Merge pull request #63 from Kyorai/cuttlefish-60-multiline-values

Support multiline configuration values

Author: michaelklishin

README.md

@@ -98,6 +98,40 @@ log.console.file = /var/log/console.log
 log.syslog = on
 ```
 
+### Multiline Values
+
+Cuttlefish supports multiline values using here-document syntax with triple
+single quotes (`'''`). This is useful for embedding structured content like
+JSON, scripts, or formatted text:
+
+```ini
+# Simple multiline value
+description = '''
+This is a multiline description
+that spans multiple lines
+and preserves formatting.
+'''
+
+# JSON configuration example
+api.config = '''
+{
+  "endpoints": {
+    "health": "/health",
+    "metrics": "/metrics"
+  },
+  "timeout": 30,
+  "retries": 3
+}
+'''
+
+# Inline multiline (no newlines)
+short.value = '''single line content'''
+```
+
+Whitespace within the `'''` delimiters is preserved exactly as written,
+including newlines and indentation. Leading and trailing newlines and carriage
+returns are automatically trimmed to provide clean configuration values.
+
 More information for users here:
 https://github.com/basho/cuttlefish/wiki/Cuttlefish-for-Application-Users
 

src/conf_parse.erl

@@ -221,7 +221,7 @@ parse(Input) when is_binary(Input) ->
 
 -spec 'setting'(input(), index()) -> parse_result().
 'setting'(Input, Index) ->
-  p(Input, Index, 'setting', fun(I,D) -> (p_seq([p_zero_or_more(fun 'ws'/2), fun 'key'/2, p_zero_or_more(fun 'ws'/2), p_string(<<"=">>), p_zero_or_more(fun 'ws'/2), p_choose([fun 'escaped_value'/2, fun 'unescaped_value'/2]), p_zero_or_more(fun 'ws'/2), p_optional(fun 'comment'/2)]))(I,D) end, fun(Node, Idx) ->
+  p(Input, Index, 'setting', fun(I,D) -> (p_seq([p_zero_or_more(fun 'ws'/2), fun 'key'/2, p_zero_or_more(fun 'ws'/2), p_string(<<"=">>), p_zero_or_more(fun 'ws'/2), p_choose([fun 'multiline_value'/2, fun 'escaped_value'/2, fun 'unescaped_value'/2]), p_zero_or_more(fun 'ws'/2), p_optional(fun 'comment'/2)]))(I,D) end, fun(Node, Idx) ->
     [ _, Key, _, _Eq, _, Value, _, _ ] = Node,
     {Key, try_unicode_characters_to_list(Value, Idx)}
  end).
@@ -233,6 +233,13 @@ parse(Input) when is_binary(Input) ->
     [try_unicode_characters_to_list(H, Idx)| [try_unicode_characters_to_list(W, Idx) || [_, W] <- T]]
  end).
 
+-spec 'multiline_value'(input(), index()) -> parse_result().
+'multiline_value'(Input, Index) ->
+  p(Input, Index, 'multiline_value', fun(I,D) -> (p_seq([p_string(<<"\'\'\'">>), p_choose([p_zero_or_more(p_seq([p_not(p_string(<<"\'\'\'">>)), p_anything()])), fun 'crlf'/2]), p_string(<<"\'\'\'">>)]))(I,D) end, fun(Node, Idx) ->
+    Node0 = string:trim(Node, both, [$',$\n,[$\r,$\n]]),
+    try_unicode_characters_to_list(Node0, Idx)
+ end).
+
 -spec 'escaped_value'(input(), index()) -> parse_result().
 'escaped_value'(Input, Index) ->
   p(Input, Index, 'escaped_value', fun(I,D) -> (p_seq([p_string(<<"\'">>), p_zero_or_more(p_seq([p_not(p_string(<<"\'">>)), p_anything()])), p_string(<<"\'">>)]))(I,D) end, fun(Node, Idx) ->

src/conf_parse.peg

@@ -56,7 +56,7 @@ line <- ((setting / include / comment / ws+) (crlf / eof)) / crlf %{
 
 %% A setting is a key and a value, joined by =, with surrounding
 %% whitespace ignored.
-setting <- ws* key ws* "=" ws* (escaped_value / unescaped_value) ws* comment? %{
+setting <- ws* key ws* "=" ws* (multiline_value / escaped_value / unescaped_value) ws* comment? %{
     [ _, Key, _, _Eq, _, Value, _, _ ] = Node,
     {Key, try_unicode_characters_to_list(Value, Idx)}
 %};
@@ -67,6 +67,12 @@ key <- head:word tail:("." word)* %{
     [try_unicode_characters_to_list(H, Idx)| [try_unicode_characters_to_list(W, Idx) || [_, W] <- T]]
 %};
 
+%% A multiline value is any character between two pairs of triple single quotes, spanning lines
+multiline_value <- "'''" ((!"'''" .)* / crlf) "'''" %{
+    Node0 = string:trim(Node, both, [$',$\n,[$\r,$\n]]),
+    try_unicode_characters_to_list(Node0 Idx)
+%};
+
 %% An escaped value is any character between single quotes except for EOF
 escaped_value <- "'" (!"'" .)* "'" %{
     Stripped = string:trim(Node, both, [$']),

src/cuttlefish_conf.erl

@@ -351,6 +351,94 @@ generate_comments_test() ->
     Comments = generate_comments(SchemaElement),
     ?assertEqual(["## Hi!", "## Bye!", "## ", "## Acceptable values:", "##   - text"], Comments).
 
+normalize_newlines(Conf) ->
+    normalize_newlines(Conf, {os_type, os:type()}).
+
+normalize_newlines(Conf, {os_type, {win32, _}}) ->
+    {ok, Regex} = re:compile("\r\n", [unicode]),
+    F = fun (V) ->
+            re:replace(V, Regex, "\n", [global, {return, list}])
+        end,
+    normalize_newlines(Conf, F, []);
+normalize_newlines(Conf, {os_type, _}) ->
+    normalize_newlines(Conf, fun (V) -> V end, []).
+
+normalize_newlines([], _F, Acc) ->
+    Acc;
+normalize_newlines({errorlist, _}=Errorlist, _F, _Acc) ->
+    throw(Errorlist);
+normalize_newlines({error, _}=Error, _F, _Acc) ->
+    throw(Error);
+normalize_newlines([{KeyList, Value}|Rest], F, Acc0) ->
+    Acc1 = [{KeyList, F(Value)} | Acc0],
+    normalize_newlines(Rest, F, Acc1).
+
+multiline_test() ->
+    Conf = normalize_newlines(file("test/multiline.conf")),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"],"    l0\n    l1\n    l2"}], Conf).
+
+multiline_empty_test() ->
+    Conf = file("test/multiline_empty.conf"),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"],[]}], Conf).
+
+multiline_same_line_test() ->
+    Conf = file("test/multiline_same_line.conf"),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"],"value"}], Conf).
+
+multiline_special_chars_test() ->
+    Conf = normalize_newlines(file("test/multiline_special_chars.conf")),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"], "key = value\n# comment\n'quoted' = \"double\""}], Conf).
+
+multiline_unclosed_test() ->
+    Conf = file("test/multiline_unclosed.conf"),
+    ?assertMatch({errorlist, [{error, {conf_syntax, {"test/multiline_unclosed.conf", _}}}]}, Conf).
+
+multiline_embedded_delimiter_test() ->
+    Conf = file("test/multiline_embedded_delimiter.conf"),
+    ?assertMatch({errorlist, [{error, {conf_syntax, {"test/multiline_embedded_delimiter.conf", _}}}]}, Conf).
+
+multiline_with_trailing_comment_test() ->
+    Conf = normalize_newlines(file("test/multiline_with_trailing_comment.conf")),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"],"value"}], Conf).
+
+multiline_multiple_test() ->
+    Conf = normalize_newlines(file("test/multiline_multiple.conf")),
+    ?assertEqual(2, length(Conf)),
+    ?assertMatch([{["b"],"l3\nl4\nl5"},
+                  {["a"],"    l0\n    l1\n    l2"}], Conf).
+
+multiline_mixed_test() ->
+    Conf = normalize_newlines(file("test/multiline_mixed.conf")),
+    ?assertEqual(3, length(Conf)),
+    ?assertMatch([{["another"],"simple"},
+                  {["regular"],"value"},
+                  {["a"],"    l0\nl1\n    l2"}], Conf).
+
+multiline_whitespace_only_test() ->
+    Conf = normalize_newlines(file("test/multiline_whitespace_only.conf")),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"],"  \t\n\r  "}], Conf).
+
+multiline_with_include_test() ->
+    Conf = normalize_newlines(file("test/multiline_with_include.conf")),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"], "include other.conf\ninclude *.conf"}], Conf).
+
+multiline_eof_test() ->
+    Conf = file("test/multiline_eof.conf"),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["a"], "value"}], Conf).
+
+multiline_unicode_test() ->
+    Conf = normalize_newlines(file("test/multiline_unicode.conf")),
+    ?assertEqual(1, length(Conf)),
+    ?assertMatch([{["greeting"], "Hello 世界! 🌍\nПривіт, світ!\nПривет, мир!\nمرحبا بالعالم\nこんにちは世界"}], Conf).
+
 duplicates_test() ->
     Conf = file("test/multi1.conf"),
     ?assertEqual(2, length(Conf)),

test/multiline.conf

@@ -0,0 +1,5 @@
+a = '''
+    l0
+    l1
+    l2
+'''

test/multiline_embedded_delimiter.conf

@@ -0,0 +1,4 @@
+a = '''foo
+bar ''' baz
+bat
+'''

test/multiline_empty.conf

@@ -0,0 +1,2 @@
+a = '''
+'''

test/multiline_eof.conf

@@ -0,0 +1 @@
+a = '''value'''

test/multiline_mixed.conf

@@ -0,0 +1,9 @@
+a = '''
+    l0
+l1
+    l2
+'''
+
+regular = value
+
+another = simple

test/multiline_multiple.conf

@@ -0,0 +1,10 @@
+a = '''
+    l0
+    l1
+    l2
+'''
+
+b = '''l3
+l4
+l5
+'''