pwsp.yaml

---
sources:
  - source: pubmed_api
    target_source: pubmed
    disabled: false
    urls:
      - url: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/entrez/eutils/esearch.fcgi
        params:
          db: pubmed
          term: $q
          retmode: json
          sort: relevance
          retmax: $max
          usehistory: y
      - url: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/entrez/eutils/efetch.fcgi
        params:
          db: pubmed
          retmode: xml
          retmax: $max
          webenv: "eval(json):json['esearchresult']['webenv']"
          query_key: "eval(json):json['esearchresult']['querykey']"
    parse:
      - item: "pubmedarticleset > pubmedarticle"
        id: "first():pubmedarticle > medlinecitation > pmid"
        title: "first():pubmedarticle > medlinecitation > article > articletitle"
        abstract: "first():pubmedarticle > medlinecitation > article > abstract"
        link: 'first():pubmedarticle > medlinecitation > pmid|re.sub(r"^", "https://pubmed.ncbi.nlm.nih.gov/", $)'
        details: "first():pubmedarticle > medlinecitation > article > articledate"

  - source: pmc_api
    target_source: pubmed
    disabled: false
    urls:
      - url: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/entrez/eutils/esearch.fcgi
        params:
          db: pmc
          term: $q
          retmode: json
          sort: relevance
          retmax: $max
          usehistory: y
      - url: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/entrez/eutils/efetch.fcgi
        params:
          db: pmc
          retmode: xml
          retmax: $max
          webenv: "eval(json):json['esearchresult']['webenv']"
          query_key: "eval(json):json['esearchresult']['querykey']"
    parse:
      - item: "pmc-articleset > article"
        id: "first():article > front > article-meta > article-id[pub-id-type='pmid']"
        title: "first():article > front > article-meta > title-group > article-title"
        abstract: "first():article > front > article-meta > abstract"
        link: 'first():article > front > article-meta > article-id[pub-id-type="pmc"]|re.sub(r"^", "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC", $)'
        details: "first():article > front > article-meta > pub-date[pub-type='pmc-release']"

  - source: clinvar_api
    target_source: pubmed
    disabled: false
    urls:
      - url: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/entrez/eutils/esearch.fcgi
        params:
          db: clinvar
          term: $q
          retmode: json
          sort: relevance
          retmax: $max
          usehistory: y
      - url: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/entrez/eutils/elink.fcgi
        params:
          db: pubmed
          dbfrom: clinvar
          retmode: json
          retmax: $max
          webenv: "eval(json):json['esearchresult']['webenv']"
          query_key: "eval(json):json['esearchresult']['querykey']"
      - url: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/entrez/eutils/efetch.fcgi
        params:
          db: pubmed
          retmode: xml
          retmax: $max
          id: "eval(json):json['linksets'][0]['linksetdbs'][0]['links']"
    parse:
      - item: "pubmedarticleset > pubmedarticle"
        id: "first():pubmedarticle > medlinecitation > pmid"
        title: "first():pubmedarticle > medlinecitation > article > articletitle"
        abstract: "first():pubmedarticle > medlinecitation > article > abstract"
        link: 'first():pubmedarticle > medlinecitation > pmid|re.sub(r"^", "https://pubmed.ncbi.nlm.nih.gov/", $)'
        details: "first():pubmedarticle > medlinecitation > article > articledate"

  - source: pubmed
    target_source: pubmed
    disabled: false
    url: https://pubmed.ncbi.nlm.nih.gov/
    params:
      term: $q
      size: $max
      sort: relevance
    parse:
      - item: "#search-results > section > div.search-results-chunks > div > article:nth-child(n + 2) > div.docsum-wrap > div.docsum-content"
        id: "first():div.docsum-content > div.docsum-citation.full-citation > span:nth-child(5) > span"
        title: "first():div.docsum-content > a"
        abstract: "first():div.docsum-content > div.docsum-snippet > div.full-view-snippet"
        link: 'attr(href):div.docsum-content > a|re.sub(r"^", "https://pubmed.ncbi.nlm.nih.gov", $)'
        details: "first():div.docsum-content > div.docsum-citation.full-citation > span.docsum-journal-citation.full-journal-citation"
      - item: "#article-details"
        id: "first():#full-view-identifiers > li:nth-child(1) > span > strong"
        title: "first():#full-view-heading > h1"
        abstract: "first():#enc-abstract"
        link: 'first():#full-view-identifiers > li:nth-child(1) > span > strong|re.sub(r"^", "https://pubmed.ncbi.nlm.nih.gov/", $)'
        details: "first():#full-view-heading > div.article-citation > div.article-source > span.cit"

  - source: pmc
    target_source: pubmed
    disabled: true
    url: https://www.ncbi.nlm.nih.gov/pmc/
    params:
      term: $q
    parse:
      - item: "#maincontent > div > div:nth-child(5) > div:nth-child(n) > div.rslt"
        id: "first():div.rslt > div:nth-child(4) > div > dl > dd"
        title: "first():div.rslt > div.title > a"
        abstract: null
        link: 'attr(href):div.rslt > div.title > a|re.sub(r"^", "https://www.ncbi.nlm.nih.gov", $)'
        details: "first():div.rslt > div.supp > div.details"

  - source: google scholar
    disabled: false
    url: https://scholar.google.com/scholar
    # query_builder: google_scholar
    params:
      q: $q
      hl: en
    parse:
      - item: "#gs_res_ccl_mid > div:nth-child(n) > div.gs_ri"
        id: "attr(href):div.gs_ri > h3 > a"
        title: "first():div.gs_ri > h3 > a"
        abstract: "first():div.gs_ri > div.gs_rs"
        link: "attr(href):div.gs_ri > h3 > a"
        details: "first():div.gs_ri > div.gs_a"