rda-python-miscs 3.0.5__tar.gz → 3.0.6__tar.gz

{rda_python_miscs-3.0.5/src/rda_python_miscs.egg-info → rda_python_miscs-3.0.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rda_python_miscs
-Version: 3.0.5
+Version: 3.0.6
 Summary: RDA Python package to hold RDA miscellaneous utility programs
 Author-email: Zaihua Ji <zji@ucar.edu>
 Project-URL: Homepage, https://github.com/NCAR/rda-python-miscs
@@ -43,6 +43,7 @@ The package provides two categories of programs:
 | Command | Alias | Connector script | Description |
 |---------|-------|-----------------|-------------|
 | `gdexcp` | `rdacp` | `setuid_gdexcp` / `setuid_rdacp` | Copy files and directories across local, remote, Object Store, or Globus endpoints |
+| `gdexdrop` | | `setuid_gdexdrop` | Copy local files into a GDEX dataset directory as gdexdata, for data providers outside the DECS group |
 | `gdexkill` | `rdakill` | `setuid_gdexkill` / `setuid_rdakill` | Kill local processes and their children, or cancel PBS batch jobs |
 | `gdexmod` | `rdamod` | `setuid_gdexmod` / `setuid_rdamod` | Change permission modes for files and directories owned by gdexdata |
 | `decsdata_storage` | | `setuid_decsdata_storage` | Move decsdata datasets into the GLADE HSM cold storage to migrate them onto tape |
@@ -105,7 +106,7 @@ pip install rda_python_miscs
 
 ## Setuid Setup
 
-The setuid programs (`gdexcp`, `gdexkill`, `gdexmod`, `decsdata_storage`,
+The setuid programs (`gdexcp`, `gdexdrop`, `gdexkill`, `gdexmod`, `decsdata_storage`,
 `decsdata_restore` and the `rda*` aliases) execute as the common user `PGLOG['COMMONUSER']` (default `gdexdata`) via
 the `rda_python_setuid` mechanism, which is pulled in automatically as a
 dependency.  After `pip install` above, choose one of the wiring options
@@ -162,6 +163,33 @@ The shared setuid setup guide is shown automatically if any `setuid_*`
 connector script is invoked directly before the setuid wrapper has been
 configured.
 
+## gdexdrop access list
+
+`gdexdrop` lets data providers who are **not** in the DECS group copy files into
+a dataset directory so that the result is owned by `gdexdata`.  It is a
+deliberately narrow alternative to `gdexcp`: the destination is always
+`/glade/campaign/collections/gdex/data/<dsid>`, a path that escapes the dataset
+directory is rejected, and every source must be readable by the calling user
+rather than only by `gdexdata`.
+
+Who may drop into which dataset is read from the access list
+`/glade/u/home/gdexdata/config/gdexdrop.conf`.  It must be owned by `gdexdata`
+and must not be writable by group or others, otherwise `gdexdrop` refuses to
+run.  Each line is a login name, a colon, and the colon-separated dataset IDs
+that login may drop into, or `all` for every dataset; `#` starts a comment:
+
+```
+# login: dsid1[:...:dsidn]
+jdoe: d123456:d654321
+asmith: d111222
+gdexhelp: all
+```
+
+The root path, the access list path and the log path are compiled into the
+program rather than taken from `PGLOG`, because `PGLOG['DSDHOME']`,
+`PGLOG['DSSHOME']` and `PGLOG['LOGPATH']` are all settable from environment
+variables of the same name, which the caller controls.
+
 ## Cold storage setup
 
 `decsdata_storage` and `decsdata_restore` drive the GLADE HSM through the

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/README.md

@@ -26,6 +26,7 @@ The package provides two categories of programs:
 | Command | Alias | Connector script | Description |
 |---------|-------|-----------------|-------------|
 | `gdexcp` | `rdacp` | `setuid_gdexcp` / `setuid_rdacp` | Copy files and directories across local, remote, Object Store, or Globus endpoints |
+| `gdexdrop` | | `setuid_gdexdrop` | Copy local files into a GDEX dataset directory as gdexdata, for data providers outside the DECS group |
 | `gdexkill` | `rdakill` | `setuid_gdexkill` / `setuid_rdakill` | Kill local processes and their children, or cancel PBS batch jobs |
 | `gdexmod` | `rdamod` | `setuid_gdexmod` / `setuid_rdamod` | Change permission modes for files and directories owned by gdexdata |
 | `decsdata_storage` | | `setuid_decsdata_storage` | Move decsdata datasets into the GLADE HSM cold storage to migrate them onto tape |
@@ -88,7 +89,7 @@ pip install rda_python_miscs
 
 ## Setuid Setup
 
-The setuid programs (`gdexcp`, `gdexkill`, `gdexmod`, `decsdata_storage`,
+The setuid programs (`gdexcp`, `gdexdrop`, `gdexkill`, `gdexmod`, `decsdata_storage`,
 `decsdata_restore` and the `rda*` aliases) execute as the common user `PGLOG['COMMONUSER']` (default `gdexdata`) via
 the `rda_python_setuid` mechanism, which is pulled in automatically as a
 dependency.  After `pip install` above, choose one of the wiring options
@@ -145,6 +146,33 @@ The shared setuid setup guide is shown automatically if any `setuid_*`
 connector script is invoked directly before the setuid wrapper has been
 configured.
 
+## gdexdrop access list
+
+`gdexdrop` lets data providers who are **not** in the DECS group copy files into
+a dataset directory so that the result is owned by `gdexdata`.  It is a
+deliberately narrow alternative to `gdexcp`: the destination is always
+`/glade/campaign/collections/gdex/data/<dsid>`, a path that escapes the dataset
+directory is rejected, and every source must be readable by the calling user
+rather than only by `gdexdata`.
+
+Who may drop into which dataset is read from the access list
+`/glade/u/home/gdexdata/config/gdexdrop.conf`.  It must be owned by `gdexdata`
+and must not be writable by group or others, otherwise `gdexdrop` refuses to
+run.  Each line is a login name, a colon, and the colon-separated dataset IDs
+that login may drop into, or `all` for every dataset; `#` starts a comment:
+
+```
+# login: dsid1[:...:dsidn]
+jdoe: d123456:d654321
+asmith: d111222
+gdexhelp: all
+```
+
+The root path, the access list path and the log path are compiled into the
+program rather than taken from `PGLOG`, because `PGLOG['DSDHOME']`,
+`PGLOG['DSSHOME']` and `PGLOG['LOGPATH']` are all settable from environment
+variables of the same name, which the caller controls.
+
 ## Cold storage setup
 
 `decsdata_storage` and `decsdata_restore` drive the GLADE HSM through the

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "rda_python_miscs"
-version = "3.0.5"
+version = "3.0.6"
 authors = [
   { name="Zaihua Ji", email="zji@ucar.edu" },
 ]
@@ -47,6 +47,7 @@ pythonpath = [
 "rdaown" = "rda_python_miscs.gdexown:main"
 "setuid_gdexcp" = "rda_python_miscs.gdexcp:main"
 "setuid_rdacp" = "rda_python_miscs.gdexcp:main"
+"setuid_gdexdrop" = "rda_python_miscs.gdexdrop:main"
 "setuid_gdexkill" = "rda_python_miscs.gdexkill:main"
 "setuid_rdakill" = "rda_python_miscs.gdexkill:main"
 "setuid_gdexmod" = "rda_python_miscs.gdexmod:main"

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/src/rda_python_miscs/gdexcp.py

@@ -378,7 +378,7 @@ def main():
    """Entry point: instantiate GdexCp, parse arguments, run, and exit."""
    from rda_python_setuid.setup_guide import show_setup_guide
    object = GdexCp()
-   show_setup_guide(object, 'rda_python_miscs', ['gdexcp', 'gdexkill', 'gdexmod'])
+   show_setup_guide(object, 'rda_python_miscs', ['gdexcp', 'gdexdrop', 'gdexkill', 'gdexmod'])
    object.read_parameters()
    object.start_actions()
    object.pgexit(0)

rda_python_miscs-3.0.6/src/rda_python_miscs/gdexdrop.py

@@ -0,0 +1,285 @@
+#!/usr/bin/env python3
+##################################################################################
+#     Title: gdexdrop
+#    Author: Zaihua Ji, zji@ucar.edu
+#      Date: 2026-09-23
+#   Purpose: copy local files into a GDEX dataset directory as 'gdexdata', for
+#            data providers who are not in the DECS group.  Unlike gdexcp, the
+#            destination is confined to <DROPROOT>/<dsid> and the caller must be
+#            listed in the gdexdrop access list.
+#    Github: https://github.com/NCAR/rda-python-miscs.git
+##################################################################################
+import re
+import os
+import sys
+from os import path as op
+from rda_python_common.pg_file import PgFile
+
+# These are hardcoded on purpose.  gdexdrop runs setuid to 'gdexdata' for callers
+# who are NOT in the DECS group, so nothing deciding WHERE it writes or WHO may
+# write there can come from the caller.  PgLOG.SETPGLOG() fills PGLOG['DSDHOME'],
+# PGLOG['DSSHOME'] and PGLOG['LOGPATH'] from environment variables of the same
+# names, which the caller controls, so those values are deliberately not used.
+DROPROOT = "/glade/campaign/collections/gdex/data"      # dataset data root path
+DROPCONF = "/glade/u/home/gdexdata/config/gdexdrop.conf"   # access list
+DROPLOG = "/glade/u/home/gdexdata/dssdb/log"            # log path
+DROPUSER = "gdexdata"                                   # owner of dropped files
+
+class GdexDrop(PgFile):
+   """Copy local files and directories into a GDEX dataset directory as 'gdexdata'.
+
+   The destination is always <DROPROOT>/<dsid>, optionally with a sub-path given
+   by -t, and is verified to stay inside that dataset directory.  The caller must
+   be granted the dataset in the gdexdrop access list, and must be able to read
+   every source path as themselves; both checks exist because the copy itself
+   runs with the privileges of 'gdexdata'.
+   """
+
+   def __init__(self):
+      """Initialize GdexDrop with default drop options and runtime state."""
+      super().__init__()
+      self.DROP = {
+         'ds': None,   # target dataset ID, such as d123456
+         'f': [],      # from file/directory names
+         'i': None,    # input file holding a list of from file names, one per line
+         't': None,    # sub-path under the dataset directory to drop the files into
+         'O': 0,       # 1 to override an existing target file of the same size
+         'F': 0o664,   # to file mode, default to 664
+         'D': 0o775,   # to directory mode, default to 775
+      }
+      self.DINFO = {
+         'tcnt': 0,           # count of files/directories copied
+         'target': None,      # resolved target directory
+         'curdir': os.getcwd()
+      }
+
+   # function to read parameters
+   def read_parameters(self):
+      """Parse command-line arguments, verify read access, then switch to 'gdexdata'.
+
+      The default option is -f (source paths); positional arguments are treated
+      as source paths wherever they appear.  -O is a boolean flag; -F and -D take
+      octal modes.  The source access check must run before set_suid(), while the
+      real user ID is still the caller, so that os.access() reports what the
+      CALLER can read rather than what 'gdexdata' can read.
+      """
+      dohelp = 0
+      argv = sys.argv[1:]
+      self.set_help_path(__file__)
+      self.PGLOG['LOGPATH'] = DROPLOG
+      self.PGLOG['LOGFILE'] = "gdexdrop.log"   # set different log file
+      defopt = option = 'f'
+      for arg in argv:
+         if re.match(r'-(h|-help)$', arg, re.I):
+            dohelp = 1
+            continue
+         ms = re.match(r'-(\w+)$', arg)
+         if ms:
+            option = ms.group(1)
+            if option not in self.DROP: self.pglog(arg + ": Unknown Option", self.LGEREX)
+            if option == 'O':
+               self.DROP[option] = 1
+               option = defopt
+            continue
+         if not option: self.pglog(arg + ": Value provided without option", self.LGEREX)
+         if option == 'f':
+            self.DROP['f'].append(arg)
+         else:
+            if option in ('F', 'D'):
+               self.DROP[option] = self.base2int(arg, 8)
+            else:
+               self.DROP[option] = arg
+            option = defopt
+      if self.DROP['i']: self.add_input_files(self.DROP['i'])
+      if dohelp or not self.DROP['ds'] or not self.DROP['f']: self.show_usage("gdexdrop")
+      self.check_source_access()
+      self.set_suid(self.PGLOG['EUID'])
+      self.cmdlog("gdexdrop {} ({})".format(' '.join(argv), self.DINFO['curdir']))
+
+   # read source paths from an input file and append them to the -f list
+   def add_input_files(self, infile):
+      """Append source paths read from an input file to the -f source list.
+
+      Each non-empty line is treated as one source path; leading/trailing
+      whitespace is stripped and lines starting with '#' are ignored.  The file
+      must be readable by the caller, not merely by 'gdexdata'.
+
+      Args:
+         infile (str): Path to the input file holding one source path per line.
+      """
+      if not os.access(infile, os.R_OK):
+         self.pglog("{}: Input file of -i not readable by {}".format(infile, self.PGLOG['CURUID']), self.LGEREX)
+      fd = open(infile, 'r')
+      for line in fd:
+         line = line.strip()
+         if not line or line[0] == '#': continue
+         self.DROP['f'].append(line)
+      fd.close()
+
+   # make sure the caller can read every source path as themselves
+   def check_source_access(self):
+      """Verify the calling user can read each source path, exiting if not.
+
+      Must be called while the real user ID is still the caller, i.e. before
+      set_suid().  Without this check a caller could name a path readable only by
+      'gdexdata', have it copied into a dataset directory they can read, and so
+      read it indirectly.
+      """
+      for file in self.DROP['f']:
+         path = re.sub(r'/$', '', file, 1)
+         if not re.match(r'^/', path): path = self.join_paths(self.DINFO['curdir'], path)
+         if not op.exists(path):
+            self.pglog("{}: {}".format(file, self.PGLOG['MISSFILE']), self.LGEREX)
+         if not self.caller_can_read(path):
+            self.pglog("{}: not readable by {}".format(file, self.PGLOG['CURUID']), self.LGEREX)
+
+   # check read access of one path, and of everything under it for a directory
+   def caller_can_read(self, path):
+      """Return 1 if the real user can read path and all of its contents.
+
+      os.access() tests the REAL user and group IDs, which is exactly what is
+      wanted here while running with an effective ID of 'gdexdata'.
+
+      Args:
+         path (str): Absolute source path to check.
+
+      Returns:
+         int: 1 if fully readable by the calling user, 0 otherwise.
+      """
+      if not os.access(path, os.R_OK): return 0
+      if op.isfile(path): return 1
+      if not os.access(path, os.X_OK): return 0
+      for root, dirs, files in os.walk(path):
+         for name in dirs:
+            if not os.access(op.join(root, name), os.R_OK|os.X_OK): return 0
+         for name in files:
+            if not os.access(op.join(root, name), os.R_OK): return 0
+      return 1
+
+   # function to start actions
+   def start_actions(self):
+      """Resolve and authorize the target dataset directory, then copy the sources.
+
+      Logs a summary count on completion.
+      """
+      self.DINFO['target'] = self.resolve_target()
+      self.validate_caller()
+      self.PGLOG['FILEMODE'] = self.DROP['F']
+      self.PGLOG['EXECMODE'] = self.DROP['D']
+      for file in self.DROP['f']:
+         self.DINFO['tcnt'] += self.drop_one(file)
+      if self.DINFO['tcnt'] > 0:
+         s = ('s' if self.DINFO['tcnt'] > 1 else '')
+         self.pglog("Total {} File{} dropped into {}".format(self.DINFO['tcnt'], s, self.DINFO['target']), self.LOGWRN)
+      else:
+         self.pglog("No File dropped into " + self.DINFO['target'], self.LOGWRN)
+      self.cmdlog()
+
+   # resolve the target directory and confine it to the dataset directory
+   def resolve_target(self):
+      """Return the target directory under DROPROOT for -ds, confined to the dataset.
+
+      The resolved real path of the target must equal, or lie under, the resolved
+      real path of the dataset directory, so that neither a '..' component in -t
+      nor an existing symbolic link can move the drop outside the dataset.
+
+      Returns:
+         str: Resolved absolute target directory.
+      """
+      dsid = self.DROP['ds']
+      if not re.match(r'^[a-z]\d{6}$', dsid):
+         self.pglog(dsid + ": Invalid Dataset ID of -ds, expecting the form d123456", self.LGEREX)
+      dsroot = op.realpath(op.join(DROPROOT, dsid))
+      if not op.isdir(dsroot):
+         self.pglog("{}: Dataset directory NOT exists under {}".format(dsid, DROPROOT), self.LGEREX)
+      target = self.confine_path(op.join(dsroot, self.DROP['t']) if self.DROP['t'] else dsroot, dsroot)
+      return target
+
+   # make sure a path stays inside the dataset directory
+   def confine_path(self, path, dsroot):
+      """Return the resolved path, exiting if it falls outside the dataset directory.
+
+      Args:
+         path (str): Candidate absolute path.
+         dsroot (str): Resolved absolute dataset directory.
+
+      Returns:
+         str: Resolved absolute path inside dsroot.
+      """
+      rpath = op.realpath(path)
+      if rpath != dsroot and not rpath.startswith(dsroot + '/'):
+         self.pglog("{}: Target path is outside of {}".format(path, dsroot), self.LGEREX)
+      return rpath
+
+   # check the caller is granted the target dataset in the access list
+   def validate_caller(self):
+      """Exit unless the calling user is granted the target dataset.
+
+      The access list must be owned by 'gdexdata' and must not be writable by
+      group or others; otherwise anyone able to edit it could grant themselves
+      access to any dataset.  Each line is 'login: dsid1[:...:dsidn]', or
+      'login: all' for every dataset; '#' starts a comment.  The check is
+      skipped when 'gdexdata' itself runs the command.
+      """
+      logname = self.PGLOG['CURUID']
+      if logname == DROPUSER: return
+      info = self.check_local_file(DROPCONF, 6, self.LOGWRN)
+      if not info:
+         self.pglog(DROPCONF + ": gdexdrop access list NOT exists", self.LGEREX)
+      if info['logname'] != DROPUSER:
+         self.pglog("{}: access list must be owned by {}, but is owned by {}".format(DROPCONF, DROPUSER, info['logname']), self.LGEREX)
+      if info['mode']&0o022:
+         self.pglog("{}: access list must not be group or world writable, mode is {:o}".format(DROPCONF, info['mode']), self.LGEREX)
+      dsids = {}
+      fd = open(DROPCONF, 'r')
+      for line in fd:
+         line = re.sub(r'#.*$', '', line).strip()
+         if not line: continue
+         ms = re.match(r'^(\S+)\s*:\s*(.*)$', line)
+         if not ms: continue
+         if ms.group(1) != logname: continue
+         for dsid in ms.group(2).split(':'):
+            dsid = dsid.strip()
+            if dsid: dsids[dsid] = 1
+      fd.close()
+      if not ('all' in dsids or self.DROP['ds'] in dsids):
+         self.pglog("{}: NOT granted Dataset {} in {}".format(logname, self.DROP['ds'], DROPCONF), self.LGEREX)
+
+   # copy one source path into the target directory
+   def drop_one(self, file):
+      """Copy one source file or directory into the target directory.
+
+      A directory is copied recursively, as the directory itself.  An existing
+      target file of the same size is left alone unless -O is given.
+
+      Args:
+         file (str): Source path as given on the command line.
+
+      Returns:
+         int: 1 if the source was copied, 0 otherwise.
+      """
+      path = re.sub(r'/$', '', file, 1)
+      if not re.match(r'^/', path): path = self.join_paths(self.DINFO['curdir'], path)
+      finfo = self.check_local_file(path, 0, self.LOGWRN)
+      if not finfo:
+         return self.pglog("{}: {}".format(file, self.PGLOG['MISSFILE']), self.LOGERR)
+      tofile = self.confine_path(op.join(self.DINFO['target'], op.basename(path)), self.DINFO['target'])
+      if finfo['isfile'] and not self.DROP['O']:
+         tinfo = self.check_local_file(tofile, 0, self.LOGWRN)
+         if tinfo and tinfo['data_size'] == finfo['data_size']:
+            self.pglog(tofile + ": Target exists with same size, skip copying", self.LOGWRN)
+            return 0
+      return (1 if self.local_copy_local(tofile, path, self.LGWNEX) else 0)
+
+# main function to execute this script
+def main():
+   """Entry point: instantiate GdexDrop, parse arguments, run, and exit."""
+   from rda_python_setuid.setup_guide import show_setup_guide
+   object = GdexDrop()
+   show_setup_guide(object, 'rda_python_miscs', ['gdexcp', 'gdexdrop', 'gdexkill', 'gdexmod'])
+   object.read_parameters()
+   object.start_actions()
+   object.pgexit(0)
+
+# call main() to start program
+if __name__ == "__main__": main()

rda_python_miscs-3.0.6/src/rda_python_miscs/gdexdrop.usg

@@ -0,0 +1,77 @@
+
+ Copy local files and directories into a GDEX dataset directory so that they end
+ up owned by 'gdexdata'.  It is a restricted alternative to gdexcp, intended for
+ data providers who are not in the DECS group: the destination is always the
+ given dataset's directory, and the calling user must be granted that dataset in
+ the gdexdrop access list.
+
+ Usage: gdexdrop -ds DatasetID [-t TargetPath] [-i InputFile] [-O] [-h]  \
+               [-D DirectoryMode] [-F FileMode] File/Directory List
+
+      - Option -ds DatasetID is mandatory.  It names the dataset to drop the
+          files into, in the form d123456.  The files are copied under
+          /glade/campaign/collections/gdex/data/DatasetID, which must already
+          exist;
+
+      - Option -t TargetPath, sub-path under the dataset directory to drop the
+          files into.  Missing sub-directories are created.  The path must stay
+          inside the dataset directory; '..' components and symbolic links
+          leading out of it are rejected.  Defaults to the dataset directory
+          itself;
+
+      - Option -f File/Directory List is the default option, so the source paths
+          may be given without it and in any position.  Relative paths are
+          resolved against the current directory.  A directory is copied
+          recursively, as the directory itself;
+
+      - Option -i InputFile, read additional source paths from a file, one per
+          line.  Blank lines and lines starting with '#' are ignored;
+
+      - Option -O, override an existing target file of the same size.  Without
+          it such a file is left alone and the copy is skipped;
+
+      - Option -D DirectoryMode, permission mode for created directories in
+          octal notation.  Defaults to 775;
+
+      - Option -F FileMode, permission mode for copied files in octal notation.
+          Defaults to 664;
+
+      - Option -h, display this help document.
+
+  Every source path must be readable by the calling user, not merely by
+  'gdexdata'; a path the caller cannot read themselves is refused.  Copied files
+  and directories are owned by 'gdexdata'.
+
+  Access List
+  -----------
+
+  The access list is /glade/u/home/gdexdata/config/gdexdrop.conf.  It must be
+  owned by 'gdexdata' and must not be writable by group or others.  Each line
+  is a login name, a colon, and the colon-separated dataset IDs that login may
+  drop into, or 'all' for every dataset.  '#' starts a comment:
+
+      # login: dsid1[:...:dsidn]
+      jdoe: d123456:d654321
+      asmith: d111222
+      gdexhelp: all
+
+  A login absent from the list cannot drop into any dataset.  'gdexdata' itself
+  is always allowed and needs no entry.
+
+  Examples:
+
+    1. Drop two files into the top of a dataset directory:
+
+          gdexdrop -ds d123456 hourly.202601.nc hourly.202602.nc
+
+    2. Drop the same files into a sub-directory of the dataset:
+
+          gdexdrop -ds d123456 -t 2026/monthly hourly.2026*.nc
+
+    3. Drop a whole directory tree, replacing files already there:
+
+          gdexdrop -ds d123456 -t archive -O /path/to/mydir
+
+    4. Drop a long list of files named in a file, with group-writable modes:
+
+          gdexdrop -ds d123456 -i filelist.txt -F 664 -D 775

rda_python_miscs-3.0.6/src/rda_python_miscs/gdexdrop_standalone.py

@@ -0,0 +1,23 @@
+#!/glade/work/zji/conda-envs/pg-gdex/bin/python
+# -*- coding: utf-8 -*-
+#  2026-09-23, zji@ucar.edu, created for a standalone version of gdexdrop
+#
+#  Copy this file to a common area, such as /glade/u/home/gdexdata/bin, so that
+#  users without the conda environment or a venv activated can run gdexdrop.
+#  The shebang and pgpath below name the environment gdexdrop is installed in.
+#
+#  Run directly, it executes as the CALLING user, so the dropped files are owned
+#  by the caller.  To have them owned by 'gdexdata' for a caller outside the DECS
+#  group, compile a 4755 cmwrapper binary that execs this script:
+#
+#     pywrapper-install -m gdexdrop -t <common area>/gdexdrop_standalone.py \
+#                       -d <common area>
+import re
+import sys
+pgpath = '/glade/work/zji/conda-envs/pg-gdex/lib/python3.12/site-packages'
+if pgpath not in sys.path: sys.path.insert(0, pgpath)
+
+from rda_python_miscs.gdexdrop import main
+if __name__ == '__main__':
+    sys.argv[0] = re.sub(r'(-script\.pyw|\.exe)?$', '', sys.argv[0])
+    sys.exit(main())

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/src/rda_python_miscs/gdexkill.py

@@ -260,7 +260,7 @@ def main():
    """Entry point: instantiate GdexKill, parse arguments, run, and exit."""
    from rda_python_setuid.setup_guide import show_setup_guide
    object = GdexKill()
-   show_setup_guide(object, 'rda_python_miscs', ['gdexcp', 'gdexkill', 'gdexmod'])
+   show_setup_guide(object, 'rda_python_miscs', ['gdexcp', 'gdexdrop', 'gdexkill', 'gdexmod'])
    object.read_parameters()
    object.start_actions()
    object.pgexit(0)

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/src/rda_python_miscs/gdexmod.py

@@ -196,7 +196,7 @@ def main():
    """Entry point: instantiate GdexMod, parse arguments, run, and exit."""
    from rda_python_setuid.setup_guide import show_setup_guide
    object = GdexMod()
-   show_setup_guide(object, 'rda_python_miscs', ['gdexcp', 'gdexkill', 'gdexmod'])
+   show_setup_guide(object, 'rda_python_miscs', ['gdexcp', 'gdexdrop', 'gdexkill', 'gdexmod'])
    object.read_parameters()
    object.start_actions()
    object.pgexit(0)

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6/src/rda_python_miscs.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rda_python_miscs
-Version: 3.0.5
+Version: 3.0.6
 Summary: RDA Python package to hold RDA miscellaneous utility programs
 Author-email: Zaihua Ji <zji@ucar.edu>
 Project-URL: Homepage, https://github.com/NCAR/rda-python-miscs
@@ -43,6 +43,7 @@ The package provides two categories of programs:
 | Command | Alias | Connector script | Description |
 |---------|-------|-----------------|-------------|
 | `gdexcp` | `rdacp` | `setuid_gdexcp` / `setuid_rdacp` | Copy files and directories across local, remote, Object Store, or Globus endpoints |
+| `gdexdrop` | | `setuid_gdexdrop` | Copy local files into a GDEX dataset directory as gdexdata, for data providers outside the DECS group |
 | `gdexkill` | `rdakill` | `setuid_gdexkill` / `setuid_rdakill` | Kill local processes and their children, or cancel PBS batch jobs |
 | `gdexmod` | `rdamod` | `setuid_gdexmod` / `setuid_rdamod` | Change permission modes for files and directories owned by gdexdata |
 | `decsdata_storage` | | `setuid_decsdata_storage` | Move decsdata datasets into the GLADE HSM cold storage to migrate them onto tape |
@@ -105,7 +106,7 @@ pip install rda_python_miscs
 
 ## Setuid Setup
 
-The setuid programs (`gdexcp`, `gdexkill`, `gdexmod`, `decsdata_storage`,
+The setuid programs (`gdexcp`, `gdexdrop`, `gdexkill`, `gdexmod`, `decsdata_storage`,
 `decsdata_restore` and the `rda*` aliases) execute as the common user `PGLOG['COMMONUSER']` (default `gdexdata`) via
 the `rda_python_setuid` mechanism, which is pulled in automatically as a
 dependency.  After `pip install` above, choose one of the wiring options
@@ -162,6 +163,33 @@ The shared setuid setup guide is shown automatically if any `setuid_*`
 connector script is invoked directly before the setuid wrapper has been
 configured.
 
+## gdexdrop access list
+
+`gdexdrop` lets data providers who are **not** in the DECS group copy files into
+a dataset directory so that the result is owned by `gdexdata`.  It is a
+deliberately narrow alternative to `gdexcp`: the destination is always
+`/glade/campaign/collections/gdex/data/<dsid>`, a path that escapes the dataset
+directory is rejected, and every source must be readable by the calling user
+rather than only by `gdexdata`.
+
+Who may drop into which dataset is read from the access list
+`/glade/u/home/gdexdata/config/gdexdrop.conf`.  It must be owned by `gdexdata`
+and must not be writable by group or others, otherwise `gdexdrop` refuses to
+run.  Each line is a login name, a colon, and the colon-separated dataset IDs
+that login may drop into, or `all` for every dataset; `#` starts a comment:
+
+```
+# login: dsid1[:...:dsidn]
+jdoe: d123456:d654321
+asmith: d111222
+gdexhelp: all
+```
+
+The root path, the access list path and the log path are compiled into the
+program rather than taken from `PGLOG`, because `PGLOG['DSDHOME']`,
+`PGLOG['DSSHOME']` and `PGLOG['LOGPATH']` are all settable from environment
+variables of the same name, which the caller controls.
+
 ## Cold storage setup
 
 `decsdata_storage` and `decsdata_restore` drive the GLADE HSM through the

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/src/rda_python_miscs.egg-info/SOURCES.txt

@@ -13,6 +13,9 @@ src/rda_python_miscs/decsdata_storage.usg
 src/rda_python_miscs/gdex_ls.py
 src/rda_python_miscs/gdexcp.py
 src/rda_python_miscs/gdexcp.usg
+src/rda_python_miscs/gdexdrop.py
+src/rda_python_miscs/gdexdrop.usg
+src/rda_python_miscs/gdexdrop_standalone.py
 src/rda_python_miscs/gdexkill.py
 src/rda_python_miscs/gdexkill.usg
 src/rda_python_miscs/gdexls.py

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/src/rda_python_miscs.egg-info/entry_points.txt

@@ -14,6 +14,7 @@ rdazip = rda_python_miscs.gdexzip:main
 setuid_decsdata_restore = rda_python_miscs.decsdata_restore:main
 setuid_decsdata_storage = rda_python_miscs.decsdata_storage:main
 setuid_gdexcp = rda_python_miscs.gdexcp:main
+setuid_gdexdrop = rda_python_miscs.gdexdrop:main
 setuid_gdexkill = rda_python_miscs.gdexkill:main
 setuid_gdexmod = rda_python_miscs.gdexmod:main
 setuid_rdacp = rda_python_miscs.gdexcp:main

{rda_python_miscs-3.0.5 → rda_python_miscs-3.0.6}/tests/test_miscs.py

@@ -8,6 +8,7 @@ def test_miscs():
    import rda_python_miscs.gdexls
    import rda_python_miscs.pgwget
    import rda_python_miscs.gdexcp
+   import rda_python_miscs.gdexdrop
    import rda_python_miscs.gdexkill
    import rda_python_miscs.gdexmod
    import rda_python_miscs.gdexown