matitos_news

conda create -n matitos_urls python=3.12
conda activate matitos_urls
# Core
pip install django==5.1 psycopg[binary] django-redis django-tasks-scheduler
# Fetcher
pip install feedparser python-dateutil newspaper4k[all] lxml[html_clean] googlenewsdecoder gnews duckduckgo_search GoogleNews langdetect
# News visualization
pip install ollama

Database
- Database initialization -> 1-DB.ipynb
From automated inspectdb

# 1) Inspect DB, generate models.py
python manage.py inspectdb

# 2) Modify models.py

# URLS:
class Urls(models.Model):
    class STATUS_ENUM(models.TextChoices):
        RAW = "raw", "Raw"
        ERROR = "error", "Error"
        VALID = "valid", "Valid"
        UNKNOWN = "unknown", "Unknown"
        INVALID = "invalid", "Invalid"
        DUPLICATE = "duplicate", "Duplicate"

    url = models.TextField(unique=True)
    ts_fetch = models.DateTimeField(auto_now_add=True)
    status = models.TextField(choices=STATUS_ENUM.choices, default=STATUS_ENUM.RAW)  # This field type is a guess.

    class Meta:
        managed = False
        db_table = 'urls'
        ordering = ["-ts_fetch"]

# SEARCH:
class Search(models.Model):
    class TYPE_ENUM(models.TextChoices):
        RSS_FEED = "rss_feed", "RSS_Feed"
        KEYWORD_SEARCH = "keyword_search", "Keyword_Search"
        URL_HOST = "url_host", "URL_Host"

    id = models.SmallAutoField(primary_key=True)
    search = models.TextField(unique=True)
    type = models.TextField(choices=TYPE_ENUM.choices)  # This field type is a guess.

# URL_CONTENT:
class UrlContent(models.Model):
    id_url = models.OneToOneField('Urls', models.DO_NOTHING, db_column='id_url', primary_key=True)
    date_published = models.DateTimeField(blank=True, null=True)
    title = models.TextField(blank=True, null=True)
    description = models.TextField(blank=True, null=True)
    content = models.TextField(blank=True, null=True)
    valid_content = models.BooleanField(blank=True, null=True)
    language = models.CharField(max_length=2, blank=True, null=True)
    keywords = ArrayField(models.TextField(blank=True, null=True))  # This field type is a guess.
    tags = ArrayField(models.TextField(blank=True, null=True))  # This field type is a guess.
    authors = ArrayField(models.TextField(blank=True, null=True))  # This field type is a guess.
    image_main_url = models.TextField(blank=True, null=True)
    images_url = ArrayField(models.TextField(blank=True, null=True))  # This field type is a guess.
    videos_url = ArrayField(models.TextField(blank=True, null=True))  # This field type is a guess.
    url_host = models.TextField(blank=True, null=True)
    site_name = models.TextField(blank=True, null=True)

# TODO: Associate db_table name with a prefix on project_name
class Meta:
    db_table = 'urls' # db_table = '{}_urls'.format(project_name)

Environment variables

# Database
DB_NAME=${DB_NAME:-matitos}
DB_USER=${DB_NAME:-supermatitos}
DB_PASSWORD=${DB_NAME:-supermatitos}
DB_HOST=${DB_NAME:-localhost}
DB_PORT=${DB_NAME:-5432}
REDIS_HOST=${REDIS_HOST:-localhost}
REDIS_PORT=${REDIS_PORT:-6379}

# Job timeout: 30 min
JOB_DEFAULT_TIMEOUT=${RQ_DEFAULT_TIMEOUT:-1800}

# Logs path
PATH_LOGS_PARAMETERIZATION="logs/log_app_fetcher_{}.log"

# Fetcher
FETCHER_GNEWS_DECODE_SLEEP=2
FETCHER_GOOGLE_GENERAL_PAGE_ITER_SLEEP=4
FETCHER_BETWEEN_SEARCHES_SLEEP=5
FETCHER_URL_HOST_SLEEP=5
FETCHER_LANGUAGE_DETECTION_MIN_CHAR=100

SELENIUM_ENDPOINT="http://selenium_app:80"

Deploy

# Migrations
python manage.py makemigrations api; python manage.py migrate --fake-initial
# Create user
python manage.py createsuperuser

# 1) Server
python manage.py runserver

# 2) Workers
python manage.py rqworker high default low

# Visualize DB
http://localhost:8080/?pgsql=matitos_db&username=supermatitos&db=matitos&ns=public&select=urls&order%5B0%5D=id

Scheduled tasks

# Import tasks
python manage.py import --filename scheduled_tasks.json

# Modify using the admin panel, then save 
# python manage.py export > scheduled_tasks.json

Utils. TODO: To endpoint...

python manage.py rqstats