From a5051054eed3f342c126d9a6bf8f63a624a69e13 Mon Sep 17 00:00:00 2001
From: Eric Joanis <eric.joanis@nrc-cnrc.gc.ca>
Date: Wed, 5 Feb 2025 15:25:51 -0500
Subject: [PATCH] perf: cache the bundles after we fetched them once (#260)

* perf: cache the bundles once we fetched them once

* test: better test coverage for bundle caching

* test: exercise the path where network times out
---
 readalongs/text/make_package.py | 37 ++++++++++++++++++++-------------
 test/test_api.py                | 33 +++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 14 deletions(-)

diff --git a/readalongs/text/make_package.py b/readalongs/text/make_package.py
index 69105310..53a13a63 100644
--- a/readalongs/text/make_package.py
+++ b/readalongs/text/make_package.py
@@ -150,6 +150,10 @@ def fetch_bundle_file(url, filename, prev_status_code):
 
 _prev_js_status_code: Any = None
 _prev_fonts_status_code: Any = None
+# Cache the bundle contents so we don't fetch it more than once when running a process
+# that might generate several HTML files via the API, e.g., the EveryVoice demo app.
+js_bundle_contents = None
+fonts_bundle_contents = None
 
 
 def create_web_component_html(
@@ -160,24 +164,29 @@ def create_web_component_html(
     subheader=DEFAULT_SUBHEADER,
     theme="light",
 ) -> str:
-    global _prev_js_status_code
-    _prev_js_status_code, js_raw = fetch_bundle_file(
-        JS_BUNDLE_URL, "bundle.js", _prev_js_status_code
-    )
-
-    global _prev_fonts_status_code
-    if _prev_fonts_status_code is None and _prev_js_status_code != 200:
-        # If fetching bundle.js failed, don't bother trying bundle.css
-        _prev_fonts_status_code = _prev_js_status_code
-    _prev_fonts_status_code, fonts_raw = fetch_bundle_file(
-        FONTS_BUNDLE_URL, "bundle.css", _prev_fonts_status_code
-    )
+    global js_bundle_contents
+    if js_bundle_contents is None:
+        global _prev_js_status_code
+        _prev_js_status_code, js_bundle_contents = fetch_bundle_file(
+            JS_BUNDLE_URL, "bundle.js", _prev_js_status_code
+        )
+        js_bundle_contents = js_bundle_contents
+
+    global fonts_bundle_contents
+    if fonts_bundle_contents is None:
+        global _prev_fonts_status_code
+        if _prev_fonts_status_code is None and _prev_js_status_code != 200:
+            # If fetching bundle.js failed, don't bother trying bundle.css
+            _prev_fonts_status_code = _prev_js_status_code
+        _prev_fonts_status_code, fonts_bundle_contents = fetch_bundle_file(
+            FONTS_BUNDLE_URL, "bundle.css", _prev_fonts_status_code
+        )
 
     return BASIC_HTML.format(
         ras=encode_from_path(ras_path),
         audio=encode_from_path(audio_path),
-        js=js_raw,
-        fonts=fonts_raw,
+        js=js_bundle_contents,
+        fonts=fonts_bundle_contents,
         title=title,
         header=header,
         subheader=subheader,
diff --git a/test/test_api.py b/test/test_api.py
index 70a1bfdc..d3779b00 100755
--- a/test/test_api.py
+++ b/test/test_api.py
@@ -155,6 +155,13 @@ def test_convert_to_readalong(self):
         self.assertEqual(readalong, align_result)
 
     def test_convert_to_offline_html(self):
+        import readalongs.text.make_package as make_package
+
+        # We want to exercise caching of the bundles, but first we need to uncache
+        # them in case some other test case already cached them.
+        make_package.fonts_bundle_contents = None
+        make_package.js_bundle_contents = None
+
         html, _ = api.convert_prealigned_text_to_offline_html(
             self.sentences_to_convert,
             str(self.data_dir / "noise.mp3"),
@@ -171,6 +178,32 @@ def test_convert_to_offline_html(self):
         self.assertIn("<span slot='read-along-header'>", html)
         self.assertIn("<span slot='read-along-subheader'>by Jove!</span>", html)
 
+        # Make sure the bundles got cached
+        self.assertIsNotNone(make_package.fonts_bundle_contents)
+        self.assertIsNotNone(make_package.js_bundle_contents)
+
+        # And convert again, this time it's going to use the cached bundles.
+        html2, _ = api.convert_prealigned_text_to_offline_html(
+            self.sentences_to_convert,
+            str(self.data_dir / "noise.mp3"),
+            subheader="by Jove!",
+        )
+        self.assertEqual(html, html2)
+
+        # Once more, this time pretend we could not fetch the first bundle
+        make_package.fonts_bundle_contents = None
+        make_package.js_bundle_contents = None
+        make_package._prev_js_status_code = "TIMEOUT"
+        make_package._prev_fonts_status_code = None
+        _, _ = api.convert_prealigned_text_to_offline_html(
+            self.sentences_to_convert,
+            str(self.data_dir / "noise.mp3"),
+            subheader="by Jove!",
+        )
+        self.assertEqual(make_package._prev_fonts_status_code, "TIMEOUT")
+        self.assertIsNotNone(make_package.fonts_bundle_contents)
+        self.assertIsNotNone(make_package.js_bundle_contents)
+
 
 if __name__ == "__main__":
     main()