Coverage for src/qdrant_loader_mcp_server/mcp/schemas/find_similar.py: 100%

1from typing import Any 

2 

3 

4def get_find_similar_tool_schema() -> dict[str, Any]: 

5 return { 

6 "name": "find_similar_documents", 

7 "description": "Find documents similar to a target document using multiple similarity metrics", 

8 "annotations": {"read-only": True}, 

9 "inputSchema": { 

10 "type": "object", 

11 "properties": { 

12 "target_query": { 

13 "type": "string", 

14 "description": "Query to find the target document", 

15 "minLength": 1, 

16 }, 

17 "comparison_query": { 

18 "type": "string", 

19 "description": "Query to get documents to compare against", 

20 "minLength": 1, 

21 }, 

22 "similarity_metrics": { 

23 "type": "array", 

24 "items": { 

25 "type": "string", 

26 "enum": [ 

27 "entity_overlap", 

28 "topic_overlap", 

29 "semantic_similarity", 

30 "metadata_similarity", 

31 "hierarchical_distance", 

32 "content_features", 

33 ], 

34 }, 

35 "description": "Similarity metrics to use", 

36 }, 

37 "max_similar": { 

38 "type": "integer", 

39 "description": "Maximum number of similar documents to return", 

40 "default": 5, 

41 }, 

42 "source_types": { 

43 "type": "array", 

44 "items": {"type": "string"}, 

45 "description": "Optional list of source types to filter by", 

46 }, 

47 "project_ids": { 

48 "type": "array", 

49 "items": {"type": "string"}, 

50 "description": "Optional list of project IDs to filter by", 

51 }, 

52 }, 

53 "required": ["target_query", "comparison_query"], 

54 "additionalProperties": False, 

55 }, 

56 "outputSchema": { 

57 "type": "object", 

58 "additionalProperties": False, 

59 "properties": { 

60 "similar_documents": { 

61 "type": "array", 

62 "items": { 

63 "type": "object", 

64 "properties": { 

65 "document_id": {"type": "string"}, 

66 "title": {"type": "string"}, 

67 "similarity_score": { 

68 "type": "number", 

69 "minimum": 0, 

70 "maximum": 1, 

71 }, 

72 "similarity_metrics": { 

73 "type": "object", 

74 "properties": { 

75 "entity_overlap": { 

76 "type": "number", 

77 "minimum": 0, 

78 "maximum": 1, 

79 }, 

80 "topic_overlap": { 

81 "type": "number", 

82 "minimum": 0, 

83 "maximum": 1, 

84 }, 

85 "semantic_similarity": { 

86 "type": "number", 

87 "minimum": 0, 

88 "maximum": 1, 

89 }, 

90 "metadata_similarity": { 

91 "type": "number", 

92 "minimum": 0, 

93 "maximum": 1, 

94 }, 

95 "hierarchical_distance": { 

96 "type": "number", 

97 "minimum": 0, 

98 "maximum": 1, 

99 }, 

100 "content_features": { 

101 "type": "number", 

102 "minimum": 0, 

103 "maximum": 1, 

104 }, 

105 }, 

106 "additionalProperties": False, 

107 }, 

108 "similarity_reason": {"type": "string"}, 

109 "content_preview": {"type": "string"}, 

110 }, 

111 "required": [ 

112 "document_id", 

113 "title", 

114 "similarity_score", 

115 ], 

116 "additionalProperties": False, 

117 }, 

118 }, 

119 "target_document": { 

120 "type": "object", 

121 "properties": { 

122 "title": {"type": "string"}, 

123 "content_preview": {"type": "string"}, 

124 "source_type": {"type": "string"}, 

125 }, 

126 }, 

127 "similarity_summary": { 

128 "type": "object", 

129 "properties": { 

130 "total_compared": {"type": "integer"}, 

131 "similar_found": {"type": "integer"}, 

132 "highest_similarity": { 

133 "type": "number", 

134 "minimum": 0, 

135 "maximum": 1, 

136 }, 

137 "metrics_used": { 

138 "type": "array", 

139 "items": {"type": "string"}, 

140 "minItems": 1, 

141 "uniqueItems": True, 

142 }, 

143 }, 

144 "required": [ 

145 "total_compared", 

146 "similar_found", 

147 "highest_similarity", 

148 "metrics_used", 

149 ], 

150 "additionalProperties": False, 

151 }, 

152 }, 

153 }, 

154 }