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/component_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
  27 std::string GetDistillerScriptWithOptions(
  28     const dom_distiller::proto::DomDistillerOptions& options) {
  29   std::string script = ResourceBundle::GetSharedInstance()
  30                            .GetRawDataResource(IDR_DISTILLER_JS)
  31                            .as_string();
  32   if (script.empty()) {
  33     return "";
  34   }
  35
  36   scoped_ptr<base::Value> options_value(
  37       dom_distiller::proto::json::DomDistillerOptions::WriteToValue(options));
  38   std::string options_json;
  39   if (!base::JSONWriter::Write(options_value.get(), &options_json)) {
  40     NOTREACHED();
  41   }
  42   size_t options_offset = script.find(kOptionsPlaceholder);
  43   DCHECK_NE(std::string::npos, options_offset);
  44   DCHECK_EQ(std::string::npos,
  45             script.find(kOptionsPlaceholder, options_offset + 1));
  46   script =
  47       script.replace(options_offset, strlen(kOptionsPlaceholder), options_json);
  48   return script;
  49 }
  50
  51 }
  52
  53 DistilledPageInfo::DistilledPageInfo() {}
  54
  55 DistilledPageInfo::~DistilledPageInfo() {}
  56
  57 DistilledPageInfo::MarkupArticle::MarkupArticle() {}
  58
  59 DistilledPageInfo::MarkupArticle::~MarkupArticle() {}
  60
  61 DistilledPageInfo::MarkupImage::MarkupImage() {}
  62
  63 DistilledPageInfo::MarkupImage::~MarkupImage() {}
  64
  65 DistilledPageInfo::MarkupInfo::MarkupInfo() {}
  66
  67 DistilledPageInfo::MarkupInfo::~MarkupInfo() {}
  68
  69 DistillerPageFactory::~DistillerPageFactory() {}
  70
  71 DistillerPage::DistillerPage() : ready_(true) {}
  72
  73 DistillerPage::~DistillerPage() {}
  74
  75 void DistillerPage::DistillPage(
  76     const GURL& gurl,
  77     const dom_distiller::proto::DomDistillerOptions options,
  78     const DistillerPageCallback& callback) {
  79   DCHECK(ready_);
  80   // It is only possible to distill one page at a time. |ready_| is reset when
  81   // the callback to OnDistillationDone happens.
  82   ready_ = false;
  83   distiller_page_callback_ = callback;
  84   DistillPageImpl(gurl, GetDistillerScriptWithOptions(options));
  85 }
  86
  87 void DistillerPage::OnDistillationDone(const GURL& page_url,
  88                                        const base::Value* value) {
  89   DCHECK(!ready_);
  90   ready_ = true;
  91
  92   scoped_ptr<DistilledPageInfo> page_info(new DistilledPageInfo());
  93   bool found_content = !value->IsType(base::Value::TYPE_NULL);
  94   if (found_content) {
  95     dom_distiller::proto::DomDistillerResult distiller_result =
  96         dom_distiller::proto::json::DomDistillerResult::ReadFromValue(value);
  97
  98     page_info->title = distiller_result.title();
  99     page_info->html = distiller_result.distilled_content().html();
 100     page_info->next_page_url = distiller_result.pagination_info().next_page();
 101     page_info->prev_page_url = distiller_result.pagination_info().prev_page();
 102     for (int i = 0; i < distiller_result.image_urls_size(); ++i) {
 103       const std::string image_url = distiller_result.image_urls(i);
 104       if (GURL(image_url).is_valid()) {
 105         page_info->image_urls.push_back(image_url);
 106       }
 107     }
 108     const dom_distiller::proto::MarkupInfo& src_markup_info =
 109         distiller_result.markup_info();
 110     DistilledPageInfo::MarkupInfo& dst_markup_info = page_info->markup_info;
 111     dst_markup_info.title = src_markup_info.title();
 112     dst_markup_info.type = src_markup_info.type();
 113     dst_markup_info.url = src_markup_info.url();
 114     dst_markup_info.description = src_markup_info.description();
 115     dst_markup_info.publisher = src_markup_info.publisher();
 116     dst_markup_info.copyright = src_markup_info.copyright();
 117     dst_markup_info.author = src_markup_info.author();
 118
 119     const dom_distiller::proto::MarkupArticle& src_article =
 120         src_markup_info.article();
 121     DistilledPageInfo::MarkupArticle& dst_article = dst_markup_info.article;
 122     dst_article.published_time = src_article.published_time();
 123     dst_article.modified_time = src_article.modified_time();
 124     dst_article.expiration_time = src_article.expiration_time();
 125     dst_article.section = src_article.section();
 126     for (int i = 0; i < src_article.authors_size(); ++i) {
 127       dst_article.authors.push_back(src_article.authors(i));
 128     }
 129
 130     for (int i = 0; i < src_markup_info.images_size(); ++i) {
 131       const dom_distiller::proto::MarkupImage& src_image =
 132           src_markup_info.images(i);
 133       DistilledPageInfo::MarkupImage dst_image;
 134       dst_image.url = src_image.url();
 135       dst_image.secure_url = src_image.secure_url();
 136       dst_image.type = src_image.type();
 137       dst_image.caption = src_image.caption();
 138       dst_image.width = src_image.width();
 139       dst_image.height = src_image.height();
 140       dst_markup_info.images.push_back(dst_image);
 141     }
 142     if (distiller_result.has_timing_info()) {
 143       const dom_distiller::proto::TimingInfo& timing =
 144           distiller_result.timing_info();
 145       if (timing.has_markup_parsing_time()) {
 146         UMA_HISTOGRAM_TIMES(
 147             "DomDistiller.Time.MarkupParsing",
 148             base::TimeDelta::FromMillisecondsD(timing.markup_parsing_time()));
 149       }
 150       if (timing.has_document_construction_time()) {
 151         UMA_HISTOGRAM_TIMES(
 152             "DomDistiller.Time.DocumentConstruction",
 153             base::TimeDelta::FromMillisecondsD(
 154                 timing.document_construction_time()));
 155       }
 156       if (timing.has_article_processing_time()) {
 157         UMA_HISTOGRAM_TIMES(
 158             "DomDistiller.Time.ArticleProcessing",
 159             base::TimeDelta::FromMillisecondsD(
 160                 timing.article_processing_time()));
 161       }
 162       if (timing.has_formatting_time()) {
 163         UMA_HISTOGRAM_TIMES(
 164             "DomDistiller.Time.Formatting",
 165             base::TimeDelta::FromMillisecondsD(timing.formatting_time()));
 166       }
 167       if (timing.has_total_time()) {
 168         UMA_HISTOGRAM_TIMES(
 169             "DomDistiller.Time.DistillationTotal",
 170             base::TimeDelta::FromMillisecondsD(timing.total_time()));
 171       }
 172     }
 173   }
 174
 175   base::MessageLoop::current()->PostTask(
 176       FROM_HERE,
 177       base::Bind(
 178           distiller_page_callback_, base::Passed(&page_info), found_content));
 179 }
 180
 181 }  // namespace dom_distiller