Use a yellow 'possible' box for confidence between 0.3 and 0.6 (closes #5).

copyvios/checker.py

@@ -9,7 +9,10 @@ from earwigbot import exceptions
 from .misc import Query, get_cache_db
 from .sites import get_site, get_sites
 
-__all__ = ["do_check"]
+__all__ = ["do_check", "T_POSSIBLE", "T_SUSPECT"]
+
+T_POSSIBLE = 0.3
+T_SUSPECT = 0.6
 
 def do_check():
     query = Query()
@@ -60,7 +63,8 @@ def _get_results(query, follow=True):
         if not query.nocache:
             query.result = _get_cached_results(page, conn, query)
         if not query.result:
-            query.result = page.copyvio_check(max_queries=10, max_time=45)
+            query.result = page.copyvio_check(
+                min_confidence=T_SUSPECT, max_queries=10, max_time=45)
             query.result.cached = False
             _cache_result(page, query.result, conn)
 
@@ -108,7 +112,7 @@ def _get_cached_results(page, conn, query):
     return result
 
 def _do_copyvio_compare(query, page, url):
-    result = page.copyvio_compare(url, max_time=30)
+    result = page.copyvio_compare(url, min_confidence=T_SUSPECT, max_time=30)
     if result.source_chain is not page.EMPTY:
         return result
     query.error = "timeout" if result.time > 30 else "no data"

static/style.css

@@ -126,6 +126,11 @@ div.green-box {
     border: 1px solid #7F7;
 }
 
+div.yellow-box {
+    background-color: #FFE;
+    border: 1px solid #FF7;
+}
+
 div.red-box {
     background-color: #FEE;
     border: 1px solid #F77;

templates/index.mako

@@ -1,5 +1,6 @@
 <%! from flask import g, request %>\
 <%include file="/support/header.mako" args="title='Earwig\'s Copyvio Detector'"/>
+<%namespace module="copyvios.checker" import="T_POSSIBLE, T_SUSPECT"/>\
 <%namespace module="copyvios.highlighter" import="highlight_delta"/>\
 <%namespace module="copyvios.misc" import="httpsfix, urlstrip"/>\
 % if query.project and query.lang and (query.title or query.oldid):
@@ -111,12 +112,13 @@
 % if result:
     <% hide_comparison = "CopyviosHideComparison" in g.cookies and g.cookies["CopyviosHideComparison"].value == "True" %>
     <div class="divider"></div>
-    <div id="cv-result" class="${'red' if result.violation else 'green'}-box">
-        % if result.violation:
+    <div id="cv-result" class="${'red' if result.confidence >= T_SUSPECT else 'yellow' if result.confidence >= T_POSSIBLE else 'green'}-box">
+        % if result.confidence >= T_POSSIBLE:
+            <% vio_type = "suspected" if result.confidence >= T_SUSPECT else "possible" %>
             % if query.oldid:
-                <h2 id="cv-result-header"><a href="${query.page.url}">${query.page.title | h}</a> @<a href="//${query.site.domain | h}/w/index.php?oldid=${query.oldid | h}">${query.oldid | h}</a> is a suspected violation of <a href="${result.url | h}">${result.url | urlstrip, h}</a>.</h2>
+                <h2 id="cv-result-header"><a href="${query.page.url}">${query.page.title | h}</a> @<a href="//${query.site.domain | h}/w/index.php?oldid=${query.oldid | h}">${query.oldid | h}</a> is a ${vio_type} violation of <a href="${result.url | h}">${result.url | urlstrip, h}</a>.</h2>
             % else:
-                <h2 id="cv-result-header"><a href="${query.page.url}">${query.page.title | h}</a> is a suspected violation of <a href="${result.url | h}">${result.url | urlstrip, h}</a>.</h2>
+                <h2 id="cv-result-header"><a href="${query.page.url}">${query.page.title | h}</a> is a ${vio_type} violation of <a href="${result.url | h}">${result.url | urlstrip, h}</a>.</h2>
             % endif
         % else:
             % if query.oldid:
@@ -126,7 +128,7 @@
             % endif
         % endif
         <ul id="cv-result-list">
-            % if not result.violation and not query.url:
+            % if result.confidence < T_POSSIBLE and not query.url:
                 % if result.url:
                     <li>Best match: <a href="${result.url | h}">${result.url | urlstrip, h}</a>.</li>
                 % else: