rsspipe/internal/rules/evaluator.go

package rules

import (
	"fmt"
	"reflect"
	"regexp"
	"strconv"
	"strings"
	"time"

	"git.pengzhan.dev/haopengzhan/rsspipe/internal/models"
)

type Evaluator struct{}

func NewEvaluator() *Evaluator {
	return &Evaluator{}
}

// Evaluate checks whether a single FeedEntry satisfies a Rule's conditions (legacy / static evaluation).
func (e *Evaluator) Evaluate(rule models.Rule, entry models.FeedEntry) bool {
	return e.EvaluateTransition(rule, nil, &entry, nil)
}

// EvaluateTransition evaluates a rule against the (OldState, NewState, EntryDiff) transition delta.
func (e *Evaluator) EvaluateTransition(rule models.Rule, oldEntry *models.FeedEntry, newEntry *models.FeedEntry, diff *models.EntryDiff) bool {
	if newEntry == nil {
		return false
	}
	if len(rule.Conditions) == 0 {
		return true
	}

	matchMode := strings.ToLower(rule.MatchMode)
	if matchMode == "" || matchMode == "all" {
		for _, cond := range rule.Conditions {
			if !e.evaluateTransitionCondition(cond, oldEntry, newEntry, diff) {
				return false
			}
		}
		return true
	}

	// matchMode == "any"
	for _, cond := range rule.Conditions {
		if e.evaluateTransitionCondition(cond, oldEntry, newEntry, diff) {
			return true
		}
	}
	return false
}

func (e *Evaluator) evaluateTransitionCondition(cond models.RuleCondition, oldEntry *models.FeedEntry, newEntry *models.FeedEntry, diff *models.EntryDiff) bool {
	opLower := strings.ToLower(strings.TrimSpace(cond.Operator))

	// 1. Transition predicates: became_present, changed, removed
	switch opLower {
	case "became_present", "added":
		return e.checkBecamePresent(cond.Field, oldEntry, newEntry, diff)
	case "changed", "updated":
		return e.checkFieldChanged(cond.Field, oldEntry, newEntry, diff)
	case "removed", "deleted":
		return e.checkFieldRemoved(cond.Field, oldEntry, newEntry, diff)
	}

	// 2. Standard state guards & value comparisons evaluated on NewState
	actualValue := e.resolveField(cond.Field, newEntry)
	return e.compare(actualValue, opLower, cond.Value)
}

func (e *Evaluator) checkBecamePresent(field string, oldEntry *models.FeedEntry, newEntry *models.FeedEntry, diff *models.EntryDiff) bool {
	if newEntry == nil {
		return false
	}

	// If diff is available, check direct addition
	if strings.HasPrefix(field, "variants.") {
		vKey := strings.TrimPrefix(field, "variants.")
		if strings.Contains(vKey, ".") {
			vKey = strings.Split(vKey, ".")[0]
		}
		if diff != nil && diff.HasVariantAddition(vKey) {
			return true
		}
		// Fallback comparison between oldEntry and newEntry
		oldHas := oldEntry != nil && oldEntry.Variants != nil && oldEntry.Variants[vKey].TextValue != ""
		newHas := newEntry.Variants != nil && newEntry.Variants[vKey].TextValue != ""
		return !oldHas && newHas
	}

	if field == "metadata.tags" || field == "tags" {
		oldTags := getTagsSlice(oldEntry)
		newTags := getTagsSlice(newEntry)
		return len(oldTags) == 0 && len(newTags) > 0
	}

	oldVal := e.resolveField(field, oldEntry)
	newVal := e.resolveField(field, newEntry)
	return (oldVal == nil || fmt.Sprintf("%v", oldVal) == "") && (newVal != nil && fmt.Sprintf("%v", newVal) != "")
}

func (e *Evaluator) checkFieldChanged(field string, oldEntry *models.FeedEntry, newEntry *models.FeedEntry, diff *models.EntryDiff) bool {
	if oldEntry == nil || newEntry == nil {
		return false
	}
	if strings.HasPrefix(field, "variants.") {
		vKey := strings.TrimPrefix(field, "variants.")
		if strings.Contains(vKey, ".") {
			vKey = strings.Split(vKey, ".")[0]
		}
		if diff != nil && (diff.HasVariantAddition(vKey) || diff.HasVariantChange(vKey)) {
			return true
		}
	}
	oldVal := e.resolveField(field, oldEntry)
	newVal := e.resolveField(field, newEntry)
	return !reflect.DeepEqual(oldVal, newVal)
}

func (e *Evaluator) checkFieldRemoved(field string, oldEntry *models.FeedEntry, newEntry *models.FeedEntry, diff *models.EntryDiff) bool {
	if oldEntry == nil {
		return false
	}
	if strings.HasPrefix(field, "variants.") {
		vKey := strings.TrimPrefix(field, "variants.")
		if diff != nil {
			for _, r := range diff.VariantsRemoved {
				if r == vKey {
					return true
				}
			}
		}
		oldHas := oldEntry.Variants != nil && oldEntry.Variants[vKey].TextValue != ""
		newHas := newEntry != nil && newEntry.Variants != nil && newEntry.Variants[vKey].TextValue != ""
		return oldHas && !newHas
	}
	oldVal := e.resolveField(field, oldEntry)
	newVal := e.resolveField(field, newEntry)
	return (oldVal != nil && fmt.Sprintf("%v", oldVal) != "") && (newVal == nil || fmt.Sprintf("%v", newVal) == "")
}

func (e *Evaluator) resolveField(field string, entry *models.FeedEntry) interface{} {
	if entry == nil {
		return nil
	}

	fieldLower := strings.ToLower(strings.TrimSpace(field))
	switch fieldLower {
	case "feed_id":
		return entry.FeedID
	case "feed_title":
		return entry.FeedTitle
	case "title":
		return entry.Title
	case "author":
		return entry.Author
	case "url":
		return entry.URL
	case "content", "raw_content", "rss_content":
		var parts []string
		if entry.Title != "" {
			parts = append(parts, entry.Title)
		}
		if entry.Content != "" {
			parts = append(parts, entry.Content)
		}
		if entry.RawContent != "" && entry.RawContent != entry.Content {
			parts = append(parts, entry.RawContent)
		}
		if entry.Variants != nil {
			if v, ok := entry.Variants["readability_full_text"]; ok && v.TextValue != "" {
				parts = append(parts, v.TextValue)
			}
		}
		return strings.Join(parts, "\n")
	case "is_read":
		return entry.IsRead
	case "is_starred":
		return entry.IsStarred
	case "published_at":
		return entry.PublishedAt
	case "created_at":
		return entry.CreatedAt
	case "tags", "metadata.tags":
		return getTagsSlice(entry)
	}

	// Variant resolution: "variants.ai_summary.word_count", "variants.ai_summary", "variants.readability_full_text"
	if strings.HasPrefix(fieldLower, "variants.") {
		path := strings.TrimPrefix(fieldLower, "variants.")
		parts := strings.Split(path, ".")
		vKey := parts[0]

		if entry.Variants != nil {
			if v, ok := entry.Variants[vKey]; ok {
				if len(parts) > 1 {
					subProp := parts[1]
					if subProp == "word_count" {
						return float64(len(strings.Fields(v.TextValue)))
					}
					if subProp == "char_count" || subProp == "length" {
						return float64(len(v.TextValue))
					}
					if v.Metadata != nil {
						return v.Metadata[subProp]
					}
				}
				return v.TextValue
			}
		}
		return nil
	}

	return nil
}

func (e *Evaluator) compare(actual interface{}, op string, expected interface{}) bool {
	opLower := strings.ToLower(strings.TrimSpace(op))

	// Handle is_empty / is_present / exists / not_exists
	if opLower == "is_empty" || opLower == "not_exists" {
		if actual == nil {
			return true
		}
		if s, ok := actual.(string); ok && strings.TrimSpace(s) == "" {
			return true
		}
		if slice, ok := actual.([]string); ok && len(slice) == 0 {
			return true
		}
		return false
	}
	if opLower == "is_present" || opLower == "exists" {
		if actual == nil {
			return false
		}
		if s, ok := actual.(string); ok && strings.TrimSpace(s) == "" {
			return false
		}
		if slice, ok := actual.([]string); ok && len(slice) == 0 {
			return false
		}
		return true
	}

	if actual == nil {
		return opLower == "neq" || opLower == "!=" || opLower == "not_in" || opLower == "not_contains"
	}

	// Handle numeric comparisons (word_count >= 500, etc.)
	if actualNum, ok := toFloat64(actual); ok == nil {
		if expNum, okExp := toFloat64(expected); okExp == nil {
			switch opLower {
			case ">=":
				return actualNum >= expNum
			case ">":
				return actualNum > expNum
			case "<=":
				return actualNum <= expNum
			case "<":
				return actualNum < expNum
			case "eq", "==":
				return actualNum == expNum
			case "neq", "!=":
				return actualNum != expNum
			}
		}
	}

	// Handle Time comparisons
	if actualTime, ok := actual.(time.Time); ok {
		return e.compareTime(actualTime, opLower, expected)
	}

	// Handle slice containment (e.g. tags contains "Tech")
	if actualSlice, ok := actual.([]string); ok {
		expectedStr := fmt.Sprintf("%v", expected)
		switch opLower {
		case "contains", "in":
			for _, item := range actualSlice {
				if strings.EqualFold(item, expectedStr) {
					return true
				}
			}
			return false
		case "contains_any":
			tokens := parseStringList(expected)
			for _, item := range actualSlice {
				for _, t := range tokens {
					if strings.EqualFold(item, t) {
						return true
					}
				}
			}
			return false
		}
	}

	actualStr := fmt.Sprintf("%v", actual)
	expectedStr := fmt.Sprintf("%v", expected)

	switch opLower {
	case "eq", "==":
		return strings.EqualFold(actualStr, expectedStr)
	case "neq", "!=":
		return !strings.EqualFold(actualStr, expectedStr)
	case "contains":
		expectedClean := strings.Trim(expectedStr, "\"'` \t\r\n")
		return strings.Contains(strings.ToLower(actualStr), strings.ToLower(expectedClean))
	case "not_contains":
		expectedClean := strings.Trim(expectedStr, "\"'` \t\r\n")
		return !strings.Contains(strings.ToLower(actualStr), strings.ToLower(expectedClean))
	case "starts_with":
		expectedClean := strings.Trim(expectedStr, "\"'` \t\r\n")
		return strings.HasPrefix(strings.ToLower(actualStr), strings.ToLower(expectedClean))
	case "ends_with":
		expectedClean := strings.Trim(expectedStr, "\"'` \t\r\n")
		return strings.HasSuffix(strings.ToLower(actualStr), strings.ToLower(expectedClean))
	case "contains_any":
		tokens := parseStringList(expected)
		for _, token := range tokens {
			if strings.Contains(strings.ToLower(actualStr), strings.ToLower(token)) {
				return true
			}
		}
		return false
	case "in":
		tokens := parseStringList(expected)
		for _, item := range tokens {
			if strings.EqualFold(actualStr, item) {
				return true
			}
		}
		return false
	case "not_in":
		tokens := parseStringList(expected)
		for _, item := range tokens {
			if strings.EqualFold(actualStr, item) {
				return false
			}
		}
		return true
	case "matches", "regex":
		re, err := regexp.Compile(expectedStr)
		if err != nil {
			return false
		}
		return re.MatchString(actualStr)
	default:
		return false
	}
}

func (e *Evaluator) compareTime(t time.Time, op string, expected interface{}) bool {
	now := time.Now()
	switch op {
	case "within_last_hours":
		hours, err := toFloat64(expected)
		if err != nil {
			return false
		}
		cutoff := now.Add(-time.Duration(hours * float64(time.Hour)))
		return t.After(cutoff)
	case "older_than_days":
		days, err := toFloat64(expected)
		if err != nil {
			return false
		}
		cutoff := now.Add(-time.Duration(days * 24 * float64(time.Hour)))
		return t.Before(cutoff)
	case ">=", "after":
		targetTime, err := parseTime(expected)
		if err != nil {
			return false
		}
		return t.After(targetTime) || t.Equal(targetTime)
	case "<=", "before":
		targetTime, err := parseTime(expected)
		if err != nil {
			return false
		}
		return t.Before(targetTime) || t.Equal(targetTime)
	}
	return false
}

func getTagsSlice(entry *models.FeedEntry) []string {
	if entry == nil || entry.Metadata == nil {
		return nil
	}
	if raw, ok := entry.Metadata["tags"]; ok {
		return parseStringList(raw)
	}
	return nil
}

func parseStringList(val interface{}) []string {
	var result []string
	clean := func(s string) string {
		s = strings.TrimSpace(s)
		s = strings.Trim(s, "\"'`")
		return strings.TrimSpace(s)
	}

	switch v := val.(type) {
	case []interface{}:
		for _, item := range v {
			itemStr := clean(fmt.Sprintf("%v", item))
			if itemStr != "" {
				result = append(result, itemStr)
			}
		}
	case []string:
		for _, item := range v {
			itemStr := clean(item)
			if itemStr != "" {
				result = append(result, itemStr)
			}
		}
	case string:
		parts := strings.Split(v, ",")
		for _, part := range parts {
			trimmed := clean(part)
			if trimmed != "" {
				result = append(result, trimmed)
			}
		}
	}
	return result
}

func toFloat64(val interface{}) (float64, error) {
	switch v := val.(type) {
	case float64:
		return v, nil
	case int:
		return float64(v), nil
	case int64:
		return float64(v), nil
	case string:
		return strconv.ParseFloat(strings.TrimSpace(v), 64)
	}
	return 0, fmt.Errorf("cannot convert %v to float64", val)
}

func parseTime(val interface{}) (time.Time, error) {
	if t, ok := val.(time.Time); ok {
		return t, nil
	}
	s := fmt.Sprintf("%v", val)
	return time.Parse(time.RFC3339, s)
}

// ValidateRuleSafety checks for potential cyclic/endless loops at rule creation time.
func ValidateRuleSafety(rule models.Rule) error {
	trigger := strings.ToLower(rule.Trigger)

	if trigger == "on_variant_updated" || trigger == "variant_change" {
		for _, action := range rule.Actions {
			if action.Type == "set_variant" {
				targetKey, _ := action.Params["variant_key"].(string)
				if targetKey == "" {
					targetKey, _ = action.Params["key"].(string)
				}

				for _, cond := range rule.Conditions {
					if cond.Field == "variants."+targetKey && (cond.Operator == "exists" || cond.Operator == "eq") {
						return fmt.Errorf("cyclic rule detected: rule triggers on variant '%s' and unconditionally writes to '%s'", targetKey, targetKey)
					}
				}
			}
		}
	}

	return nil
}