Supporting arbitrary expressions

panoramix/migrations/versions/1e2841a4128_.py

@@ -0,0 +1,26 @@
+"""empty message
+
+Revision ID: 1e2841a4128
+Revises: 5a7bad26f2a7
+Create Date: 2015-10-05 22:11:00.537054
+
+"""
+
+# revision identifiers, used by Alembic.
+revision = '1e2841a4128'
+down_revision = '5a7bad26f2a7'
+
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects import mysql
+
+def upgrade():
+    ### commands auto generated by Alembic - please adjust! ###
+    op.add_column('table_columns', sa.Column('expression', sa.Text(), nullable=True))
+    ### end Alembic commands ###
+
+
+def downgrade():
+    ### commands auto generated by Alembic - please adjust! ###
+    op.drop_column('table_columns', 'expression')
+    ### end Alembic commands ###

panoramix/models.py

@@ -11,7 +11,8 @@
 from sqlalchemy import Table
 from sqlalchemy import create_engine, MetaData, desc, select, and_
 from sqlalchemy.orm import relationship
-from sqlalchemy.sql import table, literal_column, text
+from sqlalchemy.sql import table, literal_column, text, column
+from sqlalchemy.sql.elements import ColumnClause
 from flask import request
 
 from copy import deepcopy, copy
@@ -338,6 +339,7 @@ def query(
             inner_from_dttm=None, inner_to_dttm=None,
             extras=None):
 
+        cols = {col.column_name: col for col in self.columns}
         qry_start_dttm = datetime.now()
         if not self.main_dttm_col:
             raise Exception(
@@ -360,9 +362,25 @@ def query(
 
         if groupby:
             select_exprs = [literal_column(s) for s in groupby]
-            groupby_exprs = [literal_column(s) for s in groupby]
-            inner_groupby_exprs = [
-                literal_column(s).label('__' + s) for s in groupby]
+            select_exprs = []
+            groupby_exprs = []
+            inner_select_exprs = []
+            inner_groupby_exprs = []
+            for s in groupby:
+                col = cols[s]
+                expr = col.expression
+                if expr:
+                    outer = ColumnClause(expr, is_literal=True).label(s)
+                    inner = ColumnClause(expr, is_literal=True).label('__' + s)
+                else:
+                    outer = literal_column(s).label(s)
+                    inner = literal_column(s).label('__' + s)
+
+                groupby_exprs.append(outer)
+                select_exprs.append(outer)
+                inner_groupby_exprs.append(inner)
+                inner_select_exprs.append(inner)
+
         if granularity != "all":
             select_exprs += [timestamp]
             groupby_exprs += [timestamp]
@@ -384,9 +402,14 @@ def query(
 
         where_clause_and = []
         for col, op, eq in filter:
+            col_obj = cols[col]
             if op in ('in', 'not in'):
                 values = eq.split(",")
-                cond = literal_column(col).in_(values)
+                if col_obj.expression:
+                    cond = ColumnClause(
+                        col_obj.expression, is_literal=True).in_(values)
+                else:
+                    cond = literal_column(col).in_(values)
                 if op == 'not in':
                     cond = ~cond
                 where_clause_and.append(cond)
@@ -397,23 +420,24 @@ def query(
         qry = qry.limit(row_limit)
 
         if timeseries_limit and groupby:
-            subq = select(inner_groupby_exprs)
+            subq = select(inner_select_exprs)
             subq = subq.select_from(table(self.table_name))
             subq = subq.where(and_(*(where_clause_and + inner_time_filter)))
             subq = subq.group_by(*inner_groupby_exprs)
             subq = subq.order_by(desc(main_metric_expr))
             subq = subq.limit(timeseries_limit)
             on_clause = []
-            for gb in groupby:
+            for i, gb in enumerate(groupby):
                 on_clause.append(
-                    literal_column(gb) == literal_column("__" + gb))
+                    groupby_exprs[i] == literal_column("__" + gb))
 
             from_clause = from_clause.join(subq.alias(), and_(*on_clause))
 
         qry = qry.select_from(from_clause)
 
         engine = self.database.get_sqla_engine()
         sql = str(qry.compile(engine, compile_kwargs={"literal_binds": True}))
+        print sql
         df = read_sql_query(
             sql=sql,
             con=engine
@@ -548,6 +572,7 @@ class TableColumn(Model, AuditMixinNullable):
     max = Column(Boolean, default=False)
     min = Column(Boolean, default=False)
     filterable = Column(Boolean, default=False)
+    expression = Column(Text, default='')
     description = Column(Text, default='')
 
     def __repr__(self):

panoramix/static/widgets/viz_nvd3.js

@@ -58,6 +58,8 @@ function viz_nvd3(token_name, json_callback) {
         } else if (viz_type === 'dist_bar') {
           var chart = nv.models.multiBarChart()
             .showControls(true)   //Allow user to switch between 'Grouped' and 'Stacked' mode.
+            .reduceXTicks(false)
+            .rotateLabels(45)
             .groupSpacing(0.1);   //Distance between each group of bars.
           chart.xAxis
             .showMaxMin(false);

panoramix/views.py

@@ -42,7 +42,8 @@ class TableColumnInlineView(CompactCRUDMixin, PanoramixModelView):
     can_delete = False
     edit_columns = [
         'column_name', 'description', 'groupby', 'filterable', 'table',
-        'count_distinct', 'sum', 'min', 'max']
+        'count_distinct', 'sum', 'min', 'max', 'expression']
+    add_columns = edit_columns
     list_columns = [
         'column_name', 'type', 'groupby', 'filterable', 'count_distinct',
         'sum', 'min', 'max']