文本分析器 - 免费单词计数和字符统计工具

1import re
2from collections import Counter
3
4def analyze_text(text):
5    if not text or not text.strip():
6        return {
7            'word_count': 0,
8            'char_count_with_spaces': 0,
9            'char_count_without_spaces': 0,
10            'sentence_count': 0,
11            'paragraph_count': 0,
12            'avg_words_per_sentence': 0,
13            'top_words': [],
14            'reading_time': '0 秒'
15        }
16    
17    # 单词数
18    words = text.split()
19    word_count = len(words)
20    
21    # 字符数
22    char_count_with_spaces = len(text)
23    char_count_without_spaces = len(re.sub(r'\s+', '', text))
24    
25    # 句子数
26    sentences = re.findall(r'[.!?]+', text)
27    sentence_count = max(1, len(sentences))
28    
29    # 段落数
30    paragraphs = [p for p in text.split('\n') if p.strip()]
31    paragraph_count = max(1, len(paragraphs))
32    
33    # 每句平均单词数
34    avg_words_per_sentence = round(word_count / sentence_count, 1)
35    
36    # 前5个频繁单词
37    clean_words = [re.sub(r'[^a-z0-9]', '', word.lower()) 
38                   for word in words]
39    clean_words = [w for w in clean_words if w]
40    word_freq = Counter(clean_words)
41    top_words = [{'word': word, 'count': count} 
42                 for word, count in word_freq.most_common(5)]
43    
44    # 阅读时间（每分钟225个单词）
45    minutes = word_count // 225
46    seconds = round((word_count % 225) / 225 * 60)
47    reading_time = f"{minutes} 分 {seconds} 秒" if minutes > 0 else f"{seconds} 秒"
48    
49    return {
50        'word_count': word_count,
51        'char_count_with_spaces': char_count_with_spaces,
52        'char_count_without_spaces': char_count_without_spaces,
53        'sentence_count': sentence_count,
54        'paragraph_count': paragraph_count,
55        'avg_words_per_sentence': avg_words_per_sentence,
56        'top_words': top_words,
57        'reading_time': reading_time
58    }
59
60# 使用示例：
61sample_text = "你好世界！这是一个文本分析器。它可以统计单词等信息。"
62results = analyze_text(sample_text)
63print(results)
64

1import java.util.*;
2import java.util.regex.*;
3import java.util.stream.*;
4
5public class TextAnalyzer {
6    
7    public static class AnalysisResult {
8        public int wordCount;
9        public int charCountWithSpaces;
10        public int charCountWithoutSpaces;
11        public int sentenceCount;
12        public int paragraphCount;
13        public double avgWordsPerSentence;
14        public List<WordFrequency> topWords;
15        public String readingTime;
16        
17        public static class WordFrequency {
18            public String word;
19            public int count;
20            
21            public WordFrequency(String word, int count) {
22                this.word = word;
23                this.count = count;
24            }
25        }
26    }
27    
28    public static AnalysisResult analyzeText(String text) {
29        AnalysisResult result = new AnalysisResult();
30        
31        if (text == null || text.trim().isEmpty()) {
32            result.wordCount = 0;
33            result.charCountWithSpaces = 0;
34            result.charCountWithoutSpaces = 0;
35            result.sentenceCount = 0;
36            result.paragraphCount = 0;
37            result.avgWordsPerSentence = 0;
38            result.topWords = new ArrayList<>();
39            result.readingTime = "0 秒";
40            return result;
41        }
42        
43        // 单词数
44        String[] words = text.trim().split("\\s+");
45        result.wordCount = words.length;
46        
47        // 字符数
48        result.charCountWithSpaces = text.length();
49        result.charCountWithoutSpaces = text.replaceAll("\\s+", "").length();
50        
51        // 句子数
52        Pattern sentencePattern = Pattern.compile("[.!?]+");
53        Matcher sentenceMatcher = sentencePattern.matcher(text);
54        result.sentenceCount = Math.max(1, (int) sentenceMatcher.results().count());
55        
56        // 段落数
57        String[] paragraphs = text.split("\n+");
58        result.paragraphCount = Math.max(1, 
59            (int) Arrays.stream(paragraphs).filter(p -> !p.trim().isEmpty()).count());
60        
61        // 每句平均单词数
62        result.avgWordsPerSentence = 
63            Math.round((double) result.wordCount / result.sentenceCount * 10.0) / 10.0;
64        
65        // 前5个频繁单词
66        Map<String, Integer> wordFreq = new HashMap<>();
67        for (String word : words) {
68            String cleanWord = word.toLowerCase().replaceAll("[^a-z0-9]", "");
69            if (!cleanWord.isEmpty()) {
70                wordFreq.put(cleanWord, wordFreq.getOrDefault(cleanWord, 0) + 1);
71            }
72        }
73        
74        result.topWords = wordFreq.entrySet().stream()
75            .sorted(Map.Entry.<String, Integer>comparingByValue().reversed())
76            .limit(5)
77            .map(e -> new AnalysisResult.WordFrequency(e.getKey(), e.getValue()))
78            .collect(Collectors.toList());
79        
80        // 阅读时间（每分钟225个单词）
81        int minutes = result.wordCount / 225;
82        int seconds = Math.round((result.wordCount % 225) / 225.0f * 60);
83        result.readingTime = minutes > 0 
84            ? minutes + " 分 " + seconds + " 秒"
85            : seconds + " 秒";
86        
87        return result;
88    }
89    
90    public static void main(String[] args) {
91        String sampleText = "你好世界！这是一个文本分析器。它可以统计单词等信息。";
92        AnalysisResult results = analyzeText(sampleText);
93        System.out.println("单词数：" + results.wordCount);
94        System.out.println("阅读时间：" + results.readingTime);
95    }
96}
97

1' Excel VBA 文本分析函数
2Function WordCount(text As String) As Long
3    Dim words() As String
4    If Len(Trim(text)) = 0 Then
5        WordCount = 0
6    Else
7        words = Split(Trim(text), " ")
8        WordCount = UBound(words) + 1
9    End If
10End Function
11
12Function CharCountWithSpaces(text As String) As Long
13    CharCountWithSpaces = Len(text)
14End Function
15
16Function CharCountWithoutSpaces(text As String) As Long
17    Dim textNoSpaces As String
18    textNoSpaces = Replace(text, " ", "")
19    textNoSpaces = Replace(textNoSpaces, vbTab, "")
20    textNoSpaces = Replace(textNoSpaces, vbCrLf, "")
21    textNoSpaces = Replace(textNoSpaces, vbCr, "")
22    textNoSpaces = Replace(textNoSpaces, vbLf, "")
23    CharCountWithoutSpaces = Len(textNoSpaces)
24End Function
25
26Function SentenceCount(text As String) As Long
27    Dim count As Long
28    Dim i As Long
29    count = 0
30    
31    For i = 1 To Len(text)
32        If Mid(text, i, 1) = "." Or Mid(text, i, 1) = "!" Or Mid(text, i, 1) = "?" Then
33            count = count + 1
34        End If
35    Next i
36    
37    If count = 0 And Len(Trim(text)) > 0 Then
38        count = 1
39    End If
40    
41    SentenceCount = count
42End Function
43
44' 在 Excel 中的使用：
45' =WordCount(A1)
46' =CharCountWithSpaces(A1)
47' =CharCountWithoutSpaces(A1)
48' =SentenceCount(A1)
49