feat(parser): enhance user agent parsing logic and add tests for inva… (#76)

BenjaminAbt · web-flow · commit 30066004f430 · 2025-08-25T12:59:53.000+02:00
diff --git a/README.md b/README.md
@@ -122,21 +122,21 @@ AMD Ryzen 9 9950X, 1 CPU, 32 logical and 16 physical cores
 Job=ShortRun  IterationCount=3  LaunchCount=1
 WarmupCount=3
 
-| Method             | Categories | Data         | Mean            | Error            | StdDev         | Ratio     | RatioSD | Gen0     | Gen1     | Gen2     | Allocated  | Alloc Ratio |
-|------------------- |----------- |------------- |----------------:|-----------------:|---------------:|----------:|--------:|---------:|---------:|---------:|-----------:|------------:|
-| MyCSharp           | Basic      | Chrome Win10 |       936.44 ns |       131.253 ns |       7.194 ns |      1.00 |    0.01 |   0.0029 |        - |        - |       48 B |        1.00 |
-| UAParser           | Basic      | Chrome Win10 | 9,512,347.40 ns | 3,961,045.109 ns | 217,118.249 ns | 10,158.42 |  211.89 | 656.2500 | 546.8750 | 109.3750 | 11523315 B |  240,069.06 |
-| DeviceDetector.NET | Basic      | Chrome Win10 | 5,428,530.73 ns | 5,276,988.556 ns | 289,249.550 ns |  5,797.23 |  270.29 | 296.8750 | 125.0000 |  31.2500 |  5002239 B |  104,213.31 |
-|                    |            |              |                 |                  |                |           |         |          |          |          |            |             |
-| MyCSharp           | Basic      | Google-Bot   |       165.66 ns |        21.926 ns |       1.202 ns |      1.00 |    0.01 |        - |        - |        - |          - |          NA |
-| UAParser           | Basic      | Google-Bot   | 9,737,403.12 ns | 2,336,698.462 ns | 128,082.328 ns | 58,781.92 |  764.74 | 671.8750 | 656.2500 | 109.3750 | 11877003 B |          NA |
-| DeviceDetector.NET | Basic      | Google-Bot   | 6,331,960.42 ns | 1,602,716.199 ns |  87,850.283 ns | 38,224.23 |  518.30 | 500.0000 |  62.5000 |        - |  8817013 B |          NA |
-|                    |            |              |                 |                  |                |           |         |          |          |          |            |             |
-| MyCSharp           | Cached     | Chrome Win10 |        26.75 ns |         3.749 ns |       0.205 ns |      1.00 |    0.01 |        - |        - |        - |          - |          NA |
-| UAParser           | Cached     | Chrome Win10 |   250,039.55 ns |     6,502.182 ns |     356.407 ns |  9,346.54 |   63.39 |   2.1973 |        - |        - |    37488 B |          NA |
-|                    |            |              |                 |                  |                |           |         |          |          |          |            |             |
-| MyCSharp           | Cached     | Google-Bot   |        19.66 ns |         4.312 ns |       0.236 ns |      1.00 |    0.01 |        - |        - |        - |          - |          NA |
-| UAParser           | Cached     | Google-Bot   |   184,991.85 ns |    46,235.986 ns |   2,534.350 ns |  9,408.77 |  148.82 |   2.6855 |        - |        - |    45857 B |          NA |
+| Method             | Categories | Data         | Mean            | Error            | StdDev         | Ratio     | RatioSD  | Gen0     | Gen1     | Gen2     | Allocated  | Alloc Ratio |
+|------------------- |----------- |------------- |----------------:|-----------------:|---------------:|----------:|---------:|---------:|---------:|---------:|-----------:|------------:|
+| MyCSharp           | Basic      | Chrome Win10 |       871.85 ns |       132.008 ns |       7.236 ns |      1.00 |     0.01 |   0.0029 |        - |        - |       48 B |        1.00 |
+| UAParser           | Basic      | Chrome Win10 | 8,901,909.90 ns | 3,411,259.484 ns | 186,982.644 ns | 10,210.80 |   199.60 | 656.2500 | 578.1250 | 109.3750 | 11523310 B |  240,068.96 |
+| DeviceDetector.NET | Basic      | Chrome Win10 | 5,391,412.50 ns | 8,253,446.769 ns | 452,399.269 ns |  6,184.14 |   451.58 | 296.8750 | 125.0000 |  31.2500 |  5002239 B |  104,213.31 |
+|                    |            |              |                 |                  |                |           |          |          |          |          |            |             |
+| MyCSharp           | Basic      | Google-Bot   |       158.80 ns |        19.584 ns |       1.073 ns |      1.00 |     0.01 |        - |        - |        - |          - |          NA |
+| UAParser           | Basic      | Google-Bot   | 9,666,739.32 ns | 7,566,085.041 ns | 414,722.653 ns | 60,873.62 | 2,289.43 | 671.8750 | 656.2500 | 109.3750 | 11876998 B |          NA |
+| DeviceDetector.NET | Basic      | Google-Bot   | 6,106,666.41 ns |   593,634.990 ns |  32,539.137 ns | 38,455.05 |   285.97 | 539.0625 | 117.1875 |  23.4375 |  8817078 B |          NA |
+|                    |            |              |                 |                  |                |           |          |          |          |          |            |             |
+| MyCSharp           | Cached     | Chrome Win10 |        26.43 ns |         0.132 ns |       0.007 ns |      1.00 |     0.00 |        - |        - |        - |          - |          NA |
+| UAParser           | Cached     | Chrome Win10 |   177,417.99 ns |    24,390.139 ns |   1,336.906 ns |  6,713.66 |    43.84 |   2.1973 |        - |        - |    37488 B |          NA |
+|                    |            |              |                 |                  |                |           |          |          |          |          |            |             |
+| MyCSharp           | Cached     | Google-Bot   |        17.03 ns |         1.835 ns |       0.101 ns |      1.00 |     0.01 |        - |        - |        - |          - |          NA |
+| UAParser           | Cached     | Google-Bot   |   129,445.13 ns |    21,319.059 ns |   1,168.570 ns |  7,599.76 |    70.93 |   2.6855 |        - |        - |    45857 B |          NA |
 ```
 
 ## Disclaimer
diff --git a/src/HttpUserAgentParser/HttpUserAgentParser.cs b/src/HttpUserAgentParser/HttpUserAgentParser.cs
@@ -11,7 +11,6 @@ namespace MyCSharp.HttpUserAgentParser;
 /// Parser logic for user agents
 /// </summary>
 public static class HttpUserAgentParser
-
 {
     /// <summary>
     /// Parses given <param name="userAgent">user agent</param>
@@ -48,7 +47,6 @@ public static HttpUserAgentInformation Parse(string userAgent)
     /// </summary>
     public static HttpUserAgentPlatformInformation? GetPlatform(string userAgent)
     {
-        // Fast, allocation-free token scan (keeps public statics untouched)
         ReadOnlySpan<char> ua = userAgent.AsSpan();
         foreach ((string Token, string Name, HttpUserAgentPlatformType PlatformType) platform in HttpUserAgentStatics.s_platformRules)
         {
@@ -78,6 +76,7 @@ public static bool TryGetPlatform(string userAgent, [NotNullWhen(true)] out Http
     public static (string Name, string? Version)? GetBrowser(string userAgent)
     {
         ReadOnlySpan<char> ua = userAgent.AsSpan();
+
         foreach ((string Name, string DetectToken, string? VersionToken) browserRule in HttpUserAgentStatics.s_browserRules)
         {
             if (!TryIndexOf(ua, browserRule.DetectToken, out int detectIndex))
@@ -86,7 +85,18 @@ public static (string Name, string? Version)? GetBrowser(string userAgent)
             }
 
             // Version token may differ (e.g., Safari uses "Version/")
-            int versionSearchStart = detectIndex;
+
+            int versionSearchStart;
+            // For rules without a specific version token, ensure pattern Token/<digits>
+            if (string.IsNullOrEmpty(browserRule.VersionToken))
+            {
+                int afterDetect = detectIndex + browserRule.DetectToken.Length;
+                if (afterDetect >= ua.Length || ua[afterDetect] != '/')
+                {
+                    // Likely a misspelling or partial token (e.g., Edgg, Oprea, Chromee)
+                    continue;
+                }
+            }
             if (!string.IsNullOrEmpty(browserRule.VersionToken))
             {
                 if (TryIndexOf(ua, browserRule.VersionToken!, out int vtIndex))
@@ -104,14 +114,14 @@ public static (string Name, string? Version)? GetBrowser(string userAgent)
                 versionSearchStart = detectIndex + browserRule.DetectToken.Length;
             }
 
-            string? version = null;
-            ua = ua.Slice(versionSearchStart);
-            if (TryExtractVersion(ua, out Range range))
+            ReadOnlySpan<char> search = ua.Slice(versionSearchStart);
+            if (TryExtractVersion(search, out Range range))
             {
-                version = ua[range].ToString();
+                string? version = search[range].ToString();
+                return (browserRule.Name, version);
             }
 
-            return (browserRule.Name, version);
+            // If we didn't find a version for this rule, try next rule
         }
 
         return null;
@@ -198,39 +208,43 @@ private static bool TryExtractVersion(ReadOnlySpan<char> haystack, out Range ran
 
         // Limit search window to avoid scanning entire UA string unnecessarily
         const int Window = 128;
-        if (haystack.Length >= Window)
+        if (haystack.Length > Window)
         {
             haystack = haystack.Slice(0, Window);
         }
 
-        int i = 0;
-        for (; i < haystack.Length; ++i)
+        // Find first digit
+        int start = -1;
+        for (int i = 0; i < haystack.Length; i++)
         {
             char c = haystack[i];
-            if (char.IsBetween(c, '0', '9'))
+            if (c >= '0' && c <= '9')
             {
+                start = i;
                 break;
             }
         }
 
-        int s = i;
-        haystack = haystack.Slice(i + 1);
-        for (i = 0; i < haystack.Length; ++i)
+        if (start < 0)
         {
-            char c = haystack[i];
-            if (!(char.IsBetween(c, '0', '9') || c == '.'))
-            {
-                break;
-            }
+            // No digit found => no version
+            return false;
         }
-        i += s + 1;     // shift back the previous domain
 
-        if (i == s)
+        // Consume digits and dots after first digit
+        int end = start + 1;
+        while (end < haystack.Length)
         {
-            return false;
+            char c = haystack[end];
+            if (!((c >= '0' && c <= '9') || c == '.'))
+            {
+                break;
+            }
+            end++;
         }
 
-        range = new Range(s, i);
+        // Create exclusive end range
+        range = new Range(start, end);
         return true;
     }
 }
diff --git a/src/HttpUserAgentParser/HttpUserAgentStatics.cs b/src/HttpUserAgentParser/HttpUserAgentStatics.cs
@@ -187,6 +187,7 @@ internal static readonly (string Name, string DetectToken, string? VersionToken)
         ("Opera", "OPR", null),
         ("Flock", "Flock", null),
         ("Edge", "Edge", null),
+        ("Edge", "EdgiOS", null),
         ("Edge", "EdgA", null),
         ("Edge", "Edg", null),
         ("Vivaldi", "Vivaldi", null),
@@ -208,7 +209,6 @@ internal static readonly (string Name, string DetectToken, string? VersionToken)
         ("Netscape", "Netscape", null),
         ("OmniWeb", "OmniWeb", null),
         ("Safari", "Version/", "Version/"),
-        ("Mozilla", "Mozilla", null),
         ("Konqueror", "Konqueror", null),
         ("iCab", "icab", null),
         ("Lynx", "Lynx", null),
diff --git a/tests/HttpUserAgentParser.UnitTests/HttpUserAgentParserTests.cs b/tests/HttpUserAgentParser.UnitTests/HttpUserAgentParserTests.cs
@@ -173,4 +173,54 @@ public void BotTests(string ua, string name)
         Assert.False(uaInfo.IsMobile());
         Assert.True(uaInfo.IsRobot());
     }
+
+    [Theory]
+    [InlineData("")]
+    [InlineData("???")]
+    [InlineData("NotAUserAgent")]
+    [InlineData("Mozilla")]
+    [InlineData("Mozilla/")]
+    [InlineData("()")]
+    [InlineData("UserAgent/")]
+    [InlineData("Bot/123 (")]
+    [InlineData("123456")]
+    [InlineData("curl")]
+    [InlineData("invalid/useragent")]
+    [InlineData("Mozilla (Windows)")]
+    [InlineData("Chrome/ABC")]
+    [InlineData(";;!!##")]
+    [InlineData("Safari/ ")]
+    [InlineData("Opera( )")]
+    [InlineData("Mozilla/5.0 (X11; ) Gecko")]
+    [InlineData("FakeUA/1.0 (Test)???")]
+    [InlineData("Mozilla/ (iPhone; U; CPU iPhone OS like Mac OS X) AppleWebKit/ (KHTML, like Gecko) Version/ Mobile/ Safari/")]
+    [InlineData("Mozzila/5.0 (Windows NT 10.0; Win64; x64)")]
+    [InlineData("Chorme/91.0.4472.124 (Windows NT 10.0; Win64; x64)")]
+    [InlineData("FireFoxx/89.0 (Macintosh; Intel Mac OS X 10_15_7)")]
+    [InlineData("Safarii/14.1 (iPhone; CPU iPhone OS 14_6 like Mac OS X)")]
+    [InlineData("InternetExploder/11.0 (Windows NT 6.1; WOW64)")]
+    [InlineData("Bravee/1.25.72 (Windows NT 10.0; Win64; x64)")]
+    [InlineData("Mozzila/5.0 (X11; Ubuntu; Linux x86_64; rv:89.0)")]
+    [InlineData("Chromee/99.0.4758.102 (X11; Linux x86_64)")]
+    [InlineData("FirreFox/100.0 (Windows NT 10.0; rv:100.0)")]
+    [InlineData("Saffari/605.1.15 (iPad; CPU OS 14_6 like Mac OS X)")]
+    [InlineData("Edgg/103.0.1264.37 (Macintosh; Intel Mac OS X 11_5_2)")]
+    [InlineData("Chorome/91.0.4472.124 (Linux; Android 10; SM-G973F)")]
+    [InlineData("Edgee/18.18363 (Windows 10 1909; Win64; x64)")]
+    public void InvalidUserAgent(string userAgent)
+    {
+        HttpUserAgentInformation info = HttpUserAgentInformation.Parse(userAgent);
+
+        // Invalid or malformed UAs must be classified as Unknown
+        Assert.Equal(HttpUserAgentType.Unknown, info.Type);
+        Assert.Null(info.Name);
+        Assert.Null(info.Version);
+
+        // Parser trims input via Cleanup, so compare to trimmed UA
+        Assert.Equal(userAgent.Trim(), info.UserAgent);
+
+        // Should not be considered a browser or a robot
+        Assert.False(info.IsBrowser());
+        Assert.False(info.IsRobot());
+    }
 }

Original file line number	Diff line number	Diff line change
`@@ -11,7 +11,6 @@ namespace MyCSharp.HttpUserAgentParser;`
`11`	`11`	`/// Parser logic for user agents`
`12`	`12`	`/// </summary>`
`13`	`13`	`public static class HttpUserAgentParser`
`14`		`-`
`15`	`14`	`{`
`16`	`15`	`/// <summary>`
`17`	`16`	`/// Parses given <param name="userAgent">user agent</param>`
`@@ -48,7 +47,6 @@ public static HttpUserAgentInformation Parse(string userAgent)`
`48`	`47`	`/// </summary>`
`49`	`48`	`public static HttpUserAgentPlatformInformation? GetPlatform(string userAgent)`
`50`	`49`	`{`
`51`		`- // Fast, allocation-free token scan (keeps public statics untouched)`
`52`	`50`	`ReadOnlySpan<char> ua = userAgent.AsSpan();`
`53`	`51`	`foreach ((string Token, string Name, HttpUserAgentPlatformType PlatformType) platform in HttpUserAgentStatics.s_platformRules)`
`54`	`52`	`{`
`@@ -78,6 +76,7 @@ public static bool TryGetPlatform(string userAgent, [NotNullWhen(true)] out Http`
`78`	`76`	`public static (string Name, string? Version)? GetBrowser(string userAgent)`
`79`	`77`	`{`
`80`	`78`	`ReadOnlySpan<char> ua = userAgent.AsSpan();`
	`79`	`+`
`81`	`80`	`foreach ((string Name, string DetectToken, string? VersionToken) browserRule in HttpUserAgentStatics.s_browserRules)`
`82`	`81`	`{`
`83`	`82`	`if (!TryIndexOf(ua, browserRule.DetectToken, out int detectIndex))`
`@@ -86,7 +85,18 @@ public static (string Name, string? Version)? GetBrowser(string userAgent)`
`86`	`85`	`}`
`87`	`86`
`88`	`87`	`// Version token may differ (e.g., Safari uses "Version/")`
`89`		`- int versionSearchStart = detectIndex;`
	`88`	`+`
	`89`	`+ int versionSearchStart;`
	`90`	`+ // For rules without a specific version token, ensure pattern Token/<digits>`
	`91`	`+ if (string.IsNullOrEmpty(browserRule.VersionToken))`
	`92`	`+ {`
	`93`	`+ int afterDetect = detectIndex + browserRule.DetectToken.Length;`
	`94`	`+ if (afterDetect >= ua.Length \|\| ua[afterDetect] != '/')`
	`95`	`+ {`
	`96`	`+ // Likely a misspelling or partial token (e.g., Edgg, Oprea, Chromee)`
	`97`	`+ continue;`
	`98`	`+ }`
	`99`	`+ }`
`90`	`100`	`if (!string.IsNullOrEmpty(browserRule.VersionToken))`
`91`	`101`	`{`
`92`	`102`	`if (TryIndexOf(ua, browserRule.VersionToken!, out int vtIndex))`
`@@ -104,14 +114,14 @@ public static (string Name, string? Version)? GetBrowser(string userAgent)`
`104`	`114`	`versionSearchStart = detectIndex + browserRule.DetectToken.Length;`
`105`	`115`	`}`
`106`	`116`
`107`		`- string? version = null;`
`108`		`- ua = ua.Slice(versionSearchStart);`
`109`		`- if (TryExtractVersion(ua, out Range range))`
	`117`	`+ ReadOnlySpan<char> search = ua.Slice(versionSearchStart);`
	`118`	`+ if (TryExtractVersion(search, out Range range))`
`110`	`119`	`{`
`111`		`- version = ua[range].ToString();`
	`120`	`+ string? version = search[range].ToString();`
	`121`	`+ return (browserRule.Name, version);`
`112`	`122`	`}`
`113`	`123`
`114`		`- return (browserRule.Name, version);`
	`124`	`+ // If we didn't find a version for this rule, try next rule`
`115`	`125`	`}`
`116`	`126`
`117`	`127`	`return null;`
`@@ -198,39 +208,43 @@ private static bool TryExtractVersion(ReadOnlySpan<char> haystack, out Range ran`
`198`	`208`
`199`	`209`	`// Limit search window to avoid scanning entire UA string unnecessarily`
`200`	`210`	`const int Window = 128;`
`201`		`- if (haystack.Length >= Window)`
	`211`	`+ if (haystack.Length > Window)`
`202`	`212`	`{`
`203`	`213`	`haystack = haystack.Slice(0, Window);`
`204`	`214`	`}`
`205`	`215`
`206`		`- int i = 0;`
`207`		`- for (; i < haystack.Length; ++i)`
	`216`	`+ // Find first digit`
	`217`	`+ int start = -1;`
	`218`	`+ for (int i = 0; i < haystack.Length; i++)`
`208`	`219`	`{`
`209`	`220`	`char c = haystack[i];`
`210`		`- if (char.IsBetween(c, '0', '9'))`
	`221`	`+ if (c >= '0' && c <= '9')`
`211`	`222`	`{`
	`223`	`+ start = i;`
`212`	`224`	`break;`
`213`	`225`	`}`
`214`	`226`	`}`
`215`	`227`
`216`		`- int s = i;`
`217`		`- haystack = haystack.Slice(i + 1);`
`218`		`- for (i = 0; i < haystack.Length; ++i)`
	`228`	`+ if (start < 0)`
`219`	`229`	`{`
`220`		`- char c = haystack[i];`
`221`		`- if (!(char.IsBetween(c, '0', '9') \|\| c == '.'))`
`222`		`- {`
`223`		`- break;`
`224`		`- }`
	`230`	`+ // No digit found => no version`
	`231`	`+ return false;`
`225`	`232`	`}`
`226`		`- i += s + 1; // shift back the previous domain`
`227`	`233`
`228`		`- if (i == s)`
	`234`	`+ // Consume digits and dots after first digit`
	`235`	`+ int end = start + 1;`
	`236`	`+ while (end < haystack.Length)`
`229`	`237`	`{`
`230`		`- return false;`
	`238`	`+ char c = haystack[end];`
	`239`	`+ if (!((c >= '0' && c <= '9') \|\| c == '.'))`
	`240`	`+ {`
	`241`	`+ break;`
	`242`	`+ }`
	`243`	`+ end++;`
`231`	`244`	`}`
`232`	`245`
`233`		`- range = new Range(s, i);`
	`246`	`+ // Create exclusive end range`
	`247`	`+ range = new Range(start, end);`
`234`	`248`	`return true;`
`235`	`249`	`}`
`236`	`250`	`}`