From a7bb7852f2088294602f3e35dab5bd38a94a88fa Mon Sep 17 00:00:00 2001
From: tisfeng <tisfeng@gmail.com>
Date: Thu, 25 Jan 2024 10:17:36 +0800
Subject: [PATCH] perf(ocr): do not check if new paragraph by big line spacing

---
 .../Feature/Service/Apple/EZAppleService.m    | 38 ++++++++++---------
 1 file changed, 20 insertions(+), 18 deletions(-)

diff --git a/Easydict/Feature/Service/Apple/EZAppleService.m b/Easydict/Feature/Service/Apple/EZAppleService.m
index be76679f2..e2a681b15 100644
--- a/Easydict/Feature/Service/Apple/EZAppleService.m
+++ b/Easydict/Feature/Service/Apple/EZAppleService.m
@@ -24,7 +24,7 @@
 
 static NSArray *const kAllowedCharactersInPoetryList = @[ @"《", @"》", @"〔", @"〕" ];
 
-static CGFloat const kParagraphLineHeightRatio = 1.2;
+static CGFloat const kParagraphLineHeightRatio = 1.5;
 
 static NSInteger const kShortPoetryCharacterCountOfLine = 12;
 
@@ -1211,18 +1211,6 @@ - (void)setupOCRResult:(EZOCRResult *)ocrResult
             CGFloat deltaY = prevBoundingBox.origin.y - (boundingBox.origin.y + boundingBox.size.height);
             CGFloat deltaX = boundingBox.origin.x - (prevBoundingBox.origin.x + prevBoundingBox.size.width);
             
-            // Note that line spacing is inaccurate, sometimes it's too small 😢
-            BOOL isNewParagraph = NO;
-            if (deltaY > 0) {
-                // averageLineSpacing may too small, so deltaY should be much larger than averageLineSpacing
-                BOOL isBigLineSpacing = [self isBigSpacingLineOfTextObservation:textObservation
-                                                            prevTextObservation:prevTextObservation
-                                                     greaterThanLineHeightRatio:kParagraphLineHeightRatio];
-                if (isBigLineSpacing) {
-                    isNewParagraph = YES;
-                }
-            }
-            
             // Note that sometimes the line frames will overlap a little, then deltaY will less then 0
             BOOL isNewLine = NO;
             if (deltaY > 0) {
@@ -1251,10 +1239,9 @@ - (void)setupOCRResult:(EZOCRResult *)ocrResult
                 if (isNeedRemoveLastDashOfText) {
                     mergedText = [mergedText substringToIndex:mergedText.length - 1].mutableCopy;
                 }
-            } else if (isNewParagraph || isNewLine) {
+            } else if (isNewLine) {
                 joinedString = [self joinedStringOfTextObservation:textObservation
-                                               prevTextObservation:prevTextObservation
-                                                    isNewParagraph:isNewParagraph];
+                                               prevTextObservation:prevTextObservation];
             } else {
                 joinedString = @" "; // if the same line, just join two texts
             }
@@ -1489,9 +1476,10 @@ - (BOOL)isPoetryOftextObservations:(NSArray<VNRecognizedTextObservation *> *)tex
 /// Get joined string of text, according to its last char.
 - (NSString *)joinedStringOfTextObservation:(VNRecognizedTextObservation *)textObservation
                         prevTextObservation:(VNRecognizedTextObservation *)prevTextObservation
-                             isNewParagraph:(BOOL)isNewParagraph {
+{
     NSString *joinedString = @"";
     BOOL needLineBreak = NO;
+    BOOL isNewParagraph = NO;
     
     CGRect prevBoundingBox = prevTextObservation.boundingBox;
     CGFloat prevLineLength = prevBoundingBox.size.width;
@@ -1509,7 +1497,7 @@ - (NSString *)joinedStringOfTextObservation:(VNRecognizedTextObservation *)textO
     
     BOOL hasPrevIndentation = [self hasIndentationOfTextObservation:prevTextObservation];
     BOOL hasIndentation = [self hasIndentationOfTextObservation:textObservation];
-    
+        
     BOOL isPrevLongText = [self isLongTextObservation:prevTextObservation isStrict:NO];
     
     BOOL isEqualChineseText = [self isEqualChineseTextObservation:textObservation prevTextObservation:prevTextObservation];
@@ -1895,6 +1883,20 @@ - (BOOL)hasIndentationOfTextObservation:(VNRecognizedTextObservation *)textObser
     return hasIndentation;
 }
 
+- (BOOL)hasIndentationOfTextObservation:(VNRecognizedTextObservation *)textObservation
+                    prevTextObservation:(VNRecognizedTextObservation *)prevTextObservation {
+    BOOL isEqualX = [self isEqualXOfTextObservation:textObservation prevTextObservation:prevTextObservation];
+    
+    CGFloat lineX = CGRectGetMinX(textObservation.boundingBox);
+    CGFloat prevLineX = CGRectGetMinX(prevTextObservation.boundingBox);
+    CGFloat dx = lineX - prevLineX;
+    
+    if (!isEqualX && dx < 0) {
+        return YES;
+    }
+    return NO;
+}
+
 - (BOOL)isEqualTextObservation:(VNRecognizedTextObservation *)textObservation
            prevTextObservation:(VNRecognizedTextObservation *)prevTextObservation {
     // 0.06 - 0.025