From e06e8fcf373c6ed21fa9bc1a99ca6784097b63c2 Mon Sep 17 00:00:00 2001
From: Raghugowd <84554619+Raghugowd@users.noreply.github.com>
Date: Tue, 25 Apr 2023 01:42:58 +0530
Subject: [PATCH 1/2] Homework

    Write a Python program to Scrape the pages from Python Blogs and save the data into the DB. and dockerize.
---
 Dockerfile         | 18 +++++-------------
 docker-compose.yml | 10 ++++++++++
 webScraping.py     | 32 ++++++++++++++++++++++++++++++++
 3 files changed, 47 insertions(+), 13 deletions(-)
 create mode 100644 docker-compose.yml
 create mode 100644 webScraping.py

diff --git a/Dockerfile b/Dockerfile
index f07ae2b..84f0e65 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,17 +1,9 @@
 FROM python:3.10.2-alpine3.15
-COPY . .
-# Install Postgres
-RUN apk update
-RUN apk add postgresql
-RUN chown postgres:postgres /run/postgresql/
-# Install requirements
-COPY ./requirements.txt /tmp
-RUN pip install -r /tmp/requirements.txt
-# For psycopg2
-RUN apk add --virtual postgresql-deps libpq-dev
 # Create directories
 RUN mkdir -p /root/workspace/src
-# Mount your local file
-COPY ./web_scraping_sample.py /root/workspace/src
+COPY ./webScraping.py /root/workspace/src
 # Switch to project directory
-WORKDIR /root/workspace/src
\ No newline at end of file
+WORKDIR /root/workspace/src
+# Install required packages
+RUN pip install --upgrade pip
+RUN pip install requests bs4 html5lib psycopg2-binary
diff --git a/docker-compose.yml b/docker-compose.yml
new file mode 100644
index 0000000..e3bff76
--- /dev/null
+++ b/docker-compose.yml
@@ -0,0 +1,10 @@
+
+psql-db:
+  image: 'postgres:14'
+  container_name: psql-db
+  environment:
+    - PGPASSWORD=123456
+    - POSTGRES_USER=postgres
+    - POSTGRES_PASSWORD=123456
+  ports:
+    - '5434:5432'
diff --git a/webScraping.py b/webScraping.py
new file mode 100644
index 0000000..160f7b9
--- /dev/null
+++ b/webScraping.py
@@ -0,0 +1,32 @@
+import requests
+from bs4 import BeautifulSoup
+import re
+import psycopg2
+conn = psycopg2.connect(
+host="172.17.0.2",
+port="5432",
+database="pybd",
+user="postgres",
+password="123456"
+)
+print("Connection Successful")
+cur = conn.cursor()
+res = requests.get('https://blog.python.org/')
+soup = BeautifulSoup(res.content, 'html5lib')
+titles=[]
+authors=[]
+for i in soup.find_all('h3', class_='entry-title'):
+
+ string = i.find('a').getText() 
+ titles.append(string.strip())
+for i in soup.find_all('span', class_='fn'):
+ string = i.getText()
+ authors.append(string.strip())
+for i in range(4):
+ cur.execute(
+"INSERT INTO py1(no,title,author) VALUES(%s,%s,%s)", (i+1, titles[i], authors[i])
+)
+conn.commit()
+cur.close()
+conn.close()
+

From cdd6dfb15ae583f0cbf816506f66f8421faa70c1 Mon Sep 17 00:00:00 2001
From: Raghugowd <84554619+Raghugowd@users.noreply.github.com>
Date: Tue, 25 Apr 2023 09:55:55 +0530
Subject: [PATCH 2/2] homework

---
 HOMEWORK/Dockerfile         |  9 +++++++++
 HOMEWORK/docker-compose.yml | 10 ++++++++++
 HOMEWORK/webScraping.py     | 33 +++++++++++++++++++++++++++++++++
 3 files changed, 52 insertions(+)
 create mode 100644 HOMEWORK/Dockerfile
 create mode 100644 HOMEWORK/docker-compose.yml
 create mode 100644 HOMEWORK/webScraping.py

diff --git a/HOMEWORK/Dockerfile b/HOMEWORK/Dockerfile
new file mode 100644
index 0000000..84f0e65
--- /dev/null
+++ b/HOMEWORK/Dockerfile
@@ -0,0 +1,9 @@
+FROM python:3.10.2-alpine3.15
+# Create directories
+RUN mkdir -p /root/workspace/src
+COPY ./webScraping.py /root/workspace/src
+# Switch to project directory
+WORKDIR /root/workspace/src
+# Install required packages
+RUN pip install --upgrade pip
+RUN pip install requests bs4 html5lib psycopg2-binary
diff --git a/HOMEWORK/docker-compose.yml b/HOMEWORK/docker-compose.yml
new file mode 100644
index 0000000..e3bff76
--- /dev/null
+++ b/HOMEWORK/docker-compose.yml
@@ -0,0 +1,10 @@
+
+psql-db:
+  image: 'postgres:14'
+  container_name: psql-db
+  environment:
+    - PGPASSWORD=123456
+    - POSTGRES_USER=postgres
+    - POSTGRES_PASSWORD=123456
+  ports:
+    - '5434:5432'
diff --git a/HOMEWORK/webScraping.py b/HOMEWORK/webScraping.py
new file mode 100644
index 0000000..670e58a
--- /dev/null
+++ b/HOMEWORK/webScraping.py
@@ -0,0 +1,33 @@
+import requests
+from bs4 import BeautifulSoup
+import re
+import psycopg2
+conn = psycopg2.connect(
+host="172.17.0.2",
+port="5432",
+database="pybd",
+user="postgres",
+password="123456"
+)
+print("Connection Successful")
+cur = conn.cursor()
+res = requests.get('https://blog.python.org/')
+soup = BeautifulSoup(res.content, 'html5lib')
+titles=[]
+authors=[]
+for i in soup.find_all('h3', class_='entry-title'):
+
+ string = i.find('a').getText() 
+ titles.append(string.strip())
+for i in soup.find_all('span', class_='fn'):
+ string = i.getText()
+ authors.append(string.strip())
+for i in range(4):
+ cur.execute(
+ "CREATE TABLE py1(no INT, title VARCHAR(100),author VARCHAR(100));"
+"INSERT INTO py1(no,title,author) VALUES(%s,%s,%s)", (i+1, titles[i], authors[i])
+)
+conn.commit()
+cur.close()
+conn.close()
+