// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
+#include <algorithm>
#include <map>
+#include <string>
+#include <vector>
#include "base/bind.h"
#include "base/bind_helpers.h"
#include "base/location.h"
#include "base/memory/scoped_ptr.h"
#include "base/message_loop/message_loop.h"
+#include "base/strings/string_number_conversions.h"
#include "base/values.h"
+#include "components/dom_distiller/core/article_distillation_update.h"
#include "components/dom_distiller/core/distiller.h"
#include "components/dom_distiller/core/distiller_page.h"
+#include "components/dom_distiller/core/fake_distiller_page.h"
+#include "components/dom_distiller/core/proto/distilled_article.pb.h"
#include "components/dom_distiller/core/proto/distilled_page.pb.h"
#include "net/url_request/url_request_context_getter.h"
#include "testing/gmock/include/gmock/gmock.h"
#include "testing/gtest/include/gtest/gtest.h"
+#include "third_party/dom_distiller_js/dom_distiller.pb.h"
+#include "third_party/dom_distiller_js/dom_distiller_json_converter.h"
-using::testing::Invoke;
-using::testing::Return;
-using::testing::_;
+using std::vector;
+using std::string;
+using ::testing::Invoke;
+using ::testing::Return;
+using ::testing::_;
+
+using dom_distiller::proto::DomDistillerOptions;
+using dom_distiller::proto::DomDistillerResult;
namespace {
- const char kTitle[] = "Title";
- const char kContent[] = "Content";
- const char kURL[] = "http://a.com/";
- const char kId0[] = "0";
- const char kId1[] = "1";
- const char kImageURL0[] = "http://a.com/img1.jpg";
- const char kImageURL1[] = "http://a.com/img2.jpg";
- const char kImageData0[] = { 'a', 'b', 'c', 'd', 'e', 0 };
- const char kImageData1[] = { '1', '2', '3', '4', '5', 0 };
+const char kTitle[] = "Title";
+const char kContent[] = "Content";
+const char kURL[] = "http://a.com/";
+const size_t kTotalImages = 2;
+const char* kImageURLs[kTotalImages] = {"http://a.com/img1.jpg",
+ "http://a.com/img2.jpg"};
+const char* kImageData[kTotalImages] = {"abcde", "12345"};
+const char kDebugLog[] = "Debug Log";
+
+const string GetImageName(int page_num, int image_num) {
+ return base::IntToString(page_num) + "_" + base::IntToString(image_num);
+}
+
+scoped_ptr<base::Value> CreateDistilledValueReturnedFromJS(
+ const string& title,
+ const string& content,
+ const vector<int>& image_indices,
+ const string& next_page_url,
+ const string& prev_page_url = "") {
+ DomDistillerResult result;
+ result.set_title(title);
+ result.mutable_distilled_content()->set_html(content);
+ result.mutable_pagination_info()->set_next_page(next_page_url);
+ result.mutable_pagination_info()->set_prev_page(prev_page_url);
+
+ for (size_t i = 0; i < image_indices.size(); ++i) {
+ result.add_image_urls(kImageURLs[image_indices[i]]);
+ }
+
+ return dom_distiller::proto::json::DomDistillerResult::WriteToValue(result);
+}
+
+// Return the sequence in which Distiller will distill pages.
+// Note: ignores any delays due to fetching images etc.
+vector<int> GetPagesInSequence(int start_page_num, int num_pages) {
+ // Distiller prefers distilling past pages first. E.g. when distillation
+ // starts on page 2 then pages are distilled in the order: 2, 1, 0, 3, 4.
+ vector<int> page_nums;
+ for (int page = start_page_num; page >= 0; --page)
+ page_nums.push_back(page);
+ for (int page = start_page_num + 1; page < num_pages; ++page)
+ page_nums.push_back(page);
+ return page_nums;
+}
+
+struct MultipageDistillerData {
+ public:
+ MultipageDistillerData() {}
+ ~MultipageDistillerData() {}
+ vector<string> page_urls;
+ vector<string> content;
+ vector<vector<int> > image_ids;
+ // The Javascript values returned by mock distiller.
+ ScopedVector<base::Value> distilled_values;
+
+ private:
+ DISALLOW_COPY_AND_ASSIGN(MultipageDistillerData);
+};
+
+void VerifyIncrementalUpdatesMatch(
+ const MultipageDistillerData* distiller_data,
+ int num_pages_in_article,
+ const vector<dom_distiller::ArticleDistillationUpdate>& incremental_updates,
+ int start_page_num) {
+ vector<int> page_seq =
+ GetPagesInSequence(start_page_num, num_pages_in_article);
+ // Updates should contain a list of pages. Pages in an update should be in
+ // the correct ascending page order regardless of |start_page_num|.
+ // E.g. if distillation starts at page 2 of a 3 page article, the updates
+ // will be [[2], [1, 2], [1, 2, 3]]. This example assumes that image fetches
+ // do not delay distillation of a page. There can be scenarios when image
+ // fetch delays distillation of a page (E.g. 1 is delayed due to image
+ // fetches so updates can be in this order [[2], [2,3], [1,2,3]].
+ for (size_t update_count = 0; update_count < incremental_updates.size();
+ ++update_count) {
+ const dom_distiller::ArticleDistillationUpdate& update =
+ incremental_updates[update_count];
+ EXPECT_EQ(update_count + 1, update.GetPagesSize());
+
+ vector<int> expected_page_nums_in_update(
+ page_seq.begin(), page_seq.begin() + update.GetPagesSize());
+ std::sort(expected_page_nums_in_update.begin(),
+ expected_page_nums_in_update.end());
+
+ // If we already got the first page then there is no previous page.
+ EXPECT_EQ((expected_page_nums_in_update[0] != 0), update.HasPrevPage());
+
+ // if we already got the last page then there is no next page.
+ EXPECT_EQ(
+ (*expected_page_nums_in_update.rbegin()) != num_pages_in_article - 1,
+ update.HasNextPage());
+ for (size_t j = 0; j < update.GetPagesSize(); ++j) {
+ int actual_page_num = expected_page_nums_in_update[j];
+ EXPECT_EQ(distiller_data->page_urls[actual_page_num],
+ update.GetDistilledPage(j).url());
+ EXPECT_EQ(distiller_data->content[actual_page_num],
+ update.GetDistilledPage(j).html());
+ }
+ }
+}
+
+string GenerateNextPageUrl(const std::string& url_prefix, size_t page_num,
+ size_t pages_size) {
+ return page_num + 1 < pages_size ?
+ url_prefix + base::IntToString(page_num + 1) : "";
+}
+
+string GeneratePrevPageUrl(const std::string& url_prefix, size_t page_num) {
+ return page_num > 0 ? url_prefix + base::IntToString(page_num - 1) : "";
+}
+
+scoped_ptr<MultipageDistillerData> CreateMultipageDistillerDataWithoutImages(
+ size_t pages_size) {
+ scoped_ptr<MultipageDistillerData> result(new MultipageDistillerData());
+ string url_prefix = kURL;
+ for (size_t page_num = 0; page_num < pages_size; ++page_num) {
+ result->page_urls.push_back(url_prefix + base::IntToString(page_num));
+ result->content.push_back("Content for page:" +
+ base::IntToString(page_num));
+ result->image_ids.push_back(vector<int>());
+ string next_page_url =
+ GenerateNextPageUrl(url_prefix, page_num, pages_size);
+ string prev_page_url =
+ GeneratePrevPageUrl(url_prefix, page_num);
+ scoped_ptr<base::Value> distilled_value =
+ CreateDistilledValueReturnedFromJS(kTitle,
+ result->content[page_num],
+ result->image_ids[page_num],
+ next_page_url,
+ prev_page_url);
+ result->distilled_values.push_back(distilled_value.release());
+ }
+ return result.Pass();
+}
+
+void VerifyArticleProtoMatchesMultipageData(
+ const dom_distiller::DistilledArticleProto* article_proto,
+ const MultipageDistillerData* distiller_data,
+ size_t distilled_pages_size,
+ size_t total_pages_size) {
+ ASSERT_EQ(distilled_pages_size,
+ static_cast<size_t>(article_proto->pages_size()));
+ EXPECT_EQ(kTitle, article_proto->title());
+ std::string url_prefix = kURL;
+ for (size_t page_num = 0; page_num < distilled_pages_size; ++page_num) {
+ const dom_distiller::DistilledPageProto& page =
+ article_proto->pages(page_num);
+ EXPECT_EQ(distiller_data->content[page_num], page.html());
+ EXPECT_EQ(distiller_data->page_urls[page_num], page.url());
+ EXPECT_EQ(distiller_data->image_ids[page_num].size(),
+ static_cast<size_t>(page.image_size()));
+ const vector<int>& image_ids_for_page = distiller_data->image_ids[page_num];
+ for (size_t img_num = 0; img_num < image_ids_for_page.size(); ++img_num) {
+ EXPECT_EQ(kImageData[image_ids_for_page[img_num]],
+ page.image(img_num).data());
+ EXPECT_EQ(GetImageName(page_num + 1, img_num),
+ page.image(img_num).name());
+ }
+ std::string expected_next_page_url =
+ GenerateNextPageUrl(url_prefix, page_num, total_pages_size);
+ std::string expected_prev_page_url =
+ GeneratePrevPageUrl(url_prefix, page_num);
+ EXPECT_EQ(expected_next_page_url, page.pagination_info().next_page());
+ EXPECT_EQ(expected_prev_page_url, page.pagination_info().prev_page());
+ EXPECT_FALSE(page.pagination_info().has_canonical_page());
+ }
+}
+
} // namespace
namespace dom_distiller {
+using test::MockDistillerPage;
+using test::MockDistillerPageFactory;
+
class TestDistillerURLFetcher : public DistillerURLFetcher {
public:
- TestDistillerURLFetcher() : DistillerURLFetcher(NULL) {
- responses_[kImageURL0] = std::string(kImageData0);
- responses_[kImageURL1] = std::string(kImageData1);
+ explicit TestDistillerURLFetcher(bool delay_fetch)
+ : DistillerURLFetcher(NULL), delay_fetch_(delay_fetch) {
+ responses_[kImageURLs[0]] = string(kImageData[0]);
+ responses_[kImageURLs[1]] = string(kImageData[1]);
}
- void CallCallback(std::string url, const URLFetcherCallback& callback) {
- callback.Run(responses_[url]);
+ void FetchURL(const string& url,
+ const URLFetcherCallback& callback) override {
+ ASSERT_FALSE(callback.is_null());
+ url_ = url;
+ callback_ = callback;
+ if (!delay_fetch_) {
+ PostCallbackTask();
+ }
}
- virtual void FetchURL(const std::string& url,
- const URLFetcherCallback& callback) OVERRIDE {
+ void PostCallbackTask() {
ASSERT_TRUE(base::MessageLoop::current());
+ ASSERT_FALSE(callback_.is_null());
base::MessageLoop::current()->PostTask(
- FROM_HERE,
- base::Bind(&TestDistillerURLFetcher::CallCallback,
- base::Unretained(this), url, callback));
+ FROM_HERE, base::Bind(callback_, responses_[url_]));
}
- std::map<std::string, std::string> responses_;
+ private:
+ std::map<string, string> responses_;
+ string url_;
+ URLFetcherCallback callback_;
+ bool delay_fetch_;
};
-
class TestDistillerURLFetcherFactory : public DistillerURLFetcherFactory {
public:
TestDistillerURLFetcherFactory() : DistillerURLFetcherFactory(NULL) {}
- virtual ~TestDistillerURLFetcherFactory() {}
- virtual DistillerURLFetcher* CreateDistillerURLFetcher() const OVERRIDE {
- return new TestDistillerURLFetcher();
+
+ ~TestDistillerURLFetcherFactory() override {}
+ DistillerURLFetcher* CreateDistillerURLFetcher() const override {
+ return new TestDistillerURLFetcher(false);
}
};
-
-class MockDistillerPage : public DistillerPage {
+class MockDistillerURLFetcherFactory : public DistillerURLFetcherFactory {
public:
- MOCK_METHOD0(InitImpl, void());
- MOCK_METHOD1(LoadURLImpl, void(const GURL& gurl));
- MOCK_METHOD1(ExecuteJavaScriptImpl, void(const std::string& script));
+ MockDistillerURLFetcherFactory() : DistillerURLFetcherFactory(NULL) {}
+ virtual ~MockDistillerURLFetcherFactory() {}
- explicit MockDistillerPage(DistillerPage::Delegate* delegate)
- : DistillerPage(delegate) {}
+ MOCK_CONST_METHOD0(CreateDistillerURLFetcher, DistillerURLFetcher*());
};
-
-class MockDistillerPageFactory : public DistillerPageFactory {
+class DistillerTest : public testing::Test {
public:
- MOCK_CONST_METHOD1(
- CreateDistillerPageMock,
- DistillerPage*(DistillerPage::Delegate* delegate));
+ ~DistillerTest() override {}
- virtual scoped_ptr<DistillerPage> CreateDistillerPage(
- DistillerPage::Delegate* delegate) const OVERRIDE {
- return scoped_ptr<DistillerPage>(CreateDistillerPageMock(delegate));
+ void OnDistillArticleDone(scoped_ptr<DistilledArticleProto> proto) {
+ article_proto_ = proto.Pass();
}
-};
+ void OnDistillArticleUpdate(const ArticleDistillationUpdate& article_update) {
+ in_sequence_updates_.push_back(article_update);
+ }
-class DistillerTest : public testing::Test {
- public:
- virtual ~DistillerTest() {}
- void OnDistillPageDone(scoped_ptr<DistilledPageProto> proto) {
- proto_ = proto.Pass();
+ void DistillPage(const std::string& url,
+ scoped_ptr<DistillerPage> distiller_page) {
+ distiller_->DistillPage(GURL(url),
+ distiller_page.Pass(),
+ base::Bind(&DistillerTest::OnDistillArticleDone,
+ base::Unretained(this)),
+ base::Bind(&DistillerTest::OnDistillArticleUpdate,
+ base::Unretained(this)));
}
protected:
scoped_ptr<DistillerImpl> distiller_;
- scoped_ptr<DistilledPageProto> proto_;
+ scoped_ptr<DistilledArticleProto> article_proto_;
+ std::vector<ArticleDistillationUpdate> in_sequence_updates_;
MockDistillerPageFactory page_factory_;
TestDistillerURLFetcherFactory url_fetcher_factory_;
};
-ACTION_P2(DistillerPageOnExecuteJavaScriptDone, distiller_page, list) {
- distiller_page->OnExecuteJavaScriptDone(list);
+ACTION_P3(DistillerPageOnDistillationDone, distiller_page, url, result) {
+ distiller_page->OnDistillationDone(url, result);
}
-ACTION_P2(CreateMockDistillerPage, list, kurl) {
- DistillerPage::Delegate* delegate = arg0;
- MockDistillerPage* distiller_page = new MockDistillerPage(delegate);
- EXPECT_CALL(*distiller_page, InitImpl());
- EXPECT_CALL(*distiller_page, LoadURLImpl(kurl))
- .WillOnce(testing::InvokeWithoutArgs(distiller_page,
- &DistillerPage::OnLoadURLDone));
- EXPECT_CALL(*distiller_page, ExecuteJavaScriptImpl(_))
- .WillOnce(DistillerPageOnExecuteJavaScriptDone(distiller_page, list));
- return distiller_page;
+scoped_ptr<DistillerPage> CreateMockDistillerPage(const base::Value* result,
+ const GURL& url) {
+ MockDistillerPage* distiller_page = new MockDistillerPage();
+ EXPECT_CALL(*distiller_page, DistillPageImpl(url, _))
+ .WillOnce(DistillerPageOnDistillationDone(distiller_page, url, result));
+ return scoped_ptr<DistillerPage>(distiller_page).Pass();
+}
+
+scoped_ptr<DistillerPage> CreateMockDistillerPageWithPendingJSCallback(
+ MockDistillerPage** distiller_page_ptr,
+ const GURL& url) {
+ MockDistillerPage* distiller_page = new MockDistillerPage();
+ *distiller_page_ptr = distiller_page;
+ EXPECT_CALL(*distiller_page, DistillPageImpl(url, _));
+ return scoped_ptr<DistillerPage>(distiller_page).Pass();
+}
+
+scoped_ptr<DistillerPage> CreateMockDistillerPages(
+ MultipageDistillerData* distiller_data,
+ size_t pages_size,
+ int start_page_num) {
+ MockDistillerPage* distiller_page = new MockDistillerPage();
+ {
+ testing::InSequence s;
+ vector<int> page_nums = GetPagesInSequence(start_page_num, pages_size);
+ for (size_t page_num = 0; page_num < pages_size; ++page_num) {
+ int page = page_nums[page_num];
+ GURL url = GURL(distiller_data->page_urls[page]);
+ EXPECT_CALL(*distiller_page, DistillPageImpl(url, _))
+ .WillOnce(DistillerPageOnDistillationDone(
+ distiller_page, url, distiller_data->distilled_values[page]));
+ }
+ }
+ return scoped_ptr<DistillerPage>(distiller_page).Pass();
}
TEST_F(DistillerTest, DistillPage) {
base::MessageLoopForUI loop;
- scoped_ptr<base::ListValue> list(new base::ListValue());
- list->AppendString(kTitle);
- list->AppendString(kContent);
- list->AppendString(kImageURL0);
- list->AppendString(kImageURL1);
- EXPECT_CALL(page_factory_,
- CreateDistillerPageMock(_)).WillOnce(
- CreateMockDistillerPage(list.get(), GURL(kURL)));
- distiller_.reset(new DistillerImpl(page_factory_, url_fetcher_factory_));
- distiller_->Init();
- distiller_->DistillPage(
- GURL(kURL),
- base::Bind(&DistillerTest::OnDistillPageDone, base::Unretained(this)));
- base::MessageLoop::current()->RunUntilIdle();
- EXPECT_EQ(kTitle, proto_->title());
- EXPECT_EQ(kContent, proto_->html());
- EXPECT_EQ(kURL, proto_->url());
- EXPECT_EQ(2, proto_->image_size());
- EXPECT_EQ(kImageData0, proto_->image(0).data());
- EXPECT_EQ(kId0, proto_->image(0).name());
- EXPECT_EQ(kImageData1, proto_->image(1).data());
- EXPECT_EQ(kId1, proto_->image(1).name());
+ scoped_ptr<base::Value> result =
+ CreateDistilledValueReturnedFromJS(kTitle, kContent, vector<int>(), "");
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(kURL, CreateMockDistillerPage(result.get(), GURL(kURL)).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kTitle, article_proto_->title());
+ ASSERT_EQ(article_proto_->pages_size(), 1);
+ const DistilledPageProto& first_page = article_proto_->pages(0);
+ EXPECT_EQ(kContent, first_page.html());
+ EXPECT_EQ(kURL, first_page.url());
+}
+
+TEST_F(DistillerTest, DistillPageWithDebugInfo) {
+ base::MessageLoopForUI loop;
+ DomDistillerResult dd_result;
+ dd_result.mutable_debug_info()->set_log(kDebugLog);
+ scoped_ptr<base::Value> result =
+ dom_distiller::proto::json::DomDistillerResult::WriteToValue(dd_result);
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(kURL, CreateMockDistillerPage(result.get(), GURL(kURL)).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ const DistilledPageProto& first_page = article_proto_->pages(0);
+ EXPECT_EQ(kDebugLog, first_page.debug_info().log());
+}
+
+TEST_F(DistillerTest, DistillPageWithImages) {
+ base::MessageLoopForUI loop;
+ vector<int> image_indices;
+ image_indices.push_back(0);
+ image_indices.push_back(1);
+ scoped_ptr<base::Value> result =
+ CreateDistilledValueReturnedFromJS(kTitle, kContent, image_indices, "");
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(kURL, CreateMockDistillerPage(result.get(), GURL(kURL)).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kTitle, article_proto_->title());
+ ASSERT_EQ(article_proto_->pages_size(), 1);
+ const DistilledPageProto& first_page = article_proto_->pages(0);
+ EXPECT_EQ(kContent, first_page.html());
+ EXPECT_EQ(kURL, first_page.url());
+ ASSERT_EQ(2, first_page.image_size());
+ EXPECT_EQ(kImageData[0], first_page.image(0).data());
+ EXPECT_EQ(GetImageName(1, 0), first_page.image(0).name());
+ EXPECT_EQ(kImageData[1], first_page.image(1).data());
+ EXPECT_EQ(GetImageName(1, 1), first_page.image(1).name());
+}
+
+TEST_F(DistillerTest, DistillMultiplePages) {
+ base::MessageLoopForUI loop;
+ const size_t kNumPages = 8;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kNumPages);
+
+ // Add images.
+ int next_image_number = 0;
+ for (size_t page_num = 0; page_num < kNumPages; ++page_num) {
+ // Each page has different number of images.
+ size_t tot_images = (page_num + kTotalImages) % (kTotalImages + 1);
+ vector<int> image_indices;
+ for (size_t img_num = 0; img_num < tot_images; img_num++) {
+ image_indices.push_back(next_image_number);
+ next_image_number = (next_image_number + 1) % kTotalImages;
+ }
+ distiller_data->image_ids.push_back(image_indices);
+ }
+
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(
+ distiller_data->page_urls[0],
+ CreateMockDistillerPages(distiller_data.get(), kNumPages, 0).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ VerifyArticleProtoMatchesMultipageData(
+ article_proto_.get(), distiller_data.get(), kNumPages, kNumPages);
+}
+
+TEST_F(DistillerTest, DistillLinkLoop) {
+ base::MessageLoopForUI loop;
+ // Create a loop, the next page is same as the current page. This could
+ // happen if javascript misparses a next page link.
+ scoped_ptr<base::Value> result =
+ CreateDistilledValueReturnedFromJS(kTitle, kContent, vector<int>(), kURL);
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(kURL, CreateMockDistillerPage(result.get(), GURL(kURL)).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kTitle, article_proto_->title());
+ EXPECT_EQ(article_proto_->pages_size(), 1);
+}
+
+TEST_F(DistillerTest, CheckMaxPageLimitExtraPage) {
+ base::MessageLoopForUI loop;
+ const size_t kMaxPagesInArticle = 10;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kMaxPagesInArticle);
+
+ // Note: Next page url of the last page of article is set. So distiller will
+ // try to do kMaxPagesInArticle + 1 calls if the max article limit does not
+ // work.
+ scoped_ptr<base::Value> last_page_data =
+ CreateDistilledValueReturnedFromJS(
+ kTitle,
+ distiller_data->content[kMaxPagesInArticle - 1],
+ vector<int>(),
+ "",
+ distiller_data->page_urls[kMaxPagesInArticle - 2]);
+
+ distiller_data->distilled_values.pop_back();
+ distiller_data->distilled_values.push_back(last_page_data.release());
+
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+
+ distiller_->SetMaxNumPagesInArticle(kMaxPagesInArticle);
+
+ DistillPage(distiller_data->page_urls[0],
+ CreateMockDistillerPages(
+ distiller_data.get(), kMaxPagesInArticle, 0).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kTitle, article_proto_->title());
+ EXPECT_EQ(kMaxPagesInArticle,
+ static_cast<size_t>(article_proto_->pages_size()));
+}
+
+TEST_F(DistillerTest, CheckMaxPageLimitExactLimit) {
+ base::MessageLoopForUI loop;
+ const size_t kMaxPagesInArticle = 10;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kMaxPagesInArticle);
+
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+
+ // Check if distilling an article with exactly the page limit works.
+ distiller_->SetMaxNumPagesInArticle(kMaxPagesInArticle);
+
+ DistillPage(distiller_data->page_urls[0],
+ CreateMockDistillerPages(
+ distiller_data.get(), kMaxPagesInArticle, 0).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kTitle, article_proto_->title());
+ EXPECT_EQ(kMaxPagesInArticle,
+ static_cast<size_t>(article_proto_->pages_size()));
+}
+
+TEST_F(DistillerTest, SinglePageDistillationFailure) {
+ base::MessageLoopForUI loop;
+ // To simulate failure return a null value.
+ scoped_ptr<base::Value> nullValue(base::Value::CreateNullValue());
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(kURL,
+ CreateMockDistillerPage(nullValue.get(), GURL(kURL)).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ("", article_proto_->title());
+ EXPECT_EQ(0, article_proto_->pages_size());
+}
+
+TEST_F(DistillerTest, MultiplePagesDistillationFailure) {
+ base::MessageLoopForUI loop;
+ const size_t kNumPages = 8;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kNumPages);
+
+ // The page number of the failed page.
+ size_t failed_page_num = 3;
+ // reset distilled data of the failed page.
+ distiller_data->distilled_values.erase(
+ distiller_data->distilled_values.begin() + failed_page_num);
+ distiller_data->distilled_values.insert(
+ distiller_data->distilled_values.begin() + failed_page_num,
+ base::Value::CreateNullValue());
+ // Expect only calls till the failed page number.
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(distiller_data->page_urls[0],
+ CreateMockDistillerPages(
+ distiller_data.get(), failed_page_num + 1, 0).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kTitle, article_proto_->title());
+ VerifyArticleProtoMatchesMultipageData(
+ article_proto_.get(), distiller_data.get(), failed_page_num, kNumPages);
+}
+
+TEST_F(DistillerTest, DistillPreviousPage) {
+ base::MessageLoopForUI loop;
+ const size_t kNumPages = 8;
+
+ // The page number of the article on which distillation starts.
+ int start_page_num = 3;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kNumPages);
+
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(distiller_data->page_urls[start_page_num],
+ CreateMockDistillerPages(
+ distiller_data.get(), kNumPages, start_page_num).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ VerifyArticleProtoMatchesMultipageData(
+ article_proto_.get(), distiller_data.get(), kNumPages, kNumPages);
+}
+
+TEST_F(DistillerTest, IncrementalUpdates) {
+ base::MessageLoopForUI loop;
+ const size_t kNumPages = 8;
+
+ // The page number of the article on which distillation starts.
+ int start_page_num = 3;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kNumPages);
+
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(distiller_data->page_urls[start_page_num],
+ CreateMockDistillerPages(
+ distiller_data.get(), kNumPages, start_page_num).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kTitle, article_proto_->title());
+ ASSERT_EQ(kNumPages, static_cast<size_t>(article_proto_->pages_size()));
+ EXPECT_EQ(kNumPages, in_sequence_updates_.size());
+
+ VerifyIncrementalUpdatesMatch(
+ distiller_data.get(), kNumPages, in_sequence_updates_, start_page_num);
+}
+
+TEST_F(DistillerTest, IncrementalUpdatesDoNotDeleteFinalArticle) {
+ base::MessageLoopForUI loop;
+ const size_t kNumPages = 8;
+ int start_page_num = 3;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kNumPages);
+
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(distiller_data->page_urls[start_page_num],
+ CreateMockDistillerPages(
+ distiller_data.get(), kNumPages, start_page_num).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kNumPages, in_sequence_updates_.size());
+
+ in_sequence_updates_.clear();
+
+ // Should still be able to access article and pages.
+ VerifyArticleProtoMatchesMultipageData(
+ article_proto_.get(), distiller_data.get(), kNumPages, kNumPages);
+}
+
+TEST_F(DistillerTest, DeletingArticleDoesNotInterfereWithUpdates) {
+ base::MessageLoopForUI loop;
+ const size_t kNumPages = 8;
+ scoped_ptr<MultipageDistillerData> distiller_data =
+ CreateMultipageDistillerDataWithoutImages(kNumPages);
+ // The page number of the article on which distillation starts.
+ int start_page_num = 3;
+
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(distiller_data->page_urls[start_page_num],
+ CreateMockDistillerPages(
+ distiller_data.get(), kNumPages, start_page_num).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+ EXPECT_EQ(kNumPages, in_sequence_updates_.size());
+ EXPECT_EQ(kTitle, article_proto_->title());
+ ASSERT_EQ(kNumPages, static_cast<size_t>(article_proto_->pages_size()));
+
+ // Delete the article.
+ article_proto_.reset();
+ VerifyIncrementalUpdatesMatch(
+ distiller_data.get(), kNumPages, in_sequence_updates_, start_page_num);
+}
+
+TEST_F(DistillerTest, CancelWithDelayedImageFetchCallback) {
+ base::MessageLoopForUI loop;
+ vector<int> image_indices;
+ image_indices.push_back(0);
+ scoped_ptr<base::Value> distilled_value =
+ CreateDistilledValueReturnedFromJS(kTitle, kContent, image_indices, "");
+ TestDistillerURLFetcher* delayed_fetcher = new TestDistillerURLFetcher(true);
+ MockDistillerURLFetcherFactory mock_url_fetcher_factory;
+ EXPECT_CALL(mock_url_fetcher_factory, CreateDistillerURLFetcher())
+ .WillOnce(Return(delayed_fetcher));
+ distiller_.reset(
+ new DistillerImpl(mock_url_fetcher_factory, DomDistillerOptions()));
+ DistillPage(
+ kURL, CreateMockDistillerPage(distilled_value.get(), GURL(kURL)).Pass());
+ base::MessageLoop::current()->RunUntilIdle();
+
+ // Post callback from the url fetcher and then delete the distiller.
+ delayed_fetcher->PostCallbackTask();
+ distiller_.reset();
+
+ base::MessageLoop::current()->RunUntilIdle();
+}
+
+TEST_F(DistillerTest, CancelWithDelayedJSCallback) {
+ base::MessageLoopForUI loop;
+ scoped_ptr<base::Value> distilled_value =
+ CreateDistilledValueReturnedFromJS(kTitle, kContent, vector<int>(), "");
+ MockDistillerPage* distiller_page = NULL;
+ distiller_.reset(
+ new DistillerImpl(url_fetcher_factory_, DomDistillerOptions()));
+ DistillPage(kURL,
+ CreateMockDistillerPageWithPendingJSCallback(&distiller_page,
+ GURL(kURL)));
+ base::MessageLoop::current()->RunUntilIdle();
+
+ ASSERT_TRUE(distiller_page);
+ // Post the task to execute javascript and then delete the distiller.
+ distiller_page->OnDistillationDone(GURL(kURL), distilled_value.get());
+ distiller_.reset();
+
+ base::MessageLoop::current()->RunUntilIdle();
}
} // namespace dom_distiller