X-Git-Url: https://git.xandkar.net/?p=covid-19-scrapers.git;a=blobdiff_plain;f=fetch-case-count-usa-ny;h=16cf859969241d777c71a4ebb3a1932fb90953a6;hp=808d7be4b60acebcae2c0bf2a1d789ef8dfe0287;hb=851726db1ae32895ef5a499bc59da443be33c3c0;hpb=2162fd84b7bf9f87e6318e139e9f2a687659e26e

diff --git a/fetch-case-count-usa-ny b/fetch-case-count-usa-ny
index 808d7be..16cf859 100755
--- a/fetch-case-count-usa-ny
+++ b/fetch-case-count-usa-ny
@@ -5,6 +5,21 @@
 #     - 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
@@ -19,7 +34,6 @@ VALID_LOCATIONS='
     Montgomery
     Nassau
     New York City
-    New York State (Outside of NYC)
     Onondaga
     Ontario
     Orange
@@ -32,16 +46,15 @@ VALID_LOCATIONS='
     Sullivan
     Tioga
     Tompkins
-    Total Positive Cases (Statewide)
     Ulster
     Westchester
     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"
@@ -50,30 +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
 
-# 2020-03-16
-#URI='https://health.ny.gov/diseases/communicable/coronavirus/'
-
-# 2020-03-17
-URI='https://coronavirus.health.ny.gov/county-county-breakdown-positive-cases'
-
 curl "$URI" \
 | hxpipe \
-| awk -v location="$location" '
-    BEGIN {
-        # 2020-03-16 morning-afternoon
-        #target_path = "/html/body/div/div/div/div/div/div/table/tr/td"
-
-        # 2020-03-16 late evening 22:30
-        #target_path = "/html/body/div/div/div/div/div/div/tr/td"
-
-        # 2020-03-17 late afternoon 16:35
-        target_path = "/html/body/div/div/main/div/div/div/div/div/div/div/div/table/tbody/tr/td"
-    }
-
+| awk \
+    -v target_location="$target_location" \
+    -v target_path="$TARGET_PATH" \
+    -v opt_all="$opt_all" \
+    '
     /^[\(\)]/ {
         update_node()
         next
@@ -88,17 +89,26 @@ curl "$URI" \
         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++) {
@@ -142,4 +152,6 @@ curl "$URI" \
         sub("^" $1 s $2 s, "", val)
         _hxpipe_curr_attrs[key] = val
     }
-    '
+    ' \
+| column -ts\| \
+| sort -n -k 1