auto-coder 0.1.201__py3-none-any.whl → 0.1.202__py3-none-any.whl

{auto_coder-0.1.201.dist-info → auto_coder-0.1.202.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: auto-coder
-Version: 0.1.201
+Version: 0.1.202
 Summary: AutoCoder: AutoCoder
 Author: allwefantasy
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence

{auto_coder-0.1.201.dist-info → auto_coder-0.1.202.dist-info}/RECORD

@@ -7,7 +7,7 @@ autocoder/chat_auto_coder.py,sha256=kgcD4HKKmvDd2CI048TqZwdx2hrQRbZKYPO10qyrbA8,
 autocoder/chat_auto_coder_lang.py,sha256=QYtu5gWEQmWKVovR_qUZ8plySZarNFX_Onk-1vN9IiA,8524
 autocoder/command_args.py,sha256=MBsVjZpADu5u7CjY_1F8fGUW-9dVGyNDpzWyrDIMxz8,29890
 autocoder/lang.py,sha256=Ajng6m7towmx-cvQfEHPFp43iEfddPvr8ju5GH4H8qA,13819
-autocoder/version.py,sha256=sVvR48V5CyWVyLriw3601idt1dQMOAnWtcO_tdLfmow,24
+autocoder/version.py,sha256=KbKAfOQ1mRxioaKsb1VhjDtdk7RG8UIteq2BswiztKI,24
 autocoder/agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/agent/auto_tool.py,sha256=DBzip-P_T6ZtT2eHexPcusmKYD0h7ufzp7TLwXAY10E,11554
 autocoder/agent/coder.py,sha256=x6bdJwDuETGg9ebQnYlUWCxCtQcDGg73LtI6McpWslQ,72034
@@ -63,7 +63,7 @@ autocoder/rag/api_server.py,sha256=dRbhAZVRAOlZ64Cnxf4_rKb4iJwHnrWS9Zr67IVORw0,7
 autocoder/rag/doc_filter.py,sha256=Ha0Yae_G_hF72YzvrO7NoDZcG18K4hRcqGAEqfrIwAs,9330
 autocoder/rag/document_retriever.py,sha256=5oThtxukGuRFF96o3pHKsk306a8diXbhgSrbqyU2BvM,8894
 autocoder/rag/llm_wrapper.py,sha256=sbDxCANiZyWb_ocqNgqu2oy3c2t8orPNRGleEs-Uwl8,2649
-autocoder/rag/long_context_rag.py,sha256=kB8A8WuJJ7xYeSKtWI7pSmrYZk5Yo5HpFOm6sx7A2zI,23923
+autocoder/rag/long_context_rag.py,sha256=Y0bpO0mNOBGUtQ8WBOSbrjFLKxRUcZnD-Dax1zU7q-I,24436
 autocoder/rag/rag_config.py,sha256=8LwFcTd8OJWWwi1_WY4IzjqgtT6RyE2j4PjxS5cCTDE,802
 autocoder/rag/rag_entry.py,sha256=V1RJ8RGqM30DNPmzymv64rZjNRGWn6kfc8sRy_LECg0,2451
 autocoder/rag/raw_rag.py,sha256=yS2Ur6kG0IRjhCj2_VonwxjY_xls_E62jO5Gz5j2nqE,2952
@@ -94,7 +94,7 @@ autocoder/regexproject/__init__.py,sha256=lHTpHfYkguCMtczXoH4bMr-IMNZQtXIjmtSvjt
 autocoder/suffixproject/__init__.py,sha256=bEPW9AyGSQ8kNzrgKEXyRikrUCUEuJ6b6vCLzO8Ja6g,11017
 autocoder/tsproject/__init__.py,sha256=avSnMA3uSdZmv5MxTilDPFLRFmtfzFr1NPhHaRMFhX4,11625
 autocoder/utils/__init__.py,sha256=O3n6cpsgkIbbMuwmBHSQ1dls_IBD7_7YKFFaeKNo_tc,1193
-autocoder/utils/_markitdown.py,sha256=ZITzuo8D7dH3s9K9rmriYRciqTYpHIqRinS9FZ10pOc,46691
+autocoder/utils/_markitdown.py,sha256=RU88qn4eZfYIy0GDrPxlI8oYXIypbi63VRJjdlnE0VU,47431
 autocoder/utils/coder.py,sha256=rK8e0svQBe0NOP26dIGToUXgha_hUDgxlWoC_p_r7oc,5698
 autocoder/utils/conversation_store.py,sha256=sz-hhY7sttPAUOAQU6Pze-5zJc3j0_Emj22dM_0l5ro,1161
 autocoder/utils/llm_client_interceptors.py,sha256=FEHNXoFZlCjAHQcjPRyX8FOMjo6rPXpO2AJ2zn2KTTo,901
@@ -106,9 +106,9 @@ autocoder/utils/request_event_queue.py,sha256=r3lo5qGsB1dIjzVQ05dnr0z_9Z3zOkBdP1
 autocoder/utils/request_queue.py,sha256=nwp6PMtgTCiuwJI24p8OLNZjUiprC-TsefQrhMI-yPE,3889
 autocoder/utils/rest.py,sha256=HawagAap3wMIDROGhY1730zSZrJR_EycODAA5qOj83c,8807
 autocoder/utils/tests.py,sha256=BqphrwyycGAvs-5mhH8pKtMZdObwhFtJ5MC_ZAOiLq8,1340
-auto_coder-0.1.201.dist-info/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
-auto_coder-0.1.201.dist-info/METADATA,sha256=2iejkjbZEUMhfqT1A6k7MTJmTdpc0urt1dGGXtQ3KHA,2575
-auto_coder-0.1.201.dist-info/WHEEL,sha256=GV9aMThwP_4oNCtvEC2ec3qUYutgWeAzklro_0m4WJQ,91
-auto_coder-0.1.201.dist-info/entry_points.txt,sha256=0nzHtHH4pNcM7xq4EBA2toS28Qelrvcbrr59GqD_0Ak,350
-auto_coder-0.1.201.dist-info/top_level.txt,sha256=Jqc0_uJSw2GwoFQAa9iJxYns-2mWla-9ok_Y3Gcznjk,10
-auto_coder-0.1.201.dist-info/RECORD,,
+auto_coder-0.1.202.dist-info/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
+auto_coder-0.1.202.dist-info/METADATA,sha256=JU3TFfBNRhy5U-nqB2vFy2gLKLQs-E89lUX3B0MrNj8,2575
+auto_coder-0.1.202.dist-info/WHEEL,sha256=GV9aMThwP_4oNCtvEC2ec3qUYutgWeAzklro_0m4WJQ,91
+auto_coder-0.1.202.dist-info/entry_points.txt,sha256=0nzHtHH4pNcM7xq4EBA2toS28Qelrvcbrr59GqD_0Ak,350
+auto_coder-0.1.202.dist-info/top_level.txt,sha256=Jqc0_uJSw2GwoFQAa9iJxYns-2mWla-9ok_Y3Gcznjk,10
+auto_coder-0.1.202.dist-info/RECORD,,

autocoder/rag/long_context_rag.py

@@ -200,20 +200,27 @@ class LongContextRAG:
     def _answer_question(
         self, query: str, relevant_docs: List[str]
     ) -> Generator[str, None, None]:
-        """
-        使用以下文档来回答问题。如果文档中没有相关信息，请说"我没有足够的信息来回答这个问题"。
-
+        """        
         文档：
         {% for doc in relevant_docs %}
         {{ doc }}
         {% endfor %}
 
-        问题：{{ query }}
+        使用以上文档来回答用户的问题。回答要求：
+
+        1. 严格基于文档内容回答        
+        - 如果文档提供的信息无法回答问题,请明确回复:"抱歉,文档中没有足够的信息来回答这个问题。" 
+        - 不要添加、推测或扩展文档未提及的信息
 
-        要求：
-        1. 注意相应的markdown图片如果存在也要输出,尽可能图文并茂
-        
-        回答：
+        2. 格式如 ![image](./path.png) 的 Markdown 图片处理
+        - 根据Markdown 图片前后文本内容推测改图片与问题的相关性，有相关性则在回答中输出该Markdown图片路径
+        - 根据相关图片在文档中的位置，自然融入答复内容,保持上下文连贯
+        - 完整保留原始图片路径,不省略任何部分
+
+        3. 回答格式要求
+        - 使用markdown格式提升可读性
+
+        问题：{{ query }}
         """
 
     def _get_document_retriever_class(self):

autocoder/utils/_markitdown.py

@@ -68,7 +68,8 @@ class _CustomMarkdownify(markdownify.MarkdownConverter):
     """
 
     def __init__(self, **options: Any):
-        options["heading_style"] = options.get("heading_style", markdownify.ATX)
+        options["heading_style"] = options.get(
+            "heading_style", markdownify.ATX)
         # Explicitly cast options to the expected type if necessary
         super().__init__(**options)
 
@@ -318,7 +319,7 @@ class YouTubeConverter(DocumentConverter):
                     obj_start = lines[0].find("{")
                     obj_end = lines[0].rfind("}")
                     if obj_start >= 0 and obj_end >= 0:
-                        data = json.loads(lines[0][obj_start : obj_end + 1])
+                        data = json.loads(lines[0][obj_start: obj_end + 1])
                         attrdesc = self._findKey(
                             data, "attributedDescriptionBodyText"
                         )  # type: ignore
@@ -331,7 +332,8 @@ class YouTubeConverter(DocumentConverter):
         # Start preparing the page
         webpage_text = "# YouTube\n"
 
-        title = self._get(metadata, ["title", "og:title", "name"])  # type: ignore
+        title = self._get(
+            metadata, ["title", "og:title", "name"])  # type: ignore
         assert isinstance(title, str)
 
         if title:
@@ -468,7 +470,8 @@ class BingSerpConverter(DocumentConverter):
 
                     try:
                         # RFC 4648 / Base64URL" variant, which uses "-" and "_"
-                        a["href"] = base64.b64decode(u, altchars="-_").decode("utf-8")
+                        a["href"] = base64.b64decode(
+                            u, altchars="-_").decode("utf-8")
                     except UnicodeDecodeError:
                         pass
                     except binascii.Error:
@@ -477,7 +480,8 @@ class BingSerpConverter(DocumentConverter):
             # Convert to markdown
             md_result = _markdownify.convert_soup(result).strip()
             lines = [line.strip() for line in re.split(r"\n+", md_result)]
-            results.append("\n".join([line for line in lines if len(line) > 0]))
+            results.append(
+                "\n".join([line for line in lines if len(line) > 0]))
 
         webpage_text = (
             f"## A Bing search for '{query}' found the following results:\n\n"
@@ -507,7 +511,8 @@ class PdfConverter(DocumentConverter):
         else:
             # Create output directory for images if it doesn't exist
             image_output_dir = os.path.join(
-                os.path.dirname(local_path), "_images", os.path.basename(local_path)
+                os.path.dirname(local_path), "_images", os.path.basename(
+                    local_path).replace(" ", "_")
             )
         os.makedirs(image_output_dir, exist_ok=True)
 
@@ -545,18 +550,17 @@ class PdfConverter(DocumentConverter):
         self, layout, image_output_dir: str, image_count: int
     ) -> List[str]:
         """Process the layout of a PDF page, extracting both text and images."""
-        content = []        
-        iw = ImageWriter(image_output_dir)
-
+        content = [] 
+        local_image_count = image_count       
         for lt_obj in layout:
             # Handle images
             if isinstance(lt_obj, LTImage) or (
                 isinstance(lt_obj, LTFigure) and lt_obj.name.startswith("Im")
-            ):
-                image_count += 1
+            ):                
                 image_data = None
                 image_meta = {}
-                image_path = os.path.join(image_output_dir, f"image_{image_count}.png")
+                image_path = os.path.join(
+                    image_output_dir, f"image_{local_image_count}.png")
 
                 if hasattr(lt_obj, "stream"):
                     image_data = lt_obj.stream.get_data()
@@ -566,12 +570,15 @@ class PdfConverter(DocumentConverter):
 
                 if image_data:
                     if isinstance(lt_obj, LTImage):
+                        iw = ImageWriter(image_output_dir)
                         name = iw.export_image(lt_obj)
-                        suffix = os.path.splitext(name)[1]  
+                        suffix = os.path.splitext(name)[1]
                         temp_path = os.path.join(image_output_dir, name)
-                        image_path = os.path.join(image_output_dir, f"image_{image_count}{suffix}")
+                        image_path = os.path.join(
+                            image_output_dir, f"image_{local_image_count}{suffix}")
                         os.rename(temp_path, image_path)
-                        content.append(f"![Image {image_count}]({image_path})")
+                        content.append(f"![Image {local_image_count}]({image_path})")
+                        local_image_count += 1
                         continue
                     try:
                         # Try to handle raw pixel data
@@ -580,12 +587,14 @@ class PdfConverter(DocumentConverter):
                             height = image_meta["Height"]
                             bits = image_meta["BitsPerComponent"]
                             colorspace = image_meta["ColorSpace"].name
-                            new_image_data = np.frombuffer(image_data, dtype=np.uint8)
+                            new_image_data = np.frombuffer(
+                                image_data, dtype=np.uint8)
                             # Normalize to 8-bit if necessary
                             if bits != 8:
                                 max_val = (1 << bits) - 1
                                 new_image_data = (
-                                    new_image_data.astype("float32") * 255 / max_val
+                                    new_image_data.astype(
+                                        "float32") * 255 / max_val
                                 ).astype("uint8")
 
                             if colorspace == "DeviceRGB":
@@ -595,16 +604,19 @@ class PdfConverter(DocumentConverter):
                                 img = Image.fromarray(new_image_data, "RGB")
                                 img.save(image_path)
                                 content.append(
-                                    f"![Image {image_count}]({image_path})\n"
+                                    f"![Image {local_image_count}]({image_path})\n"
                                 )
+                                local_image_count += 1
                                 continue
                             elif colorspace == "DeviceGray":
-                                new_image_data = new_image_data.reshape((height, width))
+                                new_image_data = new_image_data.reshape(
+                                    (height, width))
                                 img = Image.fromarray(new_image_data, "L")
                                 img.save(image_path)
                                 content.append(
-                                    f"![Image {image_count}]({image_path})\n"
+                                    f"![Image {local_image_count}]({image_path})\n"
                                 )
+                                local_image_count += 1
                                 continue
                     except Exception as e:
                         print(
@@ -614,7 +626,8 @@ class PdfConverter(DocumentConverter):
                     with open(image_path, "wb") as img_file:
                         img_file.write(image_data)
 
-                    content.append(f"![Image {image_count}]({image_path})\n")
+                    content.append(f"![Image {local_image_count}]({image_path})\n")
+                    local_image_count += 1
 
             # Handle text
             if hasattr(lt_obj, "get_text"):
@@ -625,7 +638,8 @@ class PdfConverter(DocumentConverter):
             # Recursively process nested layouts
             elif hasattr(lt_obj, "_objs"):
                 content.extend(
-                    self._process_layout(lt_obj._objs, image_output_dir, image_count)
+                    self._process_layout(
+                        lt_obj._objs, image_output_dir, image_count)
                 )
 
         return content
@@ -635,7 +649,7 @@ class DocxConverter(HtmlConverter):
     """
     Converts DOCX files to Markdown. Style information (e.g.m headings) and tables are preserved where possible.
     """
-    
+
     def __init__(self):
         self._image_counter = 0
         super().__init__()
@@ -644,18 +658,19 @@ class DocxConverter(HtmlConverter):
         """
         保存图片并返回相对路径，使用递增的计数器来命名文件
         """
-        # 获取图片内容和格式        
-        image_format = image.content_type.split('/')[-1] if image.content_type else 'png'
-        
+        # 获取图片内容和格式
+        image_format = image.content_type.split(
+            '/')[-1] if image.content_type else 'png'
+
         # 增加计数器并生成文件名
         self._image_counter += 1
         image_filename = f"image_{self._image_counter}.{image_format}"
-        
+
         # 保存图片
         image_path = os.path.join(output_dir, image_filename)
         with image.open() as image_content, open(image_path, 'wb') as f:
             f.write(image_content.read())
-            
+
         return image_path
 
     def convert(self, local_path, **kwargs) -> Union[None, DocumentConverterResult]:
@@ -671,7 +686,7 @@ class DocxConverter(HtmlConverter):
         else:
             # Create output directory for images if it doesn't exist
             image_output_dir = os.path.join(os.path.dirname(
-                local_path), "_images", os.path.basename(local_path))
+                local_path), "_images", os.path.basename(local_path).replace(" ", "_"))
         os.makedirs(image_output_dir, exist_ok=True)
 
         result = None
@@ -682,7 +697,7 @@ class DocxConverter(HtmlConverter):
                     "src": self._save_image(image, image_output_dir),
                     "alt": image.alt_text if image.alt_text else f"Image {self._image_counter}"
                 }
-                        
+
             # 进行转换
             result = mammoth.convert_to_html(
                 docx_file,
@@ -691,7 +706,7 @@ class DocxConverter(HtmlConverter):
             html_content = result.value
             result = self._convert(html_content)
 
-        return result    
+        return result
 
 
 class XlsxConverter(HtmlConverter):
@@ -710,7 +725,8 @@ class XlsxConverter(HtmlConverter):
         for s in sheets:
             md_content += f"## {s}\n"
             html_content = sheets[s].to_html(index=False)
-            md_content += self._convert(html_content).text_content.strip() + "\n\n"
+            md_content += self._convert(
+                html_content).text_content.strip() + "\n\n"
 
         return DocumentConverterResult(
             title=None,
@@ -745,7 +761,8 @@ class PptxConverter(HtmlConverter):
                     # https://github.com/scanny/python-pptx/pull/512#issuecomment-1713100069
                     alt_text = ""
                     try:
-                        alt_text = shape._element._nvXxPr.cNvPr.attrib.get("descr", "")
+                        alt_text = shape._element._nvXxPr.cNvPr.attrib.get(
+                            "descr", "")
                     except Exception:
                         pass
 
@@ -767,14 +784,17 @@ class PptxConverter(HtmlConverter):
                         html_table += "<tr>"
                         for cell in row.cells:
                             if first_row:
-                                html_table += "<th>" + html.escape(cell.text) + "</th>"
+                                html_table += "<th>" + \
+                                    html.escape(cell.text) + "</th>"
                             else:
-                                html_table += "<td>" + html.escape(cell.text) + "</td>"
+                                html_table += "<td>" + \
+                                    html.escape(cell.text) + "</td>"
                         html_table += "</tr>"
                         first_row = False
                     html_table += "</table></body></html>"
                     md_content += (
-                        "\n" + self._convert(html_table).text_content.strip() + "\n"
+                        "\n" +
+                        self._convert(html_table).text_content.strip() + "\n"
                     )
 
                 # Text areas
@@ -1028,7 +1048,8 @@ class ImageConverter(MediaConverter):
             }
         ]
 
-        response = client.chat.completions.create(model=model, messages=messages)
+        response = client.chat.completions.create(
+            model=model, messages=messages)
         return response.choices[0].message.content
 
 
@@ -1242,9 +1263,11 @@ class MarkItDown:
                 if res is not None:
                     # Normalize the content
                     res.text_content = "\n".join(
-                        [line.rstrip() for line in re.split(r"\r?\n", res.text_content)]
+                        [line.rstrip()
+                         for line in re.split(r"\r?\n", res.text_content)]
                     )
-                    res.text_content = re.sub(r"\n{3,}", "\n\n", res.text_content)
+                    res.text_content = re.sub(
+                        r"\n{3,}", "\n\n", res.text_content)
 
                     # Todo
                     return res

autocoder/version.py

@@ -1 +1 @@
-__version__ = "0.1.201"
+__version__ = "0.1.202"