Merge pull request #489 from aptos-labs/feat/support_partials

Move-on-Aptos: Support More Declarations and Partial Signatures

lang/semgrep-grammars/src/semgrep-move-on-aptos/grammar.js

@@ -13,7 +13,14 @@ module.exports = grammar(base_grammar, {
   name: 'move_on_aptos',
 
   conflicts: ($, previous) => previous.concat([
-    [$.typed_metavariable, $.name_access_chain]
+    [$.typed_metavariable, $.name_access_chain],
+    [$.term, $.declaration],
+    [$._module_path, $.spec_block_target]
+  ]),
+
+  precedences: ($, previous) => previous.concat([
+    [$._sequence_item, $.declaration],
+    [$._script_use_decl, $._script_constant_decl, $._script_func_decl, $._script_spec_block],
   ]),
 
   /*
@@ -24,29 +31,40 @@ module.exports = grammar(base_grammar, {
     // Semgrep components, source: semgrep-rust
     ellipsis: $ => '...',
     deep_ellipsis: $ => seq('<...', $._expr, '...>'),
-    
+
     // Typed metavariable (an expression, not a parameter)
     // This is grammatically indistinguishable from `$.type_hint_expr: $ => seq('(', $._expr, ':', $.type, ')')`.
     // This will be handled by the semgrep converter by checking the metavariable name (`$`).
     typed_metavariable: $ => seq('(', $.identifier, ':', $.type, ')'),
 
     // Alternate "entry point". Allows parsing a standalone expression.
-    semgrep_expression: $ => seq('__SEMGREP_EXPRESSION', choice(
+    semgrep_expression: $ => choice(
       $._expr,
       $.let_expr,
-    )),
+    ),
 
     // Alternate "entry point". Allows parsing a standalone list of sequence items (statements).
-    semgrep_statement: $ => seq('__SEMGREP_STATEMENT', repeat1(choice(
+    semgrep_statement: $ => repeat1(choice(
       $._sequence_item,
-      $.constant_decl,
-    ))),
+      $.declaration,
+    )),
+
+    // Alternate "entry point". Allows parsing partial declarations (signatures).
+    semgrep_partial: $ => seq(
+      optional($.attributes),
+      repeat($.module_member_modifier),
+      choice(
+        $._function_signature,
+        $._struct_signature,
+      )
+    ),
 
     // Extend the source_file rule to allow semgrep constructs
     source_file: ($, previous) => choice(
       previous,
       $.semgrep_expression,
       $.semgrep_statement,
+      $.semgrep_partial,
     ),
 
     // Module declaration
@@ -55,6 +73,19 @@ module.exports = grammar(base_grammar, {
       $.ellipsis,
     ),
 
+    // Script members
+    // We cannot mimic the `declaration` trick here, as the script members are strictly ordered.
+    _script_use_decl: ($, previous) => choice(previous, $.ellipsis),
+    _script_constant_decl: ($, previous) => choice(previous, $.ellipsis),
+    _script_func_decl: ($, previous) => choice(previous, $.ellipsis),
+    _script_spec_block: ($, previous) => choice(previous, $.ellipsis),
+
+    // Address block members
+    _address_member: ($, previous) => choice(
+      previous,
+      $.ellipsis,
+    ),
+
     // Spec block members
     _spec_block_member: ($, previous) => choice(
       previous,
@@ -95,19 +126,32 @@ module.exports = grammar(base_grammar, {
       $.ellipsis,
     ),
 
+    // attribute value
+    // (e.g. `#[attr(key = ...)]`)
+    _attribute_val: ($, previous) => choice(
+      previous,
+      $.ellipsis,
+    ),
+
     // use member
     // (e.g. `use module_ident::...;`, `use module_ident::{..., item_ident}`)
     _use_member: ($, previous) => choice(
       previous,
       $.ellipsis,
     ),
 
+    // term
+    term: ($, previous) => choice(
+      ...previous.members,
+      $.ellipsis,
+      $.deep_ellipsis,
+    ),
+
     // expression 
     _expr: ($, previous) => choice(
       ...previous.members,
       $.ellipsis,
       $.deep_ellipsis,
-      $.field_access_ellipsis_expr,
     ),
 
     // unary expression
@@ -119,6 +163,11 @@ module.exports = grammar(base_grammar, {
       $.typed_metavariable,
     ),
 
+    _dot_or_index_chain: ($, previous) => choice(
+      ...previous.members,
+      $.field_access_ellipsis_expr,
+    ),
+
     // function parameter
     // (e.g. `call( ..., arg, ...)`)
     parameter: ($, previous) => choice(

lang/semgrep-grammars/src/semgrep-move-on-aptos/test/corpus/partials_and_decls.txt

@@ -0,0 +1,151 @@
+=======================
+Function in Statement
+=======================
+
+fun $TEST (...) : ... { ... }
+
+---
+
+(source_file
+  (semgrep_statement
+    (declaration
+      (function_decl
+        (identifier)
+        (parameters
+          (parameter
+            (ellipsis)))
+        (type
+          (ellipsis))
+        (block
+          (ellipsis))))))
+
+=======================
+Declarations
+=======================
+
+use 0xabcd::ff;
+struct $STRUCT has ... { ... }
+const $CONST : ... = $FOO(...);
+...
+fun hello (...) {
+  let XXX = $STRUCT { ... };
+  ...
+}
+
+---
+
+(source_file
+  (semgrep_statement
+    (declaration
+      (use_decl
+        (module_ident
+          (numerical_addr
+            (number))
+          (identifier))))
+    (declaration
+      (struct_decl
+        (identifier)
+        (abilities
+          (ability
+            (ellipsis)))
+        (body
+          (field_annot
+            (ellipsis)))))
+      (declaration
+        (constant_decl
+          (identifier)
+          (type
+            (ellipsis))
+          (call_expr
+            (name_access_chain
+              (identifier))
+            (call_args
+              (ellipsis)))))
+    (ellipsis)
+    (declaration
+      (function_decl
+        (identifier)
+        (parameters
+          (parameter
+            (ellipsis)))
+        (block
+          (let_expr
+            (bind_list
+              (var_name
+                (identifier)))
+            (pack_expr
+              (name_access_chain
+                (identifier))
+              (expr_field
+                (ellipsis))))
+          (ellipsis))))))
+
+=======================
+Function Signature
+=======================
+
+#[test]
+public(friend) fun $FUNC (...)
+
+---
+
+(source_file
+  (semgrep_partial
+    (attributes
+      (attribute
+        (identifier)))
+    (module_member_modifier
+      (visibility))
+    (identifier)
+    (parameters
+      (parameter
+        (ellipsis)))))
+
+=======================
+Struct Signature
+=======================
+
+public native struct $STRUCT has ...
+
+---
+
+(source_file
+  (semgrep_partial
+    (module_member_modifier
+      (visibility))
+    (module_member_modifier)
+    (identifier)
+    (abilities
+      (ability
+        (ellipsis)))))
+
+=======================
+Function Attributes
+=======================
+
+#[attr(key = ...), attr2 = ..., attr3(...), ...]
+fun $FUN (...)
+
+---
+
+(source_file
+  (semgrep_partial
+    (attributes
+      (attribute
+        (identifier)
+        (attribute
+          (identifier)
+          (ellipsis)))
+      (attribute
+        (identifier)
+        (ellipsis))
+      (attribute
+        (identifier)
+        (attribute
+          (ellipsis)))
+      (attribute
+        (ellipsis)))
+    (identifier)
+    (parameters
+      (parameter
+        (ellipsis)))))