import os
import glob
import re

def sanitize_code_blocks(markdown_content):
    """Tenta inferir a linguagem de blocos de código sem linguagem definida."""
    def replace_code(match):
        lang = match.group(1).strip()
        code = match.group(2)
        
        # Lógica de inferência para Fluig
        if not lang or lang == "java" or lang == "javascript": 
            if any(x in code for x in ["DatasetBuilder", "createDataset", "displayFields", "getSelectedZoomItem", "parent_child"]):
                lang = "javascript"
            elif any(x in code for x in ["PreparedStatement", "ResultSet", "DriverManager", "import java."]):
                lang = "java"
            elif "SELECT" in code.upper() and "FROM" in code.upper():
                lang = "sql"
        
        return f"```{lang}\n{code}\n```"

    pattern = re.compile(r"```(.*?)\n(.*?)\n```", re.DOTALL)
    return pattern.sub(replace_code, markdown_content)

def reprocess_all():
    docs_dir = "fluig_rag_docs"
    files = glob.glob(os.path.join(docs_dir, "**", "*.md"), recursive=True)
    
    changed_count = 0
    print(f"Reprocessando {len(files)} arquivos...")
    
    for file_path in files:
        if "Biblioteca de Snippets" in file_path: continue
        
        with open(file_path, "r", encoding="utf-8") as f:
            content = f.read()
        
        original_content = content
        
        # 1. Sanitização de blocos de código
        content = sanitize_code_blocks(content)
        
        # 2. Correção de Admonitions residuais (> !!! -> !!!)
        content = content.replace("> !!!", "!!!")
        
        if content != original_content:
            with open(file_path, "w", encoding="utf-8") as f:
                f.write(content)
            changed_count += 1
            # print(f"Atualizado: {file_path}")

    print(f"\nFinalizado! {changed_count} arquivos foram atualizados com melhorias semânticas.")

if __name__ == "__main__":
    reprocess_all()