abstract-utilities 0.2.2.449__py3-none-any.whl → 0.2.2.476__py3-none-any.whl

abstract_utilities/file_utils/file_utils/find_collect.py

@@ -0,0 +1,154 @@
+from .imports import *
+from .filter_params import *
+from .file_filters import enumerate_source_files
+
+
+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(
+    directory: str,
+    *,
+    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:
+    """Constructs a Unix `find` command string from keyword args."""
+    cmd = [f"find {directory}"]
+
+    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}")
+
+    if file_type in ("f", "d"):
+        cmd.append(f"-type {file_type}")
+    if name:
+        cmd.append(f"-name '{name}'")
+    if size:
+        cmd.append(f"-size {size}")
+    if mtime:
+        cmd.append(f"-mtime {mtime}")
+    if perm:
+        cmd.append(f"-perm {perm}")
+    if user:
+        cmd.append(f"-user {user}")
+
+    return " ".join(cmd)
+
+
+def collect_globs(
+    directory: str,
+    cfg: Optional["ScanConfig"] = None,
+    *,
+    exts: Optional[Set[str]] = None,
+    patterns: Optional[List[str]] = None,
+    mindepth: Optional[int] = None,
+    maxdepth: Optional[int] = None,
+    depth: Optional[int] = None,
+    file_type: Optional[str] = None,
+    user_at_host: Optional[str] = None,
+    add: bool = False,
+    **kwargs
+) -> List[str]:
+    """
+    Collect file or directory paths using either:
+      - local recursive logic (rglob)
+      - or remote shell call (find via run_cmd)
+    """
+    cfg = cfg or define_defaults(add=add)
+    directory = str(directory)
+    exts = ensure_exts(exts)
+    patterns = ensure_patterns(patterns)
+
+    # Remote path via SSH
+    if user_at_host:
+        find_cmd = get_find_cmd(
+            directory,
+            mindepth=mindepth,
+            maxdepth=maxdepth,
+            depth=depth,
+            file_type=file_type,
+            **{k: v for k, v in kwargs.items() if v},
+        )
+        return run_cmd(find_cmd, user_at_host=user_at_host)
+
+    # Local path (Python-native walk)
+    root = Path(directory)
+    results = []
+    for p in root.rglob("*"):
+        if file_type == "f" and not p.is_file():
+            continue
+        if file_type == "d" and not p.is_dir():
+            continue
+        if exts and p.suffix.lower() not in exts:
+            continue
+        if patterns and not any(p.match(pat) for pat in patterns):
+            continue
+        results.append(str(p.resolve()))
+
+    return sorted(results)

abstract_utilities/file_utils/file_utils/imports/__init__.py

@@ -0,0 +1,3 @@
+from .constants import *
+from .imports import *
+from ..imports import *

abstract_utilities/file_utils/file_utils/imports/constants.py

@@ -0,0 +1,39 @@
+from .imports import *
+from .module_imports import *
+@dataclass
+class ScanConfig:
+    allowed_exts: Set[str]
+    unallowed_exts: Set[str]
+    exclude_types: Set[str]
+    exclude_dirs: List[str] = field(default_factory=list)
+    exclude_patterns: List[str] = field(default_factory=list)
+DEFAULT_ALLOWED_EXTS: Set[str] = {
+    ".py", ".pyw",                             # python
+    ".js", ".jsx", ".ts", ".tsx", ".mjs",      # JS/TS
+    ".html", ".htm", ".xml",                   # markup
+    ".css", ".scss", ".sass", ".less",         # styles
+    ".json", ".yaml", ".yml", ".toml", ".ini",  # configs
+    ".cfg", ".md", ".markdown", ".rst",        # docs
+    ".sh", ".bash", ".env",                    # scripts/env
+    ".txt"                                     # plain text
+}
+
+DEFAULT_EXCLUDE_TYPES: Set[str] = {
+    "image", "video", "audio", "presentation",
+    "spreadsheet", "archive", "executable"
+}
+
+# never want these—even if they sneak into ALLOWED
+_unallowed = set(get_media_exts(DEFAULT_EXCLUDE_TYPES)) | {'.bak', '.shp', '.cpg', '.dbf', '.shx','.geojson',".pyc",'.shx','.geojson','.prj','.sbn','.sbx'}
+DEFAULT_UNALLOWED_EXTS = {e for e in _unallowed if e not in DEFAULT_ALLOWED_EXTS}
+
+DEFAULT_EXCLUDE_DIRS: Set[str] = {
+    "node_modules", "old","__pycache__", "backups", "backup", "backs", "trash", "depriciated", "old", "__init__"
+}
+
+DEFAULT_EXCLUDE_PATTERNS: Set[str] = {
+    "__init__*", "*.tmp", "*.log", "*.lock", "*.zip","*~"
+}
+REMOTE_RE = re.compile(r"^(?P<host>[^:\s]+@[^:\s]+):(?P<path>/.*)$")
+AllowedPredicate = Optional[Callable[[str], bool]]
+DEFAULT_EXCLUDE_FILE_PATTERNS=DEFAULT_EXCLUDE_PATTERNS

abstract_utilities/file_utils/file_utils/imports/file_functions.py

@@ -0,0 +1,10 @@
+from .imports import *
+def get_caller_path():
+    i = i or 1
+    frame = inspect.stack()[i]
+    return os.path.abspath(frame.filename)
+def get_caller_dir(i=None):
+    i = i or 1
+    frame = inspect.stack()[i]
+    abspath = os.path.abspath(frame.filename)
+    return os.path.dirname(abspath)

abstract_utilities/file_utils/file_utils/imports/imports.py

@@ -0,0 +1,39 @@
+# ============================================================
+# abstract_utilities/imports/imports.py
+# Global imports hub — everything imported here will be
+# automatically available to any module that does:
+#     from ..imports import *
+# ============================================================
+# ---- Core standard library modules -------------------------
+import os, sys, re, shlex, glob, platform, textwrap, subprocess, inspect, json, time
+import tempfile, shutil, logging, pathlib, fnmatch, importlib, importlib.util, types
+from pathlib import Path
+from datetime import datetime
+from types import ModuleType
+
+# ---- Dataclasses and typing --------------------------------
+from dataclasses import dataclass, field
+from typing import (
+    Any, Optional, List, Dict, Set, Tuple,
+    Iterable, Callable, Literal, Union, TypeVar
+)
+
+# ---- Common 3rd-party dependencies --------------------------
+import pandas as pd
+import geopandas as gpd
+import pytesseract
+import pdfplumber
+import PyPDF2
+import ezodf
+from pdf2image import convert_from_path
+from werkzeug.utils import secure_filename
+from werkzeug.datastructures import FileStorage
+
+# ---- Helpers ------------------------------------------------
+import textwrap as tw
+from pprint import pprint
+
+# ============================================================
+# AUTO-EXPORT ALL NON-PRIVATE NAMES
+# ============================================================
+__all__ = [name for name in globals() if not name.startswith("_")]

abstract_utilities/file_utils/file_utils/imports/module_imports.py

@@ -0,0 +1,13 @@
+from .imports import *
+from ....string_clean import eatAll
+from ....list_utils import make_list
+from ....type_utils import get_media_exts, is_media_type, MIME_TYPES, is_str
+from ....ssh_utils import *
+from ....env_utils import *
+from ....read_write_utils import *
+from ....abstract_classes import SingletonMeta
+
+from ....class_utils import get_caller, get_caller_path, get_caller_dir
+
+
+__all__ = [name for name in globals() if not name.startswith("_")]

abstract_utilities/file_utils/file_utils/imports.py

@@ -0,0 +1,10 @@
+from ..imports import *
+from typing import *
+from dataclasses import dataclass, field
+@dataclass
+class ScanConfig:
+    allowed_exts: Set[str]
+    unallowed_exts: Set[str]
+    exclude_types: Set[str]
+    exclude_dirs: List[str] = field(default_factory=list)
+    exclude_patterns: List[str] = field(default_factory=list)

abstract_utilities/file_utils/file_utils/map_utils.py

@@ -0,0 +1,29 @@
+from .imports import MIME_TYPES,make_list,os
+from .file_filters import get_globs
+def get_file_type(file_path,types=None,default=None):
+    mime_types = {}
+    if types:
+        types = make_list(types)
+        for typ in types:
+            mime_types[typ] = MIME_TYPES.get(typ)
+    else:
+        mime_types = MIME_TYPES
+    
+    if os.path.isfile(file_path):
+        basename = os.path.basename(file_path)
+        filename,ext = os.path.splitext(basename)
+        for file_type,ext_values in mime_types.items():
+            if ext in ext_values:
+                return file_type
+def get_file_map(directory,types=None,default=None):
+    if directory and os.path.isfile(directory):
+       directory = os.path.dirname(directory)
+    all_types = {}
+    files = get_globs(directory)
+    for file in files:
+        file_type = get_file_type(file,types=types,default=default)
+        if file_type:
+            if file_type not in all_types:
+                all_types[file_type] = []
+            all_types[file_type].append(file)
+    return all_types

abstract_utilities/{robust_reader → file_utils/file_utils}/pdf_utils.py

@@ -1,12 +1,4 @@
-import PyPDF2
-from typing import *
-from pdf2image import convert_from_path
-from abstract_utilities.path_utils import (is_file, mkdirs, get_directory, 
-                                           get_base_name, split_text, 
-                                           get_ext, get_file_name)
-from abstract_utilities.type_utils import is_str
-from abstract_utilities.cmd_utils import cmd_input
-from abstract_utilities.read_write_utils import write_to_file
+from .imports import *
 def if_none_return(obj: object, obj_2: object) -> object:
     """
     Return obj if obj_2 is None, otherwise return obj_2.

abstract_utilities/file_utils/file_utils/type_checks.py

@@ -0,0 +1,82 @@
+from .imports import *
+
+
+# --- Base remote checker -----------------------------------------------------
+def _remote_test(path: str, test_flag: str, user_at_host: str, timeout: int = 5) -> bool:
+    """
+    Run a remote shell test (e.g. -f, -d) via SSH.
+    Returns True if test succeeds, False otherwise.
+    """
+    cmd = f"[ {test_flag} {shlex.quote(path)} ] && echo 1 || echo 0"
+    try:
+        result = subprocess.check_output(
+            ["ssh", user_at_host, cmd],
+            stderr=subprocess.DEVNULL,
+            text=True,
+            timeout=timeout
+        ).strip()
+        return result == "1"
+    except Exception:
+        return False
+
+
+# --- Individual path checks --------------------------------------------------
+def is_remote_file(path: str, user_at_host: str) -> bool:
+    """True if remote path is a file."""
+    return _remote_test(path, "-f", user_at_host)
+
+
+def is_remote_dir(path: str, user_at_host: str) -> bool:
+    """True if remote path is a directory."""
+    return _remote_test(path, "-d", user_at_host)
+
+
+def is_local_file(path: str) -> bool:
+    """True if local path is a file."""
+    return os.path.isfile(path)
+
+
+def is_local_dir(path: str) -> bool:
+    """True if local path is a directory."""
+    return os.path.isdir(path)
+
+
+# --- Unified interface -------------------------------------------------------
+def is_file(path: str,*args, user_at_host: Optional[str] = None,**kwargs) -> bool:
+    """Determine if path is a file (works local or remote)."""
+    if user_at_host:
+        return is_remote_file(path, user_at_host)
+    return is_local_file(path)
+
+
+def is_dir(path: str, *args,user_at_host: Optional[str] = None,**kwargs) -> bool:
+    """Determine if path is a directory (works local or remote)."""
+    if user_at_host:
+        return is_remote_dir(path, user_at_host)
+    return is_local_dir(path)
+
+
+# --- Optional: keep your original all-in-one wrapper ------------------------
+def check_path_type(
+    path: str,
+    user_at_host: Optional[str] = None,
+) -> str:
+    """
+    Return 'file', 'directory', 'missing', or 'unknown'.
+    Uses isolated is_file/is_dir functions.
+    """
+    if user_at_host:
+        if is_remote_file(path, user_at_host):
+            return "file"
+        elif is_remote_dir(path, user_at_host):
+            return "directory"
+        else:
+            return "missing"
+    else:
+        if os.path.isfile(path):
+            return "file"
+        elif os.path.isdir(path):
+            return "directory"
+        elif not os.path.exists(path):
+            return "missing"
+        return "unknown"

abstract_utilities/file_utils/imports/__init__.py

@@ -0,0 +1,4 @@
+from .constants import *
+from .module_imports import *
+from .classes import *
+from .imports import *