weave_content/

lib.rs

#![deny(unsafe_code)]
#![deny(clippy::unwrap_used)]
#![deny(clippy::expect_used)]
#![allow(clippy::missing_errors_doc)]

pub mod build_cache;
pub mod cache;
pub mod domain;
pub mod entity;
pub mod html;
pub mod nulid_gen;
pub mod output;
pub mod parser;
pub mod registry;
pub mod relationship;
pub mod tags;
pub mod timeline;
pub mod verifier;
pub mod writeback;

use std::collections::HashSet;

use crate::entity::Entity;
use crate::parser::{ParseError, ParsedCase, SectionKind};
use crate::relationship::Rel;

/// Build a case index: scan case files for front matter `id:` and H1 title.
///
/// Returns a map of `case_path → (nulid, title)` used to resolve cross-case
/// references in `## Related Cases` sections.
pub fn build_case_index(
    case_files: &[String],
    content_root: &std::path::Path,
) -> Result<std::collections::HashMap<String, (String, String)>, i32> {
    let mut map = std::collections::HashMap::new();
    for path in case_files {
        let content = std::fs::read_to_string(path).map_err(|e| {
            eprintln!("{path}: {e}");
            1
        })?;
        if let Some(id) = extract_front_matter_id(&content)
            && let Some(case_path) = case_slug_from_path(std::path::Path::new(path), content_root)
        {
            let title = extract_title(&content).unwrap_or_else(|| case_path.clone());
            map.insert(case_path, (id, title));
        }
    }
    Ok(map)
}

/// Extract the `id:` value from YAML front matter without full parsing.
fn extract_front_matter_id(content: &str) -> Option<String> {
    let content = content.strip_prefix("---\n")?;
    let end = content.find("\n---")?;
    let fm = &content[..end];
    for line in fm.lines() {
        let trimmed = line.trim();
        if let Some(id) = trimmed.strip_prefix("id:") {
            let id = id.trim().trim_matches('"').trim_matches('\'');
            if !id.is_empty() {
                return Some(id.to_string());
            }
        }
    }
    None
}

/// Extract the first H1 heading (`# Title`) after the front matter closing delimiter.
fn extract_title(content: &str) -> Option<String> {
    let content = content.strip_prefix("---\n")?;
    let end = content.find("\n---")?;
    let after_fm = &content[end + 4..];
    for line in after_fm.lines() {
        if let Some(title) = line.strip_prefix("# ") {
            let title = title.trim();
            if !title.is_empty() {
                return Some(title.to_string());
            }
        }
    }
    None
}

/// Derive a case slug from a file path relative to content root.
///
/// E.g. `content/cases/id/corruption/2002/foo.md` with content root
/// `content/` → `id/corruption/2002/foo`.
pub fn case_slug_from_path(
    path: &std::path::Path,
    content_root: &std::path::Path,
) -> Option<String> {
    let cases_dir = content_root.join("cases");
    let rel = path.strip_prefix(&cases_dir).ok()?;
    let s = rel.to_str()?;
    Some(s.strip_suffix(".md").unwrap_or(s).to_string())
}

/// Parse a case file fully: front matter, entities, relationships, timeline.
/// Returns the parsed case, inline entities, and relationships (including NEXT from timeline).
///
/// When a registry is provided, relationship and timeline names are resolved
/// against both inline events AND the global entity registry.
pub fn parse_full(
    content: &str,
    reg: Option<&registry::EntityRegistry>,
) -> Result<(ParsedCase, Vec<Entity>, Vec<Rel>), Vec<ParseError>> {
    let case = parser::parse(content)?;
    let mut errors = Vec::new();

    let mut all_entities = Vec::new();
    for section in &case.sections {
        if matches!(
            section.kind,
            SectionKind::Events | SectionKind::Documents | SectionKind::Assets
        ) {
            let entities =
                entity::parse_entities(&section.body, section.kind, section.line, &mut errors);
            all_entities.extend(entities);
        }
    }

    // Build combined name set: inline events + registry entities
    let mut entity_names: HashSet<&str> = all_entities.iter().map(|e| e.name.as_str()).collect();
    if let Some(registry) = reg {
        for name in registry.names() {
            entity_names.insert(name);
        }
    }

    let event_names: HashSet<&str> = all_entities
        .iter()
        .filter(|e| e.label == entity::Label::Event)
        .map(|e| e.name.as_str())
        .collect();

    let mut all_rels = Vec::new();
    for section in &case.sections {
        if section.kind == SectionKind::Relationships {
            let rels = relationship::parse_relationships(
                &section.body,
                section.line,
                &entity_names,
                &case.sources,
                &mut errors,
            );
            all_rels.extend(rels);
        }
    }

    for section in &case.sections {
        if section.kind == SectionKind::Timeline {
            let rels =
                timeline::parse_timeline(&section.body, section.line, &event_names, &mut errors);
            all_rels.extend(rels);
        }
    }

    if errors.is_empty() {
        Ok((case, all_entities, all_rels))
    } else {
        Err(errors)
    }
}

/// Collect registry entities referenced by relationships in this case.
/// Sets the `slug` field on each entity from the registry's file path.
pub fn collect_referenced_registry_entities(
    rels: &[Rel],
    inline_entities: &[Entity],
    reg: &registry::EntityRegistry,
) -> Vec<Entity> {
    let inline_names: HashSet<&str> = inline_entities.iter().map(|e| e.name.as_str()).collect();
    let mut referenced = Vec::new();
    let mut seen_names: HashSet<String> = HashSet::new();

    for rel in rels {
        for name in [&rel.source_name, &rel.target_name] {
            if !inline_names.contains(name.as_str())
                && seen_names.insert(name.clone())
                && let Some(entry) = reg.get_by_name(name)
            {
                let mut entity = entry.entity.clone();
                entity.slug = reg.slug_for(entry);
                referenced.push(entity);
            }
        }
    }

    referenced
}

/// Build a `CaseOutput` from a case file path.
/// Handles parsing and ID writeback.
pub fn build_case_output(
    path: &str,
    reg: &registry::EntityRegistry,
) -> Result<output::CaseOutput, i32> {
    let mut written = HashSet::new();
    build_case_output_tracked(path, reg, &mut written, &std::collections::HashMap::new())
}

/// Build a `CaseOutput` from a case file path, tracking which entity files
/// have already been written back. This avoids re-reading entity files from
/// disk when multiple cases reference the same shared entity.
#[allow(clippy::implicit_hasher)]
pub fn build_case_output_tracked(
    path: &str,
    reg: &registry::EntityRegistry,
    written_entities: &mut HashSet<std::path::PathBuf>,
    case_nulid_map: &std::collections::HashMap<String, (String, String)>,
) -> Result<output::CaseOutput, i32> {
    let content = match std::fs::read_to_string(path) {
        Ok(c) => c,
        Err(e) => {
            eprintln!("{path}: error reading file: {e}");
            return Err(2);
        }
    };

    let (case, entities, rels) = match parse_full(&content, Some(reg)) {
        Ok(result) => result,
        Err(errors) => {
            for err in &errors {
                eprintln!("{path}:{err}");
            }
            return Err(1);
        }
    };

    let referenced_entities = collect_referenced_registry_entities(&rels, &entities, reg);

    // Resolve case NULID
    let (case_nulid, case_nulid_generated) = match nulid_gen::resolve_id(case.id.as_deref(), 1) {
        Ok(result) => result,
        Err(err) => {
            eprintln!("{path}:{err}");
            return Err(1);
        }
    };
    let case_nulid_str = case_nulid.to_string();

    // Compute case slug from file path
    let case_slug = reg
        .content_root()
        .and_then(|root| registry::path_to_slug(std::path::Path::new(path), root));

    // Derive case_id from slug (filename-based) or fall back to empty string
    let case_id = case_slug
        .as_deref()
        .and_then(|s| s.rsplit('/').next())
        .unwrap_or_default();

    let build_result = match output::build_output(
        case_id,
        &case_nulid_str,
        &case.title,
        &case.summary,
        &case.tags,
        case_slug.as_deref(),
        case.case_type.as_deref(),
        case.status.as_deref(),
        case.amounts.as_deref(),
        &case.sources,
        &case.related_cases,
        case_nulid_map,
        &entities,
        &rels,
        &referenced_entities,
    ) {
        Ok(out) => out,
        Err(errors) => {
            for err in &errors {
                eprintln!("{path}:{err}");
            }
            return Err(1);
        }
    };

    let case_output = build_result.output;

    // Write back generated IDs to source case file
    let mut case_pending = build_result.case_pending;
    if case_nulid_generated {
        case_pending.push(writeback::PendingId {
            line: writeback::find_front_matter_end(&content).unwrap_or(2),
            id: case_nulid_str.clone(),
            kind: writeback::WriteBackKind::CaseId,
        });
    }
    if !case_pending.is_empty()
        && let Some(modified) = writeback::apply_writebacks(&content, &mut case_pending)
    {
        if let Err(e) = writeback::write_file(std::path::Path::new(path), &modified) {
            eprintln!("{e}");
            return Err(2);
        }
        let count = case_pending.len();
        eprintln!("{path}: wrote {count} generated ID(s) back to file");
    }

    // Write back generated IDs to entity files
    if let Some(code) =
        writeback_registry_entities(&build_result.registry_pending, reg, written_entities)
    {
        return Err(code);
    }

    eprintln!(
        "{path}: built ({} nodes, {} relationships)",
        case_output.nodes.len(),
        case_output.relationships.len()
    );
    Ok(case_output)
}

/// Write back generated IDs to registry entity files.
/// Tracks already-written paths in `written` to avoid redundant disk reads.
/// Returns `Some(exit_code)` on error, `None` on success.
fn writeback_registry_entities(
    pending: &[(String, writeback::PendingId)],
    reg: &registry::EntityRegistry,
    written: &mut HashSet<std::path::PathBuf>,
) -> Option<i32> {
    for (entity_name, pending_id) in pending {
        let Some(entry) = reg.get_by_name(entity_name) else {
            continue;
        };
        let entity_path = &entry.path;

        // Skip if this entity file was already written by a previous case.
        if !written.insert(entity_path.clone()) {
            continue;
        }

        // Also skip if the entity already has an ID in the registry
        // (loaded from file at startup).
        if entry.entity.id.is_some() {
            continue;
        }

        let entity_content = match std::fs::read_to_string(entity_path) {
            Ok(c) => c,
            Err(e) => {
                eprintln!("{}: error reading file: {e}", entity_path.display());
                return Some(2);
            }
        };

        let fm_end = writeback::find_front_matter_end(&entity_content);
        let mut ids = vec![writeback::PendingId {
            line: fm_end.unwrap_or(2),
            id: pending_id.id.clone(),
            kind: writeback::WriteBackKind::EntityFrontMatter,
        }];
        if let Some(modified) = writeback::apply_writebacks(&entity_content, &mut ids) {
            if let Err(e) = writeback::write_file(entity_path, &modified) {
                eprintln!("{e}");
                return Some(2);
            }
            eprintln!("{}: wrote generated ID back to file", entity_path.display());
        }
    }
    None
}

/// Check whether a file's YAML front matter already contains an `id:` field.
#[cfg(test)]
fn front_matter_has_id(content: &str) -> bool {
    let mut in_front_matter = false;
    for line in content.lines() {
        let trimmed = line.trim();
        if trimmed == "---" && !in_front_matter {
            in_front_matter = true;
        } else if trimmed == "---" && in_front_matter {
            return false; // end of front matter, no id found
        } else if in_front_matter && trimmed.starts_with("id:") {
            return true;
        }
    }
    false
}

/// Resolve the content root directory.
///
/// Priority: explicit `--root` flag > parent of given path > current directory.
pub fn resolve_content_root(path: Option<&str>, root: Option<&str>) -> std::path::PathBuf {
    if let Some(r) = root {
        return std::path::PathBuf::from(r);
    }
    if let Some(p) = path {
        let p = std::path::Path::new(p);
        if p.is_file() {
            if let Some(parent) = p.parent() {
                for ancestor in parent.ancestors() {
                    if ancestor.join("cases").is_dir()
                        || ancestor.join("people").is_dir()
                        || ancestor.join("organizations").is_dir()
                    {
                        return ancestor.to_path_buf();
                    }
                }
                return parent.to_path_buf();
            }
        } else if p.is_dir() {
            return p.to_path_buf();
        }
    }
    std::path::PathBuf::from(".")
}

/// Load entity registry from content root. Returns empty registry if no entity dirs exist.
pub fn load_registry(content_root: &std::path::Path) -> Result<registry::EntityRegistry, i32> {
    match registry::EntityRegistry::load(content_root) {
        Ok(reg) => Ok(reg),
        Err(errors) => {
            for err in &errors {
                eprintln!("registry: {err}");
            }
            Err(1)
        }
    }
}

/// Load tag registry from content root. Returns empty registry if no tags.yaml exists.
pub fn load_tag_registry(content_root: &std::path::Path) -> Result<tags::TagRegistry, i32> {
    match tags::TagRegistry::load(content_root) {
        Ok(reg) => Ok(reg),
        Err(errors) => {
            for err in &errors {
                eprintln!("tags: {err}");
            }
            Err(1)
        }
    }
}

/// Resolve case file paths from path argument.
/// If path is a file, returns just that file.
/// If path is a directory (or None), auto-discovers `cases/**/*.md`.
pub fn resolve_case_files(
    path: Option<&str>,
    content_root: &std::path::Path,
) -> Result<Vec<String>, i32> {
    if let Some(p) = path {
        let p_path = std::path::Path::new(p);
        if p_path.is_file() {
            return Ok(vec![p.to_string()]);
        }
        if !p_path.is_dir() {
            eprintln!("{p}: not a file or directory");
            return Err(2);
        }
    }

    let cases_dir = content_root.join("cases");
    if !cases_dir.is_dir() {
        return Ok(Vec::new());
    }

    let mut files = Vec::new();
    discover_md_files(&cases_dir, &mut files, 0);
    files.sort();
    Ok(files)
}

/// Recursively discover .md files in a directory (max 5 levels deep for cases/country/category/year/).
fn discover_md_files(dir: &std::path::Path, files: &mut Vec<String>, depth: usize) {
    const MAX_DEPTH: usize = 5;
    if depth > MAX_DEPTH {
        return;
    }

    let Ok(entries) = std::fs::read_dir(dir) else {
        return;
    };

    let mut entries: Vec<_> = entries.filter_map(Result::ok).collect();
    entries.sort_by_key(std::fs::DirEntry::file_name);

    for entry in entries {
        let path = entry.path();
        if path.is_dir() {
            discover_md_files(&path, files, depth + 1);
        } else if path.extension().and_then(|e| e.to_str()) == Some("md")
            && let Some(s) = path.to_str()
        {
            files.push(s.to_string());
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn front_matter_has_id_present() {
        let content = "---\nid: 01JABC000000000000000000AA\n---\n\n# Test\n";
        assert!(front_matter_has_id(content));
    }

    #[test]
    fn front_matter_has_id_absent() {
        let content = "---\n---\n\n# Test\n";
        assert!(!front_matter_has_id(content));
    }

    #[test]
    fn front_matter_has_id_with_other_fields() {
        let content = "---\nother: value\nid: 01JABC000000000000000000AA\n---\n\n# Test\n";
        assert!(front_matter_has_id(content));
    }

    #[test]
    fn front_matter_has_id_no_front_matter() {
        let content = "# Test\n\nNo front matter here.\n";
        assert!(!front_matter_has_id(content));
    }

    #[test]
    fn front_matter_has_id_outside_front_matter() {
        // `id:` appearing in the body should not count
        let content = "---\n---\n\n# Test\n\n- id: some-value\n";
        assert!(!front_matter_has_id(content));
    }
}