Coverage for agentos/marketplace/skills/summarize/summarize.py: 12%

1""" 

2summarize — Text summarization using extractive and heuristic methods. 

3 

4Actions: summarize, extract_keywords, bullet_points, word_count 

5No external API required — pure Python. 

6""" 

7 

8import re 

9from collections import Counter 

10from typing import Any 

11 

12 

13def run( 

14 action: str = "summarize", 

15 text: str = "", 

16 file_path: str = "", 

17 ratio: float = 0.3, 

18 **kwargs: Any, 

19) -> str: 

20 content = text 

21 if file_path: 

22 try: 

23 with open(file_path, encoding="utf-8") as f: 

24 content = f.read() 

25 except FileNotFoundError: 

26 return f"[summarize] File not found: {file_path}" 

27 except Exception as e: 

28 return f"[summarize] Error reading file: {e}" 

29 

30 if not content: 

31 return "[summarize] No text provided." 

32 

33 if action == "word_count": 

34 words = re.findall(r"\b\w+\b", content.lower()) 

35 wc = len(words) 

36 sc = len(re.split(r"[.!?]+", content)) 

37 return f"Words: {wc}, Sentences: ~{sc}, Characters: {len(content)}" 

38 

39 if action == "extract_keywords": 

40 words = re.findall(r"\b[a-zA-Z]{3,}\b", content.lower()) 

41 stop = { 

42 "the", 

43 "and", 

44 "for", 

45 "that", 

46 "with", 

47 "this", 

48 "from", 

49 "have", 

50 "are", 

51 "not", 

52 "but", 

53 "was", 

54 "you", 

55 "all", 

56 "can", 

57 "has", 

58 "had", 

59 "been", 

60 "will", 

61 "they", 

62 "its", 

63 "their", 

64 "them", 

65 "our", 

66 "than", 

67 "then", 

68 "also", 

69 "into", 

70 "just", 

71 "about", 

72 "more", 

73 "some", 

74 "when", 

75 "your", 

76 "which", 

77 "make", 

78 "like", 

79 "what", 

80 "over", 

81 "such", 

82 "here", 

83 "were", 

84 "how", 

85 "one", 

86 "two", 

87 } 

88 filtered = [w for w in words if w not in stop] 

89 top = Counter(filtered).most_common(15) 

90 return "Top keywords: " + ", ".join(f"{w}({c})" for w, c in top) 

91 

92 if action == "bullet_points": 

93 sentences = re.split(r"(?<=[.!?])\s+", content) 

94 sentences = [s.strip() for s in sentences if len(s.strip()) > 20] 

95 n = max(3, int(len(sentences) * ratio)) 

96 return "[summarize] Bullet Points:\n" + "\n".join(f"- {s}" for s in sentences[:n]) 

97 

98 # Default: summarize (extractive) 

99 sentences = re.split(r"(?<=[.!?])\s+", content) 

100 sentences = [s.strip() for s in sentences if len(s.strip()) > 10] 

101 if not sentences: 

102 return content[:500] 

103 

104 n = max(2, int(len(sentences) * ratio)) 

105 summary = " ".join(sentences[:n]) 

106 if len(summary) > 1500: 

107 summary = summary[:1500] + "..." 

108 return summary 

109 

110 

111__all__ = ["run"]