Add URL normalizing functions for use later

1 files changed, 56 insertions, 1 deletions

@@ -105,7 +105,62 @@ NOT_IMPLEMENTED() { die 200 "NOT IMPLEMENTED!!!"; }
 NOT_FULLY_IMPLEMENTED() { log 1 "NOT FULLY IMPLEMENTED!!!"; }
 
 ### gemini ###
-# normalize a gemini address
+# url functions
+# normalize a path from /../ /./ /
+normalize_path() { # normalize_path <<< PATH
+        gawk '{
+        if ($0 == "" || $0 ~ /^\/\/[^\/]/) {
+                return -1
+        }
+        split($0, path, /\//)
+        for (c in path) {
+                if (path[c] == "" || path[c] == ".") {
+                        continue
+                }
+                if (path[c] == "..") {
+                        sub(/[^\/]+$/, "", ret)
+                        continue
+                }
+                if (! ret || match(ret, /\/$/)) {
+                        slash = ""
+                } else {
+                        slash = "/"
+                }
+                ret = ret slash path[c]
+        }
+        print ret
+        }'
+}
+
+# split a url into the URL array
+split_url() {
+        gawk '{
+        if (match($0, /^[A-Za-z]+:/)) {
+                arr["scheme"] = substr($0, RSTART, RLENGTH)
+                $0 = substr($0, RLENGTH + 1)
+        }
+        if (match($0, /^\/\/[^\/?#]+?/) || (match($0, /^[^\/?#]+?/) && scheme)) {
+                arr["authority"] = substr($0, RSTART, RLENGTH)
+                $0 = substr($0, RLENGTH + 1)
+        }
+        if (match($0, /^\/?[^?#]+/)) {
+                arr["path"] = substr($0, RSTART, RLENGTH)
+                $0 = substr($0, RLENGTH + 1)
+        }
+        if (match($0, /^\?[^#]+/)) {
+                arr["query"] = substr($0, RSTART, RLENGTH)
+                $0 = substr($0, RLENGTH + 1)
+        }
+        if (match($0, /^#.*/)) {
+                arr["fragment"] = substr($0, RSTART, RLENGTH)
+                $0 = substr($0, RLENGTH + 1)
+        }
+        for (part in arr) {
+                printf "URL[\"%s\"]=\"%s\"\n", part, arr[part]
+        }
+        }'
+}
+
 # example.com => gemini://example.com/
 _address() { # _address URL
         addr="$1"