Rust: restrict extracted files queries

Author: Paolo Tranquilli

rust/ql/lib/codeql/files/FileSystem.qll

@@ -6,6 +6,7 @@ private import codeql.rust.elements.SourceFile
 private import codeql.rust.elements.AstNode
 private import codeql.rust.elements.Comment
 private import codeql.rust.Diagnostics
+private import codeql.rust.internal.ExtractorStep
 
 private module Input implements InputSig {
   abstract class ContainerBase extends @container {
@@ -36,7 +37,9 @@ class Folder = Impl::Folder;
 /** A file. */
 class File extends Container, Impl::File {
   /** Holds if this file was extracted from ordinary source code. */
-  predicate fromSource() { any() }
+  predicate fromSource() {
+    exists(ExtractorStep s | s.getAction() = "Extract" and s.getFile() = this)
+  }
 
   /**
    * Gets the number of lines containing code in this file. This value
@@ -58,11 +61,20 @@ class File extends Container, Impl::File {
   }
 }
 
+/**
+ * A source file that was extracted.
+ *
+ * TODO: rename `SourceFile` from the generated AST to give that name to this class.
+ */
+class ExtractedFile extends File {
+  ExtractedFile() { this.fromSource() }
+}
+
 /**
  * A successfully extracted file, that is, a file that was extracted and
  * contains no extraction errors or warnings.
  */
-class SuccessfullyExtractedFile extends File {
+class SuccessfullyExtractedFile extends ExtractedFile {
   SuccessfullyExtractedFile() {
     not exists(Diagnostic d |
       d.getLocation().getFile() = this and

rust/ql/lib/codeql/rust/internal/ExtractorStep.qll

@@ -7,7 +7,7 @@ import codeql.files.FileSystem
  */
 class ExtractorStep extends @extractor_step {
   /**
-   * The string representation of this extractor step.
+   * Gets the string representation of this extractor step.
    */
   string toString() {
     exists(File file, string action |
@@ -17,17 +17,17 @@ class ExtractorStep extends @extractor_step {
   }
 
   /**
-   * The action this extractor step carried out.
+   * Gets the action this extractor step carried out.
    */
   string getAction() { extractor_steps(this, result, _, _) }
 
   /**
-   * The file the extractor step was carried out on.
+   * Gets the file the extractor step was carried out on.
    */
   File getFile() { extractor_steps(this, _, result, _) }
 
   /**
-   * The duration of the extractor step in milliseconds.
+   * Gets the duration of the extractor step in milliseconds.
    */
   int getDurationMs() { extractor_steps(this, _, _, result) }
 

rust/ql/src/queries/diagnostics/ExtractedFiles.ql

@@ -8,6 +8,6 @@
 
 import rust
 
-from File f
+from ExtractedFile f
 where exists(f.getRelativePath())
 select f, "File successfully extracted."

rust/ql/src/queries/diagnostics/ExtractionErrors.ql

@@ -11,7 +11,7 @@ import codeql.files.FileSystem
 /** Gets the SARIF severity to associate with an error. */
 int getSeverity() { result = 2 }
 
-from ExtractionError error, File f
+from ExtractionError error, ExtractedFile f
 where
   f = error.getLocation().getFile() and
   exists(f.getRelativePath())

rust/ql/src/queries/summary/SummaryStats.ql

@@ -21,10 +21,13 @@ where
   or
   key = "Extraction warnings" and value = count(ExtractionWarning w)
   or
-  key = "Files extracted - total" and value = count(File f | exists(f.getRelativePath()))
+  key = "Files extracted - total" and value = count(ExtractedFile f | exists(f.getRelativePath()))
   or
   key = "Files extracted - with errors" and
-  value = count(File f | exists(f.getRelativePath()) and not f instanceof SuccessfullyExtractedFile)
+  value =
+    count(ExtractedFile f |
+      exists(f.getRelativePath()) and not f instanceof SuccessfullyExtractedFile
+    )
   or
   key = "Files extracted - without errors" and
   value = count(SuccessfullyExtractedFile f | exists(f.getRelativePath()))

rust/ql/test/extractor-tests/File/File.expected

@@ -1,3 +1,4 @@
-| a_file.rs:0:0:0:0 | a_file.rs |
-| another_file.rs:0:0:0:0 | another_file.rs |
-| lib.rs:0:0:0:0 | lib.rs |
+| Cargo.toml:0:0:0:0 | Cargo.toml | fromSource: no |
+| a_file.rs:0:0:0:0 | a_file.rs | fromSource: yes |
+| another_file.rs:0:0:0:0 | another_file.rs | fromSource: yes |
+| lib.rs:0:0:0:0 | lib.rs | fromSource: yes |

rust/ql/test/extractor-tests/File/File.ql

@@ -1,5 +1,7 @@
 import rust
 
-from File f
-where exists(f.getRelativePath())
-select f
+from File f, string fromSource
+where
+  exists(f.getRelativePath()) and
+  if f.fromSource() then fromSource = "fromSource: yes" else fromSource = "fromSource: no"
+select f, fromSource

rust/ql/test/query-tests/diagnostics/ExtractedFiles.expected

@@ -5,4 +5,3 @@
 | main.rs:0:0:0:0 | main.rs | File successfully extracted. |
 | my_macro.rs:0:0:0:0 | my_macro.rs | File successfully extracted. |
 | my_struct.rs:0:0:0:0 | my_struct.rs | File successfully extracted. |
-| options.yml:0:0:0:0 | options.yml | File successfully extracted. |

rust/ql/test/query-tests/diagnostics/LinesOfUserCodeInFiles.expected

@@ -5,4 +5,5 @@
 | lib.rs:0:0:0:0 | lib.rs | 5 |
 | does_not_compile.rs:0:0:0:0 | does_not_compile.rs | 3 |
 | error.rs:0:0:0:0 | error.rs | 3 |
+| Cargo.toml:0:0:0:0 | Cargo.toml | 0 |
 | options.yml:0:0:0:0 | options.yml | 0 |

rust/ql/test/query-tests/diagnostics/SummaryStats.expected

@@ -2,9 +2,9 @@
 | Elements unextracted | 0 |
 | Extraction errors | 0 |
 | Extraction warnings | 7 |
-| Files extracted - total | 8 |
+| Files extracted - total | 7 |
 | Files extracted - with errors | 3 |
-| Files extracted - without errors | 5 |
+| Files extracted - without errors | 4 |
 | Inconsistencies - AST | 0 |
 | Inconsistencies - CFG | 0 |
 | Inconsistencies - data flow | 0 |