diff --git a/crawler/Readers/LocalReader.cpp b/crawler/Readers/LocalReader.cpp
index 08105912fee20bfce2f947d937149adac3ba0bf3..cef70bae71e13f472349b1950a57834e7298514e 100644
--- a/crawler/Readers/LocalReader.cpp
+++ b/crawler/Readers/LocalReader.cpp
@@ -31,7 +31,7 @@ string LocalReader::PageToString ( )
ParsedUrl LocalReader::getUrl ( )
{
- ParsedUrl url(fileName); //Fixme
+ ParsedUrl url(test_url);
return url;
}
diff --git a/tests/plaintext.txt b/tests/plaintext.txt
index b570f9cfcf71339a326b71dfe96c93b56d7e4c80..5a34b1d751755671058a3879d0f7fb783e13abf6 100644
--- a/tests/plaintext.txt
+++ b/tests/plaintext.txt
@@ -2,7 +2,11 @@
<title>
Apple Ardvark Anteater Alligator
</title>
-<body>
+<p class="text-muted">
Basement Battle Bridge Bottle
-</body>
+</p>
+<p class="text-muted">
+Hello Goodbye <a href="http://veronicaday.com/" class="btn btn-yes">
+Cat Cradle
+</p>
</html>
\ No newline at end of file
diff --git a/util/Stemmer.cpp b/util/Stemmer.cpp
index 7b17f21ca769045a0d82720203451b25c7045ec9..674a93b5626a38ee8cccffd5bc7d618d6d8df605 100644
--- a/util/Stemmer.cpp
+++ b/util/Stemmer.cpp
@@ -49,6 +49,12 @@ int Stemmer::measure ( std::string word )
int m = 0;
unsigned long wordIt = 0;
unsigned long wordEnd = word.size( ) - 1;
+
+ if ( word.empty( ) )
+ {
+ return 0;
+ }
+
// Looking for CVC pattern
while ( wordIt <= wordEnd )
{
@@ -108,6 +114,12 @@ int Stemmer::measure ( std::string word )
*/
bool Stemmer::isVowelPresent ( unsigned long wordBeg, unsigned long wordEnd, string word )
{
+
+ if ( word.empty( ) )
+ {
+ return false;
+ }
+
while ( wordBeg != wordEnd && wordBeg < word.size( ) )
{
if ( !isConsonant( wordBeg, word ) )
@@ -159,6 +171,11 @@ bool Stemmer::isConsonant ( unsigned long wordIt, string word )
*/
bool Stemmer::addE ( string word )
{
+ if ( word.empty( ) )
+ {
+ return false;
+ }
+
// AT -> ATE
// BL -> BLE
// IZ -> IZE
@@ -182,6 +199,11 @@ bool Stemmer::addE ( string word )
*/
bool Stemmer::doubleCon ( string word )
{
+ if ( word.empty( ) )
+ {
+ return false;
+ }
+
unsigned long endWord = word.size( ) - 1;
if ( word.size( ) > 2 && word[ endWord ] == word[ endWord - 1 ] )
@@ -209,6 +231,12 @@ bool Stemmer::doubleCon ( string word )
*/
bool Stemmer::endCVC ( std::string word )
{
+
+ if ( word.empty( ) )
+ {
+ return false;
+ }
+
unsigned long endWord = word.size( ) - 1;
if ( word.size( ) > 2 )
@@ -236,6 +264,11 @@ bool Stemmer::endCVC ( std::string word )
std::string Stemmer::step1a ( std::string word )
{
+ if ( word.empty( ) )
+ {
+ return word;
+ }
+
// check S at end
if ( word[ word.size( ) - 1 ] == 's' && word.size() != 1)
{
@@ -286,6 +319,12 @@ std::string Stemmer::step1a ( std::string word )
*/
std::string Stemmer::step1b ( std::string word )
{
+
+ if ( word.empty( ) )
+ {
+ return word;
+ }
+
unsigned long end = word.size( ) - 1;
auto begPtr = word.begin( );
auto endPtr = begPtr + end;
@@ -349,6 +388,10 @@ std::string Stemmer::step1b ( std::string word )
string Stemmer::step1c ( string word )
{
+ if ( word.empty( ) )
+ {
+ return word;
+ }
// Y -> I
// happy -> happi
// sky -> sky
@@ -371,7 +414,7 @@ string Stemmer::step1c ( string word )
*/
string Stemmer::step2 ( std::string word )
{
- if ( measure( word ) == 0 )
+ if ( measure( word ) == 0 || word.empty( ) )
{
return word;
}
@@ -528,7 +571,7 @@ string Stemmer::step2 ( std::string word )
std::string Stemmer::step3 ( std::string word )
{
- if ( measure( word ) == 0 )
+ if ( measure( word ) == 0 || word.empty( ) )
{
return word;
}
@@ -592,7 +635,7 @@ std::string Stemmer::step3 ( std::string word )
*/
std::string Stemmer::step4 ( std::string word )
{
- if ( measure( word ) <= 2 )
+ if ( measure( word ) <= 2 || word.empty( ) )
{
return word;
}
@@ -730,6 +773,11 @@ std::string Stemmer::step4 ( std::string word )
*/
std::string Stemmer::step5a ( std::string word )
{
+ if ( word.empty( ) )
+ {
+ return word;
+ }
+
auto m = measure( word );
// E ->
// probabte -> probat
@@ -757,6 +805,11 @@ std::string Stemmer::step5a ( std::string word )
*/
std::string Stemmer::step5b ( std::string word )
{
+ if ( word.empty( ) )
+ {
+ return word;
+ }
+
if ( word.size( ) > 2 && measure( word ) > 1 && word[ word.size( ) - 1 ] == 'l' && word[ word.size( ) - 2 ] == 'l' )
{
word = subStr( word, 0, word.size( ) - 1 );
diff --git a/util/tests/stemmerTest.cpp b/util/tests/stemmerTest.cpp
index c49415b87a47bd2cf09d7056b1888ec7db828a6b..79fabfa93c4e9e56cff20f22969c92c1b8d8f34f 100644
--- a/util/tests/stemmerTest.cpp
+++ b/util/tests/stemmerTest.cpp
@@ -110,6 +110,8 @@ int main ( )
assert ( stem.execute( "are" ) == "ar" );
assert ( stem.execute( "terrible" ) == "terribl" );
+ assert ( stem.execute( "" ) == "" );
+ assert ( stem.execute( "s" ) == "s" );
cout << "\nTests passed for Stemmer :D" << endl;