ask_candid/base/retrieval/sources.py

from ask_candid.base.retrieval.schemas import ElasticSourceConfig


CandidBlogConfig = ElasticSourceConfig(
    index_name="search-semantic-candid-blog",
    semantic_fields=("content", "authors_text", "title_summary_tags")
)


CandidHelpConfig = ElasticSourceConfig(
    index_name="search-semantic-candid-help-elser_ve1",
    semantic_fields=("content", "combined_article_description")
)


CandidLearningConfig = ElasticSourceConfig(
    index_name="search-semantic-candid-learning_ve1",
    semantic_fields=("content", "title", "training_topics", "staff_recommendations")
)


CandidNewsConfig = ElasticSourceConfig(
    index_name="news_1",
    semantic_fields=("title", "content")
)


IssueLabConfig = ElasticSourceConfig(
    index_name="search-semantic-issuelab-elser_ve2",
    semantic_fields=("description", "content", "combined_issuelab_findings", "combined_item_description")
)


YoutubeConfig = ElasticSourceConfig(
    index_name="search-semantic-youtube",
    semantic_fields=("semantic_title", "semantic_description","semantic_cc_text"),
    text_fields=("title", "description", "cc_text"),
    highlight_fields=("semantic_cc_text",),
    excluded_fields=("cc_text", "semantic_cc_text", "semantic_title")
)