9 @@ bash_toml.do_parse() {
 		# directly after the `'`
 		MODE_SINGLEQUOTE_DURING_VALUE)
 			if bash_toml.is.single_quote "$char"; then
-				mode='MODE_ANY_AFTER_VALUE'
+				mode='MODE_DEFAULT_END'
 			elif bash_toml.is.newline "$char"; then
-				bash_toml.parse_fail 'VALUE_STRING_INVALID' "Newlines are not valid in single quote"
+				bash_toml.parse_fail 'UNEXPECTED_NEWLINE' "Newlines are not valid in single quote"
 				return 1
 			elif bash_toml.is.empty "$char"; then
 				bash_toml.parse_fail 'UNEXPECTED_EOF' "Must complete the literal string with a single quote"
@@ -235,7 +233,7 @@ bash_toml.do_parse() {
 				bash_toml.append_value_string "$char"
 			fi
 			;;
-		MODE_ANY_AFTER_VALUE)
+		MODE_DEFAULT_END)
 			if bash_toml.is.whitespace "$char"; then
 				:
 			elif bash_toml.is.newline "$char"; then
@@ -252,6 +250,14 @@ bash_toml.do_parse() {
 		esac
 	done
 
+	case "$mode" in
+		MODE_DEFAULT|MODE_DEFAULT_END)
+			;;
+		*)
+			bash_toml.parse_fail 'UNEXPECTED_EOF' 'Did not finish parsing construct'
+			return 1
+			;;
+	esac
 	# If we try to set an empty key with a value, then later on,
 	# we can access any value (using a non-integer key), and we
 	# will get a result that equals the original value value

pkg/lib/util/util.sh

@@ -7,6 +7,7 @@ declare -gA BASH_TOML_ERRORS=(
 	[UNICODE_INVALID]='The resulting unicode code point was invalid'
 	[KEY_ABSENT]='Key does not have a value'
 	[UNEXPECTED_EOF]='Unexpected end of line'
+	[UNEXPECTED_NEWLINE]='Unexpected newline'
 	[UNEXPECTED_CHARACTER]='An unexpected character was encountered' # Generalization of any of the following errors
 	[KEY_INVALID]='The key is not valid'
 	[VALUE_INVALID]='The value could not be parsed'

tests/key-bare.bats

@@ -2,42 +2,48 @@
 
 load './util/init.sh'
 
-@test "fails on invalid bare key" {
-	run bash_toml.do_parse <<-"EOF"
-	fox
-	EOF
+@test "fails on invalid bare key 1" {
+	run bash_toml.do_parse < <(printf 'fox')
 
 	assert_failure
-	assert_output -p 'KEY_INVALID'
+	assert_output -p 'UNEXPECTED_EOF'
 }
 
 @test "fails on invalid bare key 2" {
-	run bash_toml.do_parse <<-"EOF"
-	f ox
-	EOF
+	run bash_toml.do_parse < <(printf 'fox\n')
 
 	assert_failure
 	assert_output -p 'KEY_INVALID'
 }
 
 @test "fails on invalid bare key 3" {
-	run bash_toml.do_parse <<-"EOF"
-	f!ox
-	EOF
+	run bash_toml.do_parse < <(printf 'f ox')
 
 	assert_failure
-	assert_output -p 'UNEXPECTED_CHARACTER'
+	assert_output -p 'KEY_INVALID'
 }
 
 @test "fails on invalid bare key 4" {
-	run bash_toml.do_parse <<-"EOF"
-	fox=
-	EOF
+	run bash_toml.do_parse < <(printf 'f!ox')
+
+	assert_failure
+	assert_output -p 'UNEXPECTED_CHARACTER'
+}
+
+@test "fails on invalid bare key 5" {
+	run bash_toml.do_parse < <(printf 'fox=')
 
 	assert_failure
 	assert_output -p 'UNEXPECTED_EOF'
 }
 
+@test "fails on invalid bare key 6" {
+	run bash_toml.do_parse < <(printf 'fox=\n')
+
+	assert_failure
+	assert_output -p 'UNEXPECTED_NEWLINE'
+}
+
 @test "succeeds on valid bare key 1" {
 	bash_toml.do_parse <<-"EOF"
 	fox = 'value'

tests/key-quoted.bats

@@ -2,55 +2,75 @@
 
 load './util/init.sh'
 
-@test "fails on invalid quoted key" {
-	run bash_toml.do_parse <<< '"'
+@test "fails on invalid quoted key 1" {
+	run bash_toml.do_parse < <(printf '"')
 
 	assert_failure
-	assert_output -p 'KEY_INVALID'
+	assert_output -p 'UNEXPECTED_EOF'
 }
 
 @test "fails on invalid quoted key 2" {
-	run bash_toml.do_parse <<< $'"\n'
+	run bash_toml.do_parse < <(printf '"\n')
 
 	assert_failure
 	assert_output -p 'KEY_INVALID'
 }
 
 @test "fails on invalid quoted key 3" {
-	run bash_toml.do_parse <<< '""'
+	run bash_toml.do_parse < <(printf '""')
 
 	assert_failure
 	assert_output -p 'UNEXPECTED_EOF'
 }
 
 @test "fails on invalid quoted key 4" {
-	run bash_toml.do_parse <<< '""='
+	run bash_toml.do_parse < <(printf '""\n')
 
 	assert_failure
-	assert_output -p 'UNEXPECTED_EOF'
+	assert_output -p 'KEY_INVALID'
 }
 
 @test "fails on invalid quoted key 5" {
-	run bash_toml.do_parse <<< '"" ='
+	run bash_toml.do_parse < <(printf '""=')
 
 	assert_failure
 	assert_output -p 'UNEXPECTED_EOF'
 }
 
 @test "fails on invalid quoted key 6" {
-	run bash_toml.do_parse <<< \"\"\ =\'
+	run bash_toml.do_parse < <(printf '""=\n')
+
+	assert_failure
+	assert_output -p 'UNEXPECTED_NEWLINE'
+}
+
+@test "fails on invalid quoted key 7" {
+	run bash_toml.do_parse < <(printf '"" =')
 
 	assert_failure
 	assert_output -p 'UNEXPECTED_EOF'
 }
 
+@test "fails on invalid quoted key 8" {
+	run bash_toml.do_parse < <(printf '"" =\n')
+
+	assert_failure
+	assert_output -p 'UNEXPECTED_NEWLINE'
+}
+
 @test "succeeds on valid quoted key 1" {
 	bash_toml.do_parse <<-"EOF"
-	"" = 'UwU'
+	"" = ''
 	EOF
 }
 
 @test "succeeds on valid quoted key 2" {
+	bash_toml.do_parse <<-"EOF"
+	"" = 'UwU'
+	EOF
+}
+
+@test "succeeds on valid quoted key 3" {
 	bash_toml.do_parse <<-"EOF"
 	"fox" = 'UwU'
 	EOF
@@ -59,7 +79,7 @@ load './util/init.sh'
 	assert test_util.toml.key_has_value 'fox' 'UwU'
 }
 
-@test "succeeds on valid quoted key 3" {
+@test "succeeds on valid quoted key 4" {
 	bash_toml.do_parse <<-"EOF"
 	"123" = 'UwU'
 	EOF

tests/string-basic.bats

@@ -3,39 +3,55 @@
 load './util/init.sh'
 
 @test "fails on invalid basic string 1" {
-	run bash_toml.do_parse <<-"EOF"
-	fox = "
-	EOF
+	run bash_toml.do_parse < <(printf 'fox = "')
 
 	assert_failure
 	assert_output -p "UNEXPECTED_EOF"
 }
 
 @test "fails on invalid basic string 2" {
-	run bash_toml.do_parse <<-"EOF"
-	fox = "a
-	EOF
+	run bash_toml.do_parse < <(printf 'fox = "\n')
 
 	assert_failure
-	assert_output -p "UNEXPECTED_EOF"
+	assert_output -p "UNEXPECTED_NEWLINE"
 }
 
-# TODO: this shouldn't be
 @test "fails on invalid basic string 3" {
-	run bash_toml.do_parse <<< 'fox = "ab'
+	run bash_toml.do_parse < <(printf 'fox = "a')
 
 	assert_failure
 	assert_output -p "UNEXPECTED_EOF"
 }
 
 @test "fails on invalid basic string 4" {
+	run bash_toml.do_parse < <(printf 'fox = "a\n')
+
+	assert_failure
+	assert_output -p "UNEXPECTED_NEWLINE"
+}
+
+@test "fails on invalid basic string 5" {
+	run bash_toml.do_parse < <(printf 'fox = "ab')
+
+	assert_failure
+	assert_output -p "UNEXPECTED_EOF"
+}
+
+@test "fails on invalid basic string 6" {
+	run bash_toml.do_parse < <(printf 'fox = "ab\n')
+
+	assert_failure
+	assert_output -p "UNEXPECTED_NEWLINE"
+}
+
+@test "fails on invalid basic string 7" {
 	run bash_toml.do_parse <<< 'fox = "\j"'
 
 	assert_failure
 	assert_output -p "UNEXPECTED_CHARACTER"
 }
 
-@test "fails on invalid basic string 5" {
+@test "fails on invalid basic string 8" {
 	for hex in D800 DFFF; do
 		run bash_toml.do_parse <<< "fox = "\"\\u"$hex"\"
 
@@ -44,7 +60,7 @@ load './util/init.sh'
 	done
 }
 
-@test "fails on invalid basic string 6" {
+@test "fails on invalid basic string 9" {
 	run bash_toml.do_parse <<-"EOF"
 	fox = "\U00110000"
 	EOF

tests/string-literal.bats

@@ -1,24 +1,34 @@
 load './util/init.sh'
 
 @test "fails on invalid literal string 1" {
-	run bash_toml.do_parse <<-"EOF"
-	fox = '
-	EOF
+	run bash_toml.do_parse < <(printf "fox = '")
 
 	assert_failure
 	assert_output -p "UNEXPECTED_EOF"
 }
 
 @test "fails on invalid literal string 2" {
-	run bash_toml.do_parse <<-"EOF"
-	fox = 'a
-	EOF
+	run bash_toml.do_parse < <(printf "fox = '\n")
 
 	assert_failure
-	assert_output -p "UNEXPECTED_EOF"
+	assert_output -p "UNEXPECTED_NEWLINE"
 }
 
 @test "fails on invalid literal string 3" {
+	run bash_toml.do_parse < <(printf "fox = 'a")
+
+	assert_failure
+	assert_output -p "UNEXPECTED_EOF"
+}
+
+@test "fails on invalid literal string 4" {
+	run bash_toml.do_parse < <(printf "fox = 'a\n")
+
+	assert_failure
+	assert_output -p "UNEXPECTED_NEWLINE"
+}
+
+@test "fails on invalid literal string 5" {
 	run bash_toml.do_parse <<-"EOF"
 	fox = '''
 	EOF