feat: simplify grammar, grab more metadata

synacktiv · May 29, 2024 · 63c3736 · 63c3736
1 parent d862269
commit 63c3736
Show file tree

Hide file tree

Showing 3 changed files with 38 additions and 12 deletions.
diff --git a/src/strace/parser/mod.rs b/src/strace/parser/mod.rs
@@ -402,7 +402,7 @@ mod tests {
                             Expression::Integer(IntegerExpression {
                                 value: IntegerExpressionValue::BinaryOr(vec![
                                     IntegerExpressionValue::NamedConst("S_IFDIR".to_owned()),
-                                    IntegerExpressionValue::NamedConst("0755".to_owned())
+                                    IntegerExpressionValue::Literal(0o755)
                                 ]),
                                 metadata: None,
                             }),

diff --git a/src/strace/parser/peg.pest b/src/strace/parser/peg.pest
@@ -8,7 +8,7 @@ pid = { ASCII_DIGIT+ }
 
 rel_ts = { ASCII_DIGIT+ ~ "." ~ ASCII_DIGIT+ }
 
-name = { (ASCII_ALPHA_LOWER | ASCII_DIGIT | "_")+ }
+name = { symbol_name }
 
 arguments = { "(" ~ (expression ~ (", " ~ expression)*)? ~ ")" }
 
@@ -18,31 +18,44 @@ ret_val = { int ~ (" " ~ ANY*)? }
 // Subrules
 
 expression = {
+	#macro = macro |
 	#int = int |
 	#struct = struct |
 	#buf = buffer |
-	#macro = macro |
 	#set = set |
 	#array = array
 }
 
+symbol_name = { (ASCII_ALPHA | ASCII_DIGIT | "_")+ }
+comment = { " /* " ~ (!" */" ~ ANY)+ ~ " */" }
+
 int = {
 	#or = or |
 	#lit = literal_int |
 	#named = named_constant
 }
-literal_int = { #hex = literal_int_hex | #dec = literal_int_dec }
+literal_int = {
+	#oct = literal_int_oct |
+	#hex = literal_int_hex |
+	#dec = literal_int_dec
+}
 or = { named_constant ~ ("|" ~ int)+ }
-named_constant = { (ASCII_ALPHA_UPPER | ASCII_DIGIT | "_")+ }
+named_constant = { symbol_name ~ int_metadata? }
 
+literal_int_oct = { "0" ~ ASCII_OCT_DIGIT+ }
 literal_int_hex = { "0x" ~ ASCII_HEX_DIGIT+ }
-literal_int_dec = { literal_int_dec_val ~ int_metadata? }
+literal_int_dec = {
+	literal_int_dec_val ~
+	(
+		#mdata = int_metadata |
+		#com = comment
+	)?
+}
 literal_int_dec_val = { "-"? ~ ASCII_DIGIT+ }
 int_metadata = { "<" ~ buffer_byte+ ~ ">" }
 
 struct = { "{" ~ (struct_member ~ (", " ~ struct_member)*)? ~ "}" }
 struct_member = { symbol_name ~ "=" ~ expression }
-symbol_name = { (ASCII_ALPHA | ASCII_DIGIT | "_")+ }
 
 buffer = {
 	"@"? ~

diff --git a/src/strace/parser/peg.rs b/src/strace/parser/peg.rs
@@ -115,6 +115,7 @@ impl From<Pair<'_, Rule>> for IntegerExpression {
             Some("lit") => {
                 let pair = pair_descend(pair, 1).unwrap();
                 let (val, metadata) = match pair.as_node_tag() {
+                    Some("oct") => (i128::from_str_radix(pair.as_str(), 8).unwrap(), None),
                     Some("hex") => (
                         pair.as_str()
                             .strip_prefix("0x")
@@ -125,7 +126,14 @@ impl From<Pair<'_, Rule>> for IntegerExpression {
                     Some("dec") => {
                         let mut children = pair.into_inner();
                         let val_pair = children.next().unwrap();
-                        let metadata_pair = children.next();
+                        let mut metadata_pair = children.next();
+                        // TODO use Option::take_if if it gets stable
+                        if metadata_pair
+                            .as_ref()
+                            .is_some_and(|p| p.as_node_tag() == Some("com"))
+                        {
+                            metadata_pair = None;
+                        }
                         (
                             val_pair.as_str().parse().unwrap(),
                             metadata_pair.map(|p| BufferExpression::from(p).value),
@@ -138,10 +146,15 @@ impl From<Pair<'_, Rule>> for IntegerExpression {
                     metadata,
                 }
             }
-            Some("named") => IntegerExpression {
-                value: IntegerExpressionValue::NamedConst(pair.as_str().to_owned()),
-                metadata: None,
-            },
+            Some("named") => {
+                let mut children = pair.into_inner();
+                let val_pair = children.next().unwrap();
+                let metadata_pair = children.next();
+                IntegerExpression {
+                    value: IntegerExpressionValue::NamedConst(val_pair.as_str().to_owned()),
+                    metadata: metadata_pair.map(|p| BufferExpression::from(p).value),
+                }
+            }
             Some("or") => {
                 let mut children = pair.into_inner();
                 let mut or_elems = Vec::with_capacity(children.len());