OK, I broke through I think. I got the output that was being sent to
elasticsearch. It was:
{u'query': {u'filtered': {u'filter': {u'terms': {u'django_ct':
[u'lenspro.product']}}, u'query': {u'query_string':
{u'fuzzy_max_expansions': 50, u'auto_generate_phrase_queries': True,
u'default_operator': u'AND', u'analyze_wildcard': True, u'query':
u'(text:(Canon C100) OR brand_name:(Canon AND C100))', u'default_field':
u'text', u'fuzzy_min_sim': 0.5}}}}, u'from': 0, u'size': 10}
And that brand name is obviously wrong because nothing has C100 in the
brand name, so that boost wasn't working. When I manually remove that from
the search terms I get some meaningful differences:
u'query': {u'filtered': {u'filter': {u'terms': {u'django_ct':
[u'lenspro.product']}}, u'query': {u'query_string':
{u'fuzzy_max_expansions': 50, u'auto_generate_phrase_queries': True,
u'default_operator': u'AND', u'analyze_wildcard': True, u'query':
u'(text:(Canon C100) OR brand_name:(Canon))', u'default_field': u'text',
u'fuzzy_min_sim': 0.5}}}}, u'from': 0, u'size': 10}
1.0 10.491137 Canon : Canon CA-930 Battery Charger for C100/C100II/C300/C500
1.0 10.025042 Canon : Canon BP-970G Battery Pack for C100/C100II/C300/C500
2.0 9.703276 Canon : Canon C100 Mark II Digital Cinema Camera EF Mount
1.5 9.525196 Canon : Canon 100 f/2.8L IS Macro
1.5 3.8322778 Canon : Canon 17-55 f/2.8 IS
1.5 3.8322778 Canon : Canon 24-105 f/4L IS
But the boost that I am adding via SearchIndex `def prepare` doesn't have
ANY affect on the results. Which I hope I can do because you see from the
above output I have a 1.0, 2.0, and 1.5 boost on specific indexes that I
need.
On Thu, Sep 1, 2016 at 2:49 PM, John Obelenus <jobele...@gmail.com> wrote:
> I've just updated to Django Haystack 2.5.0 on Django 1.9.6 and am still
> seeing this problem. Whether I defined boost or not on the brand name
> field, or through the prepare method the `score` attribute on the search
> results never change.
>
> I see the boost coming through on the `prepared_results` dictionary that
> gets sent through the elasticsearch `builk`. And I see the `brand_name`
> field boost in elasticsearch as well:
>
> curl 'localhost:9200/haystack/_mapping/modelresult/field/brand_name/'
>
>
> {"haystack":{"mappings":{"modelresult":{"brand_name":{"
> full_name":"brand_name","mapping":{"brand_name":{"type"
> :"string","boost":1.75,"analyzer":"snowball"}}}}}}
>
>
> So what am I missing? Is this an AWS only issue? Is this an elasticsearch
> issue?
>
>
> On Tuesday, May 24, 2016 at 1:33:17 PM UTC-4, John Obelenus wrote:
>>
>> My project in on Django 1.9, and haystack doesnt have a compatible
>> release yet. I was working off master and then I realized that a bunch of
>> boost support for elasticsearch was being removed. So I backported my
>> project to 1.8 so I could use the 2.4.1 official release to see if I can
>> get Document and Field boosting working. It seems to me that neither are
>> working.
>>
>> Here is what I am doing:
>> from haystack.query import SearchQuerySet, SQ, AutoQuery
>> search_term = 'C100'
>> sqs = SearchQuerySet().models(Product).filter(SQ(text=AutoQuery(se
>> arch_term)) | SQ(brand_name=search_term) )
>> for s in sqs:
>> print s.boost, s.score, s.brand_name, ':', s.object.name
>>
>> And no matter how I define the indexes the score never changes. Here are
>> three definitions I am trying and you can see the output of the above code
>> in the class comment:
>> class ProductIndex(CelerySearchIndex, indexes.Indexable):
>> """
>> 1.5 0.84893364 Zacuto : Zacuto Recoil Shoulder Rig for C100, C100II & C300
>> 1.5 0.84868896 Zacuto : Zacuto Z-Finder Pro for Canon C100
>> 1.0 0.70724076 Zacuto : Zacuto Studio Baseplate with 12" Rods for C100/
>> C100II/C300/C500
>> 1.0 0.70021886 Canon : Canon CA-930 Battery Charger for C100/C100II/C300/
>> C500
>> 1.0 0.6494849 Anton Bauer : Anton Bauer QRC-CA940 Battery Plate for Canon
>> C500, C300 & C100
>> 1.0 0.6002867 Canon : Canon BP-970G Battery Pack for C100/C100II/C300/
>> C500
>> 2.0 0.5658623 Canon : Canon C100 Mark II Digital Cinema Camera EF Mount
>> 1.5 0.4332277 Canon : Canon 100 f/2.8L IS Macro
>> """
>> text = indexes.CharField(document=True, use_template=True)
>> product_id = indexes.IntegerField(model_attr='id', null=False)
>> brand_name = indexes.CharField(model_attr='brand__name', null=True,
>> weight=1.75)
>>
>>
>> def get_model(self):
>> return models.Product
>>
>>
>> def index_queryset(self, using=None):
>> return self.get_model().objects.filter(is_active=True)
>>
>>
>> def prepare(self, instance):
>> data = super(ProductIndex, self).prepare(instance)
>> data['boost'] = 1.5 # even weight everything
>> if instance.is_camera:
>> data['boost'] = 2.0 # weight cameras higher
>> elif instance.is_accessory:
>> data['boost'] = 1.0 # weight accessories lower
>> return data
>>
>>
>>
>>
>> class ProductIndex(CelerySearchIndex, indexes.Indexable):
>> """
>> None 0.84893364 Zacuto : Zacuto Recoil Shoulder Rig for C100, C100II &
>> C300
>> None 0.84868896 Zacuto : Zacuto Z-Finder Pro for Canon C100
>> None 0.70724076 Zacuto : Zacuto Studio Baseplate with 12" Rods for
>> C100/C100II/C300/C500
>> None 0.70021886 Canon : Canon CA-930 Battery Charger for
>> C100/C100II/C300/C500
>> None 0.6494849 Anton Bauer : Anton Bauer QRC-CA940 Battery Plate for
>> Canon C500, C300 & C100
>> None 0.6002867 Canon : Canon BP-970G Battery Pack for
>> C100/C100II/C300/C500
>> None 0.5658623 Canon : Canon C100 Mark II Digital Cinema Camera EF Mount
>> None 0.4332277 Canon : Canon 100 f/2.8L IS Macro
>> """
>> text = indexes.CharField(document=True, use_template=True)
>> product_id = indexes.IntegerField(model_attr='id', null=False)
>> brand_name = indexes.CharField(model_attr='brand__name', null=True,
>> weight=5.0)
>>
>>
>> def get_model(self):
>> return models.Product
>>
>>
>> def index_queryset(self, using=None):
>> return self.get_model().objects.filter(is_active=True)
>>
>>
>> class ProductIndex(CelerySearchIndex, indexes.Indexable):
>> """
>> None 0.84893364 Zacuto : Zacuto Recoil Shoulder Rig for C100, C100II &
>> C300
>> None 0.84868896 Zacuto : Zacuto Z-Finder Pro for Canon C100
>> None 0.70724076 Zacuto : Zacuto Studio Baseplate with 12" Rods for C100/
>> C100II/C300/C500
>> None 0.70021886 Canon : Canon CA-930 Battery Charger for C100/C100II/C300
>> /C500
>> None 0.6889441 Anton Bauer : Anton Bauer QRC-CA940 Battery Plate for
>> Canon C500, C300 & C100
>> None 0.6002867 Canon : Canon BP-970G Battery Pack for C100/C100II/C300/
>> C500
>> None 0.5658623 Canon : Canon C100 Mark II Digital Cinema Camera EF Mount
>> None 0.4332277 Canon : Canon 100 f/2.8L IS Macro
>> """
>> text = indexes.CharField(document=True, use_template=True)
>> product_id = indexes.IntegerField(model_attr='id', null=False)
>> brand_name = indexes.CharField(model_attr='brand__name', null=True)
>>
>>
>> def get_model(self):
>> return models.Product
>>
>>
>> def index_queryset(self, using=None):
>> return self.get_model().objects.filter(is_active=True)
>>
>>
>> The score never changes despite document boost, or field boost against
>> the non-boosted index. The haystack documentation says this stuff is
>> supported, and I'm following the documentation example of how to query
>> against boosted fields. Am I doing something entirely wrong? What else can
>> I try to get this working?
>>
>
--
John Obelenus @ Active Frequency
(617) 858-1704
--
You received this message because you are subscribed to the Google Groups
"django-haystack" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to django-haystack+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
