commonmeta 0.9.6

/*
 * Copyright © 2026 Front Matter <info@front-matter.de>
 */

use clap::{Arg, ArgAction, ArgMatches, Command};
use std::path::Path;

use crate::cmd::resolve_db_path;
use crate::doi_utils::get_doi_ra_sync;

pub fn command() -> Command {
    Command::new("convert")
        .about("Convert scholarly metadata between formats")
        .long_about(
            "Convert scholarly metadata between formats.\n\n\
            The input is a file path, a DOI/URL, or a ROR organization ID. \
            When --from is omitted the format is auto-detected: DOIs are \
            resolved via the DOI RA API; ROR URLs are detected by pattern; \
            JSON files are inspected for schema markers.\n\n\
            For ROR input, a local 'commonmeta.sqlite3' in the current \
            directory (produced by 'commonmeta list --to ror --file \
            commonmeta.sqlite3') is queried first — faster and offline. \
            The ROR API is used as a fallback when no local database exists.\n\n\
            Supported input formats:  crossref, commonmeta, ror\n\
            Supported output formats: commonmeta, csl, ror, inveniordm\n\n\
            Examples:\n\n\
            commonmeta convert 10.5555/12345678\n\
            commonmeta convert https://doi.org/10.59350/gj8re-sca95 --to csl\n\
            commonmeta convert https://ror.org/02nr0ka47\n\
            commonmeta convert https://ror.org/02nr0ka47 --to inveniordm\n\
            commonmeta convert record.json --from commonmeta --to csl --file out.json",
        )
        .arg(
            Arg::new("input")
                .help("File path, DOI, URL, or ROR ID")
                .required(true)
                .index(1),
        )
        .arg(
            Arg::new("from")
                .long("from")
                .short('f')
                .help("Input format (crossref, commonmeta, ror); auto-detected if omitted"),
        )
        .arg(
            Arg::new("to")
                .long("to")
                .short('t')
                .help("Output format (commonmeta, csl, ror, inveniordm)")
                .default_value("commonmeta"),
        )
        .arg(
            Arg::new("style")
                .long("style")
                .short('s')
                .help("CSL style name for citation output (default: apa)"),
        )
        .arg(
            Arg::new("locale")
                .long("locale")
                .short('l')
                .help("BCP 47 locale for citation output (e.g. de-DE)"),
        )
        .arg(
            Arg::new("file")
                .long("file")
                .help("Write output to this file instead of stdout"),
        )
        .arg(
            Arg::new("no-network")
                .long("no-network")
                .help("Disable all outbound network requests; fails if the operation would require network access")
                .action(ArgAction::SetTrue),
        )
        .arg(
            Arg::new("network")
                .long("network")
                .help("For ORCID input: also query Crossref and DataCite even when local results exist; results are cached in cache.sqlite3")
                .action(ArgAction::SetTrue),
        )
        .arg(
            Arg::new("number")
                .long("number")
                .short('n')
                .help("For ORCID input: number of works to fetch per source (Crossref, DataCite); 0 fetches all (default: 100)")
                .value_parser(clap::value_parser!(usize))
                .default_value("100"),
        )
        .arg(
            Arg::new("page")
                .long("page")
                .short('p')
                .help("For ORCID input: page number (1-based) for Crossref and DataCite API requests")
                .value_parser(clap::value_parser!(usize))
                .default_value("1"),
        )
}

// ─── Format detection ─────────────────────────────────────────────────────────

fn doi_prefix(s: &str) -> Option<String> {
    let doi = s
        .trim_start_matches("https://doi.org/")
        .trim_start_matches("http://doi.org/")
        .trim_start_matches("https://dx.doi.org/")
        .trim_start_matches("http://dx.doi.org/")
        .trim_start_matches("https://commonmeta.org/")
        .trim_start_matches("http://commonmeta.org/");
    if doi.starts_with("10.") {
        doi.find('/').map(|i| doi[..i].to_string())
    } else {
        None
    }
}

/// Map a DOI registration agency name (as returned by the RA API) to the
/// commonmeta reader identifier. Returns `None` for RAs that have no reader.
pub(crate) fn ra_to_reader(ra: &str) -> Option<&'static str> {
    match ra {
        "Crossref" | "crossref" => Some("crossref"),
        "DataCite" | "datacite" => Some("datacite"),
        _ => None,
    }
}

pub(crate) fn detect_format(input: &str, no_network: bool) -> String {
    // Use validate_id for structured identifiers before falling through to DOI/URL.
    let id_type = crate::utils::validate_id(input).1;
    match id_type {
        "ROR" => return "ror".to_string(),
        "ORCID" => return "orcid".to_string(),
        _ => {}
    }
    // DOI URL or bare DOI → look up registration agency (cache-backed).
    // Returns the raw RA name (e.g. "Crossref", "mEDRA") so callers can
    // distinguish "no reader available" from "unknown RA".
    if doi_prefix(input).is_some() {
        return get_doi_ra_sync(input, no_network)
            .unwrap_or_else(|| "crossref".to_string());
    }
    // OpenAlex work ID (Wxx or https://openalex.org/Wxx)
    if crate::utils::validate_openalex(input)
        .map(|id| id.starts_with('W'))
        .unwrap_or(false)
    {
        return "openalex".to_string();
    }
    // PubMed/Europe PMC URLs and PMCIDs (bare PMIDs are too ambiguous to auto-detect)
    if input.contains("pubmed.ncbi.nlm.nih.gov")
        || input.contains("europepmc.org/article")
        || input.contains("pmc.ncbi.nlm.nih.gov/articles")
        || (input.starts_with("PMC") && input[3..].chars().all(|c| c.is_ascii_digit()) && !input[3..].is_empty())
    {
        return "pubmed".to_string();
    }
    // JSON content → inspect schema markers
    if let Ok(v) = serde_json::from_str::<serde_json::Value>(input) {
        if v.get("schema_version")
            .and_then(|s| s.as_str())
            .map(|s| s.starts_with("https://commonmeta.org"))
            .unwrap_or(false)
        {
            return "commonmeta".to_string();
        }
        // Crossref API envelope
        if v.get("message-type").is_some() {
            return "crossref".to_string();
        }
        // ROR record
        if v.get("id")
            .and_then(|s| s.as_str())
            .map(|s| s.starts_with("https://ror.org/"))
            .unwrap_or(false)
            && v.get("names").is_some()
        {
            return "ror".to_string();
        }
    }
    "commonmeta".to_string()
}

// ─── Execute ─────────────────────────────────────────────────────────────────

pub fn execute(matches: &ArgMatches) -> Result<(), String> {
    let input_arg = matches.get_one::<String>("input").expect("required");
    let out_file = matches.get_one::<String>("file");
    let no_network = matches.get_flag("no-network");
    let force_network = matches.get_flag("network");
    let number = *matches.get_one::<usize>("number").unwrap_or(&10);
    let page = *matches.get_one::<usize>("page").unwrap_or(&1);
    let style = matches.get_one::<String>("style").map(String::as_str);
    let locale = matches.get_one::<String>("locale").map(String::as_str);
    let to_arg = matches.get_one::<String>("to").expect("has default");

    let is_local_file = Path::new(input_arg).exists();

    // When --no-network is set and the input is a DOI/URL, look it up in the
    // local commonmeta database instead of fetching from the API.
    if no_network && !is_local_file && doi_prefix(input_arg).is_some() {
        let bare = input_arg
            .trim_start_matches("https://doi.org/")
            .trim_start_matches("http://doi.org/")
            .trim_start_matches("https://dx.doi.org/")
            .trim_start_matches("http://dx.doi.org/");
        let doi_url = format!("https://doi.org/{}", bare);
        let db_path_str = resolve_db_path(None);
        let db_path = Path::new(&db_path_str);
        if !db_path.exists() {
            return Err(format!(
                "local database not found at '{}'; \
                run 'commonmeta import {}' or remove --no-network",
                db_path_str, input_arg
            ));
        }
        let not_found = || format!(
            "'{}' not found in local database '{}'; \
            run 'commonmeta import {}' or remove --no-network",
            input_arg, db_path_str, input_arg
        );
        let data = match crate::read_sqlite_by_id(&doi_url, db_path) {
            Ok(Some(d)) => d,
            Ok(None) => return Err(not_found()),
            // "no such table" means the DB exists but migrations haven't run yet.
            Err(e) if e.to_string().contains("no such table") => return Err(not_found()),
            Err(e) => return Err(e.to_string()),
        };
        let json = serde_json::to_string(&data).map_err(|e| e.to_string())?;
        let to_arg = to_arg.as_str();
        let output = if to_arg == "citation" {
            crate::convert_citation("commonmeta", &json, style, locale)
                .map_err(|e| e.to_string())?
        } else {
            crate::convert("commonmeta", to_arg, &json).map_err(|e| e.to_string())?
        };
        return write_output(&output, to_arg, out_file);
    }

    let input = if is_local_file {
        std::fs::read_to_string(input_arg)
            .map_err(|e| format!("failed to read '{}': {}", input_arg, e))?
    } else {
        input_arg.clone()
    };

    let from_explicit = matches.get_one::<String>("from");
    let from_detected = from_explicit.is_none().then(|| detect_format(&input, no_network));
    // When auto-detected from a DOI and the RA has no reader, error early.
    if let Some(ref ra) = from_detected {
        if doi_prefix(input_arg).is_some() && ra_to_reader(ra).is_none() {
            return Err(format!(
                "convert: no reader available for DOI registration agency '{}' (prefix {})",
                ra,
                doi_prefix(input_arg).unwrap_or_default()
            ));
        }
    }
    let from = from_explicit
        .map(|f| f.as_str().to_string())
        .or_else(|| from_detected.as_deref().and_then(ra_to_reader).map(str::to_string))
        .unwrap_or_else(|| from_detected.unwrap_or_else(|| "commonmeta".to_string()));

    let to_arg = to_arg.as_str();
    let to = to_arg;

    // ── ORCID input path ─────────────────────────────────────────────────────
    if from == "orcid" {
        let orcid_url = crate::utils::normalize_orcid(&input);
        if orcid_url.is_empty() {
            return Err(format!("'{}' is not a valid ORCID identifier", input));
        }

        // Prefer the local people database; fall back to the ORCID public API.
        // Both paths return ORCID 3.0 person JSON (orcid_schema_v3.0.json).
        let db_path_str = resolve_db_path(None);
        let db_path = Path::new(&db_path_str);
        let value = if db_path.exists() {
            match crate::fetch_orcid_person_json_sqlite(&orcid_url, db_path) {
                Ok(v) => Ok(v),
                Err(_) if no_network => return Err(format!(
                    "'{}' not found in local database '{}'; \
                    run 'commonmeta import {}' or remove --no-network",
                    input, db_path_str, input
                )),
                Err(_) => crate::fetch_orcid_person_json(&orcid_url),
            }
        } else if no_network {
            return Err(format!(
                "ORCID lookup requires network access (no local database at '{}'); \
                run 'commonmeta import {}' or remove --no-network",
                db_path_str, input
            ));
        } else {
            crate::fetch_orcid_person_json(&orcid_url)
        }
        .map_err(|e| e.to_string())?;

        let needs_affiliations = matches!(to, "commonmeta" | "inveniordm");
        let affiliations = if needs_affiliations {
            // SQLite-stored affiliations first; fall back to live API when empty.
            let sqlite_affs = if db_path.exists() {
                crate::fetch_orcid_affiliations_sqlite(&orcid_url, db_path)
            } else {
                Vec::new()
            };
            if !sqlite_affs.is_empty() {
                sqlite_affs
            } else if !no_network {
                let ror_db = if db_path.exists() { Some(db_path) } else { None };
                crate::fetch_orcid_affiliations(&orcid_url, ror_db).unwrap_or_default()
            } else {
                Vec::new()
            }
        } else {
            Vec::new()
        };

        // For commonmeta output: SQLite first, network fallback (or --network to always query).
        let works = if to == "commonmeta" {
            // 1. Query local SQLite by ORCID (page 1 only; pagination applies to network calls).
            let sqlite_works = if db_path.exists() && page == 1 {
                crate::read_sqlite_by_orcid(&orcid_url, db_path).unwrap_or_default()
            } else {
                Vec::new()
            };
            let need_network = !no_network && (sqlite_works.is_empty() || force_network || page > 1);

            // 2. Query Crossref + DataCite when local has no results or --network/--page is set.
            let network_works = if need_network {
                if number == 0 {
                    let mut cr = crate::fetch_all_crossref_by_orcid(&orcid_url).unwrap_or_default();
                    let mut dc = crate::fetch_all_datacite_by_orcid(&orcid_url).unwrap_or_default();
                    cr.append(&mut dc);
                    cr
                } else {
                    let mut cr = crate::fetch_crossref_by_orcid(&orcid_url, number, page)
                        .unwrap_or_default();
                    let mut dc = crate::fetch_datacite_by_orcid(&orcid_url, number, page)
                        .unwrap_or_default();
                    cr.append(&mut dc);
                    cr
                }
            } else {
                Vec::new()
            };

            // 3. Write network results to cache.sqlite3 for later migration.
            if !network_works.is_empty() {
                let cache_path_str = crate::cmd::resolve_cache_db_path(None);
                let cache_path = std::path::Path::new(&cache_path_str);
                if let Some(parent) = cache_path.parent() {
                    let _ = std::fs::create_dir_all(parent);
                }
                let _ = crate::upsert_sqlite(&network_works, cache_path);
            }

            // 4. Merge, deduplicate by id, sort by date desc; truncate only when a limit is set.
            let mut seen = std::collections::HashSet::new();
            let mut all: Vec<crate::Data> = sqlite_works
                .into_iter()
                .chain(network_works)
                .filter(|d| seen.insert(d.id.clone()))
                .collect();
            all.sort_by(|a, b| b.date_published.cmp(&a.date_published));
            if number > 0 {
                all.truncate(number);
            }
            all
        } else {
            Vec::new()
        };

        let output = match to {
            "inveniordm" => crate::write_orcid_inveniordm_yaml(&value, &affiliations)
                .map_err(|e| e.to_string())?,
            "orcid" => crate::write_orcid_json(&value).map_err(|e| e.to_string())?,
            _ => crate::write_orcid_commonmeta(&value, &affiliations, &works)
                .map_err(|e| e.to_string())?,
        };
        return write_output(&output, to, out_file);
    }

    // ── ROR input path ────────────────────────────────────────────────────────
    if from == "ror" {
        // Normalize the input to a full ROR URL for the SQLite lookup.
        let ror_url = crate::utils::normalize_ror(&input);
        if ror_url.is_empty() {
            return Err(format!("'{}' is not a valid ROR identifier", input));
        }

        let db_path_str = resolve_db_path(None);
        let db_path = Path::new(&db_path_str);

        // For --to ror (and any other non-commonmeta, non-inveniordm target):
        // bypass the lossy Data conversion and serialize the raw Ror struct
        // directly, enriching geonames_details from the local GeoNames SQLite
        // if available.
        if to != "commonmeta" && to != "inveniordm" {
            let mut raw = if db_path.exists() {
                match crate::fetch_ror_raw_sqlite(&ror_url, db_path) {
                    Ok(r) => Ok(r),
                    Err(_) if no_network => return Err(format!(
                        "'{}' not found in local database '{}'; run 'commonmeta import --from ror'",
                        input, db_path_str
                    )),
                    Err(_) => crate::fetch_ror_raw(&ror_url),
                }
            } else if no_network {
                return Err(format!(
                    "ROR lookup requires network access (no local database at '{}'); run 'commonmeta import --from ror'",
                    db_path_str
                ));
            } else {
                crate::fetch_ror_raw(&ror_url)
            }.map_err(|e| e.to_string())?;

            // Enrich geonames_details if local GeoNames data is available.
            if db_path.exists() {
                crate::enrich_ror_locations(&mut raw, db_path);
            }

            let output = crate::write_ror_v2_json(&raw).map_err(|e| e.to_string())?;
            return write_output(&output, to, out_file);
        }

        // Prefer the local SQLite database (COMMONMETA_DB > platform default);
        // fall back to the ROR API unless --no-network was requested.
        let data = if db_path.exists() {
            match crate::fetch_ror_sqlite(&ror_url, db_path) {
                Ok(d) => Ok(d),
                Err(_) if no_network => return Err(format!(
                    "'{}' not found in local database '{}'; \
                    run 'commonmeta import --from ror' or remove --no-network",
                    input, db_path_str
                )),
                Err(_) => crate::fetch_ror(&ror_url),
            }
        } else if no_network {
            return Err(format!(
                "ROR lookup requires network access (no local database at '{}'); \
                run 'commonmeta import --from ror' or remove --no-network",
                db_path_str
            ));
        } else {
            crate::fetch_ror(&ror_url)
        }
        .map_err(|e| e.to_string())?;

        if to == "inveniordm" {
            let output = crate::write("ror", &data).map_err(|e| e.to_string())?;
            return write_output(&output, to, out_file);
        }

        // commonmeta output: organization + works array.
        let works_db = Path::new(&db_path_str);
        let sqlite_works = if works_db.exists() && page == 1 {
            crate::read_sqlite_by_ror(&ror_url, works_db).unwrap_or_default()
        } else {
            Vec::new()
        };
        let need_network = !no_network && (sqlite_works.is_empty() || force_network || page > 1);
        let network_works = if need_network {
            let mut cr = crate::fetch_crossref_by_ror(&ror_url, number, page).unwrap_or_default();
            let mut dc = crate::fetch_datacite_by_ror(&ror_url, number, page).unwrap_or_default();
            cr.append(&mut dc);
            cr
        } else {
            Vec::new()
        };
        if !network_works.is_empty() {
            let cache_path_str = crate::cmd::resolve_cache_db_path(None);
            let cache_path = std::path::Path::new(&cache_path_str);
            if let Some(parent) = cache_path.parent() {
                let _ = std::fs::create_dir_all(parent);
            }
            let _ = crate::upsert_sqlite(&network_works, cache_path);
        }
        let mut seen = std::collections::HashSet::new();
        let mut all: Vec<crate::Data> = sqlite_works
            .into_iter()
            .chain(network_works)
            .filter(|d| seen.insert(d.id.clone()))
            .collect();
        all.sort_by(|a, b| b.date_published.cmp(&a.date_published));
        all.truncate(number);

        // Serialize as [org_entity, work1, work2, …] JSON array.
        let org_val = serde_json::to_value(&data).map_err(|e| e.to_string())?;
        let mut items = vec![org_val];
        for work in &all {
            let prepared = crate::prepare_commonmeta(work);
            items.push(serde_json::to_value(&prepared).map_err(|e| e.to_string())?);
        }
        let output = serde_json::to_vec_pretty(&items).map_err(|e| e.to_string())?;
        return write_output(&output, to, out_file);
    }

    // ── Scholarly-work input path ─────────────────────────────────────────────
    // For commonmeta output: return [work, ref1, ref2, …] array.
    if to == "commonmeta" {
        let mut data = crate::read(&from, &input).map_err(|e| e.to_string())?;
        let db_path_str = resolve_db_path(None);
        let db_path = Path::new(&db_path_str);
        crate::enrich_citations(&mut data, db_path);
        let related_works = crate::fetch_reference_works(&data, Some(db_path), no_network);
        let mut main_prepared = crate::prepare_commonmeta(&data);
        main_prepared.references.clear();
        let main = serde_json::to_value(main_prepared).map_err(|e| e.to_string())?;
        let mut items = vec![main];
        for work in &related_works {
            let mut prepared = crate::prepare_commonmeta(work);
            prepared.references.clear();
            items.push(serde_json::to_value(prepared).map_err(|e| e.to_string())?);
        }
        let output = serde_json::to_vec_pretty(&items).map_err(|e| e.to_string())?;
        return write_output(&output, to, out_file);
    }

    let output = if to == "citation" {
        crate::convert_citation(&from, &input, style, locale).map_err(|e| e.to_string())?
    } else {
        crate::convert(&from, to, &input).map_err(|e| e.to_string())?
    };

    write_output(&output, to, out_file)
}

fn write_output(output: &[u8], to: &str, out_file: Option<&String>) -> Result<(), String> {
    // JSON formats get pretty-printed; XML/YAML stay as-is.
    let formatted: Vec<u8> = if matches!(to, "inveniordm") {
        output.to_vec()
    } else {
        serde_json::from_slice::<serde_json::Value>(output)
            .ok()
            .and_then(|v| serde_json::to_vec_pretty(&v).ok())
            .unwrap_or_else(|| output.to_vec())
    };

    match out_file {
        Some(path) => std::fs::write(path, &formatted)
            .map_err(|e| format!("failed to write '{}': {}", path, e)),
        None => {
            println!("{}", String::from_utf8_lossy(&formatted));
            Ok(())
        }
    }
}

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

    fn parse_args(args: &[&str]) -> clap::ArgMatches {
        command().try_get_matches_from(args).expect("arg parse failed")
    }

    #[test]
    fn test_no_network_with_doi_uses_local_db() {
        // With --no-network and a DOI, convert must look up the local database
        // instead of calling the API. Either the record is found and returned
        // (Ok), or the DB/record is absent and a "not found" error is returned.
        // In neither case should the error be an API-fetch refusal.
        let m = parse_args(&["convert", "--no-network", "10.7554/elife.01567"]);
        match execute(&m) {
            Ok(()) => {}
            Err(e) => assert!(
                e.contains("not found") || e.contains("--no-network"),
                "expected local-db error, got: {e}"
            ),
        }
    }

    #[test]
    fn test_no_network_with_doi_url_uses_local_db() {
        let m = parse_args(&["convert", "--no-network", "https://doi.org/10.7554/elife.01567"]);
        match execute(&m) {
            Ok(()) => {}
            Err(e) => assert!(
                e.contains("not found") || e.contains("--no-network"),
                "expected local-db error, got: {e}"
            ),
        }
    }

    #[test]
    fn test_no_network_with_local_json_passes_guard() {
        // Non-DOI non-file input (inline JSON) is not blocked by the network guard.
        // Fails at parse time because the JSON is not valid commonmeta — but NOT
        // with a --no-network error.
        let m = parse_args(&["convert", "--no-network", r#"{"type":"JournalArticle"}"#]);
        let err = execute(&m).unwrap_err();
        assert!(
            !err.contains("--no-network"),
            "should not fail at network guard for inline JSON, got: {err}"
        );
    }
}