Add cashtag support.

Use NSMatchingWithoutAnchoringBounds when matching with primary regexp.
twitter-archive · May 19, 2012 · e3c1425 · e3c1425
1 parent 726557c
commit e3c1425
Show file tree

Hide file tree

Showing 6 changed files with 147 additions and 8 deletions.
diff --git a/README.rdoc b/README.rdoc
@@ -30,7 +30,7 @@ https://github.com/twitter/twitter-text-objc/issues
 
 == To Do
 
-* Cashtags support ($AAPL)
+* None
 
 == Authors
 

diff --git a/lib/TwitterText.h b/lib/TwitterText.h
@@ -18,6 +18,7 @@
 + (NSArray*)entitiesInText:(NSString*)text;
 + (NSArray*)URLsInText:(NSString*)text;
 + (NSArray*)hashtagsInText:(NSString*)text checkingURLOverlap:(BOOL)checkingURLOverlap;
++ (NSArray*)cashtagsInText:(NSString*)text checkingURLOverlap:(BOOL)checkingURLOverlap;
 + (NSArray*)mentionedScreenNamesInText:(NSString*)text;
 + (NSArray*)mentionsOrListsInText:(NSString*)text;
 + (TwitterTextEntity*)repliedScreenNameInText:(NSString*)text;

diff --git a/lib/TwitterText.m b/lib/TwitterText.m
@@ -138,7 +138,7 @@
 @"]"
 
 #define TWUHashtagBoundary \
-@"\\A|\\z|[^&a-z0-9_" \
+@"^|$|[^&a-z0-9_" \
     TWULatinAccents \
     TWUNonLatinHashtagChars \
     TWUCJKHashtagCharacters \
@@ -149,6 +149,16 @@
 
 #define TWUEndHashTagMatch      @"\\A(?:[#＃]|://)"
 
+//
+// Cashtag
+//
+
+#define TWUCashtag          @"[a-z]{1,6}(?:[._][a-z]{1,2})?"
+#define TWUValidCashtag \
+    @"(?:^|[" TWUUnicodeSpaces @"])" \
+    @"(\\$" TWUCashtag @")" \
+    @"(?=$|\\s|[" TWUPunctuationChars @"])"
+
 //
 // Mention and list name
 //
@@ -283,12 +293,14 @@
 static NSRegularExpression *validTCOURLRegexp;
 static NSRegularExpression *validHashtagRegexp;
 static NSRegularExpression *endHashtagRegexp;
+static NSRegularExpression *validCashtagRegexp;
 static NSRegularExpression *validMentionOrListRegexp;
 static NSRegularExpression *validReplyRegexp;
 static NSRegularExpression *endMentionRegexp;
 
 @interface TwitterText ()
 + (NSArray*)hashtagsInText:(NSString*)text withURLEntities:(NSArray*)urlEntities;
++ (NSArray*)cashtagsInText:(NSString*)text withURLEntities:(NSArray*)urlEntities;
 @end
 
 @implementation TwitterText
@@ -302,10 +314,14 @@ + (NSArray*)entitiesInText:(NSString*)text
     NSMutableArray *results = [NSMutableArray array];
 
     NSArray *urls = [self URLsInText:text];
-    NSArray *hashtags = [self hashtagsInText:text withURLEntities:urls];
     [results addObjectsFromArray:urls];
+
+    NSArray *hashtags = [self hashtagsInText:text withURLEntities:urls];
     [results addObjectsFromArray:hashtags];
 
+    NSArray *cashtags = [self cashtagsInText:text withURLEntities:urls];
+    [results addObjectsFromArray:cashtags];
+
     NSArray *mentionsAndLists = [self mentionsOrListsInText:text];
     NSMutableArray *addingItems = [NSMutableArray array];
 
@@ -361,7 +377,7 @@ + (NSArray*)URLsInText:(NSString*)text
 
     while (1) {
         position = NSMaxRange(allRange);
-        NSTextCheckingResult *urlResult = [validURLRegexp firstMatchInString:text options:0 range:NSMakeRange(position, len - position)];
+        NSTextCheckingResult *urlResult = [validURLRegexp firstMatchInString:text options:NSMatchingWithoutAnchoringBounds range:NSMakeRange(position, len - position)];
         if (!urlResult || urlResult.numberOfRanges < 9) {
             break;
         }
@@ -466,7 +482,7 @@ + (NSArray*)hashtagsInText:(NSString*)text withURLEntities:(NSArray*)urlEntities
     NSInteger position = 0;
 
     while (1) {
-        NSTextCheckingResult *matchResult = [validHashtagRegexp firstMatchInString:text options:0 range:NSMakeRange(position, len - position)];
+        NSTextCheckingResult *matchResult = [validHashtagRegexp firstMatchInString:text options:NSMatchingWithoutAnchoringBounds range:NSMakeRange(position, len - position)];
         if (!matchResult || matchResult.numberOfRanges < 2) {
             break;
         }
@@ -503,6 +519,61 @@ + (NSArray*)hashtagsInText:(NSString*)text withURLEntities:(NSArray*)urlEntities
     return results;
 }
 
++ (NSArray*)cashtagsInText:(NSString*)text checkingURLOverlap:(BOOL)checkingURLOverlap
+{
+    if (!text.length) {
+        return [NSArray array];
+    }
+
+    NSArray *urls = nil;
+    if (checkingURLOverlap) {
+        urls = [self URLsInText:text];
+    }
+    return [self cashtagsInText:text withURLEntities:urls];
+}
+
++ (NSArray*)cashtagsInText:(NSString*)text withURLEntities:(NSArray*)urlEntities
+{
+    if (!text.length) {
+        return [NSArray array];
+    }
+
+    if (!validCashtagRegexp) {
+        validCashtagRegexp = [[NSRegularExpression alloc] initWithPattern:TWUValidCashtag options:NSRegularExpressionCaseInsensitive error:NULL];
+    }
+
+    NSMutableArray *results = [NSMutableArray array];
+    NSInteger len = text.length;
+    NSInteger position = 0;
+
+    while (1) {
+        NSTextCheckingResult *matchResult = [validCashtagRegexp firstMatchInString:text options:NSMatchingWithoutAnchoringBounds range:NSMakeRange(position, len - position)];
+        if (!matchResult || matchResult.numberOfRanges < 2) {
+            break;
+        }
+
+        NSRange cashtagRange = [matchResult rangeAtIndex:1];
+        BOOL matchOk = YES;
+
+        // Check URL overlap
+        for (TwitterTextEntity *urlEntity in urlEntities) {
+            if (NSIntersectionRange(urlEntity.range, cashtagRange).length > 0) {
+                matchOk = NO;
+                break;
+            }
+        }
+
+        if (matchOk) {
+            TwitterTextEntity *entity = [TwitterTextEntity entityWithType:TwitterTextEntityCashtag range:cashtagRange];
+            [results addObject:entity];
+        }
+
+        position = NSMaxRange(matchResult.range);
+    }
+
+    return results;
+}
+
 + (NSArray*)mentionedScreenNamesInText:(NSString*)text
 {
     if (!text.length) {
@@ -539,7 +610,7 @@ + (NSArray*)mentionsOrListsInText:(NSString*)text
     NSInteger position = 0;
 
     while (1) {
-        NSTextCheckingResult *matchResult = [validMentionOrListRegexp firstMatchInString:text options:0 range:NSMakeRange(position, len - position)];
+        NSTextCheckingResult *matchResult = [validMentionOrListRegexp firstMatchInString:text options:NSMatchingWithoutAnchoringBounds range:NSMakeRange(position, len - position)];
         if (!matchResult || matchResult.numberOfRanges < 5) {
             break;
         }
@@ -586,7 +657,7 @@ + (TwitterTextEntity*)repliedScreenNameInText:(NSString*)text
 
     NSInteger len = text.length;
 
-    NSTextCheckingResult *matchResult = [validReplyRegexp firstMatchInString:text options:0 range:NSMakeRange(0, len)];
+    NSTextCheckingResult *matchResult = [validReplyRegexp firstMatchInString:text options:(NSMatchingWithoutAnchoringBounds | NSMatchingAnchored) range:NSMakeRange(0, len)];
     if (!matchResult || matchResult.numberOfRanges < 2) {
         return nil;
     }
@@ -657,7 +728,7 @@ + (int)remainingCharacterCount:(NSString*)text httpURLLength:(int)httpURLLength
         TwitterTextEntity *entity = [urlEntities objectAtIndex:i];
         NSRange urlRange = entity.range;
         NSString *url = [string substringWithRange:urlRange];
-        if ([url rangeOfString:@"https" options:NSCaseInsensitiveSearch | NSAnchoredSearch].location == 0) {
+        if ([url rangeOfString:@"https" options:(NSCaseInsensitiveSearch | NSAnchoredSearch)].location == 0) {
             urlLengthOffset += httpsURLLength;
         } else {
             urlLengthOffset += httpURLLength;

diff --git a/lib/TwitterTextEntity.h b/lib/TwitterTextEntity.h
@@ -17,6 +17,7 @@ typedef enum {
     TwitterTextEntityScreenName,
     TwitterTextEntityHashtag,
     TwitterTextEntityListName,
+    TwitterTextEntityCashtag,
 } TwitterTextEntityType;
 
 @interface TwitterTextEntity : NSObject

diff --git a/lib/TwitterTextEntity.m b/lib/TwitterTextEntity.m
@@ -73,6 +73,9 @@ - (NSString*)description
         case TwitterTextEntityListName:
             typeString = @"ListName";
             break;
+        case TwitterTextEntityCashtag:
+            typeString = @"Cashtag";
+            break;
     }
     return [NSString stringWithFormat:@"<%@: %@ %@>", NSStringFromClass([self class]), typeString, NSStringFromRange(range)];
 }

diff --git a/testproject/TwitterTextTests.m b/testproject/TwitterTextTests.m
@@ -51,6 +51,8 @@ - (void)testExtract
     NSArray *urlsWithIndices = [tests objectForKey:@"urls_with_indices"];
     NSArray *hashtags = [tests objectForKey:@"hashtags"];
     NSArray *hashtagsWithIndices = [tests objectForKey:@"hashtags_with_indices"];
+    NSArray *cashtags = [tests objectForKey:@"cashtags"];
+    NSArray *cashtagsWithIndices = [tests objectForKey:@"cashtags_with_indices"];
 
     //
     // Mentions
@@ -290,6 +292,67 @@ - (void)testExtract
             STFail(@"Matching count is different: %lu != %lu\n%@", expected.count, results.count, testCase);
         }
     }
+
+    //
+    // Cashtag
+    //
+
+    for (NSDictionary *testCase in cashtags) {
+        NSString *text = [testCase objectForKey:@"text"];
+        NSArray *expected = [testCase objectForKey:@"expected"];
+
+        NSArray *results = [TwitterText cashtagsInText:text checkingURLOverlap:YES];
+        if (results.count == expected.count) {
+            int count = results.count;
+            for (int i=0; i<count; i++) {
+                NSString *expectedText = [expected objectAtIndex:i];
+
+                TwitterTextEntity *entity = [results objectAtIndex:i];
+                NSRange r = entity.range;
+                r.location++;
+                r.length--;
+                NSString *actualText = [text substringWithRange:r];
+
+                STAssertEqualObjects(expectedText, actualText, @"%@", testCase);
+            }
+        } else {
+            STFail(@"Matching count is different: %lu != %lu\n%@", expected.count, results.count, testCase);
+            NSLog(@"### %@", results);
+        }
+    }
+
+    //
+    // Cashtags with indices
+    //
+    for (NSDictionary *testCase in cashtagsWithIndices) {
+        NSString *text = [testCase objectForKey:@"text"];
+        NSArray *expected = [testCase objectForKey:@"expected"];
+
+        NSArray *results = [TwitterText cashtagsInText:text checkingURLOverlap:YES];
+        if (results.count == expected.count) {
+            int count = results.count;
+            for (int i=0; i<count; i++) {
+                NSDictionary *expectedDic = [expected objectAtIndex:i];
+                NSString *expectedCashtag = [expectedDic objectForKey:@"cashtag"];
+                NSArray *expectedIndices = [expectedDic objectForKey:@"indices"];
+                int expectedStart = [[expectedIndices objectAtIndex:0] intValue];
+                int expectedEnd = [[expectedIndices objectAtIndex:1] intValue];
+                NSRange expectedRange = NSMakeRange(expectedStart, expectedEnd - expectedStart);
+
+                TwitterTextEntity *entity = [results objectAtIndex:i];
+                NSRange actualRange = entity.range;
+                NSRange r = actualRange;
+                r.location++;
+                r.length--;
+                NSString *actualText = [text substringWithRange:r];
+
+                STAssertEqualObjects(expectedCashtag, actualText, @"%@", testCase);
+                STAssertTrue(NSEqualRanges(expectedRange, actualRange), @"%@ != %@\n%@", NSStringFromRange(expectedRange), NSStringFromRange(actualRange), testCase);
+            }
+        } else {
+            STFail(@"Matching count is different: %lu != %lu\n%@", expected.count, results.count, testCase);
+        }
+    }
 }
 
 @end