[beam] branch master updated: Fixed pcoll visualization in datatable

[pa...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

pabloem pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/beam.git


The following commit(s) were added to refs/heads/master by this push:
     new 8d8cd30  Fixed pcoll visualization in datatable
     new 13c77a8  Merge pull request #12442 from Fixed pcoll visualization in datatable
8d8cd30 is described below

commit 8d8cd30f50e01396823e818b7ca14d00114314c1
Author: Ning Kang <ni...@google.com>
AuthorDate: Fri Jul 31 15:43:12 2020 -0700

    Fixed pcoll visualization in datatable
    
    The datatable now displays some info when there is no data and renders
    datatable when data comes in for continuously updated visualization.
    
    The problem was:
    If the pcoll returns empty data for a streaming pipeline, it will cause
    the header of the datatable to container only the index column. Later
    when updating, if data is not empty, new rows are added but there are no
    columns to display those data.
---
 .../interactive/display/pcoll_visualization.py     | 24 +++++++++++++++-------
 1 file changed, 17 insertions(+), 7 deletions(-)

diff --git a/sdks/python/apache_beam/runners/interactive/display/pcoll_visualization.py b/sdks/python/apache_beam/runners/interactive/display/pcoll_visualization.py
index d11cb24..ce7ec0e 100644
--- a/sdks/python/apache_beam/runners/interactive/display/pcoll_visualization.py
+++ b/sdks/python/apache_beam/runners/interactive/display/pcoll_visualization.py
@@ -47,7 +47,6 @@ try:
   from IPython.core.display import Javascript  # pylint: disable=import-error
   from IPython.core.display import display  # pylint: disable=import-error
   from IPython.core.display import display_javascript  # pylint: disable=import-error
-  from IPython.core.display import update_display  # pylint: disable=import-error
   from facets_overview.generic_feature_statistics_generator import GenericFeatureStatisticsGenerator  # pylint: disable=import-error
   from timeloop import Timeloop  # pylint: disable=import-error
 
@@ -134,6 +133,10 @@ _DATAFRAME_PAGINATION_TEMPLATE = _CSS + """
             <script>
               {script_in_jquery_with_datatable}
             </script>"""
+_NO_DATA_TEMPLATE = _CSS + """
+            <div id="no_data_{id}">No data to display.</div>"""
+_NO_DATA_REMOVAL_SCRIPT = """
+            $("#no_data_{id}").remove();"""
 
 
 def visualize(
@@ -382,16 +385,23 @@ class PCollectionVisualization(object):
     if update and not update._is_datatable_empty:
       display_javascript(Javascript(script_in_jquery_with_datatable))
     else:
-      html = _DATAFRAME_PAGINATION_TEMPLATE.format(
-          table_id=table_id,
-          script_in_jquery_with_datatable=script_in_jquery_with_datatable)
+      if data.empty:
+        html = _NO_DATA_TEMPLATE.format(id=table_id)
+      else:
+        html = _DATAFRAME_PAGINATION_TEMPLATE.format(
+            table_id=table_id,
+            script_in_jquery_with_datatable=script_in_jquery_with_datatable)
       if update:
         if not data.empty:
-          # Re-initialize a datatable to replace the existing empty datatable.
-          update_display(HTML(html), display_id=update._df_display_id)
+          # Initialize a datatable to replace the existing no data div.
+          display(
+              Javascript(
+                  ie._JQUERY_WITH_DATATABLE_TEMPLATE.format(
+                      customized_script=_NO_DATA_REMOVAL_SCRIPT.format(
+                          id=table_id))))
+          display(HTML(html), display_id=update._df_display_id)
           update._is_datatable_empty = False
       else:
-        # Initialize a datatable for the first time rendering.
         display(HTML(html), display_id=self._df_display_id)
         if not data.empty:
           self._is_datatable_empty = False