alef_e2e/codegen/

r.rs

//! R e2e test generator using testthat.

use crate::config::E2eConfig;
use crate::escape::{escape_r, sanitize_filename, sanitize_ident};
use crate::field_access::FieldResolver;
use crate::fixture::{Assertion, CallbackAction, Fixture, FixtureGroup};
use alef_core::backend::GeneratedFile;
use alef_core::config::AlefConfig;
use alef_core::hash::{self, CommentStyle};
use anyhow::Result;
use std::fmt::Write as FmtWrite;
use std::path::PathBuf;

use super::E2eCodegen;

/// R e2e code generator.
pub struct RCodegen;

impl E2eCodegen for RCodegen {
    fn generate(
        &self,
        groups: &[FixtureGroup],
        e2e_config: &E2eConfig,
        _alef_config: &AlefConfig,
    ) -> Result<Vec<GeneratedFile>> {
        let lang = self.language_name();
        let output_base = PathBuf::from(e2e_config.effective_output()).join(lang);

        let mut files = Vec::new();

        // Resolve call config with overrides.
        let call = &e2e_config.call;
        let overrides = call.overrides.get(lang);
        let module_path = overrides
            .and_then(|o| o.module.as_ref())
            .cloned()
            .unwrap_or_else(|| call.module.clone());
        let _function_name = overrides
            .and_then(|o| o.function.as_ref())
            .cloned()
            .unwrap_or_else(|| call.function.clone());
        let result_is_simple = overrides.is_some_and(|o| o.result_is_simple);
        let _result_var = &call.result_var;

        // Resolve package config.
        let r_pkg = e2e_config.resolve_package("r");
        let pkg_name = r_pkg
            .as_ref()
            .and_then(|p| p.name.as_ref())
            .cloned()
            .unwrap_or_else(|| module_path.clone());
        let pkg_path = r_pkg
            .as_ref()
            .and_then(|p| p.path.as_ref())
            .cloned()
            .unwrap_or_else(|| "../../packages/r".to_string());
        let pkg_version = r_pkg
            .as_ref()
            .and_then(|p| p.version.as_ref())
            .cloned()
            .unwrap_or_else(|| "0.1.0".to_string());

        // Generate DESCRIPTION file.
        files.push(GeneratedFile {
            path: output_base.join("DESCRIPTION"),
            content: render_description(&pkg_name, &pkg_version, e2e_config.dep_mode),
            generated_header: false,
        });

        // Generate test runner script.
        files.push(GeneratedFile {
            path: output_base.join("run_tests.R"),
            content: render_test_runner(&pkg_path, e2e_config.dep_mode),
            generated_header: true,
        });

        // Generate test files per category.
        for group in groups {
            let active: Vec<&Fixture> = group
                .fixtures
                .iter()
                .filter(|f| f.skip.as_ref().is_none_or(|s| !s.should_skip(lang)))
                .collect();

            if active.is_empty() {
                continue;
            }

            let filename = format!("test_{}.R", sanitize_filename(&group.category));
            let field_resolver = FieldResolver::new(
                &e2e_config.fields,
                &e2e_config.fields_optional,
                &e2e_config.result_fields,
                &e2e_config.fields_array,
            );
            let content = render_test_file(&group.category, &active, &field_resolver, result_is_simple, e2e_config);
            files.push(GeneratedFile {
                path: output_base.join("tests").join(filename),
                content,
                generated_header: true,
            });
        }

        Ok(files)
    }

    fn language_name(&self) -> &'static str {
        "r"
    }
}

fn render_description(pkg_name: &str, pkg_version: &str, dep_mode: crate::config::DependencyMode) -> String {
    let dep_line = match dep_mode {
        crate::config::DependencyMode::Registry => {
            format!("Imports: {pkg_name} ({pkg_version})\n")
        }
        crate::config::DependencyMode::Local => String::new(),
    };
    format!(
        r#"Package: e2e.r
Title: E2E Tests for {pkg_name}
Version: 0.1.0
Description: End-to-end test suite.
{dep_line}Suggests: testthat (>= 3.0.0)
Config/testthat/edition: 3
"#
    )
}

fn render_test_runner(pkg_path: &str, dep_mode: crate::config::DependencyMode) -> String {
    let mut out = String::new();
    out.push_str(&hash::header(CommentStyle::Hash));
    let _ = writeln!(out, "library(testthat)");
    match dep_mode {
        crate::config::DependencyMode::Registry => {
            // In registry mode, require the installed CRAN package directly.
            let _ = writeln!(out, "# Package loaded via library() from CRAN install.");
        }
        crate::config::DependencyMode::Local => {
            // Use devtools::load_all() to load the local R package without requiring
            // a full install, matching the e2e test runner convention.
            let _ = writeln!(out, "devtools::load_all(\"{pkg_path}\")");
        }
    }
    let _ = writeln!(out);
    let _ = writeln!(out, "test_dir(\"tests\")");
    out
}

fn render_test_file(
    category: &str,
    fixtures: &[&Fixture],
    field_resolver: &FieldResolver,
    result_is_simple: bool,
    e2e_config: &E2eConfig,
) -> String {
    let mut out = String::new();
    out.push_str(&hash::header(CommentStyle::Hash));
    let _ = writeln!(out, "# E2e tests for category: {category}");
    let _ = writeln!(out);

    for (i, fixture) in fixtures.iter().enumerate() {
        render_test_case(&mut out, fixture, e2e_config, field_resolver, result_is_simple);
        if i + 1 < fixtures.len() {
            let _ = writeln!(out);
        }
    }

    // Clean up trailing newlines.
    while out.ends_with("\n\n") {
        out.pop();
    }
    if !out.ends_with('\n') {
        out.push('\n');
    }
    out
}

fn render_test_case(
    out: &mut String,
    fixture: &Fixture,
    e2e_config: &E2eConfig,
    field_resolver: &FieldResolver,
    result_is_simple: bool,
) {
    let call_config = e2e_config.resolve_call(fixture.call.as_deref());
    let function_name = &call_config.function;
    let result_var = &call_config.result_var;

    let test_name = sanitize_ident(&fixture.id);
    let description = fixture.description.replace('"', "\\\"");

    let expects_error = fixture.assertions.iter().any(|a| a.assertion_type == "error");

    let args_str = build_args_string(&fixture.input, &call_config.args);

    // Build visitor setup and args if present
    let mut setup_lines = Vec::new();
    let final_args = if let Some(visitor_spec) = &fixture.visitor {
        build_r_visitor(&mut setup_lines, visitor_spec);
        if args_str.is_empty() {
            "visitor".to_string()
        } else {
            format!("{args_str}, visitor = visitor")
        }
    } else {
        args_str
    };

    if expects_error {
        let _ = writeln!(out, "test_that(\"{test_name}: {description}\", {{");
        for line in &setup_lines {
            let _ = writeln!(out, "  {line}");
        }
        let _ = writeln!(out, "  expect_error({function_name}({final_args}))");
        let _ = writeln!(out, "}})");
        return;
    }

    let _ = writeln!(out, "test_that(\"{test_name}: {description}\", {{");
    for line in &setup_lines {
        let _ = writeln!(out, "  {line}");
    }
    let _ = writeln!(out, "  {result_var} <- {function_name}({final_args})");

    for assertion in &fixture.assertions {
        render_assertion(out, assertion, result_var, field_resolver, result_is_simple, e2e_config);
    }

    let _ = writeln!(out, "}})");
}

fn build_args_string(input: &serde_json::Value, args: &[crate::config::ArgMapping]) -> String {
    if args.is_empty() {
        return json_to_r(input, true);
    }

    let parts: Vec<String> = args
        .iter()
        .filter_map(|arg| {
            let field = arg.field.strip_prefix("input.").unwrap_or(&arg.field);
            let val = input.get(field)?;
            if val.is_null() && arg.optional {
                return None;
            }
            Some(format!("{} = {}", arg.name, json_to_r(val, true)))
        })
        .collect();

    parts.join(", ")
}

fn render_assertion(
    out: &mut String,
    assertion: &Assertion,
    result_var: &str,
    field_resolver: &FieldResolver,
    result_is_simple: bool,
    _e2e_config: &E2eConfig,
) {
    // Skip assertions on fields that don't exist on the result type.
    if let Some(f) = &assertion.field {
        if !f.is_empty() && !field_resolver.is_valid_for_result(f) {
            let _ = writeln!(out, "  # skipped: field '{f}' not available on result type");
            return;
        }
    }

    // When result_is_simple, skip assertions that reference non-content fields
    // (e.g., metadata, document, structure) since the binding returns a plain value.
    if result_is_simple {
        if let Some(f) = &assertion.field {
            let f_lower = f.to_lowercase();
            if !f.is_empty()
                && f_lower != "content"
                && (f_lower.starts_with("metadata")
                    || f_lower.starts_with("document")
                    || f_lower.starts_with("structure"))
            {
                let _ = writeln!(
                    out,
                    "  # skipped: result_is_simple for field '{f}' not available on result type"
                );
                return;
            }
        }
    }

    let field_expr = if result_is_simple {
        result_var.to_string()
    } else {
        match &assertion.field {
            Some(f) if !f.is_empty() => field_resolver.accessor(f, "r", result_var),
            _ => result_var.to_string(),
        }
    };

    match assertion.assertion_type.as_str() {
        "equals" => {
            if let Some(expected) = &assertion.value {
                let r_val = json_to_r(expected, false);
                let _ = writeln!(out, "  expect_equal(trimws({field_expr}), {r_val})");
            }
        }
        "contains" => {
            if let Some(expected) = &assertion.value {
                let r_val = json_to_r(expected, false);
                let _ = writeln!(out, "  expect_true(grepl({r_val}, {field_expr}, fixed = TRUE))");
            }
        }
        "contains_all" => {
            if let Some(values) = &assertion.values {
                for val in values {
                    let r_val = json_to_r(val, false);
                    let _ = writeln!(out, "  expect_true(grepl({r_val}, {field_expr}, fixed = TRUE))");
                }
            }
        }
        "not_contains" => {
            if let Some(expected) = &assertion.value {
                let r_val = json_to_r(expected, false);
                let _ = writeln!(out, "  expect_false(grepl({r_val}, {field_expr}, fixed = TRUE))");
            }
        }
        "not_empty" => {
            let _ = writeln!(
                out,
                "  expect_true(if (is.character({field_expr})) nchar({field_expr}) > 0 else length({field_expr}) > 0)"
            );
        }
        "is_empty" => {
            let _ = writeln!(out, "  expect_equal({field_expr}, \"\")");
        }
        "contains_any" => {
            if let Some(values) = &assertion.values {
                let items: Vec<String> = values.iter().map(|v| json_to_r(v, false)).collect();
                let vec_str = items.join(", ");
                let _ = writeln!(
                    out,
                    "  expect_true(any(sapply(c({vec_str}), function(v) grepl(v, {field_expr}, fixed = TRUE))))"
                );
            }
        }
        "greater_than" => {
            if let Some(val) = &assertion.value {
                let r_val = json_to_r(val, false);
                let _ = writeln!(out, "  expect_true({field_expr} > {r_val})");
            }
        }
        "less_than" => {
            if let Some(val) = &assertion.value {
                let r_val = json_to_r(val, false);
                let _ = writeln!(out, "  expect_true({field_expr} < {r_val})");
            }
        }
        "greater_than_or_equal" => {
            if let Some(val) = &assertion.value {
                let r_val = json_to_r(val, false);
                let _ = writeln!(out, "  expect_true({field_expr} >= {r_val})");
            }
        }
        "less_than_or_equal" => {
            if let Some(val) = &assertion.value {
                let r_val = json_to_r(val, false);
                let _ = writeln!(out, "  expect_true({field_expr} <= {r_val})");
            }
        }
        "starts_with" => {
            if let Some(expected) = &assertion.value {
                let r_val = json_to_r(expected, false);
                let _ = writeln!(out, "  expect_true(startsWith({field_expr}, {r_val}))");
            }
        }
        "ends_with" => {
            if let Some(expected) = &assertion.value {
                let r_val = json_to_r(expected, false);
                let _ = writeln!(out, "  expect_true(endsWith({field_expr}, {r_val}))");
            }
        }
        "min_length" => {
            if let Some(val) = &assertion.value {
                if let Some(n) = val.as_u64() {
                    let _ = writeln!(out, "  expect_true(nchar({field_expr}) >= {n})");
                }
            }
        }
        "max_length" => {
            if let Some(val) = &assertion.value {
                if let Some(n) = val.as_u64() {
                    let _ = writeln!(out, "  expect_true(nchar({field_expr}) <= {n})");
                }
            }
        }
        "count_min" => {
            if let Some(val) = &assertion.value {
                if let Some(n) = val.as_u64() {
                    let _ = writeln!(out, "  expect_true(length({field_expr}) >= {n})");
                }
            }
        }
        "count_equals" => {
            if let Some(val) = &assertion.value {
                if let Some(n) = val.as_u64() {
                    let _ = writeln!(out, "  expect_equal(length({field_expr}), {n})");
                }
            }
        }
        "is_true" => {
            let _ = writeln!(out, "  expect_true({field_expr})");
        }
        "is_false" => {
            let _ = writeln!(out, "  expect_false({field_expr})");
        }
        "method_result" => {
            if let Some(method_name) = &assertion.method {
                let call_expr = build_r_method_call(result_var, method_name, assertion.args.as_ref());
                let check = assertion.check.as_deref().unwrap_or("is_true");
                match check {
                    "equals" => {
                        if let Some(val) = &assertion.value {
                            if val.is_boolean() {
                                if val.as_bool() == Some(true) {
                                    let _ = writeln!(out, "  expect_true({call_expr})");
                                } else {
                                    let _ = writeln!(out, "  expect_false({call_expr})");
                                }
                            } else {
                                let r_val = json_to_r(val, false);
                                let _ = writeln!(out, "  expect_equal({call_expr}, {r_val})");
                            }
                        }
                    }
                    "is_true" => {
                        let _ = writeln!(out, "  expect_true({call_expr})");
                    }
                    "is_false" => {
                        let _ = writeln!(out, "  expect_false({call_expr})");
                    }
                    "greater_than_or_equal" => {
                        if let Some(val) = &assertion.value {
                            let r_val = json_to_r(val, false);
                            let _ = writeln!(out, "  expect_true({call_expr} >= {r_val})");
                        }
                    }
                    "count_min" => {
                        if let Some(val) = &assertion.value {
                            let n = val.as_u64().unwrap_or(0);
                            let _ = writeln!(out, "  expect_true(length({call_expr}) >= {n})");
                        }
                    }
                    "is_error" => {
                        let _ = writeln!(out, "  expect_error({call_expr})");
                    }
                    "contains" => {
                        if let Some(val) = &assertion.value {
                            let r_val = json_to_r(val, false);
                            let _ = writeln!(out, "  expect_true(grepl({r_val}, {call_expr}, fixed = TRUE))");
                        }
                    }
                    other_check => {
                        panic!("R e2e generator: unsupported method_result check type: {other_check}");
                    }
                }
            } else {
                panic!("R e2e generator: method_result assertion missing 'method' field");
            }
        }
        "not_error" => {
            // Already handled — the call would stop on error.
        }
        "error" => {
            // Handled at the test level.
        }
        other => {
            panic!("R e2e generator: unsupported assertion type: {other}");
        }
    }
}

/// Convert a `serde_json::Value` to an R literal string.
///
/// # Arguments
///
/// * `value` - The JSON value to convert
/// * `lowercase_enum_values` - If true, lowercase strings starting with uppercase letter (for enum values).
///   If false, preserve original case (for assertion expected values).
fn json_to_r(value: &serde_json::Value, lowercase_enum_values: bool) -> String {
    match value {
        serde_json::Value::String(s) => {
            // Lowercase enum values (strings starting with uppercase letter) only if requested
            let normalized = if lowercase_enum_values && s.chars().next().is_some_and(|c| c.is_uppercase()) {
                s.to_lowercase()
            } else {
                s.clone()
            };
            format!("\"{}\"", escape_r(&normalized))
        }
        serde_json::Value::Bool(true) => "TRUE".to_string(),
        serde_json::Value::Bool(false) => "FALSE".to_string(),
        serde_json::Value::Number(n) => n.to_string(),
        serde_json::Value::Null => "NULL".to_string(),
        serde_json::Value::Array(arr) => {
            let items: Vec<String> = arr.iter().map(|v| json_to_r(v, lowercase_enum_values)).collect();
            format!("c({})", items.join(", "))
        }
        serde_json::Value::Object(map) => {
            let entries: Vec<String> = map
                .iter()
                .map(|(k, v)| format!("\"{}\" = {}", escape_r(k), json_to_r(v, lowercase_enum_values)))
                .collect();
            format!("list({})", entries.join(", "))
        }
    }
}

/// Build an R visitor list and add setup line.
fn build_r_visitor(setup_lines: &mut Vec<String>, visitor_spec: &crate::fixture::VisitorSpec) {
    use std::fmt::Write as FmtWrite;
    let mut visitor_obj = String::new();
    let _ = writeln!(visitor_obj, "list(");
    for (method_name, action) in &visitor_spec.callbacks {
        emit_r_visitor_method(&mut visitor_obj, method_name, action);
    }
    let _ = writeln!(visitor_obj, "  )");

    setup_lines.push(format!("visitor <- {visitor_obj}"));
}

/// Build an R call expression for a `method_result` assertion.
/// Maps method names to the appropriate R function or method calls.
fn build_r_method_call(result_var: &str, method_name: &str, args: Option<&serde_json::Value>) -> String {
    match method_name {
        "root_child_count" => format!("{result_var}$root_child_count()"),
        "root_node_type" => format!("{result_var}$root_node_type()"),
        "named_children_count" => format!("{result_var}$named_children_count()"),
        "has_error_nodes" => format!("tree_has_error_nodes({result_var})"),
        "error_count" | "tree_error_count" => format!("tree_error_count({result_var})"),
        "tree_to_sexp" => format!("tree_to_sexp({result_var})"),
        "contains_node_type" => {
            let node_type = args
                .and_then(|a| a.get("node_type"))
                .and_then(|v| v.as_str())
                .unwrap_or("");
            format!("tree_contains_node_type({result_var}, \"{node_type}\")")
        }
        "find_nodes_by_type" => {
            let node_type = args
                .and_then(|a| a.get("node_type"))
                .and_then(|v| v.as_str())
                .unwrap_or("");
            format!("find_nodes_by_type({result_var}, \"{node_type}\")")
        }
        "run_query" => {
            let query_source = args
                .and_then(|a| a.get("query_source"))
                .and_then(|v| v.as_str())
                .unwrap_or("");
            let language = args
                .and_then(|a| a.get("language"))
                .and_then(|v| v.as_str())
                .unwrap_or("");
            format!("run_query({result_var}, \"{language}\", \"{query_source}\", source)")
        }
        _ => {
            if let Some(args_val) = args {
                let arg_str = args_val
                    .as_object()
                    .map(|obj| {
                        obj.iter()
                            .map(|(k, v)| {
                                let r_val = json_to_r(v, false);
                                format!("{k} = {r_val}")
                            })
                            .collect::<Vec<_>>()
                            .join(", ")
                    })
                    .unwrap_or_default();
                format!("{result_var}${method_name}({arg_str})")
            } else {
                format!("{result_var}${method_name}()")
            }
        }
    }
}

/// Emit an R visitor method for a callback action.
fn emit_r_visitor_method(out: &mut String, method_name: &str, action: &CallbackAction) {
    use std::fmt::Write as FmtWrite;

    // R uses visit_ prefix (matches binding signature)
    let params = match method_name {
        "visit_link" => "ctx, href, text, title",
        "visit_image" => "ctx, src, alt, title",
        "visit_heading" => "ctx, level, text, id",
        "visit_code_block" => "ctx, lang, code",
        "visit_code_inline"
        | "visit_strong"
        | "visit_emphasis"
        | "visit_strikethrough"
        | "visit_underline"
        | "visit_subscript"
        | "visit_superscript"
        | "visit_mark"
        | "visit_button"
        | "visit_summary"
        | "visit_figcaption"
        | "visit_definition_term"
        | "visit_definition_description" => "ctx, text",
        "visit_text" => "ctx, text",
        "visit_list_item" => "ctx, ordered, marker, text",
        "visit_blockquote" => "ctx, content, depth",
        "visit_table_row" => "ctx, cells, is_header",
        "visit_custom_element" => "ctx, tag_name, html",
        "visit_form" => "ctx, action_url, method",
        "visit_input" => "ctx, input_type, name, value",
        "visit_audio" | "visit_video" | "visit_iframe" => "ctx, src",
        "visit_details" => "ctx, is_open",
        "visit_element_end" | "visit_table_end" | "visit_definition_list_end" | "visit_figure_end" => "ctx, output",
        "visit_list_start" => "ctx, ordered",
        "visit_list_end" => "ctx, ordered, output",
        _ => "ctx",
    };

    let _ = writeln!(out, "    {method_name} = function({params}) {{");
    match action {
        CallbackAction::Skip => {
            let _ = writeln!(out, "      \"skip\"");
        }
        CallbackAction::Continue => {
            let _ = writeln!(out, "      \"continue\"");
        }
        CallbackAction::PreserveHtml => {
            let _ = writeln!(out, "      \"preserve_html\"");
        }
        CallbackAction::Custom { output } => {
            let escaped = escape_r(output);
            let _ = writeln!(out, "      list(custom = {escaped})");
        }
        CallbackAction::CustomTemplate { template } => {
            let escaped = escape_r(template);
            let _ = writeln!(out, "      list(custom = {escaped})");
        }
    }
    let _ = writeln!(out, "    }},");
}