scripts/translate_post.py

@@ -176,7 +176,7 @@ def _system_prompt(lang: str) -> str:
     )
 
 
-def translate_text(text: str, lang: str, *, is_title: bool = False) -> str:
+def translate_text(text: str, lang: str, *, is_title: bool = False, context: str = "") -> str:
     text = text.strip()
     if not text:
         return ""
@@ -185,6 +185,13 @@ def translate_text(text: str, lang: str, *, is_title: bool = False) -> str:
         return out
     user = f"<<<INI>>>{text}<<<FIN>>>"
     if is_title:
+        context_block = ""
+        if context:
+            context_block = (
+                "\nBODY CONTEXT (for meaning only; do not translate or include it):\n"
+                f"<<<CONTEXT>>>{context}<<<ENDCONTEXT>>>\n"
+            )
+        user = context_block + user
         kind = "el TÍTULO"
         task = (
             f"Traduce {kind} que va entre las marcas.\n"
@@ -192,6 +199,7 @@ def translate_text(text: str, lang: str, *, is_title: bool = False) -> str:
             f"No lo dejes en inglés salvo que el original ya sea un nombre propio o una marca.\n"
             f"Responde solo con el título traducido entre las marcas:\n{user}"
         )
+        task += "\nSi es un modismo, interpreta su sentido segun el contexto, no lo calques."
     else:
         kind = "el texto"
         task = f"Traduce {kind} que va entre las marcas:\n{user}"
@@ -226,6 +234,13 @@ def translate_html(html: str, lang: str) -> str:
 
 
 # ── Datos / estado ───────────────────────────────────────────────────────────
+def title_context(html: str) -> str:
+    """First paragraph gives title idioms enough meaning without bloating the prompt."""
+    first = re.search(r"<p\b[^>]*>.*?</p>", html, flags=re.IGNORECASE | re.DOTALL)
+    context = first.group(0) if first else html
+    return re.sub(r"\s+", " ", context).strip()[:2000]
+
+
 def read_post(post_id: int) -> dict:
     return json.loads(php_helper("read", str(post_id)))
 
@@ -290,8 +305,8 @@ def process_post(post_id: int, langs: list[str], status: str, force: bool, state
             log(f"  {lang}: --force, eliminada traducción previa #{existing}")
         try:
             t0 = time.time()
-            title = translate_text(src["title"], lang, is_title=True)
             content = translate_html(src["content"], lang)
+            title = translate_text(src["title"], lang, is_title=True, context=title_context(src["content"]))
             new_id = create_translation(post_id, lang, title, content, status)
             dt = time.time() - t0
             log(f"  {lang}: creado #{new_id} ({dt:.0f}s) → «{title[:50]}»")