Add support for manually setting a limit on results

* Added a query parameter to allow user to specify max results, rather
  than a hard limit at 5000, for each index.
* Added support for a default max result limit in the config.  If unset,
  it remains at the former default of 5000 results per index.
* Fixed issue where no results would be returned if the limit was
  exceeded - now it simply returns the results up to the limit.

api/api.go

@@ -18,6 +18,7 @@ import (
 const (
 	defaultLinesOfContext uint = 2
 	maxLinesOfContext     uint = 20
+	maxLimit              int  = 100000
 )
 
 type Stats struct {
@@ -128,11 +129,25 @@ func parseAsUintValue(sv string, min, max, def uint) uint {
 		return max
 	}
 	if min != 0 && uint(iv) < min {
-		return max
+		return min
 	}
 	return uint(iv)
 }
 
+func parseAsIntValue(sv string, min, max, def int) int {
+	iv, err := strconv.ParseInt(sv, 10, 64)
+	if err != nil {
+		return def
+	}
+	if max != 0 && int(iv) > max {
+		return max
+	}
+	if min != 0 && int(iv) < min {
+		return min
+	}
+	return int(iv)
+}
+
 func parseRangeInt(v string, i *int) {
 	*i = 0
 	if v == "" {
@@ -159,8 +174,7 @@ func parseRangeValue(rv string) (int, int) {
 	return b, e
 }
 
-func Setup(m *http.ServeMux, idx map[string]*searcher.Searcher) {
-
+func Setup(m *http.ServeMux, idx map[string]*searcher.Searcher, defaultMaxResults int) {
 	m.HandleFunc("/api/v1/repos", func(w http.ResponseWriter, r *http.Request) {
 		res := map[string]*config.Repo{}
 		for name, srch := range idx {
@@ -181,6 +195,11 @@ func Setup(m *http.ServeMux, idx map[string]*searcher.Searcher) {
 		opt.ExcludeFileRegexp = r.FormValue("excludeFiles")
 		opt.IgnoreCase = parseAsBool(r.FormValue("i"))
 		opt.LiteralSearch = parseAsBool(r.FormValue("literal"))
+		opt.MaxResults = parseAsIntValue(
+			r.FormValue("limit"),
+			-1,
+			maxLimit,
+			defaultMaxResults)
 		opt.LinesOfContext = parseAsUintValue(
 			r.FormValue("ctx"),
 			0,

cmds/houndd/main.go

@@ -111,7 +111,7 @@ func runHttp( //nolint
 	}
 
 	m.Handle("/", h)
-	api.Setup(m, idx)
+	api.Setup(m, idx, cfg.ResultLimit)
 	return http.ListenAndServe(addr, m)
 }
 

config/config.go

@@ -17,6 +17,7 @@ const (
 	defaultBaseUrl               = "{url}/blob/{rev}/{path}{anchor}"
 	defaultAnchor                = "#L{line}"
 	defaultHealthCheckURI        = "/healthz"
+	defaultResultLimit           = 5000
 )
 
 type UrlPattern struct {
@@ -63,6 +64,7 @@ type Config struct {
 	MaxConcurrentIndexers int                       `json:"max-concurrent-indexers"`
 	HealthCheckURI        string                    `json:"health-check-uri"`
 	VCSConfigMessages     map[string]*SecretMessage `json:"vcs-config"`
+	ResultLimit           int                       `json:"result-limit"`
 }
 
 // SecretMessage is just like json.RawMessage but it will not
@@ -130,6 +132,10 @@ func initConfig(c *Config) error {
 		c.HealthCheckURI = defaultHealthCheckURI
 	}
 
+	if c.ResultLimit == 0 {
+		c.ResultLimit = defaultResultLimit
+	}
+
 	return mergeVCSConfigs(c)
 }