Implement noskip.

copyvios/checker.py

@@ -65,11 +65,12 @@ def _get_results(query, follow=True):
             return
         mode = "{0}:{1}:".format(use_engine, use_links)
         if not query.nocache:
-            query.result = _get_cached_results(page, conn, mode)
+            query.result = _get_cached_results(page, conn, mode, query.noskip)
         if not query.result:
             query.result = page.copyvio_check(
                 min_confidence=T_SUSPECT, max_queries=10, max_time=45,
-                no_searches=not use_engine, no_links=not use_links)
+                no_searches=not use_engine, no_links=not use_links,
+                short_circuit=not query.noskip)
             query.result.cached = False
             _cache_result(page, query.result, conn, mode)
     elif query.action == "compare":
@@ -110,7 +111,7 @@ def _get_page_by_revid(site, revid):
     page._load_content(res)
     return page
 
-def _get_cached_results(page, conn, mode):
+def _get_cached_results(page, conn, mode, noskip):
     query1 = """DELETE FROM cache
                 WHERE cache_time < DATE_SUB(CURRENT_TIMESTAMP, INTERVAL 3 DAY)"""
     query2 = """SELECT cache_time, cache_queries, cache_process_time
@@ -147,6 +148,8 @@ def _get_cached_results(page, conn, mode):
         return None
 
     for url, confidence, skipped in data:
+        if noskip and skipped:
+            return None
         source = CopyvioSource(None, url)
         source.confidence = confidence
         source.skipped = bool(skipped)

templates/index.mako

@@ -193,7 +193,7 @@
             <li>Redirected from <a href="${query.redirected_from.url}">${query.redirected_from.title | h}</a>. <a href="${request.url | httpsfix, h}&amp;noredirect=1">Check the original page.</a></li>
         % endif
         % if skips:
-            <li>Since a suspected source was found with a high confidence value, some URLs were skipped. <a href="javascript:alert('Not implemented yet!');">Check all URLs.</a></li>
+            <li>Since a suspected source was found with a high confidence value, some URLs were skipped. <a href="${request.url | httpsfix, h}&amp;noskip=1">Check all URLs.</a></li>
         % endif
         % if result.cached:
             <li>Results are <a id="cv-cached" href="#">cached<span>To save time (and money), this tool will retain the results of checks for up to 72 hours. This includes the URLs of the checked sources, but neither their content nor the content of the article. Future checks on the same page (assuming it remains unchanged) will not involve additional search queries, but a fresh comparison against the source URL will be made. If the page is modified, a new check will be run.</span></a> from <abbr title="${result.cache_time}">${result.cache_age} ago</abbr>. Originally generated in <span class="mono">${round(result.time, 3)}</span> seconds using <span class="mono">${result.queries}</span> queries. <a href="${request.url | httpsfix, h}&amp;nocache=1">Bypass the cache.</a></li>