Relax CSS inlining charset check to always permit ASCII-only data.

Needed for fonts.g.c due to apparent recent changes (and us not
permitting inlining there was making the corresponding continuous tests
fail).

Author: morlovich

net/instaweb/rewriter/css_inline_filter.cc

@@ -175,15 +175,34 @@ bool CssInlineFilter::ShouldInline(const ResourcePtr& resource,
     return false;
   }
 
-  // If the charset is incompatible with the HTML's, don't inline.
+  // If the charset is incompatible with the HTML's, we may not be able to
+  // inline.
   StringPiece htmls_charset(driver()->containing_charset());
   GoogleString css_charset = RewriteFilter::GetCharsetForStylesheet(
       resource.get(), attrs_charset, htmls_charset);
   if (!StringCaseEqual(htmls_charset, css_charset)) {
-    *reason = StrCat("CSS not inlined due to apparent charset incompatibility;"
-                     " we think the HTML is ", htmls_charset,
-                     " while the CSS is ", css_charset);
-    return false;
+    // Check if everything is in <= 127 range, we may still be able to
+    // inline if it keeps to the ASCII subset (also potentially dropping the
+    // BOM, since we'll strip it anyway).
+    StringPiece contents = resource->ExtractUncompressedContents();
+    StringPiece clean_contents(contents);
+    StripUtf8Bom(&clean_contents);
+
+    bool has_non_ascii = false;
+    for (int i = 0, n = clean_contents.size(); i < n; ++i) {
+      if (static_cast<unsigned char>(clean_contents[i]) >= 0x80) {
+        has_non_ascii = true;
+        break;
+      }
+    }
+
+    if (has_non_ascii) {
+      *reason = StrCat(
+          "CSS not inlined due to apparent charset incompatibility;"
+          " we think the HTML is ", htmls_charset,
+          " while the CSS is ", css_charset);
+      return false;
+    }
   }
 
   return true;

net/instaweb/rewriter/css_inline_filter_test.cc

@@ -581,8 +581,8 @@ TEST_F(CssInlineFilterTest, InlineWithCompatibleBom) {
                              "", css_with_bom, true, css, "");
 }
 
-TEST_F(CssInlineFilterTest, DoNotInlineWithIncompatibleBom) {
-  const GoogleString css = "BODY { color: red; }\n";
+TEST_F(CssInlineFilterTest, DoNotInlineWithIncompatibleBomAndNonAscii) {
+  const GoogleString css = "BODY { color: red; /* \xD2\x90 */ }\n";
   const GoogleString css_with_bom = StrCat(kUtf8Bom, css);
   TestInlineCssWithOutputUrl("http://www.example.com/index.html",
                              "  <meta charset=\"ISO-8859-1\">\n",
@@ -594,6 +594,19 @@ TEST_F(CssInlineFilterTest, DoNotInlineWithIncompatibleBom) {
                              "while the CSS is utf-8");
 }
 
+TEST_F(CssInlineFilterTest, DoInlineWithIncompatibleBomAndAscii) {
+  // Even though the content is labeled utf-8, it keeps to ASCII subset, so it's
+  // safe to inline.
+  const GoogleString css = "BODY { color: red; }\n";
+  const GoogleString css_with_bom = StrCat(kUtf8Bom, css);
+  TestInlineCssWithOutputUrl("http://www.example.com/index.html",
+                             "  <meta charset=\"ISO-8859-1\">\n",
+                             "http://www.example.com/styles.css",
+                             "http://www.example.com/styles.css",
+                             "", css_with_bom, true, css,
+                             "");
+}
+
 // See: http://www.alistapart.com/articles/alternate/
 //  and http://www.w3.org/TR/html4/present/styles.html#h-14.3.1
 TEST_F(CssInlineFilterTest, AlternateStylesheet) {