4e4e59fc-9df4-4838-9fd4-6c7ea33cdb7c

Committed 07 Apr 2025 06:36PM UTC coverage: 1.283% (-94.6%) from 95.862%

Build # 4e4e59fc-9df4-4838-9fd4-6c7ea33cdb7c

Build Type

Pull #1994

circleci

Committed by

hectorcorrea

Commit Message

Switched to use the autocomplete that we aleady use for ROR. Integrated it with the existing logic for creators

Pull Request Pull Request #1994: Started adding auto complete to contributors

Run Details

0 of 46 new or added lines in 2 files covered. (0.0%)

4806 existing lines in 74 files now uncovered.

65 of 5065 relevant lines covered (1.28%)

0.01 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0

/app/models/dataset_citation.rb

# frozen_string_literal: true

# Handles citations for datasets
# rubocop:disable Metrics/ParameterLists
# rubocop:disable Metrics/ClassLength
# rubocop:disable Style/NumericPredicate
# rubocop:disable Style/IfUnlessModifier
class DatasetCitation
  NEWLINE_INDENTED = "\r\n\t\t\t\t\t\t\t\t"

  # @param authors [<String>] Array of authors.
  # @param years [<String>] Array of years (expected 1 or 2 values).
  # @param title [String>] Title of the dataset
  # @param type [String] Type of the dataset (e.g. "Data set" or "Unpublished raw data")
  # @publisher [String] Publisher of the dataset
  # @doi [String] DOI URL
  def initialize(authors, years, title, type, publisher, doi)
    @authors = authors || []
    @years = years || []
    @title = title
    @type = type
    @publisher = publisher
    @doi = doi
  end

  def to_s(style)
    if style == "BibTeX"
      bibtex
    else
      apa
    end
  end

  # Returns a string with APA-ish citation for the dataset
  # Reference: https://libguides.usc.edu/APA7th/datasets#s-lg-box-22855503
  def apa
    apa_title = DatasetCitation.custom_strip(@title)
    apa_title += " [#{@type}]" if @type.present?
    apa_title = append_dot(apa_title)

    apa_publisher = append_dot(@publisher)
    apa_doi = @doi

    tokens = [append_dot(apa_author), append_dot(apa_year), apa_title, apa_publisher, apa_doi].compact_blank
    tokens.join(" ")
  rescue => ex
    Rails.logger.error "Error generating APA citation for (#{@title}): #{ex.message}"
    nil
  end

  # Returns a string with BibTeX citation for the dataset.
  #
  # There is no set standard for datasets and therefore the format we produce
  # was put together from a variety of sources including mimicking what Zotero
  # does and looking at examples from Zenodo (e.g. https://zenodo.org/record/6062882/export/hx#.Yiejad9OnUL)
  #
  # Notice that we use the @electronic{...} identifier instead of @dataset{...} since
  # Zotero does not recognize the later.
  def bibtex
    tokens = bibtex_checks

    text = "@electronic{#{bibtex_id},\r\n"
    text += tokens.map { |token| "\t#{token}" }.join(",\r\n") + "\r\n"
    text += "}"
    text
  rescue => ex
    Rails.logger.error "Error generating BibTex citation for (#{@title}): #{ex.message}"
    nil
  end

  # Returns an ID value for a BibTex citation
  def bibtex_id
    author_id = "unknown"
    if @authors.count > 0
      author_id = @authors.first.downcase.tr(" ", "_").gsub(/[^a-z0-9_]/, "")
    end
    year_id = @years.first&.to_s || "unknown"
    "#{author_id}_#{year_id}"
  end

  # Breaks a string into lines of at most max_length.
  # Returns an array with the lines.
  def bibtex_lines(string, max_length = 40)
    string = string.to_s # handles non-string values gracefully
    lines = []
    until string.nil?
      # TODO: it would be nice it we break on spaces rather than in the middle of a word.
      lines << string[0..max_length - 1]
      string = string[max_length..-1]
    end
    lines
  end

  # Creates a line to represent a BibTex field.
  # Breaks long lines into smaller lines.
  # Examples:
  #
  #     field_name = {{ short value }}
  #     field_name = {{ very very very
  #                 very very very very
  #                 long value }}
  #
  def bibtex_field(name, value, open_tag = "", close_tag = "")
    return nil if value.blank?

    value_trim = bibtex_lines(value).join(NEWLINE_INDENTED)
    name.ljust(12) + "= " + open_tag + value_trim + close_tag
  end

  # Creates a line to represent multiple authors in a BibTex field
  # https://en.wikibooks.org/wiki/LaTeX/Bibliography_Management#Authors
  #
  # Example:
  #
  #     author = { author1 and
  #              author2 and
  #              author3 }
  #
  def bibtex_field_author(name, authors, open_tag = "{", close_tag = "}")
    return nil if @authors.count <= 0

    value_trim = authors.join(" and #{NEWLINE_INDENTED}")
    name.ljust(12) + "= " + open_tag + value_trim + close_tag
  end

  # Appends a dot to a string if it does not end with one.
  def append_dot(value)
    return nil if value.nil?
    return "" if value.empty?
    DatasetCitation.custom_strip(value) + "."
  end

  # Strip a few specific characters that tend to mess up citations (e.g. trailing periods, commas, et cetera)
  def self.custom_strip(value)
    return nil if value.nil?
    return "" if value.empty?
    while true
      last_char = value[-1]
      break if last_char.nil? || !last_char.in?(". ,")
      value = value.chomp(last_char)
    end
    value
  end

  private

    def apa_author
      apa_author = ""
      case @authors.count
      when 0
        # do nothing
      when 1
        apa_author += @authors.first
      when 2
        apa_author += @authors.join(" & ")
      else
        apa_author += @authors[0..-2].join(", ") + ", & " + @authors[-1]
      end
      apa_author
    end

    def apa_year
      apa_year = ""
      case @years.count
      when 0
        # do nothing
      when 1
        apa_year += "(#{@years.first})"
      else
        apa_year += "(#{@years.join('-')})"
      end
      apa_year
    end

    def bibtex_checks
      tokens = []

      tokens << bibtex_field_author("author", @authors)

      tokens << bibtex_field("title", @title, "{{", "}}")

      tokens << bibtex_field("publisher", @publisher, "{{", "}}")

      tokens << bibtex_field("year", @years.first)

      tokens << bibtex_field("url", @doi, "{", "}")

      tokens.compact # remove nil values
    end
end
# rubocop:enable Metrics/ParameterLists
# rubocop:enable Metrics/ClassLength
# rubocop:enable Style/NumericPredicate
# rubocop:enable Style/IfUnlessModifier

1	# frozen_string_literal: true
2
3	# Handles citations for datasets
4	# rubocop:disable Metrics/ParameterLists
5	# rubocop:disable Metrics/ClassLength
6	# rubocop:disable Style/NumericPredicate
7	# rubocop:disable Style/IfUnlessModifier
UNCOV 8	class DatasetCitation	×
UNCOV 9	NEWLINE_INDENTED = "\r\n\t\t\t\t\t\t\t\t"	×
10
11	# @param authors [<String>] Array of authors.
12	# @param years [<String>] Array of years (expected 1 or 2 values).
13	# @param title [String>] Title of the dataset
14	# @param type [String] Type of the dataset (e.g. "Data set" or "Unpublished raw data")
15	# @publisher [String] Publisher of the dataset
16	# @doi [String] DOI URL
UNCOV 17	def initialize(authors, years, title, type, publisher, doi)	×
UNCOV 18	@authors = authors \|\| []	×
UNCOV 19	@years = years \|\| []	×
UNCOV 20	@title = title	×
UNCOV 21	@type = type	×
UNCOV 22	@publisher = publisher	×
UNCOV 23	@doi = doi	×
UNCOV 24	end	×
25
UNCOV 26	def to_s(style)	×
UNCOV 27	if style == "BibTeX"	×
UNCOV 28	bibtex	×
UNCOV 29	else	×
UNCOV 30	apa	×
UNCOV 31	end	×
UNCOV 32	end	×
33
34	# Returns a string with APA-ish citation for the dataset
35	# Reference: https://libguides.usc.edu/APA7th/datasets#s-lg-box-22855503
UNCOV 36	def apa	×
UNCOV 37	apa_title = DatasetCitation.custom_strip(@title)	×
UNCOV 38	apa_title += " [#{@type}]" if @type.present?	×
UNCOV 39	apa_title = append_dot(apa_title)	×
40
UNCOV 41	apa_publisher = append_dot(@publisher)	×
UNCOV 42	apa_doi = @doi	×
43
UNCOV 44	tokens = [append_dot(apa_author), append_dot(apa_year), apa_title, apa_publisher, apa_doi].compact_blank	×
UNCOV 45	tokens.join(" ")	×
UNCOV 46	rescue => ex	×
47	Rails.logger.error "Error generating APA citation for (#{@title}): #{ex.message}"	×
48	nil	×
UNCOV 49	end	×
50
51	# Returns a string with BibTeX citation for the dataset.
52	#
53	# There is no set standard for datasets and therefore the format we produce
54	# was put together from a variety of sources including mimicking what Zotero
55	# does and looking at examples from Zenodo (e.g. https://zenodo.org/record/6062882/export/hx#.Yiejad9OnUL)
56	#
57	# Notice that we use the @electronic{...} identifier instead of @dataset{...} since
58	# Zotero does not recognize the later.
UNCOV 59	def bibtex	×
UNCOV 60	tokens = bibtex_checks	×
61
UNCOV 62	text = "@electronic{#{bibtex_id},\r\n"	×
UNCOV 63	text += tokens.map { \|token\| "\t#{token}" }.join(",\r\n") + "\r\n"	×
UNCOV 64	text += "}"	×
UNCOV 65	text	×
UNCOV 66	rescue => ex	×
67	Rails.logger.error "Error generating BibTex citation for (#{@title}): #{ex.message}"	×
68	nil	×
UNCOV 69	end	×
70
71	# Returns an ID value for a BibTex citation
UNCOV 72	def bibtex_id	×
UNCOV 73	author_id = "unknown"	×
UNCOV 74	if @authors.count > 0	×
UNCOV 75	author_id = @authors.first.downcase.tr(" ", "_").gsub(/[^a-z0-9_]/, "")	×
UNCOV 76	end	×
UNCOV 77	year_id = @years.first&.to_s \|\| "unknown"	×
UNCOV 78	"#{author_id}_#{year_id}"	×
UNCOV 79	end	×
80
81	# Breaks a string into lines of at most max_length.
82	# Returns an array with the lines.
UNCOV 83	def bibtex_lines(string, max_length = 40)	×
UNCOV 84	string = string.to_s # handles non-string values gracefully	×
UNCOV 85	lines = []	×
UNCOV 86	until string.nil?	×
87	# TODO: it would be nice it we break on spaces rather than in the middle of a word.
UNCOV 88	lines << string[0..max_length - 1]	×
UNCOV 89	string = string[max_length..-1]	×
UNCOV 90	end	×
UNCOV 91	lines	×
UNCOV 92	end	×
93
94	# Creates a line to represent a BibTex field.
95	# Breaks long lines into smaller lines.
96	# Examples:
97	#
98	# field_name = {{ short value }}
99	# field_name = {{ very very very
100	# very very very very
101	# long value }}
102	#
UNCOV 103	def bibtex_field(name, value, open_tag = "", close_tag = "")	×
UNCOV 104	return nil if value.blank?	×
105
UNCOV 106	value_trim = bibtex_lines(value).join(NEWLINE_INDENTED)	×
UNCOV 107	name.ljust(12) + "= " + open_tag + value_trim + close_tag	×
UNCOV 108	end	×
109
110	# Creates a line to represent multiple authors in a BibTex field
111	# https://en.wikibooks.org/wiki/LaTeX/Bibliography_Management#Authors
112	#
113	# Example:
114	#
115	# author = { author1 and
116	# author2 and
117	# author3 }
118	#
UNCOV 119	def bibtex_field_author(name, authors, open_tag = "{", close_tag = "}")	×
UNCOV 120	return nil if @authors.count <= 0	×
121
UNCOV 122	value_trim = authors.join(" and #{NEWLINE_INDENTED}")	×
UNCOV 123	name.ljust(12) + "= " + open_tag + value_trim + close_tag	×
UNCOV 124	end	×
125
126	# Appends a dot to a string if it does not end with one.
UNCOV 127	def append_dot(value)	×
UNCOV 128	return nil if value.nil?	×
UNCOV 129	return "" if value.empty?	×
UNCOV 130	DatasetCitation.custom_strip(value) + "."	×
UNCOV 131	end	×
132
133	# Strip a few specific characters that tend to mess up citations (e.g. trailing periods, commas, et cetera)
UNCOV 134	def self.custom_strip(value)	×
UNCOV 135	return nil if value.nil?	×
UNCOV 136	return "" if value.empty?	×
UNCOV 137	while true	×
UNCOV 138	last_char = value[-1]	×
UNCOV 139	break if last_char.nil? \|\| !last_char.in?(". ,")	×
UNCOV 140	value = value.chomp(last_char)	×
UNCOV 141	end	×
UNCOV 142	value	×
UNCOV 143	end	×
144
UNCOV 145	private	×
146
UNCOV 147	def apa_author	×
UNCOV 148	apa_author = ""	×
UNCOV 149	case @authors.count	×
UNCOV 150	when 0	×
151	# do nothing
UNCOV 152	when 1	×
UNCOV 153	apa_author += @authors.first	×
UNCOV 154	when 2	×
UNCOV 155	apa_author += @authors.join(" & ")	×
UNCOV 156	else	×
UNCOV 157	apa_author += @authors[0..-2].join(", ") + ", & " + @authors[-1]	×
UNCOV 158	end	×
UNCOV 159	apa_author	×
UNCOV 160	end	×
161
UNCOV 162	def apa_year	×
UNCOV 163	apa_year = ""	×
UNCOV 164	case @years.count	×
UNCOV 165	when 0	×
166	# do nothing
UNCOV 167	when 1	×
UNCOV 168	apa_year += "(#{@years.first})"	×
UNCOV 169	else	×
UNCOV 170	apa_year += "(#{@years.join('-')})"	×
UNCOV 171	end	×
UNCOV 172	apa_year	×
UNCOV 173	end	×
174
UNCOV 175	def bibtex_checks	×
UNCOV 176	tokens = []	×
177
UNCOV 178	tokens << bibtex_field_author("author", @authors)	×
179
UNCOV 180	tokens << bibtex_field("title", @title, "{{", "}}")	×
181
UNCOV 182	tokens << bibtex_field("publisher", @publisher, "{{", "}}")	×
183
UNCOV 184	tokens << bibtex_field("year", @years.first)	×
185
UNCOV 186	tokens << bibtex_field("url", @doi, "{", "}")	×
187
UNCOV 188	tokens.compact # remove nil values	×
UNCOV 189	end	×
UNCOV 190	end	×
191	# rubocop:enable Metrics/ParameterLists
192	# rubocop:enable Metrics/ClassLength
193	# rubocop:enable Style/NumericPredicate
194	# rubocop:enable Style/IfUnlessModifier

pulibrary / pdc_describe / 4e4e59fc-9df4-4838-9fd4-6c7ea33cdb7c

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous