🖋️ Refactor common grammar to avoid unwanted greedy terminals

packages/parser/src/language/architecture/arch.langium

@@ -0,0 +1,2 @@
+terminal ARCH_ICON: /\([\w-:]+\)/;
+terminal ARCH_TITLE: /\[[\w ]+\]/;

packages/parser/src/language/architecture/architecture.langium

@@ -1,14 +1,15 @@
 grammar Architecture
 import "../common/common";
+import "arch";
 
 entry Architecture:
     NEWLINE*
     "architecture-beta"
     (
-    NEWLINE* TitleAndAccessibilities
+        NEWLINE
-    | NEWLINE* Statement*
+        | TitleAndAccessibilities
-    | NEWLINE*
+        | Statement
-    )
+    )*
 ;
 
 fragment Statement:
@@ -31,25 +32,21 @@ fragment Arrow:
 ;
 
 Group:
-    'group' id=ARCH_ID icon=ARCH_ICON? title=ARCH_TITLE? ('in' in=ARCH_ID)? EOL
+    'group' id=ID icon=ARCH_ICON? title=ARCH_TITLE? ('in' in=ID)? EOL
 ;
 
 Service:
-    'service' id=ARCH_ID (iconText=ARCH_TEXT_ICON | icon=ARCH_ICON)? title=ARCH_TITLE? ('in' in=ARCH_ID)? EOL
+    'service' id=ID (iconText=STRING | icon=ARCH_ICON)? title=ARCH_TITLE? ('in' in=ID)? EOL
 ;
 
 Junction:
-    'junction' id=ARCH_ID ('in' in=ARCH_ID)? EOL
+    'junction' id=ID ('in' in=ID)? EOL
 ;
 
 Edge:
-    lhsId=ARCH_ID lhsGroup?=ARROW_GROUP? Arrow rhsId=ARCH_ID rhsGroup?=ARROW_GROUP? EOL
+    lhsId=ID lhsGroup?=ARROW_GROUP? Arrow rhsId=ID rhsGroup?=ARROW_GROUP? EOL
 ;
 
 terminal ARROW_DIRECTION: 'L' | 'R' | 'T' | 'B';
-terminal ARCH_ID: /[\w]+/;
-terminal ARCH_TEXT_ICON: /\("[^"]+"\)/;
-terminal ARCH_ICON: /\([\w-:]+\)/;
-terminal ARCH_TITLE: /\[[\w ]+\]/;
 terminal ARROW_GROUP: /\{group\}/;
 terminal ARROW_INTO: /<|>/;

packages/parser/src/language/common/common.langium

@@ -1,22 +1,35 @@
-interface Common {
+// Base terminals and fragments for common language constructs
-  accDescr?: string;
+// Terminal Precedence: Lazy to Greedy
-  accTitle?: string;
+// When imported, the terminals are considered after the terminals in the importing grammar
-  title?: string;
+// Note: Hence, to add a terminal greedier than the common terminals, import the common grammar first
-}
-
-fragment TitleAndAccessibilities:
-  ((accDescr=ACC_DESCR | accTitle=ACC_TITLE | title=TITLE) EOL)+
-;
 
 fragment EOL returns string:
   NEWLINE+ | EOF
 ;
 
-terminal NEWLINE: /\r?\n/;
+fragment TitleAndAccessibilities:
+  ((accDescr=ACC_DESCR | accTitle=ACC_TITLE | title=TITLE) EOL)+
+;
+
+terminal BOOLEAN returns boolean: 'true' | 'false';
+
 terminal ACC_DESCR: /[\t ]*accDescr(?:[\t ]*:([^\n\r]*?(?=%%)|[^\n\r]*)|\s*{([^}]*)})/;
 terminal ACC_TITLE: /[\t ]*accTitle[\t ]*:(?:[^\n\r]*?(?=%%)|[^\n\r]*)/;
 terminal TITLE: /[\t ]*title(?:[\t ][^\n\r]*?(?=%%)|[\t ][^\n\r]*|)/;
 
+terminal FLOAT returns number: /[0-9]+\.[0-9]+(?!\.)/;
+terminal INT returns number: /0|[1-9][0-9]*(?!\.)/;
+terminal NUMBER returns number: FLOAT | INT;
+
+terminal STRING returns string: /"([^"\\]|\\.)*"|'([^'\\]|\\.)*'/;
+
+// Alphanumerics with underscores and dashes
+// Must start with an alphanumeric or an underscore
+// Cant end with a dash
+terminal ID returns string: /[\w]([-\w]*\w)?/;
+
+terminal NEWLINE: /\r?\n/;
+
 hidden terminal WHITESPACE: /[\t ]+/;
 hidden terminal YAML: /---[\t ]*\r?\n(?:[\S\s]*?\r?\n)?---(?:\r?\n|(?!\S))/;
 hidden terminal DIRECTIVE: /[\t ]*%%{[\S\s]*?}%%(?:\r?\n|(?!\S))/;

packages/parser/src/language/gitGraph/gitGraph.langium

@@ -1,39 +1,15 @@
 grammar GitGraph
-
+import "../common/common";
-interface Common {
+import "reference";
-  accDescr?: string;
-  accTitle?: string;
-  title?: string;
-}
-
-fragment TitleAndAccessibilities:
-  ((accDescr=ACC_DESCR | accTitle=ACC_TITLE | title=TITLE) EOL)+
-;
-
-fragment EOL returns string:
-  NEWLINE+ | EOF
-;
-
-terminal NEWLINE: /\r?\n/;
-terminal ACC_DESCR: /[\t ]*accDescr(?:[\t ]*:([^\n\r]*?(?=%%)|[^\n\r]*)|\s*{([^}]*)})/;
-terminal ACC_TITLE: /[\t ]*accTitle[\t ]*:(?:[^\n\r]*?(?=%%)|[^\n\r]*)/;
-terminal TITLE: /[\t ]*title(?:[\t ][^\n\r]*?(?=%%)|[\t ][^\n\r]*|)/;
-
-hidden terminal WHITESPACE: /[\t ]+/;
-hidden terminal YAML: /---[\t ]*\r?\n(?:[\S\s]*?\r?\n)?---(?:\r?\n|(?!\S))/;
-hidden terminal DIRECTIVE: /[\t ]*%%{[\S\s]*?}%%(?:\r?\n|(?!\S))/;
-hidden terminal SINGLE_LINE_COMMENT: /[\t ]*%%[^\n\r]*/;
 
 entry GitGraph:
   NEWLINE*
   ('gitGraph' | 'gitGraph' ':' | 'gitGraph:' | ('gitGraph' Direction ':'))
-  NEWLINE*
   (
-    NEWLINE*
+    NEWLINE
-    (TitleAndAccessibilities |
+    | TitleAndAccessibilities
-    statements+=Statement |
+    | statements+=Statement
-     NEWLINE)*
+  )*
-  )
 ;
 
 Statement
@@ -56,12 +32,12 @@ Commit:
         |'type:' type=('NORMAL' | 'REVERSE' | 'HIGHLIGHT')
     )* EOL;
 Branch:
-    'branch' name=(ID|STRING)
+    'branch' name=(REFERENCE|STRING)
     ('order:' order=INT)?
     EOL;
 
 Merge:
-    'merge' branch=(ID|STRING)
+    'merge' branch=(REFERENCE|STRING)
     (
         'id:' id=STRING
         |'tag:' tags+=STRING
@@ -69,7 +45,7 @@ Merge:
     )* EOL;
 
 Checkout:
-    ('checkout'|'switch') branch=(ID|STRING) EOL;
+    ('checkout'|'switch') branch=(REFERENCE|STRING) EOL;
 
 CherryPicking:
     'cherry-pick' 
@@ -78,10 +54,3 @@ CherryPicking:
         |'tag:' tags+=STRING
         |'parent:' parent=STRING
     )* EOL;
-
-
-
-terminal INT returns number: /[0-9]+(?=\s)/;
-terminal ID returns string: /\w([-\./\w]*[-\w])?/;
-terminal STRING: /"[^"]*"|'[^']*'/;
-

packages/parser/src/language/gitGraph/reference.langium

@@ -0,0 +1,4 @@
+// Alphanumerics with underscores, dashes, slashes, and dots
+// Must start with an alphanumeric or an underscore
+// Cant end with a dash, slash, or dot
+terminal REFERENCE returns string: /\w([-\./\w]*[-\w])?/;

packages/parser/src/language/index.ts

@@ -12,7 +12,6 @@ export {
   Commit,
   Merge,
   Statement,
-  isCommon,
   isInfo,
   isPacket,
   isPacketBlock,

packages/parser/src/language/packet/packet.langium

@@ -5,15 +5,12 @@ entry Packet:
   NEWLINE*
   "packet-beta"
   (
-    NEWLINE* TitleAndAccessibilities blocks+=PacketBlock*
+    TitleAndAccessibilities
-    | NEWLINE+ blocks+=PacketBlock+
+    | blocks+=PacketBlock
-    | NEWLINE*
+    | NEWLINE
-  )
+  )*
 ;
 
 PacketBlock:
   start=INT('-' end=INT)? ':' label=STRING EOL
-;
+;
-
-terminal INT returns number: /0|[1-9][0-9]*/;
-terminal STRING: /"[^"]*"|'[^']*'/;

packages/parser/src/language/pie/pie.langium

@@ -5,15 +5,12 @@ entry Pie:
   NEWLINE*
   "pie" showData?="showData"?
   (
-    NEWLINE* TitleAndAccessibilities sections+=PieSection*
+    TitleAndAccessibilities 
-    | NEWLINE+ sections+=PieSection+
+    | sections+=PieSection
-    | NEWLINE*
+    | NEWLINE
-  )
+  )*
 ;
 
 PieSection:
-  label=PIE_SECTION_LABEL ":" value=PIE_SECTION_VALUE EOL
+  label=STRING ":" value=NUMBER EOL
 ;
-
-terminal PIE_SECTION_LABEL: /"[^"]+"/;
-terminal PIE_SECTION_VALUE returns number: /(0|[1-9][0-9]*)(\.[0-9]+)?/;

packages/parser/src/language/radar/radar.langium

@@ -1,31 +1,5 @@
 grammar Radar
-// import "../common/common";
+import "../common/common";
-// Note: The import statement breaks TitleAndAccessibilities probably because of terminal order definition
-// TODO: May need to change the common.langium to fix this
-
-interface Common {
-  accDescr?: string;
-  accTitle?: string;
-  title?: string;
-}
-
-fragment TitleAndAccessibilities:
-  ((accDescr=ACC_DESCR | accTitle=ACC_TITLE | title=TITLE) EOL)+
-;
-
-fragment EOL returns string:
-  NEWLINE+ | EOF
-;
-
-terminal NEWLINE: /\r?\n/;
-terminal ACC_DESCR: /[\t ]*accDescr(?:[\t ]*:([^\n\r]*?(?=%%)|[^\n\r]*)|\s*{([^}]*)})/;
-terminal ACC_TITLE: /[\t ]*accTitle[\t ]*:(?:[^\n\r]*?(?=%%)|[^\n\r]*)/;
-terminal TITLE: /[\t ]*title(?:[\t ][^\n\r]*?(?=%%)|[\t ][^\n\r]*|)/;
-
-hidden terminal WHITESPACE: /[\t ]+/;
-hidden terminal YAML: /---[\t ]*\r?\n(?:[\S\s]*?\r?\n)?---(?:\r?\n|(?!\S))/;
-hidden terminal DIRECTIVE: /[\t ]*%%{[\S\s]*?}%%(?:\r?\n|(?!\S))/;
-hidden terminal SINGLE_LINE_COMMENT: /[\t ]*%%[^\n\r]*/;
 
 entry Radar:
     NEWLINE*
@@ -76,14 +50,6 @@ Option:
         | name='min' value=NUMBER
         | name='graticule' value=GRATICULE
     )
-;
+;   
 
-
+terminal GRATICULE returns string: 'circle' | 'polygon';
-terminal NUMBER returns number: /(0|[1-9][0-9]*)(\.[0-9]+)?/;
-
-terminal BOOLEAN returns boolean: 'true' | 'false';    
-
-terminal GRATICULE returns string: 'circle' | 'polygon';
-
-terminal ID returns string: /[a-zA-Z_][a-zA-Z0-9\-_]*/; 
-terminal STRING: /"[^"]*"|'[^']*'/;