privacore-open-source-searc…/test/unit/SummaryTest.cpp

#include <gtest/gtest.h>

#include "Summary.h"
#include "HttpMime.h" // CT_HTML
#include <cstdio>

#include "Xml.h"
#include "Words.h"
#include "Phrases.h"
#include "Sections.h"
#include "Pos.h"
#include "Query.h"
#include "Url.h"
#include "Matches.h"
#include "Linkdb.h"
#include "Title.h"

#define MAX_BUF_SIZE 1024
#define HTML_FORMAT "<html><head>%s</head><body>%s</body></html>"

static void generateSummary( Summary &summary, char *htmlInput, const char *queryStr, const char *urlStr ) {
	Xml xml;
	ASSERT_TRUE(xml.set(htmlInput, strlen(htmlInput), 0, CT_HTML));

	Words words;
	ASSERT_TRUE(words.set(&xml, true));

	Bits bits;
	ASSERT_TRUE(bits.set(&words));

	Url url;
	url.set(urlStr);

	Sections sections;
	ASSERT_TRUE(sections.set(&words, &bits, &url, "", CT_HTML));

	Query query;
	ASSERT_TRUE(query.set2(queryStr, langEnglish, true, true));

	LinkInfo linkInfo;
	memset ( &linkInfo , 0 , sizeof(LinkInfo) );
	linkInfo.m_lisize = sizeof(LinkInfo);

	Title title;
	ASSERT_TRUE(title.setTitle(&xml, &words, 80, &query, &linkInfo, &url, NULL, 0, CT_HTML, langEnglish));

	Pos pos;
	ASSERT_TRUE(pos.set(&words));

	Bits bitsForSummary;
	ASSERT_TRUE(bitsForSummary.setForSummary(&words));

	Phrases phrases;
	ASSERT_TRUE(phrases.set(&words, &bits));

	Matches matches;
	matches.setQuery(&query);
	ASSERT_TRUE(matches.set(&words, &phrases, &sections, &bitsForSummary, &pos, &xml, &title, &url, &linkInfo));

	summary.setSummary(&xml, &words, &sections, &pos, &query, 180, 3, 3, 180, &url, &matches, title.getTitle(), title.getTitleLen());
}

TEST( SummaryTest, StripSamePunct ) {
	const char *body =
	   "<pre>"
	   "---------------------------------------------------------------------------------\n"
	   "|                      Name                      |       Total Donations        |\n"
	   "---------------------------------------------------------------------------------\n"
	   "| JENNI STANLEY                                  |                       $10.00 |\n"
	   "---------------------------------------------------------------------------------\n"
	   "| CANDRA BUDGE                                   |                       $22.00 |\n"
	   "---------------------------------------------------------------------------------\n"
	   "| JESSE NICLEY                                   |                       $34.00 |\n"
	   "---------------------------------------------------------------------------------\n"
	   "| SHARON YOKLEY                                  |                       $45.00 |\n"
	   "---------------------------------------------------------------------------------\n"
	   "</pre>";

	char input[MAX_BUF_SIZE];
	std::sprintf(input, HTML_FORMAT, "", body);

	Summary summary;
	generateSummary(summary, input, "jesse budge", "http://www.example.com/");

	EXPECT_STREQ("CANDRA BUDGE | $22.00 | … | JESSE NICLEY | $34.00 …", summary.getSummary());
}

TEST( SummaryTest, DISABLED_BUGNoEllipsisAdded ) {
	const char *head =
		"<title>Instrument prices by Acme Inc.</title>\n"
		"<meta name=\"description\" content=\"Unorthodox musical instrument value estimation\">\n";

	const char *body =
		"<h1>Unusual saxophone valuation</h1>\n"
		"<p>Looking for knowing how much your saxophone is worth and what an appropriate insurance should be?. We provide that and other relevant information such as procedures, locations and time tables</p>\n"
		"<p>We also provide valuation for other musical instrucments.</p>\n";

	char input[MAX_BUF_SIZE];
	std::sprintf(input, HTML_FORMAT, head, body);

	Summary summary;
	generateSummary(summary, input, "saxophone", "http://www.example.com/");

	/// @todo ALC we're not adding ellipsis here due to lack of space. we should take one less word instead and add ellipsis.
	EXPECT_STREQ( "Unusual saxophone valuation. Looking for knowing how much your saxophone is worth and what an appropriate insurance should be?. We provide that and other relevant information …", summary.getSummary() );
}

TEST( SummaryTest, DISABLED_BUGEllipsisAdded ) {
	const char *body = "Giraffe on rollerblades. Penguin on skateboard. The giraffe is way faster than that plumb bird with pathetic wings.\n";

	char input[MAX_BUF_SIZE];
	std::sprintf(input, "%s", body);

	Summary summary;
	generateSummary(summary, input, "giraffe", "http://www.example.com/");

	/// @todo ALC we're adding ellipsis even with a full sentence.
	EXPECT_STREQ( "Giraffe on rollerblades. Penguin on skateboard. The giraffe is way faster than that plumb bird with pathetic wings.", summary.getSummary() );
}

TEST( SummaryTest, DefaultSummary ) {
	const char *head = "<qtitle>f1 doc</qtitle>";
	const char *body = "<p>cucumber</p>\n"
	                   "<a href=\"f3.html\">snegl</a>\n"
	                   "snegl\n";

	char input[MAX_BUF_SIZE];
	std::sprintf(input, HTML_FORMAT, head, body);

	Summary summary;
	generateSummary(summary, input, "banana", "http://www.example.com/");

	EXPECT_STREQ( "cucumber. snegl snegl", summary.getSummary() );
}
Use angle brace instead of quote to include gtest.h 2016-09-23 11:05:22 +02:00			`#include <gtest/gtest.h>`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
			`#include "Summary.h"`
			`#include "HttpMime.h" // CT_HTML`
			`#include <cstdio>`

			`#include "Xml.h"`
			`#include "Words.h"`
			`#include "Phrases.h"`
			`#include "Sections.h"`
			`#include "Pos.h"`
			`#include "Query.h"`
			`#include "Url.h"`
			`#include "Matches.h"`
			`#include "Linkdb.h"`
			`#include "Title.h"`

			`#define MAX_BUF_SIZE 1024`
			`#define HTML_FORMAT "<html><head>%s</head><body>%s</body></html>"`

Add another unit test to show bug in ellipsis 2016-05-24 16:07:46 +02:00			`static void generateSummary( Summary &summary, char htmlInput, const char queryStr, const char *urlStr ) {`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00			`Xml xml;`
Remove niceness and quickpoll from Xml 2016-09-23 00:54:04 +02:00			`ASSERT_TRUE(xml.set(htmlInput, strlen(htmlInput), 0, CT_HTML));`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
			`Words words;`
			`ASSERT_TRUE(words.set(&xml, true));`

			`Bits bits;`
Remove niceness from Bits 2016-09-23 01:15:54 +02:00			`ASSERT_TRUE(bits.set(&words));`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
			`Url url;`
			`url.set(urlStr);`

			`Sections sections;`
Fix unittest 2016-09-23 02:30:19 +02:00			`ASSERT_TRUE(sections.set(&words, &bits, &url, "", CT_HTML));`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
			`Query query;`
Fix compilation error 2016-12-22 12:29:49 +01:00			`ASSERT_TRUE(query.set2(queryStr, langEnglish, true, true));`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
			`LinkInfo linkInfo;`
Remove same punctuation in summary 2016-02-18 22:18:42 +01:00			`memset ( &linkInfo , 0 , sizeof(LinkInfo) );`
			`linkInfo.m_lisize = sizeof(LinkInfo);`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
			`Title title;`
Fix unittest 2016-09-23 02:30:19 +02:00			`ASSERT_TRUE(title.setTitle(&xml, &words, 80, &query, &linkInfo, &url, NULL, 0, CT_HTML, langEnglish));`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
			`Pos pos;`
			`ASSERT_TRUE(pos.set(&words));`

			`Bits bitsForSummary;`
			`ASSERT_TRUE(bitsForSummary.setForSummary(&words));`

Remove SectionVotingTable & passing Phrases into Sections 2016-02-23 14:35:40 +01:00			`Phrases phrases;`
Remove niceness from Phrases 2016-09-23 02:40:50 +02:00			`ASSERT_TRUE(phrases.set(&words, &bits));`
Remove SectionVotingTable & passing Phrases into Sections 2016-02-23 14:35:40 +01:00
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00			`Matches matches;`
			`matches.setQuery(&query);`
Fix unit tests 2016-09-23 03:02:09 +02:00			`ASSERT_TRUE(matches.set(&words, &phrases, &sections, &bitsForSummary, &pos, &xml, &title, &url, &linkInfo));`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00
Remove not set Msg20::ptr_termFreq variable 2016-03-01 14:49:03 +01:00			`summary.setSummary(&xml, &words, &sections, &pos, &query, 180, 3, 3, 180, &url, &matches, title.getTitle(), title.getTitleLen());`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00			`}`

Add unit test to demonstrate behaviour where ellipsis is not added. Yet another place to use horizontal ellipsis instead of three dots 2016-05-24 13:07:59 +02:00			`TEST( SummaryTest, StripSamePunct ) {`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00			`const char *body =`
			`"<pre>"`
Remove same punctuation in summary 2016-02-18 22:18:42 +01:00			`"---------------------------------------------------------------------------------\n"`
			`"\| Name \| Total Donations \|\n"`
			`"---------------------------------------------------------------------------------\n"`
			`"\| JENNI STANLEY \| $10.00 \|\n"`
			`"---------------------------------------------------------------------------------\n"`
			`"\| CANDRA BUDGE \| $22.00 \|\n"`
			`"---------------------------------------------------------------------------------\n"`
			`"\| JESSE NICLEY \| $34.00 \|\n"`
			`"---------------------------------------------------------------------------------\n"`
			`"\| SHARON YOKLEY \| $45.00 \|\n"`
			`"---------------------------------------------------------------------------------\n"`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00			`"</pre>";`

			`char input[MAX_BUF_SIZE];`
			`std::sprintf(input, HTML_FORMAT, "", body);`

Remove same punctuation in summary 2016-02-18 22:18:42 +01:00			`Summary summary;`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00			`generateSummary(summary, input, "jesse budge", "http://www.example.com/");`

Partially fix broken unit test from ellipsis changes 2016-05-13 11:31:28 +02:00			`EXPECT_STREQ("CANDRA BUDGE \| $22.00 \| … \| JESSE NICLEY \| $34.00 …", summary.getSummary());`
Initial version of Summary unit test 2016-02-18 17:10:18 +01:00			`}`
Add unit test to demonstrate behaviour where ellipsis is not added. Yet another place to use horizontal ellipsis instead of three dots 2016-05-24 13:07:59 +02:00
Make test fail (bug with ellipsis in existing code that happens in corner cases) & disable them. 2016-11-11 16:48:25 +01:00			`TEST( SummaryTest, DISABLED_BUGNoEllipsisAdded ) {`
Add unit test to demonstrate behaviour where ellipsis is not added. Yet another place to use horizontal ellipsis instead of three dots 2016-05-24 13:07:59 +02:00			`const char *head =`
			`"<title>Instrument prices by Acme Inc.</title>\n"`
			`"<meta name=\"description\" content=\"Unorthodox musical instrument value estimation\">\n";`

			`const char *body =`
			`"<h1>Unusual saxophone valuation</h1>\n"`
			`"<p>Looking for knowing how much your saxophone is worth and what an appropriate insurance should be?. We provide that and other relevant information such as procedures, locations and time tables</p>\n"`
			`"<p>We also provide valuation for other musical instrucments.</p>\n";`

			`char input[MAX_BUF_SIZE];`
			`std::sprintf(input, HTML_FORMAT, head, body);`

			`Summary summary;`
			`generateSummary(summary, input, "saxophone", "http://www.example.com/");`

Add/change todo comments 2016-06-01 11:10:06 +02:00			`/// @todo ALC we're not adding ellipsis here due to lack of space. we should take one less word instead and add ellipsis.`
Make test fail (bug with ellipsis in existing code that happens in corner cases) & disable them. 2016-11-11 16:48:25 +01:00			`EXPECT_STREQ( "Unusual saxophone valuation. Looking for knowing how much your saxophone is worth and what an appropriate insurance should be?. We provide that and other relevant information …", summary.getSummary() );`
Add another unit test to show bug in ellipsis 2016-05-24 16:07:46 +02:00			`}`

Make test fail (bug with ellipsis in existing code that happens in corner cases) & disable them. 2016-11-11 16:48:25 +01:00			`TEST( SummaryTest, DISABLED_BUGEllipsisAdded ) {`
Add another unit test to show bug in ellipsis 2016-05-24 16:07:46 +02:00			`const char *body = "Giraffe on rollerblades. Penguin on skateboard. The giraffe is way faster than that plumb bird with pathetic wings.\n";`

			`char input[MAX_BUF_SIZE];`
			`std::sprintf(input, "%s", body);`

			`Summary summary;`
			`generateSummary(summary, input, "giraffe", "http://www.example.com/");`

Add/change todo comments 2016-06-01 11:10:06 +02:00			`/// @todo ALC we're adding ellipsis even with a full sentence.`
Make test fail (bug with ellipsis in existing code that happens in corner cases) & disable them. 2016-11-11 16:48:25 +01:00			`EXPECT_STREQ( "Giraffe on rollerblades. Penguin on skateboard. The giraffe is way faster than that plumb bird with pathetic wings.", summary.getSummary() );`
Fix summary generation for default summary 2016-06-09 17:06:29 +02:00			`}`

			`TEST( SummaryTest, DefaultSummary ) {`
			`const char *head = "<qtitle>f1 doc</qtitle>";`
			`const char *body = "<p>cucumber</p>\n"`
Change sample document to simpler example 2016-06-09 18:09:05 +02:00			`"<a href=\"f3.html\">snegl</a>\n"`
Fix summary generation for default summary 2016-06-09 17:06:29 +02:00			`"snegl\n";`

			`char input[MAX_BUF_SIZE];`
			`std::sprintf(input, HTML_FORMAT, head, body);`

			`Summary summary;`
			`generateSummary(summary, input, "banana", "http://www.example.com/");`

			`EXPECT_STREQ( "cucumber. snegl snegl", summary.getSummary() );`
Remove niceness and quickpoll from Xml 2016-09-23 00:54:04 +02:00			`}`