Merge pull request #27 from rich-iannone/fix-rows-distinct-visual-display

rich-iannone · web-flow · commit 6de30f212bd2 · 2025-01-11T11:06:38.000-05:00
fix: improve visual display of `rows_distinct()` steps in validation report
diff --git a/pointblank/validate.py b/pointblank/validate.py
@@ -4933,6 +4933,13 @@ def get_tabular_report(
 
             if assertion_type[i] in ["col_schema_match", "row_count_match", "col_count_match"]:
                 columns_upd.append("&mdash;")
+            elif assertion_type[i] in ["rows_distinct"]:
+                if not column:
+                    # If there is no column subset, then all columns are used
+                    columns_upd.append("ALL COLUMNS")
+                else:
+                    # With a column subset list, format with commas between the column names
+                    columns_upd.append(", ".join(column))
             else:
                 columns_upd.append(str(column))
 
@@ -4981,8 +4988,13 @@ def get_tabular_report(
             elif assertion_type[i] in ["col_vals_in_set", "col_vals_not_in_set"]:
                 values_upd.append(str(value)[1:-1].replace("'", ""))
 
-            # If the assertion type checks for NULL or not NULL values, use an em dash
-            elif assertion_type[i] in ["col_vals_null", "col_vals_not_null", "col_exists"]:
+            # Certain assertion types don't have an associated value, so use an em dash for those
+            elif assertion_type[i] in [
+                "col_vals_null",
+                "col_vals_not_null",
+                "col_exists",
+                "rows_distinct",
+            ]:
                 values_upd.append("&mdash;")
 
             elif assertion_type[i] in ["col_schema_match"]:
diff --git a/tests/snapshots/test_validate/test_comprehensive_validation_report_html_snap/comprehensive_validation_report.html b/tests/snapshots/test_validate/test_comprehensive_validation_report_html_snap/comprehensive_validation_report.html
@@ -969,6 +969,90 @@
     <td style="height: 40px; background-color: #FCFCFC; border-right: 1px solid #D3D3D3;" class="gt_row gt_center"><span style="color: #439CFE;">&cir;</span></td>
     <td style="height: 40px;" class="gt_row gt_center">—</td>
   </tr>
+  <tr>
+    <td style="height: 40px; background-color: #FFBF00; color: transparent;font-size: 0px;" class="gt_row gt_left">#FFBF00</td>
+    <td style="height: 40px; color: #666666;font-size: 13px;font-weight: bold;" class="gt_row gt_right">23</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px;" class="gt_row gt_left">
+        <div style="margin:0;padding:0;display:inline-block;height:30px;vertical-align:middle;">
+        <!--?xml version="1.0" encoding="UTF-8"?--><?xml version="1.0" encoding="UTF-8"?>
+<svg width="30px" height="30px" viewBox="0 0 67 67" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
+    <title>rows_distinct</title>
+    <g id="All-Icons" stroke="none" stroke-width="1" fill="none" fill-rule="evenodd">
+        <g id="rows_distinct" transform="translate(0.000000, 0.482759)">
+            <path d="M56.712234,1 C59.1975153,1 61.4475153,2.00735931 63.076195,3.63603897 C64.7048747,5.26471863 65.712234,7.51471863 65.712234,10 L65.712234,10 L65.712234,65 L10.712234,65 C8.22695259,65 5.97695259,63.9926407 4.34827294,62.363961 C2.71959328,60.7352814 1.71223397,58.4852814 1.71223397,56 L1.71223397,56 L1.71223397,10 C1.71223397,7.51471863 2.71959328,5.26471863 4.34827294,3.63603897 C5.97695259,2.00735931 8.22695259,1 10.712234,1 L10.712234,1 Z" id="rectangle" stroke="#000000" stroke-width="2" fill="#FFFFFF"></path>
+            <g id="no_gemini" transform="translate(17.000000, 13.000000)">
+                <path d="M3.66705619,6.62107508 C3.12510104,6.64066386 2.67455974,7.04767444 2.60273432,7.58527682 C2.52873228,8.1228792 2.85303526,8.6343627 3.37104848,8.79760239 C4.40489909,9.15455286 6.70113553,9.87063021 9.86580595,10.364702 L9.86580595,30.7369976 C6.65978137,31.2332458 4.37225104,31.964559 3.37104848,32.3215095 C2.78991555,32.5282797 2.48520173,33.1681784 2.69197196,33.7493114 C2.8987422,34.3304443 3.53864094,34.6351581 4.11977387,34.4283879 C5.54975217,33.9190808 10.2031678,32.4608072 16.5172734,32.4608072 C22.7943781,32.4608072 27.5500901,33.8907855 29.0540706,34.4109757 C29.6352036,34.6133926 30.2707495,34.304326 30.4731664,33.7231931 C30.6755833,33.1420601 30.3665167,32.5065142 29.7853838,32.3040973 C28.7493568,31.9449704 26.4313554,31.2441283 23.2383897,30.7544098 L23.2383897,10.3821143 C26.4444143,9.88804243 28.745004,9.16108259 29.7679716,8.79760239 C30.3491045,8.59083215 30.6538184,7.95093341 30.4470481,7.36980048 C30.2402779,6.78866755 29.6003791,6.48395373 29.0192462,6.69072396 C27.5587963,7.20873774 22.9162637,8.65830464 16.6391589,8.65830464 C10.3707603,8.65830464 5.61287188,7.21309051 4.10236166,6.69072396 C3.96306391,6.6384873 3.81506005,6.61454548 3.66705619,6.62107508 Z M12.0945699,10.6432975 C13.4940771,10.789125 15.0198226,10.8870686 16.6391589,10.8870686 C18.1997289,10.8870686 19.6623552,10.7978311 21.0096257,10.6607098 L21.0096257,30.4584021 C19.623178,30.316928 18.1191975,30.2320433 16.5172734,30.2320433 C14.9327615,30.2320433 13.4570763,30.3191044 12.0945699,30.4584021 L12.0945699,10.6432975 Z" id="gemini" stroke="#000000" stroke-width="2" fill="#000000" fill-rule="nonzero"></path>
+                <path d="M16.6605354,-5.05929499 C16.9366778,-5.05929499 17.1605354,-4.83543737 17.1605354,-4.55929499 L17.1605354,44.5687509 C17.1605354,44.8448933 16.9366778,45.0687509 16.6605354,45.0687509 C16.384393,45.0687509 16.1605354,44.8448933 16.1605354,44.5687509 L16.1605354,-4.55929499 C16.1605354,-4.83543737 16.384393,-5.05929499 16.6605354,-5.05929499 Z" id="line_black" fill="#000000" transform="translate(16.660535, 20.004728) rotate(-320.000000) translate(-16.660535, -20.004728) "></path>
+                <polygon id="line_white" fill="#FFFFFF" transform="translate(18.560032, 19.158031) rotate(-320.000000) translate(-18.560032, -19.158031) " points="18.0600316 -4.45366735 19.0600316 -4.45366735 19.0600316 42.7697299 18.0600316 42.7697299"></polygon>
+            </g>
+        </g>
+    </g>
+</svg>
+        </div>
+        <span style="font-family: 'IBM Plex Mono', monospace, courier; color: black; font-size:11px;"> rows_distinct()</span>
+        </td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5; white-space: nowrap; text-overflow: ellipsis; overflow: hidden;" class="gt_row gt_left">ALL COLUMNS</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5; white-space: nowrap; text-overflow: ellipsis; overflow: hidden;" class="gt_row gt_left">&mdash;</td>
+    <td style="height: 40px; background-color: #FCFCFC; border-left: 1px solid #D3D3D3;" class="gt_row gt_center"><svg width="25px" height="25px" viewBox="0 0 25 25" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" style="vertical-align: middle;">
+    <g id="unchanged" stroke="none" stroke-width="1" fill="none" fill-rule="evenodd">
+        <g id="unchanged" transform="translate(0.500000, 0.570147)">
+            <rect id="Rectangle" x="0.125132506" y="0" width="23.749735" height="23.7894737"></rect>
+            <path d="M5.80375046,8.18194736 C3.77191832,8.18194736 2.11875046,9.83495328 2.11875046,11.8669474 C2.11875046,13.8989414 3.77191832,15.5519474 5.80375046,15.5519474 C7.8355826,15.5519474 9.48875046,13.8989414 9.48875046,11.8669474 C9.48875046,9.83495328 7.83552863,8.18194736 5.80375046,8.18194736 Z M5.80375046,14.814915 C4.17821997,14.814915 2.85578285,13.4924778 2.85578285,11.8669474 C2.85578285,10.2414169 4.17821997,8.91897975 5.80375046,8.91897975 C7.42928095,8.91897975 8.75171807,10.2414169 8.75171807,11.8669474 C8.75171807,13.4924778 7.42928095,14.814915 5.80375046,14.814915 Z" id="Shape" fill="#000000" fill-rule="nonzero"></path>
+            <path d="M13.9638189,8.699335 C13.9364621,8.70430925 13.9091059,8.71176968 13.8842359,8.71923074 C13.7822704,8.73663967 13.6877654,8.77643115 13.6056956,8.83860518 L10.2433156,11.3852598 C10.0766886,11.5046343 9.97720993,11.6986181 9.97720993,11.9025491 C9.97720993,12.1064807 10.0766886,12.3004639 10.2433156,12.4198383 L13.6056956,14.966493 C13.891697,15.1803725 14.2970729,15.1231721 14.5109517,14.8371707 C14.7248313,14.5511692 14.6676309,14.145794 14.3816294,13.9319145 L12.5313257,12.5392127 L21.8812495,12.5392127 L21.8812495,11.2658854 L12.5313257,11.2658854 L14.3816294,9.87318364 C14.6377872,9.71650453 14.7497006,9.40066014 14.6477351,9.11714553 C14.5482564,8.83363156 14.262255,8.65954352 13.9638189,8.699335 Z" id="arrow" fill="#000000" transform="translate(15.929230, 11.894737) rotate(-180.000000) translate(-15.929230, -11.894737) "></path>
+        </g>
+    </g>
+</svg></td>
+    <td style="height: 40px; background-color: #FCFCFC; border-right: 1px solid #D3D3D3;" class="gt_row gt_center"><span style="color:#4CA64C;">&check;</span></td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px;" class="gt_row gt_right">13</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5;" class="gt_row gt_right">11<br />0.85</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5;" class="gt_row gt_right">2<br />0.15</td>
+    <td style="height: 40px; background-color: #FCFCFC; border-left: 1px solid #D3D3D3;" class="gt_row gt_center"><span style="color: #E5AB00;">&#9679;</span></td>
+    <td style="height: 40px; background-color: #FCFCFC;" class="gt_row gt_center"><span style="color: #CF142B;">&cir;</span></td>
+    <td style="height: 40px; background-color: #FCFCFC; border-right: 1px solid #D3D3D3;" class="gt_row gt_center"><span style="color: #439CFE;">&cir;</span></td>
+    <td style="height: 40px;" class="gt_row gt_center">—</td>
+  </tr>
+  <tr>
+    <td style="height: 40px; background-color: #FFBF00; color: transparent;font-size: 0px;" class="gt_row gt_left">#FFBF00</td>
+    <td style="height: 40px; color: #666666;font-size: 13px;font-weight: bold;" class="gt_row gt_right">24</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px;" class="gt_row gt_left">
+        <div style="margin:0;padding:0;display:inline-block;height:30px;vertical-align:middle;">
+        <!--?xml version="1.0" encoding="UTF-8"?--><?xml version="1.0" encoding="UTF-8"?>
+<svg width="30px" height="30px" viewBox="0 0 67 67" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
+    <title>rows_distinct</title>
+    <g id="All-Icons" stroke="none" stroke-width="1" fill="none" fill-rule="evenodd">
+        <g id="rows_distinct" transform="translate(0.000000, 0.482759)">
+            <path d="M56.712234,1 C59.1975153,1 61.4475153,2.00735931 63.076195,3.63603897 C64.7048747,5.26471863 65.712234,7.51471863 65.712234,10 L65.712234,10 L65.712234,65 L10.712234,65 C8.22695259,65 5.97695259,63.9926407 4.34827294,62.363961 C2.71959328,60.7352814 1.71223397,58.4852814 1.71223397,56 L1.71223397,56 L1.71223397,10 C1.71223397,7.51471863 2.71959328,5.26471863 4.34827294,3.63603897 C5.97695259,2.00735931 8.22695259,1 10.712234,1 L10.712234,1 Z" id="rectangle" stroke="#000000" stroke-width="2" fill="#FFFFFF"></path>
+            <g id="no_gemini" transform="translate(17.000000, 13.000000)">
+                <path d="M3.66705619,6.62107508 C3.12510104,6.64066386 2.67455974,7.04767444 2.60273432,7.58527682 C2.52873228,8.1228792 2.85303526,8.6343627 3.37104848,8.79760239 C4.40489909,9.15455286 6.70113553,9.87063021 9.86580595,10.364702 L9.86580595,30.7369976 C6.65978137,31.2332458 4.37225104,31.964559 3.37104848,32.3215095 C2.78991555,32.5282797 2.48520173,33.1681784 2.69197196,33.7493114 C2.8987422,34.3304443 3.53864094,34.6351581 4.11977387,34.4283879 C5.54975217,33.9190808 10.2031678,32.4608072 16.5172734,32.4608072 C22.7943781,32.4608072 27.5500901,33.8907855 29.0540706,34.4109757 C29.6352036,34.6133926 30.2707495,34.304326 30.4731664,33.7231931 C30.6755833,33.1420601 30.3665167,32.5065142 29.7853838,32.3040973 C28.7493568,31.9449704 26.4313554,31.2441283 23.2383897,30.7544098 L23.2383897,10.3821143 C26.4444143,9.88804243 28.745004,9.16108259 29.7679716,8.79760239 C30.3491045,8.59083215 30.6538184,7.95093341 30.4470481,7.36980048 C30.2402779,6.78866755 29.6003791,6.48395373 29.0192462,6.69072396 C27.5587963,7.20873774 22.9162637,8.65830464 16.6391589,8.65830464 C10.3707603,8.65830464 5.61287188,7.21309051 4.10236166,6.69072396 C3.96306391,6.6384873 3.81506005,6.61454548 3.66705619,6.62107508 Z M12.0945699,10.6432975 C13.4940771,10.789125 15.0198226,10.8870686 16.6391589,10.8870686 C18.1997289,10.8870686 19.6623552,10.7978311 21.0096257,10.6607098 L21.0096257,30.4584021 C19.623178,30.316928 18.1191975,30.2320433 16.5172734,30.2320433 C14.9327615,30.2320433 13.4570763,30.3191044 12.0945699,30.4584021 L12.0945699,10.6432975 Z" id="gemini" stroke="#000000" stroke-width="2" fill="#000000" fill-rule="nonzero"></path>
+                <path d="M16.6605354,-5.05929499 C16.9366778,-5.05929499 17.1605354,-4.83543737 17.1605354,-4.55929499 L17.1605354,44.5687509 C17.1605354,44.8448933 16.9366778,45.0687509 16.6605354,45.0687509 C16.384393,45.0687509 16.1605354,44.8448933 16.1605354,44.5687509 L16.1605354,-4.55929499 C16.1605354,-4.83543737 16.384393,-5.05929499 16.6605354,-5.05929499 Z" id="line_black" fill="#000000" transform="translate(16.660535, 20.004728) rotate(-320.000000) translate(-16.660535, -20.004728) "></path>
+                <polygon id="line_white" fill="#FFFFFF" transform="translate(18.560032, 19.158031) rotate(-320.000000) translate(-18.560032, -19.158031) " points="18.0600316 -4.45366735 19.0600316 -4.45366735 19.0600316 42.7697299 18.0600316 42.7697299"></polygon>
+            </g>
+        </g>
+    </g>
+</svg>
+        </div>
+        <span style="font-family: 'IBM Plex Mono', monospace, courier; color: black; font-size:11px;"> rows_distinct()</span>
+        </td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5; white-space: nowrap; text-overflow: ellipsis; overflow: hidden;" class="gt_row gt_left">a, b, c</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5; white-space: nowrap; text-overflow: ellipsis; overflow: hidden;" class="gt_row gt_left">&mdash;</td>
+    <td style="height: 40px; background-color: #FCFCFC; border-left: 1px solid #D3D3D3;" class="gt_row gt_center"><svg width="25px" height="25px" viewBox="0 0 25 25" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" style="vertical-align: middle;">
+    <g id="unchanged" stroke="none" stroke-width="1" fill="none" fill-rule="evenodd">
+        <g id="unchanged" transform="translate(0.500000, 0.570147)">
+            <rect id="Rectangle" x="0.125132506" y="0" width="23.749735" height="23.7894737"></rect>
+            <path d="M5.80375046,8.18194736 C3.77191832,8.18194736 2.11875046,9.83495328 2.11875046,11.8669474 C2.11875046,13.8989414 3.77191832,15.5519474 5.80375046,15.5519474 C7.8355826,15.5519474 9.48875046,13.8989414 9.48875046,11.8669474 C9.48875046,9.83495328 7.83552863,8.18194736 5.80375046,8.18194736 Z M5.80375046,14.814915 C4.17821997,14.814915 2.85578285,13.4924778 2.85578285,11.8669474 C2.85578285,10.2414169 4.17821997,8.91897975 5.80375046,8.91897975 C7.42928095,8.91897975 8.75171807,10.2414169 8.75171807,11.8669474 C8.75171807,13.4924778 7.42928095,14.814915 5.80375046,14.814915 Z" id="Shape" fill="#000000" fill-rule="nonzero"></path>
+            <path d="M13.9638189,8.699335 C13.9364621,8.70430925 13.9091059,8.71176968 13.8842359,8.71923074 C13.7822704,8.73663967 13.6877654,8.77643115 13.6056956,8.83860518 L10.2433156,11.3852598 C10.0766886,11.5046343 9.97720993,11.6986181 9.97720993,11.9025491 C9.97720993,12.1064807 10.0766886,12.3004639 10.2433156,12.4198383 L13.6056956,14.966493 C13.891697,15.1803725 14.2970729,15.1231721 14.5109517,14.8371707 C14.7248313,14.5511692 14.6676309,14.145794 14.3816294,13.9319145 L12.5313257,12.5392127 L21.8812495,12.5392127 L21.8812495,11.2658854 L12.5313257,11.2658854 L14.3816294,9.87318364 C14.6377872,9.71650453 14.7497006,9.40066014 14.6477351,9.11714553 C14.5482564,8.83363156 14.262255,8.65954352 13.9638189,8.699335 Z" id="arrow" fill="#000000" transform="translate(15.929230, 11.894737) rotate(-180.000000) translate(-15.929230, -11.894737) "></path>
+        </g>
+    </g>
+</svg></td>
+    <td style="height: 40px; background-color: #FCFCFC; border-right: 1px solid #D3D3D3;" class="gt_row gt_center"><span style="color:#4CA64C;">&check;</span></td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px;" class="gt_row gt_right">13</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5;" class="gt_row gt_right">11<br />0.85</td>
+    <td style="height: 40px; color: black;font-family: IBM Plex Mono;font-size: 11px; border-left: 1px dashed #E5E5E5;" class="gt_row gt_right">2<br />0.15</td>
+    <td style="height: 40px; background-color: #FCFCFC; border-left: 1px solid #D3D3D3;" class="gt_row gt_center"><span style="color: #E5AB00;">&#9679;</span></td>
+    <td style="height: 40px; background-color: #FCFCFC;" class="gt_row gt_center"><span style="color: #CF142B;">&cir;</span></td>
+    <td style="height: 40px; background-color: #FCFCFC; border-right: 1px solid #D3D3D3;" class="gt_row gt_center"><span style="color: #439CFE;">&cir;</span></td>
+    <td style="height: 40px;" class="gt_row gt_center">—</td>
+  </tr>
 </tbody>
   
 
diff --git a/tests/test_validate.py b/tests/test_validate.py
@@ -4320,6 +4320,8 @@ def test_comprehensive_validation_report_html_snap(snapshot):
         .row_count_match(count=2, inverse=True)
         .col_count_match(count=8)
         .col_count_match(count=2, inverse=True)
+        .rows_distinct()
+        .rows_distinct(columns_subset=["a", "b", "c"])
         .interrogate()
     )
 

Original file line number	Diff line number	Diff line change
`@@ -4320,6 +4320,8 @@ def test_comprehensive_validation_report_html_snap(snapshot):`
`4320`	`4320`	`.row_count_match(count=2, inverse=True)`
`4321`	`4321`	`.col_count_match(count=8)`
`4322`	`4322`	`.col_count_match(count=2, inverse=True)`
	`4323`	`+ .rows_distinct()`
	`4324`	`+ .rows_distinct(columns_subset=["a", "b", "c"])`
`4323`	`4325`	`.interrogate()`
`4324`	`4326`	`)`
`4325`	`4327`