xinmi/internal/application/service/chat_pipeline/extract_entity_coverage.go

package chatpipeline

import (
	"fmt"
	"regexp"
	"strings"

	"github.com/Tencent/WeKnora/internal/types"
)

var graphStorageReachInstructionPattern = regexp.MustCompile(`^(?:请\s*)?(?:放\s*置\s*于|置\s*于|放\s*在).+不\s*易\s*触\s*及(?:\s*之)?\s*处$`)

// This bounded coverage check recognizes explicit placement instructions only.
// Unclear OCR stays verbatim; detecting a missing rule must never create an edge.
func validateGraphStorageCoverage(graph *types.GraphData, content string) error {
	for _, clause := range strings.FieldsFunc(content, func(r rune) bool {
		return strings.ContainsRune("，,。！？!?；;：:\n\r", r)
	}) {
		clause = strings.TrimSpace(clause)
		if heading := graphSectionHeadingPattern.FindStringIndex(clause); heading != nil && heading[0] == 0 {
			clause = strings.TrimSpace(clause[heading[1]:])
		}
		if !graphStorageReachInstructionPattern.MatchString(clause) {
			continue
		}
		covered := false
		for _, relation := range graph.Relation {
			if relation != nil && relation.Type == "约束" &&
				strings.Contains(relation.Node2, clause) &&
				strings.Contains(relation.Evidence, clause) &&
				strings.Contains(content, relation.Evidence) {
				covered = true
				break
			}
		}
		if !covered {
			return fmt.Errorf("%w: explicit storage restriction is not covered by a complete source-grounded constraint: %q; preserve the original clause verbatim without repairing OCR if its relationship is supported, otherwise leave this source for review; never invent an edge to satisfy coverage", ErrGraphNeedsReview, clause)
		}
	}
	return nil
}