osint/tools/migrate_external_orgs.py

#!/usr/bin/env python3
"""Fichas-indice de organizaciones EXTERNAS (proveedores, bancos, telcos, gov) en osint.

A diferencia de las empresas propias (FenixFood/Biorganic), las externas ya aparecen como
documentos dentro de personas/ y organizaciones/fenixfood-sl/. Este script:

  1. Crea una ficha organizaciones/<slug>.md (tipo: organizacion, externa: true) por cada
     empresa externa.
  2. Cross-referencia (## Relacionado) los documentos ya existentes en osint donde aparece
     la empresa, SIN moverlos (siguen bajo su dueño: FenixFood o la persona).
  3. Para las empresas con notas sueltas reales en NotasDeObsidian (move_loose=True), mueve
     esas notas a organizaciones/<slug>/<doc-slug>.md con sus attachments.

Idempotente (overwrite). Modo MOVER para las sueltas (borra original); backup tar cubre.
"""
import sys, os, re, shutil

sys.path.insert(0, "/home/enmanuel/fn_registry/python/functions")
from obsidian import (read_obsidian_note, create_obsidian_note,
                      slugify_obsidian_name, extract_obsidian_embeds, resolve_obsidian_embed,
                      list_obsidian_notes)

NOTAS = "/home/enmanuel/Obsidian/NotasDeObsidian"
OSINT = "/home/enmanuel/Obsidian/osint"
STOPWORDS = {"de", "del", "la", "las", "el", "los", "y", "a", "en"}

# (display, slug, regex_match, regex_excluir_falsos_positivos, move_loose)
EXT = [
    ("Santander", "santander", r"santander", None, False),
    ("Unicaja", "unicaja", r"unicaja", None, True),
    ("Abanca", "abanca", r"abanca", None, False),
    ("BBVA", "bbva", r"\bbbva\b", None, False),
    ("Cajamar", "cajamar", r"cajamar", None, False),
    ("Vodafone", "vodafone", r"vodafone", None, True),
    ("Orange", "orange", r"\borange\b", r"orange\s*pi", False),
    ("Endesa", "endesa", r"endesa", None, False),
    ("Ebury", "ebury", r"ebury", None, False),
    ("OVH", "ovh", r"\bovh\b", None, False),
    ("Legalitas", "legalitas", r"legalitas", None, False),
    ("Revolut", "revolut", r"revolut", r"paypal2revolut", False),
    ("Kiwa", "kiwa", r"kiwa", None, False),
    ("Martransit", "martransit", r"martransit", None, False),
    ("Transportes Nieves", "transportes-nieves", r"transportes\s*nieves", None, False),
    ("AEAT", "aeat", r"\baeat\b|agencia\s+tributaria", None, True),
]


def doc_slug(title, ntok):
    parts = [p for p in slugify_obsidian_name(title).split("-")
             if p and p not in ntok and p not in STOPWORDS]
    return "-".join(parts) or "documento"


def main():
    osint_docs = [n for n in list_obsidian_notes(OSINT)
                  if "/personas/" in n or "/organizaciones/" in n]
    notas_rest = [n for n in list_obsidian_notes(NOTAS)
                  if "/.git/" not in n and "/dist/" not in n]
    att_del, docs_del, report = set(), set(), []

    for display, slug, rx, excl, move_loose in EXT:
        rxc = re.compile(rx, re.I)
        exclc = re.compile(excl, re.I) if excl else None
        ntok = set(slug.split("-")) | {"documentos", "documento"}

        # 1. cross-refs en osint (no mover)
        xrefs = []
        for n in osint_docs:
            t = os.path.basename(n)[:-3]
            if rxc.search(t) and not (exclc and exclc.search(t)):
                rel = os.path.relpath(n, OSINT)[:-3]  # sin .md
                dueno = rel.split("/")[1] if rel.startswith(("personas/", "organizaciones/")) else ""
                xrefs.append((rel, t, dueno))

        # 2. notas sueltas reales (solo si move_loose)
        moved = []
        if move_loose:
            att_rel = f"attachments/organizaciones/{slug}"
            att_abs = f"{OSINT}/{att_rel}"
            os.makedirs(att_abs, exist_ok=True)
            for n in notas_rest:
                t = os.path.basename(n)[:-3]
                if not rxc.search(t) or (exclc and exclc.search(t)):
                    continue
                if slugify_obsidian_name(t) == slug:  # nota homonima = cuerpo de ficha, se trata aparte
                    continue
                dn = read_obsidian_note(n)
                ds = doc_slug(t, ntok)
                base, k = ds, 2
                while os.path.exists(f"{OSINT}/organizaciones/{slug}/{ds}.md"):
                    ds = f"{base}-{k}"; k += 1
                new = []
                for i, emb in enumerate(extract_obsidian_embeds(dn["body"]), 1):
                    ap = resolve_obsidian_embed(NOTAS, emb)
                    if not ap:
                        new.append(f"<!-- no encontrado: {emb} -->"); continue
                    ext = os.path.splitext(ap)[1].lower()
                    nn = f"{ds}-{i}{ext}"
                    shutil.copy2(ap, f"{att_abs}/{nn}"); att_del.add(ap)
                    new.append(f"![[{att_rel}/{nn}]]")
                text = re.sub(r'!\[\[[^\]]+\]\]', '', dn["body"]).strip()
                parts = [x for x in [text, "\n".join(new)] if x]
                create_obsidian_note(OSINT, f"organizaciones/{slug}/{ds}",
                                     body="\n\n".join(parts) if parts else "(sin contenido)",
                                     frontmatter={"tipo": "documento", "entidad": f"[[{slug}]]",
                                                  "fuente": "NotasDeObsidian/" + os.path.relpath(n, NOTAS)},
                                     overwrite=True)
                moved.append((t, ds)); docs_del.add(n)

        # 3. ficha-indice (omitir si no hay nada que referenciar)
        if not xrefs and not moved:
            continue
        bl = [f"Ficha de la organizacion externa {display}. Referencia de todo lo trabajado con ellos.", ""]
        if xrefs:
            bl += ["## Relacionado", ""]
            for rel, t, dueno in xrefs:
                bl.append(f"- [[{rel}|{t}]]" + (f"  ({dueno})" if dueno else ""))
        if moved:
            bl += ["", "## Documentos", ""] + [f"- [[organizaciones/{slug}/{ds}|{t}]]" for t, ds in moved]
        bl += ["", "## Notas", ""]
        create_obsidian_note(OSINT, f"organizaciones/{slug}", body="\n".join(bl),
                             frontmatter={"tipo": "organizacion", "nombre": display, "slug": slug,
                                          "externa": True, "tags": ["organizacion", "externa", "osint"]},
                             overwrite=True)
        report.append((display, slug, len(xrefs), len(moved)))

    for p in docs_del:
        try: os.remove(p)
        except FileNotFoundError: pass
    moved_att = 0
    for ap in att_del:
        try:
            os.remove(ap); moved_att += 1
        except FileNotFoundError:
            pass

    print(f"fichas externas creadas: {len(report)} | sueltas movidas: {len(docs_del)} | attachments: {moved_att}\n")
    for d, s, x, m in report:
        print(f"  {d:20} -> organizaciones/{s}  | xref={x} sueltas_movidas={m}")


if __name__ == "__main__":
    main()