beswarm 0.3.12__py3-none-any.whl → 0.3.13__py3-none-any.whl

beswarm/aient/aient/architext/architext/core.py

@@ -233,73 +233,116 @@ class Files(ContextProvider):
     def __init__(self, *paths: Union[str, List[str]], name: str = "files", visible: bool = True):
         super().__init__(name, visible=visible)
         self._files: Dict[str, str] = {}
+        self._file_sources: Dict[str, Dict] = {}
 
         file_paths: List[str] = []
         if paths:
-            # Handle the case where the first argument is a list of paths, e.g., Files(['a', 'b'])
             if len(paths) == 1 and isinstance(paths[0], list):
                 file_paths.extend(paths[0])
-            # Handle the case where arguments are individual string paths, e.g., Files('a', 'b')
             else:
                 file_paths.extend(paths)
 
         if file_paths:
             for path in file_paths:
-                try:
-                    with open(path, 'r', encoding='utf-8') as f:
-                        self._files[path] = f.read()
-                except FileNotFoundError:
-                    logging.warning(f"File not found during initialization: {path}. Skipping.")
-                except Exception as e:
-                    logging.error(f"Error reading file {path} during initialization: {e}")
+                self.update(path)
+
+    def _read_from_disk(self, path: str, head: Optional[int] = None) -> str:
+        """Reads content from a file on disk, respecting the head parameter."""
+        try:
+            with open(path, 'r', encoding='utf-8') as f:
+                if head is not None and head > 0:
+                    lines = []
+                    for _ in range(head):
+                        try:
+                            lines.append(next(f))
+                        except StopIteration:
+                            break
+                    return "".join(lines).rstrip('\n')
+                else:
+                    return f.read()
+        except FileNotFoundError:
+            raise
+        except Exception as e:
+            logging.error(f"Error reading file {path}: {e}")
+            return f"[Error: Could not read file at path '{path}': {e}]"
 
     async def refresh(self):
         """
-        Overrides the default refresh behavior. It synchronizes the content of
-        all tracked files with the file system. If a file is not found, its
-        content is updated to reflect the error.
+        Synchronizes content for files sourced from disk.
+        Content set manually is overwritten if the file exists, but preserved if it does not.
         """
         is_changed = False
-        for path in list(self._files.keys()):
-            try:
-                with open(path, 'r', encoding='utf-8') as f:
-                    new_content = f.read()
-                if self._files.get(path) != new_content:
-                    self._files[path] = new_content
-                    is_changed = True
-            except FileNotFoundError:
-                error_msg = f"[Error: File not found at path '{path}']"
-                if self._files.get(path) != error_msg:
-                    self._files[path] = error_msg
-                    is_changed = True
-            except Exception as e:
-                error_msg = f"[Error: Could not read file at path '{path}': {e}]"
-                if self._files.get(path) != error_msg:
-                    self._files[path] = error_msg
-                    is_changed = True
+        for path, spec in list(self._file_sources.items()):
+            if spec.get('source') == 'disk':
+                try:
+                    head = spec.get('head')
+                    new_content = self._read_from_disk(path, head)
+                    if self._files.get(path) != new_content:
+                        self._files[path] = new_content
+                        is_changed = True
+                except FileNotFoundError:
+                    error_msg = f"[Error: File not found at path '{path}']"
+                    if self._files.get(path) != error_msg:
+                        self._files[path] = error_msg
+                        is_changed = True
+            elif spec.get('source') == 'manual':
+                try:
+                    # File exists, so we must overwrite manual content.
+                    head = spec.get('head')
+                    new_content = self._read_from_disk(path, head)
+                    if self._files.get(path) != new_content:
+                        self._files[path] = new_content
+                        is_changed = True
+                    # If we are here, manual content was overwritten by disk content,
+                    # so we should update the source.
+                    self._file_sources[path] = {'source': 'disk'}
+
+                except FileNotFoundError:
+                    # File does not exist, so we keep the manual content. No change.
+                    pass
 
         if is_changed:
             self.mark_stale()
-
         await super().refresh()
 
-    def update(self, path: str, content: Optional[str] = None):
+    def update(self, path: str, content: Optional[str] = None, head: Optional[Union[int, str]] = None):
         """
-        Updates a single file. If content is provided, it updates the file in
-        memory. If content is None, it reads the file from disk.
+        Updates a single file's content and its source specification.
         """
+        if head is not None:
+            try:
+                head = int(head)
+            except (ValueError, TypeError):
+                logging.warning(f"Invalid 'head' parameter for file '{path}': {head}. Must be an integer. Ignoring.")
+                head = None
+
         if content is not None:
-            self._files[path] = content
+            # New logic: if head is also provided, decide what to do with content.
+            if head is not None and head > 0:
+                try:
+                    # If file exists, prioritize reading from disk.
+                    self._files[path] = self._read_from_disk(path, head)
+                    self._file_sources[path] = {'source': 'disk', 'head': head}
+                except FileNotFoundError:
+                    # If file does not exist, use the provided content's head.
+                    lines = content.split('\n')
+                    self._files[path] = "\n".join(lines[:head])
+                    self._file_sources[path] = {'source': 'manual', 'head': head}
+            else:
+                # Original logic for when only content is provided.
+                self._files[path] = content
+                self._file_sources[path] = {'source': 'manual'}
         else:
+            # Original logic for when only path (and optional head) is provided.
             try:
-                with open(path, 'r', encoding='utf-8') as f:
-                    self._files[path] = f.read()
+                self._files[path] = self._read_from_disk(path, head)
+                spec = {'source': 'disk'}
+                if head is not None and head > 0:
+                    spec['head'] = head
+                self._file_sources[path] = spec
             except FileNotFoundError:
-                logging.error(f"File not found for update: {path}.")
                 self._files[path] = f"[Error: File not found at path '{path}']"
-            except Exception as e:
-                logging.error(f"Error reading file for update {path}: {e}.")
-                self._files[path] = f"[Error: Could not read file at path '{path}': {e}]"
+                self._file_sources[path] = {'source': 'disk'}
         self.mark_stale()
     async def render(self) -> str:
         if not self._files: return None

beswarm/aient/aient/architext/test/test.py

@@ -1783,6 +1783,107 @@ Files: {Files(visible=True, name="files")}
         self.assertEqual(len(rendered_separated), 1, "被空消息隔开的同角色消息在渲染时应该合并")
         self.assertEqual(rendered_separated[0]['content'], "hihi2", "被空消息隔开的同角色消息合并后内容不正确")
 
+    async def test_zzc_files_update_with_head(self):
+        """测试 Files.update 是否支持 head 参数以及新的 refresh 逻辑"""
+        test_file = "test_file_head.txt"
+        file_content = "Line 1\nLine 2\nLine 3\nLine 4\nLine 5"
+        with open(test_file, "w", encoding='utf-8') as f:
+            f.write(file_content)
+
+        try:
+            messages = Messages(UserMessage(Files(name="files")))
+
+            # 1. Test update with head=3 using the provider from Messages
+            files_provider = messages.provider("files")
+            self.assertIsInstance(files_provider, Files)
+            files_provider.update(path=test_file, head=3)
+
+            rendered_head = await messages.render_latest()
+            expected_head_content = "Line 1\nLine 2\nLine 3"
+            self.assertIn(f"<file_content>{expected_head_content}</file_content>", rendered_head[0]['content'])
+            self.assertNotIn("Line 4", rendered_head[0]['content'])
+
+            # 2. Test update without head (default behavior)
+            files_provider.update(path=test_file)
+            rendered_full = await messages.render_latest()
+            self.assertIn(f"<file_content>{file_content}</file_content>", rendered_full[0]['content'])
+
+            # 3. Test that refresh overwrites manual content when file exists
+            files_provider.update(path=test_file, content="manual content")
+            rendered_overwritten = await messages.render_latest()
+            self.assertIn(f"<file_content>{file_content}</file_content>", rendered_overwritten[0]['content'])
+            self.assertNotIn("manual content", rendered_overwritten[0]['content'])
+
+            # 4. Test that manual content is kept if file does not exist
+            non_existent_file = "non_existent_for_sure.txt"
+            files_provider.update(path=non_existent_file, content="manual content to keep")
+            rendered_kept = await messages.render_latest()
+            self.assertIn("<file_content>manual content to keep</file_content>", rendered_kept[0]['content'])
+
+        finally:
+            if os.path.exists(test_file):
+                os.remove(test_file)
+
+    async def test_zzd_files_update_with_string_head(self):
+        """测试 Files.update 是否能自动将字符串 head 参数转换为整数"""
+        test_file = "test_file_string_head.txt"
+        file_content = "Line 1\nLine 2\nLine 3\nLine 4\nLine 5"
+        with open(test_file, "w", encoding='utf-8') as f:
+            f.write(file_content)
+
+        try:
+            messages = Messages(UserMessage(Files(name="files")))
+
+            # Test update with head="3"
+            files_provider = messages.provider("files")
+            self.assertIsInstance(files_provider, Files)
+            # This is the part that will likely fail before implementation
+            files_provider.update(path=test_file, head="3")
+
+            rendered_head = await messages.render_latest()
+            expected_head_content = "Line 1\nLine 2\nLine 3"
+            self.assertIn(f"<file_content>{expected_head_content}</file_content>", rendered_head[0]['content'])
+            self.assertNotIn("Line 4", rendered_head[0]['content'])
+
+        finally:
+            if os.path.exists(test_file):
+                os.remove(test_file)
+
+    async def test_zzc_files_update_with_head_and_content(self):
+        """测试update同时设置path, content, head时的行为"""
+        existing_file = "test_file_existing.txt"
+        non_existent_file = "test_file_non_existent.txt"
+        existing_content = "Line 1 (from file)\nLine 2 (from file)\nLine 3 (from file)\nLine 4 (from file)"
+        manual_content = "Line 1 (manual)\nLine 2 (manual)\nLine 3 (manual)\nLine 4 (manual)"
+
+        # --- 场景1: 文件存在 ---
+        with open(existing_file, "w", encoding='utf-8') as f:
+            f.write(existing_content)
+        try:
+            messages = Messages(UserMessage(Files(name="files")))
+            files_provider = messages.provider("files")
+            files_provider.update(path=existing_file, content=manual_content, head=2)
+            rendered_existing = await messages.render_latest()
+            expected_existing_head = "Line 1 (from file)\nLine 2 (from file)"
+            self.assertIn(f"<file_content>{expected_existing_head}</file_content>", rendered_existing[0]['content'])
+            self.assertNotIn("manual", rendered_existing[0]['content'])
+        finally:
+            if os.path.exists(existing_file):
+                os.remove(existing_file)
+
+        # --- 场景2: 文件不存在 ---
+        try:
+            messages = Messages(UserMessage(Files(name="files")))
+            files_provider = messages.provider("files")
+            files_provider.update(path=non_existent_file, content=manual_content, head=2)
+            rendered_non_existent = await messages.render_latest()
+            expected_manual_head = "Line 1 (manual)\nLine 2 (manual)"
+            self.assertIn(f"<file_content>{expected_manual_head}</file_content>", rendered_non_existent[0]['content'])
+            self.assertNotIn("from file", rendered_non_existent[0]['content'])
+        finally:
+            if os.path.exists(non_existent_file):
+                os.remove(non_existent_file)
+
 # ==============================================================================
 # 6. 演示
 # ==============================================================================

beswarm/aient/aient/models/chatgpt.py

@@ -615,7 +615,7 @@ class chatgpt(BaseLLM):
                 final_tool_response = tool_response
                 if "<tool_error>" not in tool_response:
                     if tool_name == "read_file":
-                        self.conversation[convo_id].provider("files").update(tool_info['parameter']["file_path"], tool_response)
+                        self.conversation[convo_id].provider("files").update(tool_info['parameter']["file_path"], tool_response, head=safe_get(tool_info, 'parameter', "head", default=None))
                         final_tool_response = "Read file successfully! The file content has been updated in the tag <latest_file_content>."
                     elif tool_name == "get_knowledge_graph_tree":
                         self.conversation[convo_id].provider("knowledge_graph").visible = True

beswarm/tools/read_file.py

@@ -2,7 +2,6 @@ import os
 import json
 import chardet
 from pdfminer.high_level import extract_text
-
 from ..aient.aient.plugins import register_tool
 from ..core import current_work_dir
 
@@ -52,9 +51,7 @@ Examples:
 </read_file>
     """
 
-    work_dir = current_work_dir.get()
-    if not work_dir:
-        return f"<tool_error>关键上下文 'current_work_dir' 未设置，无法确定安全的工作目录。</tool_error>"
+    work_dir = current_work_dir.get(os.getcwd())
 
     # Determine the final, absolute path for the file operation.
     if os.path.isabs(file_path):
@@ -62,12 +59,7 @@ Examples:
     else:
         final_path = os.path.join(work_dir, file_path)
 
-    # Security check: Ensure the final path is within the designated work directory.
-    # abs_work_dir = os.path.abspath(work_dir)
     abs_final_path = os.path.abspath(final_path)
-    # if not abs_final_path.startswith(abs_work_dir):
-    #     return f"<tool_error>路径遍历攻击被阻止。尝试写入的路径 '{file_path}' 解析后超出了允许的工作目录范围 '{abs_work_dir}'。</tool_error>"
-
     file_path = abs_final_path
 
     try:
@@ -125,16 +117,23 @@ Examples:
             # 更新：修改通用文件读取逻辑以支持多种编码
             # 这部分替换了原有的 else 块内容
             try:
+                # 优化：分块读取以加速 chardet
                 with open(file_path, 'rb') as file: # 以二进制模式读取
-                    raw_data = file.read()
+                    # 1. 读取用于检测编码的初始块
+                    detection_chunk = file.read(4096) # Read first 4KB
+                    # 2. 读取文件的剩余部分
+                    remaining_data = file.read()
+
+                # 将两部分数据合并为完整的文件内容
+                raw_data = detection_chunk + remaining_data
 
                 if not raw_data: # 处理空文件
                     text_content = ""
                 else:
-                    detected_info = chardet.detect(raw_data)
+                    # 3. 仅对初始块运行 chardet 以提高速度
+                    detected_info = chardet.detect(detection_chunk) # Detect on the small chunk
                     primary_encoding_to_try = detected_info['encoding']
                     confidence = detected_info['confidence']
-
                     decoded_successfully = False
 
                     # 尝试1: 使用检测到的编码 (如果置信度高且编码有效)
@@ -193,14 +192,6 @@ Examples:
                 # Invalid head value, ignore and proceed with normal logic.
                 pass
 
-        # if file_path.lower().endswith('.csv'):
-        #     lines = text_content.splitlines(True)
-        #     if len(lines) > 500:
-        #         top_lines = lines[:250]
-        #         bottom_lines = lines[-250:]
-        #         omitted_count = len(lines) - 500
-        #         text_content = "".join(top_lines) + f"\n... (中间省略了 {omitted_count} 行) ...\n" + "".join(bottom_lines)
-
         # 返回文件内容
         return text_content
 
@@ -213,7 +204,8 @@ Examples:
         return f"<tool_error>读取文件时发生错误: {e}</tool_error>"
 
 if __name__ == "__main__":
-    # python -m beswarm.aient.aient.plugins.read_file
-    result = read_file("./work/cax/Lenia Notebook.ipynb", head=10)
+    # python -m beswarm.tools.read_file
+    # result = read_file("traindata.csv", head=3)
+    result = read_file("testdata.csv", head=3)
     print(result)
-    print(len(result))
+    print(f"行数: {len(result.splitlines())}")

{beswarm-0.3.12.dist-info → beswarm-0.3.13.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: beswarm
-Version: 0.3.12
+Version: 0.3.13
 Summary: MAS
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown

{beswarm-0.3.12.dist-info → beswarm-0.3.13.dist-info}/RECORD

@@ -10,9 +10,9 @@ beswarm/agents/chatgroup.py,sha256=PzrmRcDKAbB7cxL16nMod_CzPosDV6bfTmXxQVuv-AQ,1
 beswarm/agents/planact.py,sha256=dhPkMUaZLc1dO0TrI1Orr_i60iLzJHQIYBI_QjjSJwI,18723
 beswarm/aient/aient/__init__.py,sha256=SRfF7oDVlOOAi6nGKiJIUK6B_arqYLO9iSMp-2IZZps,21
 beswarm/aient/aient/architext/architext/__init__.py,sha256=79Ih1151rfcqZdr7F8HSZSTs_iT2SKd1xCkehMsXeXs,19
-beswarm/aient/aient/architext/architext/core.py,sha256=Sid3_Bf1a988xOWIfXOrojeCSn6ElJ4Hqxy6rmDnqFY,35591
+beswarm/aient/aient/architext/architext/core.py,sha256=3io21wNZ-Za8VpQZn9nnc8qIjfON9fxMxURMC7fYCPY,37559
 beswarm/aient/aient/architext/test/openai_client.py,sha256=Dqtbmubv6vwF8uBqcayG0kbsiO65of7sgU2-DRBi-UM,4590
-beswarm/aient/aient/architext/test/test.py,sha256=c3ev6as7k9eDpV-le1xvj--x_hUN2xXH-oIzyQ5hsjA,83798
+beswarm/aient/aient/architext/test/test.py,sha256=kWsvKq8lYLuRVlU6Sqe7PyQAP15ENpemy5EV6Se4UpE,89019
 beswarm/aient/aient/architext/test/test_save_load.py,sha256=o8DqH6gDYZkFkQy-a7blqLtJTRj5e4a-Lil48pJ0V3g,3260
 beswarm/aient/aient/core/__init__.py,sha256=NxjebTlku35S4Dzr16rdSqSTWUvvwEeACe8KvHJnjPg,34
 beswarm/aient/aient/core/log_config.py,sha256=kz2_yJv1p-o3lUQOwA3qh-LSc3wMHv13iCQclw44W9c,274
@@ -27,7 +27,7 @@ beswarm/aient/aient/core/test/test_payload.py,sha256=8jBiJY1uidm1jzL-EiK0s6UGmW9
 beswarm/aient/aient/models/__init__.py,sha256=ZTiZgbfBPTjIPSKURE7t6hlFBVLRS9lluGbmqc1WjxQ,43
 beswarm/aient/aient/models/audio.py,sha256=FNW4lxG1IhxOU7L8mvcbaeC1nXk_lpUZQlg9ijQ0h_Q,1937
 beswarm/aient/aient/models/base.py,sha256=HWIGfa2A7OTccvHK0wG1-UlHB-yaWRC7hbi4oR1Mu1Y,7228
-beswarm/aient/aient/models/chatgpt.py,sha256=d1ZE12AQriIl8DF6OQ3612_ieP5cbGdBhngYVUdIhKs,43814
+beswarm/aient/aient/models/chatgpt.py,sha256=ZM0Ol_0vTZ5-5NjTolVfQzN47OyH1e8eJEDf2QRqwHA,43875
 beswarm/aient/aient/plugins/__init__.py,sha256=p3KO6Aa3Lupos4i2SjzLQw1hzQTigOAfEHngsldrsyk,986
 beswarm/aient/aient/plugins/arXiv.py,sha256=yHjb6PS3GUWazpOYRMKMzghKJlxnZ5TX8z9F6UtUVow,1461
 beswarm/aient/aient/plugins/config.py,sha256=TGgZ5SnNKZ8MmdznrZ-TEq7s2ulhAAwTSKH89bci3dA,7079
@@ -111,7 +111,7 @@ beswarm/tools/deep_search.py,sha256=O5n9L1foEdfNiwVpjnbM6_1X-MT4kQ2lSWKRvYWhbEg,
 beswarm/tools/edit_file.py,sha256=ZTJvbpsfRlp2t98kTn9XQ5qZBTdsWJVWv9t0lvK4RfU,9147
 beswarm/tools/graph.py,sha256=Ni51LCB65bLDDLC3ZyEJj_BeHAzhimhFgH9Ekr56Ll8,5871
 beswarm/tools/planner.py,sha256=vsHd7rE8RQHJrZ7BQ0ZXhbt4Fjh3DeyxU4piA5R-VPM,1253
-beswarm/tools/read_file.py,sha256=_MX2y-EtBH3wxcCBUtkAaiU4GZX34z2q3VKXjSs9gLs,10060
+beswarm/tools/read_file.py,sha256=ytXkYGzI0ryNmj3_F2kv5qxL2Yb4xWDmpFrvvel3XzI,9675
 beswarm/tools/repomap.py,sha256=w98aHmjNjtvcUVc5maWORqzKqDy2KVGLooOe__uJVCU,45235
 beswarm/tools/request_input.py,sha256=3n2UW8m8Q7dxGhd7L7hzSJ1kk4ekMbtdtNZZT3dJf20,938
 beswarm/tools/screenshot.py,sha256=hyL6F8_k9Y03Nb_X18cY-klCpWWdkqyC-iGXfKX-7jc,1007
@@ -121,8 +121,8 @@ beswarm/tools/subtasks.py,sha256=Fsf_542CkECcwFwxD-F38_IUsmk06xe1P7e6pBPTy4Y,128
 beswarm/tools/worker.py,sha256=mQ1qdrQ8MgL99byAbTvxfEByFFGN9mty3UHqHjARMQ8,2331
 beswarm/tools/write_csv.py,sha256=u0Hq18Ksfheb52MVtyLNCnSDHibITpsYBPs2ub7USYA,1466
 beswarm/tools/write_file.py,sha256=L2coBqz-aRFxPBvJBrbWbUJhu7p3oKAAGb9R144bFtk,4926
-beswarm-0.3.12.dist-info/METADATA,sha256=lsjwHlJwRdbZPGPqsihISy5xKb1Rim2bsiTcQaYJXVo,3878
-beswarm-0.3.12.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-beswarm-0.3.12.dist-info/entry_points.txt,sha256=URK7Y4PDzBgxIecQnxsWTu4O-eaFa1CoAcNTWh5R7LM,45
-beswarm-0.3.12.dist-info/top_level.txt,sha256=pJw4O87wvt5882smuSO6DfByJz7FJ8SxxT8h9fHCmpo,8
-beswarm-0.3.12.dist-info/RECORD,,
+beswarm-0.3.13.dist-info/METADATA,sha256=MQHN-a_l_axNIuLC3PR6h0r_3xWYTkMiy4dBC-K3frs,3878
+beswarm-0.3.13.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+beswarm-0.3.13.dist-info/entry_points.txt,sha256=URK7Y4PDzBgxIecQnxsWTu4O-eaFa1CoAcNTWh5R7LM,45
+beswarm-0.3.13.dist-info/top_level.txt,sha256=pJw4O87wvt5882smuSO6DfByJz7FJ8SxxT8h9fHCmpo,8
+beswarm-0.3.13.dist-info/RECORD,,