mfcli 0.2.7__tar.gz → 0.2.9__tar.gz

{mfcli-0.2.7 → mfcli-0.2.9}/MCP_SETUP.md

@@ -115,7 +115,8 @@ If the file doesn't exist, create it with this content:
       "disabled": false,
       "timeout": 60,
       "type": "stdio",
-      "command": "mfcli-mcp"
+      "command": "mfcli-mcp",
+      "autoApprove": ["query_local_rag"]
     }
   }
 }
@@ -133,12 +134,15 @@ If the file exists, add the `mfcli-mcp` entry to the existing `mcpServers` objec
       "disabled": false,
       "timeout": 60,
       "type": "stdio",
-      "command": "mfcli-mcp"
+      "command": "mfcli-mcp",
+      "autoApprove": ["query_local_rag"]
     }
   }
 }
 ```
 
+**Note about `autoApprove`:** The `autoApprove` option allows the `query_local_rag` tool to run automatically without requiring explicit user approval each time. This streamlines the workflow when working with hardware documentation, as the tool is read-only and simply queries your local knowledge base.
+
 ### Step 3: Restart Your Editor
 
 Close and reopen your AI coding assistant for the changes to take effect.
@@ -382,6 +386,32 @@ Enable verbose logging for troubleshooting:
 }
 ```
 
+### Auto-Approval for Tools
+
+The `autoApprove` option allows specific tools to run automatically without requiring user approval each time:
+
+```json
+{
+  "mfcli-mcp": {
+    "autoApprove": ["query_local_rag"]
+  }
+}
+```
+
+**Benefits:**
+- ✅ Streamlined workflow - No manual approval needed for each query
+- ✅ Faster responses - AI assistant can query documentation immediately
+- ✅ Better UX - More natural conversation flow
+- ✅ Safe for read-only tools - `query_local_rag` only reads local data
+
+**Security Note:** The `query_local_rag` tool is safe to auto-approve because:
+- It only performs read operations on your local knowledge base
+- No data is modified or transmitted externally
+- It operates within your local filesystem permissions
+- All data stays on your machine
+
+If you prefer to manually approve each tool use, simply omit the `autoApprove` field from the configuration.
+
 ## Best Practices
 
 ### Document Organization

{mfcli-0.2.7 → mfcli-0.2.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mfcli
-Version: 0.2.7
+Version: 0.2.9
 Summary: AI-powered CLI for analyzing hardware engineering documents
 Author: Multifactor AI
 License: MIT

{mfcli-0.2.7 → mfcli-0.2.9}/mfcli/constants/file_types.py

@@ -13,6 +13,10 @@ class FileTypes(BaseEnum):
     PDF = 5
     SCH = 6
     KICAD_SCH = 7
+    TXT = 8
+    RTF = 9
+    DOCX = 10
+    JPG = 11
 
 
 class FileSubtypes(BaseEnum):
@@ -47,7 +51,9 @@ SchemalessFileSubtypes: set[FileSubtypes] = {
     FileSubtypes.ERRATA,
     FileSubtypes.MCU_DATASHEET,
     FileSubtypes.USER_GUIDE,
-    FileSubtypes.REFERENCE_MANUAL
+    FileSubtypes.REFERENCE_MANUAL,
+    FileSubtypes.GENERAL_DATASHEET,
+    FileSubtypes.UNKNOWN
 }
 
 # File subtype names are for use with LLM, validating subtype response
@@ -67,7 +73,8 @@ OtherFileSubtypeNames = Literal[
     'PADS_PCB_ASCII',
     'KICAD_LEGACY_NET',
     'KICAD_SPICE',
-    'PROTEL_ALTIUM'
+    'PROTEL_ALTIUM',
+    'UNKNOWN'
 ]
 
 PDFMimeTypes = {
@@ -138,6 +145,40 @@ SupportedFileTypes = {
         "subtypes": {
             "SCHEMATIC"
         }
+    },
+    "TXT": {
+        "mime_types": {
+            "text/plain"
+        },
+        "subtypes": {
+            "UNKNOWN"
+        }
+    },
+    "RTF": {
+        "mime_types": {
+            "application/rtf",
+            "text/rtf"
+        },
+        "subtypes": {
+            "UNKNOWN"
+        }
+    },
+    "DOCX": {
+        "mime_types": {
+            "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+        },
+        "subtypes": {
+            "UNKNOWN"
+        }
+    },
+    "JPG": {
+        "mime_types": {
+            "image/jpeg",
+            "image/jpg"
+        },
+        "subtypes": {
+            "UNKNOWN"
+        }
     }
 }
 

mfcli-0.2.9/mfcli/pipeline/extractor.py

@@ -0,0 +1,85 @@
+from mfcli.constants.file_types import FileTypes
+from mfcli.models.file import File
+from mfcli.pipeline.extractors.pdf import extract_text_from_pdf
+from mfcli.utils.files import is_text_mime_type
+from mfcli.utils.logger import get_logger
+
+logger = get_logger(__name__)
+
+
+def extract_text_from_docx(file_bytes: bytes) -> str:
+    """Extract text from DOCX file using python-docx."""
+    try:
+        from io import BytesIO
+        import docx
+        
+        doc = docx.Document(BytesIO(file_bytes))
+        text_parts = []
+        for paragraph in doc.paragraphs:
+            text_parts.append(paragraph.text)
+        return '\n'.join(text_parts)
+    except ImportError:
+        logger.warning("python-docx not installed, treating DOCX as binary")
+        return file_bytes.decode(errors='ignore')
+    except Exception as e:
+        logger.error(f"Error extracting text from DOCX: {e}")
+        return file_bytes.decode(errors='ignore')
+
+
+def extract_text_from_rtf(file_bytes: bytes) -> str:
+    """Extract text from RTF file using striprtf."""
+    try:
+        from striprtf.striprtf import rtf_to_text
+        rtf_string = file_bytes.decode(errors='ignore')
+        return rtf_to_text(rtf_string)
+    except ImportError:
+        logger.warning("striprtf not installed, treating RTF as plain text")
+        return file_bytes.decode(errors='ignore')
+    except Exception as e:
+        logger.error(f"Error extracting text from RTF: {e}")
+        return file_bytes.decode(errors='ignore')
+
+
+def extract_document_text(file: File, file_bytes: bytes) -> str:
+    """
+    Extract text from a file based on its MIME type.
+    
+    For PDFs, uses PyMuPDF (fitz) for fast, reliable text extraction.
+    For text files, decodes the bytes directly.
+    For DOCX files, uses python-docx to extract text.
+    For RTF files, uses striprtf to extract text.
+    For image files (JPG, PNG), returns empty string (OCR not implemented).
+    KiCad schematics and other schematic files are treated as text files.
+    
+    Args:
+        file: File object containing metadata
+        file_bytes: Raw file content as bytes
+        
+    Returns:
+        Extracted text content
+        
+    Raises:
+        ValueError: If the file type is not supported
+    """
+    if is_text_mime_type(file.mime_type):
+        return file_bytes.decode(errors='ignore')
+    elif file.type == FileTypes.PDF:
+        return extract_text_from_pdf(file_bytes)
+    elif file.type == FileTypes.DOCX:
+        return extract_text_from_docx(file_bytes)
+    elif file.type == FileTypes.RTF:
+        return extract_text_from_rtf(file_bytes)
+    elif file.type == FileTypes.JPG:
+        # Image files don't have extractable text without OCR
+        # Return empty string for now - file will still be vectorized if it's a PDF-embedded image
+        logger.info(f"Image file detected ({file.name}), no text extraction (OCR not implemented)")
+        return ""
+    elif file.type == FileTypes.CSV:
+        # CSV files (including Excel CSV with application/vnd.ms-excel MIME type)
+        # are text-based and should be decoded directly
+        return file_bytes.decode(errors='ignore')
+    elif file.type in (FileTypes.KICAD_SCH, FileTypes.SCH, FileTypes.TXT):
+        # KiCad schematics and plain text files are S-expression text files but may have 
+        # application/octet-stream MIME type
+        return file_bytes.decode(errors='ignore')
+    raise ValueError(f"Unsupported MIME type: {file.mime_type}")

{mfcli-0.2.7 → mfcli-0.2.9}/mfcli/pipeline/sub_classifier.py

@@ -85,7 +85,9 @@ class FileSubtypeAnalyzer:
         subtype = await self._get_subtype_from_gemini(prompt, instructions, gemini_file, file_class)
         logger.debug(f"Subtype discovered: {subtype}")
         if subtype == FILE_SUBTYPE_UNKNOWN:
-            raise RuntimeError(f"Could not determine the file subtype for file: {file.name}")
+            logger.warning(f"Could not determine the file subtype for file: {file.name}, using UNKNOWN")
+            file.sub_type = FileSubtypes.UNKNOWN.value
+            return
         if not subtype in relevant_subtype_descriptions:
             raise RuntimeError(f"LLM responded with invalid subtype: {subtype}")
         file.sub_type = FileSubtypes.get(subtype)
@@ -109,6 +111,13 @@ class FileSubtypeAnalyzer:
             file.sub_type = FileSubtypes.SCHEMATIC.value
             return
         
+        # Generic document file types (TXT, RTF, DOCX, JPG) default to UNKNOWN
+        # These files will be vectorized but not analyzed for specific subtypes
+        if file.type in (FileTypes.TXT, FileTypes.RTF, FileTypes.DOCX, FileTypes.JPG):
+            file.sub_type = FileSubtypes.UNKNOWN.value
+            logger.info(f"File type {FileTypes(file.type).name} auto-assigned UNKNOWN subtype")
+            return
+        
         # Handle text MIME types for content-based detection
         if text and is_text_mime_type(file.mime_type):
             if file.type == FileTypes.NET:

{mfcli-0.2.7 → mfcli-0.2.9}/mfcli/utils/data_cleaner.py

@@ -19,7 +19,8 @@ logger = get_logger(__name__)
 warning_message = dedent(
     """
     
-    WARNING: This will permanently delete all mfcli data for the current project, including datasheets, cheat sheets, and project data. 
+    WARNING: This will permanently delete all mfcli data for the current project, including datasheets, cheat sheets, and project data.
+    The design folder and its contents will be preserved.
     Should we proceed? (Y/n):
     
     """
@@ -105,9 +106,8 @@ class DataCleaner:
                 logger.debug(f"Removing file docket: {app_dirs.file_docket_path}")
                 self._remove_file(app_dirs.file_docket_path)
             
-            # Remove directories
+            # Remove directories (excluding design folder)
             for dir_path in [
-                config_dir,
                 app_dirs.data_sheets_dir,
                 app_dirs.fw_tasks_dir,
                 app_dirs.generated_files_dir,
@@ -118,6 +118,9 @@ class DataCleaner:
                 logger.debug(f"Removing directory: {dir_path}")
                 self._remove_dir(dir_path)
             
+            # Note: design_dir is intentionally NOT removed to preserve user input files
+            logger.info(f"Design folder preserved at: {app_dirs.design_dir}")
+            
             # Delete project (CASCADE will handle related records like pipeline_runs)
             logger.debug(f"Deleting project from database: {project.name}")
             self._db.delete(project)

{mfcli-0.2.7 → mfcli-0.2.9}/mfcli/utils/mcp_configurator.py

@@ -73,7 +73,8 @@ def get_mfcli_mcp_config() -> dict:
             "disabled": False,
             "timeout": 60,
             "type": "stdio",
-            "command": "mfcli-mcp"
+            "command": "mfcli-mcp",
+            "autoApprove": ["query_local_rag"]
         }
     }
 
@@ -382,7 +383,8 @@ editor's MCP configuration file.
       "disabled": false,
       "timeout": 60,
       "type": "stdio",
-      "command": "mfcli-mcp"
+      "command": "mfcli-mcp",
+      "autoApprove": ["query_local_rag"]
     }
   }
 }

{mfcli-0.2.7 → mfcli-0.2.9}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "mfcli"
-version = "0.2.7"
+version = "0.2.9"
 description = "AI-powered CLI for analyzing hardware engineering documents"
 readme = "README.md"
 license = {text = "MIT"}

mfcli-0.2.7/mfcli/pipeline/extractor.py

@@ -1,37 +0,0 @@
-from mfcli.constants.file_types import FileTypes
-from mfcli.models.file import File
-from mfcli.pipeline.extractors.pdf import extract_text_from_pdf
-from mfcli.utils.files import is_text_mime_type
-
-
-def extract_document_text(file: File, file_bytes: bytes) -> str:
-    """
-    Extract text from a file based on its MIME type.
-    
-    For PDFs, uses PyMuPDF (fitz) for fast, reliable text extraction.
-    For text files, decodes the bytes directly.
-    KiCad schematics and other schematic files are treated as text files.
-    
-    Args:
-        file: File object containing metadata
-        file_bytes: Raw file content as bytes
-        
-    Returns:
-        Extracted text content
-        
-    Raises:
-        ValueError: If the file type is not supported
-    """
-    if is_text_mime_type(file.mime_type):
-        return file_bytes.decode(errors='ignore')
-    elif file.type == FileTypes.PDF:
-        return extract_text_from_pdf(file_bytes)
-    elif file.type == FileTypes.CSV:
-        # CSV files (including Excel CSV with application/vnd.ms-excel MIME type)
-        # are text-based and should be decoded directly
-        return file_bytes.decode(errors='ignore')
-    elif file.type in (FileTypes.KICAD_SCH, FileTypes.SCH):
-        # KiCad schematics are S-expression text files but may have 
-        # application/octet-stream MIME type
-        return file_bytes.decode(errors='ignore')
-    raise ValueError(f"Unsupported MIME type: {file.mime_type}")