Spaces:
Sleeping
Sleeping
Update app.py
Browse filesfixed open md error
app.py
CHANGED
|
@@ -25,26 +25,26 @@ def procesar_pdf(pdf_file=None, url_pdf=None, formatos=[], verbose="auto"):
|
|
| 25 |
os.makedirs(output_base, exist_ok=True)
|
| 26 |
start_time = time.time()
|
| 27 |
|
| 28 |
-
|
|
|
|
| 29 |
try:
|
| 30 |
subprocess.run(["marker_single", "--help"], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
| 31 |
except FileNotFoundError:
|
| 32 |
-
yield "
|
| 33 |
return
|
| 34 |
|
| 35 |
if not formatos:
|
| 36 |
-
yield "
|
| 37 |
return
|
| 38 |
|
| 39 |
try:
|
| 40 |
-
# Obtener el archivo
|
| 41 |
if url_pdf:
|
| 42 |
parsed = urlparse(url_pdf)
|
| 43 |
nombre_pdf = parsed.path.split("/")[-1] or "documento"
|
| 44 |
ruta_pdf = obtener_ruta_segura(output_base, nombre_pdf)
|
| 45 |
response = requests.get(url_pdf)
|
| 46 |
if response.status_code != 200:
|
| 47 |
-
yield "
|
| 48 |
return
|
| 49 |
with open(ruta_pdf, "wb") as f:
|
| 50 |
f.write(response.content)
|
|
@@ -53,17 +53,16 @@ def procesar_pdf(pdf_file=None, url_pdf=None, formatos=[], verbose="auto"):
|
|
| 53 |
ruta_pdf = obtener_ruta_segura(output_base, nombre_pdf)
|
| 54 |
shutil.copyfile(pdf_file.name, ruta_pdf)
|
| 55 |
else:
|
| 56 |
-
yield "
|
| 57 |
return
|
| 58 |
|
| 59 |
-
# Verificar cabecera PDF
|
| 60 |
with open(ruta_pdf, "rb") as f:
|
| 61 |
if f.read(4) != b"%PDF":
|
| 62 |
-
yield "
|
| 63 |
return
|
| 64 |
|
| 65 |
if os.path.getsize(ruta_pdf) > 3 * 1024 * 1024:
|
| 66 |
-
|
| 67 |
|
| 68 |
nombre_sin_ext = os.path.splitext(os.path.basename(ruta_pdf))[0]
|
| 69 |
carpeta_salida = os.path.join(output_base, nombre_sin_ext)
|
|
@@ -71,9 +70,11 @@ def procesar_pdf(pdf_file=None, url_pdf=None, formatos=[], verbose="auto"):
|
|
| 71 |
shutil.rmtree(carpeta_salida)
|
| 72 |
os.makedirs(carpeta_salida, exist_ok=True)
|
| 73 |
|
| 74 |
-
# Procesar cada formato
|
| 75 |
for fmt in formatos:
|
| 76 |
-
|
|
|
|
|
|
|
|
|
|
| 77 |
if fmt == "md":
|
| 78 |
cmd = ["marker_single", ruta_pdf, "--output_format", "markdown", "--disable_ocr", "--output_dir", carpeta_salida]
|
| 79 |
elif fmt == "md + ocr":
|
|
@@ -81,54 +82,56 @@ def procesar_pdf(pdf_file=None, url_pdf=None, formatos=[], verbose="auto"):
|
|
| 81 |
else:
|
| 82 |
cmd = ["marker_single", ruta_pdf, "--output_format", fmt, "--output_dir", carpeta_salida]
|
| 83 |
|
| 84 |
-
yield f"⏳ Procesando formato '{fmt}'... (tiempo: {tiempo_parcial} s)", None, None, None
|
| 85 |
-
|
| 86 |
for intento in range(1, 4):
|
| 87 |
try:
|
| 88 |
result = subprocess.run(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True, timeout=900)
|
| 89 |
if result.returncode == 0:
|
| 90 |
break
|
| 91 |
elif intento == 3:
|
| 92 |
-
|
|
|
|
| 93 |
return
|
| 94 |
time.sleep(2)
|
| 95 |
except Exception as e:
|
| 96 |
-
|
|
|
|
| 97 |
return
|
| 98 |
|
| 99 |
-
# Leer contenido Markdown
|
| 100 |
md_path = os.path.join(carpeta_salida, f"{nombre_sin_ext}.md")
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
|
|
|
|
|
|
|
|
|
|
| 105 |
|
| 106 |
tiempo_total = round(time.time() - start_time, 1)
|
| 107 |
palabras = len(md_content.split())
|
| 108 |
caracteres = len(md_content)
|
| 109 |
figuras = len([f for f in os.listdir(carpeta_salida) if f.lower().endswith(".jpeg")])
|
| 110 |
tablas = len(os.listdir(os.path.join(carpeta_salida, "tables"))) if os.path.exists(os.path.join(carpeta_salida, "tables")) else 0
|
| 111 |
-
tamaño_kb = os.path.getsize(md_path) / 1024
|
| 112 |
|
| 113 |
-
resumen = "\n
|
| 114 |
-
resumen += f"- Palabras
|
| 115 |
-
resumen += f"- Caracteres
|
| 116 |
-
resumen += f"- Tamaño
|
| 117 |
resumen += f"- Imágenes extraídas: {figuras}\n"
|
| 118 |
resumen += f"- Tablas detectadas: {tablas}\n"
|
| 119 |
-
resumen += f"- Tiempo total: {tiempo_total}
|
| 120 |
|
| 121 |
zip_path = os.path.join(output_base, f"{nombre_sin_ext}.zip")
|
| 122 |
shutil.make_archive(base_name=zip_path.replace(".zip", ""), format="zip", root_dir=carpeta_salida)
|
| 123 |
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
yield estado_final, gr.update(value=md_path, visible=True), gr.update(value=zip_path, visible=True), salida_md
|
| 127 |
|
| 128 |
except Exception as e:
|
| 129 |
-
|
|
|
|
| 130 |
|
| 131 |
-
# Interfaz
|
| 132 |
demo = gr.Interface(
|
| 133 |
fn=procesar_pdf,
|
| 134 |
inputs=[
|
|
@@ -141,13 +144,13 @@ demo = gr.Interface(
|
|
| 141 |
)
|
| 142 |
],
|
| 143 |
outputs=[
|
| 144 |
-
gr.Textbox(label="Estado del procesamiento", lines=
|
| 145 |
-
gr.
|
| 146 |
-
gr.File(label="Descargar
|
| 147 |
-
gr.
|
| 148 |
],
|
| 149 |
title="Marker PDF",
|
| 150 |
-
description="Convierte
|
| 151 |
flagging_mode="never"
|
| 152 |
)
|
| 153 |
|
|
|
|
| 25 |
os.makedirs(output_base, exist_ok=True)
|
| 26 |
start_time = time.time()
|
| 27 |
|
| 28 |
+
estado = ""
|
| 29 |
+
|
| 30 |
try:
|
| 31 |
subprocess.run(["marker_single", "--help"], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
| 32 |
except FileNotFoundError:
|
| 33 |
+
yield "Error: el comando 'marker_single' no está disponible.\nVerifica que marker-pdf esté en requirements.txt", "", None, None
|
| 34 |
return
|
| 35 |
|
| 36 |
if not formatos:
|
| 37 |
+
yield "Debes seleccionar al menos un formato de salida.", "", None, None
|
| 38 |
return
|
| 39 |
|
| 40 |
try:
|
|
|
|
| 41 |
if url_pdf:
|
| 42 |
parsed = urlparse(url_pdf)
|
| 43 |
nombre_pdf = parsed.path.split("/")[-1] or "documento"
|
| 44 |
ruta_pdf = obtener_ruta_segura(output_base, nombre_pdf)
|
| 45 |
response = requests.get(url_pdf)
|
| 46 |
if response.status_code != 200:
|
| 47 |
+
yield "No se pudo descargar el PDF desde la URL proporcionada.", "", None, None
|
| 48 |
return
|
| 49 |
with open(ruta_pdf, "wb") as f:
|
| 50 |
f.write(response.content)
|
|
|
|
| 53 |
ruta_pdf = obtener_ruta_segura(output_base, nombre_pdf)
|
| 54 |
shutil.copyfile(pdf_file.name, ruta_pdf)
|
| 55 |
else:
|
| 56 |
+
yield "No se proporcionó ni archivo ni URL.", "", None, None
|
| 57 |
return
|
| 58 |
|
|
|
|
| 59 |
with open(ruta_pdf, "rb") as f:
|
| 60 |
if f.read(4) != b"%PDF":
|
| 61 |
+
yield "El archivo proporcionado no es un PDF válido.", "", None, None
|
| 62 |
return
|
| 63 |
|
| 64 |
if os.path.getsize(ruta_pdf) > 3 * 1024 * 1024:
|
| 65 |
+
estado += "Aviso: el PDF es grande y puede tardar más de lo normal.\n"
|
| 66 |
|
| 67 |
nombre_sin_ext = os.path.splitext(os.path.basename(ruta_pdf))[0]
|
| 68 |
carpeta_salida = os.path.join(output_base, nombre_sin_ext)
|
|
|
|
| 70 |
shutil.rmtree(carpeta_salida)
|
| 71 |
os.makedirs(carpeta_salida, exist_ok=True)
|
| 72 |
|
|
|
|
| 73 |
for fmt in formatos:
|
| 74 |
+
tiempo = round(time.time() - start_time, 1)
|
| 75 |
+
estado += f"Procesando formato '{fmt}'... (tiempo: {tiempo} s)\n"
|
| 76 |
+
yield estado, f"{tiempo} s", None, None
|
| 77 |
+
|
| 78 |
if fmt == "md":
|
| 79 |
cmd = ["marker_single", ruta_pdf, "--output_format", "markdown", "--disable_ocr", "--output_dir", carpeta_salida]
|
| 80 |
elif fmt == "md + ocr":
|
|
|
|
| 82 |
else:
|
| 83 |
cmd = ["marker_single", ruta_pdf, "--output_format", fmt, "--output_dir", carpeta_salida]
|
| 84 |
|
|
|
|
|
|
|
| 85 |
for intento in range(1, 4):
|
| 86 |
try:
|
| 87 |
result = subprocess.run(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True, timeout=900)
|
| 88 |
if result.returncode == 0:
|
| 89 |
break
|
| 90 |
elif intento == 3:
|
| 91 |
+
estado += f"Error ejecutando marker_single (intento {intento}):\n{result.stderr}\n"
|
| 92 |
+
yield estado, "", None, None
|
| 93 |
return
|
| 94 |
time.sleep(2)
|
| 95 |
except Exception as e:
|
| 96 |
+
estado += f"Error inesperado durante el intento {intento}: {str(e)}\n"
|
| 97 |
+
yield estado, "", None, None
|
| 98 |
return
|
| 99 |
|
|
|
|
| 100 |
md_path = os.path.join(carpeta_salida, f"{nombre_sin_ext}.md")
|
| 101 |
+
if not os.path.exists(md_path):
|
| 102 |
+
estado += "No se generó el archivo Markdown. Puede que el PDF no contenga texto reconocible.\n"
|
| 103 |
+
yield estado, "", None, None
|
| 104 |
+
return
|
| 105 |
+
|
| 106 |
+
with open(md_path, "r", encoding="utf-8") as f:
|
| 107 |
+
md_content = f.read()
|
| 108 |
|
| 109 |
tiempo_total = round(time.time() - start_time, 1)
|
| 110 |
palabras = len(md_content.split())
|
| 111 |
caracteres = len(md_content)
|
| 112 |
figuras = len([f for f in os.listdir(carpeta_salida) if f.lower().endswith(".jpeg")])
|
| 113 |
tablas = len(os.listdir(os.path.join(carpeta_salida, "tables"))) if os.path.exists(os.path.join(carpeta_salida, "tables")) else 0
|
| 114 |
+
tamaño_kb = os.path.getsize(md_path) / 1024
|
| 115 |
|
| 116 |
+
resumen = f"\n---\nResumen del procesamiento:\n"
|
| 117 |
+
resumen += f"- Palabras: {palabras}\n"
|
| 118 |
+
resumen += f"- Caracteres: {caracteres}\n"
|
| 119 |
+
resumen += f"- Tamaño Markdown: {tamaño_kb:.2f} KB\n"
|
| 120 |
resumen += f"- Imágenes extraídas: {figuras}\n"
|
| 121 |
resumen += f"- Tablas detectadas: {tablas}\n"
|
| 122 |
+
resumen += f"- Tiempo total: {tiempo_total} s"
|
| 123 |
|
| 124 |
zip_path = os.path.join(output_base, f"{nombre_sin_ext}.zip")
|
| 125 |
shutil.make_archive(base_name=zip_path.replace(".zip", ""), format="zip", root_dir=carpeta_salida)
|
| 126 |
|
| 127 |
+
estado += f"Procesamiento finalizado correctamente. Tiempo total: {tiempo_total} s"
|
| 128 |
+
yield estado + resumen, f"{tiempo_total} s", gr.update(value=md_path, visible=True), gr.update(value=zip_path, visible=True)
|
|
|
|
| 129 |
|
| 130 |
except Exception as e:
|
| 131 |
+
estado += f"Error general inesperado: {str(e)}"
|
| 132 |
+
yield estado, "", None, None
|
| 133 |
|
| 134 |
+
# Interfaz 4 salidas (estado + tiempo + 2 archivos)
|
| 135 |
demo = gr.Interface(
|
| 136 |
fn=procesar_pdf,
|
| 137 |
inputs=[
|
|
|
|
| 144 |
)
|
| 145 |
],
|
| 146 |
outputs=[
|
| 147 |
+
gr.Textbox(label="Estado del procesamiento", lines=12),
|
| 148 |
+
gr.Textbox(label="Tiempo transcurrido", interactive=False),
|
| 149 |
+
gr.File(label="Descargar Markdown (.md)", visible=False),
|
| 150 |
+
gr.File(label="Descargar ZIP completo", visible=False)
|
| 151 |
],
|
| 152 |
title="Marker PDF",
|
| 153 |
+
description="Convierte artículos científicos en Markdown, HTML o JSON. Incluye OCR opcional y resumen detallado del contenido.",
|
| 154 |
flagging_mode="never"
|
| 155 |
)
|
| 156 |
|