@j0hanz/superfetch 2.1.4 → 2.1.5
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/dist/cache.d.ts +1 -0
- package/dist/cache.d.ts.map +1 -0
- package/dist/cache.js +1 -0
- package/dist/cache.js.map +1 -0
- package/dist/config.d.ts +1 -0
- package/dist/config.d.ts.map +1 -0
- package/dist/config.js +1 -0
- package/dist/config.js.map +1 -0
- package/dist/crypto.d.ts +1 -0
- package/dist/crypto.d.ts.map +1 -0
- package/dist/crypto.js +1 -0
- package/dist/crypto.js.map +1 -0
- package/dist/errors.d.ts +1 -0
- package/dist/errors.d.ts.map +1 -0
- package/dist/errors.js +1 -0
- package/dist/errors.js.map +1 -0
- package/dist/fetch.d.ts +1 -0
- package/dist/fetch.d.ts.map +1 -0
- package/dist/fetch.js +1 -0
- package/dist/fetch.js.map +1 -0
- package/dist/http.d.ts +1 -0
- package/dist/http.d.ts.map +1 -0
- package/dist/http.js +1 -0
- package/dist/http.js.map +1 -0
- package/dist/index.d.ts +1 -0
- package/dist/index.js +1 -0
- package/dist/index.js.map +1 -1
- package/dist/mcp.d.ts +1 -0
- package/dist/mcp.d.ts.map +1 -0
- package/dist/mcp.js +1 -0
- package/dist/mcp.js.map +1 -0
- package/dist/observability.d.ts +1 -0
- package/dist/observability.d.ts.map +1 -0
- package/dist/observability.js +1 -0
- package/dist/observability.js.map +1 -0
- package/dist/tools.d.ts +1 -0
- package/dist/tools.d.ts.map +1 -0
- package/dist/tools.js +1 -0
- package/dist/tools.js.map +1 -0
- package/dist/transform.d.ts +1 -0
- package/dist/transform.d.ts.map +1 -0
- package/dist/transform.js +1 -0
- package/dist/transform.js.map +1 -0
- package/dist/utils.d.ts +1 -0
- package/dist/utils.d.ts.map +1 -0
- package/dist/utils.js +1 -0
- package/dist/utils.js.map +1 -0
- package/dist/workers/transform-worker.d.ts +1 -0
- package/dist/workers/transform-worker.d.ts.map +1 -0
- package/dist/workers/transform-worker.js +1 -0
- package/dist/workers/transform-worker.js.map +1 -0
- package/package.json +5 -2
- package/dist/config/auth-config.d.ts +0 -16
- package/dist/config/auth-config.js +0 -53
- package/dist/config/constants.d.ts +0 -17
- package/dist/config/constants.d.ts.map +0 -1
- package/dist/config/constants.js +0 -22
- package/dist/config/constants.js.map +0 -1
- package/dist/config/env-parsers.d.ts +0 -8
- package/dist/config/env-parsers.js +0 -96
- package/dist/config/formatting.d.ts +0 -7
- package/dist/config/formatting.d.ts.map +0 -1
- package/dist/config/formatting.js +0 -9
- package/dist/config/formatting.js.map +0 -1
- package/dist/config/index.d.ts +0 -61
- package/dist/config/index.d.ts.map +0 -1
- package/dist/config/index.js +0 -107
- package/dist/config/index.js.map +0 -1
- package/dist/config/types/content.d.ts +0 -41
- package/dist/config/types/content.d.ts.map +0 -1
- package/dist/config/types/content.js +0 -1
- package/dist/config/types/content.js.map +0 -1
- package/dist/config/types/runtime.d.ts +0 -72
- package/dist/config/types/runtime.d.ts.map +0 -1
- package/dist/config/types/runtime.js +0 -1
- package/dist/config/types/runtime.js.map +0 -1
- package/dist/config/types/tools.d.ts +0 -28
- package/dist/config/types/tools.d.ts.map +0 -1
- package/dist/config/types/tools.js +0 -1
- package/dist/config/types/tools.js.map +0 -1
- package/dist/config/types.d.ts +0 -4
- package/dist/config/types.d.ts.map +0 -1
- package/dist/config/types.js +0 -2
- package/dist/config/types.js.map +0 -1
- package/dist/errors/app-error.d.ts +0 -7
- package/dist/errors/app-error.d.ts.map +0 -1
- package/dist/errors/app-error.js +0 -16
- package/dist/errors/app-error.js.map +0 -1
- package/dist/errors/index.d.ts +0 -2
- package/dist/errors/index.d.ts.map +0 -1
- package/dist/errors/index.js +0 -2
- package/dist/errors/index.js.map +0 -1
- package/dist/http/accept-policy.d.ts +0 -3
- package/dist/http/accept-policy.js +0 -45
- package/dist/http/async-handler.d.ts +0 -2
- package/dist/http/async-handler.js +0 -5
- package/dist/http/auth-introspection.d.ts +0 -2
- package/dist/http/auth-introspection.js +0 -141
- package/dist/http/auth-static.d.ts +0 -2
- package/dist/http/auth-static.js +0 -23
- package/dist/http/auth.d.ts +0 -3
- package/dist/http/auth.d.ts.map +0 -1
- package/dist/http/auth.js +0 -269
- package/dist/http/auth.js.map +0 -1
- package/dist/http/base-middleware.d.ts +0 -7
- package/dist/http/base-middleware.js +0 -143
- package/dist/http/cors.d.ts +0 -2
- package/dist/http/cors.d.ts.map +0 -1
- package/dist/http/cors.js +0 -9
- package/dist/http/cors.js.map +0 -1
- package/dist/http/download-routes.d.ts +0 -2
- package/dist/http/download-routes.d.ts.map +0 -1
- package/dist/http/download-routes.js +0 -104
- package/dist/http/download-routes.js.map +0 -1
- package/dist/http/error-handler.d.ts +0 -2
- package/dist/http/error-handler.js +0 -55
- package/dist/http/host-allowlist.d.ts +0 -3
- package/dist/http/host-allowlist.js +0 -117
- package/dist/http/jsonrpc-http.d.ts +0 -2
- package/dist/http/jsonrpc-http.js +0 -10
- package/dist/http/mcp-routes.d.ts +0 -9
- package/dist/http/mcp-routes.d.ts.map +0 -1
- package/dist/http/mcp-routes.js +0 -204
- package/dist/http/mcp-routes.js.map +0 -1
- package/dist/http/mcp-session-eviction.d.ts +0 -3
- package/dist/http/mcp-session-eviction.js +0 -24
- package/dist/http/mcp-session-helpers.d.ts +0 -12
- package/dist/http/mcp-session-helpers.d.ts.map +0 -1
- package/dist/http/mcp-session-helpers.js +0 -64
- package/dist/http/mcp-session-helpers.js.map +0 -1
- package/dist/http/mcp-session-init.d.ts +0 -7
- package/dist/http/mcp-session-init.js +0 -94
- package/dist/http/mcp-session-slots.d.ts +0 -17
- package/dist/http/mcp-session-slots.js +0 -55
- package/dist/http/mcp-session-transport-init.d.ts +0 -7
- package/dist/http/mcp-session-transport-init.js +0 -41
- package/dist/http/mcp-session-transport.d.ts +0 -7
- package/dist/http/mcp-session-transport.js +0 -57
- package/dist/http/mcp-session-types.d.ts +0 -5
- package/dist/http/mcp-session-types.js +0 -1
- package/dist/http/mcp-session.d.ts +0 -10
- package/dist/http/mcp-session.d.ts.map +0 -1
- package/dist/http/mcp-session.js +0 -23
- package/dist/http/mcp-session.js.map +0 -1
- package/dist/http/mcp-sessions.d.ts +0 -41
- package/dist/http/mcp-sessions.js +0 -392
- package/dist/http/mcp-validation.d.ts +0 -3
- package/dist/http/mcp-validation.d.ts.map +0 -1
- package/dist/http/mcp-validation.js +0 -14
- package/dist/http/mcp-validation.js.map +0 -1
- package/dist/http/protocol-policy.d.ts +0 -2
- package/dist/http/protocol-policy.js +0 -31
- package/dist/http/rate-limit.d.ts +0 -12
- package/dist/http/rate-limit.d.ts.map +0 -1
- package/dist/http/rate-limit.js +0 -93
- package/dist/http/rate-limit.js.map +0 -1
- package/dist/http/server-config.d.ts +0 -1
- package/dist/http/server-config.js +0 -40
- package/dist/http/server-middleware.d.ts +0 -7
- package/dist/http/server-middleware.d.ts.map +0 -1
- package/dist/http/server-middleware.js +0 -52
- package/dist/http/server-middleware.js.map +0 -1
- package/dist/http/server-shutdown.d.ts +0 -4
- package/dist/http/server-shutdown.js +0 -43
- package/dist/http/server-tuning.d.ts +0 -9
- package/dist/http/server-tuning.js +0 -45
- package/dist/http/server.d.ts +0 -3
- package/dist/http/server.d.ts.map +0 -1
- package/dist/http/server.js +0 -291
- package/dist/http/server.js.map +0 -1
- package/dist/http/session-cleanup.d.ts +0 -2
- package/dist/http/session-cleanup.d.ts.map +0 -1
- package/dist/http/session-cleanup.js +0 -40
- package/dist/http/session-cleanup.js.map +0 -1
- package/dist/http/sessions.d.ts +0 -14
- package/dist/http/sessions.d.ts.map +0 -1
- package/dist/http/sessions.js +0 -63
- package/dist/http/sessions.js.map +0 -1
- package/dist/middleware/error-handler.d.ts +0 -2
- package/dist/middleware/error-handler.d.ts.map +0 -1
- package/dist/middleware/error-handler.js +0 -56
- package/dist/middleware/error-handler.js.map +0 -1
- package/dist/middleware/rate-limiter.d.ts +0 -16
- package/dist/middleware/rate-limiter.d.ts.map +0 -1
- package/dist/middleware/rate-limiter.js +0 -111
- package/dist/middleware/rate-limiter.js.map +0 -1
- package/dist/parsers/base-html-element-parser.d.ts +0 -43
- package/dist/parsers/base-html-element-parser.d.ts.map +0 -1
- package/dist/parsers/base-html-element-parser.js +0 -59
- package/dist/parsers/base-html-element-parser.js.map +0 -1
- package/dist/parsers/heading-element-parser.d.ts +0 -14
- package/dist/parsers/heading-element-parser.d.ts.map +0 -1
- package/dist/parsers/heading-element-parser.js +0 -26
- package/dist/parsers/heading-element-parser.js.map +0 -1
- package/dist/parsers/image-element-parser.d.ts +0 -16
- package/dist/parsers/image-element-parser.d.ts.map +0 -1
- package/dist/parsers/image-element-parser.js +0 -33
- package/dist/parsers/image-element-parser.js.map +0 -1
- package/dist/parsers/link-element-parser.d.ts +0 -15
- package/dist/parsers/link-element-parser.d.ts.map +0 -1
- package/dist/parsers/link-element-parser.js +0 -28
- package/dist/parsers/link-element-parser.js.map +0 -1
- package/dist/parsers/open-graph-parser.d.ts +0 -17
- package/dist/parsers/open-graph-parser.d.ts.map +0 -1
- package/dist/parsers/open-graph-parser.js +0 -41
- package/dist/parsers/open-graph-parser.js.map +0 -1
- package/dist/parsers/schema-org-parser.d.ts +0 -17
- package/dist/parsers/schema-org-parser.d.ts.map +0 -1
- package/dist/parsers/schema-org-parser.js +0 -32
- package/dist/parsers/schema-org-parser.js.map +0 -1
- package/dist/parsers/standard-meta-parser.d.ts +0 -18
- package/dist/parsers/standard-meta-parser.d.ts.map +0 -1
- package/dist/parsers/standard-meta-parser.js +0 -32
- package/dist/parsers/standard-meta-parser.js.map +0 -1
- package/dist/parsers/twitter-card-parser.d.ts +0 -17
- package/dist/parsers/twitter-card-parser.d.ts.map +0 -1
- package/dist/parsers/twitter-card-parser.js +0 -41
- package/dist/parsers/twitter-card-parser.js.map +0 -1
- package/dist/prompts/index.d.ts +0 -3
- package/dist/prompts/index.d.ts.map +0 -1
- package/dist/prompts/index.js +0 -73
- package/dist/prompts/index.js.map +0 -1
- package/dist/resources/cached-content-params.d.ts +0 -5
- package/dist/resources/cached-content-params.js +0 -36
- package/dist/resources/cached-content.d.ts +0 -2
- package/dist/resources/cached-content.d.ts.map +0 -1
- package/dist/resources/cached-content.js +0 -132
- package/dist/resources/cached-content.js.map +0 -1
- package/dist/resources/index.d.ts +0 -2
- package/dist/resources/index.d.ts.map +0 -1
- package/dist/resources/index.js +0 -4
- package/dist/resources/index.js.map +0 -1
- package/dist/server.d.ts +0 -3
- package/dist/server.d.ts.map +0 -1
- package/dist/server.js +0 -94
- package/dist/server.js.map +0 -1
- package/dist/services/cache-events.d.ts +0 -8
- package/dist/services/cache-events.js +0 -19
- package/dist/services/cache-keys.d.ts +0 -7
- package/dist/services/cache-keys.js +0 -57
- package/dist/services/cache.d.ts +0 -17
- package/dist/services/cache.d.ts.map +0 -1
- package/dist/services/cache.js +0 -145
- package/dist/services/cache.js.map +0 -1
- package/dist/services/cache.service.d.ts +0 -52
- package/dist/services/cache.service.d.ts.map +0 -1
- package/dist/services/cache.service.js +0 -113
- package/dist/services/cache.service.js.map +0 -1
- package/dist/services/card-extractor.d.ts +0 -6
- package/dist/services/card-extractor.d.ts.map +0 -1
- package/dist/services/card-extractor.js +0 -199
- package/dist/services/card-extractor.js.map +0 -1
- package/dist/services/context.d.ts +0 -10
- package/dist/services/context.d.ts.map +0 -1
- package/dist/services/context.js +0 -14
- package/dist/services/context.js.map +0 -1
- package/dist/services/extractor.d.ts +0 -5
- package/dist/services/extractor.d.ts.map +0 -1
- package/dist/services/extractor.js +0 -142
- package/dist/services/extractor.js.map +0 -1
- package/dist/services/extractor.service.d.ts +0 -18
- package/dist/services/extractor.service.d.ts.map +0 -1
- package/dist/services/extractor.service.js +0 -75
- package/dist/services/extractor.service.js.map +0 -1
- package/dist/services/fetcher/agents.d.ts +0 -3
- package/dist/services/fetcher/agents.d.ts.map +0 -1
- package/dist/services/fetcher/agents.js +0 -100
- package/dist/services/fetcher/agents.js.map +0 -1
- package/dist/services/fetcher/dns-selection.d.ts +0 -2
- package/dist/services/fetcher/dns-selection.js +0 -72
- package/dist/services/fetcher/errors.d.ts +0 -4
- package/dist/services/fetcher/errors.d.ts.map +0 -1
- package/dist/services/fetcher/errors.js +0 -70
- package/dist/services/fetcher/errors.js.map +0 -1
- package/dist/services/fetcher/headers.d.ts +0 -2
- package/dist/services/fetcher/headers.d.ts.map +0 -1
- package/dist/services/fetcher/headers.js +0 -6
- package/dist/services/fetcher/headers.js.map +0 -1
- package/dist/services/fetcher/interceptors.d.ts +0 -10
- package/dist/services/fetcher/interceptors.d.ts.map +0 -1
- package/dist/services/fetcher/interceptors.js +0 -108
- package/dist/services/fetcher/interceptors.js.map +0 -1
- package/dist/services/fetcher/redirects.d.ts +0 -4
- package/dist/services/fetcher/redirects.d.ts.map +0 -1
- package/dist/services/fetcher/redirects.js +0 -78
- package/dist/services/fetcher/redirects.js.map +0 -1
- package/dist/services/fetcher/response.d.ts +0 -4
- package/dist/services/fetcher/response.d.ts.map +0 -1
- package/dist/services/fetcher/response.js +0 -104
- package/dist/services/fetcher/response.js.map +0 -1
- package/dist/services/fetcher/retry-policy.d.ts +0 -1
- package/dist/services/fetcher/retry-policy.d.ts.map +0 -1
- package/dist/services/fetcher/retry-policy.js +0 -131
- package/dist/services/fetcher/retry-policy.js.map +0 -1
- package/dist/services/fetcher.d.ts +0 -25
- package/dist/services/fetcher.d.ts.map +0 -1
- package/dist/services/fetcher.js +0 -607
- package/dist/services/fetcher.js.map +0 -1
- package/dist/services/fetcher.service.d.ts +0 -18
- package/dist/services/fetcher.service.d.ts.map +0 -1
- package/dist/services/fetcher.service.js +0 -122
- package/dist/services/fetcher.service.js.map +0 -1
- package/dist/services/fifo-queue.d.ts +0 -8
- package/dist/services/fifo-queue.js +0 -25
- package/dist/services/logger.d.ts +0 -5
- package/dist/services/logger.d.ts.map +0 -1
- package/dist/services/logger.js +0 -52
- package/dist/services/logger.js.map +0 -1
- package/dist/services/logger.service.d.ts +0 -5
- package/dist/services/logger.service.d.ts.map +0 -1
- package/dist/services/logger.service.js +0 -57
- package/dist/services/logger.service.js.map +0 -1
- package/dist/services/metadata-collector.d.ts +0 -2
- package/dist/services/metadata-collector.js +0 -80
- package/dist/services/parser.d.ts +0 -6
- package/dist/services/parser.d.ts.map +0 -1
- package/dist/services/parser.js +0 -278
- package/dist/services/parser.js.map +0 -1
- package/dist/services/parser.service.d.ts +0 -42
- package/dist/services/parser.service.d.ts.map +0 -1
- package/dist/services/parser.service.js +0 -209
- package/dist/services/parser.service.js.map +0 -1
- package/dist/services/session-manager.d.ts +0 -18
- package/dist/services/session-manager.d.ts.map +0 -1
- package/dist/services/session-manager.js +0 -73
- package/dist/services/session-manager.js.map +0 -1
- package/dist/services/telemetry.d.ts +0 -19
- package/dist/services/telemetry.js +0 -43
- package/dist/services/transform-worker-pool.d.ts +0 -11
- package/dist/services/transform-worker-pool.js +0 -244
- package/dist/services/transform-worker-types.d.ts +0 -32
- package/dist/services/transform-worker-types.js +0 -14
- package/dist/strategies/exponential-backoff-strategy.d.ts +0 -13
- package/dist/strategies/exponential-backoff-strategy.d.ts.map +0 -1
- package/dist/strategies/exponential-backoff-strategy.js +0 -32
- package/dist/strategies/exponential-backoff-strategy.js.map +0 -1
- package/dist/tools/handlers/fetch-links/link-extractor.d.ts +0 -4
- package/dist/tools/handlers/fetch-links/link-extractor.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-links/link-extractor.js +0 -159
- package/dist/tools/handlers/fetch-links/link-extractor.js.map +0 -1
- package/dist/tools/handlers/fetch-links.tool.d.ts +0 -5
- package/dist/tools/handlers/fetch-links.tool.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-links.tool.js +0 -98
- package/dist/tools/handlers/fetch-links.tool.js.map +0 -1
- package/dist/tools/handlers/fetch-markdown.tool.d.ts +0 -11
- package/dist/tools/handlers/fetch-markdown.tool.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-markdown.tool.js +0 -97
- package/dist/tools/handlers/fetch-markdown.tool.js.map +0 -1
- package/dist/tools/handlers/fetch-single.shared.d.ts +0 -31
- package/dist/tools/handlers/fetch-single.shared.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-single.shared.js +0 -223
- package/dist/tools/handlers/fetch-single.shared.js.map +0 -1
- package/dist/tools/handlers/fetch-url.tool.d.ts +0 -10
- package/dist/tools/handlers/fetch-url.tool.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-url.tool.js +0 -119
- package/dist/tools/handlers/fetch-url.tool.js.map +0 -1
- package/dist/tools/handlers/fetch-urls/processor.d.ts +0 -13
- package/dist/tools/handlers/fetch-urls/processor.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-urls/processor.js +0 -153
- package/dist/tools/handlers/fetch-urls/processor.js.map +0 -1
- package/dist/tools/handlers/fetch-urls/response.d.ts +0 -3
- package/dist/tools/handlers/fetch-urls/response.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-urls/response.js +0 -79
- package/dist/tools/handlers/fetch-urls/response.js.map +0 -1
- package/dist/tools/handlers/fetch-urls/validation.d.ts +0 -5
- package/dist/tools/handlers/fetch-urls/validation.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-urls/validation.js +0 -18
- package/dist/tools/handlers/fetch-urls/validation.js.map +0 -1
- package/dist/tools/handlers/fetch-urls.tool.d.ts +0 -5
- package/dist/tools/handlers/fetch-urls.tool.d.ts.map +0 -1
- package/dist/tools/handlers/fetch-urls.tool.js +0 -124
- package/dist/tools/handlers/fetch-urls.tool.js.map +0 -1
- package/dist/tools/index.d.ts +0 -3
- package/dist/tools/index.d.ts.map +0 -1
- package/dist/tools/index.js +0 -37
- package/dist/tools/index.js.map +0 -1
- package/dist/tools/schemas.d.ts +0 -12
- package/dist/tools/schemas.d.ts.map +0 -1
- package/dist/tools/schemas.js +0 -21
- package/dist/tools/schemas.js.map +0 -1
- package/dist/tools/utils/cache-vary.d.ts +0 -1
- package/dist/tools/utils/cache-vary.d.ts.map +0 -1
- package/dist/tools/utils/cache-vary.js +0 -12
- package/dist/tools/utils/cache-vary.js.map +0 -1
- package/dist/tools/utils/cached-markdown.d.ts +0 -5
- package/dist/tools/utils/cached-markdown.js +0 -46
- package/dist/tools/utils/common.d.ts +0 -5
- package/dist/tools/utils/common.d.ts.map +0 -1
- package/dist/tools/utils/common.js +0 -42
- package/dist/tools/utils/common.js.map +0 -1
- package/dist/tools/utils/content-shaping.d.ts +0 -4
- package/dist/tools/utils/content-shaping.js +0 -67
- package/dist/tools/utils/content-transform-async.d.ts +0 -6
- package/dist/tools/utils/content-transform-async.js +0 -33
- package/dist/tools/utils/content-transform-core.d.ts +0 -5
- package/dist/tools/utils/content-transform-core.js +0 -180
- package/dist/tools/utils/content-transform-workers.d.ts +0 -1
- package/dist/tools/utils/content-transform-workers.js +0 -1
- package/dist/tools/utils/content-transform.d.ts +0 -3
- package/dist/tools/utils/content-transform.d.ts.map +0 -1
- package/dist/tools/utils/content-transform.js +0 -41
- package/dist/tools/utils/content-transform.js.map +0 -1
- package/dist/tools/utils/fetch-pipeline.d.ts +0 -2
- package/dist/tools/utils/fetch-pipeline.d.ts.map +0 -1
- package/dist/tools/utils/fetch-pipeline.js +0 -98
- package/dist/tools/utils/fetch-pipeline.js.map +0 -1
- package/dist/tools/utils/frontmatter.d.ts +0 -3
- package/dist/tools/utils/frontmatter.js +0 -73
- package/dist/tools/utils/index.d.ts +0 -4
- package/dist/tools/utils/index.d.ts.map +0 -1
- package/dist/tools/utils/index.js +0 -3
- package/dist/tools/utils/index.js.map +0 -1
- package/dist/tools/utils/inline-content.d.ts +0 -10
- package/dist/tools/utils/inline-content.d.ts.map +0 -1
- package/dist/tools/utils/inline-content.js +0 -35
- package/dist/tools/utils/inline-content.js.map +0 -1
- package/dist/tools/utils/markdown-heuristics.d.ts +0 -1
- package/dist/tools/utils/markdown-heuristics.js +0 -19
- package/dist/tools/utils/markdown-signals.d.ts +0 -1
- package/dist/tools/utils/markdown-signals.js +0 -19
- package/dist/tools/utils/markdown-toc.d.ts +0 -3
- package/dist/tools/utils/markdown-toc.d.ts.map +0 -1
- package/dist/tools/utils/markdown-toc.js +0 -35
- package/dist/tools/utils/markdown-toc.js.map +0 -1
- package/dist/tools/utils/raw-markdown-frontmatter.d.ts +0 -3
- package/dist/tools/utils/raw-markdown-frontmatter.js +0 -73
- package/dist/tools/utils/raw-markdown.d.ts +0 -6
- package/dist/tools/utils/raw-markdown.js +0 -149
- package/dist/tools/utils/response-builder.d.ts +0 -3
- package/dist/tools/utils/response-builder.d.ts.map +0 -1
- package/dist/tools/utils/response-builder.js +0 -24
- package/dist/tools/utils/response-builder.js.map +0 -1
- package/dist/tools/utils/tool-response.d.ts +0 -9
- package/dist/tools/utils/tool-response.d.ts.map +0 -1
- package/dist/tools/utils/tool-response.js +0 -19
- package/dist/tools/utils/tool-response.js.map +0 -1
- package/dist/transformers/jsonl.transformer.d.ts +0 -2
- package/dist/transformers/jsonl.transformer.d.ts.map +0 -1
- package/dist/transformers/jsonl.transformer.js +0 -75
- package/dist/transformers/jsonl.transformer.js.map +0 -1
- package/dist/transformers/markdown/fenced-code-rule.d.ts +0 -2
- package/dist/transformers/markdown/fenced-code-rule.js +0 -38
- package/dist/transformers/markdown/frontmatter.d.ts +0 -2
- package/dist/transformers/markdown/frontmatter.js +0 -45
- package/dist/transformers/markdown/noise-rule.d.ts +0 -2
- package/dist/transformers/markdown/noise-rule.js +0 -80
- package/dist/transformers/markdown/turndown-instance.d.ts +0 -2
- package/dist/transformers/markdown/turndown-instance.js +0 -19
- package/dist/transformers/markdown.d.ts +0 -5
- package/dist/transformers/markdown.js +0 -314
- package/dist/transformers/markdown.transformer.d.ts +0 -2
- package/dist/transformers/markdown.transformer.d.ts.map +0 -1
- package/dist/transformers/markdown.transformer.js +0 -14
- package/dist/transformers/markdown.transformer.js.map +0 -1
- package/dist/types/content.types.d.ts +0 -63
- package/dist/types/content.types.d.ts.map +0 -1
- package/dist/types/content.types.js +0 -2
- package/dist/types/content.types.js.map +0 -1
- package/dist/types/index.d.ts +0 -2
- package/dist/types/index.d.ts.map +0 -1
- package/dist/types/index.js +0 -2
- package/dist/types/index.js.map +0 -1
- package/dist/types/schemas.d.ts +0 -49
- package/dist/types/schemas.d.ts.map +0 -1
- package/dist/types/schemas.js +0 -5
- package/dist/types/schemas.js.map +0 -1
- package/dist/utils/cached-payload.d.ts +0 -7
- package/dist/utils/cached-payload.js +0 -36
- package/dist/utils/cancellation.d.ts +0 -1
- package/dist/utils/cancellation.js +0 -18
- package/dist/utils/code-language-bash.d.ts +0 -1
- package/dist/utils/code-language-bash.js +0 -48
- package/dist/utils/code-language-core.d.ts +0 -2
- package/dist/utils/code-language-core.js +0 -13
- package/dist/utils/code-language-detectors.d.ts +0 -5
- package/dist/utils/code-language-detectors.js +0 -142
- package/dist/utils/code-language-helpers.d.ts +0 -5
- package/dist/utils/code-language-helpers.js +0 -62
- package/dist/utils/code-language-parsing.d.ts +0 -5
- package/dist/utils/code-language-parsing.js +0 -62
- package/dist/utils/code-language.d.ts +0 -2
- package/dist/utils/code-language.d.ts.map +0 -1
- package/dist/utils/code-language.js +0 -260
- package/dist/utils/code-language.js.map +0 -1
- package/dist/utils/concurrency.d.ts +0 -3
- package/dist/utils/concurrency.d.ts.map +0 -1
- package/dist/utils/concurrency.js +0 -38
- package/dist/utils/concurrency.js.map +0 -1
- package/dist/utils/content-cleaner.d.ts +0 -5
- package/dist/utils/content-cleaner.d.ts.map +0 -1
- package/dist/utils/content-cleaner.js +0 -77
- package/dist/utils/content-cleaner.js.map +0 -1
- package/dist/utils/crypto.d.ts +0 -2
- package/dist/utils/crypto.d.ts.map +0 -1
- package/dist/utils/crypto.js +0 -32
- package/dist/utils/crypto.js.map +0 -1
- package/dist/utils/download-url.d.ts +0 -16
- package/dist/utils/download-url.d.ts.map +0 -1
- package/dist/utils/download-url.js +0 -30
- package/dist/utils/download-url.js.map +0 -1
- package/dist/utils/error-details.d.ts +0 -3
- package/dist/utils/error-details.js +0 -12
- package/dist/utils/error-utils.d.ts +0 -3
- package/dist/utils/error-utils.d.ts.map +0 -1
- package/dist/utils/error-utils.js +0 -12
- package/dist/utils/error-utils.js.map +0 -1
- package/dist/utils/filename-generator.d.ts +0 -1
- package/dist/utils/filename-generator.d.ts.map +0 -1
- package/dist/utils/filename-generator.js +0 -81
- package/dist/utils/filename-generator.js.map +0 -1
- package/dist/utils/guards.d.ts +0 -1
- package/dist/utils/guards.js +0 -3
- package/dist/utils/header-normalizer.d.ts +0 -5
- package/dist/utils/header-normalizer.d.ts.map +0 -1
- package/dist/utils/header-normalizer.js +0 -31
- package/dist/utils/header-normalizer.js.map +0 -1
- package/dist/utils/host-normalizer.d.ts +0 -1
- package/dist/utils/host-normalizer.js +0 -37
- package/dist/utils/html-truncator.d.ts +0 -1
- package/dist/utils/html-truncator.d.ts.map +0 -1
- package/dist/utils/html-truncator.js +0 -13
- package/dist/utils/html-truncator.js.map +0 -1
- package/dist/utils/ip-address.d.ts +0 -4
- package/dist/utils/ip-address.js +0 -6
- package/dist/utils/language-detector.d.ts +0 -2
- package/dist/utils/language-detector.d.ts.map +0 -1
- package/dist/utils/language-detector.js +0 -39
- package/dist/utils/language-detector.js.map +0 -1
- package/dist/utils/sanitizer.d.ts +0 -2
- package/dist/utils/sanitizer.d.ts.map +0 -1
- package/dist/utils/sanitizer.js +0 -20
- package/dist/utils/sanitizer.js.map +0 -1
- package/dist/utils/tool-error-handler.d.ts +0 -3
- package/dist/utils/tool-error-handler.d.ts.map +0 -1
- package/dist/utils/tool-error-handler.js +0 -31
- package/dist/utils/tool-error-handler.js.map +0 -1
- package/dist/utils/url-redactor.d.ts +0 -1
- package/dist/utils/url-redactor.js +0 -13
- package/dist/utils/url-sanitizer.d.ts +0 -2
- package/dist/utils/url-sanitizer.d.ts.map +0 -1
- package/dist/utils/url-sanitizer.js +0 -12
- package/dist/utils/url-sanitizer.js.map +0 -1
- package/dist/utils/url-transformer.d.ts +0 -7
- package/dist/utils/url-transformer.js +0 -147
- package/dist/utils/url-validator.d.ts +0 -6
- package/dist/utils/url-validator.d.ts.map +0 -1
- package/dist/utils/url-validator.js +0 -156
- package/dist/utils/url-validator.js.map +0 -1
- package/dist/workers/content-transform.worker.d.ts +0 -1
- package/dist/workers/content-transform.worker.js +0 -40
|
@@ -1,80 +0,0 @@
|
|
|
1
|
-
import { isRecord } from '../../utils/guards.js';
|
|
2
|
-
const STRUCTURAL_TAGS = new Set([
|
|
3
|
-
'script',
|
|
4
|
-
'style',
|
|
5
|
-
'noscript',
|
|
6
|
-
'iframe',
|
|
7
|
-
'nav',
|
|
8
|
-
'footer',
|
|
9
|
-
'aside',
|
|
10
|
-
'header',
|
|
11
|
-
'form',
|
|
12
|
-
'button',
|
|
13
|
-
'input',
|
|
14
|
-
'select',
|
|
15
|
-
'textarea',
|
|
16
|
-
]);
|
|
17
|
-
const NAVIGATION_ROLES = new Set([
|
|
18
|
-
'navigation',
|
|
19
|
-
'banner',
|
|
20
|
-
'complementary',
|
|
21
|
-
'contentinfo',
|
|
22
|
-
'tree',
|
|
23
|
-
'menubar',
|
|
24
|
-
'menu',
|
|
25
|
-
]);
|
|
26
|
-
const PROMO_PATTERN = /banner|promo|announcement|cta|callout|advert|newsletter|subscribe|cookie|consent|popup|modal|overlay|toast/;
|
|
27
|
-
const FIXED_PATTERN = /\b(fixed|sticky)\b/;
|
|
28
|
-
const HIGH_Z_PATTERN = /\bz-(?:4\d|50)\b/;
|
|
29
|
-
const ISOLATE_PATTERN = /\bisolate\b/;
|
|
30
|
-
function isElement(node) {
|
|
31
|
-
return (isRecord(node) &&
|
|
32
|
-
'getAttribute' in node &&
|
|
33
|
-
typeof node.getAttribute === 'function');
|
|
34
|
-
}
|
|
35
|
-
function isStructuralNoiseTag(tagName) {
|
|
36
|
-
return (STRUCTURAL_TAGS.has(tagName) || tagName === 'svg' || tagName === 'canvas');
|
|
37
|
-
}
|
|
38
|
-
function isElementHidden(element) {
|
|
39
|
-
return (element.getAttribute('hidden') !== null ||
|
|
40
|
-
element.getAttribute('aria-hidden') === 'true');
|
|
41
|
-
}
|
|
42
|
-
function hasNoiseRole(role) {
|
|
43
|
-
return role !== null && NAVIGATION_ROLES.has(role);
|
|
44
|
-
}
|
|
45
|
-
function matchesPromoIdOrClass(className, id) {
|
|
46
|
-
const combined = `${className} ${id}`.toLowerCase();
|
|
47
|
-
return PROMO_PATTERN.test(combined);
|
|
48
|
-
}
|
|
49
|
-
function matchesHighZIsolate(className) {
|
|
50
|
-
return HIGH_Z_PATTERN.test(className) && ISOLATE_PATTERN.test(className);
|
|
51
|
-
}
|
|
52
|
-
function matchesFixedOrHighZIsolate(className) {
|
|
53
|
-
return FIXED_PATTERN.test(className) || matchesHighZIsolate(className);
|
|
54
|
-
}
|
|
55
|
-
function readElementMetadata(element) {
|
|
56
|
-
return {
|
|
57
|
-
tagName: element.tagName.toLowerCase(),
|
|
58
|
-
className: element.getAttribute('class') ?? '',
|
|
59
|
-
id: element.getAttribute('id') ?? '',
|
|
60
|
-
role: element.getAttribute('role'),
|
|
61
|
-
isHidden: isElementHidden(element),
|
|
62
|
-
};
|
|
63
|
-
}
|
|
64
|
-
function isNoiseElement(node) {
|
|
65
|
-
const metadata = readElementMetadata(node);
|
|
66
|
-
return (isStructuralNoiseTag(metadata.tagName) ||
|
|
67
|
-
metadata.isHidden ||
|
|
68
|
-
hasNoiseRole(metadata.role) ||
|
|
69
|
-
matchesFixedOrHighZIsolate(metadata.className) ||
|
|
70
|
-
matchesPromoIdOrClass(metadata.className, metadata.id));
|
|
71
|
-
}
|
|
72
|
-
function isNoiseNode(node) {
|
|
73
|
-
return isElement(node) && isNoiseElement(node);
|
|
74
|
-
}
|
|
75
|
-
export function addNoiseRule(instance) {
|
|
76
|
-
instance.addRule('removeNoise', {
|
|
77
|
-
filter: (node) => isNoiseNode(node),
|
|
78
|
-
replacement: () => '',
|
|
79
|
-
});
|
|
80
|
-
}
|
|
@@ -1,19 +0,0 @@
|
|
|
1
|
-
import TurndownService from 'turndown';
|
|
2
|
-
import { addFencedCodeRule } from './fenced-code-rule.js';
|
|
3
|
-
import { addNoiseRule } from './noise-rule.js';
|
|
4
|
-
let turndownInstance = null;
|
|
5
|
-
function createTurndownInstance() {
|
|
6
|
-
const instance = new TurndownService({
|
|
7
|
-
headingStyle: 'atx',
|
|
8
|
-
codeBlockStyle: 'fenced',
|
|
9
|
-
emDelimiter: '_',
|
|
10
|
-
bulletListMarker: '-',
|
|
11
|
-
});
|
|
12
|
-
addNoiseRule(instance);
|
|
13
|
-
addFencedCodeRule(instance);
|
|
14
|
-
return instance;
|
|
15
|
-
}
|
|
16
|
-
export function getTurndown() {
|
|
17
|
-
turndownInstance ??= createTurndownInstance();
|
|
18
|
-
return turndownInstance;
|
|
19
|
-
}
|
|
@@ -1,314 +0,0 @@
|
|
|
1
|
-
import { parseHTML } from 'linkedom';
|
|
2
|
-
import { NodeHtmlMarkdown, } from 'node-html-markdown';
|
|
3
|
-
import { CODE_BLOCK, FRONTMATTER_DELIMITER, joinLines, } from '../config/formatting.js';
|
|
4
|
-
import { FetchError } from '../errors/app-error.js';
|
|
5
|
-
import { endTransformStage, startTransformStage, } from '../services/telemetry.js';
|
|
6
|
-
import { throwIfAborted } from '../utils/cancellation.js';
|
|
7
|
-
import { detectLanguageFromCode, resolveLanguageFromAttributes, } from '../utils/code-language.js';
|
|
8
|
-
import { isRecord } from '../utils/guards.js';
|
|
9
|
-
const YAML_SPECIAL_CHARS = /[:[\]{}"\r\t'|>&*!?,#]|\n/;
|
|
10
|
-
const YAML_NUMERIC = /^[\d.]+$/;
|
|
11
|
-
const YAML_RESERVED_WORDS = /^(true|false|null|yes|no|on|off)$/i;
|
|
12
|
-
const ESCAPE_PATTERNS = {
|
|
13
|
-
backslash: /\\/g,
|
|
14
|
-
quote: /"/g,
|
|
15
|
-
newline: /\n/g,
|
|
16
|
-
tab: /\t/g,
|
|
17
|
-
};
|
|
18
|
-
const YAML_QUOTE_CHECKS = [
|
|
19
|
-
(input) => YAML_SPECIAL_CHARS.test(input),
|
|
20
|
-
(input) => input.startsWith(' ') || input.endsWith(' '),
|
|
21
|
-
(input) => input === '',
|
|
22
|
-
(input) => YAML_NUMERIC.test(input),
|
|
23
|
-
(input) => YAML_RESERVED_WORDS.test(input),
|
|
24
|
-
];
|
|
25
|
-
function needsYamlQuotes(value) {
|
|
26
|
-
return YAML_QUOTE_CHECKS.some((check) => check(value));
|
|
27
|
-
}
|
|
28
|
-
function escapeYamlValue(value) {
|
|
29
|
-
if (!needsYamlQuotes(value)) {
|
|
30
|
-
return value;
|
|
31
|
-
}
|
|
32
|
-
const escaped = value
|
|
33
|
-
.replace(ESCAPE_PATTERNS.backslash, '\\\\')
|
|
34
|
-
.replace(ESCAPE_PATTERNS.quote, '\\"')
|
|
35
|
-
.replace(ESCAPE_PATTERNS.newline, '\\n')
|
|
36
|
-
.replace(ESCAPE_PATTERNS.tab, '\\t');
|
|
37
|
-
return `"${escaped}"`;
|
|
38
|
-
}
|
|
39
|
-
function appendFrontmatterField(lines, key, value) {
|
|
40
|
-
if (!value)
|
|
41
|
-
return;
|
|
42
|
-
lines.push(`${key}: ${escapeYamlValue(value)}`);
|
|
43
|
-
}
|
|
44
|
-
function buildFrontmatter(metadata) {
|
|
45
|
-
if (!metadata)
|
|
46
|
-
return '';
|
|
47
|
-
const lines = [FRONTMATTER_DELIMITER];
|
|
48
|
-
appendFrontmatterField(lines, 'title', metadata.title);
|
|
49
|
-
appendFrontmatterField(lines, 'source', metadata.url);
|
|
50
|
-
appendFrontmatterField(lines, 'author', metadata.author);
|
|
51
|
-
appendFrontmatterField(lines, 'description', metadata.description);
|
|
52
|
-
appendFrontmatterField(lines, 'fetchedAt', metadata.fetchedAt);
|
|
53
|
-
lines.push(FRONTMATTER_DELIMITER);
|
|
54
|
-
return joinLines(lines);
|
|
55
|
-
}
|
|
56
|
-
function isElement(node) {
|
|
57
|
-
return (isRecord(node) &&
|
|
58
|
-
'getAttribute' in node &&
|
|
59
|
-
typeof node.getAttribute === 'function');
|
|
60
|
-
}
|
|
61
|
-
const STRUCTURAL_TAGS = new Set([
|
|
62
|
-
'script',
|
|
63
|
-
'style',
|
|
64
|
-
'noscript',
|
|
65
|
-
'iframe',
|
|
66
|
-
'nav',
|
|
67
|
-
'footer',
|
|
68
|
-
'aside',
|
|
69
|
-
'header',
|
|
70
|
-
'form',
|
|
71
|
-
'button',
|
|
72
|
-
'input',
|
|
73
|
-
'select',
|
|
74
|
-
'textarea',
|
|
75
|
-
]);
|
|
76
|
-
const NAVIGATION_ROLES = new Set([
|
|
77
|
-
'navigation',
|
|
78
|
-
'banner',
|
|
79
|
-
'complementary',
|
|
80
|
-
'contentinfo',
|
|
81
|
-
'tree',
|
|
82
|
-
'menubar',
|
|
83
|
-
'menu',
|
|
84
|
-
]);
|
|
85
|
-
const PROMO_PATTERN = /banner|promo|announcement|cta|callout|advert|newsletter|subscribe|cookie|consent|popup|modal|overlay|toast/;
|
|
86
|
-
const FIXED_PATTERN = /\b(fixed|sticky)\b/;
|
|
87
|
-
const HIGH_Z_PATTERN = /\bz-(?:4\d|50)\b/;
|
|
88
|
-
const ISOLATE_PATTERN = /\bisolate\b/;
|
|
89
|
-
const HTML_DOCUMENT_MARKERS = /<\s*(?:!doctype|html|head|body)\b/i;
|
|
90
|
-
const NOISE_MARKERS = [
|
|
91
|
-
'<script',
|
|
92
|
-
'<style',
|
|
93
|
-
'<noscript',
|
|
94
|
-
'<iframe',
|
|
95
|
-
'<nav',
|
|
96
|
-
'<footer',
|
|
97
|
-
'<aside',
|
|
98
|
-
'<header',
|
|
99
|
-
'<form',
|
|
100
|
-
'<button',
|
|
101
|
-
'<input',
|
|
102
|
-
'<select',
|
|
103
|
-
'<textarea',
|
|
104
|
-
'<svg',
|
|
105
|
-
'<canvas',
|
|
106
|
-
' aria-hidden="true"',
|
|
107
|
-
" aria-hidden='true'",
|
|
108
|
-
' hidden',
|
|
109
|
-
' role="navigation"',
|
|
110
|
-
" role='navigation'",
|
|
111
|
-
' role="banner"',
|
|
112
|
-
" role='banner'",
|
|
113
|
-
' role="complementary"',
|
|
114
|
-
" role='complementary'",
|
|
115
|
-
' role="contentinfo"',
|
|
116
|
-
" role='contentinfo'",
|
|
117
|
-
' role="tree"',
|
|
118
|
-
" role='tree'",
|
|
119
|
-
' role="menubar"',
|
|
120
|
-
" role='menubar'",
|
|
121
|
-
' role="menu"',
|
|
122
|
-
" role='menu'",
|
|
123
|
-
' banner',
|
|
124
|
-
' promo',
|
|
125
|
-
' announcement',
|
|
126
|
-
' cta',
|
|
127
|
-
' callout',
|
|
128
|
-
' advert',
|
|
129
|
-
' newsletter',
|
|
130
|
-
' subscribe',
|
|
131
|
-
' cookie',
|
|
132
|
-
' consent',
|
|
133
|
-
' popup',
|
|
134
|
-
' modal',
|
|
135
|
-
' overlay',
|
|
136
|
-
' toast',
|
|
137
|
-
' fixed',
|
|
138
|
-
' sticky',
|
|
139
|
-
' z-50',
|
|
140
|
-
' z-4',
|
|
141
|
-
' isolate',
|
|
142
|
-
];
|
|
143
|
-
function mayContainNoise(html) {
|
|
144
|
-
const haystack = html.toLowerCase();
|
|
145
|
-
return NOISE_MARKERS.some((marker) => haystack.includes(marker));
|
|
146
|
-
}
|
|
147
|
-
function isFullDocumentHtml(html) {
|
|
148
|
-
return HTML_DOCUMENT_MARKERS.test(html);
|
|
149
|
-
}
|
|
150
|
-
function isStructuralNoiseTag(tagName) {
|
|
151
|
-
return (STRUCTURAL_TAGS.has(tagName) || tagName === 'svg' || tagName === 'canvas');
|
|
152
|
-
}
|
|
153
|
-
function isElementHidden(element) {
|
|
154
|
-
return (element.getAttribute('hidden') !== null ||
|
|
155
|
-
element.getAttribute('aria-hidden') === 'true');
|
|
156
|
-
}
|
|
157
|
-
function hasNoiseRole(role) {
|
|
158
|
-
return role !== null && NAVIGATION_ROLES.has(role);
|
|
159
|
-
}
|
|
160
|
-
function matchesPromoIdOrClass(className, id) {
|
|
161
|
-
const combined = `${className} ${id}`.toLowerCase();
|
|
162
|
-
return PROMO_PATTERN.test(combined);
|
|
163
|
-
}
|
|
164
|
-
function matchesHighZIsolate(className) {
|
|
165
|
-
return HIGH_Z_PATTERN.test(className) && ISOLATE_PATTERN.test(className);
|
|
166
|
-
}
|
|
167
|
-
function matchesFixedOrHighZIsolate(className) {
|
|
168
|
-
return FIXED_PATTERN.test(className) || matchesHighZIsolate(className);
|
|
169
|
-
}
|
|
170
|
-
function readElementMetadata(element) {
|
|
171
|
-
return {
|
|
172
|
-
tagName: element.tagName.toLowerCase(),
|
|
173
|
-
className: element.getAttribute('class') ?? '',
|
|
174
|
-
id: element.getAttribute('id') ?? '',
|
|
175
|
-
role: element.getAttribute('role'),
|
|
176
|
-
isHidden: isElementHidden(element),
|
|
177
|
-
};
|
|
178
|
-
}
|
|
179
|
-
function isNoiseElement(node) {
|
|
180
|
-
const metadata = readElementMetadata(node);
|
|
181
|
-
return (isStructuralNoiseTag(metadata.tagName) ||
|
|
182
|
-
metadata.isHidden ||
|
|
183
|
-
hasNoiseRole(metadata.role) ||
|
|
184
|
-
matchesFixedOrHighZIsolate(metadata.className) ||
|
|
185
|
-
matchesPromoIdOrClass(metadata.className, metadata.id));
|
|
186
|
-
}
|
|
187
|
-
function removeNoiseFromHtml(html) {
|
|
188
|
-
const shouldParse = isFullDocumentHtml(html) || mayContainNoise(html);
|
|
189
|
-
if (!shouldParse)
|
|
190
|
-
return html;
|
|
191
|
-
const shouldRemove = mayContainNoise(html);
|
|
192
|
-
try {
|
|
193
|
-
const { document } = parseHTML(html);
|
|
194
|
-
if (shouldRemove) {
|
|
195
|
-
const nodes = Array.from(document.querySelectorAll('*'));
|
|
196
|
-
for (let index = nodes.length - 1; index >= 0; index -= 1) {
|
|
197
|
-
const node = nodes[index];
|
|
198
|
-
if (!node)
|
|
199
|
-
continue;
|
|
200
|
-
if (isElement(node) && isNoiseElement(node)) {
|
|
201
|
-
node.remove();
|
|
202
|
-
}
|
|
203
|
-
}
|
|
204
|
-
}
|
|
205
|
-
const { body } = document;
|
|
206
|
-
if (body?.innerHTML)
|
|
207
|
-
return body.innerHTML;
|
|
208
|
-
if (typeof document.toString ===
|
|
209
|
-
'function') {
|
|
210
|
-
return document.toString();
|
|
211
|
-
}
|
|
212
|
-
const { documentElement } = document;
|
|
213
|
-
if (documentElement?.outerHTML)
|
|
214
|
-
return documentElement.outerHTML;
|
|
215
|
-
return html;
|
|
216
|
-
}
|
|
217
|
-
catch {
|
|
218
|
-
return html;
|
|
219
|
-
}
|
|
220
|
-
}
|
|
221
|
-
function buildInlineCode(content) {
|
|
222
|
-
const runs = content.match(/`+/g);
|
|
223
|
-
const longest = runs?.sort((a, b) => b.length - a.length)[0] ?? '';
|
|
224
|
-
const delimiter = `\`${longest}`;
|
|
225
|
-
const padding = delimiter.length > 1 ? ' ' : '';
|
|
226
|
-
return `${delimiter}${padding}${content}${padding}${delimiter}`;
|
|
227
|
-
}
|
|
228
|
-
function isCodeBlock(parent) {
|
|
229
|
-
if (!isRecord(parent))
|
|
230
|
-
return false;
|
|
231
|
-
const tagName = typeof parent.tagName === 'string' ? parent.tagName.toUpperCase() : '';
|
|
232
|
-
return ['PRE', 'WRAPPED-PRE'].includes(tagName);
|
|
233
|
-
}
|
|
234
|
-
function createCodeTranslator() {
|
|
235
|
-
return {
|
|
236
|
-
code: (ctx) => {
|
|
237
|
-
if (!isRecord(ctx)) {
|
|
238
|
-
return {
|
|
239
|
-
spaceIfRepeatingChar: true,
|
|
240
|
-
noEscape: true,
|
|
241
|
-
postprocess: ({ content }) => buildInlineCode(content),
|
|
242
|
-
};
|
|
243
|
-
}
|
|
244
|
-
const { node, parent, visitor } = ctx;
|
|
245
|
-
const getAttribute = isRecord(node) && typeof node.getAttribute === 'function'
|
|
246
|
-
? node.getAttribute.bind(node)
|
|
247
|
-
: undefined;
|
|
248
|
-
if (!isCodeBlock(parent)) {
|
|
249
|
-
return {
|
|
250
|
-
spaceIfRepeatingChar: true,
|
|
251
|
-
noEscape: true,
|
|
252
|
-
postprocess: ({ content }) => buildInlineCode(content),
|
|
253
|
-
};
|
|
254
|
-
}
|
|
255
|
-
const className = getAttribute?.('class') ?? '';
|
|
256
|
-
const dataLanguage = getAttribute?.('data-language') ?? '';
|
|
257
|
-
const attributeLanguage = resolveLanguageFromAttributes(className, dataLanguage);
|
|
258
|
-
const childTranslators = isRecord(visitor) ? visitor.instance : null;
|
|
259
|
-
const codeBlockTranslators = isRecord(childTranslators) &&
|
|
260
|
-
isRecord(childTranslators
|
|
261
|
-
.codeBlockTranslators)
|
|
262
|
-
? childTranslators.codeBlockTranslators
|
|
263
|
-
: null;
|
|
264
|
-
return {
|
|
265
|
-
noEscape: true,
|
|
266
|
-
preserveWhitespace: true,
|
|
267
|
-
...(codeBlockTranslators
|
|
268
|
-
? { childTranslators: codeBlockTranslators }
|
|
269
|
-
: null),
|
|
270
|
-
postprocess: ({ content }) => {
|
|
271
|
-
const language = attributeLanguage ?? detectLanguageFromCode(content) ?? '';
|
|
272
|
-
return CODE_BLOCK.format(content, language);
|
|
273
|
-
},
|
|
274
|
-
};
|
|
275
|
-
},
|
|
276
|
-
};
|
|
277
|
-
}
|
|
278
|
-
let markdownInstance = null;
|
|
279
|
-
function createMarkdownInstance() {
|
|
280
|
-
return new NodeHtmlMarkdown({
|
|
281
|
-
codeFence: CODE_BLOCK.fence,
|
|
282
|
-
codeBlockStyle: 'fenced',
|
|
283
|
-
emDelimiter: '_',
|
|
284
|
-
bulletMarker: '-',
|
|
285
|
-
}, createCodeTranslator());
|
|
286
|
-
}
|
|
287
|
-
function getMarkdownConverter() {
|
|
288
|
-
markdownInstance ??= createMarkdownInstance();
|
|
289
|
-
return markdownInstance;
|
|
290
|
-
}
|
|
291
|
-
export function htmlToMarkdown(html, metadata, options) {
|
|
292
|
-
const url = options?.url ?? metadata?.url ?? '';
|
|
293
|
-
const frontmatter = buildFrontmatter(metadata);
|
|
294
|
-
if (!html)
|
|
295
|
-
return frontmatter;
|
|
296
|
-
try {
|
|
297
|
-
throwIfAborted(options?.signal, url, 'markdown:begin');
|
|
298
|
-
const noiseStage = startTransformStage(url, 'markdown:noise');
|
|
299
|
-
const cleanedHtml = removeNoiseFromHtml(html);
|
|
300
|
-
endTransformStage(noiseStage);
|
|
301
|
-
throwIfAborted(options?.signal, url, 'markdown:cleaned');
|
|
302
|
-
const translateStage = startTransformStage(url, 'markdown:translate');
|
|
303
|
-
const content = getMarkdownConverter().translate(cleanedHtml).trim();
|
|
304
|
-
endTransformStage(translateStage);
|
|
305
|
-
throwIfAborted(options?.signal, url, 'markdown:translated');
|
|
306
|
-
return frontmatter ? `${frontmatter}\n${content}` : content;
|
|
307
|
-
}
|
|
308
|
-
catch (error) {
|
|
309
|
-
if (error instanceof FetchError) {
|
|
310
|
-
throw error;
|
|
311
|
-
}
|
|
312
|
-
return frontmatter;
|
|
313
|
-
}
|
|
314
|
-
}
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"markdown.transformer.d.ts","sourceRoot":"","sources":["../../src/transformers/markdown.transformer.ts"],"names":[],"mappings":"AAOA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,4BAA4B,CAAC;AA+HhE,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,aAAa,GAAG,MAAM,CAa7E"}
|
|
@@ -1,14 +0,0 @@
|
|
|
1
|
-
import { buildFrontmatter } from './markdown/frontmatter.js';
|
|
2
|
-
import { getTurndown } from './markdown/turndown-instance.js';
|
|
3
|
-
export function htmlToMarkdown(html, metadata) {
|
|
4
|
-
const frontmatter = buildFrontmatter(metadata);
|
|
5
|
-
if (!html)
|
|
6
|
-
return frontmatter;
|
|
7
|
-
try {
|
|
8
|
-
const content = getTurndown().turndown(html).trim();
|
|
9
|
-
return frontmatter ? `${frontmatter}\n${content}` : content;
|
|
10
|
-
}
|
|
11
|
-
catch {
|
|
12
|
-
return frontmatter;
|
|
13
|
-
}
|
|
14
|
-
}
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"markdown.transformer.js","sourceRoot":"","sources":["../../src/transformers/markdown.transformer.ts"],"names":[],"mappings":"AAAA,OAAO,eAAe,MAAM,UAAU,CAAC;AAEvC,OAAO,EACL,UAAU,EACV,qBAAqB,EACrB,SAAS,GACV,MAAM,yBAAyB,CAAC;AAGjC,OAAO,EACL,sBAAsB,EACtB,6BAA6B,GAC9B,MAAM,2BAA2B,CAAC;AAEnC,IAAI,gBAAgB,GAA2B,IAAI,CAAC;AAEpD,SAAS,WAAW;IAClB,IAAI,gBAAgB;QAAE,OAAO,gBAAgB,CAAC;IAC9C,gBAAgB,GAAG,sBAAsB,EAAE,CAAC;IAC5C,OAAO,gBAAgB,CAAC;AAC1B,CAAC;AAED,SAAS,sBAAsB;IAC7B,MAAM,QAAQ,GAAG,IAAI,eAAe,CAAC;QACnC,YAAY,EAAE,KAAK;QACnB,cAAc,EAAE,QAAQ;QACxB,WAAW,EAAE,GAAG;QAChB,gBAAgB,EAAE,GAAG;KACtB,CAAC,CAAC;IAEH,YAAY,CAAC,QAAQ,CAAC,CAAC;IACvB,iBAAiB,CAAC,QAAQ,CAAC,CAAC;IAE5B,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,SAAS,YAAY,CAAC,QAAyB;IAC7C,QAAQ,CAAC,OAAO,CAAC,aAAa,EAAE;QAC9B,MAAM,EAAE,CAAC,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,CAAC;QAC3E,WAAW,EAAE,GAAG,EAAE,CAAC,EAAE;KACtB,CAAC,CAAC;AACL,CAAC;AAED,SAAS,iBAAiB,CAAC,QAAyB;IAClD,QAAQ,CAAC,OAAO,CAAC,6BAA6B,EAAE;QAC9C,MAAM,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CAAC,iBAAiB,CAAC,IAAI,EAAE,OAAO,CAAC;QAC3D,WAAW,EAAE,CAAC,QAAQ,EAAE,IAAI,EAAE,EAAE,CAAC,qBAAqB,CAAC,IAAI,CAAC;KAC7D,CAAC,CAAC;AACL,CAAC;AAED,SAAS,iBAAiB,CACxB,IAA0B,EAC1B,OAAgC;IAEhC,IAAI,OAAO,CAAC,cAAc,KAAK,QAAQ;QAAE,OAAO,KAAK,CAAC;IACtD,IAAI,IAAI,CAAC,QAAQ,KAAK,KAAK;QAAE,OAAO,KAAK,CAAC;IAC1C,MAAM,EAAE,UAAU,EAAE,GAAG,IAAI,CAAC;IAC5B,IAAI,CAAC,UAAU;QAAE,OAAO,KAAK,CAAC;IAC9B,OAAO,UAAU,CAAC,QAAQ,KAAK,MAAM,CAAC;AACxC,CAAC;AAED,SAAS,qBAAqB,CAAC,IAA0B;IACvD,MAAM,QAAQ,GAAG,IAAI,CAAC,UAAyB,CAAC;IAChD,MAAM,IAAI,GAAG,QAAQ,CAAC,WAAW,IAAI,EAAE,CAAC;IACxC,MAAM,QAAQ,GAAG,mBAAmB,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;IACrD,OAAO,UAAU,CAAC,MAAM,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;AAC3C,CAAC;AAED,SAAS,mBAAmB,CAAC,QAAqB,EAAE,IAAY;IAC9D,MAAM,SAAS,GAAG,QAAQ,CAAC,YAAY,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;IACvD,MAAM,QAAQ,GAAG,QAAQ,CAAC,YAAY,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;IAC9D,MAAM,iBAAiB,GAAG,6BAA6B,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC;IAC7E,OAAO,iBAAiB,IAAI,sBAAsB,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;AACjE,CAAC;AAED,MAAM,kBAAkB,GAAG,2BAA2B,CAAC;AACvD,MAAM,YAAY,GAAG,UAAU,CAAC;AAChC,MAAM,mBAAmB,GAAG,oCAAoC,CAAC;AAEjE,MAAM,eAAe,GAAG;IACtB,SAAS,EAAE,KAAK;IAChB,KAAK,EAAE,IAAI;IACX,OAAO,EAAE,KAAK;IACd,GAAG,EAAE,KAAK;CACF,CAAC;AAEX,SAAS,eAAe,CAAC,KAAa;IACpC,MAAM,MAAM,GAAG;QACb,CAAC,KAAa,EAAE,EAAE,CAAC,kBAAkB,CAAC,IAAI,CAAC,KAAK,CAAC;QACjD,CAAC,KAAa,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC;QAC/D,CAAC,KAAa,EAAE,EAAE,CAAC,KAAK,KAAK,EAAE;QAC/B,CAAC,KAAa,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC;QAC3C,CAAC,KAAa,EAAE,EAAE,CAAC,mBAAmB,CAAC,IAAI,CAAC,KAAK,CAAC;KACnD,CAAC;IAEF,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC;AAC9C,CAAC;AAED,SAAS,eAAe,CAAC,KAAa;IACpC,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,EAAE,CAAC;QAC5B,OAAO,KAAK,CAAC;IACf,CAAC;IAED,MAAM,OAAO,GAAG,KAAK;SAClB,OAAO,CAAC,eAAe,CAAC,SAAS,EAAE,MAAM,CAAC;SAC1C,OAAO,CAAC,eAAe,CAAC,KAAK,EAAE,KAAK,CAAC;SACrC,OAAO,CAAC,eAAe,CAAC,OAAO,EAAE,KAAK,CAAC;SACvC,OAAO,CAAC,eAAe,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;IAEvC,OAAO,IAAI,OAAO,GAAG,CAAC;AACxB,CAAC;AAED,SAAS,iBAAiB,CAAC,QAAuB;IAChD,MAAM,KAAK,GAAa,CAAC,qBAAqB,CAAC,CAAC;IAEhD,IAAI,QAAQ,CAAC,KAAK,EAAE,CAAC;QACnB,KAAK,CAAC,IAAI,CAAC,UAAU,eAAe,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;IAC1D,CAAC;IACD,IAAI,QAAQ,CAAC,GAAG,EAAE,CAAC;QACjB,KAAK,CAAC,IAAI,CAAC,WAAW,eAAe,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;IACzD,CAAC;IAED,KAAK,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAC;IAClC,OAAO,SAAS,CAAC,KAAK,CAAC,CAAC;AAC1B,CAAC;AAED,SAAS,qBAAqB,CAAC,IAAY;IACzC,OAAO,WAAW,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;AAC7C,CAAC;AAED,SAAS,qBAAqB,CAAC,QAAwB;IACrD,OAAO,QAAQ,CAAC,CAAC,CAAC,iBAAiB,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;AACrD,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,IAAY,EAAE,QAAwB;IACnE,MAAM,WAAW,GAAG,qBAAqB,CAAC,QAAQ,CAAC,CAAC;IAEpD,IAAI,CAAC,gBAAgB,CAAC,IAAI,CAAC,EAAE,CAAC;QAC5B,OAAO,WAAW,CAAC;IACrB,CAAC;IAED,IAAI,CAAC;QACH,MAAM,OAAO,GAAG,qBAAqB,CAAC,IAAI,CAAC,CAAC;QAC5C,OAAO,WAAW,CAAC,CAAC,CAAC,GAAG,WAAW,KAAK,OAAO,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC;IAC9D,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,WAAW,CAAC;IACrB,CAAC;AACH,CAAC;AAED,SAAS,gBAAgB,CAAC,IAAY;IACpC,OAAO,OAAO,CAAC,IAAI,IAAI,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC;AACnD,CAAC"}
|
|
@@ -1,63 +0,0 @@
|
|
|
1
|
-
type ContentBlockType = 'metadata' | 'heading' | 'paragraph' | 'list' | 'code' | 'table' | 'image';
|
|
2
|
-
interface ContentBlock {
|
|
3
|
-
type: ContentBlockType;
|
|
4
|
-
}
|
|
5
|
-
export interface MetadataBlock extends ContentBlock {
|
|
6
|
-
type: 'metadata';
|
|
7
|
-
title?: string | undefined;
|
|
8
|
-
description?: string | undefined;
|
|
9
|
-
author?: string | undefined;
|
|
10
|
-
url: string;
|
|
11
|
-
fetchedAt: string;
|
|
12
|
-
}
|
|
13
|
-
export interface HeadingBlock extends ContentBlock {
|
|
14
|
-
type: 'heading';
|
|
15
|
-
level: number;
|
|
16
|
-
text: string;
|
|
17
|
-
}
|
|
18
|
-
export interface ParagraphBlock extends ContentBlock {
|
|
19
|
-
type: 'paragraph';
|
|
20
|
-
text: string;
|
|
21
|
-
}
|
|
22
|
-
export interface ListBlock extends ContentBlock {
|
|
23
|
-
type: 'list';
|
|
24
|
-
ordered: boolean;
|
|
25
|
-
items: string[];
|
|
26
|
-
}
|
|
27
|
-
export interface CodeBlock extends ContentBlock {
|
|
28
|
-
type: 'code';
|
|
29
|
-
language?: string | undefined;
|
|
30
|
-
text: string;
|
|
31
|
-
}
|
|
32
|
-
export interface TableBlock extends ContentBlock {
|
|
33
|
-
type: 'table';
|
|
34
|
-
headers?: string[] | undefined;
|
|
35
|
-
rows: string[][];
|
|
36
|
-
}
|
|
37
|
-
export interface ImageBlock extends ContentBlock {
|
|
38
|
-
type: 'image';
|
|
39
|
-
src: string;
|
|
40
|
-
alt?: string | undefined;
|
|
41
|
-
}
|
|
42
|
-
export type ContentBlockUnion = MetadataBlock | HeadingBlock | ParagraphBlock | ListBlock | CodeBlock | TableBlock | ImageBlock;
|
|
43
|
-
export interface ExtractedArticle {
|
|
44
|
-
title?: string | undefined;
|
|
45
|
-
byline?: string | undefined;
|
|
46
|
-
content: string;
|
|
47
|
-
textContent: string;
|
|
48
|
-
excerpt?: string | undefined;
|
|
49
|
-
siteName?: string | undefined;
|
|
50
|
-
}
|
|
51
|
-
export interface CacheEntry {
|
|
52
|
-
url: string;
|
|
53
|
-
content: string;
|
|
54
|
-
fetchedAt: string;
|
|
55
|
-
expiresAt: string;
|
|
56
|
-
}
|
|
57
|
-
export interface ExtractedLink {
|
|
58
|
-
href: string;
|
|
59
|
-
text: string;
|
|
60
|
-
type: 'internal' | 'external' | 'image';
|
|
61
|
-
}
|
|
62
|
-
export {};
|
|
63
|
-
//# sourceMappingURL=content.types.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"content.types.d.ts","sourceRoot":"","sources":["../../src/types/content.types.ts"],"names":[],"mappings":"AACA,KAAK,gBAAgB,GACjB,UAAU,GACV,SAAS,GACT,WAAW,GACX,MAAM,GACN,MAAM,GACN,OAAO,GACP,OAAO,CAAC;AAGZ,UAAU,YAAY;IACpB,IAAI,EAAE,gBAAgB,CAAC;CACxB;AAGD,MAAM,WAAW,aAAc,SAAQ,YAAY;IACjD,IAAI,EAAE,UAAU,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC3B,WAAW,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACjC,MAAM,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC5B,GAAG,EAAE,MAAM,CAAC;IACZ,SAAS,EAAE,MAAM,CAAC;CACnB;AAGD,MAAM,WAAW,YAAa,SAAQ,YAAY;IAChD,IAAI,EAAE,SAAS,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,MAAM,CAAC;CACd;AAGD,MAAM,WAAW,cAAe,SAAQ,YAAY;IAClD,IAAI,EAAE,WAAW,CAAC;IAClB,IAAI,EAAE,MAAM,CAAC;CACd;AAGD,MAAM,WAAW,SAAU,SAAQ,YAAY;IAC7C,IAAI,EAAE,MAAM,CAAC;IACb,OAAO,EAAE,OAAO,CAAC;IACjB,KAAK,EAAE,MAAM,EAAE,CAAC;CACjB;AAGD,MAAM,WAAW,SAAU,SAAQ,YAAY;IAC7C,IAAI,EAAE,MAAM,CAAC;IACb,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC9B,IAAI,EAAE,MAAM,CAAC;CACd;AAGD,MAAM,WAAW,UAAW,SAAQ,YAAY;IAC9C,IAAI,EAAE,OAAO,CAAC;IACd,OAAO,CAAC,EAAE,MAAM,EAAE,GAAG,SAAS,CAAC;IAC/B,IAAI,EAAE,MAAM,EAAE,EAAE,CAAC;CAClB;AAGD,MAAM,WAAW,UAAW,SAAQ,YAAY;IAC9C,IAAI,EAAE,OAAO,CAAC;IACd,GAAG,EAAE,MAAM,CAAC;IACZ,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;CAC1B;AAGD,MAAM,MAAM,iBAAiB,GACzB,aAAa,GACb,YAAY,GACZ,cAAc,GACd,SAAS,GACT,SAAS,GACT,UAAU,GACV,UAAU,CAAC;AAGf,MAAM,WAAW,gBAAgB;IAC/B,KAAK,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC3B,MAAM,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC5B,OAAO,EAAE,MAAM,CAAC;IAChB,WAAW,EAAE,MAAM,CAAC;IACpB,OAAO,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC7B,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;CAC/B;AAGD,MAAM,WAAW,UAAU;IACzB,GAAG,EAAE,MAAM,CAAC;IACZ,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;CACnB;AAGD,MAAM,WAAW,aAAa;IAC5B,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,UAAU,GAAG,UAAU,GAAG,OAAO,CAAC;CACzC"}
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"content.types.js","sourceRoot":"","sources":["../../src/types/content.types.ts"],"names":[],"mappings":""}
|
package/dist/types/index.d.ts
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/types/index.ts"],"names":[],"mappings":"AAAA,mBAAmB,oBAAoB,CAAC"}
|
package/dist/types/index.js
DELETED
package/dist/types/index.js.map
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/types/index.ts"],"names":[],"mappings":""}
|
package/dist/types/schemas.d.ts
DELETED
|
@@ -1,49 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* Tool input types - used for type safety in tool handlers
|
|
3
|
-
*/
|
|
4
|
-
/** Common request options shared across tools */
|
|
5
|
-
export interface RequestOptions {
|
|
6
|
-
/** Custom HTTP headers for the request */
|
|
7
|
-
customHeaders?: Record<string, string> | undefined;
|
|
8
|
-
/** Request timeout in milliseconds (1000-60000) */
|
|
9
|
-
timeout?: number | undefined;
|
|
10
|
-
/** Number of retry attempts (1-10) */
|
|
11
|
-
retries?: number | undefined;
|
|
12
|
-
}
|
|
13
|
-
export interface FetchUrlInput extends RequestOptions {
|
|
14
|
-
url: string;
|
|
15
|
-
extractMainContent?: boolean | undefined;
|
|
16
|
-
includeMetadata?: boolean | undefined;
|
|
17
|
-
maxContentLength?: number | undefined;
|
|
18
|
-
format?: 'jsonl' | 'markdown' | undefined;
|
|
19
|
-
}
|
|
20
|
-
export interface FetchLinksInput extends RequestOptions {
|
|
21
|
-
url: string;
|
|
22
|
-
includeExternal?: boolean | undefined;
|
|
23
|
-
includeInternal?: boolean | undefined;
|
|
24
|
-
/** Maximum number of links to return */
|
|
25
|
-
maxLinks?: number | undefined;
|
|
26
|
-
/** Regex pattern to filter links (matches against href) */
|
|
27
|
-
filterPattern?: string | undefined;
|
|
28
|
-
/** Include image links (img src attributes) */
|
|
29
|
-
includeImages?: boolean | undefined;
|
|
30
|
-
}
|
|
31
|
-
export interface FetchMarkdownInput extends RequestOptions {
|
|
32
|
-
url: string;
|
|
33
|
-
extractMainContent?: boolean | undefined;
|
|
34
|
-
includeMetadata?: boolean | undefined;
|
|
35
|
-
/** Maximum content length in characters */
|
|
36
|
-
maxContentLength?: number | undefined;
|
|
37
|
-
/** Generate table of contents from headings */
|
|
38
|
-
generateToc?: boolean | undefined;
|
|
39
|
-
}
|
|
40
|
-
export interface FetchUrlsInput extends RequestOptions {
|
|
41
|
-
urls: string[];
|
|
42
|
-
extractMainContent?: boolean | undefined;
|
|
43
|
-
includeMetadata?: boolean | undefined;
|
|
44
|
-
maxContentLength?: number | undefined;
|
|
45
|
-
format?: 'jsonl' | 'markdown' | undefined;
|
|
46
|
-
concurrency?: number | undefined;
|
|
47
|
-
continueOnError?: boolean | undefined;
|
|
48
|
-
}
|
|
49
|
-
//# sourceMappingURL=schemas.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"schemas.d.ts","sourceRoot":"","sources":["../../src/types/schemas.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,iDAAiD;AACjD,MAAM,WAAW,cAAc;IAC7B,0CAA0C;IAC1C,aAAa,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,SAAS,CAAC;IACnD,mDAAmD;IACnD,OAAO,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC7B,sCAAsC;IACtC,OAAO,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;CAC9B;AAED,MAAM,WAAW,aAAc,SAAQ,cAAc;IACnD,GAAG,EAAE,MAAM,CAAC;IACZ,kBAAkB,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACzC,eAAe,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACtC,gBAAgB,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACtC,MAAM,CAAC,EAAE,OAAO,GAAG,UAAU,GAAG,SAAS,CAAC;CAC3C;AAED,MAAM,WAAW,eAAgB,SAAQ,cAAc;IACrD,GAAG,EAAE,MAAM,CAAC;IACZ,eAAe,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACtC,eAAe,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACtC,wCAAwC;IACxC,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC9B,2DAA2D;IAC3D,aAAa,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACnC,+CAA+C;IAC/C,aAAa,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;CACrC;AAED,MAAM,WAAW,kBAAmB,SAAQ,cAAc;IACxD,GAAG,EAAE,MAAM,CAAC;IACZ,kBAAkB,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACzC,eAAe,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACtC,2CAA2C;IAC3C,gBAAgB,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACtC,+CAA+C;IAC/C,WAAW,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;CACnC;AAED,MAAM,WAAW,cAAe,SAAQ,cAAc;IACpD,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,kBAAkB,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACzC,eAAe,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACtC,gBAAgB,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACtC,MAAM,CAAC,EAAE,OAAO,GAAG,UAAU,GAAG,SAAS,CAAC;IAC1C,WAAW,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACjC,eAAe,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;CACvC"}
|
package/dist/types/schemas.js
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"schemas.js","sourceRoot":"","sources":["../../src/types/schemas.ts"],"names":[],"mappings":"AAAA;;GAEG"}
|
|
@@ -1,7 +0,0 @@
|
|
|
1
|
-
export interface CachedPayload {
|
|
2
|
-
content?: string;
|
|
3
|
-
markdown?: string;
|
|
4
|
-
title?: string;
|
|
5
|
-
}
|
|
6
|
-
export declare function parseCachedPayload(raw: string): CachedPayload | null;
|
|
7
|
-
export declare function resolveCachedPayloadContent(payload: CachedPayload): string | null;
|
|
@@ -1,36 +0,0 @@
|
|
|
1
|
-
import { isRecord } from './guards.js';
|
|
2
|
-
export function parseCachedPayload(raw) {
|
|
3
|
-
try {
|
|
4
|
-
const parsed = JSON.parse(raw);
|
|
5
|
-
return isCachedPayload(parsed) ? parsed : null;
|
|
6
|
-
}
|
|
7
|
-
catch {
|
|
8
|
-
return null;
|
|
9
|
-
}
|
|
10
|
-
}
|
|
11
|
-
export function resolveCachedPayloadContent(payload) {
|
|
12
|
-
if (typeof payload.markdown === 'string') {
|
|
13
|
-
return payload.markdown;
|
|
14
|
-
}
|
|
15
|
-
if (typeof payload.content === 'string') {
|
|
16
|
-
return payload.content;
|
|
17
|
-
}
|
|
18
|
-
return null;
|
|
19
|
-
}
|
|
20
|
-
function hasOptionalStringProperty(value, key) {
|
|
21
|
-
const prop = value[key];
|
|
22
|
-
if (prop === undefined)
|
|
23
|
-
return true;
|
|
24
|
-
return typeof prop === 'string';
|
|
25
|
-
}
|
|
26
|
-
function isCachedPayload(value) {
|
|
27
|
-
if (!isRecord(value))
|
|
28
|
-
return false;
|
|
29
|
-
if (!hasOptionalStringProperty(value, 'content'))
|
|
30
|
-
return false;
|
|
31
|
-
if (!hasOptionalStringProperty(value, 'markdown'))
|
|
32
|
-
return false;
|
|
33
|
-
if (!hasOptionalStringProperty(value, 'title'))
|
|
34
|
-
return false;
|
|
35
|
-
return true;
|
|
36
|
-
}
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
export declare function throwIfAborted(signal: AbortSignal | undefined, url: string, stage: string): void;
|