parse episodes more safely

4 files changed, 78 insertions, 2 deletions

diff --git a/main.go b/main.go
index 2304b4b..5f990bb 100644
--- a/main.go
+++ b/main.go
@@ -133,7 +133,10 @@ func populateFeed(feed *feeds.Feed, page []byte) (err error) {
 		if len(episodeUrlRe.FindAllSubmatch(episode, -1)) > 1 {
 			return errBadEpisode
 		}
-		url, _ := parseSingle(episode, episodeUrlRe)
+		url, err := parseSingle(episode, episodeUrlRe)
+		if err != nil {
+			return errBadEpisode
+		}
 		episodeUrl := urlPrefix + string(url)
 		title, _ := parseSingle(episode, episodeTitleRe)
 		episodeTitle := string(title)
@@ -148,7 +151,7 @@ func populateFeed(feed *feeds.Feed, page []byte) (err error) {
 			Created:   date,
 		})
 	}
-	return nil
+	return
 }
 
 func parse(src []byte, re *regexp.Regexp, n int) (out [][]byte, err error) {
diff --git a/main_test.go b/main_test.go
index 2d520cb..d00a6ae 100644
--- a/main_test.go
+++ b/main_test.go
@@ -27,6 +27,7 @@ import (
 	"os/exec"
 	"path/filepath"
 	"regexp"
+	"strconv"
 	"strings"
 	"sync"
 	"testing"
@@ -91,6 +92,21 @@ func TestFeed(t *testing.T) {
 	assertGolden(t, actual, golden)
 }
 
+func TestBadEpisode(t *testing.T) {
+	feed := &feeds.Feed{
+		Link: &feeds.Link{Href: "http://www.radiorus.ru/brand/57083/episodes"},
+	}
+
+	for i := 0; i <= 1; i++ {
+		page := helperLoadBytes(t, "episodes.badep."+strconv.Itoa(i))
+		page = cleanText(page)
+
+		if err := populateFeed(feed, page); err != errBadEpisode {
+			t.Error("for sample", i, "want:", errBadEpisode, "got:", err)
+		}
+	}
+}
+
 func TestFindEpisodes(t *testing.T) {
 	var tests = []string{
 		"episodes",
diff --git a/testdata/episodes.badep.0 b/testdata/episodes.badep.0
new file mode 100644
index 0000000..13d82f2
--- /dev/null
+++ b/testdata/episodes.badep.0
@@ -0,0 +1,23 @@
+    <h2><a href="/brand/57083">&quot;Аэростат&quot;</a></h2>
+            <div class="brand-promo__header">
+            <img src="https://cdn-st4.rtr-vesti.ru/vh/pictures/xw/124/617/1.jpg" alt='&quot;Аэростат&quot;'>
+                                                                                            <div class="brand__list--wrap--item">
+                                    <div class="photo-wrap">
+                                                                                    <a href="/brand/57083/episode/2237781" class="photo-wrap__link">
+                                                <img src="https://cdn-st1.rtr-vesti.ru/vh/pictures/bw/183/795/6.jpg"
+                                                     alt="The Cure | momento mori, Malaysia">
+                                                <!-- если фото > 1 [ -->
+                                                                                            <!-- ] если фото > 1 -->
+                                            </a>
+                                                                            </div>
+                                    <a href="/brand/57083/episode/2237781"
+                                       class="brand-time brand-menu-link">19.01.2020 в 14:10</a>
+                                    <a href="/brand/57083/episode/2237781" class="more-info brand-menu-link">Подробнее</a>
+
+                                                                        <!-- если есть аудио[ -->
+                                        <div class="audio-count" data-type="audio"
+                                             data-id="2466052"></div>
+                                        <div class="add-to-list"></div>
+                                        <!-- ] если есть аудио -->
+                                    
+                                </div>
diff --git a/testdata/episodes.badep.1 b/testdata/episodes.badep.1
new file mode 100644
index 0000000..b29fcc8
--- /dev/null
+++ b/testdata/episodes.badep.1
@@ -0,0 +1,34 @@
+    <h2><a href="/brand/57083">&quot;Аэростат&quot;</a></h2>
+            <div class="brand-promo__header">
+            <img src="https://cdn-st4.rtr-vesti.ru/vh/pictures/xw/124/617/1.jpg" alt='&quot;Аэростат&quot;'>
+                                                                                                                        <div class="brand__list--wrap--item">
+                                    <div class="photo-wrap">
+                                                                                    <a href="/brand/57083/episode/2237849" class="photo-wrap__link">
+                                                <img src="https://cdn-st2.rtr-vesti.ru/vh/pictures/bw/207/010/1.jpg"
+                                                     alt="Ансамбль &quot;Pied Pipers&quot; | public domain">
+                                                <!-- если фото > 1 [ -->
+                                                                                            <!-- ] если фото > 1 -->
+                                            </a>
+                                                                            </div>
+                                    <a href="/brand/57083/episode/2237849"
+                                       class="brand-time brand-menu-link">26.01.2020 в 14:10</a>
+                                    <a href="/brand/57083/episode/2237849" class="title brand-menu-link">Новые имена 27</a>
+                                                                                    <a href="/brand/57083/episode/2237781" class="photo-wrap__link">
+                                                <img src="https://cdn-st1.rtr-vesti.ru/vh/pictures/bw/183/795/6.jpg"
+                                                     alt="The Cure | momento mori, Malaysia">
+                                                <!-- если фото > 1 [ -->
+                                                                                            <!-- ] если фото > 1 -->
+                                            </a>
+                                                                            </div>
+                                    <a href="/brand/57083/episode/2237781"
+                                       class="brand-time brand-menu-link">19.01.2020 в 14:10</a>
+                                    <a href="/brand/57083/episode/2237781" class="title brand-menu-link">The Cure</a>
+                                    <a href="/brand/57083/episode/2237781" class="more-info brand-menu-link">Подробнее</a>
+
+                                                                        <!-- если есть аудио[ -->
+                                        <div class="audio-count" data-type="audio"
+                                             data-id="2466052"></div>
+                                        <div class="add-to-list"></div>
+                                        <!-- ] если есть аудио -->
+                                    
+                                </div>