rewrite: rewrite rel urls to rel urls, both server and wombat side #123

Author: ikreymer

pywb/rewrite/test/test_url_rewriter.py

@@ -21,10 +21,10 @@
 
 # UrlRewriter tests
 >>> do_rewrite('other.html', '20131010/http://example.com/path/page.html', 'https://web.archive.org/web/')
-'https://web.archive.org/web/20131010/http://example.com/path/other.html'
+'/web/20131010/http://example.com/path/other.html'
 
 >>> do_rewrite('file.js', '20131010/http://example.com/path/page.html', 'https://web.archive.org/web/', 'js_')
-'https://web.archive.org/web/20131010js_/http://example.com/path/file.js'
+'/web/20131010js_/http://example.com/path/file.js'
 
 >>> do_rewrite('file.js', '20131010/http://example.com/', '/coll/')
 '/coll/20131010/http://example.com/file.js'
@@ -35,6 +35,9 @@
 >>> do_rewrite('file.js', '20131010/http://example.com', '/coll/', '')
 '/coll/20131010/http://example.com/file.js'
 
+>>> do_rewrite('/other.html', '20130907*/http://example.com/path/page.html', 'http://localhost:8080/coll/')
+'/coll/20130907*/http://example.com/other.html'
+
 >>> do_rewrite('/other.html', '20130907*/http://example.com/path/page.html', '/coll/')
 '/coll/20130907*/http://example.com/other.html'
 

pywb/rewrite/url_rewriter.py

@@ -30,6 +30,7 @@ def __init__(self, wburl, prefix, full_prefix=None, rel_prefix=None,
             self.prefix_scheme = self.full_prefix.split(':')[0]
         else:
             self.prefix_scheme = None
+        self.prefix_abs = self.prefix and self.prefix.startswith(self.PROTOCOLS)
         self.cookie_scope = cookie_scope
         self.rewrite_opts = rewrite_opts
 
@@ -74,8 +75,15 @@ def rewrite(self, url, mod=None):
             mod = wburl.mod
 
         final_url = self.prefix + wburl.to_str(mod=mod, url=new_url)
+
+        if not is_abs and self.prefix_abs and not self.rewrite_opts.get('no_match_rel'):
+            parts = final_url.split('/', 3)
+            final_url = '/'
+            if len(parts) == 4:
+                final_url += parts[3]
+
         # experiment for setting scheme rel url
-        if scheme_rel and self.prefix.startswith(self.PROTOCOLS):
+        elif scheme_rel and self.prefix_abs:
             final_url = final_url.split(':', 1)[1]
 
         return final_url

pywb/static/wombat.js

@@ -143,7 +143,7 @@ var wombat_internal = function($wbwindow) {
     var URL_PROPS = ["href", "hash", "pathname", "host", "hostname", "protocol", "origin", "search", "port"];
 
     //============================================
-    function rewrite_url_(url) {
+    function rewrite_url_(url, use_rel) {
         // If undefined, just return it
         if (!url) {
             return url;
@@ -215,7 +215,7 @@ var wombat_internal = function($wbwindow) {
                 return url;
             }
 
-            return wb_replay_date_prefix + wb_orig_origin + url;
+            return (use_rel ? wb_coll_prefix : wb_replay_date_prefix) + wb_orig_origin + url;
         }
 
         // If full url starting with http://, https:// or //
@@ -235,7 +235,6 @@ var wombat_internal = function($wbwindow) {
                 }
 
                 var curr_scheme = orig_protocol + '//';
-                var host = orig_host + '/';
                 var path = url.substring(prefix_host.length);
                 var rebuild = false;
 
@@ -251,7 +250,12 @@ var wombat_internal = function($wbwindow) {
                 }
 
                 if (rebuild) {
-                    url = curr_scheme + host + path;
+                    if (!use_rel) {
+                        url = curr_scheme + orig_host;
+                    } else {
+                        url = "";
+                    }
+                    url += "/" + path;
                 }
 
                 return url;
@@ -451,7 +455,8 @@ var wombat_internal = function($wbwindow) {
                     value = this._parser[prop];
                 } else {
                     prop = "href";
-                    value = rewrite_url(this._parser.href);
+                    var rel = (value == this._parser.pathname);
+                    value = rewrite_url(this._parser.href, rel);
                 }
 
                 orig_setter.call(this, prop, value);