1 files changed, 44 insertions, 27 deletions

diff --git a/common/patterns.js b/common/patterns.js
index e198482..5d84a73 100644
--- a/common/patterns.js
+++ b/common/patterns.js
@@ -41,50 +41,67 @@
  * proprietary program, I am not going to enforce this in court.
  */
 
-const MAX_URL_PATH_LEN = 12;
-const MAX_URL_PATH_CHARS = 255;
-const MAX_DOMAIN_LEN = 7;
-const MAX_DOMAIN_CHARS = 100;
+const MAX = {
+    URL_PATH_LEN:   12,
+    URL_PATH_CHARS: 255,
+    DOMAIN_LEN:     7,
+    DOMAIN_CHARS:   100
+};
 
 const proto_regex = /^(\w+):\/\/(.*)$/;
 
 const user_re = "[^/?#@]+@"
-const domain_re = "[^/?#]+";
+const domain_re = "[.*a-zA-Z0-9-]+";
 const path_re = "[^?#]*";
 const query_re = "\\??[^#]*";
 
 const http_regex = new RegExp(`^(${domain_re})(${path_re})(${query_re}).*`);
 
-const file_regex = new RegExp(`^(${path_re}).*`);
+const file_regex = new RegExp(`^(/${path_re}).*`);
 
 const ftp_regex = new RegExp(`^(${user_re})?(${domain_re})(${path_re}).*`);
 
-function deconstruct_url(url)
+function match_or_throw(regex, string, error_msg)
 {
-    const proto_match = proto_regex.exec(url);
-    if (proto_match === null)
-	return undefined;
+    const match = regex.exec(string);
+    if (match === null)
+	throw error_msg;
 
+    return match;
+}
+
+function deconstruct_url(url, use_limits=true)
+{
+    const max = MAX;
+    if (!use_limits) {
+	for (key in MAX)
+	    max[key] = Infinity;
+    }
+
+    const matcher = (re, str) => match_or_throw(re, str, `bad url '${url}'`)
+
+    const proto_match = matcher(proto_regex, url);
     const deco = {proto: proto_match[1]};
 
     if (deco.proto === "file") {
-	deco.path = file_regex.exec(proto_match[2])[1];
+	deco.path = matcher(file_regex, proto_match[2])[1];
     } else if (deco.proto === "ftp") {
-	[deco.domain, deco.path] = ftp_regex.exec(proto_match[2]).slice(2, 4);
+	[deco.domain, deco.path] =
+	    matcher(ftp_regex, proto_match[2]).slice(2, 4);
+    } else if (deco.proto === "http" || deco.proto === "https") {
+	[deco.domain, deco.path, deco.query] =
+	    matcher(http_regex, proto_match[2]).slice(1, 4);
+	deco.domain = deco.domain.toLowerCase();
     } else {
-	const http_match = http_regex.exec(proto_match[2]);
-	if (!http_match)
-	    return undefined;
-	[deco.domain, deco.path, deco.query] = http_match.slice(1, 4);
+	throw `unsupported protocol in url '${url}'`;
     }
 
-    const leading_dash = deco.path[0] === "/";
     deco.trailing_dash = deco.path[deco.path.length - 1] === "/";
 
     if (deco.domain) {
-	if (deco.domain.length > MAX_DOMAIN_CHARS) {
+	if (deco.domain.length > max.DOMAIN_CHARS) {
 	    const idx = deco.domain.indexOf(".", deco.domain.length -
-					    MAX_DOMAIN_CHARS);
+					    max.DOMAIN_CHARS);
 	    if (idx === -1)
 		deco.domain = [];
 	    else
@@ -93,7 +110,7 @@ function deconstruct_url(url)
 	    deco.domain_truncated = true;
 	}
 
-	if (deco.path.length > MAX_URL_PATH_CHARS) {
+	if (deco.path.length > max.URL_PATH_CHARS) {
 	    deco.path = deco.path.substring(0, deco.path.lastIndexOf("/"));
 	    deco.path_truncated = true;
 	}
@@ -101,16 +118,14 @@ function deconstruct_url(url)
 
     if (typeof deco.domain === "string") {
 	deco.domain = deco.domain.split(".");
-	if (deco.domain.splice(0, deco.domain.length - MAX_DOMAIN_LEN).length
+	if (deco.domain.splice(0, deco.domain.length - max.DOMAIN_LEN).length
 	    > 0)
 	    deco.domain_truncated = true;
     }
 
     deco.path = deco.path.split("/").filter(s => s !== "");
-    if (deco.domain && deco.path.splice(MAX_URL_PATH_LEN).length > 0)
+    if (deco.domain && deco.path.splice(max.URL_PATH_LEN).length > 0)
 	deco.path_truncated = true;
-    if (leading_dash || deco.path.length === 0)
-	deco.path.unshift("");
 
     return deco;
 }
@@ -132,13 +147,14 @@ function* each_domain_pattern(deco)
 
 function* each_path_pattern(deco)
 {
-    for (let slice = deco.path.length; slice > 0; slice--) {
-	const path_part = deco.path.slice(0, slice).join("/");
+    for (let slice = deco.path.length; slice >= 0; slice--) {
+	const path_part = ["", ...deco.path.slice(0, slice)].join("/");
 	const path_wildcards = [];
 	if (slice === deco.path.length && !deco.path_truncated) {
 	    if (deco.trailing_dash)
 		yield path_part + "/";
-	    yield path_part;
+	    if (slice > 0 || deco.proto !== "file")
+		yield path_part;
 	}
 	if (slice === deco.path.length - 1 && !deco.path_truncated &&
 	    deco.path[slice] !== "*")
@@ -171,5 +187,6 @@ function* each_url_pattern(url)
 /*
  * EXPORTS_START
  * EXPORT each_url_pattern
+ * EXPORT deconstruct_url
  * EXPORTS_END
  */