test(conformance): add golden artifact corpus

.github/workflows/conformance.yml

@@ -0,0 +1,71 @@
+name: Artifact conformance
+
+on:
+  pull_request:
+  release:
+    types: [published]
+  schedule:
+    - cron: "17 4 * * 2"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  fast:
+    name: Fast hermetic corpus
+    if: github.event_name == 'pull_request' || github.event_name == 'workflow_dispatch'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v7
+
+      - name: Install Rust toolchain
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          toolchain: "1.94"
+
+      - name: Run focused conformance harness
+        env:
+          RENDERFLOW_CONFORMANCE_TIER: "fast"
+          RENDERFLOW_CONFORMANCE_REPORT: "${{ runner.temp }}/renderflow-conformance-fast.json"
+        run: cargo test --package renderflow --test golden_conformance --locked
+
+      - name: Upload conformance report
+        if: always()
+        uses: actions/upload-artifact@v6
+        with:
+          name: "renderflow-conformance-fast-${{ github.run_id }}"
+          path: "${{ runner.temp }}/renderflow-conformance-fast.json"
+
+  maximal:
+    name: Maximal tool-backed corpus
+    if: github.event_name == 'schedule' || github.event_name == 'release' || github.event_name == 'workflow_dispatch'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v7
+
+      - name: Install Rust toolchain
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          toolchain: "1.94"
+
+      - name: Install optional conformance providers
+        run: |
+          sudo apt-get update
+          sudo apt-get install --yes ffmpeg pandoc
+
+      - name: Run maximal conformance harness
+        env:
+          RENDERFLOW_CONFORMANCE_TIER: "maximal"
+          RENDERFLOW_CONFORMANCE_REPORT: "${{ runner.temp }}/renderflow-conformance-maximal.json"
+        run: cargo test --package renderflow --test golden_conformance --locked
+
+      - name: Upload conformance report
+        if: always()
+        uses: actions/upload-artifact@v6
+        with:
+          name: "renderflow-conformance-maximal-${{ github.run_id }}"
+          path: "${{ runner.temp }}/renderflow-conformance-maximal.json"
+          retention-days: 90

crates/renderflow-core/src/detect.rs

@@ -83,6 +83,21 @@ pub struct DetectionConflict {
 /// The `buf` slice should contain at least the first 16 bytes of the file for
 /// reliable results, but longer slices are handled correctly.
 pub fn detect_from_bytes(buf: &[u8]) -> Option<Format> {
+    if buf.starts_with(b"RIFF") && buf.len() >= 12 {
+        match &buf[8..12] {
+            b"WAVE" => return Some(Format::Wav),
+            b"AVI " => return Some(Format::Avi),
+            b"WEBP" => return Some(Format::Webp),
+            _ => {}
+        }
+    }
+    if buf.len() >= 12 && buf.get(4..8) == Some(b"ftyp") {
+        return Some(match &buf[8..12] {
+            b"avif" | b"avis" => Format::Avif,
+            _ => Format::Mp4,
+        });
+    }
+
     // Iterate in a deterministic order so that ties between formats sharing
     // a prefix are resolved consistently.
     let registry = FormatCapabilityRegistry::global();
@@ -104,7 +119,10 @@ pub fn detect_from_bytes(buf: &[u8]) -> Option<Format> {
 
     // Prefer the match with the longest (most specific) signature to avoid
     // false positives when a shorter signature is a prefix of another.
-    candidates.sort_by(|a, b| b.1.cmp(&a.1));
+    candidates.sort_by(|a, b| {
+        b.1.cmp(&a.1)
+            .then_with(|| a.0.to_string().cmp(&b.0.to_string()))
+    });
     candidates.into_iter().next().map(|(f, _)| f)
 }
 
@@ -198,6 +216,25 @@ mod tests {
         assert_eq!(result, Some(Format::Flac));
     }
 
+    #[test]
+    fn detect_riff_formats_from_form_type() {
+        assert_eq!(detect_from_bytes(b"RIFF\0\0\0\0WAVE"), Some(Format::Wav));
+        assert_eq!(detect_from_bytes(b"RIFF\0\0\0\0AVI "), Some(Format::Avi));
+        assert_eq!(detect_from_bytes(b"RIFF\0\0\0\0WEBP"), Some(Format::Webp));
+    }
+
+    #[test]
+    fn detect_iso_base_media_from_major_brand() {
+        assert_eq!(
+            detect_from_bytes(b"\0\0\0\x18ftypisom\0\0\0\0"),
+            Some(Format::Mp4)
+        );
+        assert_eq!(
+            detect_from_bytes(b"\0\0\0\x18ftypavif\0\0\0\0"),
+            Some(Format::Avif)
+        );
+    }
+
     #[test]
     fn detect_mp3_from_id3_tag() {
         let buf = b"ID3\x03\x00\x00\x00\x00\x00\x00";

crates/renderflow-core/src/intake.rs

@@ -425,7 +425,9 @@ impl IntakeEngine {
                 .find(|descriptor| descriptor.id == "zip")
                 .copied()
         } else {
-            magic_matches.first().copied()
+            crate::detect::detect_from_bytes(bytes)
+                .and_then(|format| registry.get(format))
+                .or_else(|| magic_matches.first().copied())
         };
         if let Some(descriptor) = magic_match {
             signals.push(IntakeSignal {

crates/renderflow-core/src/validation.rs

@@ -821,10 +821,7 @@ fn conformance_row(
             .map(|tool| tool.stable_id().to_string())
             .collect(),
         validator_ids,
-        fixture_ids: match descriptor.id {
-            "png" => vec!["fixture.corrupt.png.truncated".to_string()],
-            _ => Vec::new(),
-        },
+        fixture_ids: golden_fixture_ids(descriptor.id),
         supported_platforms: vec![
             "linux".to_string(),
             "macos".to_string(),
@@ -836,6 +833,22 @@ fn conformance_row(
     }
 }
 
+fn golden_fixture_ids(format: &str) -> Vec<String> {
+    let ids: &[&str] = match format {
+        "markdown" => &["fixture.document.markdown"],
+        "pdf" => &["fixture.document.pdf"],
+        "png" => &["fixture.image.png", "fixture.mismatch.jpeg-png"],
+        "svg" => &["fixture.image.svg"],
+        "wav" => &["fixture.audio.wav"],
+        "mp4" => &["fixture.video.mp4"],
+        "zip" => &["fixture.archive.zip", "fixture.corrupt.zip"],
+        "json" => &["fixture.data.json"],
+        "srt" => &["fixture.subtitle.srt"],
+        _ => &[],
+    };
+    ids.iter().map(|id| (*id).to_string()).collect()
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -862,5 +875,14 @@ mod tests {
         let second = CapabilityConformanceMatrix::builtins();
         assert_eq!(first, second);
         assert!(!first.formats.is_empty());
+        assert_eq!(
+            first
+                .formats
+                .iter()
+                .find(|row| row.format == "wav")
+                .unwrap()
+                .fixture_ids,
+            vec!["fixture.audio.wav"]
+        );
     }
 }