abstract-utilities 0.2.2.449__py3-none-any.whl → 0.2.2.451__py3-none-any.whl

abstract_utilities/file_utils/file_utils/find_collect.py

@@ -0,0 +1,153 @@
+from .imports import *
+from .filter_params import define_defaults, ensure_patterns, ensure_exts
+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,
+) -> 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.py

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

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/imports/__init__.py

@@ -0,0 +1,5 @@
+from .constants import *
+from .imports import *
+from .module_imports import *
+from .classes import *
+from .file_functions import *

abstract_utilities/file_utils/imports/classes.py

@@ -0,0 +1,381 @@
+from .imports import *
+from .module_imports import *
+from .constants import *
+def get_item_check_cmd(path, file=True, directory=False, exists=False):
+    if (directory and file) or exists:
+        typ = "e"
+    elif file:
+        typ = "f"
+    elif directory:
+        typ = "d"
+    elif isinstance(file, str):
+        if "f" in file:
+            typ = "f"
+        elif "d" in file:
+            typ = "d"
+        else:
+            typ = "e"
+    else:
+        typ = "e"
+    return f"test -{typ} {shlex.quote(path)} && echo __OK__ || true"
+
+
+def get_all_item_check_cmd(path, file=True, directory=True, exists=True):
+    collects = []
+    out_js = {}
+
+    if file:
+        collects.append("file")
+    if directory:
+        collects.append("dir")
+    if exists:
+        collects.append("exists")
+
+    if not collects:
+        return out_js
+
+    path = shlex.quote(path)
+    for typ in collects:
+        t = typ[0]  # f, d, or e
+        out_js[typ] = f"test -{t} {path} && echo __OK__ || true"
+
+    return out_js
+        
+
+def is_file(
+    path,
+    user_at_host=None,
+    password=None,
+    key=None,
+    env_path=None,
+    **kwargs
+    ):
+    contingencies = list(set([user_at_host,password,key,env_path]))
+    len_contingencies = len(contingencies)
+    is_potential = (len_contingencies >1 or (None not in contingencies))
+    if not is_potential:
+        return os.path.isfile(path)
+    cmd = get_item_check_cmd(path,file=True)
+    return run_cmd(cmd=cmd,
+            user_at_host=user_at_host,
+            password=password,
+            key=key,
+            env_path=env_path,
+            **kwargs
+            )
+def is_dir(
+    path,
+    user_at_host=None,
+    password=None,
+    key=None,
+    env_path=None,
+    **kwargs
+    ):
+    contingencies = list(set([user_at_host,password,key,env_path]))
+    len_contingencies = len(contingencies)
+    is_potential = (len_contingencies >1 or (None not in contingencies))
+    if not is_potential:
+        return os.path.isdir(path)
+    cmd = get_item_check_cmd(path,file=False,directory=True)
+    return run_cmd(cmd=cmd,
+            user_at_host=user_at_host,
+            password=password,
+            key=key,
+            env_path=env_path,
+            **kwargs
+            )
+def is_exists(
+    path,
+    user_at_host=None,
+    password=None,
+    key=None,
+    env_path=None,
+    **kwargs
+    ):
+    contingencies = list(set([user_at_host,password,key,env_path]))
+    len_contingencies = len(contingencies)
+    is_potential = (len_contingencies >1 or (None not in contingencies))
+    if not is_potential:
+        return os.path.exists(path)
+    if is_potential == True:
+        cmd = get_item_check_cmd(path,exists=True)
+        return run_cmd(cmd=cmd,
+                user_at_host=user_at_host,
+                password=password,
+                key=key,
+                env_path=env_path,
+                **kwargs
+                )
+def is_any(
+    path,
+    user_at_host=None,
+    password=None,
+    key=None,
+    env_path=None,
+    **kwargs
+    ):
+    contingencies = list(set([user_at_host,password,key,env_path]))
+    len_contingencies = len(contingencies)
+    is_potential = (len_contingencies >1 or (None not in contingencies))
+    if not is_potential:
+        return os.path.exists(path)
+    if is_potential == True:
+        out_js = get_all_item_check_cmd(path,file=True,directory=True,exists=True)
+        for typ,cmd in out_js.items():
+            response = run_cmd(cmd=cmd,
+                    user_at_host=user_at_host,
+                    password=password,
+                    key=key,
+                    env_path=env_path,
+                    **kwargs
+                    )
+            result = "__OK__" in (response or "")
+            if result:
+                return typ
+    return None
+class PathBackend(Protocol):
+    def join(self, *parts: str) -> str: ...
+    def isfile(self, path: str) -> bool: ...
+    def isdir(self, path: str) -> bool: ...
+    def glob_recursive(self, base: str, **opts) -> List[str]: ...
+    def listdir(self, base: str) -> List[str]: ...
+
+class LocalFS:
+    def __init__(self, get_type=False, get_is_dir=False, get_is_file=False, get_is_exists=False, **kwargs):
+        self.get_type = get_type
+        self.get_is_dir = get_is_dir
+        self.get_is_file = get_is_file
+        self.get_is_exists = get_is_exists
+
+    def join(self, *parts: str) -> str:
+        return os.path.join(*parts)
+
+    def isfile(self, path: str) -> bool:
+        return os.path.isfile(path)
+
+    def isdir(self, path: str) -> bool:
+        return os.path.isdir(path)
+
+    def isexists(self, path: str) -> bool:
+        return os.path.exists(path)
+
+    def istype(self, path: str) -> str | None:
+        funcs_js = {"file": os.path.isfile, "dir": os.path.isdir, "exists": os.path.exists}
+        for key, func in funcs_js.items():
+            if func(path):
+                return key
+        return None
+
+    def is_included(self, path, **kwargs):
+        include_js = {}
+        if self.get_type:
+            include_js["typ"] = self.istype(path)
+        if self.get_is_dir:
+            include_js["dir"] = self.isdir(path)
+        if self.get_is_file:
+            include_js["file"] = self.isfile(path)
+        if self.get_is_exists:
+            include_js["exists"] = self.isexists(path)
+        return include_js
+    def glob_recursive(self, base: str, **opts) -> List[str]:
+        """
+        opts:
+          - maxdepth: int | None
+          - mindepth: int (default 1)
+          - follow_symlinks: bool
+          - include_dirs: bool
+          - include_files: bool
+          - exclude_hidden: bool
+        """
+        maxdepth = opts.get("maxdepth")
+        mindepth = opts.get("mindepth", 1)
+        follow   = opts.get("follow_symlinks", False)
+        want_d   = opts.get("include_dirs", True)
+        want_f   = opts.get("include_files", True)
+        hide     = opts.get("exclude_hidden", False)
+
+        results: List[str] = []
+        base_depth = os.path.normpath(base).count(os.sep)
+
+        for root, dirs, files in os.walk(base, followlinks=follow):
+            depth = os.path.normpath(root).count(os.sep) - base_depth
+            if maxdepth is not None and depth > maxdepth:
+                dirs[:] = []
+                continue
+            if want_d and depth >= mindepth:
+                for d in dirs:
+                    if hide and d.startswith("."): continue
+                    results.append(os.path.join(root, d))
+            if want_f and depth >= mindepth:
+                for f in files:
+                    if hide and f.startswith("."): continue
+                    results.append(os.path.join(root, f))
+        return results
+
+    def listdir(self, base: str) -> List[str]:
+        try:
+            return [os.path.join(base, name) for name in os.listdir(base)]
+        except Exception:
+            return []
+def get_spec_kwargs(
+    user_at_host=None,
+    password=None,
+    key=None,
+    env_path=None,
+    kwargs=None
+):
+    kwargs = kwargs or {}
+    kwargs["user_at_host"] = kwargs.get("user_at_host") or user_at_host
+    kwargs["password"] = kwargs.get("password") or password
+    kwargs["key"] = kwargs.get("key") or key
+    kwargs["env_path"] = kwargs.get("env_path") or env_path
+    return kwargs
+class SSHFS:
+    """Remote POSIX backend via run_remote_cmd."""
+    def __init__(self, password=None, key=None, env_path=None,
+                 get_type=False, get_is_dir=False, get_is_file=False, get_is_exists=False, **kwargs):
+        self.user_at_host = kwargs.get('user_at_host') or kwargs.get('user') or kwargs.get('host')
+        self.password = password
+        self.key = key
+        self.env_path = env_path
+        self.get_type = get_type
+        self.get_is_dir = get_is_dir
+        self.get_is_file = get_is_file
+        self.get_is_exists = get_is_exists
+
+    def cell_spec_kwargs(self, func, path, **kwargs):
+        kwargs = get_spec_kwargs(
+            user_at_host=self.user_at_host,
+            password=self.password,
+            key=self.key,
+            env_path=self.env_path,
+            kwargs=kwargs
+        )
+        return func(path, **kwargs)
+
+    def is_included(self, path, **kwargs):
+        include_js = {}
+        if self.get_type:
+            include_js["typ"] = self.istype(path, **kwargs)
+        if self.get_is_dir:
+            include_js["dir"] = self.isdir(path, **kwargs)
+        if self.get_is_file:
+            include_js["file"] = self.isfile(path, **kwargs)
+        if self.get_is_exists:
+            include_js["exists"] = self.isexists(path, **kwargs)
+        return include_js
+
+    def join(self, *parts: str) -> str:
+        return posixpath.join(*parts)
+
+    def isfile(self, path: str, **kwargs) -> bool:
+        out = self.cell_spec_kwargs(is_file, path, **kwargs)
+        return "__OK__" in (out or "")
+
+    def isdir(self, path: str, **kwargs) -> bool:
+        out = self.cell_spec_kwargs(is_dir, path, **kwargs)
+        return "__OK__" in (out or "")
+
+    def isexists(self, path: str, **kwargs) -> bool:
+        out = self.cell_spec_kwargs(is_exists, path, **kwargs)
+        return "__OK__" in (out or "")
+
+    def istype(self, path: str, **kwargs) -> str | None:
+        out = self.cell_spec_kwargs(is_any, path, **kwargs)
+        return out
+
+    def glob_recursive(self, base: str, **opts) -> List[str]:
+        maxdepth = opts.get("maxdepth")
+        mindepth = opts.get("mindepth", 1)
+        follow   = opts.get("follow_symlinks", False)
+        want_d   = opts.get("include_dirs", True)
+        want_f   = opts.get("include_files", True)
+        hide     = opts.get("exclude_hidden", False)
+
+        parts = []
+        if follow:
+            parts.append("-L")
+        parts += ["find", shlex.quote(base)]
+        if mindepth is not None:
+            parts += ["-mindepth", str(mindepth)]
+        if maxdepth is not None:
+            parts += ["-maxdepth", str(maxdepth)]
+
+        type_filters = []
+        if want_d and not want_f:
+            type_filters = ["-type", "d"]
+        elif want_f and not want_d:
+            type_filters = ["-type", "f"]
+
+        hidden_filter = []
+        if hide:
+            hidden_filter = ["!", "-regex", r".*/\..*"]
+
+        cmd = " ".join(parts + type_filters + hidden_filter + ["-printf", r"'%p\n'"]) + " 2>/dev/null"
+        out = run_remote_cmd(self.user_at_host, cmd)
+        return [line.strip().strip("'") for line in (out or "").splitlines() if line.strip()]
+
+    def listdir(self, base: str) -> List[str]:
+        cmd = f"find {shlex.quote(base)} -maxdepth 1 -mindepth 1 -printf '%p\\n' 2>/dev/null"
+        out = run_remote_cmd(self.user_at_host, cmd)
+        return [line.strip() for line in (out or "").splitlines() if line.strip()]
+
+
+
+def try_group(pre,item,strings):
+    
+    try:
+        m = pre.match(item)
+        for i,string in enumerate(strings):
+            strings[i] = m.group(string)
+        
+    except:
+        return None
+    return strings
+def normalize_items(
+    paths: Iterable[str],
+    user_at_host=None,
+    get_type=True,
+    get_is_dir=False,
+    get_is_file=False,
+    get_is_exists=False,
+    **kwargs
+) -> List[tuple[PathBackend, str, dict]]:
+    pairs: List[tuple[PathBackend, str, dict]] = []
+    host = user_at_host or kwargs.get("host") or kwargs.get("user")
+    paths = make_list(paths)
+    for item in paths:
+        if not item:
+            continue
+
+        strings = try_group(REMOTE_RE, item, ["host", "path"])
+        fs_host = None
+        nuhost = None
+
+        if (strings and None not in strings) or host:
+            if strings and None not in strings:
+                nuhost = strings[0]
+                item = strings[1] or item
+            nuhost = nuhost or host
+            fs_host = SSHFS(
+                nuhost,
+                user_at_host=user_at_host,
+                get_type=get_type,
+                get_is_dir=get_is_dir,
+                get_is_file=get_is_file,
+                get_is_exists=get_is_exists,
+                **kwargs
+            )
+        else:
+            fs_host = LocalFS(
+                get_type=get_type,
+                get_is_dir=get_is_dir,
+                get_is_file=get_is_file,
+                get_is_exists=get_is_exists
+            )
+
+        includes = fs_host.is_included(item)
+        pairs.append((fs_host, item, includes))
+    return pairs
+
+

abstract_utilities/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/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/imports/imports.py

@@ -0,0 +1,13 @@
+import pandas as pd
+from typing import *
+import geopandas as gpd
+from pathlib import Path
+from types import ModuleType
+from datetime import datetime
+from pdf2image import convert_from_path
+from dataclasses import dataclass, field
+from werkzeug.utils import secure_filename
+from werkzeug.datastructures import FileStorage
+import fnmatch, fnmatch,shlex, os, glob, platform, textwrap, pkgutil,time
+import tempfile,shutil,logging,ezodf,fnmatch,pytesseract,pdfplumber,re
+import textwrap, sys, types, importlib, importlib.util, inspect,PyPDF2

abstract_utilities/file_utils/imports/module_imports.py

@@ -0,0 +1,10 @@
+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 ...log_utils import get_logFile
+from ...class_utils import get_caller,get_caller_path,get_caller_dir
+from ...ssh_utils  import run_cmd