acs-pop.csv:
  landing_page: https://data.census.gov/cedsci/table?q=population&g=0200000US1,2,3,4&tid=ACSDT1Y2019.B01003&moe=false&hidePreview=true
  raw: ACSDT1Y2019.B01003_2020-11-10T165412.zip
  description: |
    ACS 2019 1-year Total Population table (B01003) for U.S. regions, no margin of error


acs-race.csv:
  landing_page: https://data.census.gov/cedsci/table?q=race&g=0200000US1,2,3,4&tid=ACSDT1Y2019.B02001&hidePreview=true
  raw: ACSDT1Y2019.B02001_2020-11-10T164137.zip
  description: |
    ACS 2019 1-year Race estimate for U.S. regions

acs-race-meta.csv:
  landing_page: https://data.census.gov/cedsci/table?q=race&g=0200000US1,2,3,4&tid=ACSDT1Y2019.B02001&hidePreview=true

babynames-single-year.csv:
  wrangling: |
    cp yob1907.txt babynames-single-year.csv

babynames.csv:
  wrangling: |
    csvstack *.txt -H --filenames | csvheader --AX 'year,name,sex,count' > /tmp/babynames.csv

    cat /tmp/babynames.csv \
      | xsv search '\b(?:Mary|James|Logan|Leslie|Hillary|Donald)\b' \
      > babynames-yob.csv

boardings.csv:
  landing_page: https://github.com/storydrivendatasets/faa_airport_boardings
  direct: https://raw.githubusercontent.com/storydrivendatasets/faa_airport_boardings/master/data/wrangled/boardings.csv
  wrangling: |
    csvgrep -c 5 -r 'ATL|LAX|ORD|DFW|DEN|SFO|SEA|LAS' boardings.csv


cdph-env-inspections.csv:
  landing_page: https://data.cityofchicago.org/Environment-Sustainable-Development/CDPH-Environmental-Inspections/i9rk-duva
  direct: https://data.cityofchicago.org/api/views/i9rk-duva/rows.csv?accessType=DOWNLOAD
  gsheet: 'https://docs.google.com/spreadsheets/d/1GYe3XuUI_jsnWbMz17FtT6A5xZ8GPSV1IO6srhjy1v8/edit#gid=0'
  wrangling: |
    csvcut -c 'INSPECTION ID,ADDRESS,INSPECTION CATEGORY,INSPECTION SUB CATEGORY,INSPECTION DATE,NARRATIVE,LATITUDE,LONGITUDE,MODIFIED DATE'

census-econ-survey.csv:
  landing_page: https://data.census.gov/cedsci/table?q=selected&g=0100000US.04000.001&tid=ACSDP5Y2018.DP03&hidePreview=true

census-econ-survey-meta.csv:
  landing_page: https://data.census.gov/cedsci/table?q=selected&g=0100000US.04000.001&tid=ACSDP5Y2018.DP03&hidePreview=true


chicago-crimes.csv:
  gsheet: https://docs.google.com/spreadsheets/d/1KtDbUxo2Y-IqSl_bYUwyVSorJwhRh3KMgdQD-C0-XUk/edit#gid=0


surnames.csv:
  landing_page: https://www.census.gov/topics/population/genealogy/data/2010_surnames.html
  direct: https://www2.census.gov/topics/genealogy/2010surnames/names.zip
  wrangling:
    head -n 1000 Names_2010Census.csv


potus-tweets.csv:
  landing_page: https://twitter.com/POTUS/
  cmd: |
    twarc timeline POTUS --format csv > examples/real/potus-tweets.csv
  wrangle: |
    # a sample for testing csvnorm/csvflatten
    xsv search '\n' examples/real/potus-tweets.csv  | xsv sample 10 > examples/tweets-newlines.csv
  description: |
    Latest tweets from @realDonaldTrump as of 2020-11-10

    This data was generated using the twarc command-line tool, which converts the
    Twitter JSON response to a flat CSV file.

    https://github.com/DocNow/twarc#timeline

white-house-salaries.csv:
  landing_page: https://github.com/storydrivendatasets/white_house_salaries
  direct: https://raw.githubusercontent.com/storydrivendatasets/white_house_salaries/master/data/wrangled/white_house_salaries.csv
