X-Git-Url: https://git.xandkar.net/?p=covid-19-scrapers.git;a=blobdiff_plain;f=fetch-case-count-usa-ny;h=16cf859969241d777c71a4ebb3a1932fb90953a6;hp=b3bd2515ce594e911679abb8cfe821947e30bf87;hb=a8677ce2ff31aed05e1adf8bc142b459f662b52a;hpb=cb4ba2007be59c08a5d0458ed629d20ffae69c1f

diff --git a/fetch-case-count-usa-ny b/fetch-case-count-usa-ny
index b3bd251..16cf859 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
@@ -16,25 +33,28 @@ VALID_LOCATIONS='
     Monroe
     Montgomery
     Nassau
+    New York City
+    Onondaga
+    Ontario
     Orange
     Putnam
+    Rensselaer
     Rockland
     Saratoga
     Schenectady
     Suffolk
+    Sullivan
     Tioga
     Tompkins
     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"
+        printf "Usage: %s [OPTION | LOCATION]\n" "$0"
         printf '\n'
+        printf 'OPTION = -h | --all \n'
         printf 'LOCATION ='
         printf '%s\n' "$VALID_LOCATIONS"
         printf "Default LOCATION:\n    %s\n" "$DEFAULT_LOCATION"
@@ -43,17 +63,18 @@ usage() {
 
 case "$1" in
     '-h') usage;;
-    ''  ) location="$DEFAULT_LOCATION";;
-     *  ) location="$1";;
+    '--all') opt_all=1;;
+    ''  ) target_location="$DEFAULT_LOCATION";;
+     *  ) target_location="$1";;
 esac
 
-curl 'https://health.ny.gov/diseases/communicable/coronavirus/' \
+curl "$URI" \
 | hxpipe \
-| awk -v location="$location" '
-    BEGIN {
-        target_path = "/html/body/div/div/div/div/div/div/table/tr/td"
-    }
-
+| awk \
+    -v target_location="$target_location" \
+    -v target_path="$TARGET_PATH" \
+    -v opt_all="$opt_all" \
+    '
     /^[\(\)]/ {
         update_node()
         next
@@ -68,17 +89,26 @@ curl 'https://health.ny.gov/diseases/communicable/coronavirus/' \
         XmlPayload = substr($0, 2, length($0))
     }
 
-    XmlPath == target_path && XmlPayload == location {
-        found = 1;
+    XmlPath == target_path && !location {
+        location = XmlPayload;
         next
     }
 
-    XmlPath == target_path && found {
-        print XmlPayload;
-        found = 0;
+    XmlPath == target_path && location {
+        case_counts[location] = XmlPayload;
+        location = "";
         next;
     }
 
+    END {
+        if (opt_all) {
+            for (location in case_counts)
+                printf("%d|%s\n", case_counts[location], location)
+        } else {
+                printf("%d\n", case_counts[target_location])
+        }
+    }
+
     function path_to_string(path, depth,    p, i) {
         p = ""
         for (i = 1; i <= depth; i++) {
@@ -122,4 +152,6 @@ curl 'https://health.ny.gov/diseases/communicable/coronavirus/' \
         sub("^" $1 s $2 s, "", val)
         _hxpipe_curr_attrs[key] = val
     }
-    '
+    ' \
+| column -ts\| \
+| sort -n -k 1