[grammar refactor] Consolidate eggex

oils-for-unix · Dec 12, 2023 · a82f94a · a82f94a
1 parent 6b6319e
commit a82f94a
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 16 deletions.
diff --git a/test/ysh-parse-errors.sh b/test/ysh-parse-errors.sh
@@ -1278,7 +1278,7 @@ json write (x) {
   _ysh-parse-error 'f(x)'  # test short name
 }
 
-test-eggex() {
+test-eggex-capture() {
   _should-parse '= / d+ /'
   #_should-parse '= / <d+ : date> /'
   _should-parse '= / < capture d+ as date > /'
@@ -1289,6 +1289,19 @@ test-eggex() {
   _parse-error 'var as = 42'
 }
 
+
+test-eggex-flags() {
+  _should-parse '= / d+ ; ignorecase /'
+
+  # ERE means is the default; it's POSIX ERE
+  # Other option is PCRE
+  _should-parse '= / d+ ; ignorecase ; ERE /'
+  _should-parse '= / d+ ; ; ERE /'
+
+  # typo should be parse error
+  _parse-error '= / d+ ; ignorecas /'
+}
+
 #
 # Entry Points
 #

diff --git a/ysh/expr_to_ast.py b/ysh/expr_to_ast.py
@@ -384,13 +384,6 @@ def _Atom(self, parent):
                 i += 1
             return self._Dict(parent, parent.GetChild(i))
 
-        if id_ == Id.Arith_Slash:
-            r = self._Regex(parent.GetChild(1))
-            flags = []  # type: List[Token]
-            # TODO: Parse translation preference.
-            trans_pref = None  # type: Token
-            return Eggex(parent.GetChild(0).tok, r, flags, trans_pref)
-
         if id_ == Id.Arith_Amp:
             n = parent.NumChildren()
             if n >= 3:
@@ -673,6 +666,9 @@ def Expr(self, pnode):
                 inner = self.Expr(pnode.GetChild(1))
                 return expr.Literal(inner)
 
+            elif typ == grammar_nt.eggex:
+                return self._Eggex(pnode)
+
             elif typ == grammar_nt.ysh_expr_sub:
                 return self.Expr(pnode.GetChild(0))
 
@@ -824,6 +820,12 @@ def MakeMutation(self, p_node):
         rhs = self.Expr(p_node.GetChild(2))
         return command.Mutation(None, lhs_list, op_tok, rhs)
 
+    def _Eggex(self, p_node):
+        # type: (PNode) -> Eggex
+        left = p_node.GetChild(0).tok
+        regex = self._Regex(p_node.GetChild(1))
+        return Eggex(left, regex, [], None)
+
     def YshCasePattern(self, pnode):
         # type: (PNode) -> pat_t
         assert pnode.typ == grammar_nt.ysh_case_pat, pnode
@@ -846,11 +848,8 @@ def YshCasePattern(self, pnode):
                         exprs.append(expr)
                 return pat.YshExprs(exprs)
 
-        elif typ == grammar_nt.pat_eggex:
-            # pat_eggex
-            left = pattern.GetChild(0).tok
-            re = self._Regex(pattern.GetChild(1))
-            return Eggex(left, re, [], None)
+        elif typ == grammar_nt.eggex:
+            return self._Eggex(pattern)
 
         raise NotImplementedError()
 

diff --git a/ysh/grammar.pgen2 b/ysh/grammar.pgen2
@@ -84,7 +84,6 @@ atom: (
   | '[' [testlist_comp] ']' 
     # Note: newlines are significant inside {}, unlike inside () and []
   | '{' [Op_Newline] [dict] '}'
-  | '/' regex [re_flags] '/'
   | '&' Expr_Name place_trailer*
 
     # NOTE: These atoms are are allowed in typed array literals
@@ -105,6 +104,7 @@ atom: (
   | dq_string | sq_string
     # Expr_Symbol could be %mykey
 
+  | eggex
   | literal_expr
 
   # $foo is disallowed, but $? is allowed.  Should be "$foo" to indicate a
@@ -472,14 +472,15 @@ regex: [re_alt] (('|'|'or') re_alt)*
 #                  ^-- End of pattern/beginning of case arm body
 # }
 
+eggex: '/' regex [re_flags] '/'
+
 ysh_case_pat: (
     '(' (pat_else | pat_exprs)
-  | pat_eggex
+  | eggex
 ) [Op_Newline] '{'
 
 pat_else: 'else' ')'
 pat_exprs: expr ')' [Op_Newline] ('|' [Op_Newline] '(' expr ')' [Op_Newline])*
-pat_eggex: '/' regex [re_flags] '/'
 
 # e.g. /digit+ ; multiline !ignorecase/
 #