@lingjingai/scriptctl 0.34.0 → 0.36.0
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/README.md +30 -5
- package/changes/0.35.0.md +22 -0
- package/changes/0.36.0.md +22 -0
- package/changes/unreleased.md +3 -5
- package/dist/cli.d.ts +2 -1
- package/dist/cli.js +56 -128
- package/dist/cli.js.map +1 -1
- package/dist/common.d.ts +0 -16
- package/dist/common.js +2 -342
- package/dist/common.js.map +1 -1
- package/dist/domain/ingest/apply-plan.d.ts +23 -0
- package/dist/domain/ingest/apply-plan.js +210 -0
- package/dist/domain/ingest/apply-plan.js.map +1 -0
- package/dist/domain/ingest/assembler.d.ts +57 -0
- package/dist/domain/ingest/assembler.js +181 -0
- package/dist/domain/ingest/assembler.js.map +1 -0
- package/dist/domain/ingest/chunker.d.ts +9 -0
- package/dist/domain/ingest/chunker.js +38 -0
- package/dist/domain/ingest/chunker.js.map +1 -0
- package/dist/domain/ingest/evidence.d.ts +5 -0
- package/dist/domain/ingest/evidence.js +66 -0
- package/dist/domain/ingest/evidence.js.map +1 -0
- package/dist/domain/ingest/govern.d.ts +3 -0
- package/dist/domain/ingest/govern.js +127 -0
- package/dist/domain/ingest/govern.js.map +1 -0
- package/dist/domain/ingest/jsonl-parser.d.ts +6 -0
- package/dist/domain/ingest/jsonl-parser.js +187 -0
- package/dist/domain/ingest/jsonl-parser.js.map +1 -0
- package/dist/domain/ingest/markdown-parser.d.ts +6 -0
- package/dist/domain/ingest/markdown-parser.js +186 -0
- package/dist/domain/ingest/markdown-parser.js.map +1 -0
- package/dist/domain/ingest/records.d.ts +71 -0
- package/dist/domain/ingest/records.js +9 -0
- package/dist/domain/ingest/records.js.map +1 -0
- package/dist/domain/ingest/renumber.d.ts +14 -0
- package/dist/domain/ingest/renumber.js +130 -0
- package/dist/domain/ingest/renumber.js.map +1 -0
- package/dist/domain/ingest/script-v3.d.ts +94 -0
- package/dist/domain/ingest/script-v3.js +7 -0
- package/dist/domain/ingest/script-v3.js.map +1 -0
- package/dist/domain/ingest/state-reconcile.d.ts +11 -0
- package/dist/domain/ingest/state-reconcile.js +90 -0
- package/dist/domain/ingest/state-reconcile.js.map +1 -0
- package/dist/domain/ingest/validation.d.ts +11 -0
- package/dist/domain/ingest/validation.js +168 -0
- package/dist/domain/ingest/validation.js.map +1 -0
- package/dist/domain/ingest/video-aggregate.d.ts +23 -0
- package/dist/domain/ingest/video-aggregate.js +89 -0
- package/dist/domain/ingest/video-aggregate.js.map +1 -0
- package/dist/domain/ingest/video-apply.d.ts +37 -0
- package/dist/domain/ingest/video-apply.js +236 -0
- package/dist/domain/ingest/video-apply.js.map +1 -0
- package/dist/domain/ingest/video-md-parser.d.ts +2 -0
- package/dist/domain/ingest/video-md-parser.js +233 -0
- package/dist/domain/ingest/video-md-parser.js.map +1 -0
- package/dist/domain/ingest/video-records.d.ts +56 -0
- package/dist/domain/ingest/video-records.js +9 -0
- package/dist/domain/ingest/video-records.js.map +1 -0
- package/dist/domain/loom/index.d.ts +2 -1
- package/dist/domain/loom/index.js +12 -10
- package/dist/domain/loom/index.js.map +1 -1
- package/dist/domain/loom/types.d.ts +3 -2
- package/dist/domain/script/lookups.d.ts +3 -2
- package/dist/domain/script/lookups.js +25 -8
- package/dist/domain/script/lookups.js.map +1 -1
- package/dist/domain/script/patch/apply.js +2 -2
- package/dist/domain/script/patch/apply.js.map +1 -1
- package/dist/domain/script/patch/helpers.d.ts +3 -7
- package/dist/domain/script/patch/helpers.js +23 -96
- package/dist/domain/script/patch/helpers.js.map +1 -1
- package/dist/domain/script/patch/ops-action.d.ts +0 -2
- package/dist/domain/script/patch/ops-action.js +34 -100
- package/dist/domain/script/patch/ops-action.js.map +1 -1
- package/dist/domain/script/patch/ops-asset.d.ts +1 -0
- package/dist/domain/script/patch/ops-asset.js +102 -155
- package/dist/domain/script/patch/ops-asset.js.map +1 -1
- package/dist/domain/script/patch/ops-dialogue.d.ts +1 -4
- package/dist/domain/script/patch/ops-dialogue.js +43 -220
- package/dist/domain/script/patch/ops-dialogue.js.map +1 -1
- package/dist/domain/script/patch/ops-extend.js +8 -2
- package/dist/domain/script/patch/ops-extend.js.map +1 -1
- package/dist/domain/script/patch/ops-meta.js.map +1 -1
- package/dist/domain/script/patch/ops-scene.js +26 -39
- package/dist/domain/script/patch/ops-scene.js.map +1 -1
- package/dist/domain/script/patch/ops-state.d.ts +3 -3
- package/dist/domain/script/patch/ops-state.js +33 -49
- package/dist/domain/script/patch/ops-state.js.map +1 -1
- package/dist/domain/script/patch/registry.js +9 -13
- package/dist/domain/script/patch/registry.js.map +1 -1
- package/dist/domain/script/refs.js +23 -111
- package/dist/domain/script/refs.js.map +1 -1
- package/dist/domain/script/scene-refs.d.ts +3 -0
- package/dist/domain/script/scene-refs.js +35 -0
- package/dist/domain/script/scene-refs.js.map +1 -0
- package/dist/domain/script/schema.js +17 -21
- package/dist/domain/script/schema.js.map +1 -1
- package/dist/domain/script/text-util.d.ts +7 -0
- package/dist/domain/script/text-util.js +91 -0
- package/dist/domain/script/text-util.js.map +1 -0
- package/dist/help-text.js +145 -414
- package/dist/help-text.js.map +1 -1
- package/dist/infra/converters.js +1 -1
- package/dist/infra/converters.js.map +1 -1
- package/dist/infra/episode-title.d.ts +4 -0
- package/dist/infra/episode-title.js +129 -0
- package/dist/infra/episode-title.js.map +1 -0
- package/dist/infra/ingest/review-renderer.d.ts +1 -0
- package/dist/infra/ingest/review-renderer.js +431 -0
- package/dist/infra/ingest/review-renderer.js.map +1 -0
- package/dist/infra/ingest/source-resolver.d.ts +21 -0
- package/dist/infra/ingest/source-resolver.js +51 -0
- package/dist/infra/ingest/source-resolver.js.map +1 -0
- package/dist/infra/ingest/text-reader.d.ts +3 -0
- package/dist/infra/ingest/text-reader.js +27 -0
- package/dist/infra/ingest/text-reader.js.map +1 -0
- package/dist/infra/ingest/video-discovery.d.ts +8 -0
- package/dist/infra/ingest/video-discovery.js +60 -0
- package/dist/infra/ingest/video-discovery.js.map +1 -0
- package/dist/infra/ingest/workspace-store.d.ts +18 -0
- package/dist/infra/ingest/workspace-store.js +70 -0
- package/dist/infra/ingest/workspace-store.js.map +1 -0
- package/dist/infra/llm/anthropic-model.d.ts +11 -0
- package/dist/infra/llm/anthropic-model.js +109 -0
- package/dist/infra/llm/anthropic-model.js.map +1 -0
- package/dist/infra/llm/gemini-upload.d.ts +9 -0
- package/dist/infra/llm/gemini-upload.js +94 -0
- package/dist/infra/llm/gemini-upload.js.map +1 -0
- package/dist/infra/llm/gemini-video-model.d.ts +47 -0
- package/dist/infra/llm/gemini-video-model.js +167 -0
- package/dist/infra/llm/gemini-video-model.js.map +1 -0
- package/dist/infra/llm/openai-compatible-model.d.ts +13 -0
- package/dist/infra/llm/openai-compatible-model.js +143 -0
- package/dist/infra/llm/openai-compatible-model.js.map +1 -0
- package/dist/infra/llm/proxy.d.ts +1 -0
- package/dist/infra/llm/proxy.js +30 -0
- package/dist/infra/llm/proxy.js.map +1 -0
- package/dist/infra/llm/retry.d.ts +5 -0
- package/dist/infra/llm/retry.js +20 -0
- package/dist/infra/llm/retry.js.map +1 -0
- package/dist/llm/composition.d.ts +5 -0
- package/dist/llm/composition.js +5 -0
- package/dist/llm/composition.js.map +1 -0
- package/dist/llm/config.d.ts +26 -0
- package/dist/llm/config.js +201 -0
- package/dist/llm/config.js.map +1 -0
- package/dist/llm/registry.d.ts +44 -0
- package/dist/llm/registry.js +48 -0
- package/dist/llm/registry.js.map +1 -0
- package/dist/llm/router.d.ts +23 -0
- package/dist/llm/router.js +0 -0
- package/dist/llm/router.js.map +1 -0
- package/dist/llm/tasks/ingest/asset-curate.d.ts +13 -0
- package/dist/llm/tasks/ingest/asset-curate.js +45 -0
- package/dist/llm/tasks/ingest/asset-curate.js.map +1 -0
- package/dist/llm/tasks/ingest/asset-extract.d.ts +12 -0
- package/dist/llm/tasks/ingest/asset-extract.js +51 -0
- package/dist/llm/tasks/ingest/asset-extract.js.map +1 -0
- package/dist/llm/tasks/ingest/asset-govern.d.ts +22 -0
- package/dist/llm/tasks/ingest/asset-govern.js +143 -0
- package/dist/llm/tasks/ingest/asset-govern.js.map +1 -0
- package/dist/llm/tasks/ingest/mention-resolve.d.ts +9 -0
- package/dist/llm/tasks/ingest/mention-resolve.js +39 -0
- package/dist/llm/tasks/ingest/mention-resolve.js.map +1 -0
- package/dist/llm/tasks/ingest/plot-speaker.d.ts +11 -0
- package/dist/llm/tasks/ingest/plot-speaker.js +72 -0
- package/dist/llm/tasks/ingest/plot-speaker.js.map +1 -0
- package/dist/llm/tasks/ingest/roster-merge.d.ts +46 -0
- package/dist/llm/tasks/ingest/roster-merge.js +235 -0
- package/dist/llm/tasks/ingest/roster-merge.js.map +1 -0
- package/dist/llm/tasks/ingest/state-merge.d.ts +10 -0
- package/dist/llm/tasks/ingest/state-merge.js +84 -0
- package/dist/llm/tasks/ingest/state-merge.js.map +1 -0
- package/dist/llm/tasks/ingest/state-reconcile.d.ts +8 -0
- package/dist/llm/tasks/ingest/state-reconcile.js +44 -0
- package/dist/llm/tasks/ingest/state-reconcile.js.map +1 -0
- package/dist/llm/tasks/ingest/text-normalize.d.ts +10 -0
- package/dist/llm/tasks/ingest/text-normalize.js +89 -0
- package/dist/llm/tasks/ingest/text-normalize.js.map +1 -0
- package/dist/llm/tasks/ingest/video-correct.d.ts +30 -0
- package/dist/llm/tasks/ingest/video-correct.js +144 -0
- package/dist/llm/tasks/ingest/video-correct.js.map +1 -0
- package/dist/llm/tasks/ingest/video-transcribe.d.ts +4 -0
- package/dist/llm/tasks/ingest/video-transcribe.js +105 -0
- package/dist/llm/tasks/ingest/video-transcribe.js.map +1 -0
- package/dist/llm/tasks/schemas.d.ts +5 -0
- package/dist/llm/tasks/schemas.js +107 -0
- package/dist/llm/tasks/schemas.js.map +1 -0
- package/dist/llm/types.d.ts +97 -0
- package/dist/llm/types.js +13 -0
- package/dist/llm/types.js.map +1 -0
- package/dist/usecases/doctor.d.ts +1 -1
- package/dist/usecases/doctor.js +36 -9
- package/dist/usecases/doctor.js.map +1 -1
- package/dist/usecases/ingest/command.d.ts +5 -0
- package/dist/usecases/ingest/command.js +57 -0
- package/dist/usecases/ingest/command.js.map +1 -0
- package/dist/usecases/ingest/fanout.d.ts +10 -0
- package/dist/usecases/ingest/fanout.js +23 -0
- package/dist/usecases/ingest/fanout.js.map +1 -0
- package/dist/usecases/ingest/pass4-mention-resolve.d.ts +11 -0
- package/dist/usecases/ingest/pass4-mention-resolve.js +52 -0
- package/dist/usecases/ingest/pass4-mention-resolve.js.map +1 -0
- package/dist/usecases/ingest/pass5-extract.d.ts +7 -0
- package/dist/usecases/ingest/pass5-extract.js +50 -0
- package/dist/usecases/ingest/pass5-extract.js.map +1 -0
- package/dist/usecases/ingest/pass5-govern.d.ts +13 -0
- package/dist/usecases/ingest/pass5-govern.js +155 -0
- package/dist/usecases/ingest/pass5-govern.js.map +1 -0
- package/dist/usecases/ingest/pipeline.d.ts +28 -0
- package/dist/usecases/ingest/pipeline.js +144 -0
- package/dist/usecases/ingest/pipeline.js.map +1 -0
- package/dist/usecases/ingest/publish-command.d.ts +4 -0
- package/dist/usecases/ingest/publish-command.js +77 -0
- package/dist/usecases/ingest/publish-command.js.map +1 -0
- package/dist/usecases/ingest/status.d.ts +39 -0
- package/dist/usecases/ingest/status.js +155 -0
- package/dist/usecases/ingest/status.js.map +1 -0
- package/dist/usecases/ingest/video-pass3.d.ts +26 -0
- package/dist/usecases/ingest/video-pass3.js +0 -0
- package/dist/usecases/ingest/video-pass3.js.map +1 -0
- package/dist/usecases/ingest/video-pipeline.d.ts +32 -0
- package/dist/usecases/ingest/video-pipeline.js +202 -0
- package/dist/usecases/ingest/video-pipeline.js.map +1 -0
- package/dist/usecases/ingest/view-command.d.ts +4 -0
- package/dist/usecases/ingest/view-command.js +51 -0
- package/dist/usecases/ingest/view-command.js.map +1 -0
- package/dist/usecases/script/actions.js +2 -2
- package/dist/usecases/script/actions.js.map +1 -1
- package/dist/usecases/script/actor.js.map +1 -1
- package/dist/usecases/script/actors.js +1 -0
- package/dist/usecases/script/actors.js.map +1 -1
- package/dist/usecases/script/add-actor.js.map +1 -1
- package/dist/usecases/script/add-episode.js.map +1 -1
- package/dist/usecases/script/add-location.js.map +1 -1
- package/dist/usecases/script/add-prop.js.map +1 -1
- package/dist/usecases/script/alias.js.map +1 -1
- package/dist/usecases/script/assets.js +2 -1
- package/dist/usecases/script/assets.js.map +1 -1
- package/dist/usecases/script/context.js.map +1 -1
- package/dist/usecases/script/create.js +3 -6
- package/dist/usecases/script/create.js.map +1 -1
- package/dist/usecases/script/delete.js.map +1 -1
- package/dist/usecases/script/describe.js.map +1 -1
- package/dist/usecases/script/dialogue.js +6 -13
- package/dist/usecases/script/dialogue.js.map +1 -1
- package/dist/usecases/script/do.js +1 -1
- package/dist/usecases/script/do.js.map +1 -1
- package/dist/usecases/script/episodes.js.map +1 -1
- package/dist/usecases/script/{add-speaker.d.ts → importance.d.ts} +1 -1
- package/dist/usecases/script/importance.js +17 -0
- package/dist/usecases/script/importance.js.map +1 -0
- package/dist/usecases/script/insert.js.map +1 -1
- package/dist/usecases/script/issues.js.map +1 -1
- package/dist/usecases/script/lib.d.ts +19 -6
- package/dist/usecases/script/lib.js +173 -99
- package/dist/usecases/script/lib.js.map +1 -1
- package/dist/usecases/script/locations.js +1 -0
- package/dist/usecases/script/locations.js.map +1 -1
- package/dist/usecases/script/merge.js.map +1 -1
- package/dist/usecases/script/move.js.map +1 -1
- package/dist/usecases/script/props.js +1 -0
- package/dist/usecases/script/props.js.map +1 -1
- package/dist/usecases/script/refs.js.map +1 -1
- package/dist/usecases/script/rename.js.map +1 -1
- package/dist/usecases/script/replace.js.map +1 -1
- package/dist/usecases/script/role.js.map +1 -1
- package/dist/usecases/script/scenes.js +8 -15
- package/dist/usecases/script/scenes.js.map +1 -1
- package/dist/usecases/script/script-patch.js +1 -1
- package/dist/usecases/script/script-patch.js.map +1 -1
- package/dist/usecases/script/script-validate.js +1 -1
- package/dist/usecases/script/script-validate.js.map +1 -1
- package/dist/usecases/script/session.d.ts +4 -4
- package/dist/usecases/script/session.js +57 -78
- package/dist/usecases/script/session.js.map +1 -1
- package/dist/usecases/script/split.js.map +1 -1
- package/dist/usecases/script/state-add.js.map +1 -1
- package/dist/usecases/script/state-delete.js.map +1 -1
- package/dist/usecases/script/state-rename.js.map +1 -1
- package/dist/usecases/script/states.js +58 -13
- package/dist/usecases/script/states.js.map +1 -1
- package/dist/usecases/script/summary.js +8 -0
- package/dist/usecases/script/summary.js.map +1 -1
- package/dist/usecases/script/synopsis-generate.d.ts +4 -5
- package/dist/usecases/script/synopsis-generate.js +30 -46
- package/dist/usecases/script/synopsis-generate.js.map +1 -1
- package/dist/usecases/script/transition.js.map +1 -1
- package/dist/usecases/script/type.js.map +1 -1
- package/dist/usecases/script/worldview.js.map +1 -1
- package/package.json +11 -3
- package/scripts/install-skill.mjs +120 -0
- package/skills/scriptctl/SKILL.md +293 -0
- package/skills/scriptctl/references/atomic-write-workflow.md +117 -0
- package/skills/scriptctl/references/ingest-workflow.md +85 -0
- package/skills/scriptctl/references/state-reference-repair.md +59 -0
- package/dist/domain/direct/runner.d.ts +0 -25
- package/dist/domain/direct/runner.js +0 -88
- package/dist/domain/direct/runner.js.map +0 -1
- package/dist/domain/direct/stage.d.ts +0 -108
- package/dist/domain/direct/stage.js +0 -134
- package/dist/domain/direct/stage.js.map +0 -1
- package/dist/domain/direct/stages/asset-curation.d.ts +0 -2
- package/dist/domain/direct/stages/asset-curation.js +0 -10
- package/dist/domain/direct/stages/asset-curation.js.map +0 -1
- package/dist/domain/direct/stages/batch-extract.d.ts +0 -2
- package/dist/domain/direct/stages/batch-extract.js +0 -12
- package/dist/domain/direct/stages/batch-extract.js.map +0 -1
- package/dist/domain/direct/stages/batch-plan.d.ts +0 -2
- package/dist/domain/direct/stages/batch-plan.js +0 -9
- package/dist/domain/direct/stages/batch-plan.js.map +0 -1
- package/dist/domain/direct/stages/episode-merge.d.ts +0 -2
- package/dist/domain/direct/stages/episode-merge.js +0 -11
- package/dist/domain/direct/stages/episode-merge.js.map +0 -1
- package/dist/domain/direct/stages/episode-plan.d.ts +0 -2
- package/dist/domain/direct/stages/episode-plan.js +0 -9
- package/dist/domain/direct/stages/episode-plan.js.map +0 -1
- package/dist/domain/direct/stages/episode-synopsis.d.ts +0 -2
- package/dist/domain/direct/stages/episode-synopsis.js +0 -12
- package/dist/domain/direct/stages/episode-synopsis.js.map +0 -1
- package/dist/domain/direct/stages/episode-titles.d.ts +0 -2
- package/dist/domain/direct/stages/episode-titles.js +0 -12
- package/dist/domain/direct/stages/episode-titles.js.map +0 -1
- package/dist/domain/direct/stages/index.d.ts +0 -19
- package/dist/domain/direct/stages/index.js +0 -41
- package/dist/domain/direct/stages/index.js.map +0 -1
- package/dist/domain/direct/stages/metadata.d.ts +0 -2
- package/dist/domain/direct/stages/metadata.js +0 -11
- package/dist/domain/direct/stages/metadata.js.map +0 -1
- package/dist/domain/direct/stages/script-merge.d.ts +0 -2
- package/dist/domain/direct/stages/script-merge.js +0 -15
- package/dist/domain/direct/stages/script-merge.js.map +0 -1
- package/dist/domain/direct/stages/script-synopsis.d.ts +0 -2
- package/dist/domain/direct/stages/script-synopsis.js +0 -12
- package/dist/domain/direct/stages/script-synopsis.js.map +0 -1
- package/dist/domain/direct/stages/source-prepare.d.ts +0 -2
- package/dist/domain/direct/stages/source-prepare.js +0 -9
- package/dist/domain/direct/stages/source-prepare.js.map +0 -1
- package/dist/domain/direct/stages/state-binding.d.ts +0 -2
- package/dist/domain/direct/stages/state-binding.js +0 -9
- package/dist/domain/direct/stages/state-binding.js.map +0 -1
- package/dist/domain/direct/stages/state-curation.d.ts +0 -2
- package/dist/domain/direct/stages/state-curation.js +0 -10
- package/dist/domain/direct/stages/state-curation.js.map +0 -1
- package/dist/domain/direct/stages/validate.d.ts +0 -2
- package/dist/domain/direct/stages/validate.js +0 -10
- package/dist/domain/direct/stages/validate.js.map +0 -1
- package/dist/domain/direct-core.d.ts +0 -313
- package/dist/domain/direct-core.js +0 -8516
- package/dist/domain/direct-core.js.map +0 -1
- package/dist/domain/script/validate.d.ts +0 -22
- package/dist/domain/script/validate.js +0 -997
- package/dist/domain/script/validate.js.map +0 -1
- package/dist/infra/providers.d.ts +0 -97
- package/dist/infra/providers.js +0 -1396
- package/dist/infra/providers.js.map +0 -1
- package/dist/usecases/direct.d.ts +0 -27
- package/dist/usecases/direct.js +0 -4172
- package/dist/usecases/direct.js.map +0 -1
- package/dist/usecases/parse.d.ts +0 -15
- package/dist/usecases/parse.js +0 -429
- package/dist/usecases/parse.js.map +0 -1
- package/dist/usecases/script/add-speaker.js +0 -22
- package/dist/usecases/script/add-speaker.js.map +0 -1
- package/dist/usecases/script/export.d.ts +0 -3
- package/dist/usecases/script/export.js +0 -111
- package/dist/usecases/script/export.js.map +0 -1
- package/dist/usecases/script/overlap.d.ts +0 -3
- package/dist/usecases/script/overlap.js +0 -21
- package/dist/usecases/script/overlap.js.map +0 -1
- package/dist/usecases/script/speakers.d.ts +0 -3
- package/dist/usecases/script/speakers.js +0 -28
- package/dist/usecases/script/speakers.js.map +0 -1
- package/dist/usecases/script/state-change.d.ts +0 -3
- package/dist/usecases/script/state-change.js +0 -24
- package/dist/usecases/script/state-change.js.map +0 -1
|
@@ -0,0 +1,127 @@
|
|
|
1
|
+
// Deterministic governance core (§9.6/§9.8/§9.9 minus LLM). Given the provisional
|
|
2
|
+
// initial script it: (1) assigns each asset's `default` state by highest explicit
|
|
3
|
+
// frequency (tie → codepoint-smallest state name) and (2) materializes inherited
|
|
4
|
+
// state onto every entity reference so the final product is self-contained
|
|
5
|
+
// (§15.2). Identity merge / drop / audit are LLM stages layered on separately;
|
|
6
|
+
// this keep-all core always produces a valid, materialized v3 doc.
|
|
7
|
+
const isObj = (v) => typeof v === "object" && v !== null && !Array.isArray(v);
|
|
8
|
+
const isArr = (v) => Array.isArray(v);
|
|
9
|
+
function assetIndex(list) {
|
|
10
|
+
const map = new Map();
|
|
11
|
+
if (!isArr(list))
|
|
12
|
+
return map;
|
|
13
|
+
for (const a of list) {
|
|
14
|
+
if (!isObj(a))
|
|
15
|
+
continue;
|
|
16
|
+
const id = a.actor_id ?? a.location_id ?? a.prop_id;
|
|
17
|
+
if (typeof id === "string")
|
|
18
|
+
map.set(id, a);
|
|
19
|
+
}
|
|
20
|
+
return map;
|
|
21
|
+
}
|
|
22
|
+
// Walk the doc in order, collecting entity refs (scene cast/props, scene
|
|
23
|
+
// location, dialogue speakers) as mutable holders plus their explicit state id.
|
|
24
|
+
function collectRefs(doc) {
|
|
25
|
+
const refs = [];
|
|
26
|
+
const push = (holder, idField) => {
|
|
27
|
+
if (!isObj(holder) || typeof holder[idField] !== "string")
|
|
28
|
+
return;
|
|
29
|
+
refs.push({ assetId: holder[idField], holder, explicit: typeof holder.state_id === "string" ? holder.state_id : undefined });
|
|
30
|
+
};
|
|
31
|
+
for (const ep of isArr(doc.episodes) ? doc.episodes : []) {
|
|
32
|
+
if (!isObj(ep) || !isArr(ep.scenes))
|
|
33
|
+
continue;
|
|
34
|
+
for (const sc of ep.scenes) {
|
|
35
|
+
if (!isObj(sc))
|
|
36
|
+
continue;
|
|
37
|
+
push(sc.location, "location_id");
|
|
38
|
+
for (const a of isArr(sc.actors) ? sc.actors : [])
|
|
39
|
+
push(a, "actor_id");
|
|
40
|
+
for (const p of isArr(sc.props) ? sc.props : [])
|
|
41
|
+
push(p, "prop_id");
|
|
42
|
+
for (const ac of isArr(sc.actions) ? sc.actions : []) {
|
|
43
|
+
if (isObj(ac))
|
|
44
|
+
push(ac.speaker, "actor_id");
|
|
45
|
+
}
|
|
46
|
+
}
|
|
47
|
+
}
|
|
48
|
+
return refs;
|
|
49
|
+
}
|
|
50
|
+
function pickDefault(states, counts) {
|
|
51
|
+
const candidates = states.filter((s) => (counts.get(s.state_id) ?? 0) > 0);
|
|
52
|
+
if (candidates.length === 0)
|
|
53
|
+
return null;
|
|
54
|
+
let best = candidates[0];
|
|
55
|
+
for (const s of candidates.slice(1)) {
|
|
56
|
+
const c = counts.get(s.state_id) ?? 0;
|
|
57
|
+
const bc = counts.get(best.state_id) ?? 0;
|
|
58
|
+
if (c > bc || (c === bc && s.state_name < best.state_name))
|
|
59
|
+
best = s;
|
|
60
|
+
}
|
|
61
|
+
return best.state_id;
|
|
62
|
+
}
|
|
63
|
+
export function governDeterministic(input) {
|
|
64
|
+
if (!isObj(input))
|
|
65
|
+
throw new Error("govern expects a script object");
|
|
66
|
+
const doc = JSON.parse(JSON.stringify(input)); // pure: never mutate input
|
|
67
|
+
const actors = assetIndex(doc.actors);
|
|
68
|
+
const locations = assetIndex(doc.locations);
|
|
69
|
+
const props = assetIndex(doc.props);
|
|
70
|
+
const all = new Map([...actors, ...locations, ...props]);
|
|
71
|
+
const refs = collectRefs(doc);
|
|
72
|
+
// 1. Count explicit state usage per asset.
|
|
73
|
+
const counts = new Map();
|
|
74
|
+
for (const ref of refs) {
|
|
75
|
+
if (ref.explicit === undefined)
|
|
76
|
+
continue;
|
|
77
|
+
if (!counts.has(ref.assetId))
|
|
78
|
+
counts.set(ref.assetId, new Map());
|
|
79
|
+
const m = counts.get(ref.assetId);
|
|
80
|
+
m.set(ref.explicit, (m.get(ref.explicit) ?? 0) + 1);
|
|
81
|
+
}
|
|
82
|
+
// 2. Per asset: pick default winner, rebuild states[] so exactly one 'default'.
|
|
83
|
+
const remap = new Map(); // assetId -> (oldStateId -> newStateId)
|
|
84
|
+
for (const [assetId, asset] of all) {
|
|
85
|
+
const states = (isArr(asset.states) ? asset.states : []).filter(isObj);
|
|
86
|
+
const local = new Map();
|
|
87
|
+
const winner = pickDefault(states, counts.get(assetId) ?? new Map());
|
|
88
|
+
let out;
|
|
89
|
+
if (winner === null) {
|
|
90
|
+
// No explicit states used: keep a single synthetic default.
|
|
91
|
+
out = [{ state_id: "default", state_name: states.find((s) => s.state_id === "default")?.state_name ?? "default", description: "" }];
|
|
92
|
+
}
|
|
93
|
+
else {
|
|
94
|
+
out = [];
|
|
95
|
+
for (const s of states) {
|
|
96
|
+
if (s.state_id === "default")
|
|
97
|
+
continue; // drop synthetic default
|
|
98
|
+
if (s.state_id === winner) {
|
|
99
|
+
local.set(s.state_id, "default");
|
|
100
|
+
out.unshift({ state_id: "default", state_name: s.state_name, description: s.description ?? "" });
|
|
101
|
+
}
|
|
102
|
+
else {
|
|
103
|
+
local.set(s.state_id, s.state_id);
|
|
104
|
+
out.push({ state_id: s.state_id, state_name: s.state_name, description: s.description ?? "" });
|
|
105
|
+
}
|
|
106
|
+
}
|
|
107
|
+
}
|
|
108
|
+
asset.states = out;
|
|
109
|
+
remap.set(assetId, local);
|
|
110
|
+
}
|
|
111
|
+
// 3. Materialize: walk refs in order, inherit last explicit state per asset.
|
|
112
|
+
const last = new Map();
|
|
113
|
+
for (const ref of refs) {
|
|
114
|
+
const local = remap.get(ref.assetId);
|
|
115
|
+
let resolved;
|
|
116
|
+
if (ref.explicit !== undefined) {
|
|
117
|
+
resolved = local?.get(ref.explicit) ?? ref.explicit;
|
|
118
|
+
last.set(ref.assetId, resolved);
|
|
119
|
+
}
|
|
120
|
+
else {
|
|
121
|
+
resolved = last.get(ref.assetId) ?? "default";
|
|
122
|
+
}
|
|
123
|
+
ref.holder.state_id = resolved;
|
|
124
|
+
}
|
|
125
|
+
return doc;
|
|
126
|
+
}
|
|
127
|
+
//# sourceMappingURL=govern.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"govern.js","sourceRoot":"","sources":["../../../src/domain/ingest/govern.ts"],"names":[],"mappings":"AAAA,kFAAkF;AAClF,kFAAkF;AAClF,iFAAiF;AACjF,2EAA2E;AAC3E,+EAA+E;AAC/E,mEAAmE;AAGnE,MAAM,KAAK,GAAG,CAAC,CAAU,EAAY,EAAE,CAAC,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;AACjG,MAAM,KAAK,GAAG,CAAC,CAAU,EAAkB,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;AAQ/D,SAAS,UAAU,CAAC,IAAa;IAC/B,MAAM,GAAG,GAAG,IAAI,GAAG,EAAe,CAAC;IACnC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC;QAAE,OAAO,GAAG,CAAC;IAC7B,KAAK,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC;QACrB,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;YAAE,SAAS;QACxB,MAAM,EAAE,GAAG,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,OAAO,CAAC;QACpD,IAAI,OAAO,EAAE,KAAK,QAAQ;YAAE,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;IAC7C,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,yEAAyE;AACzE,gFAAgF;AAChF,SAAS,WAAW,CAAC,GAAQ;IAC3B,MAAM,IAAI,GAAU,EAAE,CAAC;IACvB,MAAM,IAAI,GAAG,CAAC,MAAe,EAAE,OAAe,EAAE,EAAE;QAChD,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,IAAI,OAAO,MAAM,CAAC,OAAO,CAAC,KAAK,QAAQ;YAAE,OAAO;QAClE,IAAI,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,OAAO,CAAW,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,MAAM,CAAC,QAAQ,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS,EAAE,CAAC,CAAC;IACzI,CAAC,CAAC;IACF,KAAK,MAAM,EAAE,IAAI,KAAK,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;QACzD,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,MAAM,CAAC;YAAE,SAAS;QAC9C,KAAK,MAAM,EAAE,IAAI,EAAE,CAAC,MAAM,EAAE,CAAC;YAC3B,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBAAE,SAAS;YACzB,IAAI,CAAC,EAAE,CAAC,QAAQ,EAAE,aAAa,CAAC,CAAC;YACjC,KAAK,MAAM,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE;gBAAE,IAAI,CAAC,CAAC,EAAE,UAAU,CAAC,CAAC;YACvE,KAAK,MAAM,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE;gBAAE,IAAI,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;YACpE,KAAK,MAAM,EAAE,IAAI,KAAK,CAAC,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;gBACrD,IAAI,KAAK,CAAC,EAAE,CAAC;oBAAE,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;YAC9C,CAAC;QACH,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED,SAAS,WAAW,CAAC,MAAoB,EAAE,MAA2B;IACpE,MAAM,UAAU,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IAC3E,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IACzC,IAAI,IAAI,GAAG,UAAU,CAAC,CAAC,CAAE,CAAC;IAC1B,KAAK,MAAM,CAAC,IAAI,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;QACpC,MAAM,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;QACtC,MAAM,EAAE,GAAG,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;QAC1C,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,UAAU,GAAG,IAAI,CAAC,UAAU,CAAC;YAAE,IAAI,GAAG,CAAC,CAAC;IACvE,CAAC;IACD,OAAO,IAAI,CAAC,QAAQ,CAAC;AACvB,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,KAAc;IAChD,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,gCAAgC,CAAC,CAAC;IACrE,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAQ,CAAC,CAAC,2BAA2B;IAEjF,MAAM,MAAM,GAAG,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IACtC,MAAM,SAAS,GAAG,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;IAC5C,MAAM,KAAK,GAAG,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IACpC,MAAM,GAAG,GAAG,IAAI,GAAG,CAAc,CAAC,GAAG,MAAM,EAAE,GAAG,SAAS,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC;IAEtE,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC;IAE9B,2CAA2C;IAC3C,MAAM,MAAM,GAAG,IAAI,GAAG,EAA+B,CAAC;IACtD,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,IAAI,GAAG,CAAC,QAAQ,KAAK,SAAS;YAAE,SAAS;QACzC,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC;YAAE,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,EAAE,IAAI,GAAG,EAAE,CAAC,CAAC;QACjE,MAAM,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAE,CAAC;QACnC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IACtD,CAAC;IAED,gFAAgF;IAChF,MAAM,KAAK,GAAG,IAAI,GAAG,EAA+B,CAAC,CAAC,wCAAwC;IAC9F,KAAK,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,IAAI,GAAG,EAAE,CAAC;QACnC,MAAM,MAAM,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,MAAM,CAAC,KAAK,CAA4B,CAAC;QAClG,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkB,CAAC;QACxC,MAAM,MAAM,GAAG,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,GAAG,EAAE,CAAC,CAAC;QACrE,IAAI,GAAiB,CAAC;QACtB,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;YACpB,4DAA4D;YAC5D,GAAG,GAAG,CAAC,EAAE,QAAQ,EAAE,SAAS,EAAE,UAAU,EAAE,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,KAAK,SAAS,CAAC,EAAE,UAAU,IAAI,SAAS,EAAE,WAAW,EAAE,EAAE,EAAE,CAAC,CAAC;QACtI,CAAC;aAAM,CAAC;YACN,GAAG,GAAG,EAAE,CAAC;YACT,KAAK,MAAM,CAAC,IAAI,MAAM,EAAE,CAAC;gBACvB,IAAI,CAAC,CAAC,QAAQ,KAAK,SAAS;oBAAE,SAAS,CAAC,yBAAyB;gBACjE,IAAI,CAAC,CAAC,QAAQ,KAAK,MAAM,EAAE,CAAC;oBAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC;oBAAC,GAAG,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,SAAS,EAAE,UAAU,EAAE,CAAC,CAAC,UAAU,EAAE,WAAW,EAAE,CAAC,CAAC,WAAW,IAAI,EAAE,EAAE,CAAC,CAAC;gBAAC,CAAC;qBAC7J,CAAC;oBAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC;oBAAC,GAAG,CAAC,IAAI,CAAC,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,UAAU,EAAE,CAAC,CAAC,UAAU,EAAE,WAAW,EAAE,CAAC,CAAC,WAAW,IAAI,EAAE,EAAE,CAAC,CAAC;gBAAC,CAAC;YAC7I,CAAC;QACH,CAAC;QACD,KAAK,CAAC,MAAM,GAAG,GAAG,CAAC;QACnB,KAAK,CAAC,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;IAC5B,CAAC;IAED,6EAA6E;IAC7E,MAAM,IAAI,GAAG,IAAI,GAAG,EAAkB,CAAC;IACvC,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACrC,IAAI,QAAgB,CAAC;QACrB,IAAI,GAAG,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC/B,QAAQ,GAAG,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,GAAG,CAAC,QAAQ,CAAC;YACpD,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;QAClC,CAAC;aAAM,CAAC;YACN,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,SAAS,CAAC;QAChD,CAAC;QACD,GAAG,CAAC,MAAM,CAAC,QAAQ,GAAG,QAAQ,CAAC;IACjC,CAAC;IAED,OAAO,GAAG,CAAC;AACb,CAAC"}
|
|
@@ -0,0 +1,187 @@
|
|
|
1
|
+
// Strict JSONL → typed record parser (§7). Rejects unknown fields, malformed
|
|
2
|
+
// records, missing/invalid source spans, and inline-tag syntax errors. Never
|
|
3
|
+
// repairs or downgrades: a bad line fails the batch (§12.3).
|
|
4
|
+
import { NON_ACTOR_SPEAKER_KINDS, RECORD_KINDS, } from "./records.js";
|
|
5
|
+
export class JsonlParseError extends Error {
|
|
6
|
+
line;
|
|
7
|
+
constructor(message, line) {
|
|
8
|
+
super(`line ${line}: ${message}`);
|
|
9
|
+
this.line = line;
|
|
10
|
+
}
|
|
11
|
+
}
|
|
12
|
+
function isObj(v) {
|
|
13
|
+
return typeof v === "object" && v !== null && !Array.isArray(v);
|
|
14
|
+
}
|
|
15
|
+
function requireKeys(obj, allowed, line) {
|
|
16
|
+
for (const key of Object.keys(obj)) {
|
|
17
|
+
if (!allowed.includes(key))
|
|
18
|
+
throw new JsonlParseError(`unknown field '${key}'`, line);
|
|
19
|
+
}
|
|
20
|
+
}
|
|
21
|
+
function parseSource(v, line) {
|
|
22
|
+
if (!isObj(v))
|
|
23
|
+
throw new JsonlParseError("missing source", line);
|
|
24
|
+
requireKeys(v, ["chunk", "line_start", "line_end"], line);
|
|
25
|
+
const { chunk, line_start, line_end } = v;
|
|
26
|
+
if (![chunk, line_start, line_end].every((n) => typeof n === "number" && Number.isInteger(n) && n >= 0)) {
|
|
27
|
+
throw new JsonlParseError("source needs integer chunk/line_start/line_end", line);
|
|
28
|
+
}
|
|
29
|
+
if (line_start >= line_end)
|
|
30
|
+
throw new JsonlParseError("source line_start must be < line_end", line);
|
|
31
|
+
return { chunk: chunk, line_start: line_start, line_end: line_end };
|
|
32
|
+
}
|
|
33
|
+
function str(obj, key, line) {
|
|
34
|
+
const v = obj[key];
|
|
35
|
+
if (typeof v !== "string" || v === "")
|
|
36
|
+
throw new JsonlParseError(`'${key}' must be a non-empty string`, line);
|
|
37
|
+
return v;
|
|
38
|
+
}
|
|
39
|
+
function optStr(obj, key, line) {
|
|
40
|
+
if (!(key in obj) || obj[key] === undefined)
|
|
41
|
+
return undefined;
|
|
42
|
+
const v = obj[key];
|
|
43
|
+
if (typeof v !== "string" || v === "")
|
|
44
|
+
throw new JsonlParseError(`'${key}' must be a non-empty string`, line);
|
|
45
|
+
return v;
|
|
46
|
+
}
|
|
47
|
+
function parseEntityRefs(v, key, line) {
|
|
48
|
+
if (!Array.isArray(v))
|
|
49
|
+
throw new JsonlParseError(`'${key}' must be an array`, line);
|
|
50
|
+
return v.map((item) => {
|
|
51
|
+
if (!isObj(item))
|
|
52
|
+
throw new JsonlParseError(`'${key}' entries must be objects`, line);
|
|
53
|
+
requireKeys(item, ["name", "state"], line);
|
|
54
|
+
const ref = { name: str(item, "name", line) };
|
|
55
|
+
const state = optStr(item, "state", line);
|
|
56
|
+
if (state !== undefined)
|
|
57
|
+
ref.state = state;
|
|
58
|
+
return ref;
|
|
59
|
+
});
|
|
60
|
+
}
|
|
61
|
+
function parseSpeaker(v, line) {
|
|
62
|
+
if (!isObj(v))
|
|
63
|
+
throw new JsonlParseError("dialogue/inner_thought needs a speaker", line);
|
|
64
|
+
if ("kind" in v) {
|
|
65
|
+
requireKeys(v, ["kind", "label"], line);
|
|
66
|
+
const kind = v.kind;
|
|
67
|
+
if (typeof kind !== "string" || !NON_ACTOR_SPEAKER_KINDS.includes(kind)) {
|
|
68
|
+
throw new JsonlParseError(`speaker.kind must be one of ${NON_ACTOR_SPEAKER_KINDS.join("|")}`, line);
|
|
69
|
+
}
|
|
70
|
+
const ref = { kind: kind };
|
|
71
|
+
const label = optStr(v, "label", line);
|
|
72
|
+
if (label !== undefined)
|
|
73
|
+
ref.label = label;
|
|
74
|
+
return ref;
|
|
75
|
+
}
|
|
76
|
+
requireKeys(v, ["name", "state"], line);
|
|
77
|
+
const ref = { name: str(v, "name", line) };
|
|
78
|
+
const state = optStr(v, "state", line);
|
|
79
|
+
if (state !== undefined)
|
|
80
|
+
ref.state = state;
|
|
81
|
+
return ref;
|
|
82
|
+
}
|
|
83
|
+
function parseRecord(obj, line) {
|
|
84
|
+
const kind = obj.kind;
|
|
85
|
+
if (typeof kind !== "string" || !RECORD_KINDS.includes(kind)) {
|
|
86
|
+
throw new JsonlParseError(`unknown or missing kind '${String(kind)}'`, line);
|
|
87
|
+
}
|
|
88
|
+
const source = parseSource(obj.source, line);
|
|
89
|
+
switch (kind) {
|
|
90
|
+
case "episode": {
|
|
91
|
+
requireKeys(obj, ["kind", "episode", "title", "source"], line);
|
|
92
|
+
const episode = obj.episode;
|
|
93
|
+
if (typeof episode !== "number" || !Number.isInteger(episode))
|
|
94
|
+
throw new JsonlParseError("episode must be an integer", line);
|
|
95
|
+
const rec = { kind, episode, source };
|
|
96
|
+
const title = optStr(obj, "title", line);
|
|
97
|
+
if (title !== undefined)
|
|
98
|
+
rec.title = title;
|
|
99
|
+
return rec;
|
|
100
|
+
}
|
|
101
|
+
case "scene": {
|
|
102
|
+
requireKeys(obj, ["kind", "scene", "environment", "location", "actors", "props", "source"], line);
|
|
103
|
+
const rec = { kind, source };
|
|
104
|
+
if ("scene" in obj && obj.scene !== undefined) {
|
|
105
|
+
if (typeof obj.scene !== "number" || !Number.isInteger(obj.scene))
|
|
106
|
+
throw new JsonlParseError("scene must be an integer", line);
|
|
107
|
+
rec.scene = obj.scene;
|
|
108
|
+
}
|
|
109
|
+
if ("environment" in obj && obj.environment !== undefined) {
|
|
110
|
+
if (!isObj(obj.environment))
|
|
111
|
+
throw new JsonlParseError("environment must be an object", line);
|
|
112
|
+
requireKeys(obj.environment, ["space", "time"], line);
|
|
113
|
+
rec.environment = obj.environment;
|
|
114
|
+
}
|
|
115
|
+
if ("location" in obj && obj.location !== undefined) {
|
|
116
|
+
if (!isObj(obj.location))
|
|
117
|
+
throw new JsonlParseError("location must be an object", line);
|
|
118
|
+
requireKeys(obj.location, ["name", "state"], line);
|
|
119
|
+
const loc = { name: str(obj.location, "name", line) };
|
|
120
|
+
const st = optStr(obj.location, "state", line);
|
|
121
|
+
if (st !== undefined)
|
|
122
|
+
loc.state = st;
|
|
123
|
+
rec.location = loc;
|
|
124
|
+
}
|
|
125
|
+
if ("actors" in obj && obj.actors !== undefined)
|
|
126
|
+
rec.actors = parseEntityRefs(obj.actors, "actors", line);
|
|
127
|
+
if ("props" in obj && obj.props !== undefined)
|
|
128
|
+
rec.props = parseEntityRefs(obj.props, "props", line);
|
|
129
|
+
return rec;
|
|
130
|
+
}
|
|
131
|
+
case "action":
|
|
132
|
+
case "transition": {
|
|
133
|
+
requireKeys(obj, ["kind", "content", "source"], line);
|
|
134
|
+
const content = str(obj, "content", line);
|
|
135
|
+
return { kind, content, source };
|
|
136
|
+
}
|
|
137
|
+
case "dialogue":
|
|
138
|
+
case "inner_thought": {
|
|
139
|
+
requireKeys(obj, ["kind", "speaker", "content", "emotion", "source"], line);
|
|
140
|
+
const content = str(obj, "content", line);
|
|
141
|
+
const speaker = parseSpeaker(obj.speaker, line);
|
|
142
|
+
const rec = { kind, speaker, content, source };
|
|
143
|
+
const emotion = optStr(obj, "emotion", line);
|
|
144
|
+
if (emotion !== undefined)
|
|
145
|
+
rec.emotion = emotion;
|
|
146
|
+
return rec;
|
|
147
|
+
}
|
|
148
|
+
case "note": {
|
|
149
|
+
requireKeys(obj, ["kind", "content", "category", "source"], line);
|
|
150
|
+
const content = str(obj, "content", line);
|
|
151
|
+
const rec = { kind, content, source };
|
|
152
|
+
if ("category" in obj && obj.category !== undefined) {
|
|
153
|
+
const cat = obj.category;
|
|
154
|
+
if (cat !== "author_note" && cat !== "format_note" && cat !== "unclassified") {
|
|
155
|
+
throw new JsonlParseError("note.category must be author_note|format_note|unclassified", line);
|
|
156
|
+
}
|
|
157
|
+
rec.category = cat;
|
|
158
|
+
}
|
|
159
|
+
return rec;
|
|
160
|
+
}
|
|
161
|
+
default:
|
|
162
|
+
throw new JsonlParseError(`unhandled kind '${kind}'`, line);
|
|
163
|
+
}
|
|
164
|
+
}
|
|
165
|
+
// Parse full JSONL text. Blank lines are skipped; every other line must be a
|
|
166
|
+
// valid record. `lineOffset` lets callers report absolute file lines.
|
|
167
|
+
export function parseJsonl(text, lineOffset = 0) {
|
|
168
|
+
const out = [];
|
|
169
|
+
const lines = text.split("\n");
|
|
170
|
+
for (let i = 0; i < lines.length; i++) {
|
|
171
|
+
const raw = lines[i].trim();
|
|
172
|
+
if (raw === "")
|
|
173
|
+
continue;
|
|
174
|
+
let parsed;
|
|
175
|
+
try {
|
|
176
|
+
parsed = JSON.parse(raw);
|
|
177
|
+
}
|
|
178
|
+
catch {
|
|
179
|
+
throw new JsonlParseError("invalid JSON", i + 1 + lineOffset);
|
|
180
|
+
}
|
|
181
|
+
if (!isObj(parsed))
|
|
182
|
+
throw new JsonlParseError("record must be a JSON object", i + 1 + lineOffset);
|
|
183
|
+
out.push(parseRecord(parsed, i + 1 + lineOffset));
|
|
184
|
+
}
|
|
185
|
+
return out;
|
|
186
|
+
}
|
|
187
|
+
//# sourceMappingURL=jsonl-parser.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"jsonl-parser.js","sourceRoot":"","sources":["../../../src/domain/ingest/jsonl-parser.ts"],"names":[],"mappings":"AAAA,6EAA6E;AAC7E,6EAA6E;AAC7E,6DAA6D;AAE7D,OAAO,EAKL,uBAAuB,EACvB,YAAY,GACb,MAAM,cAAc,CAAC;AAEtB,MAAM,OAAO,eAAgB,SAAQ,KAAK;IACF;IAAtC,YAAY,OAAe,EAAW,IAAY;QAChD,KAAK,CAAC,QAAQ,IAAI,KAAK,OAAO,EAAE,CAAC,CAAC;QADE,SAAI,GAAJ,IAAI,CAAQ;IAElD,CAAC;CACF;AAID,SAAS,KAAK,CAAC,CAAU;IACvB,OAAO,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;AAClE,CAAC;AAED,SAAS,WAAW,CAAC,GAAQ,EAAE,OAAiB,EAAE,IAAY;IAC5D,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC;QACnC,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC;YAAE,MAAM,IAAI,eAAe,CAAC,kBAAkB,GAAG,GAAG,EAAE,IAAI,CAAC,CAAC;IACxF,CAAC;AACH,CAAC;AAED,SAAS,WAAW,CAAC,CAAU,EAAE,IAAY;IAC3C,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;QAAE,MAAM,IAAI,eAAe,CAAC,gBAAgB,EAAE,IAAI,CAAC,CAAC;IACjE,WAAW,CAAC,CAAC,EAAE,CAAC,OAAO,EAAE,YAAY,EAAE,UAAU,CAAC,EAAE,IAAI,CAAC,CAAC;IAC1D,MAAM,EAAE,KAAK,EAAE,UAAU,EAAE,QAAQ,EAAE,GAAG,CAAC,CAAC;IAC1C,IAAI,CAAC,CAAC,KAAK,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,KAAK,QAAQ,IAAI,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACxG,MAAM,IAAI,eAAe,CAAC,gDAAgD,EAAE,IAAI,CAAC,CAAC;IACpF,CAAC;IACD,IAAK,UAAqB,IAAK,QAAmB;QAAE,MAAM,IAAI,eAAe,CAAC,sCAAsC,EAAE,IAAI,CAAC,CAAC;IAC5H,OAAO,EAAE,KAAK,EAAE,KAAe,EAAE,UAAU,EAAE,UAAoB,EAAE,QAAQ,EAAE,QAAkB,EAAE,CAAC;AACpG,CAAC;AAED,SAAS,GAAG,CAAC,GAAQ,EAAE,GAAW,EAAE,IAAY;IAC9C,MAAM,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC;IACnB,IAAI,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,EAAE;QAAE,MAAM,IAAI,eAAe,CAAC,IAAI,GAAG,8BAA8B,EAAE,IAAI,CAAC,CAAC;IAC9G,OAAO,CAAC,CAAC;AACX,CAAC;AAED,SAAS,MAAM,CAAC,GAAQ,EAAE,GAAW,EAAE,IAAY;IACjD,IAAI,CAAC,CAAC,GAAG,IAAI,GAAG,CAAC,IAAI,GAAG,CAAC,GAAG,CAAC,KAAK,SAAS;QAAE,OAAO,SAAS,CAAC;IAC9D,MAAM,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC;IACnB,IAAI,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,EAAE;QAAE,MAAM,IAAI,eAAe,CAAC,IAAI,GAAG,8BAA8B,EAAE,IAAI,CAAC,CAAC;IAC9G,OAAO,CAAC,CAAC;AACX,CAAC;AAED,SAAS,eAAe,CAAC,CAAU,EAAE,GAAW,EAAE,IAAY;IAC5D,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;QAAE,MAAM,IAAI,eAAe,CAAC,IAAI,GAAG,oBAAoB,EAAE,IAAI,CAAC,CAAC;IACpF,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QACpB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC;YAAE,MAAM,IAAI,eAAe,CAAC,IAAI,GAAG,2BAA2B,EAAE,IAAI,CAAC,CAAC;QACtF,WAAW,CAAC,IAAI,EAAE,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,IAAI,CAAC,CAAC;QAC3C,MAAM,GAAG,GAAiB,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,CAAC;QAC5D,MAAM,KAAK,GAAG,MAAM,CAAC,IAAI,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;QAC1C,IAAI,KAAK,KAAK,SAAS;YAAE,GAAG,CAAC,KAAK,GAAG,KAAK,CAAC;QAC3C,OAAO,GAAG,CAAC;IACb,CAAC,CAAC,CAAC;AACL,CAAC;AAED,SAAS,YAAY,CAAC,CAAU,EAAE,IAAY;IAC5C,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;QAAE,MAAM,IAAI,eAAe,CAAC,wCAAwC,EAAE,IAAI,CAAC,CAAC;IACzF,IAAI,MAAM,IAAI,CAAC,EAAE,CAAC;QAChB,WAAW,CAAC,CAAC,EAAE,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,IAAI,CAAC,CAAC;QACxC,MAAM,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC;QACpB,IAAI,OAAO,IAAI,KAAK,QAAQ,IAAI,CAAE,uBAA6C,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;YAC/F,MAAM,IAAI,eAAe,CAAC,+BAA+B,uBAAuB,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;QACtG,CAAC;QACD,MAAM,GAAG,GAAe,EAAE,IAAI,EAAE,IAAgD,EAAE,CAAC;QACnF,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;QACvC,IAAI,KAAK,KAAK,SAAS;YAAE,GAAG,CAAC,KAAK,GAAG,KAAK,CAAC;QAC3C,OAAO,GAAG,CAAC;IACb,CAAC;IACD,WAAW,CAAC,CAAC,EAAE,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,IAAI,CAAC,CAAC;IACxC,MAAM,GAAG,GAAe,EAAE,IAAI,EAAE,GAAG,CAAC,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,CAAC;IACvD,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;IACvC,IAAI,KAAK,KAAK,SAAS;QAAE,GAAG,CAAC,KAAK,GAAG,KAAK,CAAC;IAC3C,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,WAAW,CAAC,GAAQ,EAAE,IAAY;IACzC,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC;IACtB,IAAI,OAAO,IAAI,KAAK,QAAQ,IAAI,CAAE,YAAkC,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;QACpF,MAAM,IAAI,eAAe,CAAC,4BAA4B,MAAM,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;IAC/E,CAAC;IACD,MAAM,MAAM,GAAG,WAAW,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;IAE7C,QAAQ,IAAI,EAAE,CAAC;QACb,KAAK,SAAS,CAAC,CAAC,CAAC;YACf,WAAW,CAAC,GAAG,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,QAAQ,CAAC,EAAE,IAAI,CAAC,CAAC;YAC/D,MAAM,OAAO,GAAG,GAAG,CAAC,OAAO,CAAC;YAC5B,IAAI,OAAO,OAAO,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,OAAO,CAAC;gBAAE,MAAM,IAAI,eAAe,CAAC,4BAA4B,EAAE,IAAI,CAAC,CAAC;YAC7H,MAAM,GAAG,GAAG,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAkB,CAAC;YACtD,MAAM,KAAK,GAAG,MAAM,CAAC,GAAG,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;YACzC,IAAI,KAAK,KAAK,SAAS;gBAAG,GAA0B,CAAC,KAAK,GAAG,KAAK,CAAC;YACnE,OAAO,GAAG,CAAC;QACb,CAAC;QACD,KAAK,OAAO,CAAC,CAAC,CAAC;YACb,WAAW,CAAC,GAAG,EAAE,CAAC,MAAM,EAAE,OAAO,EAAE,aAAa,EAAE,UAAU,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,CAAC,EAAE,IAAI,CAAC,CAAC;YAClG,MAAM,GAAG,GAA4B,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC;YACtD,IAAI,OAAO,IAAI,GAAG,IAAI,GAAG,CAAC,KAAK,KAAK,SAAS,EAAE,CAAC;gBAC9C,IAAI,OAAO,GAAG,CAAC,KAAK,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC;oBAAE,MAAM,IAAI,eAAe,CAAC,0BAA0B,EAAE,IAAI,CAAC,CAAC;gBAC/H,GAAG,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;YACxB,CAAC;YACD,IAAI,aAAa,IAAI,GAAG,IAAI,GAAG,CAAC,WAAW,KAAK,SAAS,EAAE,CAAC;gBAC1D,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,WAAW,CAAC;oBAAE,MAAM,IAAI,eAAe,CAAC,+BAA+B,EAAE,IAAI,CAAC,CAAC;gBAC9F,WAAW,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,EAAE,IAAI,CAAC,CAAC;gBACtD,GAAG,CAAC,WAAW,GAAG,GAAG,CAAC,WAAW,CAAC;YACpC,CAAC;YACD,IAAI,UAAU,IAAI,GAAG,IAAI,GAAG,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;gBACpD,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,QAAQ,CAAC;oBAAE,MAAM,IAAI,eAAe,CAAC,4BAA4B,EAAE,IAAI,CAAC,CAAC;gBACxF,WAAW,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,IAAI,CAAC,CAAC;gBACnD,MAAM,GAAG,GAA4B,EAAE,IAAI,EAAE,GAAG,CAAC,GAAG,CAAC,QAAQ,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,CAAC;gBAC/E,MAAM,EAAE,GAAG,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;gBAC/C,IAAI,EAAE,KAAK,SAAS;oBAAE,GAAG,CAAC,KAAK,GAAG,EAAE,CAAC;gBACrC,GAAG,CAAC,QAAQ,GAAG,GAAG,CAAC;YACrB,CAAC;YACD,IAAI,QAAQ,IAAI,GAAG,IAAI,GAAG,CAAC,MAAM,KAAK,SAAS;gBAAE,GAAG,CAAC,MAAM,GAAG,eAAe,CAAC,GAAG,CAAC,MAAM,EAAE,QAAQ,EAAE,IAAI,CAAC,CAAC;YAC1G,IAAI,OAAO,IAAI,GAAG,IAAI,GAAG,CAAC,KAAK,KAAK,SAAS;gBAAE,GAAG,CAAC,KAAK,GAAG,eAAe,CAAC,GAAG,CAAC,KAAK,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;YACrG,OAAO,GAA8B,CAAC;QACxC,CAAC;QACD,KAAK,QAAQ,CAAC;QACd,KAAK,YAAY,CAAC,CAAC,CAAC;YAClB,WAAW,CAAC,GAAG,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,EAAE,IAAI,CAAC,CAAC;YACtD,MAAM,OAAO,GAAG,GAAG,CAAC,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,CAAC;YAC1C,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAkB,CAAC;QACnD,CAAC;QACD,KAAK,UAAU,CAAC;QAChB,KAAK,eAAe,CAAC,CAAC,CAAC;YACrB,WAAW,CAAC,GAAG,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,SAAS,EAAE,SAAS,EAAE,QAAQ,CAAC,EAAE,IAAI,CAAC,CAAC;YAC5E,MAAM,OAAO,GAAG,GAAG,CAAC,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,CAAC;YAC1C,MAAM,OAAO,GAAG,YAAY,CAAC,GAAG,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;YAChD,MAAM,GAAG,GAA4B,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC;YACxE,MAAM,OAAO,GAAG,MAAM,CAAC,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,CAAC;YAC7C,IAAI,OAAO,KAAK,SAAS;gBAAE,GAAG,CAAC,OAAO,GAAG,OAAO,CAAC;YACjD,OAAO,GAA8B,CAAC;QACxC,CAAC;QACD,KAAK,MAAM,CAAC,CAAC,CAAC;YACZ,WAAW,CAAC,GAAG,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,UAAU,EAAE,QAAQ,CAAC,EAAE,IAAI,CAAC,CAAC;YAClE,MAAM,OAAO,GAAG,GAAG,CAAC,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,CAAC;YAC1C,MAAM,GAAG,GAA4B,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC;YAC/D,IAAI,UAAU,IAAI,GAAG,IAAI,GAAG,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;gBACpD,MAAM,GAAG,GAAG,GAAG,CAAC,QAAQ,CAAC;gBACzB,IAAI,GAAG,KAAK,aAAa,IAAI,GAAG,KAAK,aAAa,IAAI,GAAG,KAAK,cAAc,EAAE,CAAC;oBAC7E,MAAM,IAAI,eAAe,CAAC,4DAA4D,EAAE,IAAI,CAAC,CAAC;gBAChG,CAAC;gBACD,GAAG,CAAC,QAAQ,GAAG,GAAG,CAAC;YACrB,CAAC;YACD,OAAO,GAA8B,CAAC;QACxC,CAAC;QACD;YACE,MAAM,IAAI,eAAe,CAAC,mBAAmB,IAAI,GAAG,EAAE,IAAI,CAAC,CAAC;IAChE,CAAC;AACH,CAAC;AAED,6EAA6E;AAC7E,sEAAsE;AACtE,MAAM,UAAU,UAAU,CAAC,IAAY,EAAE,UAAU,GAAG,CAAC;IACrD,MAAM,GAAG,GAAmB,EAAE,CAAC;IAC/B,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC/B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,GAAG,GAAG,KAAK,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAC;QAC7B,IAAI,GAAG,KAAK,EAAE;YAAE,SAAS;QACzB,IAAI,MAAe,CAAC;QACpB,IAAI,CAAC;YACH,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAC3B,CAAC;QAAC,MAAM,CAAC;YACP,MAAM,IAAI,eAAe,CAAC,cAAc,EAAE,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,CAAC;QAChE,CAAC;QACD,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC;YAAE,MAAM,IAAI,eAAe,CAAC,8BAA8B,EAAE,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,CAAC;QAClG,GAAG,CAAC,IAAI,CAAC,WAAW,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC;IACpD,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC"}
|
|
@@ -0,0 +1,6 @@
|
|
|
1
|
+
import type { Chunk } from "./chunker.js";
|
|
2
|
+
import { type IngestRecord, type NameStateRef } from "./records.js";
|
|
3
|
+
export declare class MarkdownParseError extends Error {
|
|
4
|
+
}
|
|
5
|
+
export declare function parseEntityList(raw: string): NameStateRef[];
|
|
6
|
+
export declare function parseNormalizeMarkdown(md: string, chunk: Chunk): IngestRecord[];
|
|
@@ -0,0 +1,186 @@
|
|
|
1
|
+
// Deterministic normalize-markdown → records parser (§pass2). The cheap model
|
|
2
|
+
// rewrites one source chunk as a lean plain-text screenplay markdown
|
|
3
|
+
// (episode/scene headers + `- 时空/地点/人物/道具` scene bullets + `> [anchor]
|
|
4
|
+
// body` lines). Bodies are PLAIN TEXT — no inline entity tags. Scene cast/props
|
|
5
|
+
// declared via `- 人物`/`- 道具` are captured onto the scene record; entity
|
|
6
|
+
// binding is done later by pass5-extract. Source is chunk-granular.
|
|
7
|
+
import { NON_ACTOR_SPEAKER_KINDS, } from "./records.js";
|
|
8
|
+
export class MarkdownParseError extends Error {
|
|
9
|
+
}
|
|
10
|
+
const EP_RE = /^#\s*第?\s*(\d+)\s*集?\s*(.*)$/;
|
|
11
|
+
const SCENE_LABEL_RE = /^##\s*(.*)$/;
|
|
12
|
+
const ENV_RE = /^-\s*时空\s*[::]\s*(.*)$/;
|
|
13
|
+
const LOC_RE = /^-\s*地点\s*[::]\s*(.*)$/;
|
|
14
|
+
const CAST_RE = /^-\s*人物\s*[::]\s*(.*)$/;
|
|
15
|
+
const PROP_RE = /^-\s*道具\s*[::]\s*(.*)$/;
|
|
16
|
+
// dialogue/inner_thought/transition/note carry a `[label]`; everything else is
|
|
17
|
+
// a bare action line. `> ` and `[act]` are accepted but not required.
|
|
18
|
+
const ANCHOR_RE = /^(?:>\s*)?\[([^\]]*)\]\s?([\s\S]*)$/;
|
|
19
|
+
const KNOWN_LABELS = new Set(["act", "dlg", "think", "trans", "note"]);
|
|
20
|
+
const SPACE_MAP = { "内": "interior", "外": "exterior" };
|
|
21
|
+
const TIME_MAP = { "日": "day", "夜": "night" };
|
|
22
|
+
// Split a declared cast/prop list on top-level separators, keeping 【…】 (which
|
|
23
|
+
// may itself contain commas) intact, then split each item into name + 【state】.
|
|
24
|
+
export function parseEntityList(raw) {
|
|
25
|
+
const items = [];
|
|
26
|
+
let depth = 0;
|
|
27
|
+
let buf = "";
|
|
28
|
+
for (const ch of raw) {
|
|
29
|
+
if (ch === "【")
|
|
30
|
+
depth++;
|
|
31
|
+
else if (ch === "】")
|
|
32
|
+
depth = Math.max(0, depth - 1);
|
|
33
|
+
if ((ch === "," || ch === "," || ch === "、") && depth === 0) {
|
|
34
|
+
if (buf.trim())
|
|
35
|
+
items.push(buf.trim());
|
|
36
|
+
buf = "";
|
|
37
|
+
continue;
|
|
38
|
+
}
|
|
39
|
+
buf += ch;
|
|
40
|
+
}
|
|
41
|
+
if (buf.trim())
|
|
42
|
+
items.push(buf.trim());
|
|
43
|
+
const refs = [];
|
|
44
|
+
for (const item of items) {
|
|
45
|
+
const m = /^([^【]+?)\s*(?:【(.+)】)?\s*$/.exec(item);
|
|
46
|
+
if (!m)
|
|
47
|
+
continue;
|
|
48
|
+
const name = m[1].trim();
|
|
49
|
+
if (!name)
|
|
50
|
+
continue;
|
|
51
|
+
const state = m[2]?.trim();
|
|
52
|
+
refs.push(state ? { name, state } : { name });
|
|
53
|
+
}
|
|
54
|
+
return refs;
|
|
55
|
+
}
|
|
56
|
+
function parseSpeaker(field) {
|
|
57
|
+
const trimmed = field.trim();
|
|
58
|
+
const colon = trimmed.indexOf(":");
|
|
59
|
+
if (colon > 0) {
|
|
60
|
+
const kind = trimmed.slice(0, colon).trim();
|
|
61
|
+
if (NON_ACTOR_SPEAKER_KINDS.includes(kind)) {
|
|
62
|
+
const label = trimmed.slice(colon + 1).trim();
|
|
63
|
+
return label ? { kind: kind, label } : { kind: kind };
|
|
64
|
+
}
|
|
65
|
+
}
|
|
66
|
+
// A bare kind word (`group` / `system` / …) with no label is that kind, not a
|
|
67
|
+
// person named "group".
|
|
68
|
+
if (NON_ACTOR_SPEAKER_KINDS.includes(trimmed))
|
|
69
|
+
return { kind: trimmed };
|
|
70
|
+
if (!trimmed)
|
|
71
|
+
throw new MarkdownParseError("dialogue anchor missing speaker");
|
|
72
|
+
return { name: trimmed };
|
|
73
|
+
}
|
|
74
|
+
// A dialogue/inner_thought anchor may pack several simultaneous speakers
|
|
75
|
+
// (`李叔,张妈`). Split on punctuation separators only (NOT 和, which can appear
|
|
76
|
+
// inside a name); each becomes its own record with the same content.
|
|
77
|
+
function parseSpeakers(field) {
|
|
78
|
+
const parts = field.split(/[,,、;;/]/).map((s) => s.trim()).filter(Boolean);
|
|
79
|
+
if (parts.length <= 1)
|
|
80
|
+
return [parseSpeaker(field)];
|
|
81
|
+
return parts.map(parseSpeaker);
|
|
82
|
+
}
|
|
83
|
+
const NOTE_CATEGORY = { author: "author_note", format: "format_note" };
|
|
84
|
+
function anchorRecords(label, body, source) {
|
|
85
|
+
const content = body.trim();
|
|
86
|
+
const parts = label.split("|");
|
|
87
|
+
const kind = parts[0].trim();
|
|
88
|
+
switch (kind) {
|
|
89
|
+
case "act":
|
|
90
|
+
return [{ kind: "action", content, source }];
|
|
91
|
+
case "trans":
|
|
92
|
+
return [{ kind: "transition", content, source }];
|
|
93
|
+
case "dlg":
|
|
94
|
+
case "think": {
|
|
95
|
+
const emotion = parts[2]?.trim();
|
|
96
|
+
// Simultaneous speakers → one record each, same content (overlap 拆多条).
|
|
97
|
+
return parseSpeakers(parts[1] ?? "").map((speaker) => {
|
|
98
|
+
const rec = { kind: kind === "dlg" ? "dialogue" : "inner_thought", speaker, content, source };
|
|
99
|
+
if (emotion)
|
|
100
|
+
rec.emotion = emotion;
|
|
101
|
+
return rec;
|
|
102
|
+
});
|
|
103
|
+
}
|
|
104
|
+
case "note":
|
|
105
|
+
return [{ kind: "note", content, category: NOTE_CATEGORY[(parts[1] ?? "").trim()] ?? "unclassified", source }];
|
|
106
|
+
default:
|
|
107
|
+
throw new MarkdownParseError(`unknown anchor label '[${label}]'`);
|
|
108
|
+
}
|
|
109
|
+
}
|
|
110
|
+
export function parseNormalizeMarkdown(md, chunk) {
|
|
111
|
+
const source = { chunk: chunk.index, line_start: chunk.core_line_start, line_end: chunk.core_line_end };
|
|
112
|
+
const records = [];
|
|
113
|
+
const lines = stripFences(md).split("\n");
|
|
114
|
+
let openScene = null;
|
|
115
|
+
for (const raw of lines) {
|
|
116
|
+
const t = raw.trim();
|
|
117
|
+
if (t === "")
|
|
118
|
+
continue;
|
|
119
|
+
if (t.startsWith("##")) {
|
|
120
|
+
const rec = { kind: "scene", source };
|
|
121
|
+
const num = /(\d+)/.exec(SCENE_LABEL_RE.exec(t)?.[1]?.trim() ?? "")?.[1];
|
|
122
|
+
if (num)
|
|
123
|
+
rec.scene = Number(num);
|
|
124
|
+
records.push(rec);
|
|
125
|
+
openScene = rec;
|
|
126
|
+
continue;
|
|
127
|
+
}
|
|
128
|
+
if (t.startsWith("#")) {
|
|
129
|
+
const ep = EP_RE.exec(t);
|
|
130
|
+
if (ep) {
|
|
131
|
+
const title = ep[2].trim();
|
|
132
|
+
const rec = { kind: "episode", episode: Number(ep[1]), source };
|
|
133
|
+
if (title)
|
|
134
|
+
rec.title = title;
|
|
135
|
+
records.push(rec);
|
|
136
|
+
}
|
|
137
|
+
continue;
|
|
138
|
+
}
|
|
139
|
+
// `- …` scene metadata bullet (only meaningful while a scene is open).
|
|
140
|
+
if (t.startsWith("-")) {
|
|
141
|
+
if (openScene) {
|
|
142
|
+
const env = ENV_RE.exec(t);
|
|
143
|
+
if (env) {
|
|
144
|
+
const e = {};
|
|
145
|
+
for (const tok of env[1].trim().split(/\s+/)) {
|
|
146
|
+
if (SPACE_MAP[tok])
|
|
147
|
+
e.space = SPACE_MAP[tok];
|
|
148
|
+
if (TIME_MAP[tok])
|
|
149
|
+
e.time = TIME_MAP[tok];
|
|
150
|
+
}
|
|
151
|
+
if (e.space || e.time)
|
|
152
|
+
openScene.environment = e;
|
|
153
|
+
}
|
|
154
|
+
else {
|
|
155
|
+
const loc = LOC_RE.exec(t);
|
|
156
|
+
const cast = CAST_RE.exec(t);
|
|
157
|
+
const prop = PROP_RE.exec(t);
|
|
158
|
+
if (loc && loc[1].trim()) {
|
|
159
|
+
const [name, state] = loc[1].split("|");
|
|
160
|
+
openScene.location = state ? { name: name.trim(), state: state.trim() } : { name: name.trim() };
|
|
161
|
+
}
|
|
162
|
+
else if (cast && cast[1].trim())
|
|
163
|
+
openScene.actors = parseEntityList(cast[1]);
|
|
164
|
+
else if (prop && prop[1].trim())
|
|
165
|
+
openScene.props = parseEntityList(prop[1]);
|
|
166
|
+
}
|
|
167
|
+
}
|
|
168
|
+
continue;
|
|
169
|
+
}
|
|
170
|
+
// A `[dlg|…]`/`[think|…]`/`[trans]`/`[note]`/`[act]` anchor, else a bare
|
|
171
|
+
// action line (content = the whole line).
|
|
172
|
+
const anchor = ANCHOR_RE.exec(t);
|
|
173
|
+
if (anchor && KNOWN_LABELS.has(anchor[1].split("|")[0].trim()))
|
|
174
|
+
records.push(...anchorRecords(anchor[1], anchor[2] ?? "", source));
|
|
175
|
+
else
|
|
176
|
+
records.push({ kind: "action", content: t, source });
|
|
177
|
+
}
|
|
178
|
+
const BODY = new Set(["action", "dialogue", "inner_thought", "transition", "note"]);
|
|
179
|
+
return records.filter((r) => !(BODY.has(r.kind) && "content" in r && r.content.trim() === ""));
|
|
180
|
+
}
|
|
181
|
+
function stripFences(raw) {
|
|
182
|
+
const trimmed = raw.trim();
|
|
183
|
+
const fence = trimmed.match(/^```(?:markdown|md)?\n([\s\S]*?)\n```$/);
|
|
184
|
+
return fence ? fence[1] : trimmed;
|
|
185
|
+
}
|
|
186
|
+
//# sourceMappingURL=markdown-parser.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"markdown-parser.js","sourceRoot":"","sources":["../../../src/domain/ingest/markdown-parser.ts"],"names":[],"mappings":"AAAA,8EAA8E;AAC9E,qEAAqE;AACrE,uEAAuE;AACvE,gFAAgF;AAChF,wEAAwE;AACxE,oEAAoE;AAGpE,OAAO,EAKL,uBAAuB,GACxB,MAAM,cAAc,CAAC;AAEtB,MAAM,OAAO,kBAAmB,SAAQ,KAAK;CAAG;AAEhD,MAAM,KAAK,GAAG,8BAA8B,CAAC;AAC7C,MAAM,cAAc,GAAG,aAAa,CAAC;AACrC,MAAM,MAAM,GAAG,wBAAwB,CAAC;AACxC,MAAM,MAAM,GAAG,wBAAwB,CAAC;AACxC,MAAM,OAAO,GAAG,wBAAwB,CAAC;AACzC,MAAM,OAAO,GAAG,wBAAwB,CAAC;AACzC,+EAA+E;AAC/E,sEAAsE;AACtE,MAAM,SAAS,GAAG,qCAAqC,CAAC;AACxD,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC;AAEvE,MAAM,SAAS,GAA2B,EAAE,GAAG,EAAE,UAAU,EAAE,GAAG,EAAE,UAAU,EAAE,CAAC;AAC/E,MAAM,QAAQ,GAA2B,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,OAAO,EAAE,CAAC;AAEtE,8EAA8E;AAC9E,+EAA+E;AAC/E,MAAM,UAAU,eAAe,CAAC,GAAW;IACzC,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,KAAK,MAAM,EAAE,IAAI,GAAG,EAAE,CAAC;QACrB,IAAI,EAAE,KAAK,GAAG;YAAE,KAAK,EAAE,CAAC;aACnB,IAAI,EAAE,KAAK,GAAG;YAAE,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC;QACpD,IAAI,CAAC,EAAE,KAAK,GAAG,IAAI,EAAE,KAAK,GAAG,IAAI,EAAE,KAAK,GAAG,CAAC,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;YAAC,IAAI,GAAG,CAAC,IAAI,EAAE;gBAAE,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;YAAC,GAAG,GAAG,EAAE,CAAC;YAAC,SAAS;QAAC,CAAC;QAC5H,GAAG,IAAI,EAAE,CAAC;IACZ,CAAC;IACD,IAAI,GAAG,CAAC,IAAI,EAAE;QAAE,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;IACvC,MAAM,IAAI,GAAmB,EAAE,CAAC;IAChC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,CAAC,GAAG,6BAA6B,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACnD,IAAI,CAAC,CAAC;YAAE,SAAS;QACjB,MAAM,IAAI,GAAG,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAC;QAC1B,IAAI,CAAC,IAAI;YAAE,SAAS;QACpB,MAAM,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC;QAC3B,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC;IAChD,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED,SAAS,YAAY,CAAC,KAAa;IACjC,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,EAAE,CAAC;IAC7B,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IACnC,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;QACd,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC;QAC5C,IAAK,uBAA6C,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;YAClE,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YAC9C,OAAO,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAgD,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAgD,EAAE,CAAC;QAChJ,CAAC;IACH,CAAC;IACD,8EAA8E;IAC9E,wBAAwB;IACxB,IAAK,uBAA6C,CAAC,QAAQ,CAAC,OAAO,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,OAAmD,EAAE,CAAC;IAC3I,IAAI,CAAC,OAAO;QAAE,MAAM,IAAI,kBAAkB,CAAC,iCAAiC,CAAC,CAAC;IAC9E,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;AAC3B,CAAC;AAED,yEAAyE;AACzE,2EAA2E;AAC3E,qEAAqE;AACrE,SAAS,aAAa,CAAC,KAAa;IAClC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;IAC3E,IAAI,KAAK,CAAC,MAAM,IAAI,CAAC;QAAE,OAAO,CAAC,YAAY,CAAC,KAAK,CAAC,CAAC,CAAC;IACpD,OAAO,KAAK,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC;AACjC,CAAC;AAED,MAAM,aAAa,GAAmE,EAAE,MAAM,EAAE,aAAa,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;AAEvI,SAAS,aAAa,CAAC,KAAa,EAAE,IAAY,EAAE,MAAkB;IACpE,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;IAC5B,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAC/B,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAC;IAC9B,QAAQ,IAAI,EAAE,CAAC;QACb,KAAK,KAAK;YACR,OAAO,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;QAC/C,KAAK,OAAO;YACV,OAAO,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;QACnD,KAAK,KAAK,CAAC;QACX,KAAK,OAAO,CAAC,CAAC,CAAC;YACb,MAAM,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC;YACjC,uEAAuE;YACvE,OAAO,aAAa,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE;gBACnD,MAAM,GAAG,GAAiB,EAAE,IAAI,EAAE,IAAI,KAAK,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,eAAe,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAkB,CAAC;gBAC5H,IAAI,OAAO;oBAAG,GAA4B,CAAC,OAAO,GAAG,OAAO,CAAC;gBAC7D,OAAO,GAAG,CAAC;YACb,CAAC,CAAC,CAAC;QACL,CAAC;QACD,KAAK,MAAM;YACT,OAAO,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,aAAa,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,cAAc,EAAE,MAAM,EAAE,CAAC,CAAC;QACjH;YACE,MAAM,IAAI,kBAAkB,CAAC,0BAA0B,KAAK,IAAI,CAAC,CAAC;IACtE,CAAC;AACH,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,EAAU,EAAE,KAAY;IAC7D,MAAM,MAAM,GAAe,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,UAAU,EAAE,KAAK,CAAC,eAAe,EAAE,QAAQ,EAAE,KAAK,CAAC,aAAa,EAAE,CAAC;IACpH,MAAM,OAAO,GAAmB,EAAE,CAAC;IACnC,MAAM,KAAK,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAE1C,IAAI,SAAS,GAAoD,IAAI,CAAC;IAEtE,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;QACxB,MAAM,CAAC,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC;QACrB,IAAI,CAAC,KAAK,EAAE;YAAE,SAAS;QAEvB,IAAI,CAAC,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;YACvB,MAAM,GAAG,GAAG,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAA8C,CAAC;YAClF,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;YACzE,IAAI,GAAG;gBAAE,GAAG,CAAC,KAAK,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC;YACjC,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YAClB,SAAS,GAAG,GAAG,CAAC;YAChB,SAAS;QACX,CAAC;QACD,IAAI,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YACtB,MAAM,EAAE,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YACzB,IAAI,EAAE,EAAE,CAAC;gBACP,MAAM,KAAK,GAAG,EAAE,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAC;gBAC5B,MAAM,GAAG,GAAiB,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,EAAkB,CAAC;gBAC9F,IAAI,KAAK;oBAAG,GAA0B,CAAC,KAAK,GAAG,KAAK,CAAC;gBACrD,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACpB,CAAC;YACD,SAAS;QACX,CAAC;QACD,uEAAuE;QACvE,IAAI,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YACtB,IAAI,SAAS,EAAE,CAAC;gBACd,MAAM,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;gBAC3B,IAAI,GAAG,EAAE,CAAC;oBACR,MAAM,CAAC,GAAsC,EAAE,CAAC;oBAChD,KAAK,MAAM,GAAG,IAAI,GAAG,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC;wBAAC,IAAI,SAAS,CAAC,GAAG,CAAC;4BAAE,CAAC,CAAC,KAAK,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC;wBAAC,IAAI,QAAQ,CAAC,GAAG,CAAC;4BAAE,CAAC,CAAC,IAAI,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC;oBAAC,CAAC;oBAC3I,IAAI,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,IAAI;wBAAE,SAAS,CAAC,WAAW,GAAG,CAAC,CAAC;gBACnD,CAAC;qBAAM,CAAC;oBACN,MAAM,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;oBAC3B,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;oBAC7B,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;oBAC7B,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,EAAE,CAAC;wBAAC,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,GAAG,GAAG,CAAC,CAAC,CAAE,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;wBAAC,SAAS,CAAC,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAK,CAAC,IAAI,EAAE,EAAE,KAAK,EAAE,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAK,CAAC,IAAI,EAAE,EAAE,CAAC;oBAAC,CAAC;yBACtK,IAAI,IAAI,IAAI,IAAI,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE;wBAAE,SAAS,CAAC,MAAM,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC,CAAE,CAAC,CAAC;yBAC1E,IAAI,IAAI,IAAI,IAAI,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE;wBAAE,SAAS,CAAC,KAAK,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC,CAAE,CAAC,CAAC;gBAChF,CAAC;YACH,CAAC;YACD,SAAS;QACX,CAAC;QACD,yEAAyE;QACzE,0CAA0C;QAC1C,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QACjC,IAAI,MAAM,IAAI,YAAY,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAC;YAAE,OAAO,CAAC,IAAI,CAAC,GAAG,aAAa,CAAC,MAAM,CAAC,CAAC,CAAE,EAAE,MAAM,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC;;YACjI,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC;IAC5D,CAAC;IAED,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC,QAAQ,EAAE,UAAU,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,CAAC,CAAC,CAAC;IACpF,OAAO,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,SAAS,IAAI,CAAC,IAAK,CAAyB,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC;AAC1H,CAAC;AAED,SAAS,WAAW,CAAC,GAAW;IAC9B,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,wCAAwC,CAAC,CAAC;IACtE,OAAO,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAE,CAAC,CAAC,CAAC,OAAO,CAAC;AACrC,CAAC"}
|
|
@@ -0,0 +1,71 @@
|
|
|
1
|
+
export type SourceSpan = {
|
|
2
|
+
chunk: number;
|
|
3
|
+
line_start: number;
|
|
4
|
+
line_end: number;
|
|
5
|
+
};
|
|
6
|
+
export type NameStateRef = {
|
|
7
|
+
name: string;
|
|
8
|
+
state?: string;
|
|
9
|
+
};
|
|
10
|
+
export type SpeakerRef = {
|
|
11
|
+
name: string;
|
|
12
|
+
state?: string;
|
|
13
|
+
} | {
|
|
14
|
+
kind: "system" | "broadcast" | "offscreen" | "group";
|
|
15
|
+
label?: string;
|
|
16
|
+
};
|
|
17
|
+
export declare const NON_ACTOR_SPEAKER_KINDS: readonly ["system", "broadcast", "offscreen", "group"];
|
|
18
|
+
export type EpisodeRecord = {
|
|
19
|
+
kind: "episode";
|
|
20
|
+
episode: number;
|
|
21
|
+
title?: string;
|
|
22
|
+
source: SourceSpan;
|
|
23
|
+
};
|
|
24
|
+
export type SceneRecord = {
|
|
25
|
+
kind: "scene";
|
|
26
|
+
scene?: number;
|
|
27
|
+
environment?: {
|
|
28
|
+
space?: string;
|
|
29
|
+
time?: string;
|
|
30
|
+
};
|
|
31
|
+
location?: {
|
|
32
|
+
name: string;
|
|
33
|
+
state?: string;
|
|
34
|
+
};
|
|
35
|
+
actors?: NameStateRef[];
|
|
36
|
+
props?: NameStateRef[];
|
|
37
|
+
source: SourceSpan;
|
|
38
|
+
};
|
|
39
|
+
export type ActionRecord = {
|
|
40
|
+
kind: "action";
|
|
41
|
+
content: string;
|
|
42
|
+
source: SourceSpan;
|
|
43
|
+
};
|
|
44
|
+
export type DialogueRecord = {
|
|
45
|
+
kind: "dialogue";
|
|
46
|
+
speaker: SpeakerRef;
|
|
47
|
+
content: string;
|
|
48
|
+
emotion?: string;
|
|
49
|
+
source: SourceSpan;
|
|
50
|
+
};
|
|
51
|
+
export type InnerThoughtRecord = {
|
|
52
|
+
kind: "inner_thought";
|
|
53
|
+
speaker: SpeakerRef;
|
|
54
|
+
content: string;
|
|
55
|
+
emotion?: string;
|
|
56
|
+
source: SourceSpan;
|
|
57
|
+
};
|
|
58
|
+
export type TransitionRecord = {
|
|
59
|
+
kind: "transition";
|
|
60
|
+
content: string;
|
|
61
|
+
source: SourceSpan;
|
|
62
|
+
};
|
|
63
|
+
export type NoteRecord = {
|
|
64
|
+
kind: "note";
|
|
65
|
+
content: string;
|
|
66
|
+
category?: "author_note" | "format_note" | "unclassified";
|
|
67
|
+
source: SourceSpan;
|
|
68
|
+
};
|
|
69
|
+
export type IngestRecord = EpisodeRecord | SceneRecord | ActionRecord | DialogueRecord | InnerThoughtRecord | TransitionRecord | NoteRecord;
|
|
70
|
+
export declare const RECORD_KINDS: readonly ["episode", "scene", "action", "dialogue", "inner_thought", "transition", "note"];
|
|
71
|
+
export type RecordKind = (typeof RECORD_KINDS)[number];
|
|
@@ -0,0 +1,9 @@
|
|
|
1
|
+
// Typed JSONL record shapes for the ingest pipeline (§7). Records are name-based
|
|
2
|
+
// and pre-canonicalization; Pass5 assigns final ids. Content is plain text (no
|
|
3
|
+
// inline entity tags); scene-level cast/props are declared on the scene record
|
|
4
|
+
// (from source `人物:`/`道具:` lines) and augmented by the extract pass.
|
|
5
|
+
// Non-actor speaker kinds. Bit-part characters (龙套) are still actors (tagged
|
|
6
|
+
// importance=background), so they speak via actor_id — no special speaker kind.
|
|
7
|
+
export const NON_ACTOR_SPEAKER_KINDS = ["system", "broadcast", "offscreen", "group"];
|
|
8
|
+
export const RECORD_KINDS = ["episode", "scene", "action", "dialogue", "inner_thought", "transition", "note"];
|
|
9
|
+
//# sourceMappingURL=records.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"records.js","sourceRoot":"","sources":["../../../src/domain/ingest/records.ts"],"names":[],"mappings":"AAAA,iFAAiF;AACjF,+EAA+E;AAC/E,+EAA+E;AAC/E,qEAAqE;AAWrE,6EAA6E;AAC7E,gFAAgF;AAChF,MAAM,CAAC,MAAM,uBAAuB,GAAG,CAAC,QAAQ,EAAE,WAAW,EAAE,WAAW,EAAE,OAAO,CAAU,CAAC;AA2B9F,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,QAAQ,EAAE,UAAU,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,CAAU,CAAC"}
|