"""Build review dossiers for HTP-family vesicle-type location rows.

Like ../membrane_review/build_dossiers.py, this decides nothing: it selects rows and
gathers what a reviewer needs to judge each one by hand. Decisions live in
decisions_draft.py, written after reading the dossier.

Candidates: existing_annotations rows with evidence_type HDA or HTP, not negated, a
vesicle-type term (extracellular exosome, extracellular vesicle, blood microparticle,
vesicle, extracellular membrane-bounded organelle, extracellular organelle), and an action
that is not already a settled non-core or removal call: MARK_AS_OVER_ANNOTATED, UNDECIDED,
ACCEPT or unset. These are the rows the withdrawn 2026-10-06 vesicle rule changed (or
flagged, for ACCEPT).

Each dossier carries:
  * the row: index, term, reference (+ cached title, which names the sample: urine,
    plasma, cell line...), current action, summary and reason
  * UniProt: SUBCELLULAR LOCATION and FUNCTION text (verbatim, trimmed), and whether the
    entry has a signal peptide or transmembrane/lipid-anchor features
  * how many HTP-family vesicle-type rows (and distinct references) the same gene carries
    in its review: one survey vs repeated detection across independent EV studies
  * leads: the gene's other location annotations from its *-goa.tsv (non-HTP codes),
    with code and reference; pointers to evidence, not evidence in themselves

Second pass (--settled): the rows reviewers had already set to KEEP_AS_NON_CORE, which
the first pass did not cover; written to dossiers_settled.yaml.

Usage:  python3 projects/OMICS_EVIDENCE/htp/vesicle_review/build_dossiers.py [--settled]
Writes: projects/OMICS_EVIDENCE/htp/vesicle_review/dossiers.yaml (or dossiers_settled.yaml)
"""

from __future__ import annotations

import argparse
import csv
import glob
import os
import re
import sys

import yaml

HERE = os.path.dirname(os.path.abspath(__file__))
ROOT = os.path.abspath(os.path.join(HERE, "..", "..", "..", ".."))
sys.path.insert(0, os.path.dirname(HERE))
sys.path.insert(0, os.path.join(os.path.dirname(HERE), "membrane_review"))
from apply_dispositions import CODES, HTP_FAMILY, LOADER, VESICLE_TERMS  # noqa: E402
from build_dossiers import cc_block, pub_title  # noqa: E402  (membrane_review helpers)

OUT = os.path.join(HERE, "dossiers.yaml")
# rows already decided by the first manual pass carry this note and are not re-reviewed
PASS1_NOTE = "Manual re-review for OMICS_EVIDENCE, 2026-10-10"
OPEN_ACTIONS = {"MARK_AS_OVER_ANNOTATED", "UNDECIDED", "ACCEPT", "PENDING", None}


def main() -> int:
    ap = argparse.ArgumentParser()
    ap.add_argument("--settled", action="store_true", help="review rows already set to KEEP_AS_NON_CORE")
    args = ap.parse_args()
    wanted = {"KEEP_AS_NON_CORE"} if args.settled else OPEN_ACTIONS
    out_path = os.path.join(HERE, "dossiers_settled.yaml") if args.settled else OUT
    dossiers = []
    for path in sorted(glob.glob(os.path.join(ROOT, "genes", "*", "*", "*-ai-review.yaml"))):
        text = open(path).read()
        if not any(t in text for t in VESICLE_TERMS):
            continue
        doc = yaml.load(text, Loader=LOADER)
        anns = doc.get("existing_annotations") or []
        ves = [(i, a) for i, a in enumerate(anns)
               if a.get("evidence_type") in CODES and not a.get("negated")
               and (a.get("term") or {}).get("id") in VESICLE_TERMS]
        rows = [(i, a) for i, a in ves if (a.get("review") or {}).get("action") in wanted
                and PASS1_NOTE not in str((a.get("review") or {}).get("reason") or "")]
        if not rows:
            continue
        gdir = os.path.dirname(path)
        org, gene = path.split(os.sep)[-3], path.split(os.sep)[-2]
        uni = os.path.join(gdir, f"{gene}-uniprot.txt")
        goa = os.path.join(gdir, f"{gene}-goa.tsv")
        utext = open(uni).read() if os.path.exists(uni) else ""
        features = sorted({f for f in re.findall(r"^FT   (SIGNAL|TRANSMEM|INTRAMEM|LIPID) ", utext, re.M)})
        leads = {}
        if os.path.exists(goa):
            with open(goa, newline="") as fh:
                for r in csv.DictReader(fh, delimiter="\t"):
                    if r.get("GO ASPECT") != "cellular_component" or r.get("GO EVIDENCE CODE") in HTP_FAMILY:
                        continue
                    key = (r["GO TERM"], r["GO NAME"], r["QUALIFIER"])
                    leads.setdefault(key, set()).add(f'{r["GO EVIDENCE CODE"]} {r["REFERENCE"]}')
        lead_list = [{"term": f"{t} {n}", "qualifier": q, "evidence": sorted(ev)}
                     for (t, n, q), ev in sorted(leads.items())]
        for idx, a in rows:
            rv = a.get("review") or {}
            dossiers.append({
                "organism": org, "gene": gene, "row_index": idx,
                "term": f'{a["term"]["id"]} {a["term"].get("label")}',
                "reference": a.get("original_reference_id"),
                "reference_title": pub_title(a.get("original_reference_id")),
                "evidence_type": a.get("evidence_type"),
                "current_action": rv.get("action"),
                "current_reason": " ".join(str(rv.get("reason") or "").split()),
                "uniprot_subcellular_location": cc_block(utext, "SUBCELLULAR LOCATION", 600),
                "uniprot_function": cc_block(utext, "FUNCTION", 300),
                "uniprot_features": features,
                "gene_vesicle_rows": len(ves),
                "gene_vesicle_references": len({a2.get("original_reference_id") for _, a2 in ves}),
                "leads": lead_list,
            })
    with open(out_path, "w") as fh:
        fh.write("# Generated by build_dossiers.py -- review material only; decisions are in decisions_draft.py\n")
        yaml.safe_dump(dossiers, fh, sort_keys=False, width=120, allow_unicode=True)
    print(f"{len(dossiers)} dossiers for {len({(d['organism'], d['gene']) for d in dossiers})} genes -> {out_path}")
    return 0


if __name__ == "__main__":
    sys.exit(main())