1 files changed, 232 insertions, 0 deletions
diff --git a/tests/FundLab.Api.Tests/ProcessCollectorTests.fs b/tests/FundLab.Api.Tests/ProcessCollectorTests.fs
new file mode 100644
index 0000000..32009bf
--- /dev/null
+++ b/tests/FundLab.Api.Tests/ProcessCollectorTests.fs
@@ -0,0 +1,232 @@
+namespace FundLab.Api.Tests
+
+module ProcessCollectorTests =
+
+    open System
+    open System.Diagnostics
+    open System.IO
+    open System.Threading
+    open Xunit
+    open FundLab.Api
+
+    let private secretSentinel = "fund-lab-SECRET-SENTINEL"
+
+    let private fixtureDir () =
+        let dir = Path.Combine(Path.GetTempPath(), "fund-lab-collector-" + Guid.NewGuid().ToString("N"))
+
+        Directory.CreateDirectory(dir) |> ignore
+
+        dir
+
+    let private writeScript (dir: string) (name: string) (body: string) =
+        let path = Path.Combine(dir, name)
+        File.WriteAllText(path, body)
+        path
+
+    let private searchEnvelope =
+        "{\"schema_version\":\"fund-lab.akshare.v1\",\"operation\":\"search\",\"source\":\"fixture\",\"source_revision\":\"fixture/1\",\"collected_at\":\"2026-09-21T00:00:00Z\",\"_synthetic\":true,\"instruments\":[{\"code\":\"000001\",\"name\":\"fixture fund\",\"fund_type\":null}]}"
+
+    let private hangScript dir =
+        writeScript dir "hang.sh" ("sleep 60 & child=$!\necho $child > pidfile\nwait $child\n")
+
+    let private orphanScript dir =
+        writeScript
+            dir
+            "orphan.sh"
+            ("sleep 30 & child=$!\necho $child > pidfile\nprintf '%s' '" + searchEnvelope + "'\nexit 0\n")
+
+    let private recordedPid (dir: string) =
+        Int32.Parse(File.ReadAllText(Path.Combine(dir, "pidfile")).Trim())
+
+    let private pidAlive (pid: int) = Directory.Exists(sprintf "/proc/%d" pid)
+
+    let private waitForPidDeath (pid: int) =
+        let mutable dead = not (pidAlive pid)
+        let mutable waited = 0
+
+        while not dead && waited < 10000 do
+            Thread.Sleep(100)
+            waited <- waited + 100
+            dead <- not (pidAlive pid)
+
+        dead
+
+    [<Fact>]
+    let ``collector captures stdout from a successful subprocess`` () =
+        let dir = fixtureDir ()
+
+        try
+            let script = writeScript dir "success.sh" ("printf '%s' '" + searchEnvelope + "'")
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 30) :> IMarketDataCollector
+
+            match collector.Search("000001", CancellationToken.None) with
+            | Ok raw ->
+                Assert.Contains("000001", raw)
+                Assert.Contains("\"_synthetic\":true", raw)
+            | Error message -> Assert.Fail(sprintf "expected success, got %s" message)
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``collector drains concurrent stderr flood without deadlocking`` () =
+        let dir = fixtureDir ()
+
+        try
+            let body = "yes 'flood padding padding padding padding' | head -n 5000 >&2\nprintf '%s' '" + searchEnvelope + "'"
+
+            let script = writeScript dir "flood.sh" body
+
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 30) :> IMarketDataCollector
+
+            match collector.Search("000001", CancellationToken.None) with
+            | Ok raw -> Assert.Contains("000001", raw)
+            | Error message -> Assert.Fail(sprintf "expected success, got %s" message)
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``collector enforces deadline, kills owned processes and sanitizes error`` () =
+        let dir = fixtureDir ()
+
+        try
+            let script = hangScript dir
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 1) :> IMarketDataCollector
+            let stopwatch = Stopwatch.StartNew()
+
+            match collector.Search("000001", CancellationToken.None) with
+            | Error message ->
+                stopwatch.Stop()
+                Assert.Contains("did not finish within 1 seconds", message)
+                Assert.DoesNotContain(dir, message)
+                Assert.DoesNotContain(secretSentinel, message)
+            | Ok raw -> Assert.Fail(sprintf "expected timeout error, got %s" raw)
+
+            Assert.True(stopwatch.Elapsed < TimeSpan.FromSeconds(20.0))
+            Assert.True(waitForPidDeath (recordedPid dir), "backgrounded subprocess survived the deadline kill")
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``collector reports cancellation and kills owned processes`` () =
+        let dir = fixtureDir ()
+
+        try
+            let script = hangScript dir
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 30) :> IMarketDataCollector
+            use cts = new CancellationTokenSource(TimeSpan.FromMilliseconds(500.0))
+
+            match collector.Search("000001", cts.Token) with
+            | Error message ->
+                Assert.Equal("collector run was cancelled", message)
+                Assert.DoesNotContain(dir, message)
+                Assert.DoesNotContain(secretSentinel, message)
+            | Ok raw -> Assert.Fail(sprintf "expected cancellation, got %s" raw)
+
+            Assert.True(waitForPidDeath (recordedPid dir), "backgrounded subprocess survived cancellation")
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``collector recovers payload and reaps pipe-holding orphan after parent exit`` () =
+        let dir = fixtureDir ()
+
+        try
+            let script = orphanScript dir
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 5) :> IMarketDataCollector
+            let stopwatch = Stopwatch.StartNew()
+
+            let result = collector.Search("000001", CancellationToken.None)
+            stopwatch.Stop()
+
+            match result with
+            | Ok raw ->
+                Assert.Equal(searchEnvelope, raw)
+                Assert.True(stopwatch.Elapsed < TimeSpan.FromSeconds(20.0))
+            | Error message ->
+                Assert.Fail(sprintf "expected recovered payload, got error %s" message)
+
+            Assert.True(waitForPidDeath (recordedPid dir), "pipe-holding orphan survived parent exit")
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``collector surfaces malformed payload for parser to reject`` () =
+        let dir = fixtureDir ()
+
+        try
+            let script = writeScript dir "malformed.sh" "printf '%s' '{ not json at all'"
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 30) :> IMarketDataCollector
+
+            match collector.Search("000001", CancellationToken.None) with
+            | Ok raw ->
+                Assert.Equal("{ not json at all", raw)
+
+                match MarketData.parseSearchPayload raw with
+                | Error _ -> ()
+                | Ok _ -> Assert.Fail("expected malformed payload to fail parsing")
+            | Error message -> Assert.Fail(sprintf "expected raw output, got error %s" message)
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``nonzero exit keeps stderr sentinel and paths out of the public error`` () =
+        let dir = fixtureDir ()
+
+        try
+            let body =
+                "printf 'secret=%s\\n' '"
+                + secretSentinel
+                + "' >&2\nprintf 'script lives under "
+                + dir
+                + "\\n' >&2\nexit 3\n"
+
+            let script = writeScript dir "failing.sh" body
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 30) :> IMarketDataCollector
+
+            match collector.Search("000001", CancellationToken.None) with
+            | Error message ->
+                Assert.Equal("collector exited with code 3", message)
+                Assert.DoesNotContain(secretSentinel, message)
+                Assert.DoesNotContain(dir, message)
+            | Ok raw -> Assert.Fail(sprintf "expected failure, got %s" raw)
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``start failure keeps executable path out of the public error`` () =
+        let dir = fixtureDir ()
+
+        try
+            let script = writeScript dir "success.sh" ("printf '%s' '" + searchEnvelope + "'")
+            let missingExecutable = "/nonexistent-" + secretSentinel + "/python3"
+
+            let collector = ProcessMarketDataCollector(missingExecutable, script, None, 30) :> IMarketDataCollector
+
+            match collector.Search("000001", CancellationToken.None) with
+            | Error message ->
+                let sanitized =
+                    message = "could not run the AKShare collector"
+                    || message.StartsWith("collector exited with code")
+
+                Assert.True(sanitized, sprintf "unexpected error message: %s" message)
+                Assert.DoesNotContain(secretSentinel, message)
+                Assert.DoesNotContain(dir, message)
+            | Ok raw -> Assert.Fail(sprintf "expected failure, got %s" raw)
+        finally
+            Directory.Delete(dir, true)
+
+    [<Fact>]
+    let ``missing collector script yields sanitized error without paths`` () =
+        let dir = fixtureDir ()
+
+        try
+            let script = Path.Combine(dir, "missing.py")
+            let collector = ProcessMarketDataCollector("/bin/sh", script, None, 30) :> IMarketDataCollector
+
+            match collector.Search("000001", CancellationToken.None) with
+            | Error message ->
+                Assert.Equal("collector script is not deployed", message)
+                Assert.DoesNotContain(dir, message)
+            | Ok raw -> Assert.Fail(sprintf "expected error, got %s" raw)
+        finally
+            Directory.Delete(dir, true)