weave-content 0.2.32

//! High-level orchestration commands for content operations.
//!
//! These functions encapsulate the full pipeline for validate, verify,
//! and check-staleness operations. Both `weave-content` CLI and `loom-seed`
//! call into these functions.

use std::collections::{HashMap, HashSet};
use std::path::Path;

use rayon::prelude::*;

use crate::entity;
use crate::registry;
use crate::staleness;
use crate::tags;
use crate::verifier;
use crate::{load_registry, load_tag_registry, parse_full, resolve_case_files, resolve_content_root};

// ── Validate ────────────────────────────────────────────────────────────────

/// Run full validation on content files: parse, check schema, cross-file
/// duplicate detection, redirect collision checks, qualifier consistency.
///
/// When `strict` is true, warnings (e.g. filename mismatches) are treated
/// as errors and cause a non-zero exit code.
///
/// When `quiet` is true, per-entity and per-relationship detail lines are
/// suppressed. Only file-level summaries and errors/warnings are shown.
///
/// Returns exit code (0 = success).
pub fn validate(path: Option<&str>, root: Option<&str>, strict: bool, quiet: bool) -> i32 {
    let content_root = resolve_content_root(path, root);

    let redirect_slugs = load_redirect_slugs(&content_root);
    if !redirect_slugs.is_empty() {
        eprintln!("redirects: {} entries loaded", redirect_slugs.len());
        let collision_count = check_redirect_collisions(&redirect_slugs, &content_root);
        if collision_count > 0 {
            eprintln!(
                "error: {collision_count} file(s) at old redirected paths — move them or remove the redirect entry"
            );
            return 1;
        }
    }

    let reg = match load_registry(&content_root) {
        Ok(r) => r,
        Err(code) => return code,
    };

    let tag_reg = match load_tag_registry(&content_root) {
        Ok(r) => r,
        Err(code) => return code,
    };

    let case_files = match resolve_case_files(path, &content_root) {
        Ok(f) => f,
        Err(code) => return code,
    };

    if case_files.is_empty() {
        eprintln!("no case files found");
        return 1;
    }

    if !reg.is_empty() {
        eprintln!("registry: {} entities loaded", reg.len());
    }

    // In strict mode, filename mismatch warnings become errors
    if strict {
        let filename_warnings = reg.check_filenames();
        if !filename_warnings.is_empty() {
            for w in &filename_warnings {
                eprintln!("{w}");
            }
            eprintln!(
                "error: {} filename warning(s) treated as errors (--strict)",
                filename_warnings.len()
            );
            return 1;
        }
    }
    if !tag_reg.is_empty() {
        eprintln!(
            "tags: {} tags loaded across {} categories",
            tag_reg.len(),
            tag_reg.category_slugs().len()
        );
    }

    let mut entity_tag_errors = false;
    for entry in reg.entries() {
        let tag_errors = tag_reg.validate_tags(&entry.tags, 2);
        for err in &tag_errors {
            eprintln!("{}:{err}", entry.path.display());
        }
        if !tag_errors.is_empty() {
            entity_tag_errors = true;
        }
    }

    let results: Vec<ValidateResult> = case_files
        .par_iter()
        .map(|case_path| validate_single_case(case_path, &reg, &tag_reg, quiet))
        .collect();

    let mut exit_code = i32::from(entity_tag_errors);
    let mut all_events: Vec<(String, String)> = Vec::new();
    let mut all_rel_ids: Vec<(String, String, usize)> = Vec::new();

    for result in &results {
        if result.exit_code != 0 {
            exit_code = result.exit_code;
        }
        all_events.extend(result.events.iter().cloned());
        all_rel_ids.extend(result.rel_ids.iter().cloned());
    }

    if let Some(code) = check_duplicate_event_names(&all_events) {
        exit_code = code;
    }

    if let Some(code) = check_duplicate_rel_ids(&all_rel_ids) {
        exit_code = code;
    }

    check_qualifier_consistency(&reg, strict, &mut exit_code);

    // Print summary
    let ok_count = results.iter().filter(|r| r.exit_code == 0).count();
    let err_count = results.iter().filter(|r| r.exit_code != 0).count();
    let total_entities: usize = results.iter().map(|r| r.entity_count).sum();
    let total_rels: usize = results.iter().map(|r| r.rel_count).sum();
    eprintln!(
        "validate: {} case(s) ok, {} failed ({} entities, {} relationships)",
        ok_count, err_count, total_entities, total_rels
    );

    exit_code
}

struct ValidateResult {
    exit_code: i32,
    events: Vec<(String, String)>,
    rel_ids: Vec<(String, String, usize)>,
    entity_count: usize,
    rel_count: usize,
}

fn validate_single_case(
    path: &str,
    reg: &registry::EntityRegistry,
    tag_reg: &tags::TagRegistry,
    quiet: bool,
) -> ValidateResult {
    let content = match std::fs::read_to_string(path) {
        Ok(c) => c,
        Err(e) => {
            eprintln!("{path}: error reading file: {e}");
            return ValidateResult {
                exit_code: 2,
                events: Vec::new(),
                rel_ids: Vec::new(),
                entity_count: 0,
                rel_count: 0,
            };
        }
    };

    match parse_full(&content, Some(reg)) {
        Ok((case, entities, rels)) => {
            eprintln!(
                "{path}: ok -- {id}: {title} ({ent} entities, {rel} relationships, {src} sources)",
                id = case.id.as_deref().unwrap_or("(no id)"),
                title = case.title,
                ent = entities.len(),
                rel = rels.len(),
                src = case.sources.len(),
            );
            if !quiet {
                if !case.summary.is_empty() {
                    eprintln!(
                        "  summary: {}...",
                        &case.summary[..case.summary.len().min(80)]
                    );
                }
                for e in &entities {
                    let id_display = e.id.as_deref().unwrap_or("(no id)");
                    eprintln!(
                        "  line {}: {id_display} {} ({}, {} fields)",
                        e.line, e.name, e.label, e.fields.len()
                    );
                }
            }

            let events: Vec<(String, String)> = entities
                .iter()
                .filter(|e| e.label == entity::Label::Event)
                .map(|e| (e.name.clone(), path.to_string()))
                .collect();

            for r in &rels {
                if !quiet {
                    let id_display = r.id.as_deref().unwrap_or("(no id)");
                    eprintln!(
                        "  line {}: {id_display} {} -> {}: {}",
                        r.line, r.source_name, r.target_name, r.rel_type,
                    );
                }
            }

            let mut exit_code = 0;
            let tag_errors = tag_reg.validate_tags(&case.tags, 2);
            for err in &tag_errors {
                eprintln!("{path}:{err}");
            }
            if !tag_errors.is_empty() {
                exit_code = 1;
            }

            let rel_ids: Vec<(String, String, usize)> = rels
                .iter()
                .filter_map(|r| {
                    r.id.as_ref()
                        .map(|id| (id.clone(), path.to_string(), r.line))
                })
                .collect();

            ValidateResult {
                exit_code,
                events,
                rel_ids,
                entity_count: entities.len(),
                rel_count: rels.len(),
            }
        }
        Err(errors) => {
            for err in &errors {
                eprintln!("{path}:{err}");
            }
            ValidateResult {
                exit_code: 1,
                events: Vec::new(),
                rel_ids: Vec::new(),
                entity_count: 0,
                rel_count: 0,
            }
        }
    }
}

fn check_duplicate_event_names(all_events: &[(String, String)]) -> Option<i32> {
    let mut seen: HashMap<&str, &str> = HashMap::new();
    let mut has_duplicates = false;

    for (name, path) in all_events {
        if let Some(&first_path) = seen.get(name.as_str()) {
            eprintln!(
                "error: duplicate event name {name:?} in {path} (first defined in {first_path})"
            );
            has_duplicates = true;
        } else {
            seen.insert(name, path);
        }
    }

    if has_duplicates { Some(1) } else { None }
}

fn check_duplicate_rel_ids(all_rel_ids: &[(String, String, usize)]) -> Option<i32> {
    let mut seen: HashMap<&str, (&str, usize)> = HashMap::new();
    let mut has_duplicates = false;

    for (id, path, line) in all_rel_ids {
        if let Some(&(first_path, first_line)) = seen.get(id.as_str()) {
            eprintln!(
                "error: duplicate relationship id {id:?} at {path}:{line} (first defined at {first_path}:{first_line})"
            );
            has_duplicates = true;
        } else {
            seen.insert(id, (path, *line));
        }
    }

    if has_duplicates { Some(1) } else { None }
}

fn check_qualifier_consistency(reg: &registry::EntityRegistry, strict: bool, exit_code: &mut i32) {
    use entity::FieldValue;

    let mut by_lower: HashMap<String, Vec<(String, String)>> = HashMap::new();

    for entry in reg.entries() {
        let qualifier = entry
            .entity
            .fields
            .iter()
            .find(|(k, _)| k == "qualifier")
            .and_then(|(_, v)| match v {
                FieldValue::Single(s) => Some(s.as_str()),
                FieldValue::List(_) => None,
            });

        if let Some(q) = qualifier {
            by_lower
                .entry(q.to_lowercase())
                .or_default()
                .push((q.to_string(), entry.path.display().to_string()));
        }
    }

    let mut inconsistencies = 0usize;
    for occurrences in by_lower.values() {
        let first = &occurrences[0].0;
        let inconsistent: Vec<_> = occurrences.iter().filter(|(q, _)| q != first).collect();
        if !inconsistent.is_empty() {
            inconsistencies += 1;
            eprintln!(
                "warning: inconsistent qualifier casing for {:?}:",
                occurrences[0].0
            );
            for (q, path) in occurrences {
                eprintln!("  {path}: {q:?}");
            }
        }
    }

    if strict && inconsistencies > 0 {
        eprintln!(
            "error: {inconsistencies} qualifier consistency warning(s) treated as errors (--strict)"
        );
        *exit_code = 1;
    }
}

/// Load redirect `from:` slugs from `redirects.yaml` in the content root.
fn load_redirect_slugs(content_root: &Path) -> HashSet<String> {
    let path = content_root.join("redirects.yaml");
    let Ok(content) = std::fs::read_to_string(&path) else {
        return HashSet::new();
    };

    let mut slugs = HashSet::new();
    for line in content.lines() {
        if let Some(from) = line.strip_prefix("  - from: ") {
            slugs.insert(from.trim().to_string());
        }
    }
    slugs
}

fn check_redirect_collisions(redirect_slugs: &HashSet<String>, content_root: &Path) -> usize {
    if redirect_slugs.is_empty() {
        return 0;
    }

    let mut errors = 0;
    let root_str = content_root.to_string_lossy();

    for slug in redirect_slugs {
        let file_path = content_root.join(format!("{slug}.md"));
        if file_path.exists() {
            eprintln!(
                "{root_str}/{slug}.md: error: file at old redirected path (see redirects.yaml)"
            );
            errors += 1;
        }
    }

    errors
}

// ── Verify ──────────────────────────────────────────────────────────────────

/// Configuration for URL verification.
pub struct VerifyConfig {
    /// Maximum concurrent requests.
    pub concurrency: usize,
    /// Per-URL timeout in seconds.
    pub timeout: u64,
    /// Path to URL verification cache file.
    pub cache_path: Option<String>,
    /// Report all as warnings, never fail.
    pub warn_only: bool,
}

/// Run URL verification on content files.
///
/// Returns exit code (0 = success).
pub fn verify(path: Option<&str>, root: Option<&str>, config: &VerifyConfig) -> i32 {
    let content_root = resolve_content_root(path, root);
    let reg = match load_registry(&content_root) {
        Ok(r) => r,
        Err(code) => return code,
    };

    let case_files = match resolve_case_files(path, &content_root) {
        Ok(f) => f,
        Err(code) => return code,
    };

    if case_files.is_empty() {
        eprintln!("no case files found");
        return 1;
    }

    let mut exit_code = 0;
    for case_path in &case_files {
        let result = verify_single_case(case_path, &reg, config);
        if result != 0 {
            exit_code = result;
        }
    }

    let reg_result = verify_registry_thumbnails(&reg, config);
    if reg_result != 0 {
        exit_code = reg_result;
    }

    exit_code
}

fn verify_single_case(
    path: &str,
    reg: &registry::EntityRegistry,
    config: &VerifyConfig,
) -> i32 {
    let content = match std::fs::read_to_string(path) {
        Ok(c) => c,
        Err(e) => {
            eprintln!("{path}: error reading file: {e}");
            return 2;
        }
    };

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

    let mut collect_errors = Vec::new();
    let urls = verifier::collect_urls(&case.sources, &entities, &rels, &mut collect_errors);

    if !collect_errors.is_empty() {
        for err in &collect_errors {
            eprintln!("{path}:{err}");
        }
        return 1;
    }

    if urls.is_empty() {
        eprintln!("{path}: no URLs to verify");
        return 0;
    }

    run_url_verification(path, &urls, config)
}

fn verify_registry_thumbnails(
    reg: &registry::EntityRegistry,
    config: &VerifyConfig,
) -> i32 {
    let urls = verifier::collect_registry_urls(reg);
    if urls.is_empty() {
        return 0;
    }
    run_url_verification("(registry)", &urls, config)
}

fn run_url_verification(
    label: &str,
    urls: &[verifier::UrlEntry],
    config: &VerifyConfig,
) -> i32 {
    let mut verify_cache = load_verify_cache(label, config.cache_path.as_deref());

    let (cached_results, urls_to_check) = partition_cached(urls, verify_cache.as_ref());
    let check_count = urls_to_check.len();
    let cached_count = cached_results.len();

    if cached_count > 0 {
        eprintln!(
            "{label}: {cached_count} cached, {check_count} to check \
             (concurrency={}, timeout={}s)",
            config.concurrency, config.timeout
        );
    } else {
        eprintln!(
            "{label}: verifying {check_count} URLs \
             (concurrency={}, timeout={}s)",
            config.concurrency, config.timeout
        );
    }

    let fresh_results = if urls_to_check.is_empty() {
        Vec::new()
    } else {
        let rt = match tokio::runtime::Builder::new_current_thread()
            .enable_all()
            .build()
        {
            Ok(rt) => rt,
            Err(e) => {
                eprintln!("{label}: failed to create async runtime: {e}");
                return 2;
            }
        };
        rt.block_on(verifier::verify_urls(
            urls_to_check,
            config.concurrency,
            config.timeout,
        ))
    };

    if let Some(ref mut vc) = verify_cache {
        for check in &fresh_results {
            vc.put(&check.url, check.status, check.detail.as_deref());
        }
    }

    let mut all_results = cached_results;
    all_results.extend(fresh_results);

    let has_error = print_verification_results(label, &all_results);

    if let Some(ref vc) = verify_cache
        && let Err(e) = vc.save()
    {
        eprintln!("{label}: cache save warning: {e}");
    }

    i32::from(has_error && !config.warn_only)
}

fn load_verify_cache(
    label: &str,
    cache_path: Option<&str>,
) -> Option<crate::cache::VerifyCache> {
    cache_path.map(|p| match crate::cache::VerifyCache::load(p) {
        Ok(c) => {
            eprintln!("{label}: using cache {p}");
            c
        }
        Err(e) => {
            eprintln!("{label}: cache load warning: {e}");
            crate::cache::VerifyCache::load("/dev/null")
                .unwrap_or_else(|_| crate::cache::VerifyCache::empty())
        }
    })
}

fn print_verification_results(label: &str, results: &[verifier::UrlCheck]) -> bool {
    let mut has_error = false;

    for check in results {
        let detail = check.detail.as_deref().unwrap_or("");
        match check.status {
            verifier::CheckStatus::Ok => {
                eprintln!(
                    "  ok  {}{}",
                    check.url,
                    if check.is_thumbnail {
                        " [thumbnail]"
                    } else {
                        ""
                    }
                );
            }
            verifier::CheckStatus::Warn => {
                eprintln!("  warn  {} -- {detail}", check.url);
            }
            verifier::CheckStatus::Error => {
                has_error = true;
                eprintln!("  ERROR {} -- {detail}", check.url);
            }
        }
    }

    let ok_count = results
        .iter()
        .filter(|c| c.status == verifier::CheckStatus::Ok)
        .count();
    let warn_count = results
        .iter()
        .filter(|c| c.status == verifier::CheckStatus::Warn)
        .count();
    let err_count = results
        .iter()
        .filter(|c| c.status == verifier::CheckStatus::Error)
        .count();

    eprintln!("{label}: {ok_count} ok, {warn_count} warn, {err_count} error");

    has_error
}

fn partition_cached(
    urls: &[verifier::UrlEntry],
    verify_cache: Option<&crate::cache::VerifyCache>,
) -> (Vec<verifier::UrlCheck>, Vec<verifier::UrlEntry>) {
    let Some(vc) = verify_cache else {
        return (Vec::new(), urls.to_vec());
    };

    let mut cached = Vec::new();
    let mut uncached = Vec::new();

    for entry in urls {
        if let Some(cache_entry) = vc.get(entry.url()) {
            let status = match cache_entry.status.as_str() {
                "ok" => verifier::CheckStatus::Ok,
                "warn" => verifier::CheckStatus::Warn,
                _ => verifier::CheckStatus::Error,
            };
            cached.push(verifier::UrlCheck {
                url: entry.url().to_string(),
                status,
                detail: cache_entry.detail.clone(),
                is_thumbnail: entry.is_thumbnail(),
            });
        } else {
            uncached.push(entry.clone());
        }
    }

    (cached, uncached)
}

// ── Check Staleness ─────────────────────────────────────────────────────────

/// Configuration for staleness checking.
pub struct StalenessConfig {
    /// Months before an `under_investigation` case is considered stale.
    pub investigation_months: u32,
    /// Months before a `trial` case is considered stale.
    pub trial_months: u32,
    /// Months before an `appeal` case is considered stale.
    pub appeal_months: u32,
}

impl Default for StalenessConfig {
    fn default() -> Self {
        Self {
            investigation_months: 6,
            trial_months: 12,
            appeal_months: 12,
        }
    }
}

/// Run staleness checks on content files.
///
/// Returns exit code (0 = success, 1 = errors found).
pub fn check_staleness(path: Option<&str>, root: Option<&str>, config: &StalenessConfig) -> i32 {
    let content_root = resolve_content_root(path, root);
    let reg = match load_registry(&content_root) {
        Ok(r) => r,
        Err(code) => return code,
    };

    let case_files = match resolve_case_files(path, &content_root) {
        Ok(f) => f,
        Err(code) => return code,
    };

    if case_files.is_empty() {
        eprintln!("no case files found");
        return 1;
    }

    let thresholds = staleness::Thresholds {
        investigation_months: config.investigation_months,
        trial_months: config.trial_months,
        appeal_months: config.appeal_months,
    };

    let now = chrono_today();

    let mut all_findings: Vec<(String, staleness::Finding)> = Vec::new();

    for case_path in &case_files {
        let content = match std::fs::read_to_string(case_path) {
            Ok(c) => c,
            Err(e) => {
                eprintln!("{case_path}: error reading file: {e}");
                continue;
            }
        };

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

        let findings = staleness::check_case(&case, &entities, &thresholds, now);
        for finding in findings {
            all_findings.push((case_path.clone(), finding));
        }
    }

    all_findings.sort_by_key(|a| a.1.severity);

    let mut errors = 0u32;
    let mut warnings = 0u32;
    let mut infos = 0u32;

    for (path, finding) in &all_findings {
        eprintln!("{}: {path}: {}", finding.severity, finding.message);
        match finding.severity {
            staleness::Severity::Error => errors += 1,
            staleness::Severity::Warning => warnings += 1,
            staleness::Severity::Info => infos += 1,
        }
    }

    eprintln!(
        "staleness: {errors} error(s), {warnings} warning(s), {infos} info(s) across {} case(s)",
        case_files.len()
    );

    i32::from(errors > 0)
}

/// Get today's date as (year, month, day).
fn chrono_today() -> (i32, u32, u32) {
    let now = std::time::SystemTime::now();
    let since_epoch = now
        .duration_since(std::time::UNIX_EPOCH)
        .unwrap_or_default();
    let days = since_epoch.as_secs() / 86400;
    days_to_date(days)
}

/// Convert days since Unix epoch to (year, month, day).
/// Uses the algorithm from Howard Hinnant's date library.
#[allow(
    clippy::cast_possible_truncation,
    clippy::cast_sign_loss,
    clippy::cast_possible_wrap
)]
fn days_to_date(days: u64) -> (i32, u32, u32) {
    let z = i64::from(u32::try_from(days).unwrap_or(u32::MAX)) + 719_468;
    let era = if z >= 0 { z } else { z - 146_096 } / 146_097;
    let doe = (z - era * 146_097) as u32;
    let yoe = (doe - doe / 1460 + doe / 36524 - doe / 146_096) / 365;
    let y = (i64::from(yoe) + era * 400) as i32;
    let doy = doe - (365 * yoe + yoe / 4 - yoe / 100);
    let mp = (5 * doy + 2) / 153;
    let d = doy - (153 * mp + 2) / 5 + 1;
    let m = if mp < 10 { mp + 3 } else { mp - 9 };
    let y = if m <= 2 { y + 1 } else { y };
    (y, m, d)
}

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

    #[test]
    fn days_to_date_epoch() {
        assert_eq!(days_to_date(0), (1970, 1, 1));
    }

    #[test]
    fn days_to_date_known() {
        // 2025-01-01 = day 20089
        assert_eq!(days_to_date(20089), (2025, 1, 1));
    }

    #[test]
    fn redirect_slugs_empty_on_missing_file() {
        let slugs = load_redirect_slugs(Path::new("/nonexistent"));
        assert!(slugs.is_empty());
    }

    #[test]
    fn staleness_config_defaults() {
        let config = StalenessConfig::default();
        assert_eq!(config.investigation_months, 6);
        assert_eq!(config.trial_months, 12);
        assert_eq!(config.appeal_months, 12);
    }
}