progressbar cleanup

dvc/progress.py

@@ -5,6 +5,8 @@
 from copy import deepcopy
 from concurrent.futures import ThreadPoolExecutor
 
+logger = logging.getLogger(__name__)
+
 
 class TqdmThreadPoolExecutor(ThreadPoolExecutor):
     """
@@ -33,43 +35,72 @@ class Tqdm(tqdm):
     maximum-compatibility tqdm-based progressbars
     """
 
+    BAR_FMT_DEFAULT = (
+        "{percentage:3.0f}%|{bar:10}|{desc} {bar:-10b}{n}/{total}"
+        " [{elapsed}<{remaining}, {rate_fmt:>11}{postfix}]"
+    )
+    BAR_FMT_NOTOTAL = (
+        "{desc} {bar:b}{n} [{elapsed}<??:??, {rate_fmt:>11}{postfix}]"
+    )
+
     def __init__(
         self,
         iterable=None,
         disable=None,
-        bytes=False,  # pylint: disable=W0622
+        level=logging.ERROR,
+        desc=None,
         desc_truncate=None,
         leave=None,
+        level_leave=logging.DEBUG,
+        bar_format=None,
+        bytes=False,  # pylint: disable=W0622
         **kwargs
     ):
         """
         bytes   : shortcut for
             `unit='B', unit_scale=True, unit_divisor=1024, miniters=1`
-        desc_truncate  : like `desc` but will truncate to 10 chars
+        desc  : persists after `close()`
+        desc_truncate  : like `desc` but will `truncate()` and not persist
+        level  : effective logging level for determining `disable`;
+            used only if `disable` is unspecified
+        level_leave  : effective logging level for determining `leave`;
+            used only if `leave` is unspecified
         kwargs  : anything accepted by `tqdm.tqdm()`
         """
         kwargs = deepcopy(kwargs)
+        kwargs.setdefault("unit_scale", True)
         if bytes:
             for k, v in dict(
                 unit="B", unit_scale=True, unit_divisor=1024, miniters=1
             ).items():
                 kwargs.setdefault(k, v)
+        if desc is not None:
+            self.desc_persist = desc
         if desc_truncate is not None:
-            kwargs.setdefault("desc", self.truncate(desc_truncate))
+            desc = self.truncate(desc_truncate)
         if disable is None:
-            disable = (
-                logging.getLogger(__name__).getEffectiveLevel()
-                >= logging.CRITICAL
-            )
+            disable = logger.getEffectiveLevel() > level
+        if leave is None:
+            leave = logger.getEffectiveLevel() <= level_leave
+        if bar_format is None:
+            if kwargs.get("total", hasattr(iterable, "__len__")):
+                bar_format = self.BAR_FMT_DEFAULT
+            else:
+                bar_format = self.BAR_FMT_NOTOTAL
         super(Tqdm, self).__init__(
-            iterable=iterable, disable=disable, leave=leave, **kwargs
+            iterable=iterable,
+            disable=disable,
+            leave=leave,
+            desc=desc,
+            bar_format=bar_format,
+            **kwargs
         )
 
     def update_desc(self, desc, n=1, truncate=True):
         """
-        Calls `set_description(truncate(desc))` and `update(n)`
+        Calls `set_description_str(truncate(desc))` and `update(n)`
         """
-        self.set_description(
+        self.set_description_str(
             self.truncate(desc) if truncate else desc, refresh=False
         )
         self.update(n)
@@ -79,6 +110,11 @@ def update_to(self, current, total=None):
             self.total = total  # pylint: disable=W0613,W0201
         self.update(current - self.n)
 
+    def close(self):
+        if hasattr(self, "desc_persist"):
+            self.set_description_str(self.desc_persist, refresh=False)
+        super(Tqdm, self).close()
+
     @classmethod
     def truncate(cls, s, max_len=25, end=True, fill="..."):
         """

dvc/remote/base.py

@@ -638,7 +638,9 @@ def cache_exists(self, checksums, jobs=None):
         if not self.no_traverse:
             return list(set(checksums) & set(self.all()))
 
-        with Tqdm(total=len(checksums), unit="md5") as pbar:
+        with Tqdm(
+            desc="Querying remote cache", total=len(checksums), unit="md5"
+        ) as pbar:
 
             def exists_with_progress(path_info):
                 ret = self.exists(path_info)

dvc/remote/local/__init__.py

@@ -254,7 +254,9 @@ def move(self, from_info, to_info):
     def cache_exists(self, checksums, jobs=None):
         return [
             checksum
-            for checksum in Tqdm(checksums, unit="md5")
+            for checksum in Tqdm(
+                checksums, unit="md5", desc="Querying local cache"
+            )
             if not self.changed_cache_file(checksum)
         ]
 
@@ -313,13 +315,13 @@ def status(
         show_checksums=False,
         download=False,
     ):
-        logger.info(
+        logger.debug(
             "Preparing to collect status from {}".format(remote.path_info)
         )
         ret = self._group(checksum_infos, show_checksums=show_checksums) or {}
         md5s = list(ret)
 
-        logger.info("Collecting information from local cache...")
+        logger.debug("Collecting information from local cache...")
         local_exists = self.cache_exists(md5s, jobs=jobs)
 
         # This is a performance optimization. We can safely assume that,
@@ -329,7 +331,7 @@ def status(
         if download and sorted(local_exists) == sorted(md5s):
             remote_exists = local_exists
         else:
-            logger.info("Collecting information from remote cache...")
+            logger.debug("Collecting information from remote cache...")
             remote_exists = list(remote.cache_exists(md5s, jobs=jobs))
 
         self._fill_statuses(ret, local_exists, remote_exists)
@@ -377,7 +379,7 @@ def _process(
         show_checksums=False,
         download=False,
     ):
-        logger.info(
+        logger.debug(
             "Preparing to {} '{}'".format(
                 "download data from" if download else "upload data to",
                 remote.path_info,

dvc/remote/ssh/__init__.py

@@ -267,7 +267,9 @@ def cache_exists(self, checksums, jobs=None):
         if not self.no_traverse:
             return list(set(checksums) & set(self.all()))
 
-        with Tqdm(total=len(checksums), unit="md5") as pbar:
+        with Tqdm(
+            desc="Querying remote cache", total=len(checksums), unit="md5"
+        ) as pbar:
 
             def exists_with_progress(chunks):
                 return self.batch_exists(chunks, callback=pbar.update_desc)
@@ -278,5 +280,4 @@ def exists_with_progress(chunks):
                 results = executor.map(exists_with_progress, chunks)
                 in_remote = itertools.chain.from_iterable(results)
                 ret = list(itertools.compress(checksums, in_remote))
-                pbar.update_desc("", 0)  # clear path name description
                 return ret

dvc/repo/checkout.py

@@ -38,8 +38,14 @@ def checkout(self, target=None, with_deps=False, force=False, recursive=False):
     with self.state:
         _cleanup_unused_links(self, all_stages)
         total = get_all_files_numbers(stages)
+        if total == 0:
+            logger.info("Nothing to do")
         with Tqdm(
-            total=total, unit="file", desc="Checkout", disable=total == 0
+            total=total,
+            unit="file",
+            desc="Checkout",
+            disable=total == 0,
+            level_leave=logging.INFO,
         ) as pbar:
             for stage in stages:
                 if stage.locked:
@@ -51,4 +57,3 @@ def checkout(self, target=None, with_deps=False, force=False, recursive=False):
                     )
 
                 stage.checkout(force=force, progress_callback=pbar.update_desc)
-            pbar.update_desc("Checkout", 0)  # clear path name description

setup.py

@@ -61,7 +61,7 @@ def run(self):
     "funcy>=1.12",
     "pathspec>=0.5.9",
     "shortuuid>=0.5.0",
-    "tqdm>=4.34.0",
+    "tqdm>=4.35.0",
     "win-unicode-console>=0.5; sys_platform == 'win32'",
 ]