RomanNum3ral
/
android-reverse-engineering-skill
mirror of https://github.com/SimoneAvogadro/android-reverse-engineering-skill.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

feat(clone-app): scrape screenshots, feature graphic, description from Play

This commit is contained in:
fatih.bulut 2026-06-25 02:09:33 +03:00
parent e3a29974f4
commit aad064d220
3 changed files with 26 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
plugins/clone-app/skills/clone-app/scripts/scrape-play-store.py
View File

@ -7,7 +7,8 @@ Falls back to light regex for installs/updated which aren't always in ld+json.
import sys, json, re, argparse, urllib.request, ssl, subprocess, shutil import sys, json, re, argparse, urllib.request, ssl, subprocess, shutil
KEYS = ["package", "title", "rating", "rating_count", "installs", KEYS = ["package", "title", "rating", "rating_count", "installs",
"category", "developer", "updated", "source"] "category", "developer", "updated", "source",
"screenshot_urls", "feature_graphic", "description"]
UA = "Mozilla/5.0" UA = "Mozilla/5.0"
@ -61,6 +62,21 @@ def parse(html, package):
if ar.get("ratingCount") is not None: if ar.get("ratingCount") is not None:
try: out["rating_count"] = int(ar["ratingCount"]) try: out["rating_count"] = int(ar["ratingCount"])
except Exception: pass except Exception: pass
out["description"] = data.get("description")
img = data.get("image")
if isinstance(img, list):
img = img[0] if img else None
out["feature_graphic"] = img
shots = data.get("screenshot") or []
if isinstance(shots, dict):
shots = [shots]
urls = []
for s in shots:
if isinstance(s, str):
urls.append(s)
elif isinstance(s, dict) and s.get("url"):
urls.append(s["url"])
out["screenshot_urls"] = urls
break break
# installs (e.g. "1,000,000+"). Play wraps the count and the "Downloads" # installs (e.g. "1,000,000+"). Play wraps the count and the "Downloads"

5
plugins/clone-app/tests/fixtures/play-sample.html vendored
View File

@ -3,7 +3,10 @@
<script type="application/ld+json" nonce="5kflo12lNMmEYHKRtMGqHQ"> <script type="application/ld+json" nonce="5kflo12lNMmEYHKRtMGqHQ">
{"@context":"https://schema.org","@type":"SoftwareApplication","name":"Example App","author":{"name":"Example Studio"}, {"@context":"https://schema.org","@type":"SoftwareApplication","name":"Example App","author":{"name":"Example Studio"},
"aggregateRating":{"ratingValue":"4.3","ratingCount":"12873"}, "aggregateRating":{"ratingValue":"4.3","ratingCount":"12873"},
"applicationCategory":"GAME_PUZZLE"} "applicationCategory":"GAME_PUZZLE",
"description":"An example puzzle app.",
"image":"https://play-lh.googleusercontent.com/feature.png",
"screenshot":["https://play-lh.googleusercontent.com/shot1.png","https://play-lh.googleusercontent.com/shot2.png"]}
</script> </script>
</head><body> </head><body>
<div class="reAt0">Updated on</div><div class="xg1aie">Jun 1, 2026</div> <div class="reAt0">Updated on</div><div class="xg1aie">Jun 1, 2026</div>

6
plugins/clone-app/tests/test-scrape-play-store.py
View File

@ -25,8 +25,12 @@ def main():
check("installs", d["installs"] == "1,000,000+") check("installs", d["installs"] == "1,000,000+")
check("updated", d["updated"] == "Jun 1, 2026") check("updated", d["updated"] == "Jun 1, 2026")
check("source", d["source"] == "google-play") check("source", d["source"] == "google-play")
check("description", d["description"] == "An example puzzle app.")
check("feature_graphic", d["feature_graphic"] == "https://play-lh.googleusercontent.com/feature.png")
check("screenshot count", len(d["screenshot_urls"]) == 2)
check("screenshot url", d["screenshot_urls"][0] == "https://play-lh.googleusercontent.com/shot1.png")
# all expected keys present even if null # all expected keys present even if null
for k in ["package","title","rating","rating_count","installs","category","developer","updated","source"]: for k in ["package","title","rating","rating_count","installs","category","developer","updated","source","screenshot_urls","feature_graphic","description"]:
check(f"key present: {k}", k in d) check(f"key present: {k}", k in d)
sys.exit(1 if fails else 0) sys.exit(1 if fails else 0)