deepfos 1.1.67__tar.gz → 1.1.68__tar.gz

{deepfos-1.1.67 → deepfos-1.1.68}/CHANGELOG.md

@@ -1,3 +1,12 @@
+## [1.1.68] - 2025-07-29
+
+### 更新
+
+* DeepModel元素insert_df和update_df接受chunksize=None
+* DeepModel元素insert_df和update_df组织的bulk语句优化
+* 修复DataFrame.replace会引发类型推断导致的bug
+
+
 ## [1.1.67] - 2025-07-22
 
 ### 新增

{deepfos-1.1.67 → deepfos-1.1.68}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: deepfos
-Version: 1.1.67
+Version: 1.1.68
 Summary: Collecions of useful and handy tools for deepfos platform
 Home-page: http://py.deepfos.com
 Author: deepfos-python-team

{deepfos-1.1.67 → deepfos-1.1.68}/deepfos/_version.py

@@ -8,11 +8,11 @@ import json
 
 version_json = '''
 {
- "date": "2025-07-22T03:24:23+0000",
+ "date": "2025-08-07T03:10:50+0000",
  "dirty": false,
  "error": null,
- "full-revisionid": "61559f9a44520b8bcfced0acc40f356ed3c1e6d7",
- "version": "1.1.67"
+ "full-revisionid": "e1b02f36c62e37f03965c5219dc9426d19af95a8",
+ "version": "1.1.68"
 }
 '''  # END VERSION_JSON
 

{deepfos-1.1.67 → deepfos-1.1.68}/deepfos/element/deepmodel.py

@@ -1,4 +1,5 @@
 import re
+import textwrap
 import threading
 
 import numpy as np
@@ -121,6 +122,7 @@ dm_type_to_edb_scalar = {
     'uuid': 'std::str',
     'json': 'std::json',
 }
+TAB = ' ' * 4
 
 
 class ObjectElement(ObjectParam):
@@ -485,19 +487,30 @@ def _iter_single_assign(
     field: PtrInfo,
     cast_type: str,
     target_main_field: Dict[str, MainField]
-):
-    assign_string = f"{field.name} := "
+) -> str:
+    """
+    生成单字段赋值语句
+
+    Args:
+        field: 字段信息
+        cast_type: 字段类型
+        target_main_field: 目标字段信息
+
+    Returns:
+        赋值语句
+    """
+    assign = f"\n{field.name} := "
     # 设置标量值
     if field.name not in target_main_field:
         if field.is_multi:
-            return assign_string + f"json_array_unpack(item['{field.name}'])"
+            return assign + f"json_array_unpack(item['{field.name}'])"
 
-        assign_string += f"<{cast_type}>"
+        assign += f"<{cast_type}>"
 
         if cast_type in NEED_CAST_STR:
-            assign_string += '<std::str>'
+            assign += '<std::str>'
 
-        return assign_string + f"item['{field.name}']"
+        return assign + f"item['{field.name}']"
 
     # 设置link target值
     link = field.name
@@ -516,20 +529,20 @@ def _iter_single_assign(
         target = cast_type
 
     if main_field.is_multi:
-        assign_string += f"""(
-            select detached {target}
-            filter contains(
-                <array<{main_field.type}>>(json_get(item, '{link}', 'target')),
-                .{main_field.business_key}
-            )
-        )"""
+        assign += 'distinct (\n' + textwrap.indent(textwrap.dedent(f"""\
+        for each_{link} in json_array_unpack(json_get(item, '{link}', 'target'))
+        union (
+            select detached {target} 
+            filter .{main_field.business_key} = <{main_field.type}>each_{link}
+        )"""), TAB) + '\n)'
     else:
-        assign_string += f"""assert_single((
+        assign += textwrap.dedent(f"""\
+        assert_single((
             select detached {target}
             filter .{main_field.business_key} = <{main_field.type}>(json_get(item, '{link}'))
-        ))"""
+        ))""")
 
-    return assign_string
+    return assign
 
 
 def bulk_insert_by_fields(
@@ -537,21 +550,16 @@ def bulk_insert_by_fields(
     field_type: List[PtrInfo],
     target_main_field: Dict[str, MainField],
 ):
-    insert_assign_body = ','.join(
-        [
-            _iter_single_assign(field, field.type, target_main_field)
-            for field in field_type
-        ]
-    )
-
-    return f"""
-    with raw_data := <json>to_json(<std::str>${BATCH_INSERT_KW}),
-    for item in json_array_unpack(raw_data) union (
-        insert {object_name} {{
-            {insert_assign_body}
-        }}
-    )
-    """
+    insert_assign_body = ','.join([
+        _iter_single_assign(field, field.type, target_main_field)
+        for field in field_type
+    ])
+    return textwrap.dedent(f"""
+        with raw_data := <json>to_json(<std::str>${BATCH_INSERT_KW}),
+        for item in json_array_unpack(raw_data) union (
+            insert {object_name} {{{textwrap.indent(insert_assign_body, TAB * 4)}
+            }}
+        )""")
 
 
 def bulk_upsert_by_fields(
@@ -562,34 +570,27 @@ def bulk_upsert_by_fields(
     update_fields: Iterable[str]
 ):
     conflict_on_fields = map(lambda n: f'.{n}', exclusive_fields)
-
-    insert_assign_body = ','.join(
-        [
-            _iter_single_assign(field, field.type, target_main_field)
-            for field in field_type
-        ]
-    )
+    insert_assign_body = ','.join([
+        _iter_single_assign(field, field.type, target_main_field)
+        for field in field_type
+    ])
     update_assign_body = ','.join(
         [
             _iter_single_assign(field, field.type, target_main_field)
             for field in field_type if field.name in update_fields
         ]
     )
-
-    return f"""
+    return textwrap.dedent(f"""
         with raw_data := <json>to_json(<std::str>${BATCH_INSERT_KW}),
         for item in json_array_unpack(raw_data) union (
-            insert {object_name} {{
-                {insert_assign_body}
+            insert {object_name} {{{textwrap.indent(insert_assign_body, TAB * 4)}
             }}
             unless conflict on ({','.join(conflict_on_fields)})
             else (
-                update {object_name} set {{
-                    {update_assign_body}
+                update {object_name} set {{{textwrap.indent(update_assign_body, TAB * 5)}
                 }}
             )
-        )
-        """
+        )""")
 
 
 def bulk_update_by_fields(
@@ -599,32 +600,27 @@ def bulk_update_by_fields(
     match_fields: Iterable[str],
     update_fields: Iterable[str],
 ):
-    update_assign_body = ','.join(
-        [
-            _iter_single_assign(field, field.type, target_main_field)
-            for field in field_type if field.name in update_fields
-        ]
-    )
+    update_assign_body = ','.join([
+        _iter_single_assign(field, field.type, target_main_field)
+        for field in field_type if field.name in update_fields
+    ])
 
     field_type_map = {field.name: field.type for field in field_type}
-
     match_str = " and ".join(
         [
             f".{name} = <{field_type_map.get(name, 'std::str')}>item['{name}']"
             for name in match_fields
         ]
     )
-
-    return f"""
-        with raw_data := <json>to_json(<std::str>${BATCH_INSERT_KW}),
-        for item in json_array_unpack(raw_data) union (
-            update {object_name} 
-            filter {match_str}
-            set {{
-                {update_assign_body}
-            }}
-        )
-        """
+    return textwrap.dedent(f"""
+    with raw_data := <json>to_json(<std::str>${BATCH_INSERT_KW}),
+    for item in json_array_unpack(raw_data) union (
+        update {object_name} 
+        filter {match_str}
+        set {{{textwrap.indent(update_assign_body, TAB * 3)}
+        }}
+    )
+    """)
 
 
 def format_obj(obj: edgedb.Object) -> ObjectTypeFrame:
@@ -1302,6 +1298,10 @@ class AsyncDeepModel(ElementBase[DeepModelAPI]):
             name = field.name
             # 链接至其他对象，记录目标对象信息
             if is_multi:
+                if name not in relation:
+                    raise ValueError(
+                        f'对象[{object_name}]的多选链接:[{name}]未定义在relation中'
+                    )
                 link_props = set(relation[name].columns).intersection(field.props)
             else:
                 link_props = set(
@@ -1442,6 +1442,8 @@ class AsyncDeepModel(ElementBase[DeepModelAPI]):
                 bulk_ql = None
         qls = []
         self._ensure_client()
+        if chunk_size is None:
+            chunk_size = len(data)
         for i in range(0, len(data), chunk_size):
             part = structure.fit(data.iloc[i: i + chunk_size])
             ql_chunk = []

{deepfos-1.1.67 → deepfos-1.1.68}/deepfos/element/finmodel.py

@@ -824,8 +824,8 @@ class AsyncFinancialCube(ElementBase[FinancialModelAPI]):
     ):
         # replace NaN to standard None
         # NB: replace twice in case of infer None to nan happened in 2.x pandas
-        data = data.replace({None: np.nan})
-        data = data.replace({np.nan: None})
+        data[DFLT_DATA_COLUMN] = data[DFLT_DATA_COLUMN].replace({None: np.nan})
+        data[DFLT_DATA_COLUMN] = data[DFLT_DATA_COLUMN].replace({np.nan: None})
         # ensure view is capitalized
         if self._meta.autoCalculation:
             data = data.rename(columns=VIEW_DICT)

{deepfos-1.1.67 → deepfos-1.1.68}/deepfos/options.py

@@ -1,5 +1,6 @@
 import configparser
 import sys
+import threading
 import warnings
 from enum import Enum
 from functools import partial
@@ -276,6 +277,10 @@ class _Option(Generic[T_Opt]):
             self.unset_action = _Action.PASS
         self.unset = False
 
+    def _quick_set(self, instance, value):
+        setattr(instance, self._var_name, value)
+        self.unset = False
+
     def __str__(self):
         return self._display_name
 
@@ -366,7 +371,7 @@ class _Category:
         setattr(self, '_name', getattr(other, '_name'))
         for attr, option in self._options.items():
             try:
-                option.__set__(self, getattr(other, attr))
+                option._quick_set(self, getattr(other, attr))  # noqa
             except BaseOptionError:
                 continue
 
@@ -643,19 +648,31 @@ class _OptionCTX:
         boost: _Boost
         discovery: _ServiceDiscovery
         nacos: _Nacos
+        edgedb: _Edgedb
 
         def show_options(self, category=None):
             ...
 
     def __init__(self):
-        self._token = _option_ctx.set(_GlobalOptions())
+        self._create_at_main = threading.current_thread() is threading.main_thread()
+        self._primary_opt = _GlobalOptions()
+        self._token = _option_ctx.set(self._primary_opt)
 
     def create_local(self):
         self._token = _option_ctx.set(_GlobalOptions())
 
     @property
     def _option(self) -> _GlobalOptions:
-        return _option_ctx.get()
+        try:
+            return _option_ctx.get()
+        except LookupError:
+            if (
+                not self._create_at_main
+                and threading.current_thread() is threading.main_thread()
+            ):
+                _option_ctx.set(self._primary_opt.__copy__())
+                return _option_ctx.get()
+            raise
 
     def __getattr__(self, item):
         return getattr(self._option, item)

{deepfos-1.1.67 → deepfos-1.1.68}/deepfos.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: deepfos
-Version: 1.1.67
+Version: 1.1.68
 Summary: Collecions of useful and handy tools for deepfos platform
 Home-page: http://py.deepfos.com
 Author: deepfos-python-team