Question

I’m working on search for my blog. Anyway, I’m having some problems with implementing it.

I’m having two models:

class Article(models.Model):
    title = models.CharField(max_length=255)
    content = models.TextField()

class Tag(models.Model):
    article = models.ForeignKey(Article)
    content = models.CharField(max_length=255)

Actual search has two fields: phrase and tags. Phrase should look in Article.title and Article.content, but tag should look for articles that have Tag object by Tag.content.

I’m having a test too!

def test_by_phrase_and_tags(self):
    article_content = "spam"
    tag_content1 = "eggs"
    tag_content2 = "cheese"
    article1 = test_utilities.create_article(content=article_content)
    article2 = test_utilities.create_article(content=article_content)
    test_utilities.create_tag(article1, content=tag_content1)
    test_utilities.create_tag(article2, content=tag_content1)
    test_utilities.create_tag(article2, content=tag_content2)
    response = self.client.get(reverse("blog_search"), {
        "phrase": article_content,
        "tags": "{}, {}".format(tag_content1, tag_content2)
    })
    found_articles = response.context[-1]["found_articles"]
    self.assertEqual(found_articles, [article2])

It creates two articles with equal content, equal tag that’s set for both articles and unique tag that’s set only for 2nd article.

Then I request articles that have the content (both articles should match) and both tags (only 2nd article should match). Overall, I’m asserting that only 2nd article is returned.

I have tried a lot of things with raw SQL and Django ORM, but none seems to work.

With a sub-query:

SELECT * FROM blog_article
WHERE blog_article.content = "spam"
AND blog_article.id IN (
    SELECT blog_tag.article_id FROM blog_tag
    WHERE blog_tag.content = "eggs"
    OR blog_tag.content = "cheese"
);

With a join:

SELECT * FROM blog_article
JOIN blog_tag
ON blog_article.id = blog_tag.article_id
WHERE blog_article.content = "spam"
AND blog_tag.content = "eggs"
AND blog_tag.content = "cheese";

The same and other things with ORM…

So how can I get articles that have spam in title or content and have tags? I’m stumped.

Answer 1

If i understand correctly you want to filter the articles by article.title or article.content with the phrase or with tags:

First define a related_name for article in tags model:

class Tag(models.Model):
    article = models.ForeignKey(Article, related_name='tags')

Now query:

from django.db.models import Q

# supposed input
phrase = 'my search string'
tags = ['tag1', 'tag2']

articles = Article.objects.filter(
   Q(title__icontains=phrase) |
   Q(content__icontains=phrase)) \
   .filter(tags__content__in=tags) \
   .annotate(num_tags=Count('tags')) \
   .filter(num_tags=len(tags))