feat: first chunk of lexer/parser translation.

2025-02-16 20:49:20 +01:00
parent d386c50499
commit 4f8e83da9d
15 changed files with 2593 additions and 49 deletions
--- a/InkBlot/InkBlotAntlrGrammar.g4
+++ b/InkBlot/InkBlotAntlrGrammar.g4
@@ -2,6 +2,138 @@ grammar InkBlotAntlrGrammar;

 import InkBlotAntlrLexer;

-story: contentText+ ;
+/*
+ * STORY
+ */

-contentText: CONTENT_TEXT_NO_ESCAPE_SIMPLE ;
+story: topLevelStatements ;
+
+/*
+ * STATEMENTS FOR THE VARIOUS LEVELS
+ */
+
+topLevelStatements:
+    contentText+
+    ;
+
+ knotLevelStatements:
+    contentText+
+    ;
+
+ stitchLevelStatements:
+    contentText+
+    ;
+
+ innerBlockLevelStatements:
+    contentText+
+    ;
+
+/*
+ * STATEMENTS
+ */
+
+contentText: CONTENT_TEXT_NO_ESCAPE_SIMPLE ;
+
+mixedTextAndLogic:
+    // TODO: ~ is not allowed as first symbol of this (see InkParser_Content => MixedTextAndLog), let's implement it in the C# side?
+    // like in innerInlineSequenceObjects, a bit of a chaos to avoid having completely empty entries
+    contentText? (inlineLogicOrGlueOrTagStart? contentText | inlineLogicOrGlueOrTagStart contentText?)+
+    // TODO: this is valid only when not parsing a choice, see above for the code where this logic is implemented
+    multiDivert
+    // TODO: management of tag ftw O_O
+    ;
+
+inlineLogicOrGlueOrTagStart:
+    inlineLogic
+    // TODO: glue, tag start
+    ;
+
+inlineLogic:
+    INLINE_LOGIC_START
+    WS?
+    innerLogic
+    INLINE_LOGIC_END
+    ;
+    
+innerLogic:
+    WS?
+    sequenceTypeAnnotation innerSequenceObjects
+    // TODO: the rest of InkParser_Logic => InnerLogic
+    ;
+
+multiDivert:
+    WS?
+    (
+        THREAD_ARROW divertIdentifierWithArguments
+        // here be dragons: trying to express the various "Possible patterns" of InkParser_Divert => MultiDivert
+        | (DIVERT_ARROW divertIdentifierWithArguments)+ TUNNEL_ARROW divertIdentifierWithArguments
+        | (DIVERT_ARROW divertIdentifierWithArguments)+ TUNNEL_ARROW
+        | (DIVERT_ARROW divertIdentifierWithArguments)+ DIVERT_ARROW
+        | TUNNEL_ARROW
+        | DIVERT_ARROW // TODO: this is only valid in default choices ( https://github.com/inkle/ink/blob/master/Documentation/WritingWithInk.md#fallback-choices )
+    ) 
+    ;
+
+divertIdentifierWithArguments:
+    WS? identifier WS? ('.' WS? identifier WS? )*
+    WS?
+    (
+    '('
+    expression (',' expression)*
+    ')'
+    )?
+    WS?
+    ;
+
+identifier:
+    // TODO: reject numbers-only identifier - see InkParser_Logic => Identifier
+    IDENTIFIER
+    ;
+
+expression:
+    // TODO: temporary stuff here
+    IDENTIFIER
+    ;
+
+// all possible symbols or word(s) for sequencing
+sequenceTypeAnnotation:
+    SEQUENCE_TYPE_SYMBOL_ANNOTATION
+    | ONCE
+    | CYCLE
+    | SHUFFLE
+    | STOPPING
+    | SHUFFLE_ONCE
+    | SHUFFLE_STOPPING
+    ;
+
+/* a list of sequence objects, either compressed in a single line (e.g.: {a|b|c}) or expanded in multiple lines (e.g.:
+{\n- a\n- b\n- c}
+*/
+innerSequenceObjects:
+    NL innerMultilineSequenceObjects
+    | innerInlineSequenceObjects
+    ;
+
+innerMultilineSequenceObjects:
+    singleMultilineSequenceElement+
+    ;
+
+singleMultilineSequenceElement:
+    WS?
+    /* TODO: how to express this? and why is it here? InkParser_Sequences => SingleMultilineSequenceElement
+            if (ParseString ("->") != null)
+                return null;
+    */
+    '-'
+    WS?
+    (
+        innerBlockLevelStatements
+        | MULTILINE_WS
+    )
+    ;
+
+innerInlineSequenceObjects:
+    // it's a bit chaotic, in order to allow for empty mixedTextAndLogic, but always require at least one entry
+    (mixedTextAndLogic ('|' mixedTextAndLogic?)*)
+    | ('|' mixedTextAndLogic?)+
+    ;