illuma-agents 1.0.7 → 1.0.9
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +1 -5
- package/dist/cjs/common/enum.cjs +1 -2
- package/dist/cjs/common/enum.cjs.map +1 -1
- package/dist/cjs/instrumentation.cjs.map +1 -1
- package/dist/cjs/llm/anthropic/types.cjs.map +1 -1
- package/dist/cjs/llm/anthropic/utils/message_inputs.cjs +79 -2
- package/dist/cjs/llm/anthropic/utils/message_inputs.cjs.map +1 -1
- package/dist/cjs/llm/anthropic/utils/tools.cjs.map +1 -1
- package/dist/cjs/llm/bedrock/index.cjs +99 -0
- package/dist/cjs/llm/bedrock/index.cjs.map +1 -0
- package/dist/cjs/llm/fake.cjs.map +1 -1
- package/dist/cjs/llm/google/index.cjs +78 -9
- package/dist/cjs/llm/google/index.cjs.map +1 -1
- package/dist/cjs/llm/google/utils/common.cjs +185 -28
- package/dist/cjs/llm/google/utils/common.cjs.map +1 -1
- package/dist/cjs/llm/providers.cjs +13 -16
- package/dist/cjs/llm/providers.cjs.map +1 -1
- package/dist/cjs/llm/text.cjs.map +1 -1
- package/dist/cjs/messages/core.cjs +14 -14
- package/dist/cjs/messages/core.cjs.map +1 -1
- package/dist/cjs/messages/ids.cjs.map +1 -1
- package/dist/cjs/messages/prune.cjs.map +1 -1
- package/dist/cjs/run.cjs +10 -1
- package/dist/cjs/run.cjs.map +1 -1
- package/dist/cjs/splitStream.cjs.map +1 -1
- package/dist/cjs/stream.cjs +4 -1
- package/dist/cjs/stream.cjs.map +1 -1
- package/dist/cjs/tools/ToolNode.cjs +163 -55
- package/dist/cjs/tools/ToolNode.cjs.map +1 -1
- package/dist/cjs/tools/handlers.cjs +29 -25
- package/dist/cjs/tools/handlers.cjs.map +1 -1
- package/dist/cjs/tools/search/anthropic.cjs.map +1 -1
- package/dist/cjs/tools/search/content.cjs.map +1 -1
- package/dist/cjs/tools/search/firecrawl.cjs.map +1 -1
- package/dist/cjs/tools/search/format.cjs.map +1 -1
- package/dist/cjs/tools/search/highlights.cjs.map +1 -1
- package/dist/cjs/tools/search/rerankers.cjs.map +1 -1
- package/dist/cjs/tools/search/schema.cjs +25 -25
- package/dist/cjs/tools/search/schema.cjs.map +1 -1
- package/dist/cjs/tools/search/search.cjs +6 -1
- package/dist/cjs/tools/search/search.cjs.map +1 -1
- package/dist/cjs/tools/search/serper-scraper.cjs.map +1 -1
- package/dist/cjs/tools/search/tool.cjs +162 -35
- package/dist/cjs/tools/search/tool.cjs.map +1 -1
- package/dist/cjs/tools/search/utils.cjs.map +1 -1
- package/dist/cjs/utils/graph.cjs.map +1 -1
- package/dist/cjs/utils/llm.cjs +0 -1
- package/dist/cjs/utils/llm.cjs.map +1 -1
- package/dist/cjs/utils/misc.cjs.map +1 -1
- package/dist/cjs/utils/run.cjs.map +1 -1
- package/dist/cjs/utils/title.cjs +7 -7
- package/dist/cjs/utils/title.cjs.map +1 -1
- package/dist/esm/common/enum.mjs +1 -2
- package/dist/esm/common/enum.mjs.map +1 -1
- package/dist/esm/instrumentation.mjs.map +1 -1
- package/dist/esm/llm/anthropic/types.mjs.map +1 -1
- package/dist/esm/llm/anthropic/utils/message_inputs.mjs +79 -2
- package/dist/esm/llm/anthropic/utils/message_inputs.mjs.map +1 -1
- package/dist/esm/llm/anthropic/utils/tools.mjs.map +1 -1
- package/dist/esm/llm/bedrock/index.mjs +97 -0
- package/dist/esm/llm/bedrock/index.mjs.map +1 -0
- package/dist/esm/llm/fake.mjs.map +1 -1
- package/dist/esm/llm/google/index.mjs +79 -10
- package/dist/esm/llm/google/index.mjs.map +1 -1
- package/dist/esm/llm/google/utils/common.mjs +184 -30
- package/dist/esm/llm/google/utils/common.mjs.map +1 -1
- package/dist/esm/llm/providers.mjs +2 -5
- package/dist/esm/llm/providers.mjs.map +1 -1
- package/dist/esm/llm/text.mjs.map +1 -1
- package/dist/esm/messages/core.mjs +14 -14
- package/dist/esm/messages/core.mjs.map +1 -1
- package/dist/esm/messages/ids.mjs.map +1 -1
- package/dist/esm/messages/prune.mjs.map +1 -1
- package/dist/esm/run.mjs +10 -1
- package/dist/esm/run.mjs.map +1 -1
- package/dist/esm/splitStream.mjs.map +1 -1
- package/dist/esm/stream.mjs +4 -1
- package/dist/esm/stream.mjs.map +1 -1
- package/dist/esm/tools/ToolNode.mjs +164 -56
- package/dist/esm/tools/ToolNode.mjs.map +1 -1
- package/dist/esm/tools/handlers.mjs +30 -26
- package/dist/esm/tools/handlers.mjs.map +1 -1
- package/dist/esm/tools/search/anthropic.mjs.map +1 -1
- package/dist/esm/tools/search/content.mjs.map +1 -1
- package/dist/esm/tools/search/firecrawl.mjs.map +1 -1
- package/dist/esm/tools/search/format.mjs.map +1 -1
- package/dist/esm/tools/search/highlights.mjs.map +1 -1
- package/dist/esm/tools/search/rerankers.mjs.map +1 -1
- package/dist/esm/tools/search/schema.mjs +25 -25
- package/dist/esm/tools/search/schema.mjs.map +1 -1
- package/dist/esm/tools/search/search.mjs +6 -1
- package/dist/esm/tools/search/search.mjs.map +1 -1
- package/dist/esm/tools/search/serper-scraper.mjs.map +1 -1
- package/dist/esm/tools/search/tool.mjs +162 -35
- package/dist/esm/tools/search/tool.mjs.map +1 -1
- package/dist/esm/tools/search/utils.mjs.map +1 -1
- package/dist/esm/utils/graph.mjs.map +1 -1
- package/dist/esm/utils/llm.mjs +0 -1
- package/dist/esm/utils/llm.mjs.map +1 -1
- package/dist/esm/utils/misc.mjs.map +1 -1
- package/dist/esm/utils/run.mjs.map +1 -1
- package/dist/esm/utils/title.mjs +7 -7
- package/dist/esm/utils/title.mjs.map +1 -1
- package/dist/types/common/enum.d.ts +1 -2
- package/dist/types/llm/bedrock/index.d.ts +36 -0
- package/dist/types/llm/google/index.d.ts +10 -0
- package/dist/types/llm/google/types.d.ts +11 -1
- package/dist/types/llm/google/utils/common.d.ts +17 -2
- package/dist/types/tools/ToolNode.d.ts +9 -1
- package/dist/types/tools/search/types.d.ts +2 -0
- package/dist/types/types/llm.d.ts +3 -8
- package/dist/types/types/tools.d.ts +1 -1
- package/package.json +15 -11
- package/src/common/enum.ts +1 -2
- package/src/common/index.ts +1 -1
- package/src/instrumentation.ts +22 -22
- package/src/llm/anthropic/llm.spec.ts +1442 -1442
- package/src/llm/anthropic/types.ts +140 -140
- package/src/llm/anthropic/utils/message_inputs.ts +757 -660
- package/src/llm/anthropic/utils/output_parsers.ts +133 -133
- package/src/llm/anthropic/utils/tools.ts +29 -29
- package/src/llm/bedrock/index.ts +128 -0
- package/src/llm/fake.ts +133 -133
- package/src/llm/google/data/gettysburg10.wav +0 -0
- package/src/llm/google/data/hotdog.jpg +0 -0
- package/src/llm/google/index.ts +129 -14
- package/src/llm/google/llm.spec.ts +932 -0
- package/src/llm/google/types.ts +56 -43
- package/src/llm/google/utils/common.ts +873 -660
- package/src/llm/google/utils/tools.ts +160 -160
- package/src/llm/openai/types.ts +24 -24
- package/src/llm/openai/utils/isReasoningModel.test.ts +90 -90
- package/src/llm/providers.ts +2 -7
- package/src/llm/text.ts +94 -94
- package/src/messages/core.ts +463 -463
- package/src/messages/formatAgentMessages.tools.test.ts +400 -400
- package/src/messages/formatMessage.test.ts +693 -693
- package/src/messages/ids.ts +26 -26
- package/src/messages/prune.ts +567 -567
- package/src/messages/shiftIndexTokenCountMap.test.ts +81 -81
- package/src/mockStream.ts +98 -98
- package/src/prompts/collab.ts +5 -5
- package/src/prompts/index.ts +1 -1
- package/src/prompts/taskmanager.ts +61 -61
- package/src/run.ts +13 -4
- package/src/scripts/ant_web_search_edge_case.ts +162 -0
- package/src/scripts/ant_web_search_error_edge_case.ts +148 -0
- package/src/scripts/args.ts +48 -48
- package/src/scripts/caching.ts +123 -123
- package/src/scripts/code_exec_files.ts +193 -193
- package/src/scripts/empty_input.ts +137 -137
- package/src/scripts/image.ts +178 -178
- package/src/scripts/memory.ts +97 -97
- package/src/scripts/thinking.ts +149 -149
- package/src/specs/anthropic.simple.test.ts +67 -0
- package/src/specs/spec.utils.ts +3 -3
- package/src/specs/token-distribution-edge-case.test.ts +316 -316
- package/src/specs/tool-error.test.ts +193 -193
- package/src/splitStream.test.ts +691 -691
- package/src/splitStream.ts +234 -234
- package/src/stream.test.ts +94 -94
- package/src/stream.ts +4 -1
- package/src/tools/ToolNode.ts +206 -64
- package/src/tools/handlers.ts +32 -28
- package/src/tools/search/anthropic.ts +51 -51
- package/src/tools/search/content.test.ts +173 -173
- package/src/tools/search/content.ts +147 -147
- package/src/tools/search/direct-url.test.ts +530 -0
- package/src/tools/search/firecrawl.ts +210 -210
- package/src/tools/search/format.ts +250 -250
- package/src/tools/search/highlights.ts +320 -320
- package/src/tools/search/index.ts +2 -2
- package/src/tools/search/jina-reranker.test.ts +126 -126
- package/src/tools/search/output.md +2775 -2775
- package/src/tools/search/rerankers.ts +242 -242
- package/src/tools/search/schema.ts +63 -63
- package/src/tools/search/search.ts +766 -759
- package/src/tools/search/serper-scraper.ts +155 -155
- package/src/tools/search/test.html +883 -883
- package/src/tools/search/test.md +642 -642
- package/src/tools/search/test.ts +159 -159
- package/src/tools/search/tool.ts +619 -471
- package/src/tools/search/types.ts +689 -687
- package/src/tools/search/utils.ts +79 -79
- package/src/types/index.ts +6 -6
- package/src/types/llm.ts +2 -8
- package/src/types/tools.ts +80 -80
- package/src/utils/graph.ts +10 -10
- package/src/utils/llm.ts +26 -27
- package/src/utils/llmConfig.ts +5 -3
- package/src/utils/logging.ts +48 -48
- package/src/utils/misc.ts +57 -57
- package/src/utils/run.ts +100 -100
- package/src/utils/title.ts +165 -165
- package/dist/cjs/llm/ollama/index.cjs +0 -70
- package/dist/cjs/llm/ollama/index.cjs.map +0 -1
- package/dist/cjs/llm/ollama/utils.cjs +0 -158
- package/dist/cjs/llm/ollama/utils.cjs.map +0 -1
- package/dist/esm/llm/ollama/index.mjs +0 -68
- package/dist/esm/llm/ollama/index.mjs.map +0 -1
- package/dist/esm/llm/ollama/utils.mjs +0 -155
- package/dist/esm/llm/ollama/utils.mjs.map +0 -1
- package/dist/types/llm/ollama/index.d.ts +0 -8
- package/dist/types/llm/ollama/utils.d.ts +0 -7
- package/src/llm/ollama/index.ts +0 -92
- package/src/llm/ollama/utils.ts +0 -193
- package/src/proto/CollabGraph.ts +0 -269
- package/src/proto/TaskManager.ts +0 -243
- package/src/proto/collab.ts +0 -200
- package/src/proto/collab_design.ts +0 -184
- package/src/proto/collab_design_v2.ts +0 -224
- package/src/proto/collab_design_v3.ts +0 -255
- package/src/proto/collab_design_v4.ts +0 -220
- package/src/proto/collab_design_v5.ts +0 -251
- package/src/proto/collab_graph.ts +0 -181
- package/src/proto/collab_original.ts +0 -123
- package/src/proto/example.ts +0 -93
- package/src/proto/example_new.ts +0 -68
- package/src/proto/example_old.ts +0 -201
- package/src/proto/example_test.ts +0 -152
- package/src/proto/example_test_anthropic.ts +0 -100
- package/src/proto/log_stream.ts +0 -202
- package/src/proto/main_collab_community_event.ts +0 -133
- package/src/proto/main_collab_design_v2.ts +0 -96
- package/src/proto/main_collab_design_v4.ts +0 -100
- package/src/proto/main_collab_design_v5.ts +0 -135
- package/src/proto/main_collab_global_analysis.ts +0 -122
- package/src/proto/main_collab_hackathon_event.ts +0 -153
- package/src/proto/main_collab_space_mission.ts +0 -153
- package/src/proto/main_philosophy.ts +0 -210
- package/src/proto/original_script.ts +0 -126
- package/src/proto/standard.ts +0 -100
- package/src/proto/stream.ts +0 -56
- package/src/proto/tasks.ts +0 -118
- package/src/proto/tools/global_analysis_tools.ts +0 -86
- package/src/proto/tools/space_mission_tools.ts +0 -60
- package/src/proto/vertexai.ts +0 -54
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"handlers.mjs","sources":["../../../src/tools/handlers.ts"],"sourcesContent":["/* eslint-disable no-console */\n// src/tools/handlers.ts\nimport { nanoid } from 'nanoid';\nimport { ToolMessage } from '@langchain/core/messages';\nimport type { AnthropicWebSearchResultBlockParam } from '@/llm/anthropic/types';\nimport type { ToolCall, ToolCallChunk } from '@langchain/core/messages/tool';\nimport type { MultiAgentGraph, StandardGraph } from '@/graphs';\nimport type { AgentContext } from '@/agents/AgentContext';\nimport type * as t from '@/types';\nimport {\n ToolCallTypes,\n ContentTypes,\n GraphEvents,\n StepTypes,\n Providers,\n Constants,\n} from '@/common';\nimport {\n coerceAnthropicSearchResults,\n isAnthropicWebSearchResult,\n} from '@/tools/search/anthropic';\nimport { formatResultsForLLM } from '@/tools/search/format';\nimport { getMessageId } from '@/messages';\n\nexport async function handleToolCallChunks({\n graph,\n stepKey,\n toolCallChunks,\n metadata,\n}: {\n graph: StandardGraph | MultiAgentGraph;\n stepKey: string;\n toolCallChunks: ToolCallChunk[];\n metadata?: Record<string, unknown>;\n}): Promise<void> {\n let prevStepId: string;\n let prevRunStep: t.RunStep | undefined;\n try {\n prevStepId = graph.getStepIdByKey(stepKey);\n prevRunStep = graph.getRunStep(prevStepId);\n } catch {\n /** Edge Case: If no previous step exists, create a new message creation step */\n const message_id = getMessageId(stepKey, graph, true) ?? '';\n prevStepId = await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.MESSAGE_CREATION,\n message_creation: {\n message_id,\n },\n },\n metadata\n );\n prevRunStep = graph.getRunStep(prevStepId);\n }\n\n const _stepId = graph.getStepIdByKey(stepKey);\n\n /** Edge Case: Tool Call Run Step or `tool_call_ids` never dispatched */\n const tool_calls: ToolCall[] | undefined =\n prevStepId && prevRunStep && prevRunStep.type === StepTypes.MESSAGE_CREATION\n ? []\n : undefined;\n\n /** Edge Case: `id` and `name` fields cannot be empty strings */\n for (const toolCallChunk of toolCallChunks) {\n if (toolCallChunk.name === '') {\n toolCallChunk.name = undefined;\n }\n if (toolCallChunk.id === '') {\n toolCallChunk.id = undefined;\n } else if (\n tool_calls != null &&\n toolCallChunk.id != null &&\n toolCallChunk.name != null\n ) {\n tool_calls.push({\n args: {},\n id: toolCallChunk.id,\n name: toolCallChunk.name,\n type: ToolCallTypes.TOOL_CALL,\n });\n }\n }\n\n let stepId: string = _stepId;\n const alreadyDispatched =\n prevRunStep?.type === StepTypes.MESSAGE_CREATION &&\n graph.messageStepHasToolCalls.has(prevStepId);\n if (!alreadyDispatched && tool_calls?.length === toolCallChunks.length) {\n await graph.dispatchMessageDelta(prevStepId, {\n content: [\n {\n type: ContentTypes.TEXT,\n text: '',\n tool_call_ids: tool_calls.map((tc) => tc.id ?? ''),\n },\n ],\n });\n graph.messageStepHasToolCalls.set(prevStepId, true);\n stepId = await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.TOOL_CALLS,\n tool_calls,\n },\n metadata\n );\n }\n await graph.dispatchRunStepDelta(stepId, {\n type: StepTypes.TOOL_CALLS,\n tool_calls: toolCallChunks,\n });\n}\n\nexport const handleToolCalls = async (\n toolCalls?: ToolCall[],\n metadata?: Record<string, unknown>,\n graph?: StandardGraph | MultiAgentGraph\n): Promise<void> => {\n if (!graph || !metadata) {\n console.warn(`Graph or metadata not found in ${event} event`);\n return;\n }\n\n if (!toolCalls) {\n return;\n }\n\n if (toolCalls.length === 0) {\n return;\n }\n\n const stepKey = graph.getStepKey(metadata);\n\n for (const tool_call of toolCalls) {\n const toolCallId = tool_call.id ?? `toolu_${nanoid()}`;\n tool_call.id = toolCallId;\n if (!toolCallId || graph.toolCallStepIds.has(toolCallId)) {\n continue;\n }\n\n let prevStepId = '';\n let prevRunStep: t.RunStep | undefined;\n try {\n prevStepId = graph.getStepIdByKey(stepKey);\n prevRunStep = graph.getRunStep(prevStepId);\n } catch {\n // no previous step\n }\n\n const dispatchToolCallIds = async (\n lastMessageStepId: string\n ): Promise<void> => {\n await graph.dispatchMessageDelta(lastMessageStepId, {\n content: [\n {\n type: 'text',\n text: '',\n tool_call_ids: [toolCallId],\n },\n ],\n });\n };\n /* If the previous step exists and is a message creation */\n if (\n prevStepId &&\n prevRunStep &&\n prevRunStep.type === StepTypes.MESSAGE_CREATION\n ) {\n await dispatchToolCallIds(prevStepId);\n graph.messageStepHasToolCalls.set(prevStepId, true);\n /* If the previous step doesn't exist or is not a message creation */\n } else if (\n !prevRunStep ||\n prevRunStep.type !== StepTypes.MESSAGE_CREATION\n ) {\n const messageId = getMessageId(stepKey, graph, true) ?? '';\n const stepId = await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.MESSAGE_CREATION,\n message_creation: {\n message_id: messageId,\n },\n },\n metadata\n );\n await dispatchToolCallIds(stepId);\n graph.messageStepHasToolCalls.set(prevStepId, true);\n }\n\n await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.TOOL_CALLS,\n tool_calls: [tool_call],\n },\n metadata\n );\n }\n};\n\nexport const toolResultTypes = new Set([\n // 'tool_use',\n // 'server_tool_use',\n // 'input_json_delta',\n 'tool_result',\n 'web_search_result',\n 'web_search_tool_result',\n]);\n\n/**\n * Handles the result of a server tool call; in other words, a provider's built-in tool.\n * As of 2025-07-06, only Anthropic handles server tool calls with this pattern.\n */\nexport async function handleServerToolResult({\n graph,\n content,\n metadata,\n agentContext,\n}: {\n graph: StandardGraph | MultiAgentGraph;\n content?: string | t.MessageContentComplex[];\n metadata?: Record<string, unknown>;\n agentContext?: AgentContext;\n}): Promise<boolean> {\n let skipHandling = false;\n if (agentContext?.provider !== Providers.ANTHROPIC) {\n return skipHandling;\n }\n if (\n typeof content === 'string' ||\n content == null ||\n content.length === 0 ||\n (content.length === 1 &&\n (content[0] as t.ToolResultContent).tool_use_id == null)\n ) {\n return skipHandling;\n }\n\n for (const contentPart of content) {\n const toolUseId = (contentPart as t.ToolResultContent).tool_use_id;\n if (toolUseId == null || toolUseId === '') {\n continue;\n }\n const stepId = graph.toolCallStepIds.get(toolUseId);\n if (stepId == null || stepId === '') {\n console.warn(\n `Tool use ID ${toolUseId} not found in graph, cannot dispatch tool result.`\n );\n continue;\n }\n const runStep = graph.getRunStep(stepId);\n if (!runStep) {\n console.warn(\n `Run step for ${stepId} does not exist, cannot dispatch tool result.`\n );\n continue;\n } else if (runStep.type !== StepTypes.TOOL_CALLS) {\n console.warn(\n `Run step for ${stepId} is not a tool call step, cannot dispatch tool result.`\n );\n continue;\n }\n\n const toolCall =\n runStep.stepDetails.type === StepTypes.TOOL_CALLS\n ? (runStep.stepDetails.tool_calls?.find(\n (toolCall) => toolCall.id === toolUseId\n ) as ToolCall)\n : undefined;\n\n if (!toolCall) {\n continue;\n }\n\n if (\n contentPart.type === 'web_search_result' ||\n contentPart.type === 'web_search_tool_result'\n ) {\n await handleAnthropicSearchResults({\n contentPart: contentPart as t.ToolResultContent,\n toolCall,\n metadata,\n graph,\n });\n }\n\n if (!skipHandling) {\n skipHandling = true;\n }\n }\n\n return skipHandling;\n}\n\nasync function handleAnthropicSearchResults({\n contentPart,\n toolCall,\n metadata,\n graph,\n}: {\n contentPart: t.ToolResultContent;\n toolCall: Partial<ToolCall>;\n metadata?: Record<string, unknown>;\n graph: StandardGraph | MultiAgentGraph;\n}): Promise<void> {\n if (!Array.isArray(contentPart.content)) {\n console.warn(\n `Expected content to be an array, got ${typeof contentPart.content}`\n );\n return;\n }\n\n if (!isAnthropicWebSearchResult(contentPart.content[0])) {\n console.warn(\n `Expected content to be an Anthropic web search result, got ${JSON.stringify(\n contentPart.content\n )}`\n );\n return;\n }\n\n const turn = graph.invokedToolIds?.size ?? 0;\n const searchResultData = coerceAnthropicSearchResults({\n turn,\n results: contentPart.content as AnthropicWebSearchResultBlockParam[],\n });\n\n const name = toolCall.name;\n const input = toolCall.args ?? {};\n const artifact = {\n [Constants.WEB_SEARCH]: searchResultData,\n };\n const { output: formattedOutput } = formatResultsForLLM(\n turn,\n searchResultData\n );\n const output = new ToolMessage({\n name,\n artifact,\n content: formattedOutput,\n tool_call_id: toolCall.id!,\n });\n const toolEndData: t.ToolEndData = {\n input,\n output,\n };\n await graph.handlerRegistry\n ?.getHandler(GraphEvents.TOOL_END)\n ?.handle(GraphEvents.TOOL_END, toolEndData, metadata, graph);\n\n if (graph.invokedToolIds == null) {\n graph.invokedToolIds = new Set<string>();\n }\n\n graph.invokedToolIds.add(toolCall.id!);\n}\n"],"names":[],"mappings":";;;;;;;;AAAA;AACA;AAuBO,eAAe,oBAAoB,CAAC,EACzC,KAAK,EACL,OAAO,EACP,cAAc,EACd,QAAQ,GAMT,EAAA;AACC,IAAA,IAAI,UAAkB;AACtB,IAAA,IAAI,WAAkC;AACtC,IAAA,IAAI;AACF,QAAA,UAAU,GAAG,KAAK,CAAC,cAAc,CAAC,OAAO,CAAC;AAC1C,QAAA,WAAW,GAAG,KAAK,CAAC,UAAU,CAAC,UAAU,CAAC;;AAC1C,IAAA,MAAM;;AAEN,QAAA,MAAM,UAAU,GAAG,YAAY,CAAC,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,IAAI,EAAE;AAC3D,QAAA,UAAU,GAAG,MAAM,KAAK,CAAC,eAAe,CACtC,OAAO,EACP;YACE,IAAI,EAAE,SAAS,CAAC,gBAAgB;AAChC,YAAA,gBAAgB,EAAE;gBAChB,UAAU;AACX,aAAA;SACF,EACD,QAAQ,CACT;AACD,QAAA,WAAW,GAAG,KAAK,CAAC,UAAU,CAAC,UAAU,CAAC;;IAG5C,MAAM,OAAO,GAAG,KAAK,CAAC,cAAc,CAAC,OAAO,CAAC;;AAG7C,IAAA,MAAM,UAAU,GACd,UAAU,IAAI,WAAW,IAAI,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC;AAC1D,UAAE;UACA,SAAS;;AAGf,IAAA,KAAK,MAAM,aAAa,IAAI,cAAc,EAAE;AAC1C,QAAA,IAAI,aAAa,CAAC,IAAI,KAAK,EAAE,EAAE;AAC7B,YAAA,aAAa,CAAC,IAAI,GAAG,SAAS;;AAEhC,QAAA,IAAI,aAAa,CAAC,EAAE,KAAK,EAAE,EAAE;AAC3B,YAAA,aAAa,CAAC,EAAE,GAAG,SAAS;;aACvB,IACL,UAAU,IAAI,IAAI;YAClB,aAAa,CAAC,EAAE,IAAI,IAAI;AACxB,YAAA,aAAa,CAAC,IAAI,IAAI,IAAI,EAC1B;YACA,UAAU,CAAC,IAAI,CAAC;AACd,gBAAA,IAAI,EAAE,EAAE;gBACR,EAAE,EAAE,aAAa,CAAC,EAAE;gBACpB,IAAI,EAAE,aAAa,CAAC,IAAI;gBACxB,IAAI,EAAE,aAAa,CAAC,SAAS;AAC9B,aAAA,CAAC;;;IAIN,IAAI,MAAM,GAAW,OAAO;IAC5B,MAAM,iBAAiB,GACrB,WAAW,EAAE,IAAI,KAAK,SAAS,CAAC,gBAAgB;AAChD,QAAA,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,UAAU,CAAC;IAC/C,IAAI,CAAC,iBAAiB,IAAI,UAAU,EAAE,MAAM,KAAK,cAAc,CAAC,MAAM,EAAE;AACtE,QAAA,MAAM,KAAK,CAAC,oBAAoB,CAAC,UAAU,EAAE;AAC3C,YAAA,OAAO,EAAE;AACP,gBAAA;oBACE,IAAI,EAAE,YAAY,CAAC,IAAI;AACvB,oBAAA,IAAI,EAAE,EAAE;AACR,oBAAA,aAAa,EAAE,UAAU,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC;AACnD,iBAAA;AACF,aAAA;AACF,SAAA,CAAC;QACF,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC;AACnD,QAAA,MAAM,GAAG,MAAM,KAAK,CAAC,eAAe,CAClC,OAAO,EACP;YACE,IAAI,EAAE,SAAS,CAAC,UAAU;YAC1B,UAAU;SACX,EACD,QAAQ,CACT;;AAEH,IAAA,MAAM,KAAK,CAAC,oBAAoB,CAAC,MAAM,EAAE;QACvC,IAAI,EAAE,SAAS,CAAC,UAAU;AAC1B,QAAA,UAAU,EAAE,cAAc;AAC3B,KAAA,CAAC;AACJ;AAEO,MAAM,eAAe,GAAG,OAC7B,SAAsB,EACtB,QAAkC,EAClC,KAAuC,KACtB;AACjB,IAAA,IAAI,CAAC,KAAK,IAAI,CAAC,QAAQ,EAAE;AACvB,QAAA,OAAO,CAAC,IAAI,CAAC,kCAAkC,KAAK,CAAA,MAAA,CAAQ,CAAC;QAC7D;;IAGF,IAAI,CAAC,SAAS,EAAE;QACd;;AAGF,IAAA,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE;QAC1B;;IAGF,MAAM,OAAO,GAAG,KAAK,CAAC,UAAU,CAAC,QAAQ,CAAC;AAE1C,IAAA,KAAK,MAAM,SAAS,IAAI,SAAS,EAAE;QACjC,MAAM,UAAU,GAAG,SAAS,CAAC,EAAE,IAAI,CAAS,MAAA,EAAA,MAAM,EAAE,CAAA,CAAE;AACtD,QAAA,SAAS,CAAC,EAAE,GAAG,UAAU;AACzB,QAAA,IAAI,CAAC,UAAU,IAAI,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE;YACxD;;QAGF,IAAI,UAAU,GAAG,EAAE;AACnB,QAAA,IAAI,WAAkC;AACtC,QAAA,IAAI;AACF,YAAA,UAAU,GAAG,KAAK,CAAC,cAAc,CAAC,OAAO,CAAC;AAC1C,YAAA,WAAW,GAAG,KAAK,CAAC,UAAU,CAAC,UAAU,CAAC;;AAC1C,QAAA,MAAM;;;AAIR,QAAA,MAAM,mBAAmB,GAAG,OAC1B,iBAAyB,KACR;AACjB,YAAA,MAAM,KAAK,CAAC,oBAAoB,CAAC,iBAAiB,EAAE;AAClD,gBAAA,OAAO,EAAE;AACP,oBAAA;AACE,wBAAA,IAAI,EAAE,MAAM;AACZ,wBAAA,IAAI,EAAE,EAAE;wBACR,aAAa,EAAE,CAAC,UAAU,CAAC;AAC5B,qBAAA;AACF,iBAAA;AACF,aAAA,CAAC;AACJ,SAAC;;AAED,QAAA,IACE,UAAU;YACV,WAAW;AACX,YAAA,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC,gBAAgB,EAC/C;AACA,YAAA,MAAM,mBAAmB,CAAC,UAAU,CAAC;YACrC,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC;;;AAE9C,aAAA,IACL,CAAC,WAAW;AACZ,YAAA,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC,gBAAgB,EAC/C;AACA,YAAA,MAAM,SAAS,GAAG,YAAY,CAAC,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,IAAI,EAAE;YAC1D,MAAM,MAAM,GAAG,MAAM,KAAK,CAAC,eAAe,CACxC,OAAO,EACP;gBACE,IAAI,EAAE,SAAS,CAAC,gBAAgB;AAChC,gBAAA,gBAAgB,EAAE;AAChB,oBAAA,UAAU,EAAE,SAAS;AACtB,iBAAA;aACF,EACD,QAAQ,CACT;AACD,YAAA,MAAM,mBAAmB,CAAC,MAAM,CAAC;YACjC,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC;;AAGrD,QAAA,MAAM,KAAK,CAAC,eAAe,CACzB,OAAO,EACP;YACE,IAAI,EAAE,SAAS,CAAC,UAAU;YAC1B,UAAU,EAAE,CAAC,SAAS,CAAC;SACxB,EACD,QAAQ,CACT;;AAEL;AAEa,MAAA,eAAe,GAAG,IAAI,GAAG,CAAC;;;;IAIrC,aAAa;IACb,mBAAmB;IACnB,wBAAwB;AACzB,CAAA;AAED;;;AAGG;AACI,eAAe,sBAAsB,CAAC,EAC3C,KAAK,EACL,OAAO,EACP,QAAQ,EACR,YAAY,GAMb,EAAA;IACC,IAAI,YAAY,GAAG,KAAK;IACxB,IAAI,YAAY,EAAE,QAAQ,KAAK,SAAS,CAAC,SAAS,EAAE;AAClD,QAAA,OAAO,YAAY;;IAErB,IACE,OAAO,OAAO,KAAK,QAAQ;AAC3B,QAAA,OAAO,IAAI,IAAI;QACf,OAAO,CAAC,MAAM,KAAK,CAAC;AACpB,SAAC,OAAO,CAAC,MAAM,KAAK,CAAC;YAClB,OAAO,CAAC,CAAC,CAAyB,CAAC,WAAW,IAAI,IAAI,CAAC,EAC1D;AACA,QAAA,OAAO,YAAY;;AAGrB,IAAA,KAAK,MAAM,WAAW,IAAI,OAAO,EAAE;AACjC,QAAA,MAAM,SAAS,GAAI,WAAmC,CAAC,WAAW;QAClE,IAAI,SAAS,IAAI,IAAI,IAAI,SAAS,KAAK,EAAE,EAAE;YACzC;;QAEF,MAAM,MAAM,GAAG,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,SAAS,CAAC;QACnD,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,KAAK,EAAE,EAAE;AACnC,YAAA,OAAO,CAAC,IAAI,CACV,eAAe,SAAS,CAAA,iDAAA,CAAmD,CAC5E;YACD;;QAEF,MAAM,OAAO,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC;QACxC,IAAI,CAAC,OAAO,EAAE;AACZ,YAAA,OAAO,CAAC,IAAI,CACV,gBAAgB,MAAM,CAAA,6CAAA,CAA+C,CACtE;YACD;;aACK,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,CAAC,UAAU,EAAE;AAChD,YAAA,OAAO,CAAC,IAAI,CACV,gBAAgB,MAAM,CAAA,sDAAA,CAAwD,CAC/E;YACD;;QAGF,MAAM,QAAQ,GACZ,OAAO,CAAC,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC;AACrC,cAAG,OAAO,CAAC,WAAW,CAAC,UAAU,EAAE,IAAI,CACrC,CAAC,QAAQ,KAAK,QAAQ,CAAC,EAAE,KAAK,SAAS;cAEvC,SAAS;QAEf,IAAI,CAAC,QAAQ,EAAE;YACb;;AAGF,QAAA,IACE,WAAW,CAAC,IAAI,KAAK,mBAAmB;AACxC,YAAA,WAAW,CAAC,IAAI,KAAK,wBAAwB,EAC7C;AACA,YAAA,MAAM,4BAA4B,CAAC;AACjC,gBAAA,WAAW,EAAE,WAAkC;gBAC/C,QAAQ;gBACR,QAAQ;gBACR,KAAK;AACN,aAAA,CAAC;;QAGJ,IAAI,CAAC,YAAY,EAAE;YACjB,YAAY,GAAG,IAAI;;;AAIvB,IAAA,OAAO,YAAY;AACrB;AAEA,eAAe,4BAA4B,CAAC,EAC1C,WAAW,EACX,QAAQ,EACR,QAAQ,EACR,KAAK,GAMN,EAAA;IACC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,CAAC,OAAO,CAAC,EAAE;QACvC,OAAO,CAAC,IAAI,CACV,CAAwC,qCAAA,EAAA,OAAO,WAAW,CAAC,OAAO,CAAE,CAAA,CACrE;QACD;;IAGF,IAAI,CAAC,0BAA0B,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,EAAE;AACvD,QAAA,OAAO,CAAC,IAAI,CACV,CAAA,2DAAA,EAA8D,IAAI,CAAC,SAAS,CAC1E,WAAW,CAAC,OAAO,CACpB,CAAA,CAAE,CACJ;QACD;;IAGF,MAAM,IAAI,GAAG,KAAK,CAAC,cAAc,EAAE,IAAI,IAAI,CAAC;IAC5C,MAAM,gBAAgB,GAAG,4BAA4B,CAAC;QACpD,IAAI;QACJ,OAAO,EAAE,WAAW,CAAC,OAA+C;AACrE,KAAA,CAAC;AAEF,IAAA,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI;AAC1B,IAAA,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,IAAI,EAAE;AACjC,IAAA,MAAM,QAAQ,GAAG;AACf,QAAA,CAAC,SAAS,CAAC,UAAU,GAAG,gBAAgB;KACzC;AACD,IAAA,MAAM,EAAE,MAAM,EAAE,eAAe,EAAE,GAAG,mBAAmB,CACrD,IAAI,EACJ,gBAAgB,CACjB;AACD,IAAA,MAAM,MAAM,GAAG,IAAI,WAAW,CAAC;QAC7B,IAAI;QACJ,QAAQ;AACR,QAAA,OAAO,EAAE,eAAe;QACxB,YAAY,EAAE,QAAQ,CAAC,EAAG;AAC3B,KAAA,CAAC;AACF,IAAA,MAAM,WAAW,GAAkB;QACjC,KAAK;QACL,MAAM;KACP;IACD,MAAM,KAAK,CAAC;AACV,UAAE,UAAU,CAAC,WAAW,CAAC,QAAQ;AACjC,UAAE,MAAM,CAAC,WAAW,CAAC,QAAQ,EAAE,WAAW,EAAE,QAAQ,EAAE,KAAK,CAAC;AAE9D,IAAA,IAAI,KAAK,CAAC,cAAc,IAAI,IAAI,EAAE;AAChC,QAAA,KAAK,CAAC,cAAc,GAAG,IAAI,GAAG,EAAU;;IAG1C,KAAK,CAAC,cAAc,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAG,CAAC;AACxC;;;;"}
|
|
1
|
+
{"version":3,"file":"handlers.mjs","sources":["../../../src/tools/handlers.ts"],"sourcesContent":["/* eslint-disable no-console */\n// src/tools/handlers.ts\nimport { nanoid } from 'nanoid';\nimport { ToolMessage } from '@langchain/core/messages';\nimport type { AnthropicWebSearchResultBlockParam } from '@/llm/anthropic/types';\nimport type { ToolCall, ToolCallChunk } from '@langchain/core/messages/tool';\nimport type { MultiAgentGraph, StandardGraph } from '@/graphs';\nimport type { AgentContext } from '@/agents/AgentContext';\nimport type * as t from '@/types';\nimport {\n ToolCallTypes,\n GraphEvents,\n StepTypes,\n Providers,\n Constants,\n} from '@/common';\nimport {\n coerceAnthropicSearchResults,\n isAnthropicWebSearchResult,\n} from '@/tools/search/anthropic';\nimport { formatResultsForLLM } from '@/tools/search/format';\nimport { getMessageId } from '@/messages';\n\nexport async function handleToolCallChunks({\n graph,\n stepKey,\n toolCallChunks,\n metadata,\n}: {\n graph: StandardGraph | MultiAgentGraph;\n stepKey: string;\n toolCallChunks: ToolCallChunk[];\n metadata?: Record<string, unknown>;\n}): Promise<void> {\n let prevStepId: string;\n let prevRunStep: t.RunStep | undefined;\n try {\n prevStepId = graph.getStepIdByKey(stepKey);\n prevRunStep = graph.getRunStep(prevStepId);\n } catch {\n /** Edge Case: If no previous step exists, create a new message creation step */\n const message_id = getMessageId(stepKey, graph, true) ?? '';\n prevStepId = await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.MESSAGE_CREATION,\n message_creation: {\n message_id,\n },\n },\n metadata\n );\n prevRunStep = graph.getRunStep(prevStepId);\n }\n\n const _stepId = graph.getStepIdByKey(stepKey);\n\n /** Edge Case: Tool Call Run Step or `tool_call_ids` never dispatched */\n const tool_calls: ToolCall[] | undefined =\n prevStepId && prevRunStep && prevRunStep.type === StepTypes.MESSAGE_CREATION\n ? []\n : undefined;\n\n /** Edge Case: `id` and `name` fields cannot be empty strings */\n for (const toolCallChunk of toolCallChunks) {\n if (toolCallChunk.name === '') {\n toolCallChunk.name = undefined;\n }\n if (toolCallChunk.id === '') {\n toolCallChunk.id = undefined;\n } else if (\n tool_calls != null &&\n toolCallChunk.id != null &&\n toolCallChunk.name != null\n ) {\n tool_calls.push({\n args: {},\n id: toolCallChunk.id,\n name: toolCallChunk.name,\n type: ToolCallTypes.TOOL_CALL,\n });\n }\n }\n\n let stepId: string = _stepId;\n const alreadyDispatched =\n prevRunStep?.type === StepTypes.MESSAGE_CREATION &&\n graph.messageStepHasToolCalls.has(prevStepId);\n\n if (prevRunStep?.type === StepTypes.TOOL_CALLS) {\n /**\n * If previous step is already a tool_calls step, use that step ID\n * This ensures tool call deltas are dispatched to the correct step\n */\n stepId = prevStepId;\n } else if (\n !alreadyDispatched &&\n prevRunStep?.type === StepTypes.MESSAGE_CREATION\n ) {\n /**\n * Create tool_calls step as soon as we receive the first tool call chunk\n * This ensures deltas are always associated with the correct step\n *\n * NOTE: We do NOT dispatch an empty text block here because:\n * - Empty text blocks cause providers (Anthropic, Bedrock) to reject messages\n * - The tool_calls themselves are sufficient for the step\n * - Empty content with tool_call_ids gets stored in conversation history\n * and causes \"messages must have non-empty content\" errors on replay\n */\n graph.messageStepHasToolCalls.set(prevStepId, true);\n stepId = await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.TOOL_CALLS,\n tool_calls: tool_calls ?? [],\n },\n metadata\n );\n }\n await graph.dispatchRunStepDelta(stepId, {\n type: StepTypes.TOOL_CALLS,\n tool_calls: toolCallChunks,\n });\n}\n\nexport const handleToolCalls = async (\n toolCalls?: ToolCall[],\n metadata?: Record<string, unknown>,\n graph?: StandardGraph | MultiAgentGraph\n): Promise<void> => {\n if (!graph || !metadata) {\n console.warn(`Graph or metadata not found in ${event} event`);\n return;\n }\n\n if (!toolCalls) {\n return;\n }\n\n if (toolCalls.length === 0) {\n return;\n }\n\n const stepKey = graph.getStepKey(metadata);\n\n for (const tool_call of toolCalls) {\n const toolCallId = tool_call.id ?? `toolu_${nanoid()}`;\n tool_call.id = toolCallId;\n if (!toolCallId || graph.toolCallStepIds.has(toolCallId)) {\n continue;\n }\n\n let prevStepId = '';\n let prevRunStep: t.RunStep | undefined;\n try {\n prevStepId = graph.getStepIdByKey(stepKey);\n prevRunStep = graph.getRunStep(prevStepId);\n } catch {\n // no previous step\n }\n\n /**\n * NOTE: We do NOT dispatch empty text blocks with tool_call_ids because:\n * - Empty text blocks cause providers (Anthropic, Bedrock) to reject messages\n * - They get stored in conversation history and cause errors on replay:\n * \"messages must have non-empty content\" (Anthropic)\n * \"The content field in the Message object is empty\" (Bedrock)\n * - The tool_calls themselves are sufficient\n */\n\n /* If the previous step exists and is a message creation */\n if (\n prevStepId &&\n prevRunStep &&\n prevRunStep.type === StepTypes.MESSAGE_CREATION\n ) {\n graph.messageStepHasToolCalls.set(prevStepId, true);\n /* If the previous step doesn't exist or is not a message creation */\n } else if (\n !prevRunStep ||\n prevRunStep.type !== StepTypes.MESSAGE_CREATION\n ) {\n const messageId = getMessageId(stepKey, graph, true) ?? '';\n const stepId = await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.MESSAGE_CREATION,\n message_creation: {\n message_id: messageId,\n },\n },\n metadata\n );\n graph.messageStepHasToolCalls.set(stepId, true);\n }\n\n await graph.dispatchRunStep(\n stepKey,\n {\n type: StepTypes.TOOL_CALLS,\n tool_calls: [tool_call],\n },\n metadata\n );\n }\n};\n\nexport const toolResultTypes = new Set([\n // 'tool_use',\n // 'server_tool_use',\n // 'input_json_delta',\n 'tool_result',\n 'web_search_result',\n 'web_search_tool_result',\n]);\n\n/**\n * Handles the result of a server tool call; in other words, a provider's built-in tool.\n * As of 2025-07-06, only Anthropic handles server tool calls with this pattern.\n */\nexport async function handleServerToolResult({\n graph,\n content,\n metadata,\n agentContext,\n}: {\n graph: StandardGraph | MultiAgentGraph;\n content?: string | t.MessageContentComplex[];\n metadata?: Record<string, unknown>;\n agentContext?: AgentContext;\n}): Promise<boolean> {\n let skipHandling = false;\n if (agentContext?.provider !== Providers.ANTHROPIC) {\n return skipHandling;\n }\n if (\n typeof content === 'string' ||\n content == null ||\n content.length === 0 ||\n (content.length === 1 &&\n (content[0] as t.ToolResultContent).tool_use_id == null)\n ) {\n return skipHandling;\n }\n\n for (const contentPart of content) {\n const toolUseId = (contentPart as t.ToolResultContent).tool_use_id;\n if (toolUseId == null || toolUseId === '') {\n continue;\n }\n const stepId = graph.toolCallStepIds.get(toolUseId);\n if (stepId == null || stepId === '') {\n console.warn(\n `Tool use ID ${toolUseId} not found in graph, cannot dispatch tool result.`\n );\n continue;\n }\n const runStep = graph.getRunStep(stepId);\n if (!runStep) {\n console.warn(\n `Run step for ${stepId} does not exist, cannot dispatch tool result.`\n );\n continue;\n } else if (runStep.type !== StepTypes.TOOL_CALLS) {\n console.warn(\n `Run step for ${stepId} is not a tool call step, cannot dispatch tool result.`\n );\n continue;\n }\n\n const toolCall =\n runStep.stepDetails.type === StepTypes.TOOL_CALLS\n ? (runStep.stepDetails.tool_calls?.find(\n (toolCall) => toolCall.id === toolUseId\n ) as ToolCall)\n : undefined;\n\n if (!toolCall) {\n continue;\n }\n\n if (\n contentPart.type === 'web_search_result' ||\n contentPart.type === 'web_search_tool_result'\n ) {\n await handleAnthropicSearchResults({\n contentPart: contentPart as t.ToolResultContent,\n toolCall,\n metadata,\n graph,\n });\n }\n\n if (!skipHandling) {\n skipHandling = true;\n }\n }\n\n return skipHandling;\n}\n\nasync function handleAnthropicSearchResults({\n contentPart,\n toolCall,\n metadata,\n graph,\n}: {\n contentPart: t.ToolResultContent;\n toolCall: Partial<ToolCall>;\n metadata?: Record<string, unknown>;\n graph: StandardGraph | MultiAgentGraph;\n}): Promise<void> {\n if (!Array.isArray(contentPart.content)) {\n console.warn(\n `Expected content to be an array, got ${typeof contentPart.content}`\n );\n return;\n }\n\n if (!isAnthropicWebSearchResult(contentPart.content[0])) {\n console.warn(\n `Expected content to be an Anthropic web search result, got ${JSON.stringify(\n contentPart.content\n )}`\n );\n return;\n }\n\n const turn = graph.invokedToolIds?.size ?? 0;\n const searchResultData = coerceAnthropicSearchResults({\n turn,\n results: contentPart.content as AnthropicWebSearchResultBlockParam[],\n });\n\n const name = toolCall.name;\n const input = toolCall.args ?? {};\n const artifact = {\n [Constants.WEB_SEARCH]: searchResultData,\n };\n const { output: formattedOutput } = formatResultsForLLM(\n turn,\n searchResultData\n );\n const output = new ToolMessage({\n name,\n artifact,\n content: formattedOutput,\n tool_call_id: toolCall.id!,\n });\n const toolEndData: t.ToolEndData = {\n input,\n output,\n };\n await graph.handlerRegistry\n ?.getHandler(GraphEvents.TOOL_END)\n ?.handle(GraphEvents.TOOL_END, toolEndData, metadata, graph);\n\n if (graph.invokedToolIds == null) {\n graph.invokedToolIds = new Set<string>();\n }\n\n graph.invokedToolIds.add(toolCall.id!);\n}\n"],"names":[],"mappings":";;;;;;;;AAAA;AACA;AAsBO,eAAe,oBAAoB,CAAC,EACzC,KAAK,EACL,OAAO,EACP,cAAc,EACd,QAAQ,GAMT,EAAA;AACC,IAAA,IAAI,UAAkB;AACtB,IAAA,IAAI,WAAkC;AACtC,IAAA,IAAI;AACF,QAAA,UAAU,GAAG,KAAK,CAAC,cAAc,CAAC,OAAO,CAAC;AAC1C,QAAA,WAAW,GAAG,KAAK,CAAC,UAAU,CAAC,UAAU,CAAC;;AAC1C,IAAA,MAAM;;AAEN,QAAA,MAAM,UAAU,GAAG,YAAY,CAAC,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,IAAI,EAAE;AAC3D,QAAA,UAAU,GAAG,MAAM,KAAK,CAAC,eAAe,CACtC,OAAO,EACP;YACE,IAAI,EAAE,SAAS,CAAC,gBAAgB;AAChC,YAAA,gBAAgB,EAAE;gBAChB,UAAU;AACX,aAAA;SACF,EACD,QAAQ,CACT;AACD,QAAA,WAAW,GAAG,KAAK,CAAC,UAAU,CAAC,UAAU,CAAC;;IAG5C,MAAM,OAAO,GAAG,KAAK,CAAC,cAAc,CAAC,OAAO,CAAC;;AAG7C,IAAA,MAAM,UAAU,GACd,UAAU,IAAI,WAAW,IAAI,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC;AAC1D,UAAE;UACA,SAAS;;AAGf,IAAA,KAAK,MAAM,aAAa,IAAI,cAAc,EAAE;AAC1C,QAAA,IAAI,aAAa,CAAC,IAAI,KAAK,EAAE,EAAE;AAC7B,YAAA,aAAa,CAAC,IAAI,GAAG,SAAS;;AAEhC,QAAA,IAAI,aAAa,CAAC,EAAE,KAAK,EAAE,EAAE;AAC3B,YAAA,aAAa,CAAC,EAAE,GAAG,SAAS;;aACvB,IACL,UAAU,IAAI,IAAI;YAClB,aAAa,CAAC,EAAE,IAAI,IAAI;AACxB,YAAA,aAAa,CAAC,IAAI,IAAI,IAAI,EAC1B;YACA,UAAU,CAAC,IAAI,CAAC;AACd,gBAAA,IAAI,EAAE,EAAE;gBACR,EAAE,EAAE,aAAa,CAAC,EAAE;gBACpB,IAAI,EAAE,aAAa,CAAC,IAAI;gBACxB,IAAI,EAAE,aAAa,CAAC,SAAS;AAC9B,aAAA,CAAC;;;IAIN,IAAI,MAAM,GAAW,OAAO;IAC5B,MAAM,iBAAiB,GACrB,WAAW,EAAE,IAAI,KAAK,SAAS,CAAC,gBAAgB;AAChD,QAAA,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,UAAU,CAAC;IAE/C,IAAI,WAAW,EAAE,IAAI,KAAK,SAAS,CAAC,UAAU,EAAE;AAC9C;;;AAGG;QACH,MAAM,GAAG,UAAU;;AACd,SAAA,IACL,CAAC,iBAAiB;AAClB,QAAA,WAAW,EAAE,IAAI,KAAK,SAAS,CAAC,gBAAgB,EAChD;AACA;;;;;;;;;AASG;QACH,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC;AACnD,QAAA,MAAM,GAAG,MAAM,KAAK,CAAC,eAAe,CAClC,OAAO,EACP;YACE,IAAI,EAAE,SAAS,CAAC,UAAU;YAC1B,UAAU,EAAE,UAAU,IAAI,EAAE;SAC7B,EACD,QAAQ,CACT;;AAEH,IAAA,MAAM,KAAK,CAAC,oBAAoB,CAAC,MAAM,EAAE;QACvC,IAAI,EAAE,SAAS,CAAC,UAAU;AAC1B,QAAA,UAAU,EAAE,cAAc;AAC3B,KAAA,CAAC;AACJ;AAEO,MAAM,eAAe,GAAG,OAC7B,SAAsB,EACtB,QAAkC,EAClC,KAAuC,KACtB;AACjB,IAAA,IAAI,CAAC,KAAK,IAAI,CAAC,QAAQ,EAAE;AACvB,QAAA,OAAO,CAAC,IAAI,CAAC,kCAAkC,KAAK,CAAA,MAAA,CAAQ,CAAC;QAC7D;;IAGF,IAAI,CAAC,SAAS,EAAE;QACd;;AAGF,IAAA,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE;QAC1B;;IAGF,MAAM,OAAO,GAAG,KAAK,CAAC,UAAU,CAAC,QAAQ,CAAC;AAE1C,IAAA,KAAK,MAAM,SAAS,IAAI,SAAS,EAAE;QACjC,MAAM,UAAU,GAAG,SAAS,CAAC,EAAE,IAAI,CAAS,MAAA,EAAA,MAAM,EAAE,CAAA,CAAE;AACtD,QAAA,SAAS,CAAC,EAAE,GAAG,UAAU;AACzB,QAAA,IAAI,CAAC,UAAU,IAAI,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE;YACxD;;QAGF,IAAI,UAAU,GAAG,EAAE;AACnB,QAAA,IAAI,WAAkC;AACtC,QAAA,IAAI;AACF,YAAA,UAAU,GAAG,KAAK,CAAC,cAAc,CAAC,OAAO,CAAC;AAC1C,YAAA,WAAW,GAAG,KAAK,CAAC,UAAU,CAAC,UAAU,CAAC;;AAC1C,QAAA,MAAM;;;AAIR;;;;;;;AAOG;;AAGH,QAAA,IACE,UAAU;YACV,WAAW;AACX,YAAA,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC,gBAAgB,EAC/C;YACA,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC;;;AAE9C,aAAA,IACL,CAAC,WAAW;AACZ,YAAA,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC,gBAAgB,EAC/C;AACA,YAAA,MAAM,SAAS,GAAG,YAAY,CAAC,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,IAAI,EAAE;YAC1D,MAAM,MAAM,GAAG,MAAM,KAAK,CAAC,eAAe,CACxC,OAAO,EACP;gBACE,IAAI,EAAE,SAAS,CAAC,gBAAgB;AAChC,gBAAA,gBAAgB,EAAE;AAChB,oBAAA,UAAU,EAAE,SAAS;AACtB,iBAAA;aACF,EACD,QAAQ,CACT;YACD,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC;;AAGjD,QAAA,MAAM,KAAK,CAAC,eAAe,CACzB,OAAO,EACP;YACE,IAAI,EAAE,SAAS,CAAC,UAAU;YAC1B,UAAU,EAAE,CAAC,SAAS,CAAC;SACxB,EACD,QAAQ,CACT;;AAEL;AAEa,MAAA,eAAe,GAAG,IAAI,GAAG,CAAC;;;;IAIrC,aAAa;IACb,mBAAmB;IACnB,wBAAwB;AACzB,CAAA;AAED;;;AAGG;AACI,eAAe,sBAAsB,CAAC,EAC3C,KAAK,EACL,OAAO,EACP,QAAQ,EACR,YAAY,GAMb,EAAA;IACC,IAAI,YAAY,GAAG,KAAK;IACxB,IAAI,YAAY,EAAE,QAAQ,KAAK,SAAS,CAAC,SAAS,EAAE;AAClD,QAAA,OAAO,YAAY;;IAErB,IACE,OAAO,OAAO,KAAK,QAAQ;AAC3B,QAAA,OAAO,IAAI,IAAI;QACf,OAAO,CAAC,MAAM,KAAK,CAAC;AACpB,SAAC,OAAO,CAAC,MAAM,KAAK,CAAC;YAClB,OAAO,CAAC,CAAC,CAAyB,CAAC,WAAW,IAAI,IAAI,CAAC,EAC1D;AACA,QAAA,OAAO,YAAY;;AAGrB,IAAA,KAAK,MAAM,WAAW,IAAI,OAAO,EAAE;AACjC,QAAA,MAAM,SAAS,GAAI,WAAmC,CAAC,WAAW;QAClE,IAAI,SAAS,IAAI,IAAI,IAAI,SAAS,KAAK,EAAE,EAAE;YACzC;;QAEF,MAAM,MAAM,GAAG,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,SAAS,CAAC;QACnD,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,KAAK,EAAE,EAAE;AACnC,YAAA,OAAO,CAAC,IAAI,CACV,eAAe,SAAS,CAAA,iDAAA,CAAmD,CAC5E;YACD;;QAEF,MAAM,OAAO,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC;QACxC,IAAI,CAAC,OAAO,EAAE;AACZ,YAAA,OAAO,CAAC,IAAI,CACV,gBAAgB,MAAM,CAAA,6CAAA,CAA+C,CACtE;YACD;;aACK,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,CAAC,UAAU,EAAE;AAChD,YAAA,OAAO,CAAC,IAAI,CACV,gBAAgB,MAAM,CAAA,sDAAA,CAAwD,CAC/E;YACD;;QAGF,MAAM,QAAQ,GACZ,OAAO,CAAC,WAAW,CAAC,IAAI,KAAK,SAAS,CAAC;AACrC,cAAG,OAAO,CAAC,WAAW,CAAC,UAAU,EAAE,IAAI,CACrC,CAAC,QAAQ,KAAK,QAAQ,CAAC,EAAE,KAAK,SAAS;cAEvC,SAAS;QAEf,IAAI,CAAC,QAAQ,EAAE;YACb;;AAGF,QAAA,IACE,WAAW,CAAC,IAAI,KAAK,mBAAmB;AACxC,YAAA,WAAW,CAAC,IAAI,KAAK,wBAAwB,EAC7C;AACA,YAAA,MAAM,4BAA4B,CAAC;AACjC,gBAAA,WAAW,EAAE,WAAkC;gBAC/C,QAAQ;gBACR,QAAQ;gBACR,KAAK;AACN,aAAA,CAAC;;QAGJ,IAAI,CAAC,YAAY,EAAE;YACjB,YAAY,GAAG,IAAI;;;AAIvB,IAAA,OAAO,YAAY;AACrB;AAEA,eAAe,4BAA4B,CAAC,EAC1C,WAAW,EACX,QAAQ,EACR,QAAQ,EACR,KAAK,GAMN,EAAA;IACC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,CAAC,OAAO,CAAC,EAAE;QACvC,OAAO,CAAC,IAAI,CACV,CAAwC,qCAAA,EAAA,OAAO,WAAW,CAAC,OAAO,CAAE,CAAA,CACrE;QACD;;IAGF,IAAI,CAAC,0BAA0B,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,EAAE;AACvD,QAAA,OAAO,CAAC,IAAI,CACV,CAAA,2DAAA,EAA8D,IAAI,CAAC,SAAS,CAC1E,WAAW,CAAC,OAAO,CACpB,CAAA,CAAE,CACJ;QACD;;IAGF,MAAM,IAAI,GAAG,KAAK,CAAC,cAAc,EAAE,IAAI,IAAI,CAAC;IAC5C,MAAM,gBAAgB,GAAG,4BAA4B,CAAC;QACpD,IAAI;QACJ,OAAO,EAAE,WAAW,CAAC,OAA+C;AACrE,KAAA,CAAC;AAEF,IAAA,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI;AAC1B,IAAA,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,IAAI,EAAE;AACjC,IAAA,MAAM,QAAQ,GAAG;AACf,QAAA,CAAC,SAAS,CAAC,UAAU,GAAG,gBAAgB;KACzC;AACD,IAAA,MAAM,EAAE,MAAM,EAAE,eAAe,EAAE,GAAG,mBAAmB,CACrD,IAAI,EACJ,gBAAgB,CACjB;AACD,IAAA,MAAM,MAAM,GAAG,IAAI,WAAW,CAAC;QAC7B,IAAI;QACJ,QAAQ;AACR,QAAA,OAAO,EAAE,eAAe;QACxB,YAAY,EAAE,QAAQ,CAAC,EAAG;AAC3B,KAAA,CAAC;AACF,IAAA,MAAM,WAAW,GAAkB;QACjC,KAAK;QACL,MAAM;KACP;IACD,MAAM,KAAK,CAAC;AACV,UAAE,UAAU,CAAC,WAAW,CAAC,QAAQ;AACjC,UAAE,MAAM,CAAC,WAAW,CAAC,QAAQ,EAAE,WAAW,EAAE,QAAQ,EAAE,KAAK,CAAC;AAE9D,IAAA,IAAI,KAAK,CAAC,cAAc,IAAI,IAAI,EAAE;AAChC,QAAA,KAAK,CAAC,cAAc,GAAG,IAAI,GAAG,EAAU;;IAG1C,KAAK,CAAC,cAAc,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAG,CAAC;AACxC;;;;"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"anthropic.mjs","sources":["../../../../src/tools/search/anthropic.ts"],"sourcesContent":["import type {\
|
|
1
|
+
{"version":3,"file":"anthropic.mjs","sources":["../../../../src/tools/search/anthropic.ts"],"sourcesContent":["import type {\n AnthropicTextBlockParam,\n AnthropicWebSearchResultBlockParam,\n} from '@/llm/anthropic/types';\nimport type { SearchResultData, ProcessedOrganic } from './types';\nimport { getAttribution } from './utils';\n\n/**\n * Coerces Anthropic web search results to the SearchResultData format\n * @param results - Array of Anthropic web search results\n * @param turn - The turn number to associate with these results\n * @returns SearchResultData with minimal ProcessedOrganic items\n */\nexport function coerceAnthropicSearchResults({\n results,\n turn = 0,\n}: {\n results: (AnthropicTextBlockParam | AnthropicWebSearchResultBlockParam)[];\n turn?: number;\n}): SearchResultData {\n const organic: ProcessedOrganic[] = results\n .filter((result) => result.type === 'web_search_result')\n .map((result, index) => ({\n link: result.url,\n position: index + 1,\n title: result.title,\n date: result.page_age ?? undefined,\n attribution: getAttribution(result.url),\n }));\n\n return {\n turn,\n organic,\n };\n}\n\n/**\n * Helper function to check if an object is an Anthropic web search result\n */\nexport function isAnthropicWebSearchResult(\n obj: unknown\n): obj is AnthropicWebSearchResultBlockParam {\n return (\n typeof obj === 'object' &&\n obj !== null &&\n 'type' in obj &&\n obj.type === 'web_search_result' &&\n 'url' in obj &&\n typeof (obj as Record<string, unknown>).url === 'string'\n );\n}\n"],"names":[],"mappings":";;AAOA;;;;;AAKG;AACG,SAAU,4BAA4B,CAAC,EAC3C,OAAO,EACP,IAAI,GAAG,CAAC,GAIT,EAAA;IACC,MAAM,OAAO,GAAuB;SACjC,MAAM,CAAC,CAAC,MAAM,KAAK,MAAM,CAAC,IAAI,KAAK,mBAAmB;SACtD,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,MAAM;QACvB,IAAI,EAAE,MAAM,CAAC,GAAG;QAChB,QAAQ,EAAE,KAAK,GAAG,CAAC;QACnB,KAAK,EAAE,MAAM,CAAC,KAAK;AACnB,QAAA,IAAI,EAAE,MAAM,CAAC,QAAQ,IAAI,SAAS;AAClC,QAAA,WAAW,EAAE,cAAc,CAAC,MAAM,CAAC,GAAG,CAAC;AACxC,KAAA,CAAC,CAAC;IAEL,OAAO;QACL,IAAI;QACJ,OAAO;KACR;AACH;AAEA;;AAEG;AACG,SAAU,0BAA0B,CACxC,GAAY,EAAA;AAEZ,IAAA,QACE,OAAO,GAAG,KAAK,QAAQ;AACvB,QAAA,GAAG,KAAK,IAAI;AACZ,QAAA,MAAM,IAAI,GAAG;QACb,GAAG,CAAC,IAAI,KAAK,mBAAmB;AAChC,QAAA,KAAK,IAAI,GAAG;AACZ,QAAA,OAAQ,GAA+B,CAAC,GAAG,KAAK,QAAQ;AAE5D;;;;"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"content.mjs","sources":["../../../../src/tools/search/content.ts"],"sourcesContent":["import * as cheerio from 'cheerio';\
|
|
1
|
+
{"version":3,"file":"content.mjs","sources":["../../../../src/tools/search/content.ts"],"sourcesContent":["import * as cheerio from 'cheerio';\nimport type { References, MediaReference } from './types';\n\nexport function processContent(\n html: string,\n markdown: string\n): {\n markdown: string;\n} & References {\n const linkMap = new Map<string, MediaReference>();\n const imageMap = new Map<string, MediaReference>();\n const videoMap = new Map<string, MediaReference>();\n const iframeMap = new Map<string, MediaReference>();\n\n const $ = cheerio.load(html, {\n xmlMode: false,\n });\n\n // Extract all media references\n $('a[href]').each((_, el) => {\n const href = $(el).attr('href');\n if (href != null && href) {\n linkMap.set(href, {\n originalUrl: href,\n title: $(el).attr('title'),\n text: $(el).text().trim(),\n });\n }\n });\n\n $('img[src]').each((_, el) => {\n const src = $(el).attr('src');\n if (src != null && src) {\n imageMap.set(src, {\n originalUrl: src,\n title: $(el).attr('alt') ?? $(el).attr('title'),\n });\n }\n });\n\n // Handle videos (dedicated video elements and video platforms in iframes)\n $('video[src], iframe[src*=\"youtube\"], iframe[src*=\"vimeo\"]').each(\n (_, el) => {\n const src = $(el).attr('src');\n if (src != null && src) {\n videoMap.set(src, {\n originalUrl: src,\n title: $(el).attr('title'),\n });\n }\n }\n );\n\n // Handle all other generic iframes that aren't already captured as videos\n $('iframe').each((_, el) => {\n const src = $(el).attr('src');\n if (\n src != null &&\n src &&\n !src.includes('youtube') &&\n !src.includes('vimeo')\n ) {\n iframeMap.set(src, {\n originalUrl: src,\n title: $(el).attr('title'),\n });\n }\n });\n\n // Create lookup maps with indices\n const linkIndexMap = new Map<string, number>();\n const imageIndexMap = new Map<string, number>();\n const videoIndexMap = new Map<string, number>();\n const iframeIndexMap = new Map<string, number>();\n\n Array.from(linkMap.keys()).forEach((url, i) => linkIndexMap.set(url, i + 1));\n Array.from(imageMap.keys()).forEach((url, i) =>\n imageIndexMap.set(url, i + 1)\n );\n Array.from(videoMap.keys()).forEach((url, i) =>\n videoIndexMap.set(url, i + 1)\n );\n Array.from(iframeMap.keys()).forEach((url, i) =>\n iframeIndexMap.set(url, i + 1)\n );\n\n // Process the markdown\n let result = markdown;\n\n // Replace each URL one by one, starting with the longest URLs first to avoid partial matches\n const allUrls = [\n ...Array.from(imageMap.keys()).map((url) => ({\n url,\n type: 'image',\n idx: imageIndexMap.get(url),\n })),\n ...Array.from(videoMap.keys()).map((url) => ({\n url,\n type: 'video',\n idx: videoIndexMap.get(url),\n })),\n ...Array.from(iframeMap.keys()).map((url) => ({\n url,\n type: 'iframe',\n idx: iframeIndexMap.get(url),\n })),\n ...Array.from(linkMap.keys()).map((url) => ({\n url,\n type: 'link',\n idx: linkIndexMap.get(url),\n })),\n ].sort((a, b) => b.url.length - a.url.length);\n\n // Create a function to escape special characters in URLs for regex\n function escapeRegex(string: string): string {\n return string.replace(/[.*+?^${}()|[\\]\\\\]/g, '\\\\$&');\n }\n\n // Replace each URL in the markdown\n for (const { url, type, idx } of allUrls) {\n // Create a regex that captures URLs in markdown links\n const regex = new RegExp(`\\\\(${escapeRegex(url)}(?:\\\\s+\"[^\"]*\")?\\\\)`, 'g');\n\n result = result.replace(regex, (match) => {\n // Keep any title attribute that might exist\n const titleMatch = match.match(/\\s+\"([^\"]*)\"/);\n const titlePart = titleMatch ? ` \"${titleMatch[1]}\"` : '';\n\n return `(${type}#${idx}${titlePart})`;\n });\n }\n\n iframeMap.clear();\n const links = Array.from(linkMap.values());\n linkMap.clear();\n const images = Array.from(imageMap.values());\n imageMap.clear();\n const videos = Array.from(videoMap.values());\n videoMap.clear();\n\n return {\n markdown: result,\n links,\n images,\n videos,\n };\n}\n"],"names":[],"mappings":";;AAGgB,SAAA,cAAc,CAC5B,IAAY,EACZ,QAAgB,EAAA;AAIhB,IAAA,MAAM,OAAO,GAAG,IAAI,GAAG,EAA0B;AACjD,IAAA,MAAM,QAAQ,GAAG,IAAI,GAAG,EAA0B;AAClD,IAAA,MAAM,QAAQ,GAAG,IAAI,GAAG,EAA0B;AAClD,IAAA,MAAM,SAAS,GAAG,IAAI,GAAG,EAA0B;AAEnD,IAAA,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,EAAE;AAC3B,QAAA,OAAO,EAAE,KAAK;AACf,KAAA,CAAC;;IAGF,CAAC,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,KAAI;QAC1B,MAAM,IAAI,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC;AAC/B,QAAA,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,EAAE;AACxB,YAAA,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE;AAChB,gBAAA,WAAW,EAAE,IAAI;gBACjB,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC;gBAC1B,IAAI,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE;AAC1B,aAAA,CAAC;;AAEN,KAAC,CAAC;IAEF,CAAC,CAAC,UAAU,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,KAAI;QAC3B,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC;AAC7B,QAAA,IAAI,GAAG,IAAI,IAAI,IAAI,GAAG,EAAE;AACtB,YAAA,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE;AAChB,gBAAA,WAAW,EAAE,GAAG;AAChB,gBAAA,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC;AAChD,aAAA,CAAC;;AAEN,KAAC,CAAC;;IAGF,CAAC,CAAC,0DAA0D,CAAC,CAAC,IAAI,CAChE,CAAC,CAAC,EAAE,EAAE,KAAI;QACR,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC;AAC7B,QAAA,IAAI,GAAG,IAAI,IAAI,IAAI,GAAG,EAAE;AACtB,YAAA,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE;AAChB,gBAAA,WAAW,EAAE,GAAG;gBAChB,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC;AAC3B,aAAA,CAAC;;AAEN,KAAC,CACF;;IAGD,CAAC,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,KAAI;QACzB,MAAM,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC;QAC7B,IACE,GAAG,IAAI,IAAI;YACX,GAAG;AACH,YAAA,CAAC,GAAG,CAAC,QAAQ,CAAC,SAAS,CAAC;AACxB,YAAA,CAAC,GAAG,CAAC,QAAQ,CAAC,OAAO,CAAC,EACtB;AACA,YAAA,SAAS,CAAC,GAAG,CAAC,GAAG,EAAE;AACjB,gBAAA,WAAW,EAAE,GAAG;gBAChB,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC;AAC3B,aAAA,CAAC;;AAEN,KAAC,CAAC;;AAGF,IAAA,MAAM,YAAY,GAAG,IAAI,GAAG,EAAkB;AAC9C,IAAA,MAAM,aAAa,GAAG,IAAI,GAAG,EAAkB;AAC/C,IAAA,MAAM,aAAa,GAAG,IAAI,GAAG,EAAkB;AAC/C,IAAA,MAAM,cAAc,GAAG,IAAI,GAAG,EAAkB;AAEhD,IAAA,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,CAAC,KAAK,YAAY,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;AAC5E,IAAA,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,CAAC,KACzC,aAAa,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAC9B;AACD,IAAA,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,CAAC,KACzC,aAAa,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAC9B;AACD,IAAA,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,CAAC,KAC1C,cAAc,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAC/B;;IAGD,IAAI,MAAM,GAAG,QAAQ;;AAGrB,IAAA,MAAM,OAAO,GAAG;AACd,QAAA,GAAG,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,MAAM;YAC3C,GAAG;AACH,YAAA,IAAI,EAAE,OAAO;AACb,YAAA,GAAG,EAAE,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC;AAC5B,SAAA,CAAC,CAAC;AACH,QAAA,GAAG,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,MAAM;YAC3C,GAAG;AACH,YAAA,IAAI,EAAE,OAAO;AACb,YAAA,GAAG,EAAE,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC;AAC5B,SAAA,CAAC,CAAC;AACH,QAAA,GAAG,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,MAAM;YAC5C,GAAG;AACH,YAAA,IAAI,EAAE,QAAQ;AACd,YAAA,GAAG,EAAE,cAAc,CAAC,GAAG,CAAC,GAAG,CAAC;AAC7B,SAAA,CAAC,CAAC;AACH,QAAA,GAAG,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,MAAM;YAC1C,GAAG;AACH,YAAA,IAAI,EAAE,MAAM;AACZ,YAAA,GAAG,EAAE,YAAY,CAAC,GAAG,CAAC,GAAG,CAAC;AAC3B,SAAA,CAAC,CAAC;KACJ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC;;IAG7C,SAAS,WAAW,CAAC,MAAc,EAAA;QACjC,OAAO,MAAM,CAAC,OAAO,CAAC,qBAAqB,EAAE,MAAM,CAAC;;;IAItD,KAAK,MAAM,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,OAAO,EAAE;;AAExC,QAAA,MAAM,KAAK,GAAG,IAAI,MAAM,CAAC,CAAM,GAAA,EAAA,WAAW,CAAC,GAAG,CAAC,CAAA,mBAAA,CAAqB,EAAE,GAAG,CAAC;QAE1E,MAAM,GAAG,MAAM,CAAC,OAAO,CAAC,KAAK,EAAE,CAAC,KAAK,KAAI;;YAEvC,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,CAAC,cAAc,CAAC;AAC9C,YAAA,MAAM,SAAS,GAAG,UAAU,GAAG,CAAK,EAAA,EAAA,UAAU,CAAC,CAAC,CAAC,CAAG,CAAA,CAAA,GAAG,EAAE;AAEzD,YAAA,OAAO,IAAI,IAAI,CAAA,CAAA,EAAI,GAAG,CAAG,EAAA,SAAS,GAAG;AACvC,SAAC,CAAC;;IAGJ,SAAS,CAAC,KAAK,EAAE;IACjB,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;IAC1C,OAAO,CAAC,KAAK,EAAE;IACf,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,EAAE,CAAC;IAC5C,QAAQ,CAAC,KAAK,EAAE;IAChB,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,EAAE,CAAC;IAC5C,QAAQ,CAAC,KAAK,EAAE;IAEhB,OAAO;AACL,QAAA,QAAQ,EAAE,MAAM;QAChB,KAAK;QACL,MAAM;QACN,MAAM;KACP;AACH;;;;"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"firecrawl.mjs","sources":["../../../../src/tools/search/firecrawl.ts"],"sourcesContent":["import axios from 'axios';\r\nimport { processContent } from './content';\r\nimport type * as t from './types';\r\nimport { createDefaultLogger } from './utils';\r\n\r\n/**\r\n * Firecrawl scraper implementation\r\n * Uses the Firecrawl API to scrape web pages\r\n */\r\nexport class FirecrawlScraper implements t.BaseScraper {\r\n private apiKey: string;\r\n private apiUrl: string;\r\n private version: string;\r\n private defaultFormats: string[];\r\n private timeout: number;\r\n private logger: t.Logger;\r\n private includeTags?: string[];\r\n private excludeTags?: string[];\r\n private waitFor?: number;\r\n private maxAge?: number;\r\n private mobile?: boolean;\r\n private skipTlsVerification?: boolean;\r\n private blockAds?: boolean;\r\n private removeBase64Images?: boolean;\r\n private parsePDF?: boolean;\r\n private storeInCache?: boolean;\r\n private zeroDataRetention?: boolean;\r\n private headers?: Record<string, string>;\r\n private location?: { country?: string; languages?: string[] };\r\n private onlyMainContent?: boolean;\r\n private changeTrackingOptions?: object;\r\n\r\n constructor(config: t.FirecrawlScraperConfig = {}) {\r\n this.apiKey = config.apiKey ?? process.env.FIRECRAWL_API_KEY ?? '';\r\n\r\n this.version = config.version ?? 'v2';\r\n\r\n const baseUrl =\r\n config.apiUrl ??\r\n process.env.FIRECRAWL_BASE_URL ??\r\n 'https://api.firecrawl.dev';\r\n this.apiUrl = `${baseUrl.replace(/\\/+$/, '')}/${this.version}/scrape`;\r\n\r\n this.defaultFormats = config.formats ?? ['markdown', 'rawHtml'];\r\n this.timeout = config.timeout ?? 7500;\r\n\r\n this.logger = config.logger || createDefaultLogger();\r\n\r\n this.includeTags = config.includeTags;\r\n this.excludeTags = config.excludeTags;\r\n this.waitFor = config.waitFor;\r\n this.maxAge = config.maxAge;\r\n this.mobile = config.mobile;\r\n this.skipTlsVerification = config.skipTlsVerification;\r\n this.blockAds = config.blockAds;\r\n this.removeBase64Images = config.removeBase64Images;\r\n this.parsePDF = config.parsePDF;\r\n this.storeInCache = config.storeInCache;\r\n this.zeroDataRetention = config.zeroDataRetention;\r\n this.headers = config.headers;\r\n this.location = config.location;\r\n this.onlyMainContent = config.onlyMainContent;\r\n this.changeTrackingOptions = config.changeTrackingOptions;\r\n\r\n if (!this.apiKey) {\r\n this.logger.warn('FIRECRAWL_API_KEY is not set. Scraping will not work.');\r\n }\r\n\r\n this.logger.debug(\r\n `Firecrawl scraper initialized with API URL: ${this.apiUrl}`\r\n );\r\n }\r\n\r\n /**\r\n * Scrape a single URL\r\n * @param url URL to scrape\r\n * @param options Scrape options\r\n * @returns Scrape response\r\n */\r\n async scrapeUrl(\r\n url: string,\r\n options: t.FirecrawlScrapeOptions = {}\r\n ): Promise<[string, t.FirecrawlScrapeResponse]> {\r\n if (!this.apiKey) {\r\n return [\r\n url,\r\n {\r\n success: false,\r\n error: 'FIRECRAWL_API_KEY is not set',\r\n },\r\n ];\r\n }\r\n\r\n try {\r\n const payload = omitUndefined({\r\n url,\r\n formats: options.formats ?? this.defaultFormats,\r\n includeTags: options.includeTags ?? this.includeTags,\r\n excludeTags: options.excludeTags ?? this.excludeTags,\r\n headers: options.headers ?? this.headers,\r\n waitFor: options.waitFor ?? this.waitFor,\r\n timeout: options.timeout ?? this.timeout,\r\n onlyMainContent: options.onlyMainContent ?? this.onlyMainContent,\r\n maxAge: options.maxAge ?? this.maxAge,\r\n mobile: options.mobile ?? this.mobile,\r\n skipTlsVerification:\r\n options.skipTlsVerification ?? this.skipTlsVerification,\r\n parsePDF: options.parsePDF ?? this.parsePDF,\r\n location: options.location ?? this.location,\r\n removeBase64Images:\r\n options.removeBase64Images ?? this.removeBase64Images,\r\n blockAds: options.blockAds ?? this.blockAds,\r\n storeInCache: options.storeInCache ?? this.storeInCache,\r\n zeroDataRetention: options.zeroDataRetention ?? this.zeroDataRetention,\r\n changeTrackingOptions:\r\n options.changeTrackingOptions ?? this.changeTrackingOptions,\r\n });\r\n const response = await axios.post(this.apiUrl, payload, {\r\n headers: {\r\n 'Content-Type': 'application/json',\r\n Authorization: `Bearer ${this.apiKey}`,\r\n },\r\n timeout: this.timeout,\r\n });\r\n\r\n return [url, response.data];\r\n } catch (error) {\r\n const errorMessage =\r\n error instanceof Error ? error.message : String(error);\r\n return [\r\n url,\r\n {\r\n success: false,\r\n error: `Firecrawl API request failed: ${errorMessage}`,\r\n },\r\n ];\r\n }\r\n }\r\n\r\n /**\r\n * Extract content from scrape response\r\n * @param response Scrape response\r\n * @returns Extracted content or empty string if not available\r\n */\r\n extractContent(\r\n response: t.FirecrawlScrapeResponse\r\n ): [string, undefined | t.References] {\r\n if (!response.success || !response.data) {\r\n return ['', undefined];\r\n }\r\n\r\n if (response.data.markdown != null && response.data.html != null) {\r\n try {\r\n const { markdown, ...rest } = processContent(\r\n response.data.html,\r\n response.data.markdown\r\n );\r\n return [markdown, rest];\r\n } catch (error) {\r\n this.logger.error('Error processing content:', error);\r\n return [response.data.markdown, undefined];\r\n }\r\n } else if (response.data.markdown != null) {\r\n return [response.data.markdown, undefined];\r\n }\r\n\r\n // Fall back to HTML content\r\n if (response.data.html != null) {\r\n return [response.data.html, undefined];\r\n }\r\n\r\n // Fall back to raw HTML content\r\n if (response.data.rawHtml != null) {\r\n return [response.data.rawHtml, undefined];\r\n }\r\n\r\n return ['', undefined];\r\n }\r\n\r\n /**\r\n * Extract metadata from scrape response\r\n * @param response Scrape response\r\n * @returns Metadata object\r\n */\r\n extractMetadata(response: t.FirecrawlScrapeResponse): t.ScrapeMetadata {\r\n if (!response.success || !response.data || !response.data.metadata) {\r\n return {};\r\n }\r\n\r\n return response.data.metadata;\r\n }\r\n}\r\n\r\n/**\r\n * Create a Firecrawl scraper instance\r\n * @param config Scraper configuration\r\n * @returns Firecrawl scraper instance\r\n */\r\nexport const createFirecrawlScraper = (\r\n config: t.FirecrawlScraperConfig = {}\r\n): FirecrawlScraper => {\r\n return new FirecrawlScraper(config);\r\n};\r\n\r\n// Helper function to clean up payload for firecrawl\r\nfunction omitUndefined<T extends object>(obj: T): Partial<T> {\r\n return Object.fromEntries(\r\n Object.entries(obj).filter(([, v]) => v !== undefined)\r\n ) as Partial<T>;\r\n}\r\n"],"names":[],"mappings":";;;;AAKA;;;AAGG;MACU,gBAAgB,CAAA;AACnB,IAAA,MAAM;AACN,IAAA,MAAM;AACN,IAAA,OAAO;AACP,IAAA,cAAc;AACd,IAAA,OAAO;AACP,IAAA,MAAM;AACN,IAAA,WAAW;AACX,IAAA,WAAW;AACX,IAAA,OAAO;AACP,IAAA,MAAM;AACN,IAAA,MAAM;AACN,IAAA,mBAAmB;AACnB,IAAA,QAAQ;AACR,IAAA,kBAAkB;AAClB,IAAA,QAAQ;AACR,IAAA,YAAY;AACZ,IAAA,iBAAiB;AACjB,IAAA,OAAO;AACP,IAAA,QAAQ;AACR,IAAA,eAAe;AACf,IAAA,qBAAqB;AAE7B,IAAA,WAAA,CAAY,SAAmC,EAAE,EAAA;AAC/C,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,iBAAiB,IAAI,EAAE;QAElE,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,IAAI;AAErC,QAAA,MAAM,OAAO,GACX,MAAM,CAAC,MAAM;YACb,OAAO,CAAC,GAAG,CAAC,kBAAkB;AAC9B,YAAA,2BAA2B;AAC7B,QAAA,IAAI,CAAC,MAAM,GAAG,CAAG,EAAA,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAA,CAAA,EAAI,IAAI,CAAC,OAAO,SAAS;AAErE,QAAA,IAAI,CAAC,cAAc,GAAG,MAAM,CAAC,OAAO,IAAI,CAAC,UAAU,EAAE,SAAS,CAAC;QAC/D,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,IAAI;QAErC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,mBAAmB,EAAE;AAEpD,QAAA,IAAI,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW;AACrC,QAAA,IAAI,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW;AACrC,QAAA,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO;AAC7B,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM;AAC3B,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM;AAC3B,QAAA,IAAI,CAAC,mBAAmB,GAAG,MAAM,CAAC,mBAAmB;AACrD,QAAA,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ;AAC/B,QAAA,IAAI,CAAC,kBAAkB,GAAG,MAAM,CAAC,kBAAkB;AACnD,QAAA,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ;AAC/B,QAAA,IAAI,CAAC,YAAY,GAAG,MAAM,CAAC,YAAY;AACvC,QAAA,IAAI,CAAC,iBAAiB,GAAG,MAAM,CAAC,iBAAiB;AACjD,QAAA,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO;AAC7B,QAAA,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ;AAC/B,QAAA,IAAI,CAAC,eAAe,GAAG,MAAM,CAAC,eAAe;AAC7C,QAAA,IAAI,CAAC,qBAAqB,GAAG,MAAM,CAAC,qBAAqB;AAEzD,QAAA,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;AAChB,YAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,uDAAuD,CAAC;;QAG3E,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,CAA+C,4CAAA,EAAA,IAAI,CAAC,MAAM,CAAE,CAAA,CAC7D;;AAGH;;;;;AAKG;AACH,IAAA,MAAM,SAAS,CACb,GAAW,EACX,UAAoC,EAAE,EAAA;AAEtC,QAAA,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;YAChB,OAAO;gBACL,GAAG;AACH,gBAAA;AACE,oBAAA,OAAO,EAAE,KAAK;AACd,oBAAA,KAAK,EAAE,8BAA8B;AACtC,iBAAA;aACF;;AAGH,QAAA,IAAI;YACF,MAAM,OAAO,GAAG,aAAa,CAAC;gBAC5B,GAAG;AACH,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,cAAc;AAC/C,gBAAA,WAAW,EAAE,OAAO,CAAC,WAAW,IAAI,IAAI,CAAC,WAAW;AACpD,gBAAA,WAAW,EAAE,OAAO,CAAC,WAAW,IAAI,IAAI,CAAC,WAAW;AACpD,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;AACxC,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;AACxC,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;AACxC,gBAAA,eAAe,EAAE,OAAO,CAAC,eAAe,IAAI,IAAI,CAAC,eAAe;AAChE,gBAAA,MAAM,EAAE,OAAO,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM;AACrC,gBAAA,MAAM,EAAE,OAAO,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM;AACrC,gBAAA,mBAAmB,EACjB,OAAO,CAAC,mBAAmB,IAAI,IAAI,CAAC,mBAAmB;AACzD,gBAAA,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ;AAC3C,gBAAA,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ;AAC3C,gBAAA,kBAAkB,EAChB,OAAO,CAAC,kBAAkB,IAAI,IAAI,CAAC,kBAAkB;AACvD,gBAAA,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ;AAC3C,gBAAA,YAAY,EAAE,OAAO,CAAC,YAAY,IAAI,IAAI,CAAC,YAAY;AACvD,gBAAA,iBAAiB,EAAE,OAAO,CAAC,iBAAiB,IAAI,IAAI,CAAC,iBAAiB;AACtE,gBAAA,qBAAqB,EACnB,OAAO,CAAC,qBAAqB,IAAI,IAAI,CAAC,qBAAqB;AAC9D,aAAA,CAAC;AACF,YAAA,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,OAAO,EAAE;AACtD,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;gBACD,OAAO,EAAE,IAAI,CAAC,OAAO;AACtB,aAAA,CAAC;AAEF,YAAA,OAAO,CAAC,GAAG,EAAE,QAAQ,CAAC,IAAI,CAAC;;QAC3B,OAAO,KAAK,EAAE;AACd,YAAA,MAAM,YAAY,GAChB,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YACxD,OAAO;gBACL,GAAG;AACH,gBAAA;AACE,oBAAA,OAAO,EAAE,KAAK;oBACd,KAAK,EAAE,CAAiC,8BAAA,EAAA,YAAY,CAAE,CAAA;AACvD,iBAAA;aACF;;;AAIL;;;;AAIG;AACH,IAAA,cAAc,CACZ,QAAmC,EAAA;QAEnC,IAAI,CAAC,QAAQ,CAAC,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE;AACvC,YAAA,OAAO,CAAC,EAAE,EAAE,SAAS,CAAC;;AAGxB,QAAA,IAAI,QAAQ,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,EAAE;AAChE,YAAA,IAAI;gBACF,MAAM,EAAE,QAAQ,EAAE,GAAG,IAAI,EAAE,GAAG,cAAc,CAC1C,QAAQ,CAAC,IAAI,CAAC,IAAI,EAClB,QAAQ,CAAC,IAAI,CAAC,QAAQ,CACvB;AACD,gBAAA,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC;;YACvB,OAAO,KAAK,EAAE;gBACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,2BAA2B,EAAE,KAAK,CAAC;gBACrD,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,SAAS,CAAC;;;aAEvC,IAAI,QAAQ,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,EAAE;YACzC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,SAAS,CAAC;;;QAI5C,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,EAAE;YAC9B,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,SAAS,CAAC;;;QAIxC,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,IAAI,IAAI,EAAE;YACjC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,SAAS,CAAC;;AAG3C,QAAA,OAAO,CAAC,EAAE,EAAE,SAAS,CAAC;;AAGxB;;;;AAIG;AACH,IAAA,eAAe,CAAC,QAAmC,EAAA;AACjD,QAAA,IAAI,CAAC,QAAQ,CAAC,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE;AAClE,YAAA,OAAO,EAAE;;AAGX,QAAA,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ;;AAEhC;AAED;;;;AAIG;MACU,sBAAsB,GAAG,CACpC,MAAmC,GAAA,EAAE,KACjB;AACpB,IAAA,OAAO,IAAI,gBAAgB,CAAC,MAAM,CAAC;AACrC;AAEA;AACA,SAAS,aAAa,CAAmB,GAAM,EAAA;IAC7C,OAAO,MAAM,CAAC,WAAW,CACvB,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,CAAC,KAAK,SAAS,CAAC,CACzC;AACjB;;;;"}
|
|
1
|
+
{"version":3,"file":"firecrawl.mjs","sources":["../../../../src/tools/search/firecrawl.ts"],"sourcesContent":["import axios from 'axios';\nimport { processContent } from './content';\nimport type * as t from './types';\nimport { createDefaultLogger } from './utils';\n\n/**\n * Firecrawl scraper implementation\n * Uses the Firecrawl API to scrape web pages\n */\nexport class FirecrawlScraper implements t.BaseScraper {\n private apiKey: string;\n private apiUrl: string;\n private version: string;\n private defaultFormats: string[];\n private timeout: number;\n private logger: t.Logger;\n private includeTags?: string[];\n private excludeTags?: string[];\n private waitFor?: number;\n private maxAge?: number;\n private mobile?: boolean;\n private skipTlsVerification?: boolean;\n private blockAds?: boolean;\n private removeBase64Images?: boolean;\n private parsePDF?: boolean;\n private storeInCache?: boolean;\n private zeroDataRetention?: boolean;\n private headers?: Record<string, string>;\n private location?: { country?: string; languages?: string[] };\n private onlyMainContent?: boolean;\n private changeTrackingOptions?: object;\n\n constructor(config: t.FirecrawlScraperConfig = {}) {\n this.apiKey = config.apiKey ?? process.env.FIRECRAWL_API_KEY ?? '';\n\n this.version = config.version ?? 'v2';\n\n const baseUrl =\n config.apiUrl ??\n process.env.FIRECRAWL_BASE_URL ??\n 'https://api.firecrawl.dev';\n this.apiUrl = `${baseUrl.replace(/\\/+$/, '')}/${this.version}/scrape`;\n\n this.defaultFormats = config.formats ?? ['markdown', 'rawHtml'];\n this.timeout = config.timeout ?? 7500;\n\n this.logger = config.logger || createDefaultLogger();\n\n this.includeTags = config.includeTags;\n this.excludeTags = config.excludeTags;\n this.waitFor = config.waitFor;\n this.maxAge = config.maxAge;\n this.mobile = config.mobile;\n this.skipTlsVerification = config.skipTlsVerification;\n this.blockAds = config.blockAds;\n this.removeBase64Images = config.removeBase64Images;\n this.parsePDF = config.parsePDF;\n this.storeInCache = config.storeInCache;\n this.zeroDataRetention = config.zeroDataRetention;\n this.headers = config.headers;\n this.location = config.location;\n this.onlyMainContent = config.onlyMainContent;\n this.changeTrackingOptions = config.changeTrackingOptions;\n\n if (!this.apiKey) {\n this.logger.warn('FIRECRAWL_API_KEY is not set. Scraping will not work.');\n }\n\n this.logger.debug(\n `Firecrawl scraper initialized with API URL: ${this.apiUrl}`\n );\n }\n\n /**\n * Scrape a single URL\n * @param url URL to scrape\n * @param options Scrape options\n * @returns Scrape response\n */\n async scrapeUrl(\n url: string,\n options: t.FirecrawlScrapeOptions = {}\n ): Promise<[string, t.FirecrawlScrapeResponse]> {\n if (!this.apiKey) {\n return [\n url,\n {\n success: false,\n error: 'FIRECRAWL_API_KEY is not set',\n },\n ];\n }\n\n try {\n const payload = omitUndefined({\n url,\n formats: options.formats ?? this.defaultFormats,\n includeTags: options.includeTags ?? this.includeTags,\n excludeTags: options.excludeTags ?? this.excludeTags,\n headers: options.headers ?? this.headers,\n waitFor: options.waitFor ?? this.waitFor,\n timeout: options.timeout ?? this.timeout,\n onlyMainContent: options.onlyMainContent ?? this.onlyMainContent,\n maxAge: options.maxAge ?? this.maxAge,\n mobile: options.mobile ?? this.mobile,\n skipTlsVerification:\n options.skipTlsVerification ?? this.skipTlsVerification,\n parsePDF: options.parsePDF ?? this.parsePDF,\n location: options.location ?? this.location,\n removeBase64Images:\n options.removeBase64Images ?? this.removeBase64Images,\n blockAds: options.blockAds ?? this.blockAds,\n storeInCache: options.storeInCache ?? this.storeInCache,\n zeroDataRetention: options.zeroDataRetention ?? this.zeroDataRetention,\n changeTrackingOptions:\n options.changeTrackingOptions ?? this.changeTrackingOptions,\n });\n const response = await axios.post(this.apiUrl, payload, {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n timeout: this.timeout,\n });\n\n return [url, response.data];\n } catch (error) {\n const errorMessage =\n error instanceof Error ? error.message : String(error);\n return [\n url,\n {\n success: false,\n error: `Firecrawl API request failed: ${errorMessage}`,\n },\n ];\n }\n }\n\n /**\n * Extract content from scrape response\n * @param response Scrape response\n * @returns Extracted content or empty string if not available\n */\n extractContent(\n response: t.FirecrawlScrapeResponse\n ): [string, undefined | t.References] {\n if (!response.success || !response.data) {\n return ['', undefined];\n }\n\n if (response.data.markdown != null && response.data.html != null) {\n try {\n const { markdown, ...rest } = processContent(\n response.data.html,\n response.data.markdown\n );\n return [markdown, rest];\n } catch (error) {\n this.logger.error('Error processing content:', error);\n return [response.data.markdown, undefined];\n }\n } else if (response.data.markdown != null) {\n return [response.data.markdown, undefined];\n }\n\n // Fall back to HTML content\n if (response.data.html != null) {\n return [response.data.html, undefined];\n }\n\n // Fall back to raw HTML content\n if (response.data.rawHtml != null) {\n return [response.data.rawHtml, undefined];\n }\n\n return ['', undefined];\n }\n\n /**\n * Extract metadata from scrape response\n * @param response Scrape response\n * @returns Metadata object\n */\n extractMetadata(response: t.FirecrawlScrapeResponse): t.ScrapeMetadata {\n if (!response.success || !response.data || !response.data.metadata) {\n return {};\n }\n\n return response.data.metadata;\n }\n}\n\n/**\n * Create a Firecrawl scraper instance\n * @param config Scraper configuration\n * @returns Firecrawl scraper instance\n */\nexport const createFirecrawlScraper = (\n config: t.FirecrawlScraperConfig = {}\n): FirecrawlScraper => {\n return new FirecrawlScraper(config);\n};\n\n// Helper function to clean up payload for firecrawl\nfunction omitUndefined<T extends object>(obj: T): Partial<T> {\n return Object.fromEntries(\n Object.entries(obj).filter(([, v]) => v !== undefined)\n ) as Partial<T>;\n}\n"],"names":[],"mappings":";;;;AAKA;;;AAGG;MACU,gBAAgB,CAAA;AACnB,IAAA,MAAM;AACN,IAAA,MAAM;AACN,IAAA,OAAO;AACP,IAAA,cAAc;AACd,IAAA,OAAO;AACP,IAAA,MAAM;AACN,IAAA,WAAW;AACX,IAAA,WAAW;AACX,IAAA,OAAO;AACP,IAAA,MAAM;AACN,IAAA,MAAM;AACN,IAAA,mBAAmB;AACnB,IAAA,QAAQ;AACR,IAAA,kBAAkB;AAClB,IAAA,QAAQ;AACR,IAAA,YAAY;AACZ,IAAA,iBAAiB;AACjB,IAAA,OAAO;AACP,IAAA,QAAQ;AACR,IAAA,eAAe;AACf,IAAA,qBAAqB;AAE7B,IAAA,WAAA,CAAY,SAAmC,EAAE,EAAA;AAC/C,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,iBAAiB,IAAI,EAAE;QAElE,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,IAAI;AAErC,QAAA,MAAM,OAAO,GACX,MAAM,CAAC,MAAM;YACb,OAAO,CAAC,GAAG,CAAC,kBAAkB;AAC9B,YAAA,2BAA2B;AAC7B,QAAA,IAAI,CAAC,MAAM,GAAG,CAAG,EAAA,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAA,CAAA,EAAI,IAAI,CAAC,OAAO,SAAS;AAErE,QAAA,IAAI,CAAC,cAAc,GAAG,MAAM,CAAC,OAAO,IAAI,CAAC,UAAU,EAAE,SAAS,CAAC;QAC/D,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,IAAI;QAErC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,mBAAmB,EAAE;AAEpD,QAAA,IAAI,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW;AACrC,QAAA,IAAI,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW;AACrC,QAAA,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO;AAC7B,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM;AAC3B,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM;AAC3B,QAAA,IAAI,CAAC,mBAAmB,GAAG,MAAM,CAAC,mBAAmB;AACrD,QAAA,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ;AAC/B,QAAA,IAAI,CAAC,kBAAkB,GAAG,MAAM,CAAC,kBAAkB;AACnD,QAAA,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ;AAC/B,QAAA,IAAI,CAAC,YAAY,GAAG,MAAM,CAAC,YAAY;AACvC,QAAA,IAAI,CAAC,iBAAiB,GAAG,MAAM,CAAC,iBAAiB;AACjD,QAAA,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO;AAC7B,QAAA,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ;AAC/B,QAAA,IAAI,CAAC,eAAe,GAAG,MAAM,CAAC,eAAe;AAC7C,QAAA,IAAI,CAAC,qBAAqB,GAAG,MAAM,CAAC,qBAAqB;AAEzD,QAAA,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;AAChB,YAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,uDAAuD,CAAC;;QAG3E,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,CAA+C,4CAAA,EAAA,IAAI,CAAC,MAAM,CAAE,CAAA,CAC7D;;AAGH;;;;;AAKG;AACH,IAAA,MAAM,SAAS,CACb,GAAW,EACX,UAAoC,EAAE,EAAA;AAEtC,QAAA,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;YAChB,OAAO;gBACL,GAAG;AACH,gBAAA;AACE,oBAAA,OAAO,EAAE,KAAK;AACd,oBAAA,KAAK,EAAE,8BAA8B;AACtC,iBAAA;aACF;;AAGH,QAAA,IAAI;YACF,MAAM,OAAO,GAAG,aAAa,CAAC;gBAC5B,GAAG;AACH,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,cAAc;AAC/C,gBAAA,WAAW,EAAE,OAAO,CAAC,WAAW,IAAI,IAAI,CAAC,WAAW;AACpD,gBAAA,WAAW,EAAE,OAAO,CAAC,WAAW,IAAI,IAAI,CAAC,WAAW;AACpD,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;AACxC,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;AACxC,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;AACxC,gBAAA,eAAe,EAAE,OAAO,CAAC,eAAe,IAAI,IAAI,CAAC,eAAe;AAChE,gBAAA,MAAM,EAAE,OAAO,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM;AACrC,gBAAA,MAAM,EAAE,OAAO,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM;AACrC,gBAAA,mBAAmB,EACjB,OAAO,CAAC,mBAAmB,IAAI,IAAI,CAAC,mBAAmB;AACzD,gBAAA,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ;AAC3C,gBAAA,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ;AAC3C,gBAAA,kBAAkB,EAChB,OAAO,CAAC,kBAAkB,IAAI,IAAI,CAAC,kBAAkB;AACvD,gBAAA,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ;AAC3C,gBAAA,YAAY,EAAE,OAAO,CAAC,YAAY,IAAI,IAAI,CAAC,YAAY;AACvD,gBAAA,iBAAiB,EAAE,OAAO,CAAC,iBAAiB,IAAI,IAAI,CAAC,iBAAiB;AACtE,gBAAA,qBAAqB,EACnB,OAAO,CAAC,qBAAqB,IAAI,IAAI,CAAC,qBAAqB;AAC9D,aAAA,CAAC;AACF,YAAA,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,OAAO,EAAE;AACtD,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;gBACD,OAAO,EAAE,IAAI,CAAC,OAAO;AACtB,aAAA,CAAC;AAEF,YAAA,OAAO,CAAC,GAAG,EAAE,QAAQ,CAAC,IAAI,CAAC;;QAC3B,OAAO,KAAK,EAAE;AACd,YAAA,MAAM,YAAY,GAChB,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YACxD,OAAO;gBACL,GAAG;AACH,gBAAA;AACE,oBAAA,OAAO,EAAE,KAAK;oBACd,KAAK,EAAE,CAAiC,8BAAA,EAAA,YAAY,CAAE,CAAA;AACvD,iBAAA;aACF;;;AAIL;;;;AAIG;AACH,IAAA,cAAc,CACZ,QAAmC,EAAA;QAEnC,IAAI,CAAC,QAAQ,CAAC,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE;AACvC,YAAA,OAAO,CAAC,EAAE,EAAE,SAAS,CAAC;;AAGxB,QAAA,IAAI,QAAQ,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,EAAE;AAChE,YAAA,IAAI;gBACF,MAAM,EAAE,QAAQ,EAAE,GAAG,IAAI,EAAE,GAAG,cAAc,CAC1C,QAAQ,CAAC,IAAI,CAAC,IAAI,EAClB,QAAQ,CAAC,IAAI,CAAC,QAAQ,CACvB;AACD,gBAAA,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC;;YACvB,OAAO,KAAK,EAAE;gBACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,2BAA2B,EAAE,KAAK,CAAC;gBACrD,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,SAAS,CAAC;;;aAEvC,IAAI,QAAQ,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,EAAE;YACzC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,SAAS,CAAC;;;QAI5C,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,EAAE;YAC9B,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,SAAS,CAAC;;;QAIxC,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,IAAI,IAAI,EAAE;YACjC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,SAAS,CAAC;;AAG3C,QAAA,OAAO,CAAC,EAAE,EAAE,SAAS,CAAC;;AAGxB;;;;AAIG;AACH,IAAA,eAAe,CAAC,QAAmC,EAAA;AACjD,QAAA,IAAI,CAAC,QAAQ,CAAC,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE;AAClE,YAAA,OAAO,EAAE;;AAGX,QAAA,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ;;AAEhC;AAED;;;;AAIG;MACU,sBAAsB,GAAG,CACpC,MAAmC,GAAA,EAAE,KACjB;AACpB,IAAA,OAAO,IAAI,gBAAgB,CAAC,MAAM,CAAC;AACrC;AAEA;AACA,SAAS,aAAa,CAAmB,GAAM,EAAA;IAC7C,OAAO,MAAM,CAAC,WAAW,CACvB,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,CAAC,KAAK,SAAS,CAAC,CACzC;AACjB;;;;"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"format.mjs","sources":["../../../../src/tools/search/format.ts"],"sourcesContent":["import type * as t from './types';\r\nimport { getDomainName, fileExtRegex } from './utils';\r\n\r\nfunction addHighlightSection(): string[] {\r\n return ['\\n## Highlights', ''];\r\n}\r\n\r\n// Helper function to format a source (organic or top story)\r\nfunction formatSource(\r\n source: t.ValidSource,\r\n index: number,\r\n turn: number,\r\n sourceType: 'search' | 'news',\r\n references: t.ResultReference[]\r\n): string {\r\n /** Array of all lines to include in the output */\r\n const outputLines: string[] = [];\r\n\r\n // Add the title\r\n outputLines.push(\r\n `# ${sourceType.charAt(0).toUpperCase() + sourceType.slice(1)} ${index}: ${source.title != null && source.title ? `\"${source.title}\"` : '(no title)'}`\r\n );\r\n outputLines.push(`\\nAnchor: \\\\ue202turn${turn}${sourceType}${index}`);\r\n outputLines.push(`URL: ${source.link}`);\r\n\r\n // Add optional fields\r\n if ('snippet' in source && source.snippet != null) {\r\n outputLines.push(`Summary: ${source.snippet}`);\r\n }\r\n\r\n if (source.date != null) {\r\n outputLines.push(`Date: ${source.date}`);\r\n }\r\n\r\n if (source.attribution != null) {\r\n outputLines.push(`Source: ${source.attribution}`);\r\n }\r\n\r\n // Add highlight section or empty line\r\n if ((source.highlights?.length ?? 0) > 0) {\r\n outputLines.push(...addHighlightSection());\r\n } else {\r\n outputLines.push('');\r\n }\r\n\r\n // Process highlights if they exist\r\n (source.highlights ?? [])\r\n .filter((h) => h.text.trim().length > 0)\r\n .forEach((h, hIndex) => {\r\n outputLines.push(\r\n `### Highlight ${hIndex + 1} [Relevance: ${h.score.toFixed(2)}]`\r\n );\r\n outputLines.push('');\r\n outputLines.push('```text');\r\n outputLines.push(h.text.trim());\r\n outputLines.push('```');\r\n outputLines.push('');\r\n\r\n if (h.references != null && h.references.length) {\r\n let hasHeader = false;\r\n const refLines: string[] = [];\r\n\r\n for (let j = 0; j < h.references.length; j++) {\r\n const ref = h.references[j];\r\n if (ref.reference.originalUrl.includes('mailto:')) {\r\n continue;\r\n }\r\n if (ref.type !== 'link') {\r\n continue;\r\n }\r\n if (fileExtRegex.test(ref.reference.originalUrl)) {\r\n continue;\r\n }\r\n references.push({\r\n type: ref.type,\r\n link: ref.reference.originalUrl,\r\n attribution: getDomainName(ref.reference.originalUrl),\r\n title: (\r\n ((ref.reference.title ?? '') || ref.reference.text) ??\r\n ''\r\n ).split('\\n')[0],\r\n });\r\n\r\n if (!hasHeader) {\r\n refLines.push('Core References:');\r\n hasHeader = true;\r\n }\r\n\r\n refLines.push(\r\n `- ${ref.type}#${ref.originalIndex + 1}: ${ref.reference.originalUrl}`\r\n );\r\n refLines.push(\r\n `\\t- Anchor: \\\\ue202turn${turn}ref${references.length - 1}`\r\n );\r\n }\r\n\r\n if (hasHeader) {\r\n outputLines.push(...refLines);\r\n outputLines.push('');\r\n }\r\n }\r\n\r\n if (hIndex < (source.highlights?.length ?? 0) - 1) {\r\n outputLines.push('---');\r\n outputLines.push('');\r\n }\r\n });\r\n\r\n outputLines.push('');\r\n return outputLines.join('\\n');\r\n}\r\n\r\nexport function formatResultsForLLM(\r\n turn: number,\r\n results: t.SearchResultData\r\n): { output: string; references: t.ResultReference[] } {\r\n /** Array to collect all output lines */\r\n const outputLines: string[] = [];\r\n\r\n const addSection = (title: string): void => {\r\n outputLines.push('');\r\n outputLines.push(`=== ${title} ===`);\r\n outputLines.push('');\r\n };\r\n\r\n const references: t.ResultReference[] = [];\r\n\r\n // Organic (web) results\r\n if (results.organic?.length != null && results.organic.length > 0) {\r\n addSection(`Web Results, Turn ${turn}`);\r\n for (let i = 0; i < results.organic.length; i++) {\r\n const r = results.organic[i];\r\n outputLines.push(formatSource(r, i, turn, 'search', references));\r\n delete results.organic[i].highlights;\r\n }\r\n }\r\n\r\n // Top stories (news)\r\n const topStories = results.topStories ?? [];\r\n if (topStories.length) {\r\n addSection('News Results');\r\n for (let i = 0; i < topStories.length; i++) {\r\n const r = topStories[i];\r\n outputLines.push(formatSource(r, i, turn, 'news', references));\r\n if (results.topStories?.[i]?.highlights) {\r\n delete results.topStories[i].highlights;\r\n }\r\n }\r\n }\r\n\r\n // // Images\r\n // const images = results.images ?? [];\r\n // if (images.length) {\r\n // addSection('Image Results');\r\n // const imageLines = images.map((img, i) => [\r\n // `Anchor: \\ue202turn0image${i}`,\r\n // `Title: ${img.title ?? '(no title)'}`,\r\n // `Image URL: ${img.imageUrl}`,\r\n // ''\r\n // ].join('\\n'));\r\n // outputLines.push(imageLines.join('\\n'));\r\n // }\r\n\r\n // Knowledge Graph\r\n if (results.knowledgeGraph != null) {\r\n addSection('Knowledge Graph');\r\n const kgLines = [\r\n `**Title:** ${results.knowledgeGraph.title ?? '(no title)'}`,\r\n results.knowledgeGraph.type != null\r\n ? `**Type:** ${results.knowledgeGraph.type}`\r\n : '',\r\n results.knowledgeGraph.description != null\r\n ? `**Description:** ${results.knowledgeGraph.description}`\r\n : '',\r\n results.knowledgeGraph.descriptionSource != null\r\n ? `**Description Source:** ${results.knowledgeGraph.descriptionSource}`\r\n : '',\r\n results.knowledgeGraph.descriptionLink != null\r\n ? `**Description Link:** ${results.knowledgeGraph.descriptionLink}`\r\n : '',\r\n results.knowledgeGraph.imageUrl != null\r\n ? `**Image URL:** ${results.knowledgeGraph.imageUrl}`\r\n : '',\r\n results.knowledgeGraph.website != null\r\n ? `**Website:** ${results.knowledgeGraph.website}`\r\n : '',\r\n results.knowledgeGraph.attributes != null\r\n ? `**Attributes:**\\n\\`\\`\\`json\\n${JSON.stringify(\r\n results.knowledgeGraph.attributes,\r\n null,\r\n 2\r\n )}\\n\\`\\`\\``\r\n : '',\r\n '',\r\n ].filter(Boolean);\r\n\r\n outputLines.push(kgLines.join('\\n\\n'));\r\n }\r\n\r\n // Answer Box\r\n if (results.answerBox != null) {\r\n addSection('Answer Box');\r\n const abLines = [\r\n results.answerBox.title != null\r\n ? `**Title:** ${results.answerBox.title}`\r\n : '',\r\n results.answerBox.snippet != null\r\n ? `**Snippet:** ${results.answerBox.snippet}`\r\n : '',\r\n results.answerBox.snippetHighlighted != null\r\n ? `**Snippet Highlighted:** ${results.answerBox.snippetHighlighted\r\n .map((s) => `\\`${s}\\``)\r\n .join(' ')}`\r\n : '',\r\n results.answerBox.link != null\r\n ? `**Link:** ${results.answerBox.link}`\r\n : '',\r\n '',\r\n ].filter(Boolean);\r\n\r\n outputLines.push(abLines.join('\\n\\n'));\r\n }\r\n\r\n // People also ask\r\n const peopleAlsoAsk = results.peopleAlsoAsk ?? [];\r\n if (peopleAlsoAsk.length) {\r\n addSection('People Also Ask');\r\n\r\n const paaLines: string[] = [];\r\n peopleAlsoAsk.forEach((p, i) => {\r\n const questionLines = [\r\n `### Question ${i + 1}:`,\r\n `\"${p.question}\"`,\r\n `${p.snippet != null && p.snippet ? `Snippet: ${p.snippet}` : ''}`,\r\n `${p.title != null && p.title ? `Title: ${p.title}` : ''}`,\r\n `${p.link != null && p.link ? `Link: ${p.link}` : ''}`,\r\n '',\r\n ].filter(Boolean);\r\n\r\n paaLines.push(questionLines.join('\\n\\n'));\r\n });\r\n\r\n outputLines.push(paaLines.join(''));\r\n }\r\n\r\n return {\r\n output: outputLines.join('\\n').trim(),\r\n references,\r\n };\r\n}\r\n"],"names":[],"mappings":";;AAGA,SAAS,mBAAmB,GAAA;AAC1B,IAAA,OAAO,CAAC,iBAAiB,EAAE,EAAE,CAAC;AAChC;AAEA;AACA,SAAS,YAAY,CACnB,MAAqB,EACrB,KAAa,EACb,IAAY,EACZ,UAA6B,EAC7B,UAA+B,EAAA;;IAG/B,MAAM,WAAW,GAAa,EAAE;;IAGhC,WAAW,CAAC,IAAI,CACd,CAAA,EAAA,EAAK,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA,CAAA,EAAI,KAAK,CAAA,EAAA,EAAK,MAAM,CAAC,KAAK,IAAI,IAAI,IAAI,MAAM,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,KAAK,CAAG,CAAA,CAAA,GAAG,YAAY,CAAE,CAAA,CACvJ;IACD,WAAW,CAAC,IAAI,CAAC,CAAwB,qBAAA,EAAA,IAAI,CAAG,EAAA,UAAU,CAAG,EAAA,KAAK,CAAE,CAAA,CAAC;IACrE,WAAW,CAAC,IAAI,CAAC,CAAA,KAAA,EAAQ,MAAM,CAAC,IAAI,CAAE,CAAA,CAAC;;IAGvC,IAAI,SAAS,IAAI,MAAM,IAAI,MAAM,CAAC,OAAO,IAAI,IAAI,EAAE;QACjD,WAAW,CAAC,IAAI,CAAC,CAAA,SAAA,EAAY,MAAM,CAAC,OAAO,CAAE,CAAA,CAAC;;AAGhD,IAAA,IAAI,MAAM,CAAC,IAAI,IAAI,IAAI,EAAE;QACvB,WAAW,CAAC,IAAI,CAAC,CAAA,MAAA,EAAS,MAAM,CAAC,IAAI,CAAE,CAAA,CAAC;;AAG1C,IAAA,IAAI,MAAM,CAAC,WAAW,IAAI,IAAI,EAAE;QAC9B,WAAW,CAAC,IAAI,CAAC,CAAA,QAAA,EAAW,MAAM,CAAC,WAAW,CAAE,CAAA,CAAC;;;AAInD,IAAA,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;AACxC,QAAA,WAAW,CAAC,IAAI,CAAC,GAAG,mBAAmB,EAAE,CAAC;;SACrC;AACL,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;;AAItB,IAAA,CAAC,MAAM,CAAC,UAAU,IAAI,EAAE;AACrB,SAAA,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;AACtC,SAAA,OAAO,CAAC,CAAC,CAAC,EAAE,MAAM,KAAI;AACrB,QAAA,WAAW,CAAC,IAAI,CACd,iBAAiB,MAAM,GAAG,CAAC,CAAgB,aAAA,EAAA,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAA,CAAA,CAAG,CACjE;AACD,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,QAAA,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC;QAC3B,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;AAC/B,QAAA,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC;AACvB,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AAEpB,QAAA,IAAI,CAAC,CAAC,UAAU,IAAI,IAAI,IAAI,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE;YAC/C,IAAI,SAAS,GAAG,KAAK;YACrB,MAAM,QAAQ,GAAa,EAAE;AAE7B,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBAC5C,MAAM,GAAG,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC;gBAC3B,IAAI,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE;oBACjD;;AAEF,gBAAA,IAAI,GAAG,CAAC,IAAI,KAAK,MAAM,EAAE;oBACvB;;gBAEF,IAAI,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE;oBAChD;;gBAEF,UAAU,CAAC,IAAI,CAAC;oBACd,IAAI,EAAE,GAAG,CAAC,IAAI;AACd,oBAAA,IAAI,EAAE,GAAG,CAAC,SAAS,CAAC,WAAW;oBAC/B,WAAW,EAAE,aAAa,CAAC,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC;AACrD,oBAAA,KAAK,EAAE,CACL,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,IAAI,EAAE,KAAK,GAAG,CAAC,SAAS,CAAC,IAAI;wBAClD,EAAE,EACF,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AACjB,iBAAA,CAAC;gBAEF,IAAI,CAAC,SAAS,EAAE;AACd,oBAAA,QAAQ,CAAC,IAAI,CAAC,kBAAkB,CAAC;oBACjC,SAAS,GAAG,IAAI;;gBAGlB,QAAQ,CAAC,IAAI,CACX,CAAA,EAAA,EAAK,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,aAAa,GAAG,CAAC,KAAK,GAAG,CAAC,SAAS,CAAC,WAAW,CAAE,CAAA,CACvE;AACD,gBAAA,QAAQ,CAAC,IAAI,CACX,CAAA,uBAAA,EAA0B,IAAI,CAAA,GAAA,EAAM,UAAU,CAAC,MAAM,GAAG,CAAC,CAAA,CAAE,CAC5D;;YAGH,IAAI,SAAS,EAAE;AACb,gBAAA,WAAW,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC;AAC7B,gBAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;;AAIxB,QAAA,IAAI,MAAM,GAAG,CAAC,MAAM,CAAC,UAAU,EAAE,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;AACjD,YAAA,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC;AACvB,YAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;AAExB,KAAC,CAAC;AAEJ,IAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,IAAA,OAAO,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC;AAC/B;AAEgB,SAAA,mBAAmB,CACjC,IAAY,EACZ,OAA2B,EAAA;;IAG3B,MAAM,WAAW,GAAa,EAAE;AAEhC,IAAA,MAAM,UAAU,GAAG,CAAC,KAAa,KAAU;AACzC,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,QAAA,WAAW,CAAC,IAAI,CAAC,OAAO,KAAK,CAAA,IAAA,CAAM,CAAC;AACpC,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACtB,KAAC;IAED,MAAM,UAAU,GAAwB,EAAE;;AAG1C,IAAA,IAAI,OAAO,CAAC,OAAO,EAAE,MAAM,IAAI,IAAI,IAAI,OAAO,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE;AACjE,QAAA,UAAU,CAAC,CAAA,kBAAA,EAAqB,IAAI,CAAA,CAAE,CAAC;AACvC,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAC/C,MAAM,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC;AAC5B,YAAA,WAAW,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,CAAC,CAAC;YAChE,OAAO,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,UAAU;;;;AAKxC,IAAA,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,IAAI,EAAE;AAC3C,IAAA,IAAI,UAAU,CAAC,MAAM,EAAE;QACrB,UAAU,CAAC,cAAc,CAAC;AAC1B,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AAC1C,YAAA,MAAM,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC;AACvB,YAAA,WAAW,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,UAAU,CAAC,CAAC;YAC9D,IAAI,OAAO,CAAC,UAAU,GAAG,CAAC,CAAC,EAAE,UAAU,EAAE;gBACvC,OAAO,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,UAAU;;;;;;;;;;;;;;;;;AAmB7C,IAAA,IAAI,OAAO,CAAC,cAAc,IAAI,IAAI,EAAE;QAClC,UAAU,CAAC,iBAAiB,CAAC;AAC7B,QAAA,MAAM,OAAO,GAAG;AACd,YAAA,CAAA,WAAA,EAAc,OAAO,CAAC,cAAc,CAAC,KAAK,IAAI,YAAY,CAAE,CAAA;AAC5D,YAAA,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI;AAC7B,kBAAE,CAAa,UAAA,EAAA,OAAO,CAAC,cAAc,CAAC,IAAI,CAAE;AAC5C,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,WAAW,IAAI;AACpC,kBAAE,CAAoB,iBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,WAAW,CAAE;AAC1D,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,iBAAiB,IAAI;AAC1C,kBAAE,CAA2B,wBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,iBAAiB,CAAE;AACvE,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,eAAe,IAAI;AACxC,kBAAE,CAAyB,sBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,eAAe,CAAE;AACnE,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,IAAI;AACjC,kBAAE,CAAkB,eAAA,EAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,CAAE;AACrD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,OAAO,IAAI;AAChC,kBAAE,CAAgB,aAAA,EAAA,OAAO,CAAC,cAAc,CAAC,OAAO,CAAE;AAClD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,UAAU,IAAI;AACnC,kBAAE,CAAgC,6BAAA,EAAA,IAAI,CAAC,SAAS,CAC9C,OAAO,CAAC,cAAc,CAAC,UAAU,EACjC,IAAI,EACJ,CAAC,CACF,CAAU,QAAA;AACX,kBAAE,EAAE;YACN,EAAE;AACH,SAAA,CAAC,MAAM,CAAC,OAAO,CAAC;QAEjB,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;;;AAIxC,IAAA,IAAI,OAAO,CAAC,SAAS,IAAI,IAAI,EAAE;QAC7B,UAAU,CAAC,YAAY,CAAC;AACxB,QAAA,MAAM,OAAO,GAAG;AACd,YAAA,OAAO,CAAC,SAAS,CAAC,KAAK,IAAI;AACzB,kBAAE,CAAc,WAAA,EAAA,OAAO,CAAC,SAAS,CAAC,KAAK,CAAE;AACzC,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,OAAO,IAAI;AAC3B,kBAAE,CAAgB,aAAA,EAAA,OAAO,CAAC,SAAS,CAAC,OAAO,CAAE;AAC7C,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,kBAAkB,IAAI;AACtC,kBAAE,CAA4B,yBAAA,EAAA,OAAO,CAAC,SAAS,CAAC;qBAC7C,GAAG,CAAC,CAAC,CAAC,KAAK,CAAA,EAAA,EAAK,CAAC,CAAA,EAAA,CAAI;qBACrB,IAAI,CAAC,GAAG,CAAC,CAAE;AACd,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,IAAI,IAAI;AACxB,kBAAE,CAAa,UAAA,EAAA,OAAO,CAAC,SAAS,CAAC,IAAI,CAAE;AACvC,kBAAE,EAAE;YACN,EAAE;AACH,SAAA,CAAC,MAAM,CAAC,OAAO,CAAC;QAEjB,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;;;AAIxC,IAAA,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,EAAE;AACjD,IAAA,IAAI,aAAa,CAAC,MAAM,EAAE;QACxB,UAAU,CAAC,iBAAiB,CAAC;QAE7B,MAAM,QAAQ,GAAa,EAAE;QAC7B,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,KAAI;AAC7B,YAAA,MAAM,aAAa,GAAG;gBACpB,CAAgB,aAAA,EAAA,CAAC,GAAG,CAAC,CAAG,CAAA,CAAA;gBACxB,CAAI,CAAA,EAAA,CAAC,CAAC,QAAQ,CAAG,CAAA,CAAA;gBACjB,CAAG,EAAA,CAAC,CAAC,OAAO,IAAI,IAAI,IAAI,CAAC,CAAC,OAAO,GAAG,CAAA,SAAA,EAAY,CAAC,CAAC,OAAO,EAAE,GAAG,EAAE,CAAE,CAAA;gBAClE,CAAG,EAAA,CAAC,CAAC,KAAK,IAAI,IAAI,IAAI,CAAC,CAAC,KAAK,GAAG,CAAA,OAAA,EAAU,CAAC,CAAC,KAAK,EAAE,GAAG,EAAE,CAAE,CAAA;gBAC1D,CAAG,EAAA,CAAC,CAAC,IAAI,IAAI,IAAI,IAAI,CAAC,CAAC,IAAI,GAAG,CAAA,MAAA,EAAS,CAAC,CAAC,IAAI,EAAE,GAAG,EAAE,CAAE,CAAA;gBACtD,EAAE;AACH,aAAA,CAAC,MAAM,CAAC,OAAO,CAAC;YAEjB,QAAQ,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AAC3C,SAAC,CAAC;QAEF,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;;IAGrC,OAAO;QACL,MAAM,EAAE,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE;QACrC,UAAU;KACX;AACH;;;;"}
|
|
1
|
+
{"version":3,"file":"format.mjs","sources":["../../../../src/tools/search/format.ts"],"sourcesContent":["import type * as t from './types';\nimport { getDomainName, fileExtRegex } from './utils';\n\nfunction addHighlightSection(): string[] {\n return ['\\n## Highlights', ''];\n}\n\n// Helper function to format a source (organic or top story)\nfunction formatSource(\n source: t.ValidSource,\n index: number,\n turn: number,\n sourceType: 'search' | 'news',\n references: t.ResultReference[]\n): string {\n /** Array of all lines to include in the output */\n const outputLines: string[] = [];\n\n // Add the title\n outputLines.push(\n `# ${sourceType.charAt(0).toUpperCase() + sourceType.slice(1)} ${index}: ${source.title != null && source.title ? `\"${source.title}\"` : '(no title)'}`\n );\n outputLines.push(`\\nAnchor: \\\\ue202turn${turn}${sourceType}${index}`);\n outputLines.push(`URL: ${source.link}`);\n\n // Add optional fields\n if ('snippet' in source && source.snippet != null) {\n outputLines.push(`Summary: ${source.snippet}`);\n }\n\n if (source.date != null) {\n outputLines.push(`Date: ${source.date}`);\n }\n\n if (source.attribution != null) {\n outputLines.push(`Source: ${source.attribution}`);\n }\n\n // Add highlight section or empty line\n if ((source.highlights?.length ?? 0) > 0) {\n outputLines.push(...addHighlightSection());\n } else {\n outputLines.push('');\n }\n\n // Process highlights if they exist\n (source.highlights ?? [])\n .filter((h) => h.text.trim().length > 0)\n .forEach((h, hIndex) => {\n outputLines.push(\n `### Highlight ${hIndex + 1} [Relevance: ${h.score.toFixed(2)}]`\n );\n outputLines.push('');\n outputLines.push('```text');\n outputLines.push(h.text.trim());\n outputLines.push('```');\n outputLines.push('');\n\n if (h.references != null && h.references.length) {\n let hasHeader = false;\n const refLines: string[] = [];\n\n for (let j = 0; j < h.references.length; j++) {\n const ref = h.references[j];\n if (ref.reference.originalUrl.includes('mailto:')) {\n continue;\n }\n if (ref.type !== 'link') {\n continue;\n }\n if (fileExtRegex.test(ref.reference.originalUrl)) {\n continue;\n }\n references.push({\n type: ref.type,\n link: ref.reference.originalUrl,\n attribution: getDomainName(ref.reference.originalUrl),\n title: (\n ((ref.reference.title ?? '') || ref.reference.text) ??\n ''\n ).split('\\n')[0],\n });\n\n if (!hasHeader) {\n refLines.push('Core References:');\n hasHeader = true;\n }\n\n refLines.push(\n `- ${ref.type}#${ref.originalIndex + 1}: ${ref.reference.originalUrl}`\n );\n refLines.push(\n `\\t- Anchor: \\\\ue202turn${turn}ref${references.length - 1}`\n );\n }\n\n if (hasHeader) {\n outputLines.push(...refLines);\n outputLines.push('');\n }\n }\n\n if (hIndex < (source.highlights?.length ?? 0) - 1) {\n outputLines.push('---');\n outputLines.push('');\n }\n });\n\n outputLines.push('');\n return outputLines.join('\\n');\n}\n\nexport function formatResultsForLLM(\n turn: number,\n results: t.SearchResultData\n): { output: string; references: t.ResultReference[] } {\n /** Array to collect all output lines */\n const outputLines: string[] = [];\n\n const addSection = (title: string): void => {\n outputLines.push('');\n outputLines.push(`=== ${title} ===`);\n outputLines.push('');\n };\n\n const references: t.ResultReference[] = [];\n\n // Organic (web) results\n if (results.organic?.length != null && results.organic.length > 0) {\n addSection(`Web Results, Turn ${turn}`);\n for (let i = 0; i < results.organic.length; i++) {\n const r = results.organic[i];\n outputLines.push(formatSource(r, i, turn, 'search', references));\n delete results.organic[i].highlights;\n }\n }\n\n // Top stories (news)\n const topStories = results.topStories ?? [];\n if (topStories.length) {\n addSection('News Results');\n for (let i = 0; i < topStories.length; i++) {\n const r = topStories[i];\n outputLines.push(formatSource(r, i, turn, 'news', references));\n if (results.topStories?.[i]?.highlights) {\n delete results.topStories[i].highlights;\n }\n }\n }\n\n // // Images\n // const images = results.images ?? [];\n // if (images.length) {\n // addSection('Image Results');\n // const imageLines = images.map((img, i) => [\n // `Anchor: \\ue202turn0image${i}`,\n // `Title: ${img.title ?? '(no title)'}`,\n // `Image URL: ${img.imageUrl}`,\n // ''\n // ].join('\\n'));\n // outputLines.push(imageLines.join('\\n'));\n // }\n\n // Knowledge Graph\n if (results.knowledgeGraph != null) {\n addSection('Knowledge Graph');\n const kgLines = [\n `**Title:** ${results.knowledgeGraph.title ?? '(no title)'}`,\n results.knowledgeGraph.type != null\n ? `**Type:** ${results.knowledgeGraph.type}`\n : '',\n results.knowledgeGraph.description != null\n ? `**Description:** ${results.knowledgeGraph.description}`\n : '',\n results.knowledgeGraph.descriptionSource != null\n ? `**Description Source:** ${results.knowledgeGraph.descriptionSource}`\n : '',\n results.knowledgeGraph.descriptionLink != null\n ? `**Description Link:** ${results.knowledgeGraph.descriptionLink}`\n : '',\n results.knowledgeGraph.imageUrl != null\n ? `**Image URL:** ${results.knowledgeGraph.imageUrl}`\n : '',\n results.knowledgeGraph.website != null\n ? `**Website:** ${results.knowledgeGraph.website}`\n : '',\n results.knowledgeGraph.attributes != null\n ? `**Attributes:**\\n\\`\\`\\`json\\n${JSON.stringify(\n results.knowledgeGraph.attributes,\n null,\n 2\n )}\\n\\`\\`\\``\n : '',\n '',\n ].filter(Boolean);\n\n outputLines.push(kgLines.join('\\n\\n'));\n }\n\n // Answer Box\n if (results.answerBox != null) {\n addSection('Answer Box');\n const abLines = [\n results.answerBox.title != null\n ? `**Title:** ${results.answerBox.title}`\n : '',\n results.answerBox.snippet != null\n ? `**Snippet:** ${results.answerBox.snippet}`\n : '',\n results.answerBox.snippetHighlighted != null\n ? `**Snippet Highlighted:** ${results.answerBox.snippetHighlighted\n .map((s) => `\\`${s}\\``)\n .join(' ')}`\n : '',\n results.answerBox.link != null\n ? `**Link:** ${results.answerBox.link}`\n : '',\n '',\n ].filter(Boolean);\n\n outputLines.push(abLines.join('\\n\\n'));\n }\n\n // People also ask\n const peopleAlsoAsk = results.peopleAlsoAsk ?? [];\n if (peopleAlsoAsk.length) {\n addSection('People Also Ask');\n\n const paaLines: string[] = [];\n peopleAlsoAsk.forEach((p, i) => {\n const questionLines = [\n `### Question ${i + 1}:`,\n `\"${p.question}\"`,\n `${p.snippet != null && p.snippet ? `Snippet: ${p.snippet}` : ''}`,\n `${p.title != null && p.title ? `Title: ${p.title}` : ''}`,\n `${p.link != null && p.link ? `Link: ${p.link}` : ''}`,\n '',\n ].filter(Boolean);\n\n paaLines.push(questionLines.join('\\n\\n'));\n });\n\n outputLines.push(paaLines.join(''));\n }\n\n return {\n output: outputLines.join('\\n').trim(),\n references,\n };\n}\n"],"names":[],"mappings":";;AAGA,SAAS,mBAAmB,GAAA;AAC1B,IAAA,OAAO,CAAC,iBAAiB,EAAE,EAAE,CAAC;AAChC;AAEA;AACA,SAAS,YAAY,CACnB,MAAqB,EACrB,KAAa,EACb,IAAY,EACZ,UAA6B,EAC7B,UAA+B,EAAA;;IAG/B,MAAM,WAAW,GAAa,EAAE;;IAGhC,WAAW,CAAC,IAAI,CACd,CAAA,EAAA,EAAK,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA,CAAA,EAAI,KAAK,CAAA,EAAA,EAAK,MAAM,CAAC,KAAK,IAAI,IAAI,IAAI,MAAM,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,KAAK,CAAG,CAAA,CAAA,GAAG,YAAY,CAAE,CAAA,CACvJ;IACD,WAAW,CAAC,IAAI,CAAC,CAAwB,qBAAA,EAAA,IAAI,CAAG,EAAA,UAAU,CAAG,EAAA,KAAK,CAAE,CAAA,CAAC;IACrE,WAAW,CAAC,IAAI,CAAC,CAAA,KAAA,EAAQ,MAAM,CAAC,IAAI,CAAE,CAAA,CAAC;;IAGvC,IAAI,SAAS,IAAI,MAAM,IAAI,MAAM,CAAC,OAAO,IAAI,IAAI,EAAE;QACjD,WAAW,CAAC,IAAI,CAAC,CAAA,SAAA,EAAY,MAAM,CAAC,OAAO,CAAE,CAAA,CAAC;;AAGhD,IAAA,IAAI,MAAM,CAAC,IAAI,IAAI,IAAI,EAAE;QACvB,WAAW,CAAC,IAAI,CAAC,CAAA,MAAA,EAAS,MAAM,CAAC,IAAI,CAAE,CAAA,CAAC;;AAG1C,IAAA,IAAI,MAAM,CAAC,WAAW,IAAI,IAAI,EAAE;QAC9B,WAAW,CAAC,IAAI,CAAC,CAAA,QAAA,EAAW,MAAM,CAAC,WAAW,CAAE,CAAA,CAAC;;;AAInD,IAAA,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;AACxC,QAAA,WAAW,CAAC,IAAI,CAAC,GAAG,mBAAmB,EAAE,CAAC;;SACrC;AACL,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;;AAItB,IAAA,CAAC,MAAM,CAAC,UAAU,IAAI,EAAE;AACrB,SAAA,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;AACtC,SAAA,OAAO,CAAC,CAAC,CAAC,EAAE,MAAM,KAAI;AACrB,QAAA,WAAW,CAAC,IAAI,CACd,iBAAiB,MAAM,GAAG,CAAC,CAAgB,aAAA,EAAA,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAA,CAAA,CAAG,CACjE;AACD,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,QAAA,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC;QAC3B,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;AAC/B,QAAA,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC;AACvB,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AAEpB,QAAA,IAAI,CAAC,CAAC,UAAU,IAAI,IAAI,IAAI,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE;YAC/C,IAAI,SAAS,GAAG,KAAK;YACrB,MAAM,QAAQ,GAAa,EAAE;AAE7B,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBAC5C,MAAM,GAAG,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC;gBAC3B,IAAI,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE;oBACjD;;AAEF,gBAAA,IAAI,GAAG,CAAC,IAAI,KAAK,MAAM,EAAE;oBACvB;;gBAEF,IAAI,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE;oBAChD;;gBAEF,UAAU,CAAC,IAAI,CAAC;oBACd,IAAI,EAAE,GAAG,CAAC,IAAI;AACd,oBAAA,IAAI,EAAE,GAAG,CAAC,SAAS,CAAC,WAAW;oBAC/B,WAAW,EAAE,aAAa,CAAC,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC;AACrD,oBAAA,KAAK,EAAE,CACL,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,IAAI,EAAE,KAAK,GAAG,CAAC,SAAS,CAAC,IAAI;wBAClD,EAAE,EACF,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AACjB,iBAAA,CAAC;gBAEF,IAAI,CAAC,SAAS,EAAE;AACd,oBAAA,QAAQ,CAAC,IAAI,CAAC,kBAAkB,CAAC;oBACjC,SAAS,GAAG,IAAI;;gBAGlB,QAAQ,CAAC,IAAI,CACX,CAAA,EAAA,EAAK,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,aAAa,GAAG,CAAC,KAAK,GAAG,CAAC,SAAS,CAAC,WAAW,CAAE,CAAA,CACvE;AACD,gBAAA,QAAQ,CAAC,IAAI,CACX,CAAA,uBAAA,EAA0B,IAAI,CAAA,GAAA,EAAM,UAAU,CAAC,MAAM,GAAG,CAAC,CAAA,CAAE,CAC5D;;YAGH,IAAI,SAAS,EAAE;AACb,gBAAA,WAAW,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC;AAC7B,gBAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;;AAIxB,QAAA,IAAI,MAAM,GAAG,CAAC,MAAM,CAAC,UAAU,EAAE,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;AACjD,YAAA,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC;AACvB,YAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;AAExB,KAAC,CAAC;AAEJ,IAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,IAAA,OAAO,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC;AAC/B;AAEgB,SAAA,mBAAmB,CACjC,IAAY,EACZ,OAA2B,EAAA;;IAG3B,MAAM,WAAW,GAAa,EAAE;AAEhC,IAAA,MAAM,UAAU,GAAG,CAAC,KAAa,KAAU;AACzC,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,QAAA,WAAW,CAAC,IAAI,CAAC,OAAO,KAAK,CAAA,IAAA,CAAM,CAAC;AACpC,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACtB,KAAC;IAED,MAAM,UAAU,GAAwB,EAAE;;AAG1C,IAAA,IAAI,OAAO,CAAC,OAAO,EAAE,MAAM,IAAI,IAAI,IAAI,OAAO,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE;AACjE,QAAA,UAAU,CAAC,CAAA,kBAAA,EAAqB,IAAI,CAAA,CAAE,CAAC;AACvC,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAC/C,MAAM,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC;AAC5B,YAAA,WAAW,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,CAAC,CAAC;YAChE,OAAO,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,UAAU;;;;AAKxC,IAAA,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,IAAI,EAAE;AAC3C,IAAA,IAAI,UAAU,CAAC,MAAM,EAAE;QACrB,UAAU,CAAC,cAAc,CAAC;AAC1B,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AAC1C,YAAA,MAAM,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC;AACvB,YAAA,WAAW,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,UAAU,CAAC,CAAC;YAC9D,IAAI,OAAO,CAAC,UAAU,GAAG,CAAC,CAAC,EAAE,UAAU,EAAE;gBACvC,OAAO,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,UAAU;;;;;;;;;;;;;;;;;AAmB7C,IAAA,IAAI,OAAO,CAAC,cAAc,IAAI,IAAI,EAAE;QAClC,UAAU,CAAC,iBAAiB,CAAC;AAC7B,QAAA,MAAM,OAAO,GAAG;AACd,YAAA,CAAA,WAAA,EAAc,OAAO,CAAC,cAAc,CAAC,KAAK,IAAI,YAAY,CAAE,CAAA;AAC5D,YAAA,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI;AAC7B,kBAAE,CAAa,UAAA,EAAA,OAAO,CAAC,cAAc,CAAC,IAAI,CAAE;AAC5C,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,WAAW,IAAI;AACpC,kBAAE,CAAoB,iBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,WAAW,CAAE;AAC1D,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,iBAAiB,IAAI;AAC1C,kBAAE,CAA2B,wBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,iBAAiB,CAAE;AACvE,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,eAAe,IAAI;AACxC,kBAAE,CAAyB,sBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,eAAe,CAAE;AACnE,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,IAAI;AACjC,kBAAE,CAAkB,eAAA,EAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,CAAE;AACrD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,OAAO,IAAI;AAChC,kBAAE,CAAgB,aAAA,EAAA,OAAO,CAAC,cAAc,CAAC,OAAO,CAAE;AAClD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,UAAU,IAAI;AACnC,kBAAE,CAAgC,6BAAA,EAAA,IAAI,CAAC,SAAS,CAC9C,OAAO,CAAC,cAAc,CAAC,UAAU,EACjC,IAAI,EACJ,CAAC,CACF,CAAU,QAAA;AACX,kBAAE,EAAE;YACN,EAAE;AACH,SAAA,CAAC,MAAM,CAAC,OAAO,CAAC;QAEjB,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;;;AAIxC,IAAA,IAAI,OAAO,CAAC,SAAS,IAAI,IAAI,EAAE;QAC7B,UAAU,CAAC,YAAY,CAAC;AACxB,QAAA,MAAM,OAAO,GAAG;AACd,YAAA,OAAO,CAAC,SAAS,CAAC,KAAK,IAAI;AACzB,kBAAE,CAAc,WAAA,EAAA,OAAO,CAAC,SAAS,CAAC,KAAK,CAAE;AACzC,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,OAAO,IAAI;AAC3B,kBAAE,CAAgB,aAAA,EAAA,OAAO,CAAC,SAAS,CAAC,OAAO,CAAE;AAC7C,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,kBAAkB,IAAI;AACtC,kBAAE,CAA4B,yBAAA,EAAA,OAAO,CAAC,SAAS,CAAC;qBAC7C,GAAG,CAAC,CAAC,CAAC,KAAK,CAAA,EAAA,EAAK,CAAC,CAAA,EAAA,CAAI;qBACrB,IAAI,CAAC,GAAG,CAAC,CAAE;AACd,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,IAAI,IAAI;AACxB,kBAAE,CAAa,UAAA,EAAA,OAAO,CAAC,SAAS,CAAC,IAAI,CAAE;AACvC,kBAAE,EAAE;YACN,EAAE;AACH,SAAA,CAAC,MAAM,CAAC,OAAO,CAAC;QAEjB,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;;;AAIxC,IAAA,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,EAAE;AACjD,IAAA,IAAI,aAAa,CAAC,MAAM,EAAE;QACxB,UAAU,CAAC,iBAAiB,CAAC;QAE7B,MAAM,QAAQ,GAAa,EAAE;QAC7B,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,KAAI;AAC7B,YAAA,MAAM,aAAa,GAAG;gBACpB,CAAgB,aAAA,EAAA,CAAC,GAAG,CAAC,CAAG,CAAA,CAAA;gBACxB,CAAI,CAAA,EAAA,CAAC,CAAC,QAAQ,CAAG,CAAA,CAAA;gBACjB,CAAG,EAAA,CAAC,CAAC,OAAO,IAAI,IAAI,IAAI,CAAC,CAAC,OAAO,GAAG,CAAA,SAAA,EAAY,CAAC,CAAC,OAAO,EAAE,GAAG,EAAE,CAAE,CAAA;gBAClE,CAAG,EAAA,CAAC,CAAC,KAAK,IAAI,IAAI,IAAI,CAAC,CAAC,KAAK,GAAG,CAAA,OAAA,EAAU,CAAC,CAAC,KAAK,EAAE,GAAG,EAAE,CAAE,CAAA;gBAC1D,CAAG,EAAA,CAAC,CAAC,IAAI,IAAI,IAAI,IAAI,CAAC,CAAC,IAAI,GAAG,CAAA,MAAA,EAAS,CAAC,CAAC,IAAI,EAAE,GAAG,EAAE,CAAE,CAAA;gBACtD,EAAE;AACH,aAAA,CAAC,MAAM,CAAC,OAAO,CAAC;YAEjB,QAAQ,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AAC3C,SAAC,CAAC;QAEF,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;;IAGrC,OAAO;QACL,MAAM,EAAE,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE;QACrC,UAAU;KACX;AACH;;;;"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"highlights.mjs","sources":["../../../../src/tools/search/highlights.ts"],"sourcesContent":["import type * as t from './types';\r\n\r\n// 2. Pre-compile all regular expressions (only do this once)\r\n// Group patterns by priority for early returns\r\nconst priorityPatterns = [\r\n // High priority patterns (structural)\r\n [\r\n { regex: /\\n\\n/g }, // Double newline (paragraph break)\r\n { regex: /\\n/g }, // Single newline\r\n { regex: /={3,}\\s*\\n|-{3,}\\s*\\n/g }, // Section separators\r\n ],\r\n // Medium priority (semantic)\r\n [\r\n { regex: /[.!?][\")\\]]?\\s/g }, // End of sentence\r\n { regex: /;\\s/g }, // Semicolon\r\n { regex: /:\\s/g }, // Colon\r\n ],\r\n // Low priority (any breaks)\r\n [\r\n { regex: /,\\s/g }, // Comma\r\n { regex: /\\s-\\s/g }, // Dash surrounded by spaces\r\n { regex: /\\s/g }, // Any space\r\n ],\r\n];\r\n\r\nfunction findFirstMatch(text: string, regex: RegExp): number {\r\n // Reset regex\r\n regex.lastIndex = 0;\r\n\r\n // For very long texts, try chunking\r\n if (text.length > 10000) {\r\n const chunkSize = 2000;\r\n let position = 0;\r\n\r\n while (position < text.length) {\r\n const chunk = text.substring(position, position + chunkSize);\r\n regex.lastIndex = 0;\r\n\r\n const match = regex.exec(chunk);\r\n if (match) {\r\n return position + match.index;\r\n }\r\n\r\n // Move to next chunk with some overlap\r\n position += chunkSize - 100;\r\n if (position >= text.length) break;\r\n }\r\n return -1;\r\n }\r\n\r\n // For shorter texts, normal regex search\r\n const match = regex.exec(text);\r\n return match ? match.index : -1;\r\n}\r\n\r\n// 3. Optimized boundary finding functions\r\nfunction findLastMatch(text: string, regex: RegExp): number {\r\n // Reset regex state\r\n regex.lastIndex = 0;\r\n\r\n let lastIndex = -1;\r\n let lastLength = 0;\r\n let match;\r\n\r\n // For very long texts, use a different approach to avoid regex engine slowdowns\r\n if (text.length > 10000) {\r\n // Try dividing the text into chunks for faster processing\r\n const chunkSize = 2000;\r\n let startPosition = Math.max(0, text.length - chunkSize);\r\n\r\n while (startPosition >= 0) {\r\n const chunk = text.substring(startPosition, startPosition + chunkSize);\r\n regex.lastIndex = 0;\r\n\r\n let chunkLastIndex = -1;\r\n let chunkLastLength = 0;\r\n\r\n while ((match = regex.exec(chunk)) !== null) {\r\n chunkLastIndex = match.index;\r\n chunkLastLength = match[0].length;\r\n }\r\n\r\n if (chunkLastIndex !== -1) {\r\n return startPosition + chunkLastIndex + chunkLastLength;\r\n }\r\n\r\n // Move to previous chunk with some overlap\r\n startPosition = Math.max(0, startPosition - chunkSize + 100) - 1;\r\n if (startPosition <= 0) break;\r\n }\r\n return -1;\r\n }\r\n\r\n // For shorter texts, normal regex search\r\n while ((match = regex.exec(text)) !== null) {\r\n lastIndex = match.index;\r\n lastLength = match[0].length;\r\n }\r\n\r\n return lastIndex === -1 ? -1 : lastIndex + lastLength;\r\n}\r\n\r\n// 4. Find the best boundary with priority groups\r\nfunction findBestBoundary(text: string, direction = 'backward'): number {\r\n if (!text || text.length === 0) return 0;\r\n\r\n // Try each priority group\r\n for (const patternGroup of priorityPatterns) {\r\n for (const pattern of patternGroup) {\r\n const position =\r\n direction === 'backward'\r\n ? findLastMatch(text, pattern.regex)\r\n : findFirstMatch(text, pattern.regex);\r\n\r\n if (position !== -1) {\r\n return position;\r\n }\r\n }\r\n }\r\n\r\n // No match found, use character boundary\r\n return direction === 'backward' ? text.length : 0;\r\n}\r\n\r\n/**\r\n * Tracks references used in a highlight without changing their numbers\r\n */\r\nfunction trackReferencesInHighlight(\r\n text: string,\r\n sourceResult: t.ValidSource // Source containing the original references\r\n): {\r\n references: {\r\n type: 'link' | 'image' | 'video';\r\n originalIndex: number;\r\n reference: t.MediaReference; // Original reference object\r\n }[];\r\n} {\r\n // Track used references\r\n const references: {\r\n type: 'link' | 'image' | 'video';\r\n originalIndex: number;\r\n reference: t.MediaReference;\r\n }[] = [];\r\n\r\n if (!text || text.length === 0 || !text.includes('#')) {\r\n return { references }; // Early return\r\n }\r\n\r\n // Quick check for reference markers\r\n if (\r\n !text.includes('link#') &&\r\n !text.includes('image#') &&\r\n !text.includes('video#')\r\n ) {\r\n return { references };\r\n }\r\n\r\n // Get references from the source if available\r\n const sourceRefs = sourceResult.references || {\r\n links: [],\r\n images: [],\r\n videos: [],\r\n };\r\n\r\n // Find references but don't modify text\r\n const refRegex = /\\((link|image|video)#(\\d+)(?:\\s+\"([^\"]*)\")?\\)/g;\r\n let match;\r\n\r\n while ((match = refRegex.exec(text)) !== null) {\r\n const [, type, indexStr] = match;\r\n const originalIndex = parseInt(indexStr, 10) - 1; // Convert to 0-based\r\n\r\n // Get the source array for this type\r\n const refType = type as 'link' | 'image' | 'video';\r\n const sourceArray = sourceRefs[`${refType}s`] as\r\n | t.MediaReference[]\r\n | undefined;\r\n\r\n // Skip if invalid reference\r\n if (\r\n !sourceArray ||\r\n originalIndex < 0 ||\r\n originalIndex >= sourceArray.length\r\n ) {\r\n continue; // Skip invalid references\r\n }\r\n\r\n // Get original reference\r\n const reference = sourceArray[originalIndex];\r\n\r\n // Track if not already tracked\r\n const alreadyTracked = references.some(\r\n (ref) => ref.type === refType && ref.originalIndex === originalIndex\r\n );\r\n\r\n if (!alreadyTracked) {\r\n references.push({\r\n type: refType,\r\n originalIndex,\r\n reference,\r\n });\r\n }\r\n }\r\n\r\n return { references };\r\n}\r\n\r\n/**\r\n * Expand highlights in search results using smart boundary detection.\r\n *\r\n * This implementation finds natural text boundaries like paragraphs, sentences,\r\n * and phrases to provide context while maintaining readability.\r\n *\r\n * @param searchResults - Search results object\r\n * @param mainExpandBy - Primary expansion size on each side (default: 300)\r\n * @param separatorExpandBy - Additional range to look for separators (default: 150)\r\n * @returns Copy of search results with expanded highlights and tracked references\r\n */\r\nexport function expandHighlights(\r\n searchResults: t.SearchResultData,\r\n mainExpandBy = 300,\r\n separatorExpandBy = 150\r\n): t.SearchResultData {\r\n // Avoid deep copy - only copy what we modify\r\n const resultCopy = { ...searchResults };\r\n if (resultCopy.organic) resultCopy.organic = [...resultCopy.organic];\r\n if (resultCopy.topStories) resultCopy.topStories = [...resultCopy.topStories];\r\n\r\n // Process the results efficiently\r\n const processResultTypes = ['organic', 'topStories'] as const;\r\n\r\n for (const resultType of processResultTypes) {\r\n if (!resultCopy[resultType as 'organic' | 'topStories']) continue;\r\n\r\n // Map results to new array with modified highlights\r\n resultCopy[resultType] = resultCopy[resultType]?.map((result) => {\r\n if (\r\n result.content == null ||\r\n result.content === '' ||\r\n !result.highlights ||\r\n result.highlights.length === 0\r\n ) {\r\n return result; // No modification needed\r\n }\r\n\r\n // Create a shallow copy with expanded highlights\r\n const resultCopy = { ...result };\r\n const content = result.content;\r\n const highlights = [];\r\n // Process each highlight\r\n for (const highlight of result.highlights) {\r\n const { references } = trackReferencesInHighlight(\r\n highlight.text,\r\n result\r\n );\r\n\r\n let startPos = content.indexOf(highlight.text);\r\n let highlightLen = highlight.text.length;\r\n\r\n if (startPos === -1) {\r\n // Try with stripped whitespace\r\n const strippedHighlight = highlight.text.trim();\r\n startPos = content.indexOf(strippedHighlight);\r\n\r\n if (startPos === -1) {\r\n highlights.push({\r\n text: highlight.text,\r\n score: highlight.score,\r\n references,\r\n });\r\n continue;\r\n }\r\n highlightLen = strippedHighlight.length;\r\n }\r\n\r\n // Calculate boundaries\r\n const mainStart = Math.max(0, startPos - mainExpandBy);\r\n const mainEnd = Math.min(\r\n content.length,\r\n startPos + highlightLen + mainExpandBy\r\n );\r\n\r\n const separatorStart = Math.max(0, mainStart - separatorExpandBy);\r\n const separatorEnd = Math.min(\r\n content.length,\r\n mainEnd + separatorExpandBy\r\n );\r\n\r\n // Extract text segments\r\n const headText = content.substring(separatorStart, mainStart);\r\n const tailText = content.substring(mainEnd, separatorEnd);\r\n\r\n // Find natural boundaries\r\n const bestHeadBoundary = findBestBoundary(headText, 'backward');\r\n const bestTailBoundary = findBestBoundary(tailText, 'forward');\r\n\r\n // Calculate final positions\r\n const finalStart = separatorStart + bestHeadBoundary;\r\n const finalEnd = mainEnd + bestTailBoundary;\r\n\r\n // Extract the expanded highlight\r\n const expandedHighlightText = content\r\n .substring(finalStart, finalEnd)\r\n .trim();\r\n highlights.push({\r\n text: expandedHighlightText,\r\n score: highlight.score,\r\n references,\r\n });\r\n }\r\n\r\n resultCopy.highlights = highlights;\r\n delete resultCopy.content;\r\n delete resultCopy.references;\r\n return resultCopy;\r\n });\r\n }\r\n\r\n return resultCopy;\r\n}\r\n"],"names":[],"mappings":"AAEA;AACA;AACA,MAAM,gBAAgB,GAAG;;AAEvB,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,OAAO,EAAE;AAClB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AAChB,QAAA,EAAE,KAAK,EAAE,wBAAwB,EAAE;AACpC,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,iBAAiB,EAAE;AAC5B,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AAClB,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,QAAQ,EAAE;AACnB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AACjB,KAAA;CACF;AAED,SAAS,cAAc,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEjD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;;AAGnB,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;QACvB,MAAM,SAAS,GAAG,IAAI;QACtB,IAAI,QAAQ,GAAG,CAAC;AAEhB,QAAA,OAAO,QAAQ,GAAG,IAAI,CAAC,MAAM,EAAE;AAC7B,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,CAAC;AAC5D,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;YAEnB,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;YAC/B,IAAI,KAAK,EAAE;AACT,gBAAA,OAAO,QAAQ,GAAG,KAAK,CAAC,KAAK;;;AAI/B,YAAA,QAAQ,IAAI,SAAS,GAAG,GAAG;AAC3B,YAAA,IAAI,QAAQ,IAAI,IAAI,CAAC,MAAM;gBAAE;;QAE/B,OAAO,EAAE;;;IAIX,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;AAC9B,IAAA,OAAO,KAAK,GAAG,KAAK,CAAC,KAAK,GAAG,EAAE;AACjC;AAEA;AACA,SAAS,aAAa,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEhD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,IAAA,IAAI,SAAS,GAAG,EAAE;IAClB,IAAI,UAAU,GAAG,CAAC;AAClB,IAAA,IAAI,KAAK;;AAGT,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;;QAEvB,MAAM,SAAS,GAAG,IAAI;AACtB,QAAA,IAAI,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC;AAExD,QAAA,OAAO,aAAa,IAAI,CAAC,EAAE;AACzB,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,aAAa,EAAE,aAAa,GAAG,SAAS,CAAC;AACtE,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,YAAA,IAAI,cAAc,GAAG,EAAE;YACvB,IAAI,eAAe,GAAG,CAAC;AAEvB,YAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE;AAC3C,gBAAA,cAAc,GAAG,KAAK,CAAC,KAAK;AAC5B,gBAAA,eAAe,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAGnC,YAAA,IAAI,cAAc,KAAK,EAAE,EAAE;AACzB,gBAAA,OAAO,aAAa,GAAG,cAAc,GAAG,eAAe;;;AAIzD,YAAA,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,GAAG,SAAS,GAAG,GAAG,CAAC,GAAG,CAAC;YAChE,IAAI,aAAa,IAAI,CAAC;gBAAE;;QAE1B,OAAO,EAAE;;;AAIX,IAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,EAAE;AAC1C,QAAA,SAAS,GAAG,KAAK,CAAC,KAAK;AACvB,QAAA,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAG9B,IAAA,OAAO,SAAS,KAAK,EAAE,GAAG,EAAE,GAAG,SAAS,GAAG,UAAU;AACvD;AAEA;AACA,SAAS,gBAAgB,CAAC,IAAY,EAAE,SAAS,GAAG,UAAU,EAAA;AAC5D,IAAA,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;AAAE,QAAA,OAAO,CAAC;;AAGxC,IAAA,KAAK,MAAM,YAAY,IAAI,gBAAgB,EAAE;AAC3C,QAAA,KAAK,MAAM,OAAO,IAAI,YAAY,EAAE;AAClC,YAAA,MAAM,QAAQ,GACZ,SAAS,KAAK;kBACV,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK;kBACjC,cAAc,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK,CAAC;AAEzC,YAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;AACnB,gBAAA,OAAO,QAAQ;;;;;AAMrB,IAAA,OAAO,SAAS,KAAK,UAAU,GAAG,IAAI,CAAC,MAAM,GAAG,CAAC;AACnD;AAEA;;AAEG;AACH,SAAS,0BAA0B,CACjC,IAAY,EACZ,YAA2B;;;IAS3B,MAAM,UAAU,GAIV,EAAE;AAER,IAAA,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE;AACrD,QAAA,OAAO,EAAE,UAAU,EAAE,CAAC;;;AAIxB,IAAA,IACE,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC;AACvB,QAAA,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC;AACxB,QAAA,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EACxB;QACA,OAAO,EAAE,UAAU,EAAE;;;AAIvB,IAAA,MAAM,UAAU,GAAG,YAAY,CAAC,UAAU,IAAI;AAC5C,QAAA,KAAK,EAAE,EAAE;AACT,QAAA,MAAM,EAAE,EAAE;AACV,QAAA,MAAM,EAAE,EAAE;KACX;;IAGD,MAAM,QAAQ,GAAG,gDAAgD;AACjE,IAAA,IAAI,KAAK;AAET,IAAA,OAAO,CAAC,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,EAAE;QAC7C,MAAM,GAAG,IAAI,EAAE,QAAQ,CAAC,GAAG,KAAK;AAChC,QAAA,MAAM,aAAa,GAAG,QAAQ,CAAC,QAAQ,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC;;QAGjD,MAAM,OAAO,GAAG,IAAkC;QAClD,MAAM,WAAW,GAAG,UAAU,CAAC,GAAG,OAAO,CAAA,CAAA,CAAG,CAE/B;;AAGb,QAAA,IACE,CAAC,WAAW;AACZ,YAAA,aAAa,GAAG,CAAC;AACjB,YAAA,aAAa,IAAI,WAAW,CAAC,MAAM,EACnC;AACA,YAAA,SAAS;;;AAIX,QAAA,MAAM,SAAS,GAAG,WAAW,CAAC,aAAa,CAAC;;QAG5C,MAAM,cAAc,GAAG,UAAU,CAAC,IAAI,CACpC,CAAC,GAAG,KAAK,GAAG,CAAC,IAAI,KAAK,OAAO,IAAI,GAAG,CAAC,aAAa,KAAK,aAAa,CACrE;QAED,IAAI,CAAC,cAAc,EAAE;YACnB,UAAU,CAAC,IAAI,CAAC;AACd,gBAAA,IAAI,EAAE,OAAO;gBACb,aAAa;gBACb,SAAS;AACV,aAAA,CAAC;;;IAIN,OAAO,EAAE,UAAU,EAAE;AACvB;AAEA;;;;;;;;;;AAUG;AACG,SAAU,gBAAgB,CAC9B,aAAiC,EACjC,YAAY,GAAG,GAAG,EAClB,iBAAiB,GAAG,GAAG,EAAA;;AAGvB,IAAA,MAAM,UAAU,GAAG,EAAE,GAAG,aAAa,EAAE;IACvC,IAAI,UAAU,CAAC,OAAO;QAAE,UAAU,CAAC,OAAO,GAAG,CAAC,GAAG,UAAU,CAAC,OAAO,CAAC;IACpE,IAAI,UAAU,CAAC,UAAU;QAAE,UAAU,CAAC,UAAU,GAAG,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC;;AAG7E,IAAA,MAAM,kBAAkB,GAAG,CAAC,SAAS,EAAE,YAAY,CAAU;AAE7D,IAAA,KAAK,MAAM,UAAU,IAAI,kBAAkB,EAAE;AAC3C,QAAA,IAAI,CAAC,UAAU,CAAC,UAAsC,CAAC;YAAE;;AAGzD,QAAA,UAAU,CAAC,UAAU,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,MAAM,KAAI;AAC9D,YAAA,IACE,MAAM,CAAC,OAAO,IAAI,IAAI;gBACtB,MAAM,CAAC,OAAO,KAAK,EAAE;gBACrB,CAAC,MAAM,CAAC,UAAU;AAClB,gBAAA,MAAM,CAAC,UAAU,CAAC,MAAM,KAAK,CAAC,EAC9B;gBACA,OAAO,MAAM,CAAC;;;AAIhB,YAAA,MAAM,UAAU,GAAG,EAAE,GAAG,MAAM,EAAE;AAChC,YAAA,MAAM,OAAO,GAAG,MAAM,CAAC,OAAO;YAC9B,MAAM,UAAU,GAAG,EAAE;;AAErB,YAAA,KAAK,MAAM,SAAS,IAAI,MAAM,CAAC,UAAU,EAAE;AACzC,gBAAA,MAAM,EAAE,UAAU,EAAE,GAAG,0BAA0B,CAC/C,SAAS,CAAC,IAAI,EACd,MAAM,CACP;gBAED,IAAI,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,SAAS,CAAC,IAAI,CAAC;AAC9C,gBAAA,IAAI,YAAY,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM;AAExC,gBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;;oBAEnB,MAAM,iBAAiB,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,EAAE;AAC/C,oBAAA,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,iBAAiB,CAAC;AAE7C,oBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;wBACnB,UAAU,CAAC,IAAI,CAAC;4BACd,IAAI,EAAE,SAAS,CAAC,IAAI;4BACpB,KAAK,EAAE,SAAS,CAAC,KAAK;4BACtB,UAAU;AACX,yBAAA,CAAC;wBACF;;AAEF,oBAAA,YAAY,GAAG,iBAAiB,CAAC,MAAM;;;AAIzC,gBAAA,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,GAAG,YAAY,CAAC;AACtD,gBAAA,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CACtB,OAAO,CAAC,MAAM,EACd,QAAQ,GAAG,YAAY,GAAG,YAAY,CACvC;AAED,gBAAA,MAAM,cAAc,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,SAAS,GAAG,iBAAiB,CAAC;AACjE,gBAAA,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAC3B,OAAO,CAAC,MAAM,EACd,OAAO,GAAG,iBAAiB,CAC5B;;gBAGD,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,cAAc,EAAE,SAAS,CAAC;gBAC7D,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,OAAO,EAAE,YAAY,CAAC;;gBAGzD,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,UAAU,CAAC;gBAC/D,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,SAAS,CAAC;;AAG9D,gBAAA,MAAM,UAAU,GAAG,cAAc,GAAG,gBAAgB;AACpD,gBAAA,MAAM,QAAQ,GAAG,OAAO,GAAG,gBAAgB;;gBAG3C,MAAM,qBAAqB,GAAG;AAC3B,qBAAA,SAAS,CAAC,UAAU,EAAE,QAAQ;AAC9B,qBAAA,IAAI,EAAE;gBACT,UAAU,CAAC,IAAI,CAAC;AACd,oBAAA,IAAI,EAAE,qBAAqB;oBAC3B,KAAK,EAAE,SAAS,CAAC,KAAK;oBACtB,UAAU;AACX,iBAAA,CAAC;;AAGJ,YAAA,UAAU,CAAC,UAAU,GAAG,UAAU;YAClC,OAAO,UAAU,CAAC,OAAO;YACzB,OAAO,UAAU,CAAC,UAAU;AAC5B,YAAA,OAAO,UAAU;AACnB,SAAC,CAAC;;AAGJ,IAAA,OAAO,UAAU;AACnB;;;;"}
|
|
1
|
+
{"version":3,"file":"highlights.mjs","sources":["../../../../src/tools/search/highlights.ts"],"sourcesContent":["import type * as t from './types';\n\n// 2. Pre-compile all regular expressions (only do this once)\n// Group patterns by priority for early returns\nconst priorityPatterns = [\n // High priority patterns (structural)\n [\n { regex: /\\n\\n/g }, // Double newline (paragraph break)\n { regex: /\\n/g }, // Single newline\n { regex: /={3,}\\s*\\n|-{3,}\\s*\\n/g }, // Section separators\n ],\n // Medium priority (semantic)\n [\n { regex: /[.!?][\")\\]]?\\s/g }, // End of sentence\n { regex: /;\\s/g }, // Semicolon\n { regex: /:\\s/g }, // Colon\n ],\n // Low priority (any breaks)\n [\n { regex: /,\\s/g }, // Comma\n { regex: /\\s-\\s/g }, // Dash surrounded by spaces\n { regex: /\\s/g }, // Any space\n ],\n];\n\nfunction findFirstMatch(text: string, regex: RegExp): number {\n // Reset regex\n regex.lastIndex = 0;\n\n // For very long texts, try chunking\n if (text.length > 10000) {\n const chunkSize = 2000;\n let position = 0;\n\n while (position < text.length) {\n const chunk = text.substring(position, position + chunkSize);\n regex.lastIndex = 0;\n\n const match = regex.exec(chunk);\n if (match) {\n return position + match.index;\n }\n\n // Move to next chunk with some overlap\n position += chunkSize - 100;\n if (position >= text.length) break;\n }\n return -1;\n }\n\n // For shorter texts, normal regex search\n const match = regex.exec(text);\n return match ? match.index : -1;\n}\n\n// 3. Optimized boundary finding functions\nfunction findLastMatch(text: string, regex: RegExp): number {\n // Reset regex state\n regex.lastIndex = 0;\n\n let lastIndex = -1;\n let lastLength = 0;\n let match;\n\n // For very long texts, use a different approach to avoid regex engine slowdowns\n if (text.length > 10000) {\n // Try dividing the text into chunks for faster processing\n const chunkSize = 2000;\n let startPosition = Math.max(0, text.length - chunkSize);\n\n while (startPosition >= 0) {\n const chunk = text.substring(startPosition, startPosition + chunkSize);\n regex.lastIndex = 0;\n\n let chunkLastIndex = -1;\n let chunkLastLength = 0;\n\n while ((match = regex.exec(chunk)) !== null) {\n chunkLastIndex = match.index;\n chunkLastLength = match[0].length;\n }\n\n if (chunkLastIndex !== -1) {\n return startPosition + chunkLastIndex + chunkLastLength;\n }\n\n // Move to previous chunk with some overlap\n startPosition = Math.max(0, startPosition - chunkSize + 100) - 1;\n if (startPosition <= 0) break;\n }\n return -1;\n }\n\n // For shorter texts, normal regex search\n while ((match = regex.exec(text)) !== null) {\n lastIndex = match.index;\n lastLength = match[0].length;\n }\n\n return lastIndex === -1 ? -1 : lastIndex + lastLength;\n}\n\n// 4. Find the best boundary with priority groups\nfunction findBestBoundary(text: string, direction = 'backward'): number {\n if (!text || text.length === 0) return 0;\n\n // Try each priority group\n for (const patternGroup of priorityPatterns) {\n for (const pattern of patternGroup) {\n const position =\n direction === 'backward'\n ? findLastMatch(text, pattern.regex)\n : findFirstMatch(text, pattern.regex);\n\n if (position !== -1) {\n return position;\n }\n }\n }\n\n // No match found, use character boundary\n return direction === 'backward' ? text.length : 0;\n}\n\n/**\n * Tracks references used in a highlight without changing their numbers\n */\nfunction trackReferencesInHighlight(\n text: string,\n sourceResult: t.ValidSource // Source containing the original references\n): {\n references: {\n type: 'link' | 'image' | 'video';\n originalIndex: number;\n reference: t.MediaReference; // Original reference object\n }[];\n} {\n // Track used references\n const references: {\n type: 'link' | 'image' | 'video';\n originalIndex: number;\n reference: t.MediaReference;\n }[] = [];\n\n if (!text || text.length === 0 || !text.includes('#')) {\n return { references }; // Early return\n }\n\n // Quick check for reference markers\n if (\n !text.includes('link#') &&\n !text.includes('image#') &&\n !text.includes('video#')\n ) {\n return { references };\n }\n\n // Get references from the source if available\n const sourceRefs = sourceResult.references || {\n links: [],\n images: [],\n videos: [],\n };\n\n // Find references but don't modify text\n const refRegex = /\\((link|image|video)#(\\d+)(?:\\s+\"([^\"]*)\")?\\)/g;\n let match;\n\n while ((match = refRegex.exec(text)) !== null) {\n const [, type, indexStr] = match;\n const originalIndex = parseInt(indexStr, 10) - 1; // Convert to 0-based\n\n // Get the source array for this type\n const refType = type as 'link' | 'image' | 'video';\n const sourceArray = sourceRefs[`${refType}s`] as\n | t.MediaReference[]\n | undefined;\n\n // Skip if invalid reference\n if (\n !sourceArray ||\n originalIndex < 0 ||\n originalIndex >= sourceArray.length\n ) {\n continue; // Skip invalid references\n }\n\n // Get original reference\n const reference = sourceArray[originalIndex];\n\n // Track if not already tracked\n const alreadyTracked = references.some(\n (ref) => ref.type === refType && ref.originalIndex === originalIndex\n );\n\n if (!alreadyTracked) {\n references.push({\n type: refType,\n originalIndex,\n reference,\n });\n }\n }\n\n return { references };\n}\n\n/**\n * Expand highlights in search results using smart boundary detection.\n *\n * This implementation finds natural text boundaries like paragraphs, sentences,\n * and phrases to provide context while maintaining readability.\n *\n * @param searchResults - Search results object\n * @param mainExpandBy - Primary expansion size on each side (default: 300)\n * @param separatorExpandBy - Additional range to look for separators (default: 150)\n * @returns Copy of search results with expanded highlights and tracked references\n */\nexport function expandHighlights(\n searchResults: t.SearchResultData,\n mainExpandBy = 300,\n separatorExpandBy = 150\n): t.SearchResultData {\n // Avoid deep copy - only copy what we modify\n const resultCopy = { ...searchResults };\n if (resultCopy.organic) resultCopy.organic = [...resultCopy.organic];\n if (resultCopy.topStories) resultCopy.topStories = [...resultCopy.topStories];\n\n // Process the results efficiently\n const processResultTypes = ['organic', 'topStories'] as const;\n\n for (const resultType of processResultTypes) {\n if (!resultCopy[resultType as 'organic' | 'topStories']) continue;\n\n // Map results to new array with modified highlights\n resultCopy[resultType] = resultCopy[resultType]?.map((result) => {\n if (\n result.content == null ||\n result.content === '' ||\n !result.highlights ||\n result.highlights.length === 0\n ) {\n return result; // No modification needed\n }\n\n // Create a shallow copy with expanded highlights\n const resultCopy = { ...result };\n const content = result.content;\n const highlights = [];\n // Process each highlight\n for (const highlight of result.highlights) {\n const { references } = trackReferencesInHighlight(\n highlight.text,\n result\n );\n\n let startPos = content.indexOf(highlight.text);\n let highlightLen = highlight.text.length;\n\n if (startPos === -1) {\n // Try with stripped whitespace\n const strippedHighlight = highlight.text.trim();\n startPos = content.indexOf(strippedHighlight);\n\n if (startPos === -1) {\n highlights.push({\n text: highlight.text,\n score: highlight.score,\n references,\n });\n continue;\n }\n highlightLen = strippedHighlight.length;\n }\n\n // Calculate boundaries\n const mainStart = Math.max(0, startPos - mainExpandBy);\n const mainEnd = Math.min(\n content.length,\n startPos + highlightLen + mainExpandBy\n );\n\n const separatorStart = Math.max(0, mainStart - separatorExpandBy);\n const separatorEnd = Math.min(\n content.length,\n mainEnd + separatorExpandBy\n );\n\n // Extract text segments\n const headText = content.substring(separatorStart, mainStart);\n const tailText = content.substring(mainEnd, separatorEnd);\n\n // Find natural boundaries\n const bestHeadBoundary = findBestBoundary(headText, 'backward');\n const bestTailBoundary = findBestBoundary(tailText, 'forward');\n\n // Calculate final positions\n const finalStart = separatorStart + bestHeadBoundary;\n const finalEnd = mainEnd + bestTailBoundary;\n\n // Extract the expanded highlight\n const expandedHighlightText = content\n .substring(finalStart, finalEnd)\n .trim();\n highlights.push({\n text: expandedHighlightText,\n score: highlight.score,\n references,\n });\n }\n\n resultCopy.highlights = highlights;\n delete resultCopy.content;\n delete resultCopy.references;\n return resultCopy;\n });\n }\n\n return resultCopy;\n}\n"],"names":[],"mappings":"AAEA;AACA;AACA,MAAM,gBAAgB,GAAG;;AAEvB,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,OAAO,EAAE;AAClB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AAChB,QAAA,EAAE,KAAK,EAAE,wBAAwB,EAAE;AACpC,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,iBAAiB,EAAE;AAC5B,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AAClB,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,QAAQ,EAAE;AACnB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AACjB,KAAA;CACF;AAED,SAAS,cAAc,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEjD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;;AAGnB,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;QACvB,MAAM,SAAS,GAAG,IAAI;QACtB,IAAI,QAAQ,GAAG,CAAC;AAEhB,QAAA,OAAO,QAAQ,GAAG,IAAI,CAAC,MAAM,EAAE;AAC7B,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,CAAC;AAC5D,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;YAEnB,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;YAC/B,IAAI,KAAK,EAAE;AACT,gBAAA,OAAO,QAAQ,GAAG,KAAK,CAAC,KAAK;;;AAI/B,YAAA,QAAQ,IAAI,SAAS,GAAG,GAAG;AAC3B,YAAA,IAAI,QAAQ,IAAI,IAAI,CAAC,MAAM;gBAAE;;QAE/B,OAAO,EAAE;;;IAIX,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;AAC9B,IAAA,OAAO,KAAK,GAAG,KAAK,CAAC,KAAK,GAAG,EAAE;AACjC;AAEA;AACA,SAAS,aAAa,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEhD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,IAAA,IAAI,SAAS,GAAG,EAAE;IAClB,IAAI,UAAU,GAAG,CAAC;AAClB,IAAA,IAAI,KAAK;;AAGT,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;;QAEvB,MAAM,SAAS,GAAG,IAAI;AACtB,QAAA,IAAI,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC;AAExD,QAAA,OAAO,aAAa,IAAI,CAAC,EAAE;AACzB,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,aAAa,EAAE,aAAa,GAAG,SAAS,CAAC;AACtE,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,YAAA,IAAI,cAAc,GAAG,EAAE;YACvB,IAAI,eAAe,GAAG,CAAC;AAEvB,YAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE;AAC3C,gBAAA,cAAc,GAAG,KAAK,CAAC,KAAK;AAC5B,gBAAA,eAAe,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAGnC,YAAA,IAAI,cAAc,KAAK,EAAE,EAAE;AACzB,gBAAA,OAAO,aAAa,GAAG,cAAc,GAAG,eAAe;;;AAIzD,YAAA,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,GAAG,SAAS,GAAG,GAAG,CAAC,GAAG,CAAC;YAChE,IAAI,aAAa,IAAI,CAAC;gBAAE;;QAE1B,OAAO,EAAE;;;AAIX,IAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,EAAE;AAC1C,QAAA,SAAS,GAAG,KAAK,CAAC,KAAK;AACvB,QAAA,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAG9B,IAAA,OAAO,SAAS,KAAK,EAAE,GAAG,EAAE,GAAG,SAAS,GAAG,UAAU;AACvD;AAEA;AACA,SAAS,gBAAgB,CAAC,IAAY,EAAE,SAAS,GAAG,UAAU,EAAA;AAC5D,IAAA,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;AAAE,QAAA,OAAO,CAAC;;AAGxC,IAAA,KAAK,MAAM,YAAY,IAAI,gBAAgB,EAAE;AAC3C,QAAA,KAAK,MAAM,OAAO,IAAI,YAAY,EAAE;AAClC,YAAA,MAAM,QAAQ,GACZ,SAAS,KAAK;kBACV,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK;kBACjC,cAAc,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK,CAAC;AAEzC,YAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;AACnB,gBAAA,OAAO,QAAQ;;;;;AAMrB,IAAA,OAAO,SAAS,KAAK,UAAU,GAAG,IAAI,CAAC,MAAM,GAAG,CAAC;AACnD;AAEA;;AAEG;AACH,SAAS,0BAA0B,CACjC,IAAY,EACZ,YAA2B;;;IAS3B,MAAM,UAAU,GAIV,EAAE;AAER,IAAA,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE;AACrD,QAAA,OAAO,EAAE,UAAU,EAAE,CAAC;;;AAIxB,IAAA,IACE,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC;AACvB,QAAA,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC;AACxB,QAAA,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EACxB;QACA,OAAO,EAAE,UAAU,EAAE;;;AAIvB,IAAA,MAAM,UAAU,GAAG,YAAY,CAAC,UAAU,IAAI;AAC5C,QAAA,KAAK,EAAE,EAAE;AACT,QAAA,MAAM,EAAE,EAAE;AACV,QAAA,MAAM,EAAE,EAAE;KACX;;IAGD,MAAM,QAAQ,GAAG,gDAAgD;AACjE,IAAA,IAAI,KAAK;AAET,IAAA,OAAO,CAAC,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,EAAE;QAC7C,MAAM,GAAG,IAAI,EAAE,QAAQ,CAAC,GAAG,KAAK;AAChC,QAAA,MAAM,aAAa,GAAG,QAAQ,CAAC,QAAQ,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC;;QAGjD,MAAM,OAAO,GAAG,IAAkC;QAClD,MAAM,WAAW,GAAG,UAAU,CAAC,GAAG,OAAO,CAAA,CAAA,CAAG,CAE/B;;AAGb,QAAA,IACE,CAAC,WAAW;AACZ,YAAA,aAAa,GAAG,CAAC;AACjB,YAAA,aAAa,IAAI,WAAW,CAAC,MAAM,EACnC;AACA,YAAA,SAAS;;;AAIX,QAAA,MAAM,SAAS,GAAG,WAAW,CAAC,aAAa,CAAC;;QAG5C,MAAM,cAAc,GAAG,UAAU,CAAC,IAAI,CACpC,CAAC,GAAG,KAAK,GAAG,CAAC,IAAI,KAAK,OAAO,IAAI,GAAG,CAAC,aAAa,KAAK,aAAa,CACrE;QAED,IAAI,CAAC,cAAc,EAAE;YACnB,UAAU,CAAC,IAAI,CAAC;AACd,gBAAA,IAAI,EAAE,OAAO;gBACb,aAAa;gBACb,SAAS;AACV,aAAA,CAAC;;;IAIN,OAAO,EAAE,UAAU,EAAE;AACvB;AAEA;;;;;;;;;;AAUG;AACG,SAAU,gBAAgB,CAC9B,aAAiC,EACjC,YAAY,GAAG,GAAG,EAClB,iBAAiB,GAAG,GAAG,EAAA;;AAGvB,IAAA,MAAM,UAAU,GAAG,EAAE,GAAG,aAAa,EAAE;IACvC,IAAI,UAAU,CAAC,OAAO;QAAE,UAAU,CAAC,OAAO,GAAG,CAAC,GAAG,UAAU,CAAC,OAAO,CAAC;IACpE,IAAI,UAAU,CAAC,UAAU;QAAE,UAAU,CAAC,UAAU,GAAG,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC;;AAG7E,IAAA,MAAM,kBAAkB,GAAG,CAAC,SAAS,EAAE,YAAY,CAAU;AAE7D,IAAA,KAAK,MAAM,UAAU,IAAI,kBAAkB,EAAE;AAC3C,QAAA,IAAI,CAAC,UAAU,CAAC,UAAsC,CAAC;YAAE;;AAGzD,QAAA,UAAU,CAAC,UAAU,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,MAAM,KAAI;AAC9D,YAAA,IACE,MAAM,CAAC,OAAO,IAAI,IAAI;gBACtB,MAAM,CAAC,OAAO,KAAK,EAAE;gBACrB,CAAC,MAAM,CAAC,UAAU;AAClB,gBAAA,MAAM,CAAC,UAAU,CAAC,MAAM,KAAK,CAAC,EAC9B;gBACA,OAAO,MAAM,CAAC;;;AAIhB,YAAA,MAAM,UAAU,GAAG,EAAE,GAAG,MAAM,EAAE;AAChC,YAAA,MAAM,OAAO,GAAG,MAAM,CAAC,OAAO;YAC9B,MAAM,UAAU,GAAG,EAAE;;AAErB,YAAA,KAAK,MAAM,SAAS,IAAI,MAAM,CAAC,UAAU,EAAE;AACzC,gBAAA,MAAM,EAAE,UAAU,EAAE,GAAG,0BAA0B,CAC/C,SAAS,CAAC,IAAI,EACd,MAAM,CACP;gBAED,IAAI,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,SAAS,CAAC,IAAI,CAAC;AAC9C,gBAAA,IAAI,YAAY,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM;AAExC,gBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;;oBAEnB,MAAM,iBAAiB,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,EAAE;AAC/C,oBAAA,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,iBAAiB,CAAC;AAE7C,oBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;wBACnB,UAAU,CAAC,IAAI,CAAC;4BACd,IAAI,EAAE,SAAS,CAAC,IAAI;4BACpB,KAAK,EAAE,SAAS,CAAC,KAAK;4BACtB,UAAU;AACX,yBAAA,CAAC;wBACF;;AAEF,oBAAA,YAAY,GAAG,iBAAiB,CAAC,MAAM;;;AAIzC,gBAAA,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,GAAG,YAAY,CAAC;AACtD,gBAAA,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CACtB,OAAO,CAAC,MAAM,EACd,QAAQ,GAAG,YAAY,GAAG,YAAY,CACvC;AAED,gBAAA,MAAM,cAAc,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,SAAS,GAAG,iBAAiB,CAAC;AACjE,gBAAA,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAC3B,OAAO,CAAC,MAAM,EACd,OAAO,GAAG,iBAAiB,CAC5B;;gBAGD,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,cAAc,EAAE,SAAS,CAAC;gBAC7D,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,OAAO,EAAE,YAAY,CAAC;;gBAGzD,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,UAAU,CAAC;gBAC/D,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,SAAS,CAAC;;AAG9D,gBAAA,MAAM,UAAU,GAAG,cAAc,GAAG,gBAAgB;AACpD,gBAAA,MAAM,QAAQ,GAAG,OAAO,GAAG,gBAAgB;;gBAG3C,MAAM,qBAAqB,GAAG;AAC3B,qBAAA,SAAS,CAAC,UAAU,EAAE,QAAQ;AAC9B,qBAAA,IAAI,EAAE;gBACT,UAAU,CAAC,IAAI,CAAC;AACd,oBAAA,IAAI,EAAE,qBAAqB;oBAC3B,KAAK,EAAE,SAAS,CAAC,KAAK;oBACtB,UAAU;AACX,iBAAA,CAAC;;AAGJ,YAAA,UAAU,CAAC,UAAU,GAAG,UAAU;YAClC,OAAO,UAAU,CAAC,OAAO;YACzB,OAAO,UAAU,CAAC,UAAU;AAC5B,YAAA,OAAO,UAAU;AACnB,SAAC,CAAC;;AAGJ,IAAA,OAAO,UAAU;AACnB;;;;"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"rerankers.mjs","sources":["../../../../src/tools/search/rerankers.ts"],"sourcesContent":["import axios from 'axios';\r\nimport type * as t from './types';\r\nimport { createDefaultLogger } from './utils';\r\n\r\nexport abstract class BaseReranker {\r\n protected apiKey: string | undefined;\r\n protected logger: t.Logger;\r\n\r\n constructor(logger?: t.Logger) {\r\n // Each specific reranker will set its API key\r\n this.logger = logger || createDefaultLogger();\r\n }\r\n\r\n abstract rerank(\r\n query: string,\r\n documents: string[],\r\n topK?: number\r\n ): Promise<t.Highlight[]>;\r\n\r\n protected getDefaultRanking(\r\n documents: string[],\r\n topK: number\r\n ): t.Highlight[] {\r\n return documents\r\n .slice(0, Math.min(topK, documents.length))\r\n .map((doc) => ({ text: doc, score: 0 }));\r\n }\r\n}\r\n\r\nexport class JinaReranker extends BaseReranker {\r\n private apiUrl: string;\r\n\r\n constructor({\r\n apiKey = process.env.JINA_API_KEY,\r\n apiUrl = process.env.JINA_API_URL || 'https://api.jina.ai/v1/rerank',\r\n logger,\r\n }: {\r\n apiKey?: string;\r\n apiUrl?: string;\r\n logger?: t.Logger;\r\n }) {\r\n super(logger);\r\n this.apiKey = apiKey;\r\n this.apiUrl = apiUrl;\r\n }\r\n\r\n async rerank(\r\n query: string,\r\n documents: string[],\r\n topK: number = 5\r\n ): Promise<t.Highlight[]> {\r\n this.logger.debug(`Reranking ${documents.length} chunks with Jina using API URL: ${this.apiUrl}`);\r\n\r\n try {\r\n if (this.apiKey == null || this.apiKey === '') {\r\n this.logger.warn('JINA_API_KEY is not set. Using default ranking.');\r\n return this.getDefaultRanking(documents, topK);\r\n }\r\n\r\n const requestData = {\r\n model: 'jina-reranker-v2-base-multilingual',\r\n query: query,\r\n top_n: topK,\r\n documents: documents,\r\n return_documents: true,\r\n };\r\n\r\n const response = await axios.post<t.JinaRerankerResponse | undefined>(\r\n this.apiUrl,\r\n requestData,\r\n {\r\n headers: {\r\n 'Content-Type': 'application/json',\r\n Authorization: `Bearer ${this.apiKey}`,\r\n },\r\n }\r\n );\r\n\r\n this.logger.debug('Jina API Model:', response.data?.model);\r\n this.logger.debug('Jina API Usage:', response.data?.usage);\r\n\r\n if (response.data && response.data.results.length) {\r\n return response.data.results.map((result) => {\r\n const docIndex = result.index;\r\n const score = result.relevance_score;\r\n let text = '';\r\n\r\n // If return_documents is true, the document field will be present\r\n if (result.document != null) {\r\n const doc = result.document;\r\n if (typeof doc === 'object' && 'text' in doc) {\r\n text = doc.text;\r\n } else if (typeof doc === 'string') {\r\n text = doc;\r\n }\r\n } else {\r\n // Otherwise, use the index to get the document\r\n text = documents[docIndex];\r\n }\r\n\r\n return { text, score };\r\n });\r\n } else {\r\n this.logger.warn(\r\n 'Unexpected response format from Jina API. Using default ranking.'\r\n );\r\n return this.getDefaultRanking(documents, topK);\r\n }\r\n } catch (error) {\r\n this.logger.error('Error using Jina reranker:', error);\r\n // Fallback to default ranking on error\r\n return this.getDefaultRanking(documents, topK);\r\n }\r\n }\r\n}\r\n\r\nexport class CohereReranker extends BaseReranker {\r\n constructor({\r\n apiKey = process.env.COHERE_API_KEY,\r\n logger,\r\n }: {\r\n apiKey?: string;\r\n logger?: t.Logger;\r\n }) {\r\n super(logger);\r\n this.apiKey = apiKey;\r\n }\r\n\r\n async rerank(\r\n query: string,\r\n documents: string[],\r\n topK: number = 5\r\n ): Promise<t.Highlight[]> {\r\n this.logger.debug(`Reranking ${documents.length} chunks with Cohere`);\r\n\r\n try {\r\n if (this.apiKey == null || this.apiKey === '') {\r\n this.logger.warn('COHERE_API_KEY is not set. Using default ranking.');\r\n return this.getDefaultRanking(documents, topK);\r\n }\r\n\r\n const requestData = {\r\n model: 'rerank-v3.5',\r\n query: query,\r\n top_n: topK,\r\n documents: documents,\r\n };\r\n\r\n const response = await axios.post<t.CohereRerankerResponse | undefined>(\r\n 'https://api.cohere.com/v2/rerank',\r\n requestData,\r\n {\r\n headers: {\r\n 'Content-Type': 'application/json',\r\n Authorization: `Bearer ${this.apiKey}`,\r\n },\r\n }\r\n );\r\n\r\n this.logger.debug('Cohere API ID:', response.data?.id);\r\n this.logger.debug('Cohere API Meta:', response.data?.meta);\r\n\r\n if (response.data && response.data.results.length) {\r\n return response.data.results.map((result) => {\r\n const docIndex = result.index;\r\n const score = result.relevance_score;\r\n const text = documents[docIndex];\r\n return { text, score };\r\n });\r\n } else {\r\n this.logger.warn(\r\n 'Unexpected response format from Cohere API. Using default ranking.'\r\n );\r\n return this.getDefaultRanking(documents, topK);\r\n }\r\n } catch (error) {\r\n this.logger.error('Error using Cohere reranker:', error);\r\n // Fallback to default ranking on error\r\n return this.getDefaultRanking(documents, topK);\r\n }\r\n }\r\n}\r\n\r\nexport class InfinityReranker extends BaseReranker {\r\n constructor(logger?: t.Logger) {\r\n super(logger);\r\n // No API key needed for the placeholder implementation\r\n }\r\n\r\n async rerank(\r\n query: string,\r\n documents: string[],\r\n topK: number = 5\r\n ): Promise<t.Highlight[]> {\r\n this.logger.debug(\r\n `Reranking ${documents.length} chunks with Infinity (placeholder)`\r\n );\r\n // This would be replaced with actual Infinity reranker implementation\r\n return this.getDefaultRanking(documents, topK);\r\n }\r\n}\r\n\r\n/**\r\n * Creates the appropriate reranker based on type and configuration\r\n */\r\nexport const createReranker = (config: {\r\n rerankerType: t.RerankerType;\r\n jinaApiKey?: string;\r\n jinaApiUrl?: string;\r\n cohereApiKey?: string;\r\n logger?: t.Logger;\r\n}): BaseReranker | undefined => {\r\n const { rerankerType, jinaApiKey, jinaApiUrl, cohereApiKey, logger } = config;\r\n\r\n // Create a default logger if none is provided\r\n const defaultLogger = logger || createDefaultLogger();\r\n\r\n switch (rerankerType.toLowerCase()) {\r\n case 'jina':\r\n return new JinaReranker({ apiKey: jinaApiKey, apiUrl: jinaApiUrl, logger: defaultLogger });\r\n case 'cohere':\r\n return new CohereReranker({\r\n apiKey: cohereApiKey,\r\n logger: defaultLogger,\r\n });\r\n case 'infinity':\r\n return new InfinityReranker(defaultLogger);\r\n case 'none':\r\n defaultLogger.debug('Skipping reranking as reranker is set to \"none\"');\r\n return undefined;\r\n default:\r\n defaultLogger.warn(\r\n `Unknown reranker type: ${rerankerType}. Defaulting to InfinityReranker.`\r\n );\r\n return new JinaReranker({ apiKey: jinaApiKey, apiUrl: jinaApiUrl, logger: defaultLogger });\r\n }\r\n};\r\n\r\n// Example usage:\r\n// const jinaReranker = new JinaReranker();\r\n// const cohereReranker = new CohereReranker();\r\n// const infinityReranker = new InfinityReranker();\r\n"],"names":[],"mappings":";;;MAIsB,YAAY,CAAA;AACtB,IAAA,MAAM;AACN,IAAA,MAAM;AAEhB,IAAA,WAAA,CAAY,MAAiB,EAAA;;AAE3B,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,IAAI,mBAAmB,EAAE;;IASrC,iBAAiB,CACzB,SAAmB,EACnB,IAAY,EAAA;AAEZ,QAAA,OAAO;AACJ,aAAA,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,MAAM,CAAC;AACzC,aAAA,GAAG,CAAC,CAAC,GAAG,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;;AAE7C;AAEK,MAAO,YAAa,SAAQ,YAAY,CAAA;AACpC,IAAA,MAAM;IAEd,WAAY,CAAA,EACV,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,YAAY,EACjC,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,YAAY,IAAI,+BAA+B,EACpE,MAAM,GAKP,EAAA;QACC,KAAK,CAAC,MAAM,CAAC;AACb,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;AACpB,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;AAEhB,QAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,oCAAoC,IAAI,CAAC,MAAM,CAAA,CAAE,CAAC;AAEjG,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,iDAAiD,CAAC;gBACnE,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,oCAAoC;AAC3C,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;AACpB,gBAAA,gBAAgB,EAAE,IAAI;aACvB;AAED,YAAA,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,IAAI,CAAC,MAAM,EACX,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;AAED,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,iBAAiB,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;AAC1D,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,iBAAiB,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;AAE1D,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;oBACpC,IAAI,IAAI,GAAG,EAAE;;AAGb,oBAAA,IAAI,MAAM,CAAC,QAAQ,IAAI,IAAI,EAAE;AAC3B,wBAAA,MAAM,GAAG,GAAG,MAAM,CAAC,QAAQ;wBAC3B,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,MAAM,IAAI,GAAG,EAAE;AAC5C,4BAAA,IAAI,GAAG,GAAG,CAAC,IAAI;;AACV,6BAAA,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE;4BAClC,IAAI,GAAG,GAAG;;;yBAEP;;AAEL,wBAAA,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;;AAG5B,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,kEAAkE,CACnE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;YACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,4BAA4B,EAAE,KAAK,CAAC;;YAEtD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,cAAe,SAAQ,YAAY,CAAA;IAC9C,WAAY,CAAA,EACV,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,cAAc,EACnC,MAAM,GAIP,EAAA;QACC,KAAK,CAAC,MAAM,CAAC;AACb,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,CAAqB,mBAAA,CAAA,CAAC;AAErE,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,mDAAmD,CAAC;gBACrE,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,aAAa;AACpB,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;aACrB;YAED,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,kCAAkC,EAClC,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;AAED,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,gBAAgB,EAAE,QAAQ,CAAC,IAAI,EAAE,EAAE,CAAC;AACtD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,kBAAkB,EAAE,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC;AAE1D,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;AACpC,oBAAA,MAAM,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;AAChC,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,oEAAoE,CACrE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;YACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,8BAA8B,EAAE,KAAK,CAAC;;YAExD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,gBAAiB,SAAQ,YAAY,CAAA;AAChD,IAAA,WAAA,CAAY,MAAiB,EAAA;QAC3B,KAAK,CAAC,MAAM,CAAC;;;IAIf,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,CAAqC,mCAAA,CAAA,CACnE;;QAED,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAEjD;AAED;;AAEG;AACU,MAAA,cAAc,GAAG,CAAC,MAM9B,KAA8B;AAC7B,IAAA,MAAM,EAAE,YAAY,EAAE,UAAU,EAAE,UAAU,EAAE,YAAY,EAAE,MAAM,EAAE,GAAG,MAAM;;AAG7E,IAAA,MAAM,aAAa,GAAG,MAAM,IAAI,mBAAmB,EAAE;AAErD,IAAA,QAAQ,YAAY,CAAC,WAAW,EAAE;AAClC,QAAA,KAAK,MAAM;AACT,YAAA,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;AAC5F,QAAA,KAAK,QAAQ;YACX,OAAO,IAAI,cAAc,CAAC;AACxB,gBAAA,MAAM,EAAE,YAAY;AACpB,gBAAA,MAAM,EAAE,aAAa;AACtB,aAAA,CAAC;AACJ,QAAA,KAAK,UAAU;AACb,YAAA,OAAO,IAAI,gBAAgB,CAAC,aAAa,CAAC;AAC5C,QAAA,KAAK,MAAM;AACT,YAAA,aAAa,CAAC,KAAK,CAAC,iDAAiD,CAAC;AACtE,YAAA,OAAO,SAAS;AAClB,QAAA;AACE,YAAA,aAAa,CAAC,IAAI,CAChB,0BAA0B,YAAY,CAAA,iCAAA,CAAmC,CAC1E;AACD,YAAA,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;;AAE9F;AAEA;AACA;AACA;AACA;;;;"}
|
|
1
|
+
{"version":3,"file":"rerankers.mjs","sources":["../../../../src/tools/search/rerankers.ts"],"sourcesContent":["import axios from 'axios';\nimport type * as t from './types';\nimport { createDefaultLogger } from './utils';\n\nexport abstract class BaseReranker {\n protected apiKey: string | undefined;\n protected logger: t.Logger;\n\n constructor(logger?: t.Logger) {\n // Each specific reranker will set its API key\n this.logger = logger || createDefaultLogger();\n }\n\n abstract rerank(\n query: string,\n documents: string[],\n topK?: number\n ): Promise<t.Highlight[]>;\n\n protected getDefaultRanking(\n documents: string[],\n topK: number\n ): t.Highlight[] {\n return documents\n .slice(0, Math.min(topK, documents.length))\n .map((doc) => ({ text: doc, score: 0 }));\n }\n}\n\nexport class JinaReranker extends BaseReranker {\n private apiUrl: string;\n\n constructor({\n apiKey = process.env.JINA_API_KEY,\n apiUrl = process.env.JINA_API_URL || 'https://api.jina.ai/v1/rerank',\n logger,\n }: {\n apiKey?: string;\n apiUrl?: string;\n logger?: t.Logger;\n }) {\n super(logger);\n this.apiKey = apiKey;\n this.apiUrl = apiUrl;\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n this.logger.debug(`Reranking ${documents.length} chunks with Jina using API URL: ${this.apiUrl}`);\n\n try {\n if (this.apiKey == null || this.apiKey === '') {\n this.logger.warn('JINA_API_KEY is not set. Using default ranking.');\n return this.getDefaultRanking(documents, topK);\n }\n\n const requestData = {\n model: 'jina-reranker-v2-base-multilingual',\n query: query,\n top_n: topK,\n documents: documents,\n return_documents: true,\n };\n\n const response = await axios.post<t.JinaRerankerResponse | undefined>(\n this.apiUrl,\n requestData,\n {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n }\n );\n\n this.logger.debug('Jina API Model:', response.data?.model);\n this.logger.debug('Jina API Usage:', response.data?.usage);\n\n if (response.data && response.data.results.length) {\n return response.data.results.map((result) => {\n const docIndex = result.index;\n const score = result.relevance_score;\n let text = '';\n\n // If return_documents is true, the document field will be present\n if (result.document != null) {\n const doc = result.document;\n if (typeof doc === 'object' && 'text' in doc) {\n text = doc.text;\n } else if (typeof doc === 'string') {\n text = doc;\n }\n } else {\n // Otherwise, use the index to get the document\n text = documents[docIndex];\n }\n\n return { text, score };\n });\n } else {\n this.logger.warn(\n 'Unexpected response format from Jina API. Using default ranking.'\n );\n return this.getDefaultRanking(documents, topK);\n }\n } catch (error) {\n this.logger.error('Error using Jina reranker:', error);\n // Fallback to default ranking on error\n return this.getDefaultRanking(documents, topK);\n }\n }\n}\n\nexport class CohereReranker extends BaseReranker {\n constructor({\n apiKey = process.env.COHERE_API_KEY,\n logger,\n }: {\n apiKey?: string;\n logger?: t.Logger;\n }) {\n super(logger);\n this.apiKey = apiKey;\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n this.logger.debug(`Reranking ${documents.length} chunks with Cohere`);\n\n try {\n if (this.apiKey == null || this.apiKey === '') {\n this.logger.warn('COHERE_API_KEY is not set. Using default ranking.');\n return this.getDefaultRanking(documents, topK);\n }\n\n const requestData = {\n model: 'rerank-v3.5',\n query: query,\n top_n: topK,\n documents: documents,\n };\n\n const response = await axios.post<t.CohereRerankerResponse | undefined>(\n 'https://api.cohere.com/v2/rerank',\n requestData,\n {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n }\n );\n\n this.logger.debug('Cohere API ID:', response.data?.id);\n this.logger.debug('Cohere API Meta:', response.data?.meta);\n\n if (response.data && response.data.results.length) {\n return response.data.results.map((result) => {\n const docIndex = result.index;\n const score = result.relevance_score;\n const text = documents[docIndex];\n return { text, score };\n });\n } else {\n this.logger.warn(\n 'Unexpected response format from Cohere API. Using default ranking.'\n );\n return this.getDefaultRanking(documents, topK);\n }\n } catch (error) {\n this.logger.error('Error using Cohere reranker:', error);\n // Fallback to default ranking on error\n return this.getDefaultRanking(documents, topK);\n }\n }\n}\n\nexport class InfinityReranker extends BaseReranker {\n constructor(logger?: t.Logger) {\n super(logger);\n // No API key needed for the placeholder implementation\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n this.logger.debug(\n `Reranking ${documents.length} chunks with Infinity (placeholder)`\n );\n // This would be replaced with actual Infinity reranker implementation\n return this.getDefaultRanking(documents, topK);\n }\n}\n\n/**\n * Creates the appropriate reranker based on type and configuration\n */\nexport const createReranker = (config: {\n rerankerType: t.RerankerType;\n jinaApiKey?: string;\n jinaApiUrl?: string;\n cohereApiKey?: string;\n logger?: t.Logger;\n}): BaseReranker | undefined => {\n const { rerankerType, jinaApiKey, jinaApiUrl, cohereApiKey, logger } = config;\n\n // Create a default logger if none is provided\n const defaultLogger = logger || createDefaultLogger();\n\n switch (rerankerType.toLowerCase()) {\n case 'jina':\n return new JinaReranker({ apiKey: jinaApiKey, apiUrl: jinaApiUrl, logger: defaultLogger });\n case 'cohere':\n return new CohereReranker({\n apiKey: cohereApiKey,\n logger: defaultLogger,\n });\n case 'infinity':\n return new InfinityReranker(defaultLogger);\n case 'none':\n defaultLogger.debug('Skipping reranking as reranker is set to \"none\"');\n return undefined;\n default:\n defaultLogger.warn(\n `Unknown reranker type: ${rerankerType}. Defaulting to InfinityReranker.`\n );\n return new JinaReranker({ apiKey: jinaApiKey, apiUrl: jinaApiUrl, logger: defaultLogger });\n }\n};\n\n// Example usage:\n// const jinaReranker = new JinaReranker();\n// const cohereReranker = new CohereReranker();\n// const infinityReranker = new InfinityReranker();\n"],"names":[],"mappings":";;;MAIsB,YAAY,CAAA;AACtB,IAAA,MAAM;AACN,IAAA,MAAM;AAEhB,IAAA,WAAA,CAAY,MAAiB,EAAA;;AAE3B,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,IAAI,mBAAmB,EAAE;;IASrC,iBAAiB,CACzB,SAAmB,EACnB,IAAY,EAAA;AAEZ,QAAA,OAAO;AACJ,aAAA,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,MAAM,CAAC;AACzC,aAAA,GAAG,CAAC,CAAC,GAAG,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;;AAE7C;AAEK,MAAO,YAAa,SAAQ,YAAY,CAAA;AACpC,IAAA,MAAM;IAEd,WAAY,CAAA,EACV,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,YAAY,EACjC,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,YAAY,IAAI,+BAA+B,EACpE,MAAM,GAKP,EAAA;QACC,KAAK,CAAC,MAAM,CAAC;AACb,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;AACpB,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;AAEhB,QAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,oCAAoC,IAAI,CAAC,MAAM,CAAA,CAAE,CAAC;AAEjG,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,iDAAiD,CAAC;gBACnE,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,oCAAoC;AAC3C,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;AACpB,gBAAA,gBAAgB,EAAE,IAAI;aACvB;AAED,YAAA,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,IAAI,CAAC,MAAM,EACX,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;AAED,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,iBAAiB,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;AAC1D,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,iBAAiB,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;AAE1D,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;oBACpC,IAAI,IAAI,GAAG,EAAE;;AAGb,oBAAA,IAAI,MAAM,CAAC,QAAQ,IAAI,IAAI,EAAE;AAC3B,wBAAA,MAAM,GAAG,GAAG,MAAM,CAAC,QAAQ;wBAC3B,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,MAAM,IAAI,GAAG,EAAE;AAC5C,4BAAA,IAAI,GAAG,GAAG,CAAC,IAAI;;AACV,6BAAA,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE;4BAClC,IAAI,GAAG,GAAG;;;yBAEP;;AAEL,wBAAA,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;;AAG5B,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,kEAAkE,CACnE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;YACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,4BAA4B,EAAE,KAAK,CAAC;;YAEtD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,cAAe,SAAQ,YAAY,CAAA;IAC9C,WAAY,CAAA,EACV,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,cAAc,EACnC,MAAM,GAIP,EAAA;QACC,KAAK,CAAC,MAAM,CAAC;AACb,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,CAAqB,mBAAA,CAAA,CAAC;AAErE,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,mDAAmD,CAAC;gBACrE,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,aAAa;AACpB,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;aACrB;YAED,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,kCAAkC,EAClC,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;AAED,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,gBAAgB,EAAE,QAAQ,CAAC,IAAI,EAAE,EAAE,CAAC;AACtD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,kBAAkB,EAAE,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC;AAE1D,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;AACpC,oBAAA,MAAM,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;AAChC,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,oEAAoE,CACrE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;YACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,8BAA8B,EAAE,KAAK,CAAC;;YAExD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,gBAAiB,SAAQ,YAAY,CAAA;AAChD,IAAA,WAAA,CAAY,MAAiB,EAAA;QAC3B,KAAK,CAAC,MAAM,CAAC;;;IAIf,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,CAAqC,mCAAA,CAAA,CACnE;;QAED,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAEjD;AAED;;AAEG;AACU,MAAA,cAAc,GAAG,CAAC,MAM9B,KAA8B;AAC7B,IAAA,MAAM,EAAE,YAAY,EAAE,UAAU,EAAE,UAAU,EAAE,YAAY,EAAE,MAAM,EAAE,GAAG,MAAM;;AAG7E,IAAA,MAAM,aAAa,GAAG,MAAM,IAAI,mBAAmB,EAAE;AAErD,IAAA,QAAQ,YAAY,CAAC,WAAW,EAAE;AAClC,QAAA,KAAK,MAAM;AACT,YAAA,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;AAC5F,QAAA,KAAK,QAAQ;YACX,OAAO,IAAI,cAAc,CAAC;AACxB,gBAAA,MAAM,EAAE,YAAY;AACpB,gBAAA,MAAM,EAAE,aAAa;AACtB,aAAA,CAAC;AACJ,QAAA,KAAK,UAAU;AACb,YAAA,OAAO,IAAI,gBAAgB,CAAC,aAAa,CAAC;AAC5C,QAAA,KAAK,MAAM;AACT,YAAA,aAAa,CAAC,KAAK,CAAC,iDAAiD,CAAC;AACtE,YAAA,OAAO,SAAS;AAClB,QAAA;AACE,YAAA,aAAa,CAAC,IAAI,CAChB,0BAA0B,YAAY,CAAA,iCAAA,CAAmC,CAC1E;AACD,YAAA,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;;AAE9F;AAEA;AACA;AACA;AACA;;;;"}
|
|
@@ -8,32 +8,32 @@ var DATE_RANGE;
|
|
|
8
8
|
DATE_RANGE["PAST_MONTH"] = "m";
|
|
9
9
|
DATE_RANGE["PAST_YEAR"] = "y";
|
|
10
10
|
})(DATE_RANGE || (DATE_RANGE = {}));
|
|
11
|
-
const DEFAULT_QUERY_DESCRIPTION = `
|
|
12
|
-
GUIDELINES:
|
|
13
|
-
- Start broad, then narrow: Begin with key concepts, then refine with specifics
|
|
14
|
-
- Think like sources: Use terminology experts would use in the field
|
|
15
|
-
- Consider perspective: Frame queries from different viewpoints for better results
|
|
16
|
-
- Quality over quantity: A precise 3-4 word query often beats lengthy sentences
|
|
17
|
-
|
|
18
|
-
TECHNIQUES (combine for power searches):
|
|
19
|
-
- EXACT PHRASES: Use quotes ("climate change report")
|
|
20
|
-
- EXCLUDE TERMS: Use minus to remove unwanted results (-wikipedia)
|
|
21
|
-
- SITE-SPECIFIC: Restrict to websites (site:edu research)
|
|
22
|
-
- FILETYPE: Find specific documents (filetype:pdf study)
|
|
23
|
-
- OR OPERATOR: Find alternatives (electric OR hybrid cars)
|
|
24
|
-
- DATE RANGE: Recent information (data after:2020)
|
|
25
|
-
- WILDCARDS: Use * for unknown terms (how to * bread)
|
|
26
|
-
- SPECIFIC QUESTIONS: Use who/what/when/where/why/how
|
|
27
|
-
- DOMAIN TERMS: Include technical terminology for specialized topics
|
|
28
|
-
- CONCISE TERMS: Prioritize keywords over sentences
|
|
11
|
+
const DEFAULT_QUERY_DESCRIPTION = `
|
|
12
|
+
GUIDELINES:
|
|
13
|
+
- Start broad, then narrow: Begin with key concepts, then refine with specifics
|
|
14
|
+
- Think like sources: Use terminology experts would use in the field
|
|
15
|
+
- Consider perspective: Frame queries from different viewpoints for better results
|
|
16
|
+
- Quality over quantity: A precise 3-4 word query often beats lengthy sentences
|
|
17
|
+
|
|
18
|
+
TECHNIQUES (combine for power searches):
|
|
19
|
+
- EXACT PHRASES: Use quotes ("climate change report")
|
|
20
|
+
- EXCLUDE TERMS: Use minus to remove unwanted results (-wikipedia)
|
|
21
|
+
- SITE-SPECIFIC: Restrict to websites (site:edu research)
|
|
22
|
+
- FILETYPE: Find specific documents (filetype:pdf study)
|
|
23
|
+
- OR OPERATOR: Find alternatives (electric OR hybrid cars)
|
|
24
|
+
- DATE RANGE: Recent information (data after:2020)
|
|
25
|
+
- WILDCARDS: Use * for unknown terms (how to * bread)
|
|
26
|
+
- SPECIFIC QUESTIONS: Use who/what/when/where/why/how
|
|
27
|
+
- DOMAIN TERMS: Include technical terminology for specialized topics
|
|
28
|
+
- CONCISE TERMS: Prioritize keywords over sentences
|
|
29
29
|
`.trim();
|
|
30
|
-
const DEFAULT_COUNTRY_DESCRIPTION = `Country code to localize search results.
|
|
31
|
-
Use standard 2-letter country codes: "us", "uk", "ca", "de", "fr", "jp", "br", etc.
|
|
32
|
-
Provide this when the search should return results specific to a particular country.
|
|
33
|
-
Examples:
|
|
34
|
-
- "us" for United States (default)
|
|
35
|
-
- "de" for Germany
|
|
36
|
-
- "in" for India
|
|
30
|
+
const DEFAULT_COUNTRY_DESCRIPTION = `Country code to localize search results.
|
|
31
|
+
Use standard 2-letter country codes: "us", "uk", "ca", "de", "fr", "jp", "br", etc.
|
|
32
|
+
Provide this when the search should return results specific to a particular country.
|
|
33
|
+
Examples:
|
|
34
|
+
- "us" for United States (default)
|
|
35
|
+
- "de" for Germany
|
|
36
|
+
- "in" for India
|
|
37
37
|
`.trim();
|
|
38
38
|
const querySchema = z.string().describe(DEFAULT_QUERY_DESCRIPTION);
|
|
39
39
|
const dateSchema = z
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"schema.mjs","sources":["../../../../src/tools/search/schema.ts"],"sourcesContent":["import { z } from 'zod';\
|
|
1
|
+
{"version":3,"file":"schema.mjs","sources":["../../../../src/tools/search/schema.ts"],"sourcesContent":["import { z } from 'zod';\n\nexport enum DATE_RANGE {\n PAST_HOUR = 'h',\n PAST_24_HOURS = 'd',\n PAST_WEEK = 'w',\n PAST_MONTH = 'm',\n PAST_YEAR = 'y',\n}\n\nexport const DEFAULT_QUERY_DESCRIPTION = `\nGUIDELINES:\n- Start broad, then narrow: Begin with key concepts, then refine with specifics\n- Think like sources: Use terminology experts would use in the field\n- Consider perspective: Frame queries from different viewpoints for better results\n- Quality over quantity: A precise 3-4 word query often beats lengthy sentences\n\nTECHNIQUES (combine for power searches):\n- EXACT PHRASES: Use quotes (\"climate change report\")\n- EXCLUDE TERMS: Use minus to remove unwanted results (-wikipedia)\n- SITE-SPECIFIC: Restrict to websites (site:edu research)\n- FILETYPE: Find specific documents (filetype:pdf study)\n- OR OPERATOR: Find alternatives (electric OR hybrid cars)\n- DATE RANGE: Recent information (data after:2020)\n- WILDCARDS: Use * for unknown terms (how to * bread)\n- SPECIFIC QUESTIONS: Use who/what/when/where/why/how\n- DOMAIN TERMS: Include technical terminology for specialized topics\n- CONCISE TERMS: Prioritize keywords over sentences\n`.trim();\n\nexport const DEFAULT_COUNTRY_DESCRIPTION =\n `Country code to localize search results.\nUse standard 2-letter country codes: \"us\", \"uk\", \"ca\", \"de\", \"fr\", \"jp\", \"br\", etc.\nProvide this when the search should return results specific to a particular country.\nExamples:\n- \"us\" for United States (default)\n- \"de\" for Germany\n- \"in\" for India\n`.trim();\n\nexport const querySchema = z.string().describe(DEFAULT_QUERY_DESCRIPTION);\nexport const dateSchema = z\n .nativeEnum(DATE_RANGE)\n .optional()\n .describe('Date range for search results.');\nexport const countrySchema = z\n .string()\n .optional()\n .describe(DEFAULT_COUNTRY_DESCRIPTION);\nexport const imagesSchema = z\n .boolean()\n .optional()\n .describe('Whether to also run an image search.');\n\nexport const videosSchema = z\n .boolean()\n .optional()\n .describe('Whether to also run a video search.');\n\nexport const newsSchema = z\n .boolean()\n .optional()\n .describe('Whether to also run a news search.');\n"],"names":[],"mappings":";;IAEY;AAAZ,CAAA,UAAY,UAAU,EAAA;AACpB,IAAA,UAAA,CAAA,WAAA,CAAA,GAAA,GAAe;AACf,IAAA,UAAA,CAAA,eAAA,CAAA,GAAA,GAAmB;AACnB,IAAA,UAAA,CAAA,WAAA,CAAA,GAAA,GAAe;AACf,IAAA,UAAA,CAAA,YAAA,CAAA,GAAA,GAAgB;AAChB,IAAA,UAAA,CAAA,WAAA,CAAA,GAAA,GAAe;AACjB,CAAC,EANW,UAAU,KAAV,UAAU,GAMrB,EAAA,CAAA,CAAA;AAEY,MAAA,yBAAyB,GAAG;;;;;;;;;;;;;;;;;;CAkBxC,CAAC,IAAI;AAEO,MAAA,2BAA2B,GACtC,CAAA;;;;;;;CAOD,CAAC,IAAI;AAEC,MAAM,WAAW,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,yBAAyB;AACjE,MAAM,UAAU,GAAG;KACvB,UAAU,CAAC,UAAU;AACrB,KAAA,QAAQ;KACR,QAAQ,CAAC,gCAAgC;AACrC,MAAM,aAAa,GAAG;AAC1B,KAAA,MAAM;AACN,KAAA,QAAQ;KACR,QAAQ,CAAC,2BAA2B;AAChC,MAAM,YAAY,GAAG;AACzB,KAAA,OAAO;AACP,KAAA,QAAQ;KACR,QAAQ,CAAC,sCAAsC;AAE3C,MAAM,YAAY,GAAG;AACzB,KAAA,OAAO;AACP,KAAA,QAAQ;KACR,QAAQ,CAAC,qCAAqC;AAE1C,MAAM,UAAU,GAAG;AACvB,KAAA,OAAO;AACP,KAAA,QAAQ;KACR,QAAQ,CAAC,oCAAoC;;;;"}
|
|
@@ -424,7 +424,7 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
|
|
|
424
424
|
});
|
|
425
425
|
}
|
|
426
426
|
};
|
|
427
|
-
const processSources = async ({ result, numElements, query, news, proMode = true, onGetHighlights, }) => {
|
|
427
|
+
const processSources = async ({ result, numElements, query, news, proMode = true, onGetHighlights, skipScraping = false, }) => {
|
|
428
428
|
try {
|
|
429
429
|
if (!result.data) {
|
|
430
430
|
return {
|
|
@@ -437,6 +437,11 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
|
|
|
437
437
|
else if (!result.data.organic) {
|
|
438
438
|
return result.data;
|
|
439
439
|
}
|
|
440
|
+
// If content was already extracted directly (e.g., direct URL extraction), skip scraping
|
|
441
|
+
if (skipScraping) {
|
|
442
|
+
logger_.debug('Skipping additional scraping - content already extracted');
|
|
443
|
+
return result.data;
|
|
444
|
+
}
|
|
440
445
|
if (!proMode) {
|
|
441
446
|
const wikiSources = result.data.organic.filter((source) => source.link.includes('wikipedia.org'));
|
|
442
447
|
if (!wikiSources.length) {
|