Skip to content

Latest commit

 

History

History
228 lines (143 loc) · 8.19 KB

README.md

File metadata and controls

228 lines (143 loc) · 8.19 KB
Raw

news_fetcher

Script to fetch news URLs from news websites to database using API.

Installation

Install Python 3.8 or higher, install poetry, run poetry install --no-dev.

Then you can just run poetry run COMMAND to run specific commands under python virtual environment created by poetry.

Or you can enter poetry shell (by running poetry shell) and then type script commands.

You can also use pip.

Installation example

Assuming Python 3.8 or higher and poetry are installed.

Initialise and update virtual environment (assuming you are in the folder with this README file):

poetry install --no-dev

Run script:

poetry run python news_fetcher/news_fetcher.py --help

Windows installation example

Assuming Python 3.8 or higher is installed.

Install poetry (in Windows PowerShell):

(Invoke-WebRequest -Uri https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py -UseBasicParsing).Content | python

You may need to restart PowerShell or reboot your computer.

To install or update libraries, run batch file update.bat.

Run script:

poetry run python news_fetcher/news_fetcher.py --help

Files

  • run_all.sh is the Shell script for running all steps. It requires that environment variables are set in .env file: MEDIAWIKI_CREDENTIALS, DATABASE_URL, WIKI_TOOL_DIRECTORY, DATA_FILE, SOURCE_PATH, SOURCE_NAME, TARGET_API_URL, WIKI_PREFIX, BOT_NAME, REQUESTS_INTERVAL.
  • news_fetcher/news_fetcher.py is the script entry point.
  • news_fetcher/db.py is the DB initialization module.
  • news_fetcher/models.py is the module with DB models.
  • news_fetcher/module.py is the module with base class for "source modules" which are used to grab news from different sources.

Prostoprosport source module

This modules fetches news using prostoprosport.ru API.

  • news_fetcher/prostoprosport.py is the source module.
  • data/categories_from_js.json is a category URL data grabbed from JS.
  • data/categories_bonus.json is an additional category URL data grabbed from RSS.

RSS source module

This modules fetches news using RSS.

  • news_fetcher/rss.py is the source module.

DB models

Source

Source website.

  • slug_name — string website ID (primary key), for example: birmingham-post.

Tag

Tag for news articles.

  • tag_id — numerical ID (primary key).
  • title — tag text, for example: Sport (must be unique).

Article

News article from source website.

  • article_id — numerical ID (primary key).
  • source — source website (foreign key).
  • slug_name — string identifier (must be unique per source website), for example: sir-stanley-matthews-1915-2000-a-potteries-hero.
  • title — human-readable article title, for example: Sir Stanley Matthews 1915-2000: A Potteries hero; Stanley stayed loyal to his beloved.
  • date — publication date, for example: 2020-02-24T00:00:00.
  • source_url — full article URL, for example: https://www.thefreelibrary.com/Sir+Stanley+Matthews+1915-2000%3A+A+Potteries+hero%3B+Stanley+stayed...-a060517953.
  • source_url_ok — true if URL can be retrieved, false if it can not, null if it was not checked yet.
  • author_name — human-readable author name, may be null.
  • wikitext_paragraphs — article content converted into wiki-text stored as JSON list of paragraphs, may be null if not fetched yet.
  • misc_data — miscellaneous data stored as JSON, specific format and structure is module-dependent.
  • tags — article tags (many-to-many relation with Tag model through technical ArticleTag model with table named article_m2m_tag).

Usage

Getting help

--help Show help message and exit. If this option is used with command, then help message for that specific command will be printed.

Common options

  • --source-module TEXT (required) — source module name, can be prostoprosport or rss

Prostoprosport module options

  • --data-file FILENAME — file with categories data (can be built using process-categories command)
  • --source-path TEXT — API method name, can be news or main_news

RSS module options

  • --data-file FILENAME — JSON file with configuration, should contain folllowing keys:
    • css_selector — CSS selector for article paragraphs on web page
    • source_title — source title
    • source_template_name — template name for generated wiki-pages (optional, source title is used by default)
    • removed_last_lines — count of paragraphs at the end of article that should be skipped (optional, 0 by default)
    • disable_bold_font — true to avoid bold font in generated page (optional, false by default)
    • extra_first_lines — array of strings to add at the beginning of generated page (optional, empty by default)
  • --source-name (required) — source slug name (identifier) for DB
  • --source-path TEXT (required) — RSS feed URL

Command fetch-news

Fetch news for page range and write data to DB. Pages are numbered from most recent (1) to least recent. Note that page numbers are now used in Prostoprosport source module only.

Options

  • --first-page INTEGER — number of first page to load, should not be less than 1
  • --last-page INTEGER — number of last page to load, should not be less than 1. If it is less than first page number, no data will be fetched

Example 1

Fetch most recent page (1):

python news_fetcher/prostoprosport_news_fetcher.py fetch-news

Example 2

Fetch pages 5 most recent pages (5 to 1):

python news_fetcher/prostoprosport_news_fetcher.py fetch-news --last-page 5

Example 3

Fetch pages 11 to 20:

python news_fetcher/prostoprosport_news_fetcher.py fetch-news --first-page 11 --last-page 20

Notes

  • (OBSOLETE) Prostoprosport.ru API did not provide URLs, only category slugs and IDs, category-to-URL mappings are grabbed from JavaScript on website. Therefore URLs were not guaranteed to be correct.
  • Now all news are placed under /post/ URL path, without category URL.

Command fetch-news-pages

Fetch news pages contents for pages which were:

  1. From current source
  2. Not marked as "invalid URL" during previous fetch
  3. Not already fetched

Example

python news_fetcher/prostoprosport_news_fetcher.py fetch-news-pages

Command generate-wiki-pages

Generate MediaWiki pages as text files for fetched news pages not marked as uploaded.

Options

  • --output-file FILE — output JSON file with list of generated pages, it contains dictionary, where keys are page titles, and values are page file paths
  • --output-directory FILE — directory to place generated MediaWiki page files
  • --bot-name STRING — name of bot user account to use in page template

Example

python news_fetcher/prostoprosport_news_fetcher.py generate-wiki-pages --output-file ../data/pages.json --output-directory ../data/pages/

Command mark-uploaded-pages

Mark news articles as uploaded in database.

Options

  • --input-file FILE input JSON file generated by generate-wiki-pages command

Example

python news_fetcher/prostoprosport_news_fetcher.py mark-uploaded-pages --input-file ../data/pages.json

Module-specific command process-categories in prostoprosport source module

Build categories mapping file. It will contain data about base URL for category slugs and IDs. For example, category rpl have base URL (without leading slash) football/russia/rpl.

Options

  • --input-from-js-file FILE — JSON file with categories data grabbed from JavaScript, default is data/categories_from_js.json
  • --input-bonus-file FILE — JSON file with additional data, default is data/categories_bonus.json
  • --input-colors-file FILE — JSON file with ID-to-color mapping data grabbed from JavaScript, default is data/category_colors.json
  • --output-file FILE — output JSON file, default is data1/categories_data.json

Example

python news_fetcher/prostoprosport.py process-categories