Fix build: Bundle schwab_scraper source and use local dependencies

schwab_scraper/storage/cache.py

@@ -0,0 +1,74 @@
+import os
+from typing import Optional
+
+CACHE_DIR = "data/morningstar_pdfs"
+TRANSACTION_CACHE_DIR = "data/transaction_csvs"
+
+
+def ensure_cache_dir() -> str:
+    os.makedirs(CACHE_DIR, exist_ok=True)
+    return CACHE_DIR
+
+
+def ensure_transaction_cache_dir() -> str:
+    os.makedirs(TRANSACTION_CACHE_DIR, exist_ok=True)
+    return TRANSACTION_CACHE_DIR
+
+
+def cache_filename(ticker: str, formatted_date: str) -> str:
+    ensure_cache_dir()
+    # Sanitize date string to remove slashes that would create subdirectories
+    safe_date = formatted_date.replace('/', '_').replace('\\', '_')
+    return os.path.join(CACHE_DIR, f"{ticker.upper()}_{safe_date}.pdf")
+
+
+def transaction_cache_filename(account_label: str, timestamp_str: str) -> str:
+    """Return a path like data/transaction_csvs/<account_label>/<account_label>_Transactions_<timestamp>.csv
+
+    account_label examples: "Joint_XXX604", "IRA_XXX873". Timestamp is usually YYYYMMDD-HHMMSS.
+    """
+    ensure_transaction_cache_dir()
+    safe_label = account_label.replace("/", "_")
+    account_dir = os.path.join(TRANSACTION_CACHE_DIR, safe_label)
+    os.makedirs(account_dir, exist_ok=True)
+    return os.path.join(account_dir, f"{safe_label}_Transactions_{timestamp_str}.csv")
+
+
+def read_cached_pdf(ticker: str) -> Optional[bytes]:
+    ensure_cache_dir()
+    files = [f for f in os.listdir(CACHE_DIR) if f.startswith(ticker.upper()) and f.endswith(".pdf")]
+    if not files:
+        return None
+    with open(os.path.join(CACHE_DIR, files[0]), "rb") as f:
+        return f.read()
+
+
+def read_cached_transaction_csv(account_label: str) -> Optional[bytes]:
+    """Return latest cached CSV bytes for an account label, if any."""
+    ensure_transaction_cache_dir()
+    safe_label = account_label.replace("/", "_")
+    account_dir = os.path.join(TRANSACTION_CACHE_DIR, safe_label)
+    if not os.path.isdir(account_dir):
+        return None
+    files = [f for f in os.listdir(account_dir) if f.endswith('.csv')]
+    if not files:
+        return None
+    # Pick most recent by name (timestamp in filename)
+    files.sort(reverse=True)
+    with open(os.path.join(account_dir, files[0]), 'rb') as f:
+        return f.read()
+
+
+def write_cached_pdf(ticker: str, formatted_date: str, pdf_bytes: bytes) -> str:
+    ensure_cache_dir()
+    path = cache_filename(ticker, formatted_date)
+    with open(path, "wb") as f:
+        f.write(pdf_bytes)
+    return path
+
+
+def write_cached_transaction_csv(account_label: str, timestamp_str: str, csv_bytes: bytes) -> str:
+    path = transaction_cache_filename(account_label, timestamp_str)
+    with open(path, 'wb') as f:
+        f.write(csv_bytes)
+    return path