From fa348dca5d3a1ae9c9f299c3e3494c84a51e8bbe Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Alo=C3=AFs=20Micard?= <alois@micard.lu>
Date: Tue, 22 Sep 2020 16:07:31 +0200
Subject: [PATCH] Last cleanups

- API: implement pagination for search endpoints
- Crawler: do not save body when code > 302
- Scripts: add stop.sh
---
 README.md                   | 10 ++++--
 internal/api/api.go         | 61 ++++++++++++++++++++++++++++++++++++-
 internal/crawler/crawler.go |  2 ++
 scripts/stop.sh             |  3 ++
 4 files changed, 73 insertions(+), 3 deletions(-)
 create mode 100755 scripts/stop.sh

diff --git a/README.md b/README.md
index 544e36b..cec6b2e 100644
--- a/README.md
+++ b/README.md
@@ -50,7 +50,13 @@ this will schedule given URL for crawling.
 
 ## How to view results
 
-At the moment there is no Trandoshan dashboard.
-You can use the Kibana dashboard available at http://localhost:15004.
+## Using trandoshanctl
 
+```sh
+trandoshanctl search <term>
+```
+
+## Using kibana
+
+You can use the Kibana dashboard available at http://localhost:15004.
 You will need to create an index pattern named 'resources', and when it asks for the time field, choose 'time'.
diff --git a/internal/api/api.go b/internal/api/api.go
index dab14da..4eb534a 100644
--- a/internal/api/api.go
+++ b/internal/api/api.go
@@ -14,13 +14,28 @@ import (
 	"github.com/rs/zerolog/log"
 	"github.com/urfave/cli/v2"
 	"net/http"
+	"strconv"
 	"time"
 )
 
 var (
 	resourcesIndex = "resources"
+
+	paginationPageHeader     = "X-Pagination-Page"
+	paginationSizeHeader     = "X-Pagination-Size"
+	paginationCountHeader    = "X-Pagination-Count"
+	paginationPageQueryParam = "pagination-page"
+	paginationSizeQueryParam = "pagination-size"
+
+	defaultPaginationSize = 50
+	maxPaginationSize     = 100
 )
 
+type pagination struct {
+	page int
+	size int
+}
+
 // Represent a resource in elasticsearch
 type resourceIndex struct {
 	URL   string    `json:"url"`
@@ -115,11 +130,26 @@ func searchResources(es *elastic.Client) echo.HandlerFunc {
 			return c.NoContent(http.StatusUnprocessableEntity)
 		}
 
-		// Perform the search request.
+		// Acquire pagination
+		p := readPagination(c)
+		from := (p.page - 1) * p.size
+
+		// Build up search query
 		query := buildSearchQuery(string(b), c.QueryParam("keyword"))
+
+		// Get total count
+		totalCount, err := es.Count(resourcesIndex).Query(query).Do(context.Background())
+		if err != nil {
+			log.Err(err).Msg("Error while counting on ES")
+			return c.NoContent(http.StatusInternalServerError)
+		}
+
+		// Perform the search request.
 		res, err := es.Search().
 			Index(resourcesIndex).
 			Query(query).
+			From(from).
+			Size(p.size).
 			Do(context.Background())
 		if err != nil {
 			log.Err(err).Msg("Error while searching on ES")
@@ -142,6 +172,9 @@ func searchResources(es *elastic.Client) echo.HandlerFunc {
 			resources = append(resources, resource)
 		}
 
+		// Write pagination
+		writePagination(c, p, totalCount)
+
 		return c.JSON(http.StatusOK, resources)
 	}
 }
@@ -238,3 +271,29 @@ func setupElasticSearch(ctx context.Context, es *elastic.Client) error {
 
 	return nil
 }
+
+func readPagination(c echo.Context) pagination {
+	paginationPage, err := strconv.Atoi(c.QueryParam(paginationPageQueryParam))
+	if err != nil {
+		paginationPage = 1
+	}
+	paginationSize, err := strconv.Atoi(c.QueryParam(paginationSizeQueryParam))
+	if err != nil {
+		paginationSize = defaultPaginationSize
+	}
+	// Prevent too much results from being returned
+	if paginationSize > maxPaginationSize {
+		paginationSize = maxPaginationSize
+	}
+
+	return pagination{
+		page: paginationPage,
+		size: paginationSize,
+	}
+}
+
+func writePagination(c echo.Context, p pagination, totalCount int64) {
+	c.Response().Header().Set(paginationPageHeader, strconv.Itoa(p.page))
+	c.Response().Header().Set(paginationSizeHeader, strconv.Itoa(p.size))
+	c.Response().Header().Set(paginationCountHeader, strconv.FormatInt(totalCount, 10))
+}
diff --git a/internal/crawler/crawler.go b/internal/crawler/crawler.go
index 7e4ce65..6c3df52 100644
--- a/internal/crawler/crawler.go
+++ b/internal/crawler/crawler.go
@@ -129,6 +129,8 @@ func crawURL(httpClient *fasthttp.Client, url string, allowedContentTypes []stri
 	}
 
 	switch code := resp.StatusCode(); {
+	case code > 302:
+		return "", fmt.Errorf("non-managed error code %d", code)
 	// follow redirect
 	case code == 301 || code == 302:
 		if location := string(resp.Header.Peek("Location")); location != "" {
diff --git a/scripts/stop.sh b/scripts/stop.sh
new file mode 100755
index 0000000..4c7d11f
--- /dev/null
+++ b/scripts/stop.sh
@@ -0,0 +1,3 @@
+#!/bin/bash
+
+./scripts/exec.sh stop "$@"