fluidattacks_blends/

graphs.rs

//! Top-level orchestration: a path to its set of graphs.

use std::path::Path;

use blends_domain::graph_set::GraphSet;

use crate::ast::get_ast_graph;
use crate::content::Content;
use crate::syntax::get_syntax_graph;

#[must_use]
pub fn get_graphs_from_path(
    path: &Path,
    with_cfg: Option<bool>,
    with_metadata: Option<bool>,
) -> GraphSet {
    let Some(content) = Content::from_path(path, None) else {
        return GraphSet::default();
    };

    let Some(ast) = get_ast_graph(&content) else {
        return GraphSet::default();
    };

    let Some(syntax) = get_syntax_graph(&ast, &content, with_cfg, with_metadata) else {
        return GraphSet {
            ast: Some(ast),
            syntax: None,
        };
    };

    GraphSet {
        ast: Some(ast),
        syntax: Some(syntax),
    }
}

#[cfg(test)]
mod tests {
    use super::get_graphs_from_path;
    use crate::attrs::{
        ast_edge_attrs, ast_node_attrs, sorted_object, syntax_edge_attrs, syntax_node_attrs,
    };
    use blends_domain::ast::AstGraph;
    use blends_domain::syntax::{SyntaxGraph, SyntaxNode};
    use blends_domain::NodeId;
    use serde_json::{Map, Value};
    use std::collections::{BTreeMap, BTreeSet};
    use std::fs;
    use std::path::{Path, PathBuf};
    use test_case::test_case;

    fn fixtures_dir() -> PathBuf {
        Path::new(env!("CARGO_MANIFEST_DIR")).join("../../test/data/test_files/syntax_graph")
    }

    fn results_dir() -> PathBuf {
        Path::new(env!("CARGO_MANIFEST_DIR")).join("../../test/data/results")
    }

    fn output_dir() -> PathBuf {
        Path::new(env!("CARGO_MANIFEST_DIR")).join("../../test/data/output")
    }

    fn export_ast_graph_as_json(graph: &AstGraph) -> Value {
        let mut nodes = Map::new();
        for (id, node) in &graph.nodes {
            nodes.insert(id.0.to_string(), sorted_object(ast_node_attrs(node)));
        }

        let mut edges = Map::new();
        for (from, targets) in &graph.edges {
            let mut inner = Map::new();
            for (to, edge) in targets {
                inner.insert(to.0.to_string(), sorted_object(ast_edge_attrs(*edge)));
            }
            edges.insert(from.0.to_string(), Value::Object(inner));
        }

        let mut root = BTreeMap::new();
        root.insert("edges".to_owned(), Value::Object(edges));
        root.insert("nodes".to_owned(), Value::Object(nodes));
        sorted_object(root)
    }

    fn export_syntax_graph_as_json(graph: &SyntaxGraph) -> Value {
        let mut nodes = Map::new();
        for (id, node) in &graph.nodes {
            let attrs = syntax_node_attrs(node).unwrap_or_else(|| {
                panic!("syntax export not implemented for {}", node.label_type())
            });
            nodes.insert(id.0.to_string(), sorted_object(attrs));
        }

        let mut edges = Map::new();
        for (from, targets) in &graph.edges {
            let mut inner = Map::new();
            for (to, edge) in targets {
                inner.insert(to.0.to_string(), sorted_object(syntax_edge_attrs(*edge)));
            }
            edges.insert(from.0.to_string(), Value::Object(inner));
        }

        let mut root = BTreeMap::new();
        root.insert("edges".to_owned(), Value::Object(edges));
        root.insert("nodes".to_owned(), Value::Object(nodes));
        sorted_object(root)
    }

    #[test]
    fn empty_set_for_unsupported_file() {
        let dir = tempfile::tempdir().unwrap();
        let path = dir.path().join("a.unknown");
        fs::write(&path, b"whatever").unwrap();

        assert!(get_graphs_from_path(&path, None, None).ast.is_none());
    }

    #[test]
    fn empty_set_for_malformed_supported_file() {
        let dir = tempfile::tempdir().unwrap();
        let path = dir.path().join("a.java");
        fs::write(&path, b"class A {").unwrap();

        assert!(get_graphs_from_path(&path, None, None).ast.is_none());
    }

    fn rename_field_key(key: &str) -> String {
        key.strip_prefix("label_field_")
            .map_or_else(|| key.to_owned(), |field| format!("{field}_id"))
    }

    fn rename_node_attrs(attrs: &Value) -> Value {
        let Some(attrs) = attrs.as_object() else {
            return attrs.clone();
        };
        let mut renamed = Map::new();
        for (key, value) in attrs {
            renamed.insert(rename_field_key(key), value.clone());
        }
        Value::Object(renamed)
    }

    // The python golden keys node fields as `label_field_<field>`; the Rust engine
    // keys them as `<field>_id`. Rewrite the golden's keys in memory so the two
    // graphs compare on content. The `results/` file on disk is untouched.
    fn normalize_field_keys(graph: &Value) -> Value {
        let mut nodes = Map::new();
        if let Some(original) = graph.get("nodes").and_then(Value::as_object) {
            for (id, attrs) in original {
                nodes.insert(id.clone(), rename_node_attrs(attrs));
            }
        }

        let mut result = Map::new();
        if let Some(edges) = graph.get("edges") {
            result.insert("edges".to_owned(), edges.clone());
        }
        result.insert("nodes".to_owned(), Value::Object(nodes));
        Value::Object(result)
    }

    fn write_rust_output(suffix: &str, relative: &str, ast: &Value, syntax: Option<&Value>) {
        let mut entry = Map::new();
        entry.insert("graph".to_owned(), ast.clone());
        if let Some(syntax) = syntax {
            entry.insert("syntax_graph".to_owned(), syntax.clone());
        }
        let mut by_path = Map::new();
        by_path.insert(relative.to_owned(), Value::Object(entry));
        let mut root = Map::new();
        root.insert("graphs".to_owned(), Value::Object(by_path));

        let pretty = serde_json::to_string_pretty(&Value::Object(root)).expect("serialize output");
        let dir = output_dir();
        fs::create_dir_all(&dir).expect("create output dir");
        fs::write(dir.join(format!("root-graph_{suffix}.json")), pretty).expect("write output");
    }

    fn section(graph: &Value, key: &str) -> Map<String, Value> {
        graph
            .get(key)
            .and_then(Value::as_object)
            .cloned()
            .unwrap_or_default()
    }

    // The python golden dumps the ast after the syntax readers run, and some
    // readers overwrite `label_l` (c# class/method declaration take the line of
    // their identifier). Until those readers exist in rust, drop only `label_l`
    // on those node types from both sides, so just the mutated line is ignored
    // while every other attribute (line column, type, fields) is still compared.
    fn ignore_line_for(nodes: Map<String, Value>, skip_types: &[&str]) -> Map<String, Value> {
        nodes
            .into_iter()
            .map(|(id, mut attrs)| {
                let skip = attrs
                    .get("label_type")
                    .and_then(Value::as_str)
                    .is_some_and(|kind| skip_types.contains(&kind));
                if let Some(node) = attrs.as_object_mut().filter(|_| skip) {
                    node.remove("label_l");
                }
                (id, attrs)
            })
            .collect()
    }

    // Concise per-entry diff: only the ids whose content differs, rust vs python.
    fn diff_section(
        kind: &str,
        rust: &Map<String, Value>,
        python: &Map<String, Value>,
    ) -> Vec<String> {
        let mut diffs = Vec::new();
        for (id, rust_entry) in rust {
            match python.get(id) {
                None => diffs.push(format!(
                    "{kind} {id}: in rust output, missing in python golden"
                )),
                Some(python_entry) if python_entry != rust_entry => diffs.push(format!(
                    "{kind} {id} differs:\n  rust:   {rust_entry}\n  python: {python_entry}"
                )),
                Some(_) => {}
            }
        }
        for id in python.keys() {
            if !rust.contains_key(id) {
                diffs.push(format!(
                    "{kind} {id}: in python golden, missing in rust output"
                ));
            }
        }
        diffs
    }

    // `diff_section` is what makes every graph_generation parity assertion
    // meaningful: if it stopped reporting a category of mismatch, the suite
    // would keep passing while the rust and python graphs drifted apart.
    // Passing fixtures agree, so the mismatch arms are only reachable here.
    #[test_case(&[("n1", 1)], &[], "in rust output, missing in python golden" ; "only rust produced the entry")]
    #[test_case(&[("n1", 1)], &[("n1", 2)], "node n1 differs" ; "both produced it with different content")]
    #[test_case(&[], &[("n1", 1)], "in python golden, missing in rust output" ; "only the python golden has it")]
    fn diff_section_reports_each_kind_of_mismatch(
        rust_entries: &[(&str, i32)],
        python_entries: &[(&str, i32)],
        expected: &str,
    ) {
        let to_map = |entries: &[(&str, i32)]| -> Map<String, Value> {
            entries
                .iter()
                .map(|(id, value)| ((*id).to_owned(), Value::from(*value)))
                .collect()
        };

        let diffs = diff_section("node", &to_map(rust_entries), &to_map(python_entries));

        let [only] = diffs.as_slice() else {
            panic!("expected exactly one diff, got {diffs:?}");
        };
        assert!(
            only.contains(expected),
            "diff {only:?} does not report {expected:?}"
        );
    }

    #[test]
    fn diff_section_reports_nothing_when_both_graphs_agree() {
        let mut entries = Map::new();
        entries.insert("n1".to_owned(), Value::from(1));

        assert!(diff_section("node", &entries, &entries).is_empty());
    }

    const MAX_REPORTED_DIFFS: usize = 30;

    // Fixture suffixes whose language has no rust syntax dispatcher yet.
    const SYNTAX_NOT_YET_MIGRATED: &[&str] = &["elixir", "hcl", "kotlin", "rust", "scala", "swift"];

    // Fixture suffixes whose language dispatcher exists but still degrades
    // some node types to MissingNode: implemented nodes compare exactly,
    // diffs anchored at the missing placeholders are ignored.
    const SYNTAX_IN_PROGRESS: &[&str] = &[];

    fn ast_diffs(rust_ast: &Value, golden: &Value, suffix: &str) -> Vec<String> {
        let expected = golden
            .get("graph")
            .map(normalize_field_keys)
            .expect("locate graph block in python golden");

        // c# alone has syntax readers that overwrite `label_l` on class/method
        // declarations; ignore that attribute only for c#, never other languages.
        let line_skip: &[&str] = match suffix {
            "c_sharp" => &["class_declaration", "method_declaration"],
            _ => &[],
        };
        let mut diffs = diff_section(
            "node",
            &ignore_line_for(section(rust_ast, "nodes"), line_skip),
            &ignore_line_for(section(&expected, "nodes"), line_skip),
        );
        diffs.extend(diff_section(
            "edge",
            &section(rust_ast, "edges"),
            &section(&expected, "edges"),
        ));
        diffs
    }

    fn syntax_diffs(generated_syntax: &Value, golden: &Value) -> Vec<String> {
        let expected = golden
            .get("syntax_graph")
            .cloned()
            .expect("locate syntax_graph block in python golden");

        let mut diffs = diff_section(
            "syntax node",
            &section(generated_syntax, "nodes"),
            &section(&expected, "nodes"),
        );
        diffs.extend(diff_section(
            "syntax edge",
            &section(generated_syntax, "edges"),
            &section(&expected, "edges"),
        ));
        diffs
    }

    // The ids the rust engine degraded to MissingNode: subtrees whose reader
    // is pending, plus punctuation children the python readers consumed
    // without creating a syntax node.
    fn missing_ids(generated_syntax: &Value) -> BTreeSet<String> {
        section(generated_syntax, "nodes")
            .into_iter()
            .filter(|(_, attrs)| {
                attrs.get("label_type").and_then(Value::as_str) == Some("MissingNode")
            })
            .map(|(id, _)| id)
            .collect()
    }

    fn edge_target_ids(edges: &Map<String, Value>, from: &str) -> Vec<String> {
        edges
            .get(from)
            .and_then(Value::as_object)
            .map(|targets| targets.keys().cloned().collect())
            .unwrap_or_default()
    }

    fn pending_subtree_ids(generated_syntax: &Value) -> BTreeSet<String> {
        let edges = section(generated_syntax, "edges");
        let mut skip = missing_ids(generated_syntax);
        let mut stack: Vec<String> = skip.iter().cloned().collect();
        while let Some(from) = stack.pop() {
            let fresh: Vec<String> = edge_target_ids(&edges, &from)
                .into_iter()
                .filter(|to| skip.insert(to.clone()))
                .collect();
            stack.extend(fresh);
        }
        skip
    }

    fn drop_missing_nodes(
        nodes: Map<String, Value>,
        skip: &BTreeSet<String>,
    ) -> Map<String, Value> {
        nodes
            .into_iter()
            .filter(|(id, _)| !skip.contains(id))
            .collect()
    }

    fn drop_missing_targets(targets: &Value, skip: &BTreeSet<String>) -> Map<String, Value> {
        targets
            .as_object()
            .cloned()
            .unwrap_or_default()
            .into_iter()
            .filter(|(to, _)| !skip.contains(to))
            .collect()
    }

    fn drop_missing_edges(
        edges: Map<String, Value>,
        skip: &BTreeSet<String>,
    ) -> Map<String, Value> {
        edges
            .into_iter()
            .filter(|(from, _)| !skip.contains(from))
            .map(|(from, targets)| (from, drop_missing_targets(&targets, skip)))
            .filter(|(_, kept)| !kept.is_empty())
            .map(|(from, kept)| (from, Value::Object(kept)))
            .collect()
    }

    // In-progress compare: everything not anchored at a MissingNode id must
    // match the golden exactly, on both sides — a golden-only entry between
    // implemented nodes is still a real diff.
    fn syntax_diffs_partial(generated_syntax: &Value, golden: &Value) -> Vec<String> {
        let expected = golden
            .get("syntax_graph")
            .cloned()
            .expect("locate syntax_graph block in python golden");
        let skip = pending_subtree_ids(generated_syntax);

        let mut diffs = diff_section(
            "syntax node",
            &drop_missing_nodes(section(generated_syntax, "nodes"), &skip),
            &drop_missing_nodes(section(&expected, "nodes"), &skip),
        );
        diffs.extend(diff_section(
            "syntax edge",
            &drop_missing_edges(section(generated_syntax, "edges"), &skip),
            &drop_missing_edges(section(&expected, "edges"), &skip),
        ));
        diffs
    }

    #[test_case("c_sharp.cs", "c_sharp")]
    #[test_case("elixir.ex", "elixir")]
    #[test_case("go.go", "go")]
    #[test_case("terraform.tf", "hcl")]
    #[test_case("java.java", "java")]
    #[test_case("javascript.js", "javascript")]
    #[test_case("json.json", "json")]
    #[test_case("kotlin.kt", "kotlin")]
    #[test_case("python.py", "python")]
    #[test_case("php.php", "php")]
    #[test_case("ruby.rb", "ruby")]
    #[test_case("rust.rs", "rust")]
    #[test_case("scala.scala", "scala")]
    #[test_case("swift.swift", "swift")]
    #[test_case("syntax_cfg.ts", "typescript")]
    #[test_case("yaml.yaml", "yaml")]
    #[test_case("templates/helm_configmap.yaml", "helm_configmap_yaml")]
    #[test_case("templates/helm_configmap.json", "helm_configmap_json")]
    #[test_case("flow_mapping.yaml", "flow_mapping")]
    #[test_case("flow_sequence.yaml", "flow_sequence")]
    fn graph_generation(test_file: &str, suffix: &str) {
        let path = fixtures_dir().join(test_file);
        let graph_set = get_graphs_from_path(&path, None, None);

        assert!(
            !(SYNTAX_NOT_YET_MIGRATED.contains(&suffix) && SYNTAX_IN_PROGRESS.contains(&suffix)),
            "suffix {suffix} cannot be pending and in progress at the same time"
        );
        assert_eq!(
            graph_set.syntax.is_none(),
            SYNTAX_NOT_YET_MIGRATED.contains(&suffix),
            "\n[Syntax Parity Error] Inconsistency detected for language extension: .{suffix}\n\
             - Was syntax graph generated (None)? -> {}\n\
             - Is it marked as pending in SYNTAX_NOT_YET_MIGRATED? -> {}\n\
             👉 Hint: If it was generated but is marked as pending, move '.{suffix}' to \
             SYNTAX_IN_PROGRESS (partial compare) or drop it from both lists (strict compare).\n\
             👉 Hint: If it is not pending but returned None, a regression occurred in the language dispatcher.",
            graph_set.syntax.is_none(),
            SYNTAX_NOT_YET_MIGRATED.contains(&suffix)
        );

        let generated_ast = graph_set
            .ast
            .as_ref()
            .map(export_ast_graph_as_json)
            .expect("AST graph should be built for the fixture");

        let generated_syntax = graph_set.syntax.as_ref().map(export_syntax_graph_as_json);

        let relative = format!("test/data/test_files/{test_file}");
        write_rust_output(suffix, &relative, &generated_ast, generated_syntax.as_ref());

        let python_results: Value = serde_json::from_str(
            &fs::read_to_string(results_dir().join(format!("root-graph_{suffix}.json")))
                .expect("read python golden"),
        )
        .expect("parse python golden");
        let golden = python_results
            .get("graphs")
            .and_then(|graphs| graphs.get(&relative))
            .expect("locate the fixture entry in python golden");

        let mut diffs = ast_diffs(&generated_ast, golden, suffix);
        if let Some(generated_syntax) = &generated_syntax {
            if SYNTAX_IN_PROGRESS.contains(&suffix) {
                diffs.extend(syntax_diffs_partial(generated_syntax, golden));
            } else {
                diffs.extend(syntax_diffs(generated_syntax, golden));
            }
        }

        assert_graph_parity(suffix, &diffs);
    }

    #[test_case("java.java", "java")]
    fn graph_generation_with_metadata(test_file: &str, suffix: &str) {
        let path = fixtures_dir().join(test_file);
        let mut graph_set = get_graphs_from_path(&path, None, Some(true));

        let relative_fixture = format!("test/data/test_files/syntax_graph/{test_file}");
        if let Some(syntax) = graph_set.syntax.as_mut() {
            if let Some(SyntaxNode::Metadata {
                path: metadata_path,
                ..
            }) = syntax.nodes.get_mut(&NodeId(0))
            {
                *metadata_path = relative_fixture;
            }
        }

        let generated_ast = graph_set
            .ast
            .as_ref()
            .map(export_ast_graph_as_json)
            .expect("AST graph should be built for the fixture");
        let generated_syntax = graph_set
            .syntax
            .as_ref()
            .map(export_syntax_graph_as_json)
            .expect("syntax graph should be built with metadata");

        let relative = format!("test/data/test_files/{test_file}");
        write_rust_output(
            &format!("metadata_{suffix}"),
            &relative,
            &generated_ast,
            Some(&generated_syntax),
        );

        let python_results: Value = serde_json::from_str(
            &fs::read_to_string(results_dir().join(format!("root-graph-metadata_{suffix}.json")))
                .expect("read python golden"),
        )
        .expect("parse python golden");
        let golden = python_results
            .get("graphs")
            .and_then(|graphs| graphs.get(&relative))
            .expect("locate the fixture entry in python golden");

        let mut diffs = ast_diffs(&generated_ast, golden, suffix);
        diffs.extend(syntax_diffs(&generated_syntax, golden));
        assert_graph_parity(suffix, &diffs);
    }

    fn assert_graph_parity(suffix: &str, diffs: &[String]) {
        let shown = diffs
            .iter()
            .take(MAX_REPORTED_DIFFS)
            .cloned()
            .collect::<Vec<_>>()
            .join("\n");
        let extra = diffs.len().saturating_sub(MAX_REPORTED_DIFFS);
        let more = if extra > 0 {
            format!("\n… and {extra} more differing entries")
        } else {
            String::new()
        };

        assert!(
            diffs.is_empty(),
            "graph parity mismatch for {suffix} ({} differing entries):\n{shown}{more}",
            diffs.len()
        );
    }
}