Reject unreopenable source before checked save stages files

crates/oa-core/src/persistence.rs

@@ -3,6 +3,9 @@ use serde::{Deserialize, Serialize};
 use std::{io::Write, path::Path};
 use thiserror::Error;
 mod json_codec;
+#[cfg(test)]
+#[path = "persistence/source_save_tests.rs"]
+mod source_save_tests;
 #[path = "migration_v2/mod.rs"]
 pub(crate) mod v2;
 pub(crate) use v2::adapter::migrate as migrate_v2_limited;
@@ -948,19 +951,54 @@ fn save_checked_profile(
     expected: Option<&[u8]>,
     profile: json_codec::Profile,
     depth_128: bool,
+) -> Result<Vec<u8>, FileError> {
+    save_checked_profile_with_source(path, p, expected, profile, depth_128, |project| {
+        crate::language::format(project).into_bytes()
+    })
+}
+// Keep the source writer replaceable only at this private seam so regressions
+// can prove that malformed or lossy output never reaches filesystem staging.
+fn save_checked_profile_with_source(
+    path: &Path,
+    p: &Project,
+    expected: Option<&[u8]>,
+    profile: json_codec::Profile,
+    depth_128: bool,
+    source: impl FnOnce(&Project) -> Vec<u8>,
 ) -> Result<Vec<u8>, FileError> {
     validate_project(p)?;
-    let mut bytes = if path
+    let is_source = path
         .extension()
-        .is_some_and(|ext| ext.eq_ignore_ascii_case("arcadia"))
-    {
-        crate::language::format(p).into_bytes()
+        .is_some_and(|ext| ext.eq_ignore_ascii_case("arcadia"));
+    let mut bytes = if is_source {
+        source(p)
     } else {
         json_codec::encode(p, profile, depth_128)?
     };
     if !bytes.ends_with(b"\n") {
         bytes.push(b'\n');
     }
+    if is_source {
+        // Check exactly the final bytes, including their newline, through the
+        // public reader before creating a temp file or replacing a destination.
+        let decoded = decode_with_report_limited(
+            &bytes,
+            crate::language::ParseLimits {
+                max_source_bytes: profile.bytes,
+                ..crate::language::ParseLimits::UNLIMITED
+            },
+        )?;
+        if decoded.project != *p {
+            return Err(FileError::Invalid(
+                "Generated source does not reproduce the complete requested project.".into(),
+            ));
+        }
+        if decoded.migration.is_some() {
+            return Err(FileError::Invalid(
+                "Generated source unexpectedly requires migration.".into(),
+            ));
+        }
+    }
     let parent = path
         .parent()
         .filter(|p| !p.as_os_str().is_empty())

crates/oa-core/src/persistence/source_save_tests.rs

@@ -0,0 +1,145 @@
+//! Containment while the complete source8 adapter remains unavailable.
+use super::*;
+use crate::analysis::{Definition, DefinitionBody, QuantityAlgebra, QuantityKindDefinition};
+use crate::analysis_numeric::Dimension;
+use std::fs;
+
+const ORDINARY_SOURCE7: &str = concat!(
+    "arcadia 7;\n",
+    "project saved \"Saved model\" id \"4210b9f2-7046-47f0-aa3b-153f626a579d\" {\n",
+    "    description \"Retain café and exact identity\";\n",
+    "}",
+);
+
+fn ordinary() -> Project {
+    let decoded = decode_with_report(ORDINARY_SOURCE7.as_bytes()).unwrap();
+    assert!(decoded.migration.is_some());
+    decoded.project
+}
+
+fn directory_entries(path: &Path) -> Vec<std::ffi::OsString> {
+    let mut names = fs::read_dir(path)
+        .unwrap()
+        .map(|entry| entry.unwrap().file_name())
+        .collect::<Vec<_>>();
+    names.sort();
+    names
+}
+
+// A missing parent makes a mistakenly early NamedTempFile creation observable
+// as Io, even when the temp would otherwise be removed again on error.
+fn rejected_output(
+    project: &Project,
+    output: &[u8],
+    profile: json_codec::Profile,
+    check: impl Fn(&FileError),
+) {
+    let root = tempfile::tempdir().unwrap();
+    let existing = root.path().join("existing.arcadia");
+    let sentinel = b"external bytes must survive\r\n";
+    fs::write(&existing, sentinel).unwrap();
+    for path in [
+        existing.clone(),
+        root.path().join("new.arcadia"),
+        root.path().join("missing-parent/new.arcadia"),
+    ] {
+        let before = directory_entries(root.path());
+        let expected = (path == existing).then_some(sentinel.as_slice());
+        let error =
+            save_checked_profile_with_source(&path, project, expected, profile, false, |_| {
+                output.to_vec()
+            })
+            .unwrap_err();
+        check(&error);
+        assert_eq!(directory_entries(root.path()), before);
+        assert_eq!(fs::read(&existing).unwrap(), sentinel);
+        if path != existing {
+            assert!(!path.exists());
+        }
+    }
+}
+
+#[test]
+fn malformed_generated_source_rejects_before_any_filesystem_staging() {
+    rejected_output(
+        &ordinary(),
+        b"arcadia",
+        json_codec::Profile::DOCUMENT,
+        |error| assert!(matches!(error, FileError::Text(_)), "{error}"),
+    );
+}
+
+#[test]
+fn parseable_source_omitting_a_definition_cannot_replace_the_whole_project() {
+    let mut project = ordinary();
+    let id = Id::parse_str("188b911c-5f8b-431c-9eea-2d41f3953767").unwrap();
+    project.definitions.insert(
+        id,
+        Definition {
+            id,
+            alias: "duration".into(),
+            name: "Duration".into(),
+            description: "Unused definitions are still authored model data.".into(),
+            body: DefinitionBody::QuantityKind(QuantityKindDefinition {
+                dimension: Dimension::new([0, 0, 1, 0, 0, 0, 0]).unwrap(),
+                algebra: QuantityAlgebra::Linear,
+            }),
+        },
+    );
+    validate_project(&project).unwrap();
+    let decoded = decode(ORDINARY_SOURCE7.as_bytes()).unwrap();
+    assert_eq!(decoded.id, project.id);
+    assert_eq!(decoded.name, project.name);
+    assert_eq!(decoded.elements, project.elements);
+    assert!(decoded.definitions.is_empty());
+    assert_ne!(decoded, project);
+    rejected_output(
+        &project,
+        ORDINARY_SOURCE7.as_bytes(),
+        json_codec::Profile::DOCUMENT,
+        |error| {
+            assert!(
+                matches!(error, FileError::Invalid(message)
+                if message == "Generated source does not reproduce the complete requested project."),
+                "{error}"
+            );
+        },
+    );
+}
+
+#[test]
+fn equal_project_from_generated_legacy_source_still_rejects_migration() {
+    rejected_output(
+        &ordinary(),
+        ORDINARY_SOURCE7.as_bytes(),
+        json_codec::Profile::DOCUMENT,
+        |error| {
+            assert!(
+                matches!(error, FileError::Invalid(message)
+                if message == "Generated source unexpectedly requires migration."),
+                "{error}"
+            );
+        },
+    );
+}
+
+#[test]
+fn generated_source_admission_checks_final_newline_under_the_effective_byte_limit() {
+    let project = ordinary();
+    let profile = json_codec::Profile::with_byte_limit(ORDINARY_SOURCE7.len());
+    rejected_output(&project, ORDINARY_SOURCE7.as_bytes(), profile, |error| {
+        assert!(matches!(error, FileError::Limit(_)), "{error}");
+    });
+    let mut terminated = ORDINARY_SOURCE7.as_bytes().to_vec();
+    terminated.push(b'\n');
+    let profile = json_codec::Profile::with_byte_limit(terminated.len());
+    // Existing final newline is not added twice: decoding now reaches the
+    // independent migration refusal, rather than exceeding this exact limit.
+    rejected_output(&project, &terminated, profile, |error| {
+        assert!(
+            matches!(error, FileError::Invalid(message)
+            if message == "Generated source unexpectedly requires migration."),
+            "{error}"
+        );
+    });
+}