Require the message body to be indented

stasm · stasm · commit d694e1eef202 · 2017-02-24T16:29:37.000+01:00
Fix #12, #17, #18. With this change, the entire body of a message must be indented. This makes error recovery very easy: finding the next message definition is as simple as finding the next identifier with no indentation. It also opens up a number of opportunities: we can remove the `|` syntax for multiline blocks of text and allow line breaks inside of placeables safely. The change also allows the value to be defined on a new line, making the following examples equivalent: lipsum = Lorem ipsum dolor sit amet, consectetur adipiscing elit. Morbi pellentesque congue metus, non mattis sem faucibus sit amet. lipsum = Lorem ipsum dolor sit amet, consectetur adipiscing elit. Morbi pellentesque congue metus, non mattis sem faucibus sit amet. Lastly, quoted patterns are only available inside of placeables, cannot contain aother placeables and cannot be used directly as values. The exact semantics of \ escapes will be defined in #22.
diff --git a/guide/builtins.md b/guide/builtins.md
@@ -5,7 +5,7 @@ emails = You have { $unreadEmails } unread emails.
 emails2 = You have { NUMBER($unreadEmails) } unread emails.
 
 last-notice =
-    | Last checked: { DATETIME($lastChecked, day: "numeric", month: "long") }.
+    Last checked: { DATETIME($lastChecked, day: "numeric", month: "long") }.
 ```
 
 ```json
diff --git a/guide/functions.md b/guide/functions.md
@@ -126,16 +126,16 @@ the purpose of choosing the correct plural category:
 
 ```
 liked-count = { $num ->
-  [0]     No likes yet.
-  [one]   One person liked your message
- *[other] { $num } people liked your message
-}
+        [0]     No likes yet.
+        [one]   One person liked your message
+       *[other] { $num } people liked your message
+    }
 
 liked-count2 = { NUMBER($num) ->
-  [0]     No likes yet.
-  [one]   One person liked your message
- *[other] { $num } people liked your message
-}
+        [0]     No likes yet.
+        [one]   One person liked your message
+       *[other] { $num } people liked your message
+    }
 ```
 
 ### `DATETIME`  
diff --git a/guide/selectors.md b/guide/selectors.md
@@ -2,9 +2,9 @@
 
 ```
 emails = { $unreadEmails ->
-    [one] You have one unread email.
-   *[other] You have { $unreadEmails } unread emails.
-}
+        [one] You have one unread email.
+       *[other] You have { $unreadEmails } unread emails.
+    }
 ```
 
 ```json
@@ -34,7 +34,7 @@ unformatted number:
 
 ```
 your-score = { NUMBER($score, minimumFractionDigits: 1) ->
-    [0.0]   You scored zero points. What happened?
-   *[other] You scored { NUMBER($score, minimumFractionDigits: 1) } points.
-}
+        [0.0]   You scored zero points. What happened?
+       *[other] You scored { NUMBER($score, minimumFractionDigits: 1) } points.
+    }
 ```
diff --git a/guide/text.md b/guide/text.md
@@ -1,23 +1,16 @@
-# Working With Text: Multiline and Quotes
+# Working With Text
 
 ```
 about = About Our Software
 description =
-    | Loki is a simple micro-blogging
-    | app written entirely in <i>HTML5</i>.
-    | It uses FTL to implement localization.
-more-info = "  Read more about us! "
+    Loki is a simple micro-blogging
+    app written entirely in <i>HTML5</i>.
+    It uses FTL to implement localization.
 
 ```
 
-The value of an FTL message is usually a simple string.
+The value of an FTL message is usually a simple string.  It begins after the
+`=` and may continue over multiple lines as long as it is indented by at least
+one space.
 
-By default, a string begins after a `=` and ends with the end of line. You can
-also define easy-to-read, multiline strings with a pipe mark-up, as can be seen
-in the `description` message.
-
-FTL ignores leading whitespaces in front of the value allowing localizers to
-align their messages for readability. For multiline strings, whitespaces both
-before and after the pipe are ignored. In rare cases where leading whitespaces
-should be part of the value, FTL allows for special quote delimited strings as
-can be seen in the `more-info` message.
+Leading and trailing white-space is ignored.
diff --git a/guide/variants.md b/guide/variants.md
@@ -2,13 +2,13 @@
 
 ```
 brand-name = {
-   *[nominative] Aurora
-    [genitive] Aurore
-    [dative] Aurori
-    [accusative] Auroro
-    [locative] Aurori
-    [instrumental] Auroro
-}
+       *[nominative] Aurora
+        [genitive] Aurore
+        [dative] Aurori
+        [accusative] Auroro
+        [locative] Aurori
+        [instrumental] Auroro
+    }
 
 about = O { brand-name[locative] }
 ```
diff --git a/spec/CHANGELOG.md b/spec/CHANGELOG.md
@@ -2,6 +2,13 @@
 
 ## Unreleased
 
+  - Require the message body to be indented.
+
+    Quoted strings are now only valid in placeables and cannot contain other
+    placeables.
+
+    Remove `|` for multiline blocks.
+
   - (f176deb) Fix #30. Allow more characters in keys, and trim whitespace
     around them
 
diff --git a/spec/fluent.asdl b/spec/fluent.asdl
@@ -32,8 +32,7 @@ module Fluent
     pat = Pattern(expr* elements, bool quoted)
 
     -- Expressions
-    expr = Pattern(pat)
-         | String(string)
+    expr = String(string)
          | Number(number)
          | MessageReference(iden id)
          | ExternalArgument(iden id)
diff --git a/spec/fluent.ebnf b/spec/fluent.ebnf
@@ -1,41 +1,50 @@
-body                 ::= (entry NL)* entry? EOF
+body                 ::= (_* NL)* (entry NL)* entry? EOF
 entry                ::= comment
                        | section
                        | message
 
-comment              ::= ('//' [^#xA#xD]* NL)+
-section              ::= '[[' __ word (__ word)* __ ']]'
+comment              ::= ('//' (char - NL)* )+
+section              ::= '[[' _? word (_ word)* _? ']]'
 
 char                 ::= [https://www.w3.org/TR/REC-xml/#NT-Char]
-__                   ::= [#x20#x9]*   /* space, tab */
-NL                   ::= [#xA#xD]+    /* line feed, carriage return */
+
+/* special chars: { } [ \ ] */
+special              ::= [#x5b#x5c#x5d#x7b#x7d]
+
+/* line feed, carriage return; space, tab */
+line-break           ::= [#xA#xD]+
+inline-space         ::= [#x20#x9]+
+
+/* if you break lines you need to indent afterwards */
+break-indent         ::= (line-break* inline-space)+
+
+_                    ::= inline-space
+NL                   ::= line-break
+__                   ::= break-indent
 
 identifier           ::= [a-zA-Z_?-] [a-zA-Z0-9_?-]*
 external             ::= '$' identifier
-word                 ::= [^#x20#x9#xA#xD#x5b#x5c#x5d]+   /* exclude: white space, [, \, ] */
+word                 ::= (((char - line-break) - inline-space) - special)+
 builtin              ::= [A-Z_?-]+
 number               ::= [0-9]+ ('.' [0-9]+)?
 
 variant-key          ::= number | variant-symbol
-variant-symbol       ::= word (__ word)*
-variant              ::= '[' variant-key ']' __ pattern NL
-default-variant      ::= '*' variant
-variant-list         ::= NL (__ variant)* __ default-variant (__ variant)*
-
-attribute            ::= '.' identifier __ '=' __ pattern NL
-attribute-list       ::= NL (__ attribute)+
-
-message              ::= identifier __ '=' __ (pattern attribute-list? | attribute-list)
-pattern              ::= unquoted-pattern
-                       | quoted-pattern
-unquoted-pattern     ::= (unquoted-text | placeable | block-text)+
-quoted-pattern       ::= '"' (quoted-text | placeable)* '"'
-unquoted-text        ::= ([^{] | '\{')+
-quoted-text          ::= ([^{"] | '\{' | '\"')+
-block-text           ::= NL __ '|' unquoted-pattern
-
-placeable            ::= '{' __ (expression | select-expression | variant-list) __ '}'
-expression           ::= quoted-pattern
+variant-symbol       ::= word (_ word)*
+variant              ::= NL __ '[' _? variant-key _? ']' __ pattern
+default-variant      ::= NL __ '*[' _? variant-key _? ']' __ pattern
+variant-list         ::= variant* default-variant variant*
+
+attribute            ::= NL __ '.' identifier value
+attribute-list       ::= attribute+
+
+message              ::= identifier (value attribute-list? | attribute-list)
+value                ::= _? '=' __? pattern
+pattern              ::= (text | placeable)+
+text                 ::= ((char - line-break) - special | break-indent | '\' special)+
+quoted-text          ::= '"' (text | '\"')+ '"'
+
+placeable            ::= '{' __? (expression | select-expression | variant-list) __? '}'
+expression           ::= quoted-text
                        | number
                        | identifier
                        | external
@@ -44,10 +53,10 @@ expression           ::= quoted-pattern
                        | call-expression
                        | placeable
 
-select-expression    ::= expression __ ' ->' __ variant-list
+select-expression    ::= expression __ '->' __ variant-list
 attribute-expression ::= identifier '.' identifier
-variant-expression   ::= identifier '[' __ variant-key __ ']'
-call-expression      ::= builtin '(' __ (argument ( __ ',' __ argument)*)? __ ')'
+variant-expression   ::= identifier '[' _? variant-key _? ']'
+call-expression      ::= builtin '(' __? (argument ( __? ',' __? argument)*)? __? ')'
 argument             ::= expression
                        | named-argument
-named-argument       ::= identifier __ ':' __ ('"' quoted-text? '"' | number)
+named-argument       ::= identifier __? ':' __? (quoted-text | number)
