abstract-utilities 0.2.2.513__py3-none-any.whl → 0.2.2.627__py3-none-any.whl

abstract_utilities/file_utils/src/file_utils.py

@@ -52,38 +52,13 @@ def get_all_allowed_dirs(items,allowed=None):
     for directory in dirs:
         nu_dirs += get_filtered_dirs(directory,allowed=allowed,dirs=nu_dirs)
     return nu_dirs
-def get_files_and_dirs(
-    directory: str,
-    cfg: Optional["ScanConfig"] = None,
-    allowed_exts: Optional[Set[str]] = False,
-    unallowed_exts: Optional[Set[str]] = False,
-    allowed_types: Optional[Set[str]] = False,
-    exclude_types: Optional[Set[str]] = False,
-    allowed_dirs: Optional[List[str]] = False,
-    exclude_dirs: Optional[List[str]] = False,
-    allowed_patterns: Optional[List[str]] = False,
-    exclude_patterns: Optional[List[str]] = False,
-    add = False,
-    recursive: bool = True,
-    include_files: bool = True,
-    **kwargs
-    ):
-    cfg = cfg or define_defaults(
-            allowed_exts = allowed_exts,
-            unallowed_exts = unallowed_exts,
-            allowed_types = allowed_types,
-            exclude_types = exclude_types,
-            allowed_dirs = allowed_dirs,
-            exclude_dirs = exclude_dirs,
-            allowed_patterns = allowed_patterns,
-            exclude_patterns = exclude_patterns,
-        add=add
-        )
-    allowed = make_allowed_predicate(cfg)
+def get_files_and_dirs(*args,**kwargs)-> List[str]:
+    directories,cfg,allowed,include_files,recursive = get_file_filters(*args,**kwargs)
     items=[]
     files =[]
     if recursive:
-        items = get_globs(directory,recursive=recursive,allowed=allowed)
+        for directory in directories:
+            items += get_globs(directories,recursive=recursive,allowed=allowed)
     else:
         directories = make_list(directory)
         for directory in directories:
@@ -92,68 +67,7 @@ def get_files_and_dirs(
     if include_files:
         files = get_allowed_files(items,allowed=allowed)
     return dirs,files
-def make_allowed_predicate(cfg: ScanConfig) -> Callable[[str], bool]:
-    """
-    Build a predicate that returns True if a given path is considered allowed
-    under the given ScanConfig. Applies allowed_* and exclude_* logic symmetrically.
-    """
-    def allowed(path: str=None,p=None) -> bool:
-        p = p or Path(path)
-        name = p.name.lower()
-        path_str = str(p).lower()
-
-        # --------------------
-        # A) directory filters
-        # --------------------
-        if cfg.exclude_dirs:
-            for dpat in cfg.exclude_dirs:
-                dpat_l = dpat.lower()
-                if dpat_l in path_str or fnmatch.fnmatch(name, dpat_l):
-                    if p.is_dir() or dpat_l in path_str:
-                        return False
-
-        if cfg.allowed_dirs and cfg.allowed_dirs != ["*"]:
-            # must be in at least one allowed dir
-            if not any(
-                fnmatch.fnmatch(path_str, f"*{dpat.lower()}*") for dpat in cfg.allowed_dirs
-            ):
-                return False
 
-        # --------------------
-        # B) pattern filters
-        # --------------------
-        if cfg.allowed_patterns and cfg.allowed_patterns != ["*"]:
-            if not any(fnmatch.fnmatch(name, pat.lower()) for pat in cfg.allowed_patterns):
-                return False
-
-        if cfg.exclude_patterns:
-            for pat in cfg.exclude_patterns:
-                if fnmatch.fnmatch(name, pat.lower()):
-                    return False
-
-        # --------------------
-        # C) extension filters
-        # --------------------
-        if p.is_file():
-            ext = p.suffix.lower()
-            if cfg.allowed_exts and ext not in cfg.allowed_exts:
-                return False
-            if cfg.unallowed_exts and ext in cfg.unallowed_exts:
-                return False
-
-        # --------------------
-        # D) type filters (optional)
-        # --------------------
-        if cfg.allowed_types and cfg.allowed_types != {"*"}:
-            if not any(t in path_str for t in cfg.allowed_types):
-                return False
-        if cfg.exclude_types and cfg.exclude_types != {"*"}:
-            if any(t in path_str for t in cfg.exclude_types):
-                return False
-
-        return True
-
-    return allowed
 def correct_kwargs(**kwargs):
     for key,values in kwargs.items():
         if key.startswith('excluded'):
@@ -163,47 +77,10 @@ def correct_kwargs(**kwargs):
             kwargs[correct_key]=combine_params(correct_vals,values)
             del kwargs[key]
     
-def collect_filepaths(
-    directory: List[str],
-    cfg: ScanConfig=None,
-    allowed_exts: Optional[Set[str]] = False,
-    unallowed_exts: Optional[Set[str]] = False,
-    allowed_types: Optional[Set[str]] = False,
-    exclude_types: Optional[Set[str]] = False,
-    allowed_dirs: Optional[List[str]] = False,
-    exclude_dirs: Optional[List[str]] = False,
-    allowed_patterns: Optional[List[str]] = False,
-    exclude_patterns: Optional[List[str]] = False,
-    add=False,
-    allowed: Optional[Callable[[str], bool]] = None,
-    **kwargs
-    ) -> List[str]:
-    kwargs = correct_kwargs(allowed_exts = allowed_exts,
-        unallowed_exts = unallowed_exts,
-        allowed_types = allowed_types,
-        exclude_types = exclude_types,
-        allowed_dirs = allowed_dirs,
-        exclude_dirs = exclude_dirs,
-        allowed_patterns = allowed_patterns,
-        exclude_patterns = exclude_patterns,
-        **kwargs)
-    cfg = cfg or define_defaults(
-            allowed_exts = allowed_exts,
-            unallowed_exts = unallowed_exts,
-            allowed_types = allowed_types,
-            exclude_types = exclude_types,
-            allowed_dirs = allowed_dirs,
-            exclude_dirs = exclude_dirs,
-            allowed_patterns = allowed_patterns,
-            exclude_patterns = exclude_patterns,
-                                add = add
-                                )
-    allowed = allowed or make_allowed_predicate(cfg)
-    directories = make_list(directory)
-    roots = [r for r in directories if r]
-
+def collect_filepaths(*args,**kwargs)-> List[str]:
+    directories,cfg,allowed,include_files,recursive = get_file_filters(*args,**kwargs)
     # your existing helpers (get_dirs, get_globs, etc.) stay the same
-    original_dirs = get_allowed_dirs(roots, allowed=allowed)
+    original_dirs = get_allowed_dirs(directories, allowed=allowed)
     original_globs = get_globs(original_dirs)
     files = get_allowed_files(original_globs, allowed=allowed)
 

abstract_utilities/file_utils/src/filter_params.py

@@ -5,68 +5,78 @@ def combine_params(*values,typ=None):
     nu_values = None
     for value in values:
         if value is not None:
+            typ = typ or type(value)
             if nu_values is None:
-                typ = typ or type(value)
                 nu_values = typ()
-            else:
-                if val_type is set:
-                    nu_values = nu_values | typ(value)
-                if val_type is list:
-                    nu_values += typ(value)
+            
+            if typ is set:
+                nu_values = nu_values | typ(value)
+            if typ is list:
+                nu_values += typ(value)
     return nu_values
-# -------------------------
-# Default sets
-# -------------------------
-
-DEFAULT_ALLOWED_EXTS: Set[str] = {
-    ".py", ".pyw", ".js", ".jsx", ".ts", ".tsx", ".mjs",
-    ".html", ".htm", ".xml", ".css", ".scss", ".sass", ".less",
-    ".json", ".yaml", ".yml", ".toml", ".ini", ".cfg",
-    ".md", ".markdown", ".rst", ".sh", ".bash", ".env", ".txt"
-}
-
-DEFAULT_EXCLUDE_TYPES: Set[str] = {
-    "image", "video", "audio", "presentation",
-    "spreadsheet", "archive", "executable"
-}
-
-_unallowed = set(get_media_exts(DEFAULT_EXCLUDE_TYPES)) | {
-    ".bak", ".shp", ".cpg", ".dbf", ".shx", ".geojson",
-    ".pyc", ".prj", ".sbn", ".sbx"
-}
-DEFAULT_UNALLOWED_EXTS = {e for e in _unallowed if e not in DEFAULT_ALLOWED_EXTS}
-
-DEFAULT_EXCLUDE_DIRS: List[str] = [
-    "node_modules", "__pycache__", "backups", "backup",
-    "trash", "deprecated", "old", "__init__"
-]
-
-DEFAULT_EXCLUDE_PATTERNS: List[str] = [
-    "__init__*", "*.tmp", "*.log", "*.lock", "*.zip", "*~"
-]
-
-DEFAULT_ALLOWED_PATTERNS: List[str] = ["*"]
-DEFAULT_ALLOWED_DIRS: List[str] = ["*"]
-DEFAULT_ALLOWED_TYPES: List[str] = ["*"]
-
-REMOTE_RE = re.compile(r"^(?P<host>[^:\s]+@[^:\s]+):(?P<path>/.*)$")
-AllowedPredicate = Optional[Callable[[str], bool]]
-
-# -------------------------
-# Config dataclass
-# -------------------------
+def get_safe_kwargs(canonical_map, **kwargs):
+    # Lowercase all keys for safety
+    canonical_map = canonical_map or CANONICAL_MAP
+    norm_kwargs = {k.lower(): v for k, v in kwargs.items() if v is not None}
+
+    # Inverse lookup: alias → canonical key
+    alias_lookup = {
+        alias: canon
+        for canon, aliases in canonical_map.items()
+        if aliases
+        for alias in aliases
+    }
 
-@dataclass
-class ScanConfig:
-    allowed_exts: Set[str]
-    unallowed_exts: Set[str]
-    allowed_types: Set[str]
-    exclude_types: Set[str]
-    allowed_dirs: List[str] = field(default_factory=list)
-    exclude_dirs: List[str] = field(default_factory=list)
-    allowed_patterns: List[str] = field(default_factory=list)
-    exclude_patterns: List[str] = field(default_factory=list)
+    # Preserve correctly named keys
+    safe_kwargs = {k: v for k, v in norm_kwargs.items() if k in canonical_map}
 
+    for k, v in norm_kwargs.items():
+        if k in alias_lookup:
+            canonical_key = alias_lookup[k]
+            prev = safe_kwargs.get(canonical_key)
+            if prev is None:
+                safe_kwargs[canonical_key] = v
+            else:
+                # merge intelligently if both exist
+                if isinstance(prev, (set, list)) and isinstance(v, (set, list)):
+                    safe_kwargs[canonical_key] = list(set(prev) | set(v))
+                else:
+                    safe_kwargs[canonical_key] = v  # overwrite for non-iterables
+
+    # fill defaults if missing
+    for canon in canonical_map:
+        safe_kwargs.setdefault(canon, None)
+
+    return safe_kwargs
+
+def create_canonical_map(*args,canonical_map=None):
+    keys = [arg for arg in args if arg]
+    if not keys:
+        return CANONICAL_MAP
+    canonical_map = canonical_map or CANONICAL_MAP
+
+    return {key:canonical_map.get(key) for key in keys}
+def get_safe_canonical_kwargs(*args,canonical_map=None,**kwargs):
+    canonical_map = canonical_map or create_canonical_map(*args)
+    
+    return get_safe_kwargs(canonical_map=canonical_map,**kwargs)    
+def get_dir_filter_kwargs(**kwargs):
+    canonical_map = create_canonical_map("directories")
+    return get_safe_kwargs(canonical_map=canonical_map,**kwargs)
+def get_file_filter_kwargs(**kwargs):
+    """
+    Normalize arbitrary keyword arguments for file scanning configuration.
+    
+    Examples:
+      - 'excluded_ext' or 'unallowed_exts' → 'exclude_exts'
+      - 'include_dirs' or 'allow_dir' → 'allowed_dirs'
+      - 'excludePattern' or 'excluded_patterns' → 'exclude_patterns'
+      - 'allowed_type' or 'include_types' → 'allowed_types'
+    """
+    # Canonical keys and aliases
+    canonical_keys =["allowed_exts","exclude_exts","allowed_types","exclude_types","allowed_dirs","exclude_dirs","allowed_patterns","exclude_patterns"]
+   
+    return get_safe_canonical_kwargs(*canonical_keys,**kwargs)
 # -------------------------
 # Utility functions
 # -------------------------
@@ -110,10 +120,12 @@ def ensure_patterns(patterns):
 
 def _get_default_modular(value, default, add=False, typ=set):
     """Merge user and default values intelligently."""
-    if value in [True, None]:
+    if value == None:
+        value = add
+    if value in [True]:
         return default
     if value is False:
-        return typ()
+        return value
     if add:
         return combine_params(value,default,typ=None)
 
@@ -122,33 +134,63 @@ def _get_default_modular(value, default, add=False, typ=set):
 # -------------------------
 # Default derivation logic
 # -------------------------
-
-def derive_file_defaults(
-    allowed_exts=False, unallowed_exts=False,
-    allowed_types=False, exclude_types=False,
-    allowed_dirs=False, exclude_dirs=False,
-    allowed_patterns=False, exclude_patterns=False,
-    add=False
-):
-    allowed_exts = _get_default_modular(ensure_exts(allowed_exts), DEFAULT_ALLOWED_EXTS, add, set)
-    unallowed_exts = _get_default_modular(ensure_exts(unallowed_exts), DEFAULT_UNALLOWED_EXTS, add, set)
-    allowed_types = _get_default_modular(_normalize_listlike(allowed_types, set), DEFAULT_ALLOWED_TYPES, add, set)
-    exclude_types = _get_default_modular(_normalize_listlike(exclude_types, set), DEFAULT_EXCLUDE_TYPES, add, set)
-    allowed_dirs = _get_default_modular(_normalize_listlike(allowed_dirs, list), DEFAULT_ALLOWED_DIRS, add, list)
-    exclude_dirs = _get_default_modular(_normalize_listlike(exclude_dirs, list), DEFAULT_EXCLUDE_DIRS, add, list)
-    allowed_patterns = _get_default_modular(ensure_patterns(allowed_patterns), DEFAULT_ALLOWED_PATTERNS, add, list)
-    exclude_patterns = _get_default_modular(ensure_patterns(exclude_patterns), DEFAULT_EXCLUDE_PATTERNS, add, list)
-
-    return {
-        "allowed_exts": allowed_exts,
-        "unallowed_exts": unallowed_exts,
-        "allowed_types": allowed_types,
-        "exclude_types": exclude_types,
-        "allowed_dirs": allowed_dirs,
-        "exclude_dirs": exclude_dirs,
-        "allowed_patterns": allowed_patterns,
-        "exclude_patterns": exclude_patterns,
-    }
+def _get_default_modular(value, default, add=None, typ=set):
+    """Merge user and default values intelligently."""
+    add = add or False
+    if value == None:
+        value = add
+    if value in [True]:
+        return default
+    if value is False:
+        return value
+    if add:
+        return combine_params(value,default,typ=None)
+    return typ(value)
+def derive_all_defaults(**kwargs):
+    kwargs = get_safe_canonical_kwargs(**kwargs)
+    add = kwargs.get("add",False)
+    nu_defaults = {}
+    for key,values in DEFAULT_CANONICAL_MAP.items():
+        default = values.get("default")
+        typ = values.get("type")
+        key_value = kwargs.get(key)
+        if key in DEFAULT_ALLOWED_EXCLUDE_MAP:
+
+            if key.endswith('exts'):
+                input_value = ensure_exts(key_value)
+            if key.endswith('patterns'):
+                input_value = ensure_patterns(key_value)
+            else:
+                input_value = _normalize_listlike(key_value, typ)
+            nu_defaults[key] = _get_default_modular(input_value, default, add, typ)
+        else:
+            value = default if key_value is None else key_value
+            if typ == list:
+                value = make_list(value)
+            elif typ == bool:
+                value = bool(value)
+            nu_defaults[key] = value
+   
+    return nu_defaults
+# -------------------------
+# Default derivation logic
+# -------------------------
+def derive_file_defaults(**kwargs):
+    kwargs = derive_all_defaults(**kwargs)
+    add = kwargs.get("add",False)
+    nu_defaults = {}
+    for key,values in DEFAULT_ALLOWED_EXCLUDE_MAP.items():
+        default = values.get("default")
+        typ = values.get("type")
+        key_value = kwargs.get(key)
+        if key.endswith('exts'):
+            input_value = ensure_exts(key_value)
+        if key.endswith('patterns'):
+            input_value = ensure_patterns(key_value)
+        else:
+            input_value = _normalize_listlike(key_value, typ)
+        nu_defaults[key] = _get_default_modular(input_value, default, add, typ)
+    return nu_defaults
 
 def define_defaults(**kwargs):
     defaults = derive_file_defaults(**kwargs)