Update core.py
Browse files
core.py
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
-
# Versión
|
| 2 |
import os
|
| 3 |
import zipfile
|
| 4 |
import uuid
|
| 5 |
import subprocess
|
| 6 |
import difflib
|
| 7 |
-
import io
|
| 8 |
import pdfplumber
|
| 9 |
import pandas as pd
|
| 10 |
from pypdf import PdfWriter, PdfReader, Transformation
|
|
@@ -14,14 +14,13 @@ from PIL import Image
|
|
| 14 |
from pptx import Presentation
|
| 15 |
from pptx.util import Inches
|
| 16 |
|
| 17 |
-
# ReportLab para generar PDFs (Informes y Capas de texto)
|
| 18 |
from reportlab.lib.pagesizes import A4, letter
|
| 19 |
from reportlab.lib import colors
|
| 20 |
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer
|
| 21 |
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
|
| 22 |
from reportlab.lib.enums import TA_JUSTIFY
|
| 23 |
-
from reportlab.pdfgen import canvas
|
| 24 |
-
from reportlab.lib.units import inch
|
| 25 |
|
| 26 |
from config import TEMP_DIR
|
| 27 |
|
|
@@ -102,29 +101,25 @@ class PDFEngine:
|
|
| 102 |
except ValueError: continue
|
| 103 |
return sorted(list(set(key_pages)))
|
| 104 |
|
| 105 |
-
# ---
|
| 106 |
-
|
| 107 |
def merge_pdfs(self, file_paths: list, order_indices: list = None, use_numbering: bool = False) -> str:
|
| 108 |
if not file_paths: raise ValueError("No hay archivos.")
|
| 109 |
|
| 110 |
-
# 1. Ordenar
|
| 111 |
ordered = []
|
| 112 |
if order_indices and len(order_indices) == len(file_paths):
|
| 113 |
try: ordered = [file_paths[int(i)] for i in order_indices]
|
| 114 |
except: ordered = file_paths
|
| 115 |
else: ordered = file_paths
|
| 116 |
|
| 117 |
-
# 2. Unir
|
| 118 |
m = PdfWriter()
|
| 119 |
for p in ordered: m.append(p)
|
| 120 |
|
| 121 |
temp_out = self._get_output_path("temp_unido.pdf")
|
| 122 |
with open(temp_out, "wb") as f: m.write(f)
|
| 123 |
|
| 124 |
-
# 3. Numerar (Si se solicitó)
|
| 125 |
if use_numbering:
|
| 126 |
final_out = self._add_page_numbers(temp_out)
|
| 127 |
-
# Intentar borrar el intermedio para no acumular basura (aunque el cron lo haría)
|
| 128 |
try: os.remove(temp_out)
|
| 129 |
except: pass
|
| 130 |
return final_out
|
|
@@ -137,25 +132,16 @@ class PDFEngine:
|
|
| 137 |
num_pages = len(reader.pages)
|
| 138 |
|
| 139 |
for i, page in enumerate(reader.pages):
|
| 140 |
-
# Crear un PDF en memoria con solo el número de página
|
| 141 |
packet = io.BytesIO()
|
| 142 |
-
# Usamos canvas de reportlab
|
| 143 |
can = canvas.Canvas(packet, pagesize=letter)
|
| 144 |
-
|
| 145 |
-
# Obtener tamaño de la página original para centrar el número
|
| 146 |
page_width = float(page.mediabox.width)
|
| 147 |
-
|
| 148 |
-
# Dibujar texto: "Página X de Y" en el centro inferior
|
| 149 |
text = f"Página {i+1} de {num_pages}"
|
| 150 |
can.setFont("Helvetica", 10)
|
| 151 |
-
can.drawCentredString(page_width / 2.0, 20, text)
|
| 152 |
can.save()
|
| 153 |
|
| 154 |
-
# Mover al inicio del buffer
|
| 155 |
packet.seek(0)
|
| 156 |
new_pdf = PdfReader(packet)
|
| 157 |
-
|
| 158 |
-
# Fusionar la capa del número con la página original
|
| 159 |
page.merge_page(new_pdf.pages[0])
|
| 160 |
writer.add_page(page)
|
| 161 |
|
|
@@ -169,16 +155,14 @@ class PDFEngine:
|
|
| 169 |
reader = PdfReader(file_path)
|
| 170 |
writer = PdfWriter()
|
| 171 |
|
| 172 |
-
# Crear la marca de agua en memoria una sola vez
|
| 173 |
packet = io.BytesIO()
|
| 174 |
can = canvas.Canvas(packet, pagesize=letter)
|
| 175 |
can.setFont("Helvetica-Bold", 50)
|
| 176 |
-
can.setFillColorRGB(0.5, 0.5, 0.5, 0.3)
|
| 177 |
|
| 178 |
-
# Guardar estado, rotar y escribir en el centro (aprox)
|
| 179 |
can.saveState()
|
| 180 |
-
can.translate(300, 400)
|
| 181 |
-
can.rotate(45)
|
| 182 |
can.drawCentredString(0, 0, text)
|
| 183 |
can.restoreState()
|
| 184 |
can.save()
|
|
@@ -187,9 +171,7 @@ class PDFEngine:
|
|
| 187 |
watermark_pdf = PdfReader(packet)
|
| 188 |
watermark_page = watermark_pdf.pages[0]
|
| 189 |
|
| 190 |
-
# Aplicar a todas las páginas
|
| 191 |
for page in reader.pages:
|
| 192 |
-
# Es necesario clonar el objeto watermark para cada pagina o pypdf se queja a veces
|
| 193 |
page.merge_page(watermark_page)
|
| 194 |
writer.add_page(page)
|
| 195 |
|
|
@@ -197,21 +179,16 @@ class PDFEngine:
|
|
| 197 |
with open(out, "wb") as f: writer.write(f)
|
| 198 |
return out
|
| 199 |
|
| 200 |
-
# --- NUEVO: Reparador (Ghostscript) ---
|
| 201 |
def repair_pdf(self, file_path: str) -> str:
|
| 202 |
if not file_path: raise ValueError("Falta archivo.")
|
| 203 |
|
| 204 |
out = self._get_output_path("reparado.pdf")
|
| 205 |
-
|
| 206 |
-
# Comando mágico de Ghostscript para regenerar PDFs
|
| 207 |
-
# -sDEVICE=pdfwrite : Reescribe el PDF
|
| 208 |
-
# -dPDFSETTINGS=/default : Calidad normal
|
| 209 |
cmd = [
|
| 210 |
"gs",
|
| 211 |
-
"-o", out,
|
| 212 |
"-sDEVICE=pdfwrite",
|
| 213 |
"-dPDFSETTINGS=/default",
|
| 214 |
-
"-dInteract=N",
|
| 215 |
"-dNOPAUSE", "-dQUIET", "-dBATCH",
|
| 216 |
file_path
|
| 217 |
]
|
|
|
|
| 1 |
+
## Versión Alfa 1.0
|
| 2 |
import os
|
| 3 |
import zipfile
|
| 4 |
import uuid
|
| 5 |
import subprocess
|
| 6 |
import difflib
|
| 7 |
+
import io
|
| 8 |
import pdfplumber
|
| 9 |
import pandas as pd
|
| 10 |
from pypdf import PdfWriter, PdfReader, Transformation
|
|
|
|
| 14 |
from pptx import Presentation
|
| 15 |
from pptx.util import Inches
|
| 16 |
|
|
|
|
| 17 |
from reportlab.lib.pagesizes import A4, letter
|
| 18 |
from reportlab.lib import colors
|
| 19 |
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer
|
| 20 |
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
|
| 21 |
from reportlab.lib.enums import TA_JUSTIFY
|
| 22 |
+
from reportlab.pdfgen import canvas
|
| 23 |
+
from reportlab.lib.units import inch
|
| 24 |
|
| 25 |
from config import TEMP_DIR
|
| 26 |
|
|
|
|
| 101 |
except ValueError: continue
|
| 102 |
return sorted(list(set(key_pages)))
|
| 103 |
|
| 104 |
+
# --- HERRAMIENTAS PRINCIPALES ---
|
| 105 |
+
|
| 106 |
def merge_pdfs(self, file_paths: list, order_indices: list = None, use_numbering: bool = False) -> str:
|
| 107 |
if not file_paths: raise ValueError("No hay archivos.")
|
| 108 |
|
|
|
|
| 109 |
ordered = []
|
| 110 |
if order_indices and len(order_indices) == len(file_paths):
|
| 111 |
try: ordered = [file_paths[int(i)] for i in order_indices]
|
| 112 |
except: ordered = file_paths
|
| 113 |
else: ordered = file_paths
|
| 114 |
|
|
|
|
| 115 |
m = PdfWriter()
|
| 116 |
for p in ordered: m.append(p)
|
| 117 |
|
| 118 |
temp_out = self._get_output_path("temp_unido.pdf")
|
| 119 |
with open(temp_out, "wb") as f: m.write(f)
|
| 120 |
|
|
|
|
| 121 |
if use_numbering:
|
| 122 |
final_out = self._add_page_numbers(temp_out)
|
|
|
|
| 123 |
try: os.remove(temp_out)
|
| 124 |
except: pass
|
| 125 |
return final_out
|
|
|
|
| 132 |
num_pages = len(reader.pages)
|
| 133 |
|
| 134 |
for i, page in enumerate(reader.pages):
|
|
|
|
| 135 |
packet = io.BytesIO()
|
|
|
|
| 136 |
can = canvas.Canvas(packet, pagesize=letter)
|
|
|
|
|
|
|
| 137 |
page_width = float(page.mediabox.width)
|
|
|
|
|
|
|
| 138 |
text = f"Página {i+1} de {num_pages}"
|
| 139 |
can.setFont("Helvetica", 10)
|
| 140 |
+
can.drawCentredString(page_width / 2.0, 20, text)
|
| 141 |
can.save()
|
| 142 |
|
|
|
|
| 143 |
packet.seek(0)
|
| 144 |
new_pdf = PdfReader(packet)
|
|
|
|
|
|
|
| 145 |
page.merge_page(new_pdf.pages[0])
|
| 146 |
writer.add_page(page)
|
| 147 |
|
|
|
|
| 155 |
reader = PdfReader(file_path)
|
| 156 |
writer = PdfWriter()
|
| 157 |
|
|
|
|
| 158 |
packet = io.BytesIO()
|
| 159 |
can = canvas.Canvas(packet, pagesize=letter)
|
| 160 |
can.setFont("Helvetica-Bold", 50)
|
| 161 |
+
can.setFillColorRGB(0.5, 0.5, 0.5, 0.3)
|
| 162 |
|
|
|
|
| 163 |
can.saveState()
|
| 164 |
+
can.translate(300, 400)
|
| 165 |
+
can.rotate(45)
|
| 166 |
can.drawCentredString(0, 0, text)
|
| 167 |
can.restoreState()
|
| 168 |
can.save()
|
|
|
|
| 171 |
watermark_pdf = PdfReader(packet)
|
| 172 |
watermark_page = watermark_pdf.pages[0]
|
| 173 |
|
|
|
|
| 174 |
for page in reader.pages:
|
|
|
|
| 175 |
page.merge_page(watermark_page)
|
| 176 |
writer.add_page(page)
|
| 177 |
|
|
|
|
| 179 |
with open(out, "wb") as f: writer.write(f)
|
| 180 |
return out
|
| 181 |
|
|
|
|
| 182 |
def repair_pdf(self, file_path: str) -> str:
|
| 183 |
if not file_path: raise ValueError("Falta archivo.")
|
| 184 |
|
| 185 |
out = self._get_output_path("reparado.pdf")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 186 |
cmd = [
|
| 187 |
"gs",
|
| 188 |
+
"-o", out,
|
| 189 |
"-sDEVICE=pdfwrite",
|
| 190 |
"-dPDFSETTINGS=/default",
|
| 191 |
+
"-dInteract=N",
|
| 192 |
"-dNOPAUSE", "-dQUIET", "-dBATCH",
|
| 193 |
file_path
|
| 194 |
]
|