fix: validate AniList reports against version two schema

Checks/progress-reporting.mm

@@ -37,25 +37,37 @@ static NSString *nativePath(const std::string &path) {
     return [NSString stringWithUTF8String:path.c_str()];
 }
 
-static NSArray<NSDictionary *> *records(const std::string &path) {
+static NSArray<NSDictionary *> *records(const std::string &path, int version = 1) {
     NSString *text = [NSString stringWithContentsOfFile:nativePath(path)
         encoding:NSUTF8StringEncoding error:nullptr];
     require(text != nil && [text hasSuffix:@"\n"], "finished JSONL has a complete final line");
     NSMutableArray<NSDictionary *> *result = [NSMutableArray array];
     NSArray<NSString *> *lines = [text componentsSeparatedByString:@"\n"];
-    NSSet *keys = [NSSet setWithArray:@[@"version", @"event", @"media_url", @"position_seconds",
-                                       @"duration_seconds", @"completion_eligible"]];
+    NSSet *keys = [NSSet setWithArray:@[@"version", @"event", version == 2 ? @"media_identity" : @"media_url",
+                                       @"position_seconds", @"duration_seconds", @"completion_eligible"]];
     for (NSUInteger i = 0; i + 1 < lines.count; ++i) {
         id object = [NSJSONSerialization JSONObjectWithData:[lines[i] dataUsingEncoding:NSUTF8StringEncoding]
                                                   options:0 error:nullptr];
         require([object isKindOfClass:[NSDictionary class]], "consumer decodes every JSONL record");
         NSDictionary *row = object;
         require([[NSSet setWithArray:row.allKeys] isEqual:keys], "records have exactly the six specified fields");
-        require([row[@"version"] isKindOfClass:[NSNumber class]] && [row[@"version"] intValue] == 1,
-                "schema version is numeric one");
+        require([row[@"version"] isKindOfClass:[NSNumber class]] && [row[@"version"] intValue] == version,
+                "schema version matches the expected source contract");
         require([@[@"progress", @"ended", @"closed"] containsObject:row[@"event"]], "known event type");
-        require([row[@"media_url"] isKindOfClass:[NSString class]] &&
-                ProgressReporter::validMediaURL([row[@"media_url"] UTF8String]), "HTTP(S) consumer identity");
+        if (version == 2) {
+            NSDictionary *identity = row[@"media_identity"];
+            require([identity isKindOfClass:[NSDictionary class]] &&
+                    [[NSSet setWithArray:identity.allKeys] isEqual:
+                        [NSSet setWithArray:@[@"provider", @"media_id", @"episode"]]] &&
+                    [identity[@"provider"] isKindOfClass:[NSString class]] &&
+                    [identity[@"media_id"] isKindOfClass:[NSString class]] &&
+                    [identity[@"episode"] isKindOfClass:[NSNumber class]], "structured episode consumer identity");
+            NSString *key = [NSString stringWithFormat:@"%@:%@:%@", identity[@"provider"], identity[@"media_id"], identity[@"episode"]];
+            require(ProgressReporter::validMediaIdentity(key.UTF8String), "canonical episode consumer identity");
+        } else {
+            require([row[@"media_url"] isKindOfClass:[NSString class]] &&
+                    ProgressReporter::validMediaURL([row[@"media_url"] UTF8String]), "HTTP(S) consumer identity");
+        }
         require([row[@"position_seconds"] isKindOfClass:[NSNumber class]] &&
                 std::isfinite([row[@"position_seconds"] doubleValue]) &&
                 [row[@"position_seconds"] doubleValue] >= 0, "position is finite and nonnegative");
@@ -316,7 +328,7 @@ int main(int argc, char **argv) {
             reporter.finish();
             require(reporter.error().empty(), "AniList progress drains successfully");
         }
-        rows = records(anilistPath);
+        rows = records(anilistPath, 2);
         require(rows.count == 2, "AniList initial and closed snapshots");
         for (NSDictionary *row in rows) {
             require([row[@"version"] intValue] == 2 && row[@"media_url"] == nil &&

Checks/zoko.swift

@@ -40,9 +40,13 @@ struct ZokoCheck {
         try require(source.title == "AniList 1 · Episode 1 (sub)" && source.subtitles.count == 1 && source.subtitles[0].isDefault,
                     "honest fallback title and VTT metadata")
         try require(source.headers == ZokoResolver.referer && source.subtitles[0].headers == ZokoResolver.referer, "only fixed public Referer scoped to declared CDN")
+        let semicolonless = try page(config).replacingOccurrences(of: "';</script>", with: "'</script>")
+        let deployedSource = try ZokoResolver.parse(page: semicolonless, mediaID: "1", episode: 1)
+        try require(deployedSource.identity == source.identity, "deployed semicolonless static assignment is accepted")
         let missing = "<span>// Error</span><span>404</span><h1>Not <span>found.</span></h1><p>We couldn't find anything to play here.</p>"
         try expect(.unavailable) { _ = try ZokoResolver.parse(page: missing, mediaID: "1", episode: 27) }
-        for protected in ["<title>Just a moment</title>", "<script src='/cdn-cgi/challenge-platform/x'></script>"] {
+        for protected in ["<title>Just a moment</title>", "<script src='/cdn-cgi/challenge-platform/x'></script>",
+                          "<h1>Sorry, you have been blocked</h1>"] {
             try expect(.protection) { _ = try ZokoResolver.parse(page: protected, mediaID: "1", episode: 1) }
         }
         for malformed in ["<html></html>", "window.__P='broken';", try page(config) + page(config), String(repeating: "x", count: ZokoResolver.bodyLimit + 1)] {

Sources/App/ZokoResolver.swift

@@ -76,7 +76,7 @@ enum ZokoResolver {
 
     static func parse(page: String, mediaID: String, episode: Int) throws -> ResolvedSource {
         guard page.utf8.count <= bodyLimit, ResolvedSource.validAniListIdentity(mediaID: mediaID, episode: episode) else { throw ZokoError.malformed }
-        let pattern = #"window\.__P\s*=\s*[\"']([A-Za-z0-9+/=]+)[\"']\s*;"#
+        let pattern = #"window\.__P\s*=\s*[\"']([A-Za-z0-9+/=]+)[\"']\s*;?\s*</script\s*>"#
         let regex = try NSRegularExpression(pattern: pattern)
         let matches = regex.matches(in: page, range: NSRange(page.startIndex..., in: page))
         guard matches.count == 1, let match = matches.first, let range = Range(match.range(at: 1), in: page) else {
@@ -84,7 +84,8 @@ enum ZokoResolver {
             let lower = page.lowercased()
             let visible = lower.replacingOccurrences(of: "<[^>]*>", with: " ", options: .regularExpression)
             if visible.contains("error"), visible.contains("404"), visible.contains("couldn't find anything to play here") { throw ZokoError.unavailable }
-            if lower.contains("cf-chl-") || lower.contains("challenge-platform") || lower.contains("just a moment") || lower.contains("captcha") { throw ZokoError.protection }
+            if lower.contains("cf-chl-") || lower.contains("challenge-platform") || lower.contains("just a moment") ||
+                lower.contains("captcha") || lower.contains("sorry, you have been blocked") { throw ZokoError.protection }
             throw ZokoError.malformed
         }
         guard let encoded = Data(base64Encoded: String(page[range])), !encoded.isEmpty else { throw ZokoError.malformed }