Do not pipe into `grep -q`, because that stops reading as soon as a

[arvados.git] / tools / crunchstat-summary / crunchstat_summary / reader.py

diff --git a/tools/crunchstat-summary/crunchstat_summary/reader.py b/tools/crunchstat-summary/crunchstat_summary/reader.py
index 98dda673d5a3ab70d65ab1d3989b49f539959b69..8ccdbc2fcf04e45ca3ab3ec6e2270933d050ea1c 100644 (file)
--- a/tools/crunchstat-summary/crunchstat_summary/reader.py
+++ b/tools/crunchstat-summary/crunchstat_summary/reader.py
@@ -2,12 +2,10 @@
 #
 # SPDX-License-Identifier: AGPL-3.0
 
-from __future__ import print_function
-
 import arvados
-import Queue
+import itertools
+import queue
 import threading
-import _strptime
 
 from crunchstat_summary import logger
 
@@ -16,7 +14,7 @@ class CollectionReader(object):
     def __init__(self, collection_id):
         self._collection_id = collection_id
         self._label = collection_id
-        self._reader = None
+        self._readers = []
 
     def __str__(self):
         return self._label
@@ -25,21 +23,25 @@ class CollectionReader(object):
         logger.debug('load collection %s', self._collection_id)
         collection = arvados.collection.CollectionReader(self._collection_id)
         filenames = [filename for filename in collection]
-        if len(filenames) == 1:
-            filename = filenames[0]
-        else:
-            filename = 'crunchstat.txt'
-        self._label = "{}/{}".format(self._collection_id, filename)
-        self._reader = collection.open(filename)
-        return iter(self._reader)
+        # Crunch2 has multiple stats files
+        if len(filenames) > 1:
+            filenames = ['crunchstat.txt', 'arv-mount.txt']
+        for filename in filenames:
+            try:
+                self._readers.append(collection.open(filename))
+            except IOError:
+                logger.warn('Unable to open %s', filename)
+        self._label = "{}/{}".format(self._collection_id, filenames[0])
+        return itertools.chain(*[iter(reader) for reader in self._readers])
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
-        if self._reader:
-            self._reader.close()
-            self._reader = None
+        if self._readers:
+            for reader in self._readers:
+                reader.close()
+            self._readers = []
 
 
 class LiveLogReader(object):
@@ -83,19 +85,21 @@ class LiveLogReader(object):
             self._queue.put(self.EOF)
 
     def __iter__(self):
-        self._queue = Queue.Queue()
+        self._queue = queue.Queue()
         self._thread = threading.Thread(target=self._get_all_pages)
         self._thread.daemon = True
         self._thread.start()
         return self
 
-    def next(self):
+    def __next__(self):
         line = self._queue.get()
         if line is self.EOF:
             self._thread.join()
             raise StopIteration
         return line
 
+    next = __next__ # for Python 2
+
     def __enter__(self):
         return self