X-Git-Url: https://git.xandkar.net/?p=covid-19-scrapers.git;a=blobdiff_plain;f=fetch-case-count-usa-ny;h=e346935c14551c49381e5b8679d448a028e0a65f;hp=95a039f1f12a9909288c28c721a5f4d05cc026b7;hb=d5c6a8b1018d19ec0f3d4ab8c7f57f17aa62e8d1;hpb=0b60ba947fe0490f018137cbb9c51b01bdcb8b39 diff --git a/fetch-case-count-usa-ny b/fetch-case-count-usa-ny index 95a039f..e346935 100755 --- a/fetch-case-count-usa-ny +++ b/fetch-case-count-usa-ny @@ -5,9 +5,26 @@ # - awk # - hxpipe (packaged in html-xml-utils on Debian and Ubuntu) +# 2020-03-16 morning-afternoon +#URI='https://health.ny.gov/diseases/communicable/coronavirus/' +#TARGET_PATH='/html/body/div/div/div/div/div/div/table/tr/td' +#DEFAULT_LOCATION='New York City:' + +# 2020-03-16 late evening 22:30 +#URI='https://health.ny.gov/diseases/communicable/coronavirus/' +#TARGET_PATH='/html/body/div/div/div/div/div/div/tr/td' +#DEFAULT_LOCATION='New York City:' + +# 2020-03-17 late afternoon 16:35 +URI='https://coronavirus.health.ny.gov/county-county-breakdown-positive-cases' +TARGET_PATH='/html/body/div/div/main/div/div/div/div/div/div/div/div/table/tbody/tr/td' +DEFAULT_LOCATION='New York City' + VALID_LOCATIONS=' Albany + Allegany Broome + Clinton Delaware Dutchess Erie @@ -16,21 +33,25 @@ VALID_LOCATIONS=' Monroe Montgomery Nassau + New York City + New York State (Outside of NYC) + Onondaga + Ontario Orange Putnam + Rensselaer Rockland Saratoga Schenectady Suffolk + Sullivan Tioga Tompkins + Total Positive Cases (Statewide) Ulster Westchester - New York State (Outside of NYC) - New York City: - Total Positive Cases (Statewide) + Wyoming ' -DEFAULT_LOCATION='New York City:' usage() { printf "Usage: %s [LOCATION]\n" "$0" @@ -47,9 +68,12 @@ case "$1" in * ) location="$1";; esac -curl 'https://health.ny.gov/diseases/communicable/coronavirus/' \ +curl "$URI" \ | hxpipe \ -| awk -v location="$location" ' +| awk \ + -v location="$location" \ + -v target_path="$TARGET_PATH" \ + ' /^[\(\)]/ { update_node() next @@ -64,12 +88,12 @@ curl 'https://health.ny.gov/diseases/communicable/coronavirus/' \ XmlPayload = substr($0, 2, length($0)) } - XmlPath == "/html/body/div/div/div/div/div/div/table/tr/td" && XmlPayload == location { + XmlPath == target_path && XmlPayload == location { found = 1; next } - XmlPath == "/html/body/div/div/div/div/div/div/table/tr/td" && found { + XmlPath == target_path && found { print XmlPayload; found = 0; next;