[j8] json read now uses the j8 parser, not yajl

TODO:

- make it translate it to C++.
- get rid of remaining yajl usage, for compexport

builtin/json_ysh.py

@@ -16,9 +16,7 @@
 from frontend import typed_args
 from mycpp import mylib
 from mycpp.mylib import log
-from ysh import cpython
 
-import yajl
 import posix_ as posix
 
 from typing import TYPE_CHECKING
@@ -42,9 +40,11 @@ def __init__(self, mem, errfmt, is_j8):
         self.mem = mem
         self.errfmt = errfmt
 
-        self.stdout_ = mylib.Stdout()
         self.is_j8 = is_j8
+        self.name = 'j8' if is_j8 else 'json'  # for error messages
+
         self.j8print = j8.Printer()
+        self.stdout_ = mylib.Stdout()
 
     def Run(self, cmd_val):
         # type: (cmd_value.Argv) -> int
@@ -124,28 +124,20 @@ def Run(self, cmd_val):
                                          posix.strerror(e.err_num))
                 return 1
 
-            if self.is_j8:
-            #if True:
-                if mylib.PYTHON:
-                    p = j8.Parser(contents)
+            if mylib.PYTHON:
+                p = j8.Parser(contents)
+                try:
                     if self.is_j8:
                         val = p.ParseJ8()
                     else:
                         val = p.ParseJson()
-                    self.mem.SetPlace(place, val, blame_loc)
+                except error.Decode as err:
+                    # TODO: Need to show position info
+                    self.errfmt.Print_('%s read: %s' % (self.name, err),
+                                       blame_loc=action_loc)
+                    return 1
 
-            else:
-                if mylib.PYTHON:
-                    try:
-                        obj = yajl.loads(contents)
-                    except ValueError as e:
-                        self.errfmt.Print_('json read: %s' % e,
-                                           blame_loc=action_loc)
-                        return 1
-
-                    # TODO: use token directly
-                    val = cpython._PyObjToValue(obj)
-                    self.mem.SetPlace(place, val, blame_loc)
+                self.mem.SetPlace(place, val, blame_loc)
 
         else:
             raise error.Usage(_JSON_ACTION_ERROR, action_loc)

core/error.py

@@ -213,7 +213,12 @@ class Decode(Exception):
       - u"" doesn't accept \\yff
       - "" doesn't accept \\yff or \\u{123456}
     """
-    pass
+    def __init__(self, msg, s, start_pos, end_pos):
+        # type: (str, str, int, int) -> None
+        self.msg = msg
+        self.s = s  # string being decoded
+        self.start_pos = start_pos
+        self.end_pos = end_pos
 
 
 class Encode(Exception):

data_lang/j8.py

@@ -51,6 +51,7 @@
 from _devbuild.gen.value_asdl import (value, value_e, value_t)
 
 from asdl import format as fmt
+from core import error
 from core import vm
 from data_lang import pyj8
 from data_lang import qsn
@@ -421,16 +422,20 @@ def _Eat(self, tok_id):
             if self.tok_id != tok_id:
                 #log('position %r %d-%d %r', self.s, self.start_pos,
                 #    self.end_pos, self.s[self.start_pos:self.end_pos])
-                raise AssertionError("Expected %s, got %s" %
-                                     (Id_str(tok_id), Id_str(self.tok_id)))
+                raise self._Error("Expected %s, got %s" % (Id_str(tok_id),
+                                                           Id_str(self.tok_id)))
             self._Next()
 
+        def _Error(self, msg):
+            # type: (str) -> error.Decode
+            return error.Decode(msg, self.s, self.start_pos, self.end_pos)
+
         def _ParsePair(self):
             # type: () -> Tuple[str, value_t]
-            if self.tok_id != Id.J8_AnyString:
-                raise AssertionError(Id_str(self.tok_id))
-            k = self.decoded
-            self._Next()
+
+            k = self.decoded  # Save the potential string value
+            self._Eat(Id.J8_AnyString)  # Check that it's a string
+            assert k is not None
 
             self._Eat(Id.J8_Colon)
 
@@ -522,7 +527,11 @@ def _ParseValue(self):
                 self._Next()
                 return str_val
 
+            elif self.tok_id == Id.Eol_Tok:
+                raise self._Error('Unexpected EOF while parsing JSON')
+
             else:
+                # This should never happen
                 part = self.s[self.start_pos:self.end_pos]
                 raise AssertionError('Unexpected token %s %r' %
                                      (Id_str(self.tok_id), part))

data_lang/json-errors.sh

@@ -33,6 +33,8 @@ test-json() {
   #echo OSH=$OSH
   #set +o errexit
 
+  _error-case-X 1 'echo "" | json read'
+
   _error-case-X 1 'echo { | json read'
 
   _error-case-X 1 'echo { | j8 read'