chore: fix the JS test suite not to rely on npx --package

Author: honzajavorek

sources/academy/webscraping/scraping_basics_javascript/exercises/products.json

@@ -1,115 +1,114 @@
 setup_file() {
   cd "$BATS_TEST_DIRNAME"
-  export npm_config_yes=true
+  npm init --yes
+  npm install cheerio crawlee
+}
+
+teardown() {
+  rm -rf products.json storage dataset.json
 }
 
 teardown_file() {
-  rm -rf storage node_modules package.json package-lock.json dataset.json
+  rm -rf node_modules package.json package-lock.json
 }
 
-# retry_run() {
-#   for attempt in 1 2 3; do
-#     run "$@"
-#     (( status == 0 )) && return 0
-#     sleep 1
-#   done
-#   return "$status"
-# }
 
 @test "outputs the HTML with Star Wars products" {
-  run npx node lego.mjs
+  run node lego.mjs
 
   [[ "$output" == *"Millennium Falcon"* ]]
 }
 
 @test "counts the number of F1 Academy teams" {
-  run npx --package=cheerio node f1academy_teams.mjs
+  run node f1academy_teams.mjs
 
   [[ "$output" == "6" ]]
 }
 
 @test "counts the number of F1 Academy drivers" {
-  run npx --package=cheerio node f1academy_drivers.mjs
+  run node f1academy_drivers.mjs
 
   [[ "$output" == "18" ]]
 }
 
 @test "lists African countries" {
-  run npx --package=cheerio node wikipedia_countries.mjs
+  run node wikipedia_countries.mjs
 
   [[ "$output" == *$'Comoros\nDemocratic Republic of the Congo\n'* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "lists African countries with a single selector" {
-  run npx --package=cheerio node wikipedia_countries_single_selector.mjs
+  run node wikipedia_countries_single_selector.mjs
 
   [[ "$output" == *$'Comoros\nDemocratic Republic of the Congo\n'* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "lists Guardian F1 article titles" {
-  run npx --package=cheerio node guardian_f1_titles.mjs
+  run node guardian_f1_titles.mjs
 
   [[ "$output" == *' F1 '* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "prints warehouse stock counts" {
-  run npx --package=cheerio node warehouse_units.mjs
+  run node warehouse_units.mjs
 
   [[ "$output" == *$'JBL Flip 4 Waterproof Portable Bluetooth Speaker | 672\n'* ]]
   [[ "$output" == *$'Sony XBR-950G BRAVIA 4K HDR Ultra HD TV | 77\n'* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "prints warehouse stock counts using regex" {
-  run npx --package=cheerio node warehouse_units_regex.mjs
+  run node warehouse_units_regex.mjs
 
   [[ "$output" == *$'JBL Flip 4 Waterproof Portable Bluetooth Speaker | 672\n'* ]]
   [[ "$output" == *$'Sony XBR-950G BRAVIA 4K HDR Ultra HD TV | 77\n'* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "prints Guardian F1 titles with publish dates" {
-  run npx --package=cheerio node guardian_publish_dates.mjs
+  run node guardian_publish_dates.mjs
 
   [[ "$output" == *' F1 '* ]]
   [[ "$output" == *' | Sun '* ]]  # has info about date, Sundays are very likely
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "filters products from JSON" {
-  run npx node process_products_json.mjs
+  echo '[{"title":"Premium Speakers","minPrice":75000,"price":75000},{"title":"Budget Headphones","minPrice":25000,"price":25000}]' > products.json
+
+  run node process_products_json.mjs
 
   [[ "$output" == "{ title: 'Premium Speakers', minPrice: 75000, price: 75000 }" ]]
 }
 
 @test "lists Wikipedia country links" {
-  run npx --package=cheerio node wikipedia_country_links.mjs
+  run node wikipedia_country_links.mjs
 
   [[ "$output" == *$'https://en.wikipedia.org/wiki/Algeria\nhttps://en.wikipedia.org/wiki/Angola\n'* ]]
   [[ "$output" == *$'https://en.wikipedia.org/wiki/R%C3%A9union\n'* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "lists Guardian F1 article links" {
-  run npx --package=cheerio node guardian_f1_links.mjs
+  run node guardian_f1_links.mjs
 
   [[ "$output" == *'https://www.theguardian.com/sport/'* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "prints Wikipedia calling codes" {
-  run npx --package=cheerio node wikipedia_calling_codes.mjs
+  run node wikipedia_calling_codes.mjs
 
   [[ "$output" == *$'https://en.wikipedia.org/wiki/Comoros +269\n'* ]]
   [[ "$output" == *$'https://en.wikipedia.org/wiki/Sahrawi_Arab_Democratic_Republic null\n'* ]]
   [[ $(echo "$output" | wc -l) -gt 5 ]]
 }
 
 @test "lists Guardian F1 authors" {
-  run npx --package=cheerio node guardian_f1_authors.mjs
+  run node guardian_f1_authors.mjs
 
   [[ "$output" == *' F1 '* ]]
   [[ "$output" == *'Giles Richards: '* ]]  # writes most of them (we'll have to change this if they fire'him)
@@ -118,37 +117,33 @@ teardown_file() {
 }
 
 @test "lists npm LLM packages" {
-  run npx --package=cheerio node npm_llm_packages.mjs
+  run node npm_llm_packages.mjs
 
   (( status == 0 ))
   [[ -n "$output" ]]
 }
 
 @test "finds the shortest CNN sports article" {
-  run npx --package=cheerio node cnn_sports_shortest_article.mjs
+  run node cnn_sports_shortest_article.mjs
 
   [[ "$output" == 'https://edition.cnn.com/'* ]]
 }
 
 @test "scrapes F1 Academy driver details with Crawlee" {
-  npm init --yes
-  npm install crawlee
   run node crawlee_f1_drivers.mjs
 
   (( status == 0 ))
-  [[ -n "$output" || -f dataset.json ]]
+  [[ -f dataset.json ]]
   [[ $(cat dataset.json | jq '. | length') == "18" ]]
   [[ $(cat dataset.json | jq -c '.[0] | keys') == '["dob","instagram_url","name","nationality","team","url"]' ]]
   [[ $(cat dataset.json | jq '.[].url') == *"https://www.f1academy.com/Racing-Series/Drivers/"* ]]
 }
 
 @test "scrapes Netflix ratings with Crawlee" {
-  npm init --yes
-  npm install crawlee
   run node crawlee_netflix_ratings.mjs
 
   (( status == 0 ))
-  [[ -n "$output" || -f dataset.json ]]
+  [[ -f dataset.json ]]
   [[ $(cat dataset.json | jq '. | length') == "10" ]]
   [[ $(cat dataset.json | jq -c '.[0] | keys') == '["url","title","rating"]' ]]
   [[ $(cat dataset.json | jq '.[].url') == *"https://www.imdb.com/title/"* ]]