sift/internal/search/apply.go

package search

import (
	"net/url"
	"strings"
)

func resultHost(raw string) string {
	u, err := url.Parse(raw)
	if err != nil {
		return ""
	}
	h := strings.ToLower(u.Hostname())
	return strings.TrimSuffix(h, ".")
}

func hostMatch(host, d string) bool {
	return host == d || strings.HasSuffix(host, "."+d)
}

func filterDomains(hits []Hit, f Filter) []Hit {
	if len(f.IncludeDomains) == 0 && len(f.ExcludeDomains) == 0 {
		return hits
	}
	mode := f.IncludeDomainsMode
	if mode == "" {
		mode = "restrict"
	}
	restrict := mode != "prefer" && len(f.IncludeDomains) > 0
	out := make([]Hit, 0, len(hits))
	for _, h := range hits {
		host := resultHost(h.URL)
		if host == "" {
			if restrict {
				continue
			}
		} else {
			drop := false
			for _, d := range f.ExcludeDomains {
				if hostMatch(host, d) {
					drop = true
					break
				}
			}
			if drop {
				continue
			}
			if restrict {
				ok := false
				for _, d := range f.IncludeDomains {
					if hostMatch(host, d) {
						ok = true
						break
					}
				}
				if !ok {
					continue
				}
			}
		}
		out = append(out, h)
	}
	return out
}

func filterMinScore(hits []Hit, min float64) []Hit {
	if min <= 0 {
		return hits
	}
	out := make([]Hit, 0, len(hits))
	for _, h := range hits {
		if h.Score >= min {
			out = append(out, h)
		}
	}
	return out
}

func dedupeStrings(in []string) []string {
	seen := map[string]bool{}
	out := make([]string, 0, len(in))
	for _, s := range in {
		if s == "" || seen[s] {
			continue
		}
		seen[s] = true
		out = append(out, s)
	}
	return out
}

func searxQueryAndNotes(q string, f Filter) (string, []string) {
	var notes []string
	if f.IncludeDomainsMode == "prefer" && len(f.IncludeDomains) > 0 {
		notes = append(notes, "searxng: include_domains_mode=prefer not applied")
	} else if n := len(f.IncludeDomains); n == 1 {
		q = "site:" + f.IncludeDomains[0] + " " + q
	} else if n > 1 {
		max := n
		if max > searxSiteCap {
			max = searxSiteCap
			notes = append(notes, "searxng: extra include_domains dropped")
		}
		var b strings.Builder
		for i := 0; i < max; i++ {
			if i > 0 {
				b.WriteString(" OR ")
			}
			b.WriteString("site:")
			b.WriteString(f.IncludeDomains[i])
		}
		b.WriteByte(' ')
		b.WriteString(q)
		q = b.String()
	}
	if f.Topic == "finance" {
		notes = append(notes, "searxng: topic ignored")
	}
	if f.StartDate != "" || f.EndDate != "" {
		notes = append(notes, "searxng: start_date/end_date not applied")
	}
	if f.MinScore > 0 {
		notes = append(notes, "searxng: min_score not applied")
	}
	return q, notes
}

func braveNotes(f Filter) []string {
	var notes []string
	if f.TimeRange != "" {
		notes = append(notes, "brave_llm: time_range not applied")
	}
	if f.Topic != "" && f.Topic != "general" {
		notes = append(notes, "brave_llm: topic not applied")
	}
	if f.ExactMatch {
		notes = append(notes, "brave_llm: exact_match not applied")
	}
	if f.IncludeDomainsMode == "prefer" && len(f.IncludeDomains) > 0 {
		notes = append(notes, "brave_llm: include_domains_mode=prefer not applied")
	}
	if f.MinScore > 0 {
		notes = append(notes, "brave_llm: min_score not applied")
	}
	return notes
}

func FilterKinds(f Filter) string {
	var k []string
	if len(f.IncludeDomains) > 0 || len(f.ExcludeDomains) > 0 {
		k = append(k, "domains")
	}
	if f.TimeRange != "" || f.StartDate != "" || f.EndDate != "" {
		k = append(k, "time")
	}
	if f.Topic != "" && f.Topic != "general" {
		k = append(k, "topic")
	}
	if f.ExactMatch {
		k = append(k, "exact")
	}
	if f.MinScore > 0 {
		k = append(k, "score")
	}
	return strings.Join(k, ",")
}