abstract-utilities 0.2.2.476__py3-none-any.whl → 0.2.2.540__py3-none-any.whl

abstract_utilities/file_utils/src/filter_params.py

@@ -0,0 +1,197 @@
+
+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)
+# -------------------------
+# Utility functions
+# -------------------------
+
+def _normalize_listlike(value, typ=list, sep=','):
+    """Normalize comma-separated or iterable values into the desired type."""
+    if value in [True, None, False]:
+        return value
+    if isinstance(value, str):
+        value = [v.strip() for v in value.split(sep) if v.strip()]
+    return typ(value)
+
+def ensure_exts(exts):
+    if exts in [True, None, False]:
+        return exts
+    out = []
+    for ext in _normalize_listlike(exts, list):
+        if not ext.startswith('.'):
+            ext = f".{ext}"
+        out.append(ext)
+    return set(out)
+
+def ensure_patterns(patterns):
+    """Normalize pattern list and ensure they are valid globs."""
+    if patterns in [True, None, False]:
+        return patterns
+    patterns = _normalize_listlike(patterns, list)
+    out = []
+    for pattern in patterns:
+        if not pattern:
+            continue
+        if '*' not in pattern and '?' not in pattern:
+            # Implicitly make it a prefix match
+            if pattern.startswith('.') or pattern.startswith('~'):
+                pattern = f"*{pattern}"
+            else:
+                pattern = f"{pattern}*"
+        out.append(pattern)
+    return out
+
+
+def _get_default_modular(value, default, add=False, typ=set):
+    """Merge user and default values intelligently."""
+    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)
+
+# -------------------------
+# Default derivation logic
+# -------------------------
+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)
+    return ScanConfig(**defaults)

abstract_utilities/file_utils/src/find_collect.py

@@ -0,0 +1,266 @@
+from ..imports import *
+from .file_filters import *
+from .filter_params import *
+from pathlib import Path
+from typing import Optional, List, Set
+
+def get_proper_type_str(string):
+    if not string:
+        return None
+    string_lower = string.lower()
+    items = {
+        "d":["dir","dirs","directory","directories","d","dirname"],
+        "f":["file","filepath","file_path","files","filepaths","file_paths","f"]
+     }
+    for key,values in items.items():
+        if string_lower in values:
+            return key
+    init = string_lower[0] if len(string_lower)>0 else None
+    if init in items:
+        return init
+def check_path_type(
+    path: str,
+    user: Optional[str] = None,
+    host: Optional[str] = None,
+    user_as_host: Optional[str] = None,
+    use_shell: bool = False
+) -> Literal["file", "directory", "missing", "unknown"]:
+    """
+    Determine whether a given path is a file, directory, or missing.
+    Works locally or remotely (via SSH).
+
+    Args:
+        path: The path to check.
+        user, host, user_as_host: SSH parameters if remote.
+        use_shell: Force shell test instead of Python os.path.
+    Returns:
+        One of: 'file', 'directory', 'missing', or 'unknown'
+    """
+
+    # --- remote check if user/host is given ---
+    if user_as_host or (user and host):
+        remote_target = user_as_host or f"{user}@{host}"
+        cmd = f"if [ -f '{path}' ]; then echo file; elif [ -d '{path}' ]; then echo directory; else echo missing; fi"
+        try:
+            result = subprocess.check_output(
+                ["ssh", remote_target, cmd],
+                stderr=subprocess.DEVNULL,
+                text=True,
+                timeout=5
+            ).strip()
+            return result if result in ("file", "directory", "missing") else "unknown"
+        except Exception:
+            return "unknown"
+
+    # --- local check ---
+    if not use_shell:
+        if os.path.isfile(path):
+            return "file"
+        elif os.path.isdir(path):
+            return "directory"
+        elif not os.path.exists(path):
+            return "missing"
+        return "unknown"
+    else:
+        # fallback using shell tests (useful for sandboxed contexts)
+        cmd = f"if [ -f '{path}' ]; then echo file; elif [ -d '{path}' ]; then echo directory; else echo missing; fi"
+        try:
+            output = subprocess.check_output(
+                cmd, shell=True, stderr=subprocess.DEVNULL, text=True
+            ).strip()
+            return output if output in ("file", "directory", "missing") else "unknown"
+        except Exception:
+            return "unknown"
+
+
+
+
+
+def get_find_cmd(
+    *args,
+    mindepth: Optional[int] = None,
+    maxdepth: Optional[int] = None,
+    depth: Optional[int] = None,
+    file_type: Optional[str] = None,  # 'f' or 'd'
+    name: Optional[str] = None,
+    size: Optional[str] = None,
+    mtime: Optional[str] = None,
+    perm: Optional[str] = None,
+    user: Optional[str] = None,
+    **kwargs
+) -> str:
+    """
+    Construct a Unix `find` command string that supports multiple directories.
+    Accepts filtering via ScanConfig-compatible kwargs.
+    """
+    # Normalize inputs into canonical form
+    kwargs = get_safe_canonical_kwargs(*args, **kwargs)
+    cfg = define_defaults(**kwargs)
+
+    # Get directory list (may come from args or kwargs)
+    directories = [r for r in make_list(kwargs.get("directories")) if r]
+    if not directories:
+        return []
+
+    # Build base command for all directories
+    dir_expr = " ".join(shlex.quote(d) for d in directories)
+    cmd = [f"find {dir_expr}"]
+
+    # --- depth filters ---
+    if depth is not None:
+        cmd += [f"-mindepth {depth}", f"-maxdepth {depth}"]
+    else:
+        if mindepth is not None:
+            cmd.append(f"-mindepth {mindepth}")
+        if maxdepth is not None:
+            cmd.append(f"-maxdepth {maxdepth}")
+
+    # --- file type ---
+    if file_type in ("f", "d"):
+        cmd.append(f"-type {file_type}")
+
+    # --- basic attributes ---
+    if name:
+        cmd.append(f"-name {shlex.quote(name)}")
+    if size:
+        cmd.append(f"-size {shlex.quote(size)}")
+    if mtime:
+        cmd.append(f"-mtime {shlex.quote(mtime)}")
+    if perm:
+        cmd.append(f"-perm {shlex.quote(perm)}")
+    if user:
+        cmd.append(f"-user {shlex.quote(user)}")
+
+    # --- cfg-based filters ---
+    if cfg:
+        # Allowed extensions
+        if cfg.allowed_exts and cfg.allowed_exts != {"*"}:
+            ext_expr = " -o ".join(
+                [f"-name '*{e}'" for e in cfg.allowed_exts if e]
+            )
+            cmd.append(f"\\( {ext_expr} \\)")
+
+        # Excluded extensions
+        if cfg.exclude_exts:
+            for e in cfg.exclude_exts:
+                cmd.append(f"! -name '*{e}'")
+
+        # Allowed directories
+        if cfg.allowed_dirs and cfg.allowed_dirs != ["*"]:
+            dir_expr = " -o ".join(
+                [f"-path '*{d}*'" for d in cfg.allowed_dirs if d]
+            )
+            cmd.append(f"\\( {dir_expr} \\)")
+
+        # Excluded directories
+        if cfg.exclude_dirs:
+            for d in cfg.exclude_dirs:
+                cmd.append(f"! -path '*{d}*'")
+
+        # Allowed patterns
+        if cfg.allowed_patterns and cfg.allowed_patterns != ["*"]:
+            pat_expr = " -o ".join(
+                [f"-name '{p}'" for p in cfg.allowed_patterns if p]
+            )
+            cmd.append(f"\\( {pat_expr} \\)")
+
+        # Excluded patterns
+        if cfg.exclude_patterns:
+            for p in cfg.exclude_patterns:
+                cmd.append(f"! -name '{p}'")
+
+        # Allowed types (semantic, not `-type`)
+        if cfg.allowed_types and cfg.allowed_types != {"*"}:
+            type_expr = " -o ".join(
+                [f"-path '*{t}*'" for t in cfg.allowed_types if t]
+            )
+            cmd.append(f"\\( {type_expr} \\)")
+
+        # Excluded types
+        if cfg.exclude_types:
+            for t in cfg.exclude_types:
+                cmd.append(f"! -path '*{t}*'")
+
+    return " ".join(cmd)
+
+def ensure_directories(*args,**kwargs):
+    directories = []
+    for arg in args:
+        arg_str = str(arg)
+        if is_dir(arg_str,**kwargs):
+            directories.append(arg_str)
+        elif is_file(arg_str,**kwargs):
+            dirname = os.path.dirname(arg_str)
+            directories.append(dirname)
+    safe_directories = get_dir_filter_kwargs(**kwargs)
+    directories+= make_list(safe_directories.get('directories',[]))
+    return list(set([r for r in directories if r]))
+
+def collect_globs(
+    *args,
+    mindepth: Optional[int] = None,
+    maxdepth: Optional[int] = None,
+    depth: Optional[int] = None,
+    file_type: Optional[str] = None,   # "f", "d", or None
+    allowed: Optional[Callable[[str], bool]] = None,
+    **kwargs
+) -> List[str] | dict:
+    """
+    Collect file or directory paths recursively.
+
+    - If file_type is None → returns {"f": [...], "d": [...]}
+    - If file_type is "f" or "d" → returns a list of that type
+    - Supports SSH mode via `user_at_host`
+    """
+    kwargs["directories"] = ensure_directories(*args, **kwargs)
+    kwargs= get_safe_canonical_kwargs(**kwargs)
+    kwargs["cfg"] = define_defaults(**kwargs)
+    
+    type_strs = {"f":"files","d":"dirs"}
+    file_type = get_proper_type_str(file_type)
+    file_types = make_list(file_type)
+    if file_type == None:
+        file_types = ["f","d"]
+    return_results = {}
+    return_result=[]
+    for file_type in file_types:
+        type_str = type_strs.get(file_type)
+        # Remote path (SSH)
+        find_cmd = get_find_cmd(
+                mindepth=mindepth,
+                maxdepth=maxdepth,
+                depth=depth,
+                file_type=file_type,
+                **{k: v for k, v in kwargs.items() if v},
+            )
+        result = run_pruned_func(run_cmd,find_cmd,
+            **kwargs
+            
+            )
+        return_result = [res for res in result.split('\n') if res]
+        return_results[type_str]=return_result
+    if len(file_types) == 1:
+        return return_result
+    return return_results
+def get_files_and_dirs(
+    *args,
+    recursive: bool = True,
+    include_files: bool = True,
+    **kwargs
+    ):
+    if recursive == False:
+        kwargs['maxdepth']=1
+    if include_files == False:
+        kwargs['file_type']='d'
+    result = collect_globs(*args,**kwargs)
+    if include_files == False:
+        return result,[]
+    dirs = result.get("dirs")
+    files = result.get("files")
+    return dirs,files
+def collect_filepaths(
+    *args,
+    **kwargs
+    ) -> List[str]:
+    kwargs['file_type']='f'
+    return collect_globs(*args,**kwargs)