Update app.py

app.py

@@ -11,7 +11,7 @@
 # en_core_web_sm @ https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.7.1/en_core_web_sm-3.7.1-py3-none-any.whl
 
 import os, re, json, base64, zlib
-from typing import Dict, Optional, List, Any  # <- FIX: List/Any importados
+from typing import Dict, Optional, List, Any
 import gradio as gr
 
 # ------------ Archivos esperados ------------
@@ -370,7 +370,7 @@ def _es_conj(lemma, tense, person):
         return tab[tense].get(person, tab[tense]["3s"])
     if lemma=="estar":
         tab={"Pres":{"1s":"estoy","2s":"estás","3s":"está","1p":"estamos","2p":"estáis","3p":"están"},
-             "Past":{"1s":"estuve","2s":"estuviste","3s":"estuvo","1p":"estuvimos","2p":"estuvisteis","3p":"estuvieron"},
+             "Past":{"1s":"estuviste","2s":"estuviste","3s":"estuvo","1p":"estuvimos","2p":"estuvisteis","3p":"estuvieron"},
              "Fut":{"1s":"estaré","2s":"estarás","3s":"estará","1p":"estaremos","2p":"estaréis","3p":"estarán"}}
         return tab[tense].get(person, tab[tense]["3s"])
     if lemma=="ir":
@@ -406,6 +406,70 @@ def _en_conj(lemma, tense, person):
         return lemma+"ed"
     return lemma
 
+# ================= Helper de construcción/translate =================
+def _build_with_spacy(text: str, src_lang: str, target: str,
+                      drop_articles: bool, zero_copula: bool,
+                      semi_lossless: bool, remove_pronouns: bool) -> str:
+    nlp = nlp_es if src_lang=="Español" else nlp_en
+    doc = nlp(text)
+    if target == "Minimax-ASCII":
+        return realize_minimax(doc, src_lang, drop_articles, zero_copula, semi_lossless, remove_pronouns=remove_pronouns)
+    else:
+        return realize_komin(doc, src_lang, drop_articles, zero_copula, semi_lossless, remove_pronouns=remove_pronouns)
+
+def build_sentence(text: str, src_lang: str, target: str,
+                   drop_articles: bool, zero_copula: bool,
+                   mode: str, max_comp_exact: bool = False, remove_pronouns: bool = False) -> str:
+    if not text.strip(): return ""
+    semi = True  # siempre semi-lossless en construcción
+    core = _build_with_spacy(text, src_lang, target, drop_articles, zero_copula and not semi, semi, remove_pronouns) if USE_SPACY else encode_simple(text, src_lang, target)
+    if max_comp_exact:
+        return custom_sidecar_enc(core, text)
+    return core
+
+def universal_translate(text: str, src: str, tgt: str,
+                        drop_articles: bool, zero_copula: bool,
+                        mode: str, max_comp_exact: bool = False, remove_pronouns: bool = False) -> str:
+    if not text.strip(): return ""
+    if src == tgt: return text
+
+    # Natural → Conlang
+    if src in ("Español","English") and tgt in ("Minimax-ASCII","Kōmín-CJK"):
+        return build_sentence(text, src, tgt, drop_articles, zero_copula, mode, max_comp_exact, remove_pronouns)
+
+    # Conlang → Natural (considera sidecars)
+    if src in ("Minimax-ASCII","Kōmín-CJK") and tgt in ("Español","English"):
+        orig = extract_custom_sidecar(text)
+        if orig is not None: return orig
+        orig = extract_sidecar_b85(text)
+        if orig is not None: return orig
+        return decode_simple(strip_custom_sidecar(strip_sidecar_b85(text)), src, tgt)
+
+    # Natural ↔ Natural (lemas muy simples)
+    if src in ("Español","English") and tgt in ("Español","English"):
+        return text
+
+    # Conlang ↔ Conlang (simple)
+    if src in ("Minimax-ASCII","Kōmín-CJK") and tgt in ("Minimax-ASCII","Kōmín-CJK"):
+        core = strip_custom_sidecar(text)
+        es_lemmas = decode_simple(core, src, "Español")
+        words = re.findall(r"\w+|[^\w\s]+", es_lemmas)
+        out=[]
+        for w in words:
+            if re.fullmatch(r"\w+", w):
+                code = ES2MINI.get(norm_es(w)) if tgt=="Minimax-ASCII" else ES2KOMI.get(norm_es(w))
+                if not code:
+                    code = enc_oov_minimax(w) if tgt=="Minimax-ASCII" else enc_oov_komin(w)
+                out.append(code)
+            else:
+                out.append(w)
+        out_text = " ".join(out)
+        if extract_custom_sidecar(text) is not None:
+            return custom_sidecar_enc(out_text, extract_custom_sidecar(text) or "")
+        return out_text
+
+    return "[No soportado]"
+
 # =====================================================================================
 # ========================= UI bilingüe y explicaciones claras ========================
 # =====================================================================================
@@ -466,7 +530,7 @@ EXPLAIN_TAB_ROUNDTRIP_EN = "Runs **(ES/EN→Conlang)→(Conlang→ES/EN)** to ve
 EXPLAIN_CHECKBOX_EN = "Drop articles ~10–15%, Zero copula ~5–10% extra, Remove pronouns variable, Max Exact 40–60% for >100 chars."
 LEXICON_FRIENDLY_EN = "We use **WordNet (OMW)**, pair ES words with EN, clean & sort by frequency, assign short **codes** (Minimax/Kōmín), and save three JSONs so the app can encode/decode compactly."
 
-# >>> NUEVO <<< ——— Explicación directa de los “botones / modos” (ES/EN)
+# >>> Explicación directa de los “botones / modos”
 EXPLAIN_MODES_ES = """
 **¿Qué hace cada botón / modo?** *(Marca uno en “🧭 Modo de uso”)*
 
@@ -632,18 +696,16 @@ def make_panel_roundtrip(lang="ES"):
             c.change(run, [text, src, tgt, mode_hidden, exact], [out1, out2])
     return g
 
-# ---------- Página (ES/EN), con “modos” como CHECKBOX (mutuamente excluyentes) ----------
+# ---------- Página (ES/EN), modos como CheckboxGroup (selección única) ----------
 with gr.Blocks(title="Universal Conlang Translator", theme=gr.themes.Soft()) as demo:
     gr.Markdown("## 🌍 Idioma / Language")
     lang = gr.Radio(["ES","EN"], value="ES", label="Selecciona / Select")
 
-    # >>> NUEVO <<< — Acordeones con explicación de los botones/modos (mismo nivel)
+    # Acordeones “¿qué hace cada botón?” (ES/EN)
     acc_modes_es = gr.Accordion("📖 ¿Qué hace cada botón / modo? (ES)", open=False, visible=True)
-    with acc_modes_es:
-        gr.Markdown(EXPLAIN_MODES_ES)
+    with acc_modes_es: gr.Markdown(EXPLAIN_MODES_ES)
     acc_modes_en = gr.Accordion("📖 What does each button / mode do? (EN)", open=False, visible=False)
-    with acc_modes_en:
-        gr.Markdown(EXPLAIN_MODES_EN)
+    with acc_modes_en: gr.Markdown(EXPLAIN_MODES_EN)
 
     # Acordeones intro + léxico (mismo nivel)
     acc_intro_es = gr.Accordion("☑️ Opciones y compactación — guía rápida (ES)", open=False, visible=True)
@@ -665,14 +727,20 @@ with gr.Blocks(title="Universal Conlang Translator", theme=gr.themes.Soft()) as
         table_en = gr.Dataframe(headers=["lemma_es","lemma_en","minimax","komin"], row_count=1, interactive=False)
         gr.Button("Refresh").click(lambda n: master_preview(int(n)), [n_rows_en], [table_en])
 
-    # “Modos” como CHECKBOX (mutuamente excluyentes)
+    # Modo de uso (CheckboxGroup selección única)
     gr.Markdown("### 🧭 Modo de uso (marca **uno**)")
-    row_modes = gr.Row()
-    with row_modes:
-        cb_tr = gr.Checkbox(True,  label="🔁 Traducir / Translate")
-        cb_bu = gr.Checkbox(False, label="🛠️ Construir (ES/EN → Conlang) / Build")
-        cb_de = gr.Checkbox(False, label="🗝️ Decodificar (Conlang → ES/EN) / Decode")
-        cb_rt = gr.Checkbox(False, label="🔄 Prueba ida→vuelta / Round-trip")
+    cb_modes = gr.CheckboxGroup(
+        choices=[
+            "🔁 Traducir / Translate",
+            "🛠️ Construir (ES/EN → Conlang) / Build",
+            "🗝️ Decodificar (Conlang → ES/EN) / Decode",
+            "🔄 Prueba ida→vuelta / Round-trip",
+        ],
+        value=["🔁 Traducir / Translate"],
+        label=None,
+    )
+    # Estados para mantener la API existente (tr, bu, de, rt)
+    cb_tr = gr.State(True); cb_bu = gr.State(False); cb_de = gr.State(False); cb_rt = gr.State(False)
 
     # Paneles por modo y por idioma
     gr.Markdown("### 🧪 Área de trabajo")
@@ -683,9 +751,19 @@ with gr.Blocks(title="Universal Conlang Translator", theme=gr.themes.Soft()) as
 
     def _vis(yes): return gr.update(visible=bool(yes))
 
-    # Enforce: sólo 1 checkbox activo + visibilidad de paneles/accordions por idioma y modo
+    def _modes_to_flags(selected_list):
+        sel = selected_list or []
+        order = [
+            "🔁 Traducir / Translate",
+            "🛠️ Construir (ES/EN → Conlang) / Build",
+            "🗝️ Decodificar (Conlang → ES/EN) / Decode",
+            "🔄 Prueba ida→vuelta / Round-trip",
+        ]
+        chosen = next((o for o in order if o in sel), order[0])
+        return [chosen == o for o in order], [chosen]
+
+    # No tocamos tu lógica de visibilidad: sólo la usamos
     def switch_everything(lang_code, tr, bu, de, rt):
-        # forzar exclusividad (si varias marcadas, prioriza la primera en orden tr>bu>de>rt)
         tr2, bu2, de2, rt2 = False, False, False, False
         if tr or (not bu and not de and not rt): tr2 = True
         elif bu: bu2 = True
@@ -693,49 +771,49 @@ with gr.Blocks(title="Universal Conlang Translator", theme=gr.themes.Soft()) as
         else:    rt2 = True
 
         is_en = (lang_code == "EN")
-        # Accordions ES/EN
         vis_es = not is_en; vis_en = is_en
         updates = [
-            _vis(vis_es), _vis(vis_en),  # >>> NUEVO <<< acc_modes_es, acc_modes_en
-            _vis(vis_es), _vis(vis_en),  # intro accordions
-            _vis(vis_es), _vis(vis_en),  # lexicon accordions
+            _vis(vis_es), _vis(vis_en),  # modos ES/EN
+            _vis(vis_es), _vis(vis_en),  # intro
+            _vis(vis_es), _vis(vis_en),  # léxico
         ]
-        # Panels ES
         updates += [
-            _vis(vis_es and tr2), _vis(vis_es and bu2), _vis(vis_es and de2), _vis(vis_es and rt2)
+            _vis(vis_es and tr2), _vis(vis_es and bu2), _vis(vis_es and de2), _vis(vis_es and rt2),
+            _vis(vis_en and tr2), _vis(vis_en and bu2), _vis(vis_en and de2), _vis(vis_en and rt2),
         ]
-        # Panels EN
-        updates += [
-            _vis(vis_en and tr2), _vis(vis_en and bu2), _vis(vis_en and de2), _vis(vis_en and rt2)
-        ]
-        # Checkbox state (exclusivo)
         updates += [tr2, bu2, de2, rt2]
         return updates
 
+    def _lang_or_modes(lang_code, selected_list, tr, bu, de, rt):
+        flags, normalized = _modes_to_flags(selected_list)
+        updates = switch_everything(lang_code, *flags)
+        return updates + flags + [normalized]
+
+    # Reacciones
     lang.change(
-        switch_everything,
-        [lang, cb_tr, cb_bu, cb_de, cb_rt],
+        _lang_or_modes,
+        [lang, cb_modes, cb_tr, cb_bu, cb_de, cb_rt],
         [
-            acc_modes_es, acc_modes_en,             # >>> NUEVO <<<
+            acc_modes_es, acc_modes_en,
             acc_intro_es, acc_intro_en, acc_lex_es, acc_lex_en,
             panel_tr_es, panel_bu_es, panel_de_es, panel_rt_es,
             panel_tr_en, panel_bu_en, panel_de_en, panel_rt_en,
-            cb_tr, cb_bu, cb_de, cb_rt
+            cb_tr, cb_bu, cb_de, cb_rt,
+            cb_modes,
+        ],
+    )
+    cb_modes.change(
+        _lang_or_modes,
+        [lang, cb_modes, cb_tr, cb_bu, cb_de, cb_rt],
+        [
+            acc_modes_es, acc_modes_en,
+            acc_intro_es, acc_intro_en, acc_lex_es, acc_lex_en,
+            panel_tr_es, panel_bu_es, panel_de_es, panel_rt_es,
+            panel_tr_en, panel_bu_en, panel_de_en, panel_rt_en,
+            cb_tr, cb_bu, cb_de, cb_rt,
+            cb_modes,
         ],
     )
-
-    for box in (cb_tr, cb_bu, cb_de, cb_rt):
-        box.change(
-            switch_everything,
-            [lang, cb_tr, cb_bu, cb_de, cb_rt],
-            [
-                acc_modes_es, acc_modes_en,         # >>> NUEVO <<<
-                acc_intro_es, acc_intro_en, acc_lex_es, acc_lex_en,
-                panel_tr_es, panel_bu_es, panel_de_es, panel_rt_es,
-                panel_tr_en, panel_bu_en, panel_de_en, panel_rt_en,
-                cb_tr, cb_bu, cb_de, cb_rt
-            ],
-        )
 
 if __name__ == "__main__":
     demo.launch()