2martens
/
allris-scraper
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Added support for multiple districts

This commit is contained in:
Jim Martens 2020-01-09 23:46:40 +01:00
parent 9f89a78763
commit 7ade635756
4 changed files with 89 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
README.md
View File

@ -2,7 +2,7 @@
![Apache Licence 2.0](https://img.shields.io/pypi/l/twomartens.allrisscraper)
![Supports Python 3.7 and 3.8](https://img.shields.io/pypi/pyversions/twomartens.allrisscraper)
![version 0.2.2](https://img.shields.io/pypi/v/twomartens.allrisscraper)
![version 0.3.0](https://img.shields.io/pypi/v/twomartens.allrisscraper)
This scraper requires your username and password and performs the following tasks for you:
@ -10,7 +10,8 @@ This scraper requires your username and password and performs the following task
- download of all agendas and motions related to upcoming meetings of committees and plenary sessions
- Only considers meetings where you already have been invited formally through ALLRIS.
**IMPORTANT:** As of now only the Hamburg district of Eimsbüttel is supported.
**IMPORTANT:** All districts are supported but official committe abbreviations will only work for
Eimsbüttel as of now.
## Requirements
@ -25,6 +26,22 @@ This scraper requires your username and password and performs the following task
2. Run ALLRIS scraper a first time ``tm-allrisscraper`` (creates config ini in your current working directory)
3. Fill out the config file with your login credentials and an absolute path on your system to store PDFs of files
## Configuration
```ini
[Default]
; possible values for district: Altona, Bergedorf, Eimsbüttel, Hamburg-Nord,
; Hamburg-Mitte, Harburg, Wandsbek
district = Eimsbüttel
; if you are not from Eimsbüttel your domain ending will differ
username = max.mustermann@eimsbuettel.de
; password is stored in clear text, therefore ini file should have most
; restrictive read permissions
password = VerySecurePassword
; location for storage of PDFs (trailing slash is IMPORTANT)
pdflocation = /path/to/storage/of/PDFs/
```
## Usage after initial setup
Run ALLRIS scraper: ``tm-allrisscraper`` (takes a few seconds to finish)

2
setup.py
View File

@ -30,7 +30,7 @@ setup(
author="Jim Martens",
author_email="github@2martens.de",
url="https://git.2martens.de/2martens/allris-scraper",
version="0.2.2",
version="0.3.0",
namespace_packages=["twomartens"],
packages=find_packages('src', exclude=["*.tests", "*.tests.*", "tests.*", "tests"]),
package_dir={'': 'src'},

42
src/twomartens/allrisscraper/definitions.py
View File

@ -16,11 +16,43 @@
ABBREVIATIONS = {
"Altona": {
"Haupt": "HA",
},
"Bergedorf": {
"Haupt": "HA",
},
"Eimsbüttel": {
"RaLNS": "RaLoNiS",
"HKS": "HaKuSp",
"GNUVWD": "GNUVWDi",
"AS": "StaPla",
"AU": "Uni"
"Haupt": "HA",
"Kerngebiet": "KGA",
"RaLNS": "RaLoNiS",
"HKS": "HaKuS",
"GNUVWD": "GNUVWDi",
"SAIBGGSG": "SR",
"AS": "StaPla",
"AU": "Uni"
},
"Hamburg-Mitte": {
"Haupt": "HA",
"Stadtplanungs": "StaPla"
},
"Hamburg-Nord": {
"Haupt": "HA",
},
"Harburg": {
"Haupt": "HA",
},
"Wandsbek": {
"Haupt": "HA",
}
}
BASE_LINKS = {
"Altona": "https://sitzungsdienst-altona.hamburg.de/ri",
"Bergedorf": "https://sitzungsdienst-bergedorf.hamburg.de/ri",
"Eimsbüttel": "https://sitzungsdienst-eimsbuettel.hamburg.de/ri",
"Hamburg-Mitte": "https://sitzungsdienst-hamburg-mitte.hamburg.de/ri",
"Hamburg-Nord": "https://sitzungsdienst-hamburg-nord.hamburg.de/ri",
"Harburg": "https://sitzungsdienst-harburg.hamburg.de/ri",
"Wandsbek": "https://sitzungsdienst-wandsbek.hamburg.de/ri",
}

51
src/twomartens/allrisscraper/main.py
View File

@ -35,9 +35,10 @@ ALLRIS_OPEN: str = "https://2martens.de/bezirk-eimsbüttel"
user_agent = 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/78.0.3112.50 Safari/537.36'
_CONFIG_PROPS = {
"Default": {
"username": "",
"password": "",
"pdflocation": ""
"district": "Eimsbüttel",
"username": "max.mustermann@eimsbuettel.de",
"password": "SehrSicheresPasswort",
"pdflocation": "/Pfad/zum/Ablegen/der/PDFs/"
}
}
@ -62,9 +63,11 @@ def main() -> None:
config = configparser.ConfigParser()
config.read(config_file)
district = config["Default"]["district"]
username = config["Default"]["username"]
password = config["Default"]["password"]
pdf_location = config["Default"]["pdflocation"]
base_url = definitions.BASE_LINKS[district]
options = Options()
options.headless = True
@ -74,9 +77,9 @@ def main() -> None:
driver.get(ALLRIS_LOGIN)
login(driver, username=username, password=password)
driver.get("https://gateway.hamburg.de/HamburgGateway/Service/StartService/113")
driver.get("https://sitzungsdienst-eimsbuettel.hamburg.de/ri/si012.asp")
driver.get(f"{base_url}/ri/si012.asp")
meetings = get_meetings(driver)
download_documents(driver, meetings, pdf_location)
download_documents(driver, meetings, pdf_location, base_url, district)
driver.close()
@ -105,10 +108,11 @@ def get_meetings(driver: webdriver.WebDriver) -> List[meeting.Meeting]:
return meetings
def download_documents(driver: webdriver.WebDriver, meetings: List[meeting.Meeting], pdf_location: str) -> None:
base_link = "https://sitzungsdienst-eimsbuettel.hamburg.de/ri/do027.asp"
for meeting in meetings:
driver.get(meeting.link)
def download_documents(driver: webdriver.WebDriver, meetings: List[meeting.Meeting],
pdf_location: str, base_url: str, district: str) -> None:
base_link = f"{base_url}/do027.asp"
for _meeting in meetings:
driver.get(_meeting.link)
td = driver.find_element(By.XPATH, "//table[@class='tk1']//td[@class='me1']")
form_elements = td.find_elements_by_tag_name("form")
agenda_item = form_elements[0]
@ -118,11 +122,15 @@ def download_documents(driver: webdriver.WebDriver, meetings: List[meeting.Meeti
invitation_item = form_elements[2]
invitation_link = f"{base_link}?DOLFDNR={invitation_item.find_element_by_name('DOLFDNR').get_property('value')}&options=64"
driver.get(agenda_link)
save_pdf(driver.current_url, f"{pdf_location}{meeting.date.isoformat()}_{get_abbreviated_committee_name(meeting.name)}/Tagesordnung.pdf")
save_pdf(driver.current_url, f"{get_formatted_filename(pdf_location, _meeting, district)}/Tagesordnung.pdf")
driver.get(total_link)
save_pdf(driver.current_url, f"{pdf_location}{meeting.date.isoformat()}_{get_abbreviated_committee_name(meeting.name)}/Mappe.pdf")
save_pdf(driver.current_url, f"{get_formatted_filename(pdf_location, _meeting, district)}/Mappe.pdf")
driver.get(invitation_link)
save_pdf(driver.current_url, f"{pdf_location}{meeting.date.isoformat()}_{get_abbreviated_committee_name(meeting.name)}/Einladung.pdf")
save_pdf(driver.current_url, f"{get_formatted_filename(pdf_location, _meeting, district)}/Einladung.pdf")
def get_formatted_filename(pdf_location: str, meeting_obj: meeting.Meeting, district: str) -> str:
return f"{pdf_location}{meeting_obj.date.isoformat()}_{get_abbreviated_committee_name(meeting_obj.name, district)}"
def save_pdf(url: str, dest: str) -> None:
@ -138,29 +146,34 @@ def get_day(date_str: str) -> date:
return date(int(date_elements[-1]), int(date_elements[-2]), int(date_elements[-3]))
def get_abbreviated_committee_name(name: str) -> str:
def get_abbreviated_committee_name(name: str, district: str) -> str:
start_committee = "Sitzung des Ausschusses"
start_regional_committee = "Sitzung des Regionalausschusses"
start_plenary = "Sitzung der Bezirksversammlung"
start_youth_help_committee = "Sitzung des Jugendhilfeausschusses"
start_other_committee = "Sitzung des"
end_other_committee = "ausschusses"
abbreviated_name = ""
if start_plenary in name:
if name.startswith(start_plenary):
abbreviated_name = "BV"
elif start_committee in name:
elif name.startswith(start_committee):
second_part = name[len(start_committee):]
second_split = second_part.split(sep=",")
abbreviated_name = get_abbreviation(second_split)
if len(abbreviated_name) == 1:
abbreviated_name = f"A{abbreviated_name}"
elif start_regional_committee in name:
elif name.startswith(start_regional_committee):
second_part = name[len(start_regional_committee):]
second_split = second_part.split(sep="/")
abbreviated_name = f"Ra{get_abbreviation(second_split)}"
elif start_youth_help_committee in name:
elif name.startswith(start_youth_help_committee):
abbreviated_name = "JHA"
elif name.startswith(start_other_committee) and name.endswith(end_other_committee):
core_name = name[len(start_other_committee):-len(end_other_committee)]
abbreviated_name = core_name
if abbreviated_name in definitions.ABBREVIATIONS["Eimsbüttel"]:
abbreviated_name = definitions.ABBREVIATIONS["Eimsbüttel"][abbreviated_name]
if abbreviated_name in definitions.ABBREVIATIONS[district]:
abbreviated_name = definitions.ABBREVIATIONS[district][abbreviated_name]
return abbreviated_name