update stats in rules

Author: rekby

tools/greenplum-to-pg-tests/.gitignore

@@ -1,2 +1,3 @@
 authorized_key.json
 stat.yaml
+issues_stat.yaml

tools/greenplum-to-pg-tests/cmd/extraxtSessions.go

@@ -32,6 +32,8 @@ var extractSessionsConfig struct {
 	ydbConnectionString       string
 	limitRequests             int
 	rulesFile                 string
+	writeRulesWithStat        string
+	sortRulesByCount          bool
 	printKnownIssues          bool
 	printQueryForKnownIssue   bool
 	filterReason              string
@@ -53,6 +55,8 @@ func init() {
 	extractSessionsCmd.PersistentFlags().StringVar(&extractSessionsConfig.ydbConnectionString, "ydb-connection", "grpc://localhost:2136/local", "Connection string to ydb server for check queries")
 	extractSessionsCmd.PersistentFlags().IntVar(&extractSessionsConfig.limitRequests, "requests-limit", 100, "Limit number of parse requests, 0 mean unlimited")
 	extractSessionsCmd.PersistentFlags().StringVar(&extractSessionsConfig.rulesFile, "rules-file", "issues.yaml", "Rules for detect issue. Set empty for skip read rules.")
+	extractSessionsCmd.PersistentFlags().StringVar(&extractSessionsConfig.writeRulesWithStat, "write-updated-rules", "", "Write rules with updated stats, may be same or other file as for rules-file")
+	extractSessionsCmd.PersistentFlags().BoolVar(&extractSessionsConfig.sortRulesByCount, "sort-updates-rules-by-count", true, "")
 	extractSessionsCmd.PersistentFlags().BoolVar(&extractSessionsConfig.printKnownIssues, "print-known-issues", false, "Print known issues instead of unknown")
 	extractSessionsCmd.PersistentFlags().BoolVar(&extractSessionsConfig.printQueryForKnownIssue, "print-query-for-known-issues", true, "Print query for known issues")
 	extractSessionsCmd.PersistentFlags().IntVar(&extractSessionsConfig.errorLimit, "print-errors-limit", 0, "Limit of printed errors. 0 mean infinite")
@@ -105,8 +109,15 @@ var extractSessionsCmd = &cobra.Command{
 		sessions := readSessions()
 
 		log.Println("Start check queries")
-		checkQueries(rules, schema, db, sessions)
+		var stats SessionStats
+		checkQueries(rules, &stats, schema, db, sessions)
 
+		if extractSessionsConfig.writeRulesWithStat != "" {
+			rules.UpdateFromStats(stats, extractSessionsConfig.sortRulesByCount)
+			if err := rules.WriteToFile(extractSessionsConfig.writeRulesWithStat); err != nil {
+				log.Printf("Failed to update rules stat: %v", err)
+			}
+		}
 	},
 }
 
@@ -204,7 +215,7 @@ readLoop:
 	return res
 }
 
-func checkQueries(rules Rules, pgSchema *internal.PgSchema, db *ydb.Driver, sessions []internal.Session) {
+func checkQueries(rules Rules, stats *SessionStats, pgSchema *internal.PgSchema, db *ydb.Driver, sessions []internal.Session) {
 	reasonFilter := regexp.MustCompile(extractSessionsConfig.filterReason)
 	checked := map[string]bool{}
 
@@ -222,7 +233,6 @@ func checkQueries(rules Rules, pgSchema *internal.PgSchema, db *ydb.Driver, sess
 
 	queryIndex := 0
 
-	var stats SessionStats
 	for _, session := range sessions {
 		for _, transaction := range session.Transactions {
 			for _, pgQuery := range transaction.Queries {
@@ -235,7 +245,7 @@ func checkQueries(rules Rules, pgSchema *internal.PgSchema, db *ydb.Driver, sess
 				}
 				checked[pgQuery.Text] = true
 
-				reason, checkResult := checkQuery(&stats, rules, db, pgQuery.Text)
+				reason, checkResult := checkQuery(stats, rules, db, pgQuery.Text)
 				if !reasonFilter.MatchString(reason) {
 					continue
 				}
@@ -297,7 +307,7 @@ func checkQuery(stat *SessionStats, rules Rules, db *ydb.Driver, queryText strin
 	}
 
 	if err == nil {
-		stat.CountOK()
+		stat.CountASOK(queryText)
 		return "", checkResultOK
 	}
 
@@ -307,13 +317,13 @@ func checkQuery(stat *SessionStats, rules Rules, db *ydb.Driver, queryText strin
 	issues := internal.ExtractIssues(err)
 
 	if reason = rules.FindKnownIssue(queryText, issues); reason != "" {
-		stat.CountKnown(reason, queryText)
+		stat.CountAsKnown(reason, queryText)
 		return reason, checkResultErrKnown
 	}
 
 	reason = fmt.Sprintf("%v (%v): %#v", ydbErr.Name(), ydbErr.Code(), issues)
 
-	stat.CountUnknown(issues, queryText)
+	stat.CountAsUnknown(issues, queryText)
 	return reason, checkResultErrUnknown
 }
 
@@ -348,11 +358,15 @@ type SessionStats struct {
 	UnknownProblems map[internal.YdbIssue]*CounterWithExample[internal.YdbIssue]
 }
 
-func (s *SessionStats) CountOK() {
+func (s *SessionStats) GetOkPercent() float64 {
+	return float64(s.OkCount) / float64(s.TotalCount) * 100
+}
+
+func (s *SessionStats) CountASOK(query string) {
 	s.OkCount++
 }
 
-func (s *SessionStats) CountKnown(ruleName string, query string) {
+func (s *SessionStats) CountAsKnown(ruleName string, query string) {
 	s.TotalCount++
 	if s.MatchToRules == nil {
 		s.MatchToRules = make(map[string]*CounterWithExample[string])
@@ -371,7 +385,7 @@ func (s *SessionStats) CountKnown(ruleName string, query string) {
 	stat.Count++
 }
 
-func (s *SessionStats) CountUnknown(issues []internal.YdbIssue, query string) {
+func (s *SessionStats) CountAsUnknown(issues []internal.YdbIssue, query string) {
 	s.TotalCount++
 	if s.UnknownProblems == nil {
 		s.UnknownProblems = make(map[internal.YdbIssue]*CounterWithExample[internal.YdbIssue])
@@ -456,7 +470,7 @@ func (s *SessionStats) SaveToFile(path string) error {
 
 	statFile.TotalCount = s.TotalCount
 	statFile.OkCount = s.OkCount
-	statFile.OkPercent = float64(s.OkCount) / float64(s.TotalCount) * 100
+	statFile.OkPercent = s.GetOkPercent()
 	statFile.UnknownIssues = s.GetTopUnknown(math.MaxInt)
 	statFile.KnownIssues = s.GetTopKnown(math.MaxInt)
 

tools/greenplum-to-pg-tests/cmd/issue_rules.go

@@ -1,19 +1,27 @@
 package cmd
 
 import (
+	"bytes"
 	"errors"
 	"fmt"
+	"math"
 	"os"
 	"regexp"
+	"slices"
 
 	"gopkg.in/yaml.v3"
 
 	"github.com/ydb-platform/postgres-compatibility-tests/tools/greenplum-to-pg-tests/internal"
 )
 
 type Rules struct {
-	StatTotalCount int `yaml:"stat_total_count"`
-	Issues         []PgIssueRules
+	TotalStat struct {
+		TotalCount int     `yaml:"total_checked_queries,omitempty"`
+		TotalOk    int     `yaml:"total_ok,omitempty"`
+		OkPercent  float64 `yaml:"ok_percent,omitempty"`
+	} `yaml:"stat"`
+
+	Issues []PgIssueRules
 }
 
 func (r *Rules) LoadFromFile(path string) error {
@@ -51,6 +59,21 @@ func (r *Rules) LoadFromFile(path string) error {
 	return nil
 }
 
+func (r *Rules) WriteToFile(path string) error {
+	buf := &bytes.Buffer{}
+	encoder := yaml.NewEncoder(buf)
+	err := encoder.Encode(r)
+	if err != nil {
+		return fmt.Errorf("failed to encode rules: %w", err)
+	}
+
+	err = os.WriteFile(path, buf.Bytes(), 0666)
+	if err != nil {
+		return fmt.Errorf("failed write file to update rules")
+	}
+	return nil
+}
+
 func (r *Rules) FindKnownIssue(queryText string, ydbIssues []internal.YdbIssue) string {
 	for _, ydbIssue := range ydbIssues {
 		for _, item := range r.Issues {
@@ -63,15 +86,36 @@ func (r *Rules) FindKnownIssue(queryText string, ydbIssues []internal.YdbIssue)
 	return ""
 }
 
+func (r *Rules) UpdateFromStats(stats SessionStats, sortByCount bool) {
+	r.TotalStat.TotalCount = stats.TotalCount
+	r.TotalStat.TotalOk = stats.OkCount
+	r.TotalStat.OkPercent = math.Round(stats.GetOkPercent()*100) / 100
+
+	okStats := stats.GetTopKnown(math.MaxInt)
+	for _, stat := range okStats {
+		for issueIndex, issue := range r.Issues {
+			if issue.Name == stat.ID {
+				r.Issues[issueIndex].Count = stat.Count
+			}
+		}
+	}
+
+	if sortByCount {
+		slices.SortFunc(r.Issues, func(a, b PgIssueRules) int {
+			return b.Count - a.Count
+		})
+	}
+}
+
 type PgIssueRules struct {
 	Name        string           `yaml:"name"`
-	Tag         OneOrSliceString `yaml:"tag"`
-	IssueLink   string           `yaml:"issue_link"`
-	IssueRegexp OneOrSliceString `yaml:"issue_regexp"`
-	QueryRegexp OneOrSliceString `yaml:"query_regexp"`
-	Example     string           `yaml:"example"`
-	Comment     string           `yaml:"comment"`
 	Count       int              `yaml:"count"`
+	Tag         OneOrSliceString `yaml:"tag,omitempty"`
+	IssueLink   string           `yaml:"issue_link,omitempty"`
+	IssueRegexp OneOrSliceString `yaml:"issue_regexp,omitempty"`
+	QueryRegexp OneOrSliceString `yaml:"query_regexp,omitempty"`
+	Example     string           `yaml:"example,omitempty"`
+	Comment     string           `yaml:"comment,omitempty"`
 
 	issuesRegexpCompiled []*regexp.Regexp
 	queryRegexpCompiled  []*regexp.Regexp

tools/greenplum-to-pg-tests/issues.yaml

@@ -1,50 +1,97 @@
+stat:
+  total_checked_queries: 23
+  total_ok: 15
+  ok_percent: 65.21739130434783
 issues:
   - name: Skip unknown tables
-    issue_regexp: Cannot find table
-    tag: skip
-  - name: Rollback in query service (OK)
-    issue_regexp: ROLLBACK not supported inside YDB query
-    tag: skip
-  - name: Commit in query service (OK)
-    issue_regexp: COMMIT not supported inside YDB query
-    tag: skip
+    count: 11
+    tag:
+      - skip
+    issue_regexp:
+      - Cannot find table
+  - name: drop external table
+    count: 2
+    tag:
+      - greenplum
+    issue_regexp:
+      - 'ERROR:  syntax error at or near "external"'
+    query_regexp:
+      - drop external table
   - name: Create table as
-    issue_regexp: "^RawStmt: alternative is not implemented yet : 277$"
-    tag: YQLParser
+    count: 2
+    tag:
+      - YQLParser
+    issue_regexp:
+      - '^RawStmt: alternative is not implemented yet : 277$'
     example: create table t as select 1
-  - name: Unimplemented Discard
-    issue_regexp: "^RawStmt: alternative is not implemented yet : 282$"
-    example: DISCARD ALL
-  - name: Generator functions in SELECT
-    issue_regexp: "^Generator functions are not allowed in: SELECT$"
-    example: SELECT UNNEST($1)
-  - name: Call function from own schema
-    issue_regexp: "FuncCall: expected pg_catalog, but got: "
-  - name: Partial request
-    issue_regexp: "ERROR:  syntax error at end of input"
-    tag: skip
-  - name: "YQL: Expected type cast node as is_local arg, but got node with tag"
-    issue_regexp: "Expected type cast node as is_local arg, but got node with tag"
-  - name: "PG: unrecognized configuration parameter"
-    issue_regexp: unrecognized configuration parameter
-  - name: "PG: InsertStmt: not supported onConflictClause"
-    issue_regexp: "InsertStmt: not supported onConflictClause"
-  - name: "PG: RangeFunction: unsupported coldeflist"
-    issue_regexp: "RangeFunction: unsupported coldeflist"
-  - name: "PG: Support NOT DISTINCT"
-    issue_regexp: "A_Expr_Kind unsupported value: 4"
-  - name: drop external table
-    tag: greenplum
-    issue_regexp: "ERROR:  syntax error at or near \"external\""
-    query_regexp: "drop external table"
+  - name: Distributed randomly
+    count: 2
+    tag:
+      - greenplum
+    issue_regexp:
+      - 'ERROR:  syntax error at or near "RANDOMLY"'
+      - 'ERROR:  syntax error at or near "DISTRIBUTED"'
+    query_regexp:
+      - (?i)DISTRIBUTED\s+RANDOMLY
   - name: Stored procedures
-    issue_regexp: "No such proc: "
-  - name: "Distributed randomly"
-    tag: greenplum
+    count: 2
+    issue_regexp:
+      - 'No such proc: '
+  - name: Rollback in query service (OK)
+    count: 1
+    tag:
+      - skip
+    issue_regexp:
+      - ROLLBACK not supported inside YDB query
+  - name: Commit in query service (OK)
+    count: 1
+    tag:
+      - skip
     issue_regexp:
-      - "ERROR:  syntax error at or near \"RANDOMLY\""
-      - "ERROR:  syntax error at or near \"DISTRIBUTED\""
-    query_regexp: "(?i)DISTRIBUTED\\s+RANDOMLY"
+      - COMMIT not supported inside YDB query
   - name: Not supported set greenplum var
-    issue_regexp: "VariableSetStmt, not supported name: gp_"
-    query_regexp: ^SET gp_
+    count: 1
+    issue_regexp:
+      - 'VariableSetStmt, not supported name: gp_'
+    query_regexp:
+      - ^SET gp_
+  - name: Partial request
+    count: 1
+    tag:
+      - skip
+    issue_regexp:
+      - 'ERROR:  syntax error at end of input'
+  - name: 'YQL: Expected type cast node as is_local arg, but got node with tag'
+    count: 0
+    issue_regexp:
+      - Expected type cast node as is_local arg, but got node with tag
+  - name: 'PG: InsertStmt: not supported onConflictClause'
+    count: 0
+    issue_regexp:
+      - 'InsertStmt: not supported onConflictClause'
+  - name: 'PG: RangeFunction: unsupported coldeflist'
+    count: 0
+    issue_regexp:
+      - 'RangeFunction: unsupported coldeflist'
+  - name: 'PG: Support NOT DISTINCT'
+    count: 0
+    issue_regexp:
+      - 'A_Expr_Kind unsupported value: 4'
+  - name: 'PG: unrecognized configuration parameter'
+    count: 0
+    issue_regexp:
+      - unrecognized configuration parameter
+  - name: Call function from own schema
+    count: 0
+    issue_regexp:
+      - 'FuncCall: expected pg_catalog, but got: '
+  - name: Generator functions in SELECT
+    count: 0
+    issue_regexp:
+      - '^Generator functions are not allowed in: SELECT$'
+    example: SELECT UNNEST($1)
+  - name: Unimplemented Discard
+    count: 0
+    issue_regexp:
+      - '^RawStmt: alternative is not implemented yet : 282$'
+    example: DISCARD ALL

untitled/go.mod

@@ -0,0 +1,3 @@
+module untitled
+
+go 1.22