feat: support operator: field-extractor

Author: sean2077

.gitignore

@@ -127,4 +127,6 @@ dmypy.json
 
 # Pyre type checker
 .pyre/
-.vscode
+.vscode
+
+tmp.py

README.md

@@ -5,8 +5,8 @@
   - [Features](#features)
   - [Examples](#examples)
     - [Filter](#filter)
-    - [Sort](#sort)
-    - [Output mode: FIELD (In progress)](#output-mode-field-in-progress)
+    - [Sorter](#sorter)
+    - [Field-Extractor](#field-extractor)
 
 <!-- END doctoc generated TOC please keep comment here to allow auto update -->
 
@@ -17,12 +17,12 @@ A more powerful JSONPath implementations in modern python.
 ## Features
 
 - [x] Light. (No need to install third-party dependencies.)
-- [x] Powerful filtering function, including multi-selection, inverse-selection filtering.
-- [x] Powerful sorting function, including sorting by multiple fields, ascending and descending order.
-- [x] Support output mode: VALUE
-- [ ] Support output mode: PATH
-- [ ] Support output mode: FIELD
-- [ ] Support parent operator
+- [x] Support basic semantics of JSONPath.
+- [x] Support output modes: VALUE, PATH.
+- [x] Support filter operator, including multi-selection, inverse-selection filtering.
+- [x] Support sorter operator, including sorting by multiple fields, ascending and descending order.
+- [ ] Support parent operator.
+- [ ] Support user-defined function.
 
 ## Examples
 
@@ -122,33 +122,29 @@ data = {
 ]
 ```
 
-### Field-Extractor (In progress)
+### Field-Extractor
 
 ```python
->>> JSONPath("$.store.book[*][title,price]",result_type="FIELD").parse(data)
+>>> JSONPath("$.store.book[*](title,price)",result_type="FIELD").parse(data)
 ```
 
 ```json
-{
-  "store": {
-    "book": [
-      {
-        "title": "Sayings of the Century",
-        "price": 8.95
-      },
-      {
-        "title": "Sword of Honour",
-        "price": 12.99
-      },
-      {
-        "title": "Moby Dick",
-        "price": 8.99
-      },
-      {
-        "title": "The Lord of the Rings",
-        "price": 22.99
-      }
-    ]
+[
+  {
+    "title": "Sayings of the Century",
+    "price": 8.95
+  },
+  {
+    "title": "Sword of Honour",
+    "price": 12.99
+  },
+  {
+    "title": "Moby Dick",
+    "price": 8.99
+  },
+  {
+    "title": "The Lord of the Rings",
+    "price": 22.99
   }
-}
+]
 ```

jsonpath/__init__.py

@@ -1,17 +1,17 @@
 """
-Author       : zhangxianbing1
+Author       : zhangxianbing
 Date         : 2020-12-27 09:22:14
-LastEditors  : zhangxianbing1
-LastEditTime : 2021-01-04 14:34:35
+LastEditors  : zhangxianbing
+LastEditTime : 2021-01-04 15:25:08
 Description  : JSONPath
 """
 __version__ = "0.0.3"
 __author__ = "zhangxianbing"
 
 import json
+import logging
 import os
 import re
-import logging
 from collections import defaultdict
 from typing import Union
 
@@ -46,7 +46,6 @@ class ExprSyntaxError(Exception):
 class JSONPath:
     RESULT_TYPE = {
         "VALUE": "A list of specific values.",
-        "FIELD": "A dict with specific fields.",
         "PATH": "All path of specific values.",
     }
 
@@ -70,7 +69,7 @@ class JSONPath:
     steps: list
     lpath: int
     subx = defaultdict(list)
-    result: Union[list, dict]
+    result: list
     result_type: str
 
     def __init__(self, expr: str):
@@ -82,16 +81,14 @@ def __init__(self, expr: str):
     def parse(self, obj, result_type="VALUE"):
         if not isinstance(obj, (list, dict)):
             raise TypeError("obj must be a list or a dict.")
+
         if result_type not in JSONPath.RESULT_TYPE:
             raise ValueError(
                 f"result_type must be one of {tuple(JSONPath.RESULT_TYPE.keys())}"
             )
         self.result_type = result_type
-        if self.result_type == "FIELD":
-            self.result = {}
-        else:
-            self.result = []
 
+        self.result = []
         self._trace(obj, 0, "$")
 
         return self.result
@@ -188,8 +185,6 @@ def _trace(self, obj, i: int, path):
                 self.result.append(obj)
             elif self.result_type == "PATH":
                 self.result.append(path)
-            elif self.result_type == "FIELD":
-                pass
             LOG.debug(f"path: {path} | value: {obj}")
             return
 
@@ -240,7 +235,7 @@ def _trace(self, obj, i: int, path):
             self._traverse(self._filter, obj, i + 1, path, step)
             return
 
-        # sort
+        # sorter
         if step.startswith("/(") and step.endswith(")"):
             if isinstance(obj, list):
                 obj = list(enumerate(obj))
@@ -253,12 +248,24 @@ def _trace(self, obj, i: int, path):
                 for k, v in obj:
                     self._trace(v, i + 1, f"{path}{JSONPath.SEP}{k}")
             else:
-                raise ExprSyntaxError("sort operate must acting on list or dict")
+                raise ExprSyntaxError("sorter must acting on list or dict")
             return
 
+        # field-extractor
+        if step.startswith("(") and step.endswith(")"):
+            if isinstance(obj, dict):
+                obj_ = {}
+                for k in step[1:-1].split(","):
+                    obj_[k] = obj.get(k)
+                self._trace(obj_, i + 1, path)
+            else:
+                raise ExprSyntaxError("field-extractor must acting on list or dict")
+
 
 if __name__ == "__main__":
     with open("test/data/2.json", "rb") as f:
         d = json.load(f)
-    D = JSONPath("$.book[/(price)].price").parse(d, "PATH")
+    D = JSONPath("$.book[*].(title)").parse(d, "VALUE")
     print(D)
+    for v in D:
+        print(v)

test/conftest.py

@@ -48,6 +48,37 @@
             ["v0.0.1", "v1.0.0", "v1.0.2", "v1.0.3"],
         ),
         TestCase("$.scores[/(score)].score", data, [60, 85, 90, 95, 100]),
+        TestCase(
+            "$.scores[/(score)].(score)",
+            data,
+            [
+                {"score": 60},
+                {"score": 85},
+                {"score": 90},
+                {"score": 95},
+                {"score": 100},
+            ],
+        ),
+        TestCase(
+            "$.book[*].(title)",
+            data,
+            [
+                {"title": "Sayings of the Century"},
+                {"title": "Sword of Honour"},
+                {"title": "Moby Dick"},
+                {"title": "The Lord of the Rings"},
+            ],
+        ),
+        TestCase(
+            "$.book[/(category,price)].(title,price)",
+            data,
+            [
+                {"title": "Moby Dick", "price": 8.99},
+                {"title": "Sword of Honour", "price": 12.99},
+                {"title": "The Lord of the Rings", "price": 22.99},
+                {"title": "Sayings of the Century", "price": 8.95},
+            ],
+        ),
     ]
 )
 def value_cases(request):