pdf_processing.py

import re

import pdfplumber


def _clean_text(text: str) -> str:
    if not text:
        return ""
    text = re.sub(r'\n\s*\n\s*\n+', '\n\n', text)
    lines = [line.rstrip() for line in text.split('\n')]
    return '\n'.join(lines).strip()


def _table_to_markdown(table: list) -> str:
    if not table:
        return ""
    cleaned = [[str(c).strip() if c else "" for c in row] for row in table]
    num_cols = len(cleaned[0])
    col_widths = [0] * num_cols
    for row in cleaned:
        for i, cell in enumerate(row[:num_cols]):
            col_widths[i] = max(col_widths[i], len(cell))
    lines = []
    header = cleaned[0]
    lines.append("| " + " | ".join(c.ljust(col_widths[i]) for i, c in enumerate(header[:num_cols])) + " |")
    lines.append("| " + " | ".join("-" * w for w in col_widths) + " |")
    for row in cleaned[1:]:
        lines.append("| " + " | ".join(c.ljust(col_widths[i]) for i, c in enumerate(row[:num_cols])) + " |")
    return "\n".join(lines)


def pdf_to_markdown(pdf_path: str) -> str:
    parts = []
    with pdfplumber.open(pdf_path) as pdf:
        total = len(pdf.pages)
        for n, page in enumerate(pdf.pages, 1):
            parts.append(f"\n---\n## Trang {n}/{total}\n")
            tables = page.extract_tables()
            text = page.extract_text()
            if text:
                parts.append(_clean_text(text))
                parts.append("\n")
            if tables:
                for i, tbl in enumerate(tables, 1):
                    if tbl:
                        parts.append(f"\n**Bảng {i}:**\n")
                        parts.append(_table_to_markdown(tbl))
                        parts.append("\n")
    return _clean_text("\n".join(parts))


def format_chat_history(messages: list[dict]) -> str:
    if not messages:
        return "(Không có lịch sử trò chuyện)"
    lines = []
    for m in messages:
        sender  = m.get("senderName") or m.get("sender_id") or "?"
        content = m.get("content") or m.get("message") or ""
        ts      = m.get("timestamp") or m.get("created_at") or ""
        prefix  = f"[{ts}] " if ts else ""
        lines.append(f"{prefix}{sender}: {content}")
    return "\n".join(lines)