abstract-utilities 0.2.2.480__py3-none-any.whl → 0.2.2.688__py3-none-any.whl

abstract_utilities/file_utils/src/file_filters/filter_utils.py

@@ -0,0 +1,78 @@
+from ...imports import *
+import re
+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:
+                nu_values = typ()
+            
+            if typ is set:
+                nu_values = nu_values | typ(value)
+            if typ is list:
+                nu_values += typ(value)
+    return nu_values
+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
+    }
+
+    # 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)

abstract_utilities/file_utils/src/file_filters/predicate_utils.py

@@ -0,0 +1,44 @@
+from .ensure_utils import *
+
+def get_globs(items,recursive: bool = True,allowed=None,cfg=None,**kwargs):
+    glob_paths = []
+    allowed = get_allowed_predicate(allowed=allowed,cfg=cfg,**kwargs)
+    items = [item for item in make_list(items) if item]
+    for item in items:
+        pattern = os.path.join(item, "**/*")  # include all files recursively\n
+        nuItems = glob.glob(pattern, recursive=recursive)
+        if allowed:
+            nuItems = [nuItem for nuItem in nuItems if nuItem and allowed(nuItem)]
+        glob_paths += nuItems
+    return glob_paths
+def get_allowed_files(items,allowed=True,cfg=None,**kwargs):
+    allowed = get_allowed_predicate(allowed=allowed,cfg=cfg,**kwargs)
+    return [item for item in items if item and os.path.isfile(item) and allowed(item)]
+def get_allowed_dirs(items,allowed=False,cfg=None,**kwargs):
+    allowed = get_allowed_predicate(allowed=allowed,cfg=cfg,**kwargs)
+    return [item for item in items if item and os.path.isdir(item) and allowed(item)]
+
+def get_filtered_files(items,allowed=None,files = [],cfg=None,**kwargs):
+    allowed = get_allowed_predicate(allowed=allowed,cfg=cfg,**kwargs)
+    glob_paths = get_globs(items,allowed=allowed,cfg=cfg,**kwargs)
+    return [glob_path for glob_path in glob_paths if glob_path and os.path.isfile(glob_path) and glob_path not in files and allowed(glob_path)]
+def get_filtered_dirs(items,allowed=None,dirs = [],cfg=None,**kwargs):
+    allowed = get_allowed_predicate(allowed=allowed,cfg=cfg,**kwargs)
+    glob_paths = get_globs(items,allowed=allowed,cfg=cfg,**kwargs)
+    return [glob_path for glob_path in glob_paths if glob_path and os.path.isdir(glob_path) and glob_path not in dirs and allowed(glob_path)]
+
+def get_all_allowed_files(items,allowed=None,cfg=None,**kwargs):
+    dirs = get_all_allowed_dirs(items,allowed=allowed,cfg=cfg,**kwargs)
+    files = get_allowed_files(items,allowed=allowed,cfg=cfg,**kwargs)
+    nu_files = []
+    for directory in dirs:
+        files += get_filtered_files(directory,allowed=allowed,files=files,cfg=cfg,**kwargs)
+    return files
+def get_all_allowed_dirs(items,allowed=None,cfg=None,**kwargs):
+    allowed = get_allowed_predicate(allowed=allowed,cfg=cfg,**kwargs)
+    dirs = get_allowed_dirs(items,allowed=allowed,cfg=cfg,**kwargs)
+    nu_dirs=[]
+    for directory in dirs:
+        nu_dirs += get_filtered_dirs(directory,allowed=allowed,dirs=nu_dirs,cfg=cfg,**kwargs)
+    return nu_dirs
+

abstract_utilities/file_utils/src/file_filters.py

@@ -0,0 +1,177 @@
+from .filter_params import *
+from ..imports import *
+##from abstract_utilities import make_list,get_media_exts, is_media_type
+def get_allowed_predicate(allowed=None):
+    if allowed != False:
+        if allowed == True:
+            allowed = None
+        allowed = allowed or make_allowed_predicate()
+    else:
+        def allowed(*args):
+            return True
+        allowed = allowed
+    return allowed
+def get_globs(items,recursive: bool = True,allowed=None):
+    glob_paths = []
+    items = [item for item in make_list(items) if item]
+    for item in items:
+        pattern = os.path.join(item, "**/*")  # include all files recursively\n
+        nuItems = glob.glob(pattern, recursive=recursive)
+        if allowed:
+            nuItems = [nuItem for nuItem in nuItems if nuItem and allowed(nuItem)]
+        glob_paths += nuItems
+    return glob_paths
+def get_allowed_files(items,allowed=True):
+    allowed = get_allowed_predicate(allowed=allowed)
+    return [item for item in items if item and os.path.isfile(item) and allowed(item)]
+def get_allowed_dirs(items,allowed=False):
+    allowed = get_allowed_predicate(allowed=allowed)
+    return [item for item in items if item and os.path.isdir(item) and allowed(item)]
+
+def get_filtered_files(items,allowed=None,files = []):
+    allowed = get_allowed_predicate(allowed=allowed)
+    glob_paths = get_globs(items)
+    return [glob_path for glob_path in glob_paths if glob_path and os.path.isfile(glob_path) and glob_path not in files and allowed(glob_path)]
+def get_filtered_dirs(items,allowed=None,dirs = []):
+    allowed = get_allowed_predicate(allowed=allowed)
+    glob_paths = get_globs(items)
+    return [glob_path for glob_path in glob_paths if glob_path and os.path.isdir(glob_path) and glob_path not in dirs and allowed(glob_path)]
+
+def get_all_allowed_files(items,allowed=None):
+    dirs = get_all_allowed_dirs(items)
+    files = get_allowed_files(items)
+    nu_files = []
+    for directory in dirs:
+        files += get_filtered_files(directory,allowed=allowed,files=files)
+    return files
+def get_all_allowed_dirs(items,allowed=None):
+    allowed = get_allowed_predicate(allowed=allowed)
+    dirs = get_allowed_dirs(items)
+    nu_dirs=[]
+    for directory in dirs:
+        nu_dirs += get_filtered_dirs(directory,allowed=allowed,dirs=nu_dirs)
+    return nu_dirs
+
+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.exclude_exts and ext in cfg.exclude_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 _fast_walk(
+    root: Path,
+    exts: Iterable[str],
+    skip_dirs: Iterable[str] = (),
+    skip_patterns: Iterable[str] = (),
+) -> List[Path]:
+    exts = tuple(exts)
+    skip_dirs = set(sd.lower() for sd in skip_dirs or ())
+    skip_patterns = tuple(sp.lower() for sp in (skip_patterns or ()))
+
+    out = []
+    for p in root.rglob("*"):
+        # skip directories by name hit
+        if p.is_dir():
+            name = p.name.lower()
+            if name in skip_dirs:
+                # rglob doesn't let us prune mid-iteration cleanly; we just won't collect under it
+                continue
+            # nothing to collect for dirs
+            continue
+
+        # file filters
+        name = p.name.lower()
+        if any(fnmatch.fnmatch(name, pat) for pat in skip_patterns):
+            continue
+        if p.suffix.lower() in exts:
+            out.append(p)
+
+    # de-dup and normalize
+    return sorted({pp.resolve() for pp in out})
+
+
+def enumerate_source_files(
+    src_root: Path,
+    cfg: Optional["ScanConfig"] = None,
+    *,
+    exts: Optional[Iterable[str]] = None,
+    fast_skip_dirs: Optional[Iterable[str]] = None,
+    fast_skip_patterns: Optional[Iterable[str]] = None,
+) -> List[Path]:
+    """
+    Unified enumerator:
+      - If `cfg` is provided: use collect_filepaths(...) with full rules.
+      - Else: fast walk using rglob over `exts` (defaults to EXTS) with optional light excludes.
+    """
+    src_root = Path(src_root)
+
+    if cfg is not None:
+        files = collect_filepaths([str(src_root)], cfg=cfg)
+        return sorted({Path(f).resolve() for f in files})
+
+    # Fast mode
+    return _fast_walk(
+        src_root,
+        exts or EXTS,
+        skip_dirs=fast_skip_dirs or (),
+        skip_patterns=fast_skip_patterns or (),
+    )