phase 0

skill.src/modules/translator.py

@@ -0,0 +1,107 @@
+"""
+translator.py — Chinese to English part name translation via Claude API.
+
+Detects CJK unicode range. Batches all names in a single API call per file.
+Flags uncertain translations in the notes column.
+"""
+
+import json
+import logging
+import os
+import re
+
+import pandas as pd
+
+logger = logging.getLogger("step_processor.translator")
+
+CJK_PATTERN = re.compile(r'[一-鿿㐀-䶿]')
+
+SYSTEM_PROMPT = (
+    "You are a mechanical engineering translator specializing in Chinese "
+    "manufacturing CAD files for display and enclosure products. "
+    "Translate the following part names from Chinese to English. "
+    "Preserve technical precision. Use standard hardware/manufacturing terminology. "
+    "Output ONLY a JSON object mapping original Chinese to translated English, nothing else.\n"
+    'Example: {"安装支架": "Mounting Bracket", "螺钉M4": "M4 Screw", "前面板": "Front Panel"}'
+)
+
+
+def has_chinese(text: str) -> bool:
+    """Return True if text contains CJK characters."""
+    return bool(CJK_PATTERN.search(str(text)))
+
+
+def translate_bom(df: pd.DataFrame, model_name: str = "") -> pd.DataFrame:
+    """Detect Chinese part names and translate via Claude API."""
+    needs_translation = df["part_name_original"].apply(has_chinese)
+    chinese_names = df.loc[needs_translation, "part_name_original"].unique().tolist()
+    if not chinese_names:
+        logger.info("No Chinese part names detected — translation skipped")
+        return df
+    logger.info(f"Translating {len(chinese_names)} Chinese part names...")
+    translation_map = _call_claude_api(chinese_names, model_name)
+    if not translation_map:
+        logger.warning("Translation API returned no results — retaining original names")
+        df.loc[needs_translation, "notes"] = (
+            df.loc[needs_translation, "notes"].apply(
+                lambda n: (n + "; " if n else "") + "translation-failed"))
+        return df
+    for idx, row in df.iterrows():
+        original = row["part_name_original"]
+        if has_chinese(original):
+            translated = translation_map.get(original)
+            if translated:
+                df.at[idx, "part_name_english"] = translated
+                note_tag = "ambiguous-translation" if "[?]" in translated else "machine-translated"
+            else:
+                df.at[idx, "part_name_english"] = original
+                note_tag = "translation-missing"
+            existing = row["notes"]
+            df.at[idx, "notes"] = (existing + "; " if existing else "") + note_tag
+    logger.info(f"Translated {needs_translation.sum()} parts")
+    return df
+
+
+def get_translation_map(df: pd.DataFrame) -> dict:
+    """Return dict of original → english for all translated rows."""
+    mask = df["part_name_original"] != df["part_name_english"]
+    return dict(zip(df.loc[mask, "part_name_original"],
+                    df.loc[mask, "part_name_english"]))
+
+
+def _call_claude_api(names: list, model_name: str = "") -> dict:
+    """Single batched Claude API call. Returns original→translated dict."""
+    api_key = os.environ.get("ANTHROPIC_API_KEY")
+    if not api_key:
+        logger.error("ANTHROPIC_API_KEY not set — translation unavailable")
+        return {}
+    try:
+        import anthropic
+    except ImportError:
+        logger.error("anthropic package not installed — pip install anthropic")
+        return {}
+    names_json = json.dumps(names, ensure_ascii=False)
+    user_msg = f"Translate these part names from Chinese to English:\n{names_json}"
+    if model_name:
+        user_msg += f"\n\nContext: Parts from a {model_name} display enclosure assembly."
+    try:
+        client = anthropic.Anthropic(api_key=api_key)
+        response = client.messages.create(
+            model="claude-haiku-4-5-20251001",
+            max_tokens=2048,
+            system=SYSTEM_PROMPT,
+            messages=[{"role": "user", "content": user_msg}],
+        )
+        text = response.content[0].text.strip()
+        json_match = re.search(r'\{.*\}', text, re.DOTALL)
+        if json_match:
+            text = json_match.group(0)
+        result = json.loads(text)
+        logger.info(f"API returned {len(result)} translations")
+        return result
+    except json.JSONDecodeError as e:
+        logger.error(f"Translation API JSON parse error: {e}")
+        return {}
+    except Exception as e:
+        logger.error(f"Translation API error: {type(e).__name__}: {e}")
+        return {}