Add ability of scip encoder to merge multiple scips

Summary: Modify scip encoder so it can take multiple scip files as input. It will combine the results into the output glean json, skipping any file that appears more than once.

Reviewed By: josh-gordon-fb

Differential Revision: D78521570

fbshipit-source-id: 644694b1eb00730aab008cde60944b0e66a188b9

Author: Robert Grosse

glean/lang/scip/indexer/scip_to_glean/src/angle.rs

@@ -110,15 +110,20 @@ impl Env {
         path_prefix: Option<&str>,
         doc: Document,
     ) -> Result<()> {
-        let src_file_id = self.next_id();
         // todo - adjust filepath with prefix and suffix
         let mut filepath = doc.relative_path.to_owned();
         if let Some(path_prefix) = path_prefix {
             filepath = format!("{}{}", path_prefix, filepath);
         }
         let filepath = filepath.into_boxed_str();
 
-        self.set_def_fact(StringPredicate::File, filepath.clone(), src_file_id);
+        // Skip files if the same file has already been seen.
+        // Note that this differs from the Haskell version, which does not have this check.
+        let (src_file_id, already_seen) =
+            self.get_or_set_fact(StringPredicate::File, filepath.clone());
+        if already_seen {
+            return Ok(());
+        }
 
         self.out.src_file(src_file_id, filepath.clone());
         let lang_file_id = self.next_id();

glean/lang/scip/indexer/scip_to_glean/src/main.rs

@@ -38,7 +38,7 @@ mod output;
 )]
 struct BuildJsonArgs {
     #[arg(short, long)]
-    input: PathBuf,
+    input: Vec<PathBuf>,
     #[arg(short, long)]
     output: PathBuf,
 
@@ -99,13 +99,15 @@ fn build_json(args: BuildJsonArgs) -> Result<()> {
         .and_then(|s| LanguageId::new(s).known());
 
     let mut env = Env::new();
-    decode_scip_data(
-        &mut env,
-        &args.input,
-        default_language,
-        args.infer_language,
-        args.root_prefix.as_deref(),
-    )?;
+    for input in &args.input {
+        decode_scip_data(
+            &mut env,
+            input,
+            default_language,
+            args.infer_language,
+            args.root_prefix.as_deref(),
+        )?;
+    }
 
     let output_facts = env.output();
     info!("Found {} facts total", output_facts.total_facts_count());
@@ -213,7 +215,7 @@ mod tests {
         let output_json = NamedTempFile::new().expect("unable to create temp file");
 
         let args = BuildJsonArgs {
-            input: scip_file.path().to_path_buf(),
+            input: vec![scip_file.path().to_path_buf()],
             output: output_json.path().to_path_buf(),
             infer_language: true,
             language: None,
@@ -235,7 +237,7 @@ mod tests {
         let output_json_dir = tempfile::TempDir::new().expect("Unable to create temp dir");
 
         let args = BuildJsonArgs {
-            input: scip_file.path().to_path_buf(),
+            input: vec![scip_file.path().to_path_buf()],
             output: output_json_dir.path().to_path_buf(),
             infer_language: true,
             language: None,