LanguageConverter fix of empty and numeric strings

Bug: T51072
Bug: T48634
Bug: T53551
Change-Id: I2c88f1cf7c0014bebf5c798916b660b334a0b78b

diff --git a/includes/libs/ReplacementArray.php b/includes/libs/ReplacementArray.php
index 7fdb309..185914c 100644 (file)
--- a/includes/libs/ReplacementArray.php
+++ b/includes/libs/ReplacementArray.php
@@ -76,7 +76,7 @@ class ReplacementArray {
         * @param array $data
         */
        public function mergeArray( $data ) {
-               $this->data = array_merge( $this->data, $data );
+               $this->data = $data + $this->data;
                $this->fss = false;
        }
 
@@ -84,7 +84,7 @@ class ReplacementArray {
         * @param ReplacementArray $other
         */
        public function merge( ReplacementArray $other ) {
-               $this->data = array_merge( $this->data, $other->data );
+               $this->data = $other->data + $this->data;
                $this->fss = false;
        }
 

diff --git a/languages/ConverterRule.php b/languages/ConverterRule.php
index e6625c1..e3a4f77 100644 (file)
--- a/languages/ConverterRule.php
+++ b/languages/ConverterRule.php
@@ -155,18 +155,20 @@ class ConverterRule {
                        $to = trim( $v[1] );
                        $v = trim( $v[0] );
                        $u = explode( '=>', $v, 2 );
-                       // if $to is empty, strtr() could return a wrong result
-                       if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
+                       // if $to is empty (which is also used as $from in bidtable),
+                       // strtr() could return a wrong result.
+                       if ( count( $u ) == 1 && $to !== '' && in_array( $v, $variants ) ) {
                                $bidtable[$v] = $to;
                        } elseif ( count( $u ) == 2 ) {
                                $from = trim( $u[0] );
                                $v = trim( $u[1] );
+                               // if $from is empty, strtr() could return a wrong result.
                                if ( array_key_exists( $v, $unidtable )
                                        && !is_array( $unidtable[$v] )
-                                       && $to
+                                       && $from !== ''
                                        && in_array( $v, $variants ) ) {
                                        $unidtable[$v] = array( $from => $to );
-                               } elseif ( $to && in_array( $v, $variants ) ) {
+                               } elseif ( $from !== '' && in_array( $v, $variants ) ) {
                                        $unidtable[$v][$from] = $to;
                                }
                        }
@@ -220,17 +222,17 @@ class ConverterRule {
                        // display current variant in bidirectional array
                        $disp = $this->getTextInBidtable( $variant );
                        // or display current variant in fallbacks
-                       if ( !$disp ) {
+                       if ( $disp === false ) {
                                $disp = $this->getTextInBidtable(
                                        $this->mConverter->getVariantFallbacks( $variant ) );
                        }
                        // or display current variant in unidirectional array
-                       if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
+                       if ( $disp === false && array_key_exists( $variant, $unidtable ) ) {
                                $disp = array_values( $unidtable[$variant] );
                                $disp = $disp[0];
                        }
                        // or display frist text under disable manual convert
-                       if ( !$disp && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
+                       if ( $disp === false && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
                                if ( count( $bidtable ) > 0 ) {
                                        $disp = array_values( $bidtable );
                                        $disp = $disp[0];
@@ -325,7 +327,7 @@ class ConverterRule {
                                && isset( $unidtable[$v] )
                        ) {
                                if ( isset( $this->mConvTable[$v] ) ) {
-                                       $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
+                                       $this->mConvTable[$v] = $unidtable[$v] + $this->mConvTable[$v];
                                } else {
                                        $this->mConvTable[$v] = $unidtable[$v];
                                }
@@ -383,9 +385,11 @@ class ConverterRule {
 
                if ( !$this->mBidtable && !$this->mUnidtable ) {
                        if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
-                               // fill all variants if text in -{A/H/-|text} without rules
-                               foreach ( $this->mConverter->mVariants as $v ) {
-                                       $this->mBidtable[$v] = $rules;
+                               // fill all variants if text in -{A/H/-|text}- is non-empty but without rules
+                               if ( $rules !== '' ) {
+                                       foreach ( $this->mConverter->mVariants as $v ) {
+                                               $this->mBidtable[$v] = $rules;
+                                       }
                                }
                        } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
                                $this->mFlags = $flags = array( 'R' => true );

diff --git a/languages/LanguageConverter.php b/languages/LanguageConverter.php
index a6b687c..7879309 100644 (file)
--- a/languages/LanguageConverter.php
+++ b/languages/LanguageConverter.php
@@ -502,13 +502,9 @@ class LanguageConverter {
                        }
 
                        if ( $action == 'add' ) {
+                               // More efficient than array_merge(), about 2.5 times.
                                foreach ( $pair as $from => $to ) {
-                                       // to ensure that $from and $to not be left blank
-                                       // so $this->translate() could always return a string
-                                       if ( $from || $to ) {
-                                               // more efficient than array_merge(), about 2.5 times.
-                                               $this->mTables[$variant]->setPair( $from, $to );
-                                       }
+                                       $this->mTables[$variant]->setPair( $from, $to );
                                }
                        } elseif ( $action == 'remove' ) {
                                $this->mTables[$variant]->removeArray( $pair );
@@ -996,7 +992,7 @@ class LanguageConverter {
                if ( $recursive ) {
                        foreach ( $sublinks as $link ) {
                                $s = $this->parseCachedTable( $code, $link, $recursive );
-                               $ret = array_merge( $ret, $s );
+                               $ret = $s + $ret;
                        }
                }
 

diff --git a/tests/parser/parserTests.txt b/tests/parser/parserTests.txt
index e965352..18d9aa8 100644 (file)
--- a/tests/parser/parserTests.txt
+++ b/tests/parser/parserTests.txt
@@ -18293,6 +18293,61 @@ Raw: -{R|zh:China;zh-tw:Taiwan}-
 </p>
 !! end
 
+!! test
+Strings evaluating false shouldn't be ignored by Language converter (T51072)
+!! options
+language=zh variant=zh-cn
+!! input
+-{zh-cn:0;zh-sg:1;zh-tw:2;zh-hk:3}-
+!! result
+<p>0
+</p>
+!! end
+
+!! test
+Conversion rules from [numeric-only string] to [something else] (T48634)
+!! options
+language=zh variant=zh-cn
+!! input
+-{H|0=>zh-cn:B}--{H|0=>zh-cn:C;0=>zh-cn:D}--{H|0=>zh-hans:A}-012345-{A|zh-tw:0;zh-cn:E;}-012345
+!! result
+<p>D12345EE12345
+</p>
+!! end
+
+!! test
+Bidirectional converter rule entries with an empty value should be ignored (T53551)
+!! options
+language=zh variant=zh-cn
+!! input
+-{H|zh-cn:foo;zh-tw:;}-foobar
+!! result
+<p>foobar
+</p>
+!! end
+
+!! test
+Unidirectional converter rule entries with an empty "from" string should be ignored (T53551)
+!! options
+language=zh variant=zh-cn
+!! input
+-{H|=>zh-cn:foo;}-foobar
+!! result
+<p>foobar
+</p>
+!! end
+
+!! test
+Empty converter rule entries shouldn't be inserted into the conversion table (T53551)
+!! options
+language=zh variant=zh-cn
+!! input
+-{H|}-foobar
+!! result
+<p>foobar
+</p>
+!! end
+
 !! test
 Nested using of manual convert syntax
 !! options