Starting to lex array assignment like 'words[0]=1'.

Also moved a few tokens out of _UNQUOTED and into OUTER.

- cmd_parse.py: Clean up logic for looking ahead to a function def
- Audit of git-completion.bash language usage

core/id_kind.py

@@ -148,7 +148,8 @@ def AddKinds(spec):
   spec.AddKind('WS', ['Space'])
 
   spec.AddKind('Lit', [
-      'Chars', 'VarLike', 'Other', 'EscapedChar',
+      'Chars', 'VarLike', 'ArrayLhsOpen', 'ArrayLhsClose',
+      'Other', 'EscapedChar',
       # Either brace expansion or keyword for { and }
       'LBrace', 'RBrace', 'Comma',
       'DRightBracket',     # the ]] that matches [[, NOT a keyword

osh/cmd_parse.py

@@ -1333,11 +1333,13 @@ def ParseCommand(self, cur_aliases=None):
     self._Peek()
 
     if self.c_id in NOT_FIRST_WORDS:
-      p_die('Unexpected word when parsing cmomand', word=self.cur_word)
+      p_die('Unexpected word when parsing command', word=self.cur_word)
 
     if self.c_id == Id.KW_Function:
       return self.ParseKshFunctionDef()
 
+    # NOTE: We should have another Kind for "initial keywords".  And then
+    # NOT_FIRST_WORDS are "secondary keywords".
     if self.c_id in (
         Id.KW_DLeftBracket, Id.Op_DLeftParen, Id.Op_LParen, Id.Lit_LBrace,
         Id.KW_For, Id.KW_While, Id.KW_Until, Id.KW_If, Id.KW_Case, Id.KW_Time):
@@ -1359,14 +1361,13 @@ def ParseCommand(self, cur_aliases=None):
       return self.ParseSimpleCommand(cur_aliases)
 
     if self.c_kind == Kind.Word:
-      if self.w_parser.LookAhead() == Id.Op_LParen:  # (
-        kov = word.LooksLikeAssignment(self.cur_word)
-        if kov:
-          return self.ParseSimpleCommand(cur_aliases)  # f=(a b c)  # array
-        else:
+      if (self.w_parser.LookAhead() == Id.Op_LParen and
+          not word.LooksLikeAssignment(self.cur_word)):
           return self.ParseFunctionDef()  # f() { echo; }  # function
-
-      return self.ParseSimpleCommand(cur_aliases)  # echo foo
+      # echo foo
+      # f=(a b c)  # array
+      # array[1+2]+=1
+      return self.ParseSimpleCommand(cur_aliases)
 
     if self.c_kind == Kind.Eof:
       p_die("Unexpected EOF while parsing command", word=self.cur_word)

osh/lex.py

@@ -190,9 +190,6 @@
   # NOTE: We could add anything 128 and above to this character class?  So
   # utf-8 characters don't get split?
   R(r'[a-zA-Z0-9_/.-]+', Id.Lit_Chars),
-  # e.g. beginning of NAME=val, which will always be longer than the above
-  # Id.Lit_Chars.
-  R(r'[a-zA-Z_][a-zA-Z0-9_]*\+?=', Id.Lit_VarLike),
 
   # For tilde expansion. The list of chars is Lit_Chars, but WITHOUT the /.  We
   # want the next token after the tilde TildeLike token start with a /.
@@ -304,7 +301,15 @@ def IsKeyword(name):
 # Keywords have to be checked before _UNQUOTED so we get <KW_If "if"> instead
 # of <Lit_Chars "if">.
 LEXER_DEF[lex_mode_e.OUTER] = [
-  C('((', Id.Op_DLeftParen),  # not allowed within [[
+  # These four are not allowed within [[, so they are in OUTER but not
+  # _UNQUOTED.
+
+  # e.g. beginning of NAME=val, which will always be longer than the above
+  # Id.Lit_Chars.
+  R(r'[a-zA-Z_][a-zA-Z0-9_]*\+?=', Id.Lit_VarLike),
+  R(r'[a-zA-Z_][a-zA-Z0-9_]*\[', Id.Lit_ArrayLhsOpen),
+  R(r'\]\+?=', Id.Lit_ArrayLhsClose),
+  C('((', Id.Op_DLeftParen),
 ] + _KEYWORDS + _MORE_KEYWORDS + _UNQUOTED + _EXTGLOB_BEGIN
 
 # DBRACKET: can be like OUTER, except:

scripts/complete.sh

@@ -8,7 +8,23 @@ set -o pipefail
 set -o errexit
 
 git-audit() {
-  grep -E -w 'complete|compgen|compopt' testdata/completion/git-completion.bash
+  local file=testdata/completion/git-completion.bash
+
+  echo
+  echo --
+  # Search for completion builtin usage
+  grep -E -w --color 'complete|compgen|compopt' $file
+
+  echo
+  echo --
+  # Search for special complation var usage
+  grep -E --color 'COMP_[A-Z]+' $file
+
+  echo
+  echo --
+  # Search for array usage
+  grep -E --color ']=' $file
+  grep -E --color ']+=' $file
 }
 
 "$@"