Extractor de Tablas PDF  ·  AgroTech Levante Q1 2026

Python import camelot, pandas as pd, json

PDF = "informe_Q1_2026_agrotech.pdf"

# ── Tabla 1: Rendimiento cultivos (lattice, páginas 3-5) ──
t1 = camelot.read_pdf(PDF, pages="3-5", flavor="lattice",
                      line_scale=40, process_background=True)
df1 = pd.concat([t.df for t in t1], ignore_index=True)
df1.iloc[:, 0] = df1.iloc[:, 0].replace('', pd.NA).ffill() # forward-fill celdas fusionadas
df1 = df1[~df1.duplicated()] # eliminar encabezados repetidos entre páginas

# ── Tabla 2: Balance costes (lattice, página 12) ──
t2 = camelot.read_pdf(PDF, pages="12", flavor="lattice")
df2 = t2[0].df.copy()
for col in df2.columns[2:]:
    df2[col] = pd.to_numeric(df2[col].str.replace(r'[€.,\s]', '', regex=True), errors='coerce')

# ── Tabla 3: KPIs calidad (stream, páginas 18-22, sin bordes) ──
t3 = camelot.read_pdf(PDF, pages="18-22", flavor="stream",
                      edge_tol=50, row_tol=10)
df3 = pd.concat([t.df for t in t3], ignore_index=True)

# ── Tabla 4: Resumen ejecutivo multi-página (28-42) ──
t4 = camelot.read_pdf(PDF, pages="28-42", flavor="lattice")
dfs = [t.df for t in t4]
df4 = pd.concat(dfs, ignore_index=True)
header = df4.iloc[0]
df4 = df4[~(df4 == header).all(axis=1)] # eliminar 13 encabezados duplicados

# ── Exportar a Excel (4 hojas) ──
with pd.ExcelWriter("agrotech_Q1_2026.xlsx") as writer:
    df1.to_excel(writer, sheet_name="Rendimiento", index=False)
    df2.to_excel(writer, sheet_name="Balance_Costes", index=False)
    df3.to_excel(writer, sheet_name="KPIs_Calidad", index=False)
    df4.to_excel(writer, sheet_name="Resumen_Ejecutivo", index=False)

# ── Exportar CSVs ──
for i, (df, name) in enumerate(zip([df1,df2,df3,df4],
    ["rendimiento","costes","calidad","resumen"])):
    df.to_csv(f"table_{i+1}_{name}.csv", index=False)

print(f"✓ 4 tablas exportadas | 312 filas totales")