components/dom_distiller/core/distiller_page.cc

   1 // Copyright 2013 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "components/dom_distiller/core/distiller_page.h"
   6
   7 #include "base/bind.h"
   8 #include "base/json/json_writer.h"
   9 #include "base/logging.h"
  10 #include "base/message_loop/message_loop.h"
  11 #include "base/metrics/histogram.h"
  12 #include "base/strings/string_util.h"
  13 #include "base/strings/utf_string_conversions.h"
  14 #include "base/time/time.h"
  15 #include "grit/components_resources.h"
  16 #include "third_party/dom_distiller_js/dom_distiller.pb.h"
  17 #include "third_party/dom_distiller_js/dom_distiller_json_converter.h"
  18 #include "ui/base/resource/resource_bundle.h"
  19 #include "url/gurl.h"
  20
  21 namespace dom_distiller {
  22
  23 namespace {
  24
  25 const char* kOptionsPlaceholder = "$$OPTIONS";
  26 const char* kStringifyPlaceholder = "$$STRINGIFY";
  27 const char* kNewContextPlaceholder = "$$NEW_CONTEXT";
  28
  29 std::string GetDistillerScriptWithOptions(
  30     const dom_distiller::proto::DomDistillerOptions& options,
  31     bool stringify_output,
  32     bool create_new_context) {
  33   std::string script = ResourceBundle::GetSharedInstance()
  34                            .GetRawDataResource(IDR_DISTILLER_JS)
  35                            .as_string();
  36   if (script.empty()) {
  37     return "";
  38   }
  39
  40   scoped_ptr<base::Value> options_value(
  41       dom_distiller::proto::json::DomDistillerOptions::WriteToValue(options));
  42   std::string options_json;
  43   if (!base::JSONWriter::Write(*options_value, &options_json)) {
  44     NOTREACHED();
  45   }
  46   size_t options_offset = script.find(kOptionsPlaceholder);
  47   DCHECK_NE(std::string::npos, options_offset);
  48   DCHECK_EQ(std::string::npos,
  49             script.find(kOptionsPlaceholder, options_offset + 1));
  50   script =
  51       script.replace(options_offset, strlen(kOptionsPlaceholder), options_json);
  52
  53   std::string stringify = stringify_output ? "true" : "false";
  54   size_t stringify_offset = script.find(kStringifyPlaceholder);
  55   DCHECK_NE(std::string::npos, stringify_offset);
  56   DCHECK_EQ(std::string::npos,
  57             script.find(kStringifyPlaceholder, stringify_offset + 1));
  58   script = script.replace(stringify_offset,
  59                           strlen(kStringifyPlaceholder),
  60                           stringify);
  61
  62   std::string new_context = create_new_context ? "true" : "false";
  63   size_t new_context_offset = script.find(kNewContextPlaceholder);
  64   DCHECK_NE(std::string::npos, new_context_offset);
  65   DCHECK_EQ(std::string::npos,
  66             script.find(kNewContextPlaceholder, new_context_offset + 1));
  67   script = script.replace(new_context_offset,
  68                           strlen(kNewContextPlaceholder),
  69                           new_context);
  70
  71   return script;
  72 }
  73
  74 }
  75
  76 DistillerPageFactory::~DistillerPageFactory() {}
  77
  78 DistillerPage::DistillerPage() : ready_(true) {}
  79
  80 DistillerPage::~DistillerPage() {}
  81
  82 void DistillerPage::DistillPage(
  83     const GURL& gurl,
  84     const dom_distiller::proto::DomDistillerOptions options,
  85     const DistillerPageCallback& callback) {
  86   DCHECK(ready_);
  87   // It is only possible to distill one page at a time. |ready_| is reset when
  88   // the callback to OnDistillationDone happens.
  89   ready_ = false;
  90   distiller_page_callback_ = callback;
  91   distillation_start_ = base::TimeTicks::Now();
  92   DistillPageImpl(gurl, GetDistillerScriptWithOptions(options,
  93                                                       StringifyOutput(),
  94                                                       CreateNewContext()));
  95 }
  96
  97 void DistillerPage::OnDistillationDone(const GURL& page_url,
  98                                        const base::Value* value) {
  99   DCHECK(!ready_);
 100   ready_ = true;
 101
 102   scoped_ptr<dom_distiller::proto::DomDistillerResult> distiller_result(
 103       new dom_distiller::proto::DomDistillerResult());
 104   bool found_content;
 105   if (value->IsType(base::Value::TYPE_NULL)) {
 106     found_content = false;
 107   } else {
 108     found_content =
 109         dom_distiller::proto::json::DomDistillerResult::ReadFromValue(
 110             value, distiller_result.get());
 111     if (!found_content) {
 112       DVLOG(1) << "Unable to parse DomDistillerResult.";
 113     } else {
 114       base::TimeDelta distillation_time =
 115           base::TimeTicks::Now() - distillation_start_;
 116       UMA_HISTOGRAM_TIMES("DomDistiller.Time.DistillPage", distillation_time);
 117       VLOG(1) << "DomDistiller.Time.DistillPage = " << distillation_time;
 118
 119       if (distiller_result->has_timing_info()) {
 120         const dom_distiller::proto::TimingInfo& timing =
 121             distiller_result->timing_info();
 122         if (timing.has_markup_parsing_time()) {
 123           UMA_HISTOGRAM_TIMES(
 124               "DomDistiller.Time.MarkupParsing",
 125               base::TimeDelta::FromMillisecondsD(timing.markup_parsing_time()));
 126         }
 127         if (timing.has_document_construction_time()) {
 128           UMA_HISTOGRAM_TIMES(
 129               "DomDistiller.Time.DocumentConstruction",
 130               base::TimeDelta::FromMillisecondsD(
 131                   timing.document_construction_time()));
 132         }
 133         if (timing.has_article_processing_time()) {
 134           UMA_HISTOGRAM_TIMES(
 135               "DomDistiller.Time.ArticleProcessing",
 136               base::TimeDelta::FromMillisecondsD(
 137                   timing.article_processing_time()));
 138         }
 139         if (timing.has_formatting_time()) {
 140           UMA_HISTOGRAM_TIMES(
 141               "DomDistiller.Time.Formatting",
 142               base::TimeDelta::FromMillisecondsD(timing.formatting_time()));
 143         }
 144         if (timing.has_total_time()) {
 145           UMA_HISTOGRAM_TIMES(
 146               "DomDistiller.Time.DistillationTotal",
 147               base::TimeDelta::FromMillisecondsD(timing.total_time()));
 148           VLOG(1) << "DomDistiller.Time.DistillationTotal = " <<
 149               base::TimeDelta::FromMillisecondsD(timing.total_time());
 150         }
 151       }
 152       if (distiller_result->has_statistics_info()) {
 153         const dom_distiller::proto::StatisticsInfo& statistics =
 154             distiller_result->statistics_info();
 155         if (statistics.has_word_count()) {
 156           UMA_HISTOGRAM_CUSTOM_COUNTS(
 157             "DomDistiller.Statistics.WordCount",
 158             statistics.word_count(),
 159             1, 4000, 50);
 160         }
 161       }
 162     }
 163   }
 164
 165   base::MessageLoop::current()->PostTask(
 166       FROM_HERE,
 167       base::Bind(distiller_page_callback_,
 168                  base::Passed(&distiller_result),
 169                  found_content));
 170 }
 171
 172 }  // namespace dom_distiller