rda-python-common 3.0.27__tar.gz → 3.0.29__tar.gz

{rda_python_common-3.0.27/src/rda_python_common.egg-info → rda_python_common-3.0.29}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rda_python_common
-Version: 3.0.27
+Version: 3.0.29
 Summary: RDA Python common library codes shared by other RDA python packages
 Author-email: Zaihua Ji <zji@ucar.edu>
 Project-URL: Homepage, https://github.com/NCAR/rda-python-common
@@ -189,7 +189,7 @@ PgLOG.pglog("hello", PgLOG.LOGWRN)
 python -c "import rda_python_common; print(rda_python_common.__version__)"
 ```
 
-You should see the installed version (currently `3.0.27`).  If the import
+You should see the installed version (currently `3.0.29`).  If the import
 fails, double-check that the active Python environment is the one where you
 ran `pip install`.
 

{rda_python_common-3.0.27 → rda_python_common-3.0.29}/README.md

@@ -165,7 +165,7 @@ PgLOG.pglog("hello", PgLOG.LOGWRN)
 python -c "import rda_python_common; print(rda_python_common.__version__)"
 ```
 
-You should see the installed version (currently `3.0.27`).  If the import
+You should see the installed version (currently `3.0.29`).  If the import
 fails, double-check that the active Python environment is the one where you
 ran `pip install`.
 

{rda_python_common-3.0.27 → rda_python_common-3.0.29}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "rda_python_common"
-version = "3.0.27"
+version = "3.0.29"
 authors = [
   { name="Zaihua Ji", email="zji@ucar.edu" },
 ]

{rda_python_common-3.0.27 → rda_python_common-3.0.29}/src/rda_python_common/__init__.py

@@ -22,7 +22,7 @@ object that existing callers expect.
 
 from . import PgLOG, PgUtil, PgDBI, PgFile, PgLock, PgCMD, PgSIG, PgOPT, PgSplit
 
-__version__ = "3.0.27"
+__version__ = "3.0.29"
 
 __all__ = [
    "PgLOG",

{rda_python_common-3.0.27 → rda_python_common-3.0.29}/src/rda_python_common/pg_file.py

@@ -410,6 +410,92 @@ class PgFile(PgUtil, PgSIG):
          self.errlog("Error Execute: {}\n{}".format(cmd, buf), 'O', loop, logact)
       return self.FAILURE
 
+   # Upload a local directory to object under a key prefix
+   #    todir - target object key prefix, without the trailing '/'
+   #  fromdir - source local directory name
+   def local_copy_object_directory(self, todir, fromdir, bucket = None, meta = None, logact = 0):
+      """Upload every file under a local directory to an object key prefix.
+
+      The whole directory goes up in one recursive call, the upload twin of
+      object_copy_local_directory(): one process per file costs several seconds
+      of start-up each, which dominates for a directory of many files.  Each file
+      lands at '{todir}/{path relative to fromdir}'.
+
+      Nothing is uploaded when every file is already under the prefix at its local
+      size (unless OVRIDE is set); otherwise the whole directory is uploaded again,
+      since the recursive call cannot be limited to the missing files.
+
+      Args:
+         todir (str): Object key prefix (destination path in the bucket); not empty.
+         fromdir (str): Source local directory path.
+         bucket (str | None): Target bucket; defaults to PGLOG['OBJCTBKT'].
+         meta (dict | None): Extra metadata key/value pairs to attach to each object.
+         logact (int): Logging action flags; default 0.
+
+      Returns:
+         int: self.SUCCESS on success, self.FAILURE on error.
+      """
+      if not bucket: bucket = self.PGLOG['OBJCTBKT']
+      if meta is None: meta = {}
+      if 'user' not in meta: meta['user'] = self.PGLOG['CURUID']
+      if 'group' not in meta: meta['group'] = self.PGLOG['GDEXGRP']
+      uinfo = json.dumps(meta)
+      ms = re.match(r'^(.+)/$', fromdir)
+      if ms: fromdir = ms.group(1)
+      ms = re.match(r'^(.+)/$', todir)
+      if ms: todir = ms.group(1)
+      # the same walk isd_s3_cli makes to pick the files: no symlinked directories
+      lfiles = {}
+      for path, dirs, files in os.walk(fromdir):
+         for fname in files:
+            lfile = op.join(path, fname)
+            lfiles[op.relpath(lfile, fromdir)] = op.getsize(lfile)
+      if not lfiles:
+         return self.lmsg(fromdir, "{} to copy to {}-{}".format(self.PGLOG['MISSFILE'], self.OHOST, todir), logact)
+      prefix = todir + '/'
+      if not logact&self.OVRIDE:
+         flist = self.object_glob(todir, bucket, 0, logact)
+         if flist == self.FAILURE: return self.FAILURE
+         miss = self.object_upload_missing(prefix, lfiles, flist)
+         if not miss:
+            self.pglog("{}-{}-{}: all {} file(s) exist already, skip uploading".format(
+                       self.OHOST, bucket, todir, len(lfiles)), logact&~self.EXITLG)
+            return self.SUCCESS
+      # -db too: an upload goes to the default bucket whatever -b says
+      cmd = "{} -db {} um -ld {} -b {} -kp {} -r -md '{}'".format(
+            self.OBJCTCMD, bucket, fromdir, bucket, todir, uinfo)
+      buf = self.pgsystem(cmd, logact, self.CMDBTH)
+      flist = self.object_glob(todir, bucket, 0, logact)
+      if flist == self.FAILURE: return self.FAILURE
+      miss = self.object_upload_missing(prefix, lfiles, flist)
+      if miss:
+         return self.errlog("{}-{}: Error upload {} of {} file(s) to {}-{}\n{}".format(
+                            fromdir, miss[0], len(miss), len(lfiles), self.OHOST, todir, buf), 'O', 1, logact)
+      self.pglog("{}: {} file(s) uploaded to {}-{}-{}".format(
+                 fromdir, len(lfiles), self.OHOST, bucket, todir), logact&~self.EXITLG)
+      return self.SUCCESS
+
+   # List the local files not uploaded under an object key prefix yet
+   #   prefix - target object key prefix, including the trailing '/'
+   #   lfiles - local file sizes, keyed by path relative to the source directory
+   #    flist - object_glob() info, keyed by object key
+   def object_upload_missing(self, prefix, lfiles, flist):
+      """Return the local files that have no object of the same size under a key prefix.
+
+      Args:
+         prefix (str): Object key prefix including the trailing '/'.
+         lfiles (dict): Mapping of relative path → local file size.
+         flist (dict): Mapping of key → object info from object_glob().
+
+      Returns:
+         list: The relative paths still to upload.
+      """
+      miss = []
+      for name in lfiles:
+         oinfo = flist.get(prefix + name)
+         if not (oinfo and oinfo['data_size'] == lfiles[name]): miss.append(name)
+      return miss
+
    # Copy multiple files from a Globus endpoint to another
    #   tofiles - target file name list, echo name leading with /dsnnn.n/ on Quasar and
    #             leading with /data/ or /decsdata/ on local glade disk

{rda_python_common-3.0.27 → rda_python_common-3.0.29}/src/rda_python_common/pg_sig.py

@@ -52,7 +52,11 @@ class PgSIG(PgDBI):
          'MPROC': 1,    # default number of multiple processes
          'BPROC': 1,    # default number of multiple background processes
          'ETIME': 20,   # default error waiting time (in seconds)
-         'WTIME': 120,  # default waiting time (in seconds)
+         'WTIME': 120,  # default waiting time (in seconds), the cap of an adaptive wait
+         'WTMIN': 5,    # shortest adaptive wait (in seconds) for child/background processes
+         'WTINC': 5,    # seconds added to the adaptive wait after each loop still waiting
+         'CWTIME': 0,   # current adaptive wait (in seconds), 0 before the first wait
+         'WLOOP': -1,   # loop count of the last adaptive wait sleep
          'DTIME': 600,  # the daemon record refresh time (in seconds)
          'RTIME': 2400, # the web rda config unlocking and unconfigured system down waiting time (in seconds)
          'CTIME': 4800, # the lock cleaning & configued system down waiting time (in seconds)
@@ -1281,21 +1285,52 @@ class PgSIG(PgDBI):
       if wtime: time.sleep(wtime)
       return wtime
 
-   # show wait message every dintv and then sleep for PGSIG['WTIME']
+   # show wait message every 30 loops and then sleep for an adaptive wait time
    def show_wait_message(self, loop, msg, logact=None, dowait=0):
       """Log a wait message every 30 loops and optionally sleep.
 
       Args:
-         loop (int): Current loop iteration count.
+         loop (int): Current loop iteration count, 0 for the first sleep of a wait.
          msg (str): Message to log.
          logact (int, optional): Log action flags. Defaults to ``LOGWRN``.
-         dowait (int, optional): If non-zero, sleep for ``PGSIG['WTIME']``
-            seconds. Defaults to 0.
+         dowait (int, optional): If non-zero, sleep for the adaptive wait time
+            from ``adaptive_wait_time()``, at most ``PGSIG['WTIME']`` seconds.
+            Defaults to 0.
       """
       if logact is None: logact = self.LOGWRN
       if loop > 0 and (loop%30) == 0:
          self.pglog("{} at {}".format(msg, self.current_datetime()), logact)
-      if dowait: time.sleep(self.PGSIG['WTIME'])
+      if dowait: time.sleep(self.adaptive_wait_time(loop))
+
+   # get the adaptive time to sleep for one loop of a wait for processes
+   def adaptive_wait_time(self, loop):
+      """Return how long to sleep for one loop of a wait for child/background processes.
+
+      The wait starts at ``PGSIG['WTMIN']`` seconds and grows by ``PGSIG['WTINC']``
+      after each loop that still finds no process done, capped at the polling time
+      ``PGSIG['WTIME']``, so long-running processes are not checked too often while
+      short ones are not waited on for the full polling time.  The current wait
+      carries over to the next wait; it goes back to ``WTMIN`` when the previous
+      wait was over after its first sleep, since the processes are short.  Set
+      ``PGSIG['WTMIN']`` to ``WTIME`` or larger for the old fixed wait.
+
+      Args:
+         loop (int): Loop count of the current wait, 0 for its first sleep.
+
+      Returns:
+         int: Seconds to sleep.
+      """
+      wmax = self.PGSIG['WTIME']
+      wmin = min(self.PGSIG['WTMIN'], wmax)
+      wtime = self.PGSIG['CWTIME']
+      if loop == 0:
+         if not wtime or self.PGSIG['WLOOP'] == 0: wtime = wmin
+      else:
+         wtime += self.PGSIG['WTINC']
+      wtime = max(wmin, min(wtime, wmax))
+      self.PGSIG['CWTIME'] = wtime
+      self.PGSIG['WLOOP'] = loop
+      return wtime
 
    # register a time out function to raise a time out error
    @contextmanager

{rda_python_common-3.0.27 → rda_python_common-3.0.29/src/rda_python_common.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rda_python_common
-Version: 3.0.27
+Version: 3.0.29
 Summary: RDA Python common library codes shared by other RDA python packages
 Author-email: Zaihua Ji <zji@ucar.edu>
 Project-URL: Homepage, https://github.com/NCAR/rda-python-common
@@ -189,7 +189,7 @@ PgLOG.pglog("hello", PgLOG.LOGWRN)
 python -c "import rda_python_common; print(rda_python_common.__version__)"
 ```
 
-You should see the installed version (currently `3.0.27`).  If the import
+You should see the installed version (currently `3.0.29`).  If the import
 fails, double-check that the active Python environment is the one where you
 ran `pip install`.