@inneranimalmedia/agentsam-sdk 2.6.4 → 2.6.7

package/native/agentsam-machine/core/src/frontend.rs

@@ -0,0 +1,442 @@
+//! Deterministic frontend-source candidate extraction (TS/JS/TSX/JSX).
+//! Emits observed literals and generic *_candidate findings only.
+//! Does NOT invent CMS theme identities (hero.cinematic, etc.) — refinery owns that.
+
+use crate::MachineReceipt;
+use serde_json::{json, Value};
+use std::collections::BTreeSet;
+use std::fs;
+use std::path::Path;
+
+const ZONES: &[&str] = &["HEADER", "BODY", "FOOTER", "TEMPLATE"];
+
+pub fn enrich_frontend_source_candidates(root: &Path, receipt: &mut MachineReceipt) {
+    let ts_paths: Vec<String> = receipt
+        .facts
+        .iter()
+        .filter(|f| matches!(f.source.type_id.as_str(), "typescript" | "javascript"))
+        .map(|f| f.path.clone())
+        .collect();
+    if ts_paths.is_empty() {
+        return;
+    }
+
+    let mut route_candidates: BTreeSet<String> = BTreeSet::new();
+    let mut shell_candidates: BTreeSet<String> = BTreeSet::new();
+    let mut section_keys: BTreeSet<String> = BTreeSet::new();
+    let mut section_candidates: Vec<Value> = Vec::new();
+    let mut component_candidates: BTreeSet<String> = BTreeSet::new();
+    let mut interaction_candidates: BTreeSet<String> = BTreeSet::new();
+    let mut design_tokens: BTreeSet<String> = BTreeSet::new();
+    let mut repeater_candidates: BTreeSet<String> = BTreeSet::new();
+    let mut evidence_fact_ids: BTreeSet<String> = BTreeSet::new();
+
+    for rel in &ts_paths {
+        let abs = root.join(rel);
+        let Ok(text) = fs::read_to_string(&abs) else {
+            continue;
+        };
+        evidence_fact_ids.insert(format!("file:{rel}"));
+
+        if let Some(route) = app_router_route_from_path(rel) {
+            route_candidates.insert(route);
+        }
+
+        for slug in extract_slug_literals(&text) {
+            route_candidates.insert(slug);
+        }
+
+        for zone in ZONES {
+            if text.contains(&format!("\"{zone}\"")) || text.contains(&format!("'{zone}'")) {
+                shell_candidates.insert((*zone).to_string());
+            }
+        }
+
+        for cand in extract_section_object_candidates(&text) {
+            let key = cand.to_string();
+            if section_keys.insert(key) {
+                section_candidates.push(cand);
+            }
+        }
+
+        for name in extract_quoted_string_array_after(&text, "components") {
+            component_candidates.insert(name);
+        }
+        for name in extract_quoted_string_array_after(&text, "templateCards") {
+            component_candidates.insert(name);
+        }
+
+        for msg in extract_cms_message_types(&text) {
+            interaction_candidates.insert(msg);
+        }
+
+        for token in extract_css_custom_properties(&text) {
+            design_tokens.insert(token);
+        }
+        for color in extract_hex_colors(&text) {
+            design_tokens.insert(color);
+        }
+
+        if text.contains(".map(") && (text.contains("sections") || text.contains("pages")) {
+            repeater_candidates.insert(format!("file:{rel}"));
+        }
+    }
+
+    if route_candidates.is_empty()
+        && shell_candidates.is_empty()
+        && section_candidates.is_empty()
+        && component_candidates.is_empty()
+        && interaction_candidates.is_empty()
+        && design_tokens.is_empty()
+    {
+        return;
+    }
+
+    let route_list: Vec<String> = route_candidates.into_iter().collect();
+    let shell_list: Vec<String> = shell_candidates.into_iter().collect();
+    let component_list: Vec<String> = component_candidates.into_iter().collect();
+    let interaction_list: Vec<String> = interaction_candidates.into_iter().collect();
+    let token_list: Vec<String> = design_tokens.into_iter().take(128).collect();
+    let repeater_list: Vec<String> = repeater_candidates.into_iter().collect();
+    let fact_ids: Vec<String> = evidence_fact_ids.into_iter().collect();
+
+    receipt.findings.push(json!({
+        "id": "finding:frontend_source_candidates",
+        "kind": "frontend_source_candidates",
+        "schema": "agentsam.machine.finding.v1",
+        "basis": "observed",
+        "evidence": {
+            "fact_ids": fact_ids,
+            "note": "Generic candidates from TS/JS literals only; CMS identity mapping belongs to the refinery.",
+        },
+        "candidates": {
+            "route_candidate": route_list.clone(),
+            "shell_candidate": shell_list.clone(),
+            "section_candidate": section_candidates.clone(),
+            "component_candidate": component_list.clone(),
+            "repeater_candidate": repeater_list.clone(),
+            "interaction_candidate": interaction_list.clone(),
+            "design_token_evidence": token_list.clone(),
+        }
+    }));
+
+    // Merge into / refresh frontend_manifest artifact when present.
+    let mut merged = false;
+    for art in &mut receipt.artifacts {
+        if art.get("kind").and_then(|v| v.as_str()) == Some("frontend_manifest") {
+            if let Some(data) = art.get_mut("data").and_then(|d| d.as_object_mut()) {
+                data.insert("route_candidates".into(), json!(route_list));
+                data.insert("shell_candidates".into(), json!(shell_list));
+                data.insert("section_candidates".into(), json!(section_candidates));
+                data.insert("component_candidates".into(), json!(component_list));
+                data.insert("interaction_candidates".into(), json!(interaction_list));
+                data.insert("design_token_evidence".into(), json!(token_list));
+                data.insert("repeater_candidates".into(), json!(repeater_list));
+                merged = true;
+            }
+        }
+    }
+    if !merged {
+        receipt.artifacts.push(json!({
+            "kind": "frontend_manifest",
+            "schema": "agentsam.machine.frontend.v1",
+            "data": {
+                "route_candidates": route_list,
+                "shell_candidates": shell_list,
+                "section_candidates": section_candidates,
+                "component_candidates": component_list,
+                "interaction_candidates": interaction_list,
+                "design_token_evidence": token_list,
+                "repeater_candidates": repeater_list,
+                "note": "Candidates from frontend source literals — not normalized CMS primitives.",
+            }
+        }));
+    }
+}
+
+fn app_router_route_from_path(rel: &str) -> Option<String> {
+    let norm = rel.replace('\\', "/");
+    let marker = "/app/";
+    let idx = norm.find(marker).or_else(|| {
+        if norm.starts_with("app/") {
+            Some(0)
+        } else {
+            None
+        }
+    })?;
+    let after = if norm[idx..].starts_with("app/") {
+        &norm[idx + 4..]
+    } else {
+        &norm[idx + marker.len()..]
+    };
+    let file = after.rsplit('/').next().unwrap_or(after);
+    if !(file == "page.tsx"
+        || file == "page.jsx"
+        || file == "page.ts"
+        || file == "page.js"
+        || file == "route.ts"
+        || file == "route.js")
+    {
+        return None;
+    }
+    let dir = after.rsplit_once('/').map(|(d, _)| d).unwrap_or("");
+    if dir.is_empty() {
+        return Some("/".to_string());
+    }
+    let mut parts = Vec::new();
+    for seg in dir.split('/') {
+        if seg.is_empty() || seg.starts_with('(') && seg.ends_with(')') {
+            continue; // route groups
+        }
+        if seg.starts_with('@') {
+            continue; // parallel routes
+        }
+        if seg.starts_with('[') && seg.ends_with(']') {
+            parts.push(format!(
+                ":{}",
+                seg.trim_matches(|c| c == '[' || c == ']' || c == '.')
+            ));
+            continue;
+        }
+        parts.push(seg.to_string());
+    }
+    if parts.is_empty() {
+        Some("/".to_string())
+    } else {
+        Some(format!("/{}", parts.join("/")))
+    }
+}
+
+fn extract_slug_literals(text: &str) -> Vec<String> {
+    let mut out = Vec::new();
+    for (prefix, quote) in [
+        ("slug: \"", '"'),
+        ("slug: '", '\''),
+        ("slug:\"", '"'),
+        ("slug:'", '\''),
+    ] {
+        let mut search = text;
+        while let Some(idx) = search.find(prefix) {
+            let rest = &search[idx + prefix.len()..];
+            if let Some(end) = rest.find(quote) {
+                let slug = &rest[..end];
+                if slug.starts_with('/') && slug.len() < 128 && !slug.contains(' ') {
+                    out.push(slug.to_string());
+                }
+                search = &rest[end + 1..];
+            } else {
+                break;
+            }
+        }
+    }
+    out
+}
+
+fn floor_char_boundary(text: &str, index: usize) -> usize {
+    let mut boundary = index.min(text.len());
+    while boundary > 0 && !text.is_char_boundary(boundary) {
+        boundary -= 1;
+    }
+    boundary
+}
+
+fn extract_section_object_candidates(text: &str) -> Vec<Value> {
+    // Look for compact object literals with type + zone nearby: type: "Hero", zone: "BODY"
+    let mut out = Vec::new();
+    let mut search = text;
+    while let Some(type_idx) = find_type_literal(search) {
+        let window_start = floor_char_boundary(search, type_idx.saturating_sub(80));
+        let window_end = floor_char_boundary(search, (type_idx + 200).min(search.len()));
+        let window = &search[window_start..window_end];
+        let Some((section_type, _)) = quoted_after(window, "type:") else {
+            search = &search[type_idx + 5..];
+            continue;
+        };
+        let zone = quoted_after(window, "zone:")
+            .map(|(z, _)| z)
+            .filter(|z| ZONES.contains(&z.as_str()));
+        let name = quoted_after(window, "name:").map(|(n, _)| n);
+        // Require an explicit layout zone so Toast/Page status types are not candidates.
+        if let Some(z) = zone {
+            let mut obj = serde_json::Map::new();
+            obj.insert("type".into(), json!(section_type));
+            obj.insert("zone".into(), json!(z));
+            if let Some(n) = name {
+                obj.insert("name".into(), json!(n));
+            }
+            out.push(Value::Object(obj));
+        }
+        search = &search[type_idx + 5..];
+        if out.len() >= 64 {
+            break;
+        }
+    }
+    out
+}
+
+fn find_type_literal(text: &str) -> Option<usize> {
+    text.find("type: \"")
+        .or_else(|| text.find("type: '"))
+        .or_else(|| text.find("type:\"").or_else(|| text.find("type:'")))
+}
+
+fn quoted_after(text: &str, key: &str) -> Option<(String, usize)> {
+    let idx = text.find(key)?;
+    let rest = text[idx + key.len()..].trim_start();
+    let quote = rest.chars().next()?;
+    if quote != '"' && quote != '\'' {
+        return None;
+    }
+    let inner = &rest[1..];
+    let end = inner.find(quote)?;
+    Some((inner[..end].to_string(), idx))
+}
+
+fn extract_quoted_string_array_after(text: &str, binding: &str) -> Vec<String> {
+    let patterns = [
+        format!("const {binding} = ["),
+        format!("const {binding}=["),
+        format!("let {binding} = ["),
+        format!("{binding}: ["),
+    ];
+    let mut out = Vec::new();
+    for pat in &patterns {
+        if let Some(idx) = text.find(pat) {
+            let rest = &text[idx + pat.len()..];
+            let end = rest.find(']').unwrap_or(rest.len().min(2000));
+            let body = &rest[..end];
+            out.extend(extract_quoted_strings(body));
+            break;
+        }
+    }
+    out
+}
+
+fn extract_quoted_strings(text: &str) -> Vec<String> {
+    let mut out = Vec::new();
+    let bytes = text.as_bytes();
+    let mut i = 0;
+    while i < bytes.len() {
+        let b = bytes[i];
+        if b == b'"' || b == b'\'' {
+            let q = b;
+            i += 1;
+            let start = i;
+            while i < bytes.len() && bytes[i] != q {
+                if bytes[i] == b'\\' && i + 1 < bytes.len() {
+                    i += 2;
+                    continue;
+                }
+                i += 1;
+            }
+            if i <= bytes.len() {
+                let s = &text[start..i.min(text.len())];
+                if !s.is_empty() && s.len() < 120 {
+                    out.push(s.to_string());
+                }
+            }
+        }
+        i += 1;
+    }
+    out
+}
+
+fn extract_cms_message_types(text: &str) -> Vec<String> {
+    let mut out = Vec::new();
+    let mut search = text;
+    while let Some(idx) = search.find("cms:") {
+        let rest = &search[idx..];
+        let end = rest
+            .find(|c: char| !(c.is_ascii_alphanumeric() || c == ':' || c == '-' || c == '_'))
+            .unwrap_or(rest.len().min(48));
+        let token = &rest[..end];
+        if token.starts_with("cms:") && token.len() > 4 {
+            out.push(token.to_string());
+        }
+        search = &search[idx + 4..];
+    }
+    out
+}
+
+fn extract_css_custom_properties(text: &str) -> Vec<String> {
+    let mut out = Vec::new();
+    let mut search = text;
+    while let Some(idx) = search.find("--") {
+        let rest = &search[idx..];
+        let name_end = rest[2..]
+            .find(|c: char| !(c.is_ascii_alphanumeric() || c == '-' || c == '_'))
+            .map(|i| i + 2)
+            .unwrap_or(rest.len().min(64));
+        if name_end > 3 {
+            let name = &rest[..name_end];
+            if name
+                .chars()
+                .all(|c| c.is_ascii_alphanumeric() || c == '-' || c == '_')
+            {
+                out.push(name.to_string());
+            }
+        }
+        search = &search[idx + 2..];
+        if out.len() >= 64 {
+            break;
+        }
+    }
+    out
+}
+
+fn extract_hex_colors(text: &str) -> Vec<String> {
+    let mut out = Vec::new();
+    let bytes = text.as_bytes();
+    let mut i = 0;
+    while i < bytes.len() {
+        if bytes[i] == b'#' {
+            let start = i;
+            i += 1;
+            let mut len = 0;
+            while i < bytes.len() && bytes[i].is_ascii_hexdigit() && len < 8 {
+                i += 1;
+                len += 1;
+            }
+            if len == 3 || len == 6 || len == 8 {
+                out.push(text[start..start + 1 + len].to_ascii_lowercase());
+            }
+            continue;
+        }
+        i += 1;
+        if out.len() >= 64 {
+            break;
+        }
+    }
+    out
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn app_router_route_from_studio_page() {
+        assert_eq!(
+            app_router_route_from_path("app/studio/page.tsx").as_deref(),
+            Some("/studio")
+        );
+        assert_eq!(
+            app_router_route_from_path("app/page.tsx").as_deref(),
+            Some("/")
+        );
+    }
+
+    #[test]
+    fn section_candidate_windows_are_utf8_safe() {
+        for tail in ["a", "\u{00a0}", "é", "🙂"] {
+            let source = format!(
+                "const section = {{ type: \"Hero\", zone: \"BODY\", name: \"Café\" }};{}",
+                tail.repeat(220)
+            );
+            let candidates = extract_section_object_candidates(&source);
+            assert_eq!(candidates.len(), 1, "failed for tail {tail:?}");
+            assert_eq!(candidates[0]["type"], "Hero");
+            assert_eq!(candidates[0]["zone"], "BODY");
+        }
+    }
+}