abstract-utilities 0.2.2.540__py3-none-any.whl → 0.2.2.593__py3-none-any.whl

abstract_utilities/class_utils/caller_utils.py

@@ -1,4 +1,22 @@
 from .imports import *
+
+def get_initial_caller() -> str:
+    """
+    Return the TRUE original caller: the entrypoint script that launched the program.
+    """
+    main_mod = sys.modules.get('__main__')
+    
+    # interactive environments (REPL) may have no __file__
+    if not main_mod or not hasattr(main_mod, '__file__'):
+        return None
+
+    return os.path.realpath(main_mod.__file__)
+def get_initial_caller_dir() -> str:
+    """
+    Return the directory of the TRUE original entrypoint script.
+    """
+    caller = get_initial_caller()
+    return os.path.dirname(caller) if caller else None
 def get_caller(i: Optional[int] = None) -> str:
     """
     Return the filename of the calling frame.

abstract_utilities/class_utils/global_utils.py

@@ -48,9 +48,10 @@ def change_glob(var: str, val: any, glob: dict = return_globals()) -> any:
 def get_globes(string:str='',glob:dict=return_globals()):
     if string in glob:
         return glob[string]
-def if_none_default(string:str, default:any,glob:dict=return_globals()):
+def if_none_default(string:str, default:any,glob:dict=return_globals(),typ=None):
+    
     piece = get_globes(string=string,glob=glob)
-    if piece is None:
+    if piece is None or (typ and not isinstance(piece,typ)):
         piece = default
     return change_glob(var=string,val=piece,glob=glob)
 

abstract_utilities/class_utils/imports/imports.py

@@ -1,2 +1,2 @@
-from ...imports import inspect,os,json,functools,inspect,glob
+from ...imports import inspect,os,json,functools,inspect,glob,sys
 from typing import *

abstract_utilities/directory_utils/src/directory_utils.py

@@ -1,5 +1,21 @@
-
+from pathlib import Path
 from .utils import *
+
+def get_shortest_path(*args):
+    shortest = None
+    args = {arg:str(arg).split('/') for arg in args if arg}
+    for path,spl in args.items():
+        if shortest is None or (len(spl)<len(shortest[-1])):
+            shortest=[path,spl]
+    return shortest[0]
+def get_common_root(paths):
+    """
+    Compute the deepest common parent directory of all sysroots.
+    """
+    paths = [Path(p).resolve() for p in paths]
+    common = os.path.commonpath(paths)
+    return common
+
 def get_dirs(path):
     """
     Get List of Immediate Subdirectories in a Path

abstract_utilities/file_utils/imports/imports.py

@@ -8,10 +8,6 @@
 
 from ...imports import *
 from pathlib import Path
-
-
-
-
 from typing import *
 
 from typing import *

abstract_utilities/file_utils/imports/module_imports.py

@@ -4,5 +4,5 @@ from ...ssh_utils import *
 from ...env_utils import *
 from ...read_write_utils import read_from_file,write_to_file
 from ...log_utils import get_logFile
-from ...class_utils import get_caller, get_caller_path, get_caller_dir,SingletonMeta,run_pruned_func
+from ...class_utils import if_none_default,if_none_change,get_caller,get_initial_caller_dir, get_caller_path, get_caller_dir,SingletonMeta,run_pruned_func
 

abstract_utilities/file_utils/src/__init__.py

@@ -1,8 +1,7 @@
 from .file_filters import *
-from .filter_params import *
 from .map_utils import *
-from .pdf_utils import *
-from .file_reader import *
+from .reader_utils import *
 from .find_collect import *
+from .find_content import *
 from .initFunctionsGen import call_for_all_tabs,get_for_all_tabs
 

abstract_utilities/file_utils/src/file_filters/__init__.py

@@ -0,0 +1,4 @@
+from .ensure_utils import *
+from .filter_params import *
+from .filter_utils import *
+from .predicate_utils import *

abstract_utilities/file_utils/src/file_filters/ensure_utils.py

@@ -0,0 +1,118 @@
+from .filter_utils import *
+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 ensure_directories(*args,**kwargs):
+    directories = []
+    for arg in args:
+        arg_str = str(arg)
+        
+        if run_pruned_func(is_dir,arg_str,**kwargs):
+            directories.append(arg_str)
+        elif run_pruned_func(is_file,arg_str,**kwargs):
+            dirname = os.path.dirname(arg_str)
+            directories.append(dirname)
+    safe_directories = get_dir_filter_kwargs(**kwargs)
+    safe_dirs = safe_directories.get('directories')
+    safe_dirs = if_none_change(safe_dirs or None,get_initial_caller_dir())
+    directories+= make_list(safe_dirs)
+    return list(set([r for r in directories if r]))
+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"

abstract_utilities/file_utils/src/file_filters/filter_params.py

@@ -0,0 +1,86 @@
+from .predicate_utils import *
+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",True)
+    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)
+
+def get_file_filters(*args,**kwargs):
+    directories = ensure_directories(*args,**kwargs)
+    recursive = kwargs.get('recursive',True)
+    include_files = kwargs.get('include_files',True)
+    cfg = define_defaults(**kwargs)
+    allowed = kwargs.get("allowed") or make_allowed_predicate(cfg)
+    return directories,cfg,allowed,include_files,recursive

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 +1,114 @@
+from .ensure_utils import *
+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

abstract_utilities/file_utils/src/file_reader.py

@@ -1,7 +1,6 @@
 # file_reader.py
 from ..imports import *
 # -------- Public API drop-ins that mirror your originals --------
-from .filter_params import *
 from .file_filters import *
 from .pdf_utils import *
 # ---------------------------------------------------------------------------

abstract_utilities/file_utils/src/find_collect.py

@@ -1,76 +1,8 @@
 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"
 
 
 
@@ -95,15 +27,15 @@ def get_find_cmd(
     """
     # Normalize inputs into canonical form
     kwargs = get_safe_canonical_kwargs(*args, **kwargs)
-    cfg = define_defaults(**kwargs)
+    cfg = kwargs.get('cfg') or 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:
+    kwargs["directories"] = ensure_directories(*args, **kwargs)
+    if not kwargs["directories"]:
         return []
 
     # Build base command for all directories
-    dir_expr = " ".join(shlex.quote(d) for d in directories)
+    dir_expr = " ".join(shlex.quote(d) for d in kwargs["directories"])
     cmd = [f"find {dir_expr}"]
 
     # --- depth filters ---
@@ -183,18 +115,7 @@ def get_find_cmd(
 
     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,
@@ -212,9 +133,10 @@ def collect_globs(
     - If file_type is "f" or "d" → returns a list of that type
     - Supports SSH mode via `user_at_host`
     """
+    user_pass_host_key = get_user_pass_host_key(**kwargs)
     kwargs["directories"] = ensure_directories(*args, **kwargs)
     kwargs= get_safe_canonical_kwargs(**kwargs)
-    kwargs["cfg"] = define_defaults(**kwargs)
+    kwargs["cfg"] = kwargs.get('cfg') or define_defaults(**kwargs)
     
     type_strs = {"f":"files","d":"dirs"}
     file_type = get_proper_type_str(file_type)
@@ -227,11 +149,13 @@ def collect_globs(
         type_str = type_strs.get(file_type)
         # Remote path (SSH)
         find_cmd = get_find_cmd(
+            directories=kwargs.get("directories"),
+            cfg=kwargs.get('cfg'),
                 mindepth=mindepth,
                 maxdepth=maxdepth,
                 depth=depth,
                 file_type=file_type,
-                **{k: v for k, v in kwargs.items() if v},
+                **user_pass_host_key,
             )
         result = run_pruned_func(run_cmd,find_cmd,
             **kwargs