v0.1 of send and receive scripts

slack-bot/traffic-scraper/dev/scraper.py

@@ -0,0 +1,70 @@
+from bs4 import BeautifulSoup
+from selenium import webdriver
+import emoji
+from datetime import datetime
+import re
+
+url = 'https://www.trafficdelays.co.uk/m62-traffic-delays/'
+# prepare the  option for the chrome driver
+options = webdriver.ChromeOptions()
+options.add_argument('headless')
+
+# start the chrome driver
+browser = webdriver.Chrome(options=options)
+browser.get(url)
+html = browser.page_source
+
+
+congestion = browser.find_element_by_xpath('//*[@id="congestion"]')
+accident = browser.find_element_by_xpath('//*[@id="accident"]')
+
+congestionText = congestion.text
+congestionHTML = congestion.get_attribute('innerHTML')
+congestionCount = congestionHTML.count('<li>')
+pattern = r".*\<li\>.*title=\".*\".*\>((.|\n)*?)\<br\>"
+congestionRegexExtraction = re.findall(pattern, congestionHTML)
+
+accidentText = accident.text
+accidentHTML = accident.get_attribute('innerHTML')
+accidentCount = accidentHTML.count('<li>')
+accidentRegexExtraction = re.findall(pattern, accidentHTML)
+
+
+def printBreak():
+    print('\n')
+
+
+currentTime = datetime.now().strftime('%H:%M:%S')
+
+print(emoji.emojize('Did someone say M62 :anguished:!? Let'
+                    '\'s check the latest updates from Highways'
+                    f' England! as of {currentTime}'
+                    ':police_car::rotating_light:',
+                    use_aliases=True))
+printBreak()
+if congestionCount == 0:
+    print(emoji.emojize(f'There are currently no reported congestions on the'
+                        f' M62 :thinking_face:', use_aliases=True))
+if congestionCount != 0:
+    print(emoji.emojize(f'There are currently {congestionCount} incident(s)'
+                        f' on the M62 :scream:', use_aliases=True))
+    for i in range(0, congestionCount):
+        print(congestionRegexExtraction[i][0] + '\n')
+printBreak()
+if accidentCount == 0:
+    print(emoji.emojize(f'There are currently no reported accidents on the'
+                        f' M62 :thinking_face:', use_aliases=True))
+if accidentCount != 0:
+    print(emoji.emojize(f'There are currently {accidentCount} incident(s)'
+                        f' on the M62 :scream:', use_aliases=True))
+    for i in range(0, accidentCount):
+        print(accidentRegexExtraction[i][0] + '\n')
+printBreak()
+print(f'Hey Andy, have you thought about getting the train?'
+      + emoji.emojize(f' :bullettrain_front:', use_aliases=True))
+print(f'Hey Andy, maybe flying would be quicker?'
+      + emoji.emojize(f' :helicopter:', use_aliases=True))
+print(f'Don\'t fret, he can always work from home!'
+      + emoji.emojize(f' :house_with_garden:', use_aliases=True))
+
+browser.quit()

slack-bot/traffic-scraper/dev/scraper.py.old

@@ -0,0 +1,71 @@
+from bs4 import BeautifulSoup
+from selenium import webdriver
+import emoji
+from datetime import datetime
+import re
+
+url = 'https://www.trafficdelays.co.uk/southern-england/'
+# prepare the  option for the chrome driver
+options = webdriver.ChromeOptions()
+options.add_argument('headless')
+
+# start the chrome driver
+browser = webdriver.Chrome(options=options)
+browser.get(url)
+html = browser.page_source
+
+soup = BeautifulSoup(html, features='lxml')
+# soup.find_all(class_='alerts-severity-Severe')
+
+
+def printBreak():
+    print('\n')
+
+
+# table = soup.find_all('td')
+
+# list = []
+
+# for item in table:
+#     list.append(item)
+
+# for i in range(0, 4):
+#     print(list[i])
+
+# printBreak()
+
+# totalItems = int(len(list) / 4)
+
+# for i in range(0, 4):
+#     print(list[i].string)
+#     printBreak()
+
+# newList = ([x.text for x in soup.find_all('td')])
+
+currentTime = datetime.now().strftime('%H:%M:%S')
+
+print(emoji.emojize('Did someone say M62 :anguished:!? Let'
+                    '\'s check the latest updates from Highways'
+                    ' England! :police_car::rotating_light:',
+                    use_aliases=True))
+print(f'As of {currentTime}, there is currently {newList[2]} {newList[1]}'
+      f' on the {newList[0]}')
+
+# print(list[3].prettify())
+desc = str(list[3])
+descSplit = desc.split('<')
+
+pattern = r"\>(.+?)\s+?\:\s+?(.+?)$"
+
+for item in descSplit:
+    # print(item)
+    regex = re.findall(pattern, item)
+    # print(f'Type: {type(regex)}')
+    # print(f'Matches: {regex}')
+    try:
+        printBreak()
+        print(f'itemOne: {regex[0][0]}')
+        print(f'itemTwo: {regex[0][1]}')
+    except IndexError:
+        pass
+    # print(f'itemTwo: {regex[1]}')

slack-bot/traffic-scraper/dev/sendToSQS_old.py

@@ -0,0 +1,64 @@
+import boto3
+import os
+import sys
+sys.path.append(os.getcwd())
+sys.path.append('/home/dtomlinson/projects/slack-bot/traffic-scraper/prd')
+from pullTrafficInfo import getTrafficInfo
+import base64
+# import emoji
+import hashlib
+from datetime import datetime
+
+motorway = 'A3'
+currentTime = datetime.now().strftime('%H:%M:%S')
+
+session = boto3.Session(profile_name='plex-aws')
+sqs = session.client('sqs')
+
+# queue = sqs.create_queue(QueueName='test', Attributes={'DelaySeconds': '5'})
+
+inst = getTrafficInfo.getTrafficURL(motorway).findIncidents() \
+    .getIncidentInformation().generateOutput()
+
+# encoded = base64.b64encode(inst.output[0].encode())
+# decoded = base64.b64decode(encoded).decode()
+
+queueURL = 'https://sqs.eu-west-1.amazonaws.com/745437999005/slack-bot.fifo'
+
+# message = sqs.send_message(QueueUrl=queueURL,
+#                            MessageBody='string',
+#                            MessageGroupId='slack-bot-motorway')
+
+# response = sqs.receive_message(QueueUrl=queueURL,
+#                                MaxNumberOfMessages=5)
+# if 'Messages' in response:
+#     for message in response['Messages']:
+#         print(message['Body'])
+# else:
+#     print('Queue is empty')
+
+# for item in inst.output:
+#     print(item)
+#     encoded = (base64.b64encode(item.encode())).decode()
+#     dedupId = hashlib.md5((item + currentTime).encode()).hexdigest()
+#     msg = sqs.send_message(QueueUrl=queueURL,
+#                            MessageBody=encoded,
+#                            MessageGroupId=f'slack-bot-{motorway}',
+#                            MessageDeduplicationId=dedupId)
+#     if msg is not None:
+#         print(msg["MessageId"])
+#     # print(encoded)
+#     # print(encoded.encode())
+#     # print(base64.b64decode(encoded.encode()).decode())
+
+
+response = sqs.receive_message(QueueUrl=queueURL,
+                               MaxNumberOfMessages=10)
+if 'Messages' in response:
+    for message in response['Messages']:
+        # print(message['Body'])
+        # break
+        decoded = base64.b64decode(message['Body'].encode())
+        print(decoded.decode())
+else:
+    print('Queue is empty')