searchgit/query.go

package main

import (
	"fmt"
	"net/url"
	"regexp"
	"strconv"
	"strings"
	"time"
)

// Criteria are the search filters chosen in the UI. They are turned into a
// GitHub search query (https://docs.github.com/search-github/searching-on-github/searching-for-repositories).
type Criteria struct {
	Text      string // free text, may already contain GitHub qualifiers (topic:cli user:foo…)
	Language  string
	MinStars  int
	MaxStars  int
	MinForks  int
	Topic     string
	License   string        // SPDX key in lower case: mit, apache-2.0…
	PushedIn  time.Duration // last push less than this ago (0 = any)
	CreatedIn time.Duration // created less than this ago (0 = any)
	In        string        // where the text is searched: "", "name", "description", "readme"
	Archived  bool          // include archived repositories
	Forks     bool          // include forks
	GoodFirst bool          // has issues labeled "good first issue"
	Sort      string        // "", "stars", "forks", "updated", "help-wanted-issues"
	Order     string        // "desc" (default) or "asc"
	Page      int
	PerPage   int
}

var (
	sorts     = map[string]bool{"": true, "stars": true, "forks": true, "updated": true, "help-wanted-issues": true}
	ins       = map[string]bool{"": true, "name": true, "description": true, "readme": true}
	safeValue = regexp.MustCompile(`^[\w.+#-]+$`)  // topic, license
	safeLang  = regexp.MustCompile(`^[\w.+# -]+$`) // "Jupyter Notebook", "C#", "C++"
	durations = map[string]time.Duration{
		"1w": 7 * 24 * time.Hour, "1m": 30 * 24 * time.Hour, "3m": 91 * 24 * time.Hour,
		"6m": 182 * 24 * time.Hour, "1y": 365 * 24 * time.Hour, "2y": 730 * 24 * time.Hour,
		"5y": 1826 * 24 * time.Hour,
	}
)

// ParseCriteria reads the criteria from the /api/search query string.
func ParseCriteria(v url.Values) (Criteria, error) {
	c := Criteria{
		Text:      strings.TrimSpace(v.Get("q")),
		Language:  strings.TrimSpace(v.Get("language")),
		Topic:     strings.ToLower(strings.TrimSpace(v.Get("topic"))),
		License:   strings.ToLower(strings.TrimSpace(v.Get("license"))),
		In:        v.Get("in"),
		Archived:  v.Get("archived") == "1",
		Forks:     v.Get("forks") == "1",
		GoodFirst: v.Get("goodfirst") == "1",
		Sort:      v.Get("sort"),
		Order:     v.Get("order"),
		Page:      1,
		PerPage:   30,
	}
	var err error
	if c.MinStars, err = intParam(v, "stars"); err != nil {
		return c, err
	}
	if c.MaxStars, err = intParam(v, "maxstars"); err != nil {
		return c, err
	}
	if c.MinForks, err = intParam(v, "minforks"); err != nil {
		return c, err
	}
	if p, err := intParam(v, "page"); err != nil {
		return c, err
	} else if p > 0 {
		c.Page = p
	}
	if n, err := intParam(v, "per_page"); err != nil {
		return c, err
	} else if n > 0 {
		c.PerPage = min(n, 100)
	}
	if c.PushedIn, err = durationParam(v, "pushed"); err != nil {
		return c, err
	}
	if c.CreatedIn, err = durationParam(v, "created"); err != nil {
		return c, err
	}
	if c.Language != "" && !safeLang.MatchString(c.Language) {
		return c, &codedError{code: "bad_param", msg: "invalid language", detail: "language"}
	}
	for name, val := range map[string]string{"topic": c.Topic, "license": c.License} {
		if val != "" && !safeValue.MatchString(val) {
			return c, &codedError{code: "bad_param", msg: "invalid " + name, detail: name}
		}
	}
	if !sorts[c.Sort] {
		return c, &codedError{code: "bad_param", msg: "invalid sort", detail: "sort"}
	}
	if !ins[c.In] {
		return c, &codedError{code: "bad_param", msg: "invalid in", detail: "in"}
	}
	if c.Order != "asc" {
		c.Order = "desc"
	}
	// GitHub never returns more than the first 1000 results.
	if c.Page*c.PerPage > 1000 {
		return c, &codedError{code: "too_far", msg: "GitHub only returns the first 1000 results"}
	}
	return c, nil
}

func intParam(v url.Values, key string) (int, error) {
	s := strings.TrimSpace(v.Get(key))
	if s == "" {
		return 0, nil
	}
	n, err := strconv.Atoi(s)
	if err != nil || n < 0 {
		return 0, &codedError{code: "bad_param", msg: "invalid " + key, detail: key}
	}
	return n, nil
}

func durationParam(v url.Values, key string) (time.Duration, error) {
	s := v.Get(key)
	if s == "" {
		return 0, nil
	}
	d, ok := durations[s]
	if !ok {
		return 0, &codedError{code: "bad_param", msg: "invalid " + key, detail: key}
	}
	return d, nil
}

// Query builds the GitHub "q" parameter. now is passed in for tests.
func (c Criteria) Query(now time.Time) string {
	var parts []string
	if c.Text != "" {
		parts = append(parts, c.Text)
	}
	if c.In != "" && c.Text != "" {
		parts = append(parts, "in:"+c.In)
	}
	if c.Language != "" {
		parts = append(parts, "language:"+quoteIfNeeded(c.Language))
	}
	if c.Topic != "" {
		parts = append(parts, "topic:"+c.Topic)
	}
	if c.License != "" {
		parts = append(parts, "license:"+c.License)
	}
	switch {
	case c.MinStars > 0 && c.MaxStars > 0:
		parts = append(parts, fmt.Sprintf("stars:%d..%d", c.MinStars, c.MaxStars))
	case c.MinStars > 0:
		parts = append(parts, fmt.Sprintf("stars:>=%d", c.MinStars))
	case c.MaxStars > 0:
		parts = append(parts, fmt.Sprintf("stars:<=%d", c.MaxStars))
	}
	if c.MinForks > 0 {
		parts = append(parts, fmt.Sprintf("forks:>=%d", c.MinForks))
	}
	if c.PushedIn > 0 {
		parts = append(parts, "pushed:>="+now.Add(-c.PushedIn).UTC().Format("2006-01-02"))
	}
	if c.CreatedIn > 0 {
		parts = append(parts, "created:>="+now.Add(-c.CreatedIn).UTC().Format("2006-01-02"))
	}
	if !c.Archived {
		parts = append(parts, "archived:false")
	}
	if c.Forks {
		parts = append(parts, "fork:true")
	}
	if c.GoodFirst {
		parts = append(parts, "good-first-issues:>0")
	}
	// GitHub refuses an empty query: search everything with at least one star.
	if c.Text == "" && c.Language == "" && c.Topic == "" && c.License == "" && c.MinStars == 0 && c.MaxStars == 0 {
		parts = append(parts, "stars:>=1")
	}
	return strings.Join(parts, " ")
}

func quoteIfNeeded(s string) string {
	if strings.ContainsAny(s, " ") {
		return `"` + s + `"`
	}
	return s
}