From d6d750ae33a952e18feececaf31d63033bf5b96d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Andr=C3=A9s=20Moran?= <moranandres79@gmail.com>
Date: Wed, 11 Dec 2024 03:42:22 +0100
Subject: [PATCH] Scrapping

---
 .../panel_izquierdo.cpython-313.pyc           | Bin 0 -> 1580 bytes
 app/__pycache__/pomodoro.cpython-313.pyc      | Bin 8390 -> 8390 bytes
 app/__pycache__/scraping.cpython-313.pyc      | Bin 0 -> 3922 bytes
 app/panel_izquierdo.py                        |  21 +++++
 app/scraping.py                               |  79 ++++++++++++++++++
 main.py                                       |  15 ++--
 6 files changed, 110 insertions(+), 5 deletions(-)
 create mode 100644 app/__pycache__/panel_izquierdo.cpython-313.pyc
 create mode 100644 app/__pycache__/scraping.cpython-313.pyc
 create mode 100644 app/panel_izquierdo.py

diff --git a/app/__pycache__/panel_izquierdo.cpython-313.pyc b/app/__pycache__/panel_izquierdo.cpython-313.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..a2a58ac95c8873887469c945ba7f5dcf3a39a9e1
GIT binary patch
literal 1580
zcmZ`3O=}xRboNVH$se+5B)h6iWnq^tq*6Z|ii0qugcLW#xS9wI7M-qF<MrCBU1fGu
z)js7^8d~Vdmr|&Y?Ja+z(4!p+Hj@?#?Zr2zN=SR_o0a4=E`G4@`@MNzQg3W5jR4mE
zS#y6<5&A<4L(m6IZvmKngb`M5Bc{YuWhzkhw)T<E^dg!@C4{v+!umDHB7qsMVi~8B
zU9l8lMh>&vUf?=XGX!P8^zy*$qeh$^GAm5OY928iYmi5`3^BfH2h?BQyVvn3Ylf!=
z1UT9-56}dBQtqjc^b>U-txfpn<~pjK^Dmg%f8z&gtY8&ui)s;F5+fhJ0MPWHkxQeJ
z7eYY922S9ld}-YrT5It#hEkuFP+h@hU0qOcDv!2JoL<xdt*+I^UiOm`9>JrFN)frp
z8e0)5-*dP4hToy5KZDUyg}UJiqw6*K)-5LYbSmqH?IgNXy6#Ob2cF~Etcc~d{<kjK
z`z!GEwgD|E`)D0#A8LcO4`oS9XnlA>s3!Mw0kCoxEvQ&sRBNdVCJ?U;RCCDER`zmR
zoVTOvl}fkUE!!=7FAT~wsz6T(ZZns1p#zEFQc`@2QM(DJCK7jJmdQPfg4gej3@?%F
z5-*4-bG$GRMg(a%jADe-&pCk^LC#2y=f2o<t~cu77&TfUvxCOX(CM^jz@x^y<mQb#
z4NTc*Ft#u%(xAv_v$49%x5A)t%L{COm4(~X;WvUNqtS<V;sh|-MqA!9@rDOoZtn`S
z>f5c2roAMrm#@7{pi++$Qm*=;WBbt(Fn!kvs4%Jk`ujfmZuz)y^-<yKuT!rdPgRem
zswcVf!;!;m^@*X)j6YAH%#{c9EB7p^&E$FzA}>Oh?TVxmwpuonD7OSd(A#!%*GdXw
zli4jQQcf6bdTxhNG43+2N!l!I+pf*2$T*?jX$3DFRK6n;A9Of}p+XvG$I^u!QGb)o
zKyH==Br+QznEQ*6M;6I5DHcC1vYa%V!*<2gC3BEvMjQ}y{d+e5b^62{{U-e&eQXwv
z%)-IQ!}ks)_x`zh_Q;$)nE%Cm^ITIpGD`>V{P6Li30&pReBpcRTkB+W@?>)Esj6jj
zPm;;k(qCrICQxd6kaiZ5J~HRntMIU;#Qm&T30ZCcQzAs932BARjxYI9Lhf~Je_+WF
zvgxsi`(8kUkP!BUY&1&<oE`TZVsp;CjSi;~AwQz{g?JWB{^VjLnMnSPe1V>+8DqZh
zBehUMlJ9r{r;N=?_enrPgCmO%zn|hQX^@2vMqKR|y2RcFojmhs8qk@hD9U4W<uRK0
Q4NX7SCzScW5TH2nU$E72x&QzG

literal 0
HcmV?d00001

diff --git a/app/__pycache__/pomodoro.cpython-313.pyc b/app/__pycache__/pomodoro.cpython-313.pyc
index c148c317c10350a1ed94aaf032e0e7c1c9f90d15..31f56c0e584f66cdccf9ec79a15db1a9a44b236a 100644
GIT binary patch
delta 22
ccmX@+c+8RaGcPX}0}!0N7m>bVBkxWH09bwp*Z=?k

delta 22
ccmX@+c+8RaGcPX}0}!n74o`ovk$0y808|<WX8-^I

diff --git a/app/__pycache__/scraping.cpython-313.pyc b/app/__pycache__/scraping.cpython-313.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..1f8fcf776009df41d8b980a94e021e513aaa7adc
GIT binary patch
literal 3922
zcmai1O>7&-6`tkpQcEsHO5*Crl0~f?*%4h^@{fLOCACV)l%-gr94;9+m0)5`u4Ee2
zE;GAKY+a%R1uCNl*$o26y~LLSF3^L4UVP}GjxR=gC~3(hu~ymw?Ik@amJfk*>6;~&
zlocf&NHeqZ-h4AVeBU>3x9jWO2+HraFD|Beg#JSw?Bb|&wx^-<Ig*e>oky6oj`P$z
z4(t%iV>G5M8-rPB>GRHa>Tn&{n0ACYiG|uFIiYq-bx`w*RFM0IKu3a<SwE)8IU~K0
z%Sc)-3j^%CIVGngD@$7q2Q>}<+kb}Ya}+>x)-b7}_8zMvFvjk9sB|Sta!9nqNUY>M
z&8jGldQkW4aGXEHsY7xbLMk=qvTagbobsZRjzOwCBBCxh6_pB6i<NePT0lRiyU+r>
zcf@pMw4|K5qUlEZe;$NP25XvOGRIGx956XOiRElsT@11&LtIQIt6`{fx`ItEE9?44
z8cvy9N;c%nvaT%mhB4N#C}+f^rYcE8#-t-=WRaM~6#U7CrjLk#$Sz&iKggIaa3HQ4
zHtQ-~leJ(>HjC4$5p<XwLA#dLRFhUP*2y7@qM1mJ%*SCT`uviHWp)0nmdq_Fs-e%H
zNSqDL%u5RXFr7?mx~QlkR#Nj9t{Yc0b$%kP%9#sTy9%p@)D%|qb2AozoXyVLuiTft
zZZ>F_4MkOO!d@P?!pbB-brKLOXv@K~!lu9N_UNtAqW^Hgf4J1Rztq@VYHTevca)mD
zws<7`YzsLB_bRvLMGbAMzV9~qhPBZfqeWgU@ZvUd00LK^Z#eb9|A5VV&unm`n?C>d
zTPzq}>dmmurw0zA-v>e+^h4SkYM~#tu+UGE2($}{3@aj^z_0%Lh&%xCX2*w0$LT0z
z1JzoI8o7v&tRV#UJt9wKYB3jQB*$PS@>Mp@MuEAzOF7z4JN#&J7$VjgcS=q|3>D%v
zTEE*HR;{ylD5AKdx8jh3Rb~!=)e-gma1O+Nox~jlgS{IO&74MvM<SYQwEZPl4S8|r
zxO?xDc$H0yuky6XMy<6`q1Ly16`pcCcEQKFSCK8Aav{#*DYQFy>WO^w7@(H<E5=f$
zFDqlhKT3{)J~3CYvVc3l3fxdrCgsk-`JnJM1Ix)|`q7_M!a7CG$Vo*9GA?B^iXo@u
zzU7W8SHV$9DL`97)Od7yL|pDJ8&f$TmZ6YwUCRKOj9f~bjKwYhgL#v~$_F_`H*}L;
zR1A~X4cW-)z?PI^GKK<MAVFzz3u!f#kTV&R&gKl$nO0%@2KE!1Te+50vIbCFHyfg5
zxUfAGbeVJxXG|ukWn_~lh>4HVDYBLeSeqql%cd9D2{cWov;;XGlhHxBO=tO)VFBya
z6S@H>rszZ`MZ4e@vIQU6s0=Kx91=bTqwVmo{~anqaxXi)DeS*J{oCog{&nHts;k5c
zW%}M7U+3Q}bsqZi(w$4IkxjpNH*mjm-9Nf|w$vW@^5~tT_vrgQUmRU;AHL}<b-r<T
z;?AX;?oxwro4dt*(Z12p^Dlqk=36D9t0){Q2#4<G?p=K_Q0N=|hwzJUU8v*K3pc87
z$~O%?^$h>l)0h|f@4Je}!i8hueE&qg>D*J#B*=-s$Oj61pvd<Z`2KA~!ETA59KYXv
z-=Fsm7P+AUH}w4$4ThI`54_W-p(A6*(8CiwP=D3hM(SgM@k8_@uXntge$>rEKUujy
zz#Sn%R7+p}YU#^d)i!9E$qf{zsvHIDA_7*E8M{i=-vJEX^IoiRGXN?Tt=k0vk5vp%
zF3Dl36s$Nq0RDSpg5LmDQv}#q^|q{`+o+%vRrQjXIQ=R|XU<#NLqc~~sm%xcpS175
zch=t4(iOOX%O&cyd$l&&Q0v>himos_Fxfcs>YgoKVM0tK0eJwlJ9MQkvV3nm8V<$6
zVk|T^9Tul1#K>$+3}2j*Vp3H#dfhf-cDL!2I2DPBp?GXIF%<!*nJ_$Fh)&IfqVJ36
z!|xyY(Wzy&iSLD?<CCFi?}@>|pcs!#y&DhjIVEV)6H}4p{;7x*j>g2PbCKC7>^Bmd
zt!=Xm83fGfcvup959-0?Lt%^mVp(+Lltz*Pkh5L+$Z~sC4J)3wf;%GS4#ROz5z#8^
z=^g+gNSvHWFB4%6x~(+9bS86HCs~bhO-beq1$O}qjA~0u>9YD^AB<q~vtn|Om~1km
z=}M5rgi4rPBlsk#$c=`Zq0$Lutlyw4T`l%rZS6`}ITeN*79An}e}ifTZTh;_ecdaQ
z|K$B8&in8B=C#C)#Cm;TWn#;TSpL_MPa<207WTbBG~2%AM!XMlnA>f)+CFPvb(Z+X
zoAKX_mOPDH4C@*yd79QDHzN6t>Bk>FPH%WFmOR4R?2XwCPv<64sF4Rl4|?-W;~So{
z-!=l$RvTGi`jhEjMM{i^EVaQ1n~W>(9(u|QfA@k%4b5vG-}t!b>nZqpK)qZ;7Ea$i
z{hpF<8Yp@O3!XuuUap~+IwZU3Vg2#3Bj~Fmf$;<MBe!?FlYZ36LO)r_K)?l|YDU!a
zV~XXNtNQPi{~SPb)vV{DBTfy_5_Q_4(yyJxS?fa;cMMS098i(HGNK)d!Mw&!@0B6#
zn7?;6wBs8f&ipjOA_~$GNVJ`hv6JhSfkwwC;4Ok2CfT9sG9b~AQ}8ihI%8JXx+kqr
zGg-KY$#=n3Ev@1=$SfAW2@MXCsUC8D+YxHkUrA@Qgq>!aUhC0{V@TE2-4VV&twPoz
z<LZRO^+eSfj{!U=GSUsGAkO`5pAD}}Zwj47q34Ov^T*l0oXo!`=f^G=gyhQ1W^<s}
zeDF#0!HuKikKcaWUugd23It_ik!dY3tsBg~64O*<+6qkD2Geehw-lI`4W_NcG!z-3
zzz8?v8%&4gMP@u@_HDzuDCoffXiWODK44`$_$~72v$)t-`KHmZ#ded0j47vBXKKkr
z?G&xI`2)h^Fu6=`?(HQlmCGn+@Fa|pOIIh#09>G`XUO+8>U@S8o}renQR{QJhZ3HP
TK5F=RPX~2s`wU05TS)&04xL2I

literal 0
HcmV?d00001

diff --git a/app/panel_izquierdo.py b/app/panel_izquierdo.py
new file mode 100644
index 0000000..92f7025
--- /dev/null
+++ b/app/panel_izquierdo.py
@@ -0,0 +1,21 @@
+import tkinter as tk
+import threading
+from app import scraping
+
+class PanelIzquierdo:
+    def __init__(self, frame, text_widget):
+        # Asociar el frame del panel izquierdo
+        self.frame = frame
+        self.frame.configure(bg="lightblue", width=200)
+        self.frame.grid_propagate(False)
+        self.frame.columnconfigure(0, weight=1)
+        
+        boton_scrapping = tk.Button(
+            frame,
+            text="Iniciar Scrapping",
+            command=lambda: threading.Thread(
+                target=scraping.iniciar_scraping_y_insercion, 
+                args=("https://www.amazon.es/", text_widget) 
+            ).start()
+        )
+        boton_scrapping.pack(pady=5)
\ No newline at end of file
diff --git a/app/scraping.py b/app/scraping.py
index e69de29..8792a34 100644
--- a/app/scraping.py
+++ b/app/scraping.py
@@ -0,0 +1,79 @@
+import requests
+from bs4 import BeautifulSoup
+import threading
+import mysql.connector  
+from queue import Queue
+
+def obtener_conexion():
+    try:
+        conexion = mysql.connector.connect(
+            host='localhost',  
+            port=3306,
+            user='root',  
+            password='1234', 
+            database='scraping'  
+        )
+        return conexion
+    except mysql.connector.Error as err:  
+        print(f"Error al conectar con la base de datos: {err}")
+        return None
+
+
+# Función para extraer enlaces y enviarlos a la cola
+def extraer_enlaces(url, cola, text_widget):
+    try:
+        respuesta = requests.get(url)
+        if respuesta.status_code == 200:
+            contenido_html = respuesta.text
+            soup = BeautifulSoup(contenido_html, 'html.parser')
+            for enlace in soup.find_all('a', href=True):
+                enlace_str = enlace['href']
+                cola.put(enlace_str)  
+                if text_widget:  
+                    text_widget.insert('end', enlace_str + "\n")
+            print("Extracción de enlaces completada.")
+        else:
+            print(f"Error al acceder a la URL: {respuesta.status_code}")
+    except requests.exceptions.RequestException as e:
+        print(f"Error durante la solicitud HTTP: {e}")
+
+# Función para insertar enlaces en la base de datos
+def insertar_enlaces_mysql(cola):
+    conexion = obtener_conexion()
+    if not conexion:
+        return
+
+    cursor = conexion.cursor()
+
+    cursor.execute('''CREATE TABLE IF NOT EXISTS enlaces (
+                        id INT AUTO_INCREMENT PRIMARY KEY,
+                        enlace VARCHAR(255) UNIQUE
+                    )''')
+
+    while True:
+        enlace = cola.get()  
+        if enlace == "FIN":
+            break 
+
+        try:
+            cursor.execute("INSERT IGNORE INTO enlaces (enlace) VALUES (%s)", (enlace,))
+            conexion.commit()
+            print(f"Enlace insertado en la base de datos: {enlace}")
+        except mysql.connector.Error as err:  
+            print(f"Error al insertar enlace: {err}")
+
+    cursor.close()
+    conexion.close()
+    print("Inserción en la base de datos finalizada.")
+
+
+# Función para iniciar el scraping y la inserción en hilos
+def iniciar_scraping_y_insercion(url, text_widget):
+    cola_enlaces = Queue()
+    hilo_scraping = threading.Thread(target=extraer_enlaces, args=(url, cola_enlaces, text_widget))
+    hilo_insercion = threading.Thread(target=insertar_enlaces_mysql, args=(cola_enlaces,))
+    hilo_scraping.start()
+    hilo_insercion.start()
+    hilo_scraping.join()
+    cola_enlaces.put("FIN")
+    hilo_insercion.join()
\ No newline at end of file
diff --git a/main.py b/main.py
index 768ef98..e50504f 100644
--- a/main.py
+++ b/main.py
@@ -4,11 +4,14 @@ from tkinter import ttk # Importar el widget ttk
 from tkinter import messagebox
 from app import monitorization
 from app import panel_derecho
+from app import panel_izquierdo
 from app import pomodoro
 import threading
 import time
 import datetime
 import psutil
+from app import scraping
+
 
 def update_time(status_bar):
         while True:
@@ -73,8 +76,6 @@ frame_derecho.grid(row=0, column=2, sticky="ns")
 frame_izquierdo.grid_propagate(False)
 frame_derecho.grid_propagate(False)
 
-panel_izquierdo = panel_derecho.PanelDerecho(frame_derecho)
-
     # Dividir el frame central en dos partes (superior variable e inferior fija)
 frame_central.rowconfigure(0, weight=1)  # Parte superior, tamaño variable
 frame_central.rowconfigure(1, weight=0)  # Parte inferior, tamaño fijo
@@ -102,12 +103,13 @@ style.configure("CustomNotebook.TNotebook.Tab", font=("Arial", 12, "bold"))
 notebook = ttk.Notebook(frame_superior, style="CustomNotebook.TNotebook")
 notebook.pack(fill="both", expand=True)
 
-
     # Crear cinco solapas
 for i in range(1, 6):
     tab = ttk.Frame(notebook)
     if i == 1:
         notebook.add(tab, text="Scrapping", padding=4)
+        text_scrapping = tk.Text(tab, wrap="word")
+        text_scrapping.pack(fill="both", expand=True)
     elif i == 2:  # Identificar la solapa 2
         notebook.add(tab, text="Pomodoro", padding=4)
         pomodoro.PomodoroTimer(tab)
@@ -125,8 +127,6 @@ for i in range(1, 6):
 
 
     # Usar pack para alinear los labels horizontalmente
-
-
 label_cpu_used = tk.Label(barra_estado, text="CPU: 0%", font=("Helvetica", 11), bd=1, fg="blue", relief="sunken", anchor="w", width=20, padx=10)
 label_ram_used = tk.Label(barra_estado, text="RAM: 0%", font=("Helvetica", 11), bd=1, fg="blue", relief="sunken", anchor="w", width=20, padx=10)
 label_pc_battery = tk.Label(barra_estado, text="Temperatura CPU: ", font=("Helvetica", 11), bd=1, fg="blue", relief="sunken", anchor="w", width=20, padx=10)
@@ -140,6 +140,7 @@ label_network_used.pack(side="left", fill="x", expand=True)
 label_fecha_hora.pack(side="right", fill="x", expand=True)
 
 
+#Monitorizacion barra horizontal de abajo
 update_thread = threading.Thread(target=update_time, args=(label_fecha_hora,))
 update_thread.daemon = True  
 update_thread.start()
@@ -160,6 +161,10 @@ thread_network_used_monitor = threading.Thread(target=monitorization.monitor_net
 thread_network_used_monitor.daemon = True
 thread_network_used_monitor.start()
 
+#Lados verticulas
+panel_d = panel_derecho.PanelDerecho(frame_derecho)
+panel_i = panel_izquierdo.PanelIzquierdo(frame_izquierdo,  text_scrapping) 
+
 
     # Ejecución de la aplicación
 root.mainloop()