chore(ci): Update table diff (#16415)

bbernays · web-flow · commit 9eaedb7d6a67 · 2024-02-01T15:17:12.000Z
#### Summary

This creates a special case where it will only log a message that says that all PKs have been replaced with `_cq_id` for a specific table...
diff --git a/scripts/table_diff/changes/changes.go b/scripts/table_diff/changes/changes.go
@@ -15,6 +15,9 @@ import (
 var (
 	columnRegex = regexp.MustCompile(`^\|(?P<name>.*)\|(?P<dataType>.*)\|`)
 	pkRegex     = regexp.MustCompile(`^The composite primary key for this table is \(([^)]+)\)\.`)
+
+	// There is a different message for single PKs
+	singlePKRegex = regexp.MustCompile(`^The primary key for this table is ([^)]+)\.`)
 )
 
 type change struct {
@@ -75,12 +78,17 @@ func parseColumnChange(line string) (name string, col column) {
 }
 
 func parsePKChange(line string) (names []string) {
-	match := pkRegex.FindStringSubmatch(line)
-	if len(match) != 2 {
-		return nil
+	matchMulti := pkRegex.FindStringSubmatch(line)
+	matchSingle := singlePKRegex.FindStringSubmatch(line)
+	if len(matchMulti) == 2 {
+		for _, part := range strings.Split(matchMulti[1], ", ") {
+			names = append(names, strings.Trim(part, "*"))
+		}
 	}
-	for _, part := range strings.Split(match[1], ", ") {
-		names = append(names, strings.Trim(part, "*"))
+	if len(matchSingle) == 2 {
+		for _, part := range strings.Split(matchSingle[1], ", ") {
+			names = append(names, strings.Trim(part, "*"))
+		}
 	}
 	return
 }
@@ -141,14 +149,14 @@ func getColumnChanges(file *gitdiff.File, table string) (changes []change) {
 			continue
 		}
 
-		if added.pk() && !deleted.pk() {
+		if added.pk() && !deleted.pk() && !(len(addedPK) == 1 && addedPK[0] == "_cq_id" && len(deletedPK) > 0) {
 			changes = append(changes, change{
 				Text:     fmt.Sprintf("Table %s: primary key constraint added to column %s", backtickStrings(table, name)...),
 				Breaking: true,
 			})
 		}
 
-		if !added.pk() && deleted.pk() {
+		if !added.pk() && deleted.pk() && !(len(addedPK) == 1 && addedPK[0] == "_cq_id" && len(deletedPK) > 0) {
 			changes = append(changes, change{
 				Text:     fmt.Sprintf("Table %s: primary key constraint removed from column %s", backtickStrings(table, name)...),
 				Breaking: true,
@@ -216,6 +224,14 @@ func getColumnChanges(file *gitdiff.File, table string) (changes []change) {
 			return chI.Text < chJ.Text
 		}
 	})
+
+	if len(addedPK) == 1 && addedPK[0] == "_cq_id" && len(deletedPK) > 0 {
+		changes = append(changes, change{
+			Text:     fmt.Sprintf("Table %s: all existing primary key constraints have been removed and a primary key new constraint has been added to `_cq_id`", backtickStrings(table)...),
+			Breaking: true,
+		})
+	}
+
 	return changes
 }
 
diff --git a/scripts/table_diff/changes/changes_test.go b/scripts/table_diff/changes/changes_test.go
@@ -261,6 +261,16 @@ func Test_getChanges(t *testing.T) {
 				},
 			},
 		},
+		{
+			name:         "Special Case - don't report removal of pks, only end state",
+			diffDataFile: "testdata/pr_100644_diff.txt",
+			wantChanges: []change{
+				{
+					Text:     "Table `aws_accessanalyzer_analyzers`: all existing primary key constraints have been removed and a primary key new constraint has been added to `_cq_id`",
+					Breaking: true,
+				},
+			},
+		},
 		{
 			name:         "Should handle no backticks -> backticks",
 			diffDataFile: "testdata/pr_11034_diff.txt",
diff --git a/scripts/table_diff/changes/testdata/pr_100644_diff.txt b/scripts/table_diff/changes/testdata/pr_100644_diff.txt
@@ -0,0 +1,29 @@
+diff --git a/plugins/source/aws/docs/tables/aws_accessanalyzer_analyzers.md b/plugins/source/aws/docs/tables/aws_accessanalyzer_analyzers.md
+index 76569347f50fb9..88070ce42a78d0 100644
+--- a/plugins/source/aws/docs/tables/aws_accessanalyzer_analyzers.md
++++ b/plugins/source/aws/docs/tables/aws_accessanalyzer_analyzers.md
+@@ -4,8 +4,8 @@ This table shows data for AWS Identity and Access Management (IAM) Access Analyz
+ 
+ https://docs.aws.amazon.com/access-analyzer/latest/APIReference/API_AnalyzerSummary.html
+ 
+-The primary key for this table is **arn**.
+-
++The primary key for this table is **_cq_id**.
++The following field is used to calculate the value of `_cq_id`: **arn**.
+ ## Relations
+ 
+ The following tables depend on aws_accessanalyzer_analyzers:
+@@ -16,11 +16,11 @@ The following tables depend on aws_accessanalyzer_analyzers:
+ 
+ | Name          | Type          |
+ | ------------- | ------------- |
+-|_cq_id|`uuid`|
++|_cq_id (PK)|`uuid`|
+ |_cq_parent_id|`uuid`|
+ |account_id|`utf8`|
+ |region|`utf8`|
+-|arn (PK)|`utf8`|
++|arn|`utf8`|
+ |created_at|`timestamp[us, tz=UTC]`|
+ |name|`utf8`|
+ |status|`utf8`|
