cohere-transcribe 0.1.0
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.
- checksums.yaml +7 -0
- data/CHANGELOG.md +21 -0
- data/LICENSE.txt +201 -0
- data/NOTICE +5 -0
- data/README.md +265 -0
- data/THIRD_PARTY_NOTICES.md +67 -0
- data/exe/cohere-transcribe +8 -0
- data/exe/cohere-transcribe-doctor +8 -0
- data/ext/cohere_transcribe_native/CMakeLists.txt +262 -0
- data/ext/cohere_transcribe_native/README.md +26 -0
- data/ext/cohere_transcribe_native/audio_abi.cpp +1416 -0
- data/ext/cohere_transcribe_native/audio_exports.macos +5 -0
- data/ext/cohere_transcribe_native/audio_exports.map +10 -0
- data/ext/cohere_transcribe_native/cohere_abi.cpp +1209 -0
- data/ext/cohere_transcribe_native/exports.macos +40 -0
- data/ext/cohere_transcribe_native/exports.map +45 -0
- data/ext/cohere_transcribe_native/extconf.rb +89 -0
- data/ext/cohere_transcribe_native/test/abi_smoke.rb +232 -0
- data/ext/cohere_transcribe_native/test/audio_matrix_smoke.cpp +121 -0
- data/ext/cohere_transcribe_native/test/audio_reliability_smoke.cpp +198 -0
- data/ext/cohere_transcribe_native/test/native_abi_reliability_smoke.cpp +186 -0
- data/ext/cohere_transcribe_native/test/native_batch_projection_probe.rb +81 -0
- data/ext/cohere_transcribe_native/test/native_cancellation_smoke.cpp +194 -0
- data/lib/cohere/transcribe/alignment/ATTRIBUTION.md +23 -0
- data/lib/cohere/transcribe/alignment/LICENSE.ctc-forced-aligner +407 -0
- data/lib/cohere/transcribe/alignment/LICENSE.torchaudio +25 -0
- data/lib/cohere/transcribe/alignment/LICENSE.uroman +26 -0
- data/lib/cohere/transcribe/alignment/aligner.rb +476 -0
- data/lib/cohere/transcribe/alignment/ctc.rb +224 -0
- data/lib/cohere/transcribe/alignment/text.rb +237 -0
- data/lib/cohere/transcribe/alignment/uroman_data.rb +4950 -0
- data/lib/cohere/transcribe/api.rb +173 -0
- data/lib/cohere/transcribe/asr/batching.rb +472 -0
- data/lib/cohere/transcribe/asr/failure_policy.rb +112 -0
- data/lib/cohere/transcribe/asr/native.rb +676 -0
- data/lib/cohere/transcribe/audio/ATTRIBUTION.md +8 -0
- data/lib/cohere/transcribe/audio/LICENSE.auditok +21 -0
- data/lib/cohere/transcribe/audio/decoder.rb +315 -0
- data/lib/cohere/transcribe/audio/ffmpeg_native.rb +248 -0
- data/lib/cohere/transcribe/audio/segmentation.rb +260 -0
- data/lib/cohere/transcribe/cli.rb +727 -0
- data/lib/cohere/transcribe/configuration.rb +282 -0
- data/lib/cohere/transcribe/constants.rb +14 -0
- data/lib/cohere/transcribe/dense_converter.rb +548 -0
- data/lib/cohere/transcribe/doctor.rb +576 -0
- data/lib/cohere/transcribe/errors.rb +57 -0
- data/lib/cohere/transcribe/gguf_writer.rb +268 -0
- data/lib/cohere/transcribe/hub.rb +436 -0
- data/lib/cohere/transcribe/input.rb +110 -0
- data/lib/cohere/transcribe/licenses/crispasr.txt +21 -0
- data/lib/cohere/transcribe/loader.rb +128 -0
- data/lib/cohere/transcribe/model_identity.rb +440 -0
- data/lib/cohere/transcribe/output/publication.rb +1118 -0
- data/lib/cohere/transcribe/output/rendering.rb +105 -0
- data/lib/cohere/transcribe/output/timing.rb +86 -0
- data/lib/cohere/transcribe/python_text.rb +70 -0
- data/lib/cohere/transcribe/pytorch_checkpoint.rb +1180 -0
- data/lib/cohere/transcribe/runtime/engine.rb +1676 -0
- data/lib/cohere/transcribe/runtime/model_provider.rb +390 -0
- data/lib/cohere/transcribe/runtime/precision.rb +57 -0
- data/lib/cohere/transcribe/runtime/preparation.rb +215 -0
- data/lib/cohere/transcribe/runtime/resources.rb +165 -0
- data/lib/cohere/transcribe/runtime/word_pipeline.rb +364 -0
- data/lib/cohere/transcribe/safetensors.rb +579 -0
- data/lib/cohere/transcribe/state/checkpoint.rb +224 -0
- data/lib/cohere/transcribe/state/contracts.rb +141 -0
- data/lib/cohere/transcribe/state/io.rb +727 -0
- data/lib/cohere/transcribe/state/locking.rb +211 -0
- data/lib/cohere/transcribe/state/manifest.rb +155 -0
- data/lib/cohere/transcribe/state.rb +7 -0
- data/lib/cohere/transcribe/types.rb +535 -0
- data/lib/cohere/transcribe/vad/ATTRIBUTION.md +14 -0
- data/lib/cohere/transcribe/vad/LICENSE.faster-whisper +21 -0
- data/lib/cohere/transcribe/vad/LICENSE.silero-vad +21 -0
- data/lib/cohere/transcribe/vad/silero.rb +344 -0
- data/lib/cohere/transcribe/vad/silero_vad_v6.onnx +0 -0
- data/lib/cohere/transcribe/vad/timestamps.rb +219 -0
- data/lib/cohere/transcribe/version.rb +7 -0
- data/lib/cohere/transcribe.rb +26 -0
- data/sig/cohere/transcribe.rbs +250 -0
- data/vendor/crispasr/AUTHORS +510 -0
- data/vendor/crispasr/LICENSE +21 -0
- data/vendor/crispasr/UPSTREAM.md +9 -0
- data/vendor/crispasr/VERSION +1 -0
- data/vendor/crispasr/ggml/AUTHORS +335 -0
- data/vendor/crispasr/ggml/CMakeLists.txt +512 -0
- data/vendor/crispasr/ggml/LICENSE +21 -0
- data/vendor/crispasr/ggml/README.md +49 -0
- data/vendor/crispasr/ggml/cmake/FindNCCL.cmake +36 -0
- data/vendor/crispasr/ggml/cmake/GitVars.cmake +22 -0
- data/vendor/crispasr/ggml/cmake/common.cmake +50 -0
- data/vendor/crispasr/ggml/cmake/ggml-config.cmake.in +191 -0
- data/vendor/crispasr/ggml/ggml.pc.in +10 -0
- data/vendor/crispasr/ggml/include/ggml-alloc.h +85 -0
- data/vendor/crispasr/ggml/include/ggml-backend.h +431 -0
- data/vendor/crispasr/ggml/include/ggml-blas.h +25 -0
- data/vendor/crispasr/ggml/include/ggml-cann.h +123 -0
- data/vendor/crispasr/ggml/include/ggml-cpp.h +39 -0
- data/vendor/crispasr/ggml/include/ggml-cpu.h +151 -0
- data/vendor/crispasr/ggml/include/ggml-cuda.h +50 -0
- data/vendor/crispasr/ggml/include/ggml-hexagon.h +19 -0
- data/vendor/crispasr/ggml/include/ggml-metal.h +61 -0
- data/vendor/crispasr/ggml/include/ggml-opencl.h +26 -0
- data/vendor/crispasr/ggml/include/ggml-openvino.h +37 -0
- data/vendor/crispasr/ggml/include/ggml-opt.h +256 -0
- data/vendor/crispasr/ggml/include/ggml-rpc.h +35 -0
- data/vendor/crispasr/ggml/include/ggml-sycl.h +49 -0
- data/vendor/crispasr/ggml/include/ggml-virtgpu.h +14 -0
- data/vendor/crispasr/ggml/include/ggml-vulkan.h +29 -0
- data/vendor/crispasr/ggml/include/ggml-webgpu.h +19 -0
- data/vendor/crispasr/ggml/include/ggml-zdnn.h +17 -0
- data/vendor/crispasr/ggml/include/ggml-zendnn.h +22 -0
- data/vendor/crispasr/ggml/include/ggml.h +2887 -0
- data/vendor/crispasr/ggml/include/gguf.h +204 -0
- data/vendor/crispasr/ggml/src/CMakeLists.txt +493 -0
- data/vendor/crispasr/ggml/src/ggml-alloc.c +1323 -0
- data/vendor/crispasr/ggml/src/ggml-backend-dl.cpp +48 -0
- data/vendor/crispasr/ggml/src/ggml-backend-dl.h +44 -0
- data/vendor/crispasr/ggml/src/ggml-backend-impl.h +275 -0
- data/vendor/crispasr/ggml/src/ggml-backend-meta.cpp +2145 -0
- data/vendor/crispasr/ggml/src/ggml-backend-reg.cpp +586 -0
- data/vendor/crispasr/ggml/src/ggml-backend.cpp +2437 -0
- data/vendor/crispasr/ggml/src/ggml-common.h +1900 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/CMakeLists.txt +718 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/amx/amx.cpp +249 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/amx/amx.h +8 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/amx/common.h +115 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/amx/mmq.cpp +2512 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/amx/mmq.h +10 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/arm/cpu-feats.cpp +98 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/arm/quants.c +4244 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/arm/repack.cpp +5156 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/loongarch/quants.c +2158 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/powerpc/cpu-feats.cpp +82 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/powerpc/quants.c +2304 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/riscv/cpu-feats.cpp +38 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/riscv/quants.c +4455 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/riscv/repack.cpp +1703 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp +50 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/s390/quants.c +1465 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/wasm/quants.c +1220 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/x86/cpu-feats.cpp +327 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/x86/quants.c +3970 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch/x86/repack.cpp +6407 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/arch-fallback.h +349 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/binary-ops.cpp +154 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/binary-ops.h +16 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/cmake/FindSIMD.cmake +100 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/common.h +95 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/ggml-cpu-impl.h +539 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/ggml-cpu.c +3791 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/ggml-cpu.cpp +703 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/hbm.cpp +55 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/hbm.h +8 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/kleidiai/kernels.cpp +939 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/kleidiai/kernels.h +90 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/kleidiai/kleidiai.cpp +1513 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/kleidiai/kleidiai.h +17 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/llamafile/sgemm.cpp +4051 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/llamafile/sgemm.h +25 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/ops.cpp +11662 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/ops.h +121 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/quants.c +1288 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/quants.h +103 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/repack.cpp +4836 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/repack.h +245 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/simd-gemm.h +226 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/simd-mappings.h +1329 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/spacemit/ime.cpp +1025 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/spacemit/ime.h +13 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/spacemit/ime1_kernels.cpp +3196 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/spacemit/ime_kernels.h +26 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/traits.cpp +36 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/traits.h +38 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/unary-ops.cpp +336 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/unary-ops.h +35 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/vec.cpp +681 -0
- data/vendor/crispasr/ggml/src/ggml-cpu/vec.h +1606 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/CMakeLists.txt +272 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/acc.cu +61 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/acc.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/add-id.cu +58 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/add-id.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/arange.cu +34 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/arange.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/argmax.cu +91 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/argmax.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/argsort.cu +265 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/argsort.cuh +19 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/binbcast.cu +534 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/binbcast.cuh +12 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/clamp.cu +45 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/clamp.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/col2im-1d.cu +81 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/col2im-1d.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/common.cuh +1489 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/concat.cu +204 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/concat.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv-transpose-1d.cu +97 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv-transpose-1d.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv2d-dw.cu +161 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv2d-dw.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv2d-transpose.cu +115 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv2d-transpose.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv2d.cu +166 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/conv2d.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/convert.cu +892 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/convert.cuh +66 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/count-equal.cu +64 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/count-equal.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cp-async.cuh +57 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cpy-utils.cuh +217 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cpy.cu +581 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cpy.cuh +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cross-entropy-loss.cu +177 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cross-entropy-loss.cuh +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cumsum.cu +307 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/cumsum.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/dequantize.cuh +99 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/diag.cu +77 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/diag.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/diagmask.cu +40 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/diagmask.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn-common.cuh +1212 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn-mma-f16.cuh +1860 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn-tile.cu +57 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn-tile.cuh +1309 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn-vec.cuh +600 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn-wmma-f16.cu +696 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn-wmma-f16.cuh +51 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn.cu +620 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fattn.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fill.cu +37 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/fill.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/gated_delta_net.cu +273 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/gated_delta_net.cuh +4 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/getrows.cu +332 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/getrows.cuh +15 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/ggml-cuda.cu +5580 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/gla.cu +93 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/gla.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/im2col.cu +274 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/im2col.cuh +6 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mean.cu +75 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mean.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mma.cuh +1333 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmf.cu +191 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmf.cuh +908 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmid.cu +164 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmid.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmq.cu +372 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmq.cuh +4175 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmvf.cu +862 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmvf.cuh +14 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmvq.cu +1161 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/mmvq.cuh +16 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/norm.cu +756 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/norm.cuh +20 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/opt-step-adamw.cu +78 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/opt-step-adamw.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/opt-step-sgd.cu +49 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/opt-step-sgd.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/out-prod.cu +68 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/out-prod.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/pad.cu +106 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/pad.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/pad_reflect_1d.cu +91 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/pad_reflect_1d.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/pool2d.cu +94 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/pool2d.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/quantize.cu +443 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/quantize.cuh +41 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/reduce_rows.cuh +39 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/roll.cu +67 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/roll.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/rope.cu +665 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/rope.cuh +9 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/scale.cu +34 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/scale.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/set-rows.cu +330 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/set-rows.cuh +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/set.cu +39 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/set.cuh +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/softcap.cu +34 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/softcap.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/softmax.cu +472 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/softmax.cuh +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/solve_tri.cu +275 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/solve_tri.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/ssm-conv.cu +197 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/ssm-conv.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/ssm-scan.cu +342 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/ssm-scan.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/sum.cu +41 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/sum.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/sumrows.cu +43 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/sumrows.cuh +4 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu +6 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu +11 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu +12 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu +6 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu +12 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu +11 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu +12 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu +11 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu +10 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu +12 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu +11 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq320-dv256.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq512-dv512.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-bf16-bf16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-bf16-f16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-bf16-q4_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-bf16-q4_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-bf16-q5_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-bf16-q5_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-bf16-q8_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-bf16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q5_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q5_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q8_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-bf16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-f16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q5_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q5_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q8_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-bf16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-f16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q4_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q4_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q5_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q5_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q8_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-bf16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-f16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q4_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q4_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q5_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q5_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q8_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-bf16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-f16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q4_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q4_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q5_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q5_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q8_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-bf16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-f16.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q4_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q4_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q5_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q5_1.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu +7 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-nvfp4.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q1_0.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/top-k.cu +95 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/top-k.cuh +3 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/topk-moe.cu +415 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/topk-moe.cuh +27 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/tri.cu +136 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/tri.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/tsembd.cu +47 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/tsembd.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/unary.cu +644 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/unary.cuh +116 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/upscale.cu +293 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/upscale.cuh +5 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/vecdotq.cuh +1317 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/vendors/cuda.h +28 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/vendors/hip.h +306 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/vendors/musa.h +148 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/wkv.cu +199 -0
- data/vendor/crispasr/ggml/src/ggml-cuda/wkv.cuh +7 -0
- data/vendor/crispasr/ggml/src/ggml-impl.h +783 -0
- data/vendor/crispasr/ggml/src/ggml-metal/CMakeLists.txt +124 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-common.cpp +457 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-common.h +52 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-context.h +41 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-context.m +778 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-device.cpp +2117 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-device.h +299 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-device.m +2093 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-impl.h +1267 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-ops.cpp +5023 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal-ops.h +111 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal.cpp +954 -0
- data/vendor/crispasr/ggml/src/ggml-metal/ggml-metal.metal +11756 -0
- data/vendor/crispasr/ggml/src/ggml-opt.cpp +1094 -0
- data/vendor/crispasr/ggml/src/ggml-quants.c +5491 -0
- data/vendor/crispasr/ggml/src/ggml-quants.h +112 -0
- data/vendor/crispasr/ggml/src/ggml-threading.cpp +12 -0
- data/vendor/crispasr/ggml/src/ggml-threading.h +14 -0
- data/vendor/crispasr/ggml/src/ggml.c +7925 -0
- data/vendor/crispasr/ggml/src/ggml.cpp +26 -0
- data/vendor/crispasr/ggml/src/gguf.cpp +1556 -0
- data/vendor/crispasr/src/cohere-arch.h +137 -0
- data/vendor/crispasr/src/cohere.cpp +5642 -0
- data/vendor/crispasr/src/cohere.h +327 -0
- data/vendor/crispasr/src/cohere_batch_planner.h +82 -0
- data/vendor/crispasr/src/cohere_chunking.h +64 -0
- data/vendor/crispasr/src/cohere_decoder_batch_layout.h +60 -0
- data/vendor/crispasr/src/cohere_encoder_padded_layout.h +27 -0
- data/vendor/crispasr/src/cohere_frontend.cpp +189 -0
- data/vendor/crispasr/src/cohere_frontend.h +31 -0
- data/vendor/crispasr/src/cohere_ragged_controller.h +149 -0
- data/vendor/crispasr/src/cohere_token_renderer.h +181 -0
- data/vendor/crispasr/src/core/attention.h +924 -0
- data/vendor/crispasr/src/core/audio_chunking.h +97 -0
- data/vendor/crispasr/src/core/beam_decode.h +486 -0
- data/vendor/crispasr/src/core/cpu_ops.h +135 -0
- data/vendor/crispasr/src/core/gguf_loader.cpp +1021 -0
- data/vendor/crispasr/src/core/gguf_loader.h +216 -0
- data/vendor/crispasr/src/core/gpu_backend_pref.h +119 -0
- data/vendor/crispasr/src/core/mel.cpp +519 -0
- data/vendor/crispasr/src/core/mel.h +265 -0
- data/vendor/crispasr/src/core/ngram_loop_fix.h +173 -0
- data/vendor/crispasr/src/core/repetition_loop_guard.h +54 -0
- data/vendor/crispasr/src/crispasr_imatrix.cpp +255 -0
- data/vendor/crispasr/src/crispasr_imatrix.h +38 -0
- metadata +596 -0
|
@@ -0,0 +1,211 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
require "digest"
|
|
4
|
+
require "fileutils"
|
|
5
|
+
require "tmpdir"
|
|
6
|
+
|
|
7
|
+
module Cohere
|
|
8
|
+
module Transcribe
|
|
9
|
+
module State
|
|
10
|
+
LOCK_DIRECTORY_PREFIX = "cohere-transcribe"
|
|
11
|
+
LOCK_SUFFIX = ".lock"
|
|
12
|
+
|
|
13
|
+
OutputLockTarget = Data.define(:path, :identity) do
|
|
14
|
+
def sort_key
|
|
15
|
+
[path.to_s.downcase, identity.downcase]
|
|
16
|
+
end
|
|
17
|
+
end
|
|
18
|
+
|
|
19
|
+
class OutputSetLock
|
|
20
|
+
@guard = Mutex.new
|
|
21
|
+
@active = {}
|
|
22
|
+
|
|
23
|
+
class << self
|
|
24
|
+
attr_reader :guard, :active
|
|
25
|
+
|
|
26
|
+
def acquire(target, blocking: false, operation_hook: nil)
|
|
27
|
+
guard.synchronize do
|
|
28
|
+
handle = nil
|
|
29
|
+
acquired = false
|
|
30
|
+
key = target.path.to_s
|
|
31
|
+
if active.key?(key)
|
|
32
|
+
raise TranscriptionRuntimeError,
|
|
33
|
+
"Another transcription job owns output set #{target.identity}"
|
|
34
|
+
end
|
|
35
|
+
|
|
36
|
+
handle = State.open_lock_file(target.path)
|
|
37
|
+
operation = File::LOCK_EX
|
|
38
|
+
operation |= File::LOCK_NB unless blocking
|
|
39
|
+
unless handle.flock(operation)
|
|
40
|
+
handle.close
|
|
41
|
+
raise TranscriptionRuntimeError,
|
|
42
|
+
"Another transcription process owns output set #{target.identity} (lock #{target.path})"
|
|
43
|
+
end
|
|
44
|
+
operation_hook&.call(:after_flock, target)
|
|
45
|
+
State.verify_lock_identity!(
|
|
46
|
+
target.path,
|
|
47
|
+
handle,
|
|
48
|
+
message: "Output lock changed while acquiring #{target.identity}"
|
|
49
|
+
)
|
|
50
|
+
|
|
51
|
+
lock = new(target, handle, key)
|
|
52
|
+
active[key] = lock
|
|
53
|
+
acquired = true
|
|
54
|
+
lock
|
|
55
|
+
rescue Errno::EWOULDBLOCK, Errno::EAGAIN
|
|
56
|
+
handle&.close
|
|
57
|
+
raise TranscriptionRuntimeError,
|
|
58
|
+
"Another transcription process owns output set #{target.identity} (lock #{target.path})"
|
|
59
|
+
rescue Interrupt, SystemExit, StandardError
|
|
60
|
+
handle&.close
|
|
61
|
+
raise
|
|
62
|
+
ensure
|
|
63
|
+
handle.close if handle && !acquired && !handle.closed?
|
|
64
|
+
end
|
|
65
|
+
end
|
|
66
|
+
|
|
67
|
+
def release_all
|
|
68
|
+
guard.synchronize { active.values.dup }.each(&:release)
|
|
69
|
+
end
|
|
70
|
+
end
|
|
71
|
+
|
|
72
|
+
attr_reader :target
|
|
73
|
+
|
|
74
|
+
def initialize(target, handle, key)
|
|
75
|
+
@target = target
|
|
76
|
+
@handle = handle
|
|
77
|
+
@key = key
|
|
78
|
+
@released = false
|
|
79
|
+
end
|
|
80
|
+
|
|
81
|
+
def release
|
|
82
|
+
self.class.guard.synchronize do
|
|
83
|
+
return if @released
|
|
84
|
+
|
|
85
|
+
begin
|
|
86
|
+
@handle.flock(File::LOCK_UN)
|
|
87
|
+
ensure
|
|
88
|
+
@handle.close
|
|
89
|
+
@released = true
|
|
90
|
+
self.class.active.delete(@key)
|
|
91
|
+
end
|
|
92
|
+
end
|
|
93
|
+
nil
|
|
94
|
+
end
|
|
95
|
+
|
|
96
|
+
def released?
|
|
97
|
+
@released
|
|
98
|
+
end
|
|
99
|
+
end
|
|
100
|
+
|
|
101
|
+
module_function
|
|
102
|
+
|
|
103
|
+
def lock_target_for_outputs(output_paths)
|
|
104
|
+
parent, stem = output_parent_and_stem(output_paths)
|
|
105
|
+
identity = parent.join(stem).expand_path.cleanpath.to_s
|
|
106
|
+
digest = Digest::SHA256.hexdigest(identity.downcase)
|
|
107
|
+
OutputLockTarget.new(
|
|
108
|
+
path: lock_registry_directory.join("#{digest}#{LOCK_SUFFIX}").freeze,
|
|
109
|
+
identity: identity.freeze
|
|
110
|
+
)
|
|
111
|
+
end
|
|
112
|
+
|
|
113
|
+
def with_output_lock(target, blocking: false)
|
|
114
|
+
lock = OutputSetLock.acquire(target, blocking: blocking)
|
|
115
|
+
yield lock
|
|
116
|
+
ensure
|
|
117
|
+
lock&.release
|
|
118
|
+
end
|
|
119
|
+
|
|
120
|
+
def lock_registry_directory
|
|
121
|
+
scope = Process.respond_to?(:uid) ? Process.uid.to_s : Digest::SHA256.hexdigest(Dir.home)[0, 16]
|
|
122
|
+
Pathname(Dir.tmpdir).join("#{LOCK_DIRECTORY_PREFIX}-#{scope}")
|
|
123
|
+
end
|
|
124
|
+
|
|
125
|
+
def open_lock_file(path)
|
|
126
|
+
path = Pathname(path)
|
|
127
|
+
succeeded = false
|
|
128
|
+
validate_lock_directory!(path.dirname)
|
|
129
|
+
inspect_lock_path!(path)
|
|
130
|
+
flags = File::RDWR | File::CREAT
|
|
131
|
+
flags |= File::NOFOLLOW if defined?(File::NOFOLLOW)
|
|
132
|
+
flags |= File::CLOEXEC if defined?(File::CLOEXEC)
|
|
133
|
+
descriptor = ::IO.sysopen(path.to_s, flags, 0o600)
|
|
134
|
+
handle = File.new(descriptor, "r+", autoclose: true)
|
|
135
|
+
descriptor = nil
|
|
136
|
+
opened = verify_lock_identity!(
|
|
137
|
+
path,
|
|
138
|
+
handle,
|
|
139
|
+
message: "Output lock changed while it was being opened or is not regular"
|
|
140
|
+
)
|
|
141
|
+
validate_owned_private_file!(opened, path)
|
|
142
|
+
succeeded = true
|
|
143
|
+
handle
|
|
144
|
+
rescue Errno::ELOOP, Errno::EISDIR, Errno::ENXIO => e
|
|
145
|
+
raise TranscriptionRuntimeError, "Output lock is not a regular file: #{path}", cause: e
|
|
146
|
+
ensure
|
|
147
|
+
if !succeeded && handle
|
|
148
|
+
handle.close
|
|
149
|
+
elsif descriptor
|
|
150
|
+
::IO.new(descriptor).close
|
|
151
|
+
end
|
|
152
|
+
end
|
|
153
|
+
|
|
154
|
+
def validate_lock_directory!(path)
|
|
155
|
+
begin
|
|
156
|
+
Dir.mkdir(path, 0o700)
|
|
157
|
+
rescue Errno::EEXIST
|
|
158
|
+
nil
|
|
159
|
+
rescue SystemCallError => e
|
|
160
|
+
raise TranscriptionRuntimeError, "Cannot prepare output lock directory #{path}: #{e.message}"
|
|
161
|
+
end
|
|
162
|
+
stat = path.lstat
|
|
163
|
+
raise TranscriptionRuntimeError, "Output lock directory is not a real directory: #{path}" unless stat.directory? && !stat.symlink?
|
|
164
|
+
|
|
165
|
+
return if Gem.win_platform?
|
|
166
|
+
|
|
167
|
+
if Process.respond_to?(:uid) && stat.uid != Process.uid
|
|
168
|
+
raise TranscriptionRuntimeError,
|
|
169
|
+
"Output lock directory is not owned by the current user: #{path}"
|
|
170
|
+
end
|
|
171
|
+
return if stat.mode.nobits?(0o077)
|
|
172
|
+
|
|
173
|
+
raise TranscriptionRuntimeError,
|
|
174
|
+
"Output lock directory permissions must be private (0700): #{path}"
|
|
175
|
+
rescue Errno::ENOENT, Errno::ENOTDIR => e
|
|
176
|
+
raise TranscriptionRuntimeError, "Cannot prepare output lock directory #{path}: #{e.message}"
|
|
177
|
+
end
|
|
178
|
+
|
|
179
|
+
def inspect_lock_path!(path)
|
|
180
|
+
stat = path.lstat
|
|
181
|
+
return if stat.file? && !stat.symlink?
|
|
182
|
+
|
|
183
|
+
raise TranscriptionRuntimeError, "Output lock is not a regular file: #{path}"
|
|
184
|
+
rescue Errno::ENOENT
|
|
185
|
+
nil
|
|
186
|
+
end
|
|
187
|
+
|
|
188
|
+
def validate_owned_private_file!(stat, path)
|
|
189
|
+
return if Gem.win_platform?
|
|
190
|
+
|
|
191
|
+
if Process.respond_to?(:uid) && stat.uid != Process.uid
|
|
192
|
+
raise TranscriptionRuntimeError, "Output lock is not owned by the current user: #{path}"
|
|
193
|
+
end
|
|
194
|
+
return if stat.mode.nobits?(0o077)
|
|
195
|
+
|
|
196
|
+
raise TranscriptionRuntimeError,
|
|
197
|
+
"Output lock permissions must be private (0600): #{path}"
|
|
198
|
+
end
|
|
199
|
+
|
|
200
|
+
def verify_lock_identity!(path, handle, message:)
|
|
201
|
+
opened = handle.stat
|
|
202
|
+
current = Pathname(path).lstat
|
|
203
|
+
return opened if opened.file? && same_file?(opened, current)
|
|
204
|
+
|
|
205
|
+
raise TranscriptionRuntimeError, "#{message}: #{path}"
|
|
206
|
+
rescue SystemCallError => e
|
|
207
|
+
raise TranscriptionRuntimeError, "#{message}: #{path}", cause: e
|
|
208
|
+
end
|
|
209
|
+
end
|
|
210
|
+
end
|
|
211
|
+
end
|
|
@@ -0,0 +1,155 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
require "digest"
|
|
4
|
+
|
|
5
|
+
module Cohere
|
|
6
|
+
module Transcribe
|
|
7
|
+
module State
|
|
8
|
+
PublicationVerification = Data.define(:verified, :generation_id, :reason) do
|
|
9
|
+
def verified?
|
|
10
|
+
verified
|
|
11
|
+
end
|
|
12
|
+
end
|
|
13
|
+
|
|
14
|
+
module_function
|
|
15
|
+
|
|
16
|
+
def published_manifest_content(source_snapshot:, output_paths:, contents:, asr_contract_key:,
|
|
17
|
+
render_contract_key:, generation_id:)
|
|
18
|
+
payload = published_payload(
|
|
19
|
+
source_snapshot: source_snapshot,
|
|
20
|
+
output_paths: output_paths,
|
|
21
|
+
contents: contents,
|
|
22
|
+
asr_contract_key: asr_contract_key,
|
|
23
|
+
render_contract_key: render_contract_key,
|
|
24
|
+
generation_id: generation_id
|
|
25
|
+
)
|
|
26
|
+
JSON.pretty_generate(envelope(payload)) << "\n"
|
|
27
|
+
end
|
|
28
|
+
|
|
29
|
+
def published_payload(source_snapshot:, output_paths:, contents:, asr_contract_key:,
|
|
30
|
+
render_contract_key:, generation_id:)
|
|
31
|
+
raise ArgumentError, "generation ID is invalid" unless generation_id.is_a?(String) && !generation_id.empty?
|
|
32
|
+
|
|
33
|
+
{
|
|
34
|
+
"kind" => "published",
|
|
35
|
+
"generation_id" => generation_id,
|
|
36
|
+
"asr_contract_key" => asr_contract_key,
|
|
37
|
+
"render_contract_key" => render_contract_key,
|
|
38
|
+
"source" => source_snapshot.payload,
|
|
39
|
+
"updated_unix_seconds" => Time.now.to_f,
|
|
40
|
+
"outputs" => output_paths.keys.sort.to_h do |format|
|
|
41
|
+
content = contents.fetch(format)
|
|
42
|
+
[
|
|
43
|
+
format,
|
|
44
|
+
{
|
|
45
|
+
"name" => Pathname(output_paths.fetch(format)).basename.to_s,
|
|
46
|
+
"size" => content.bytesize,
|
|
47
|
+
"sha256" => Digest::SHA256.hexdigest(content)
|
|
48
|
+
}
|
|
49
|
+
]
|
|
50
|
+
end
|
|
51
|
+
}
|
|
52
|
+
end
|
|
53
|
+
|
|
54
|
+
def verify_published_outputs(source_snapshot:, output_paths:, state_path:, asr_contract_key:,
|
|
55
|
+
render_contract_key:, directory_binding: nil,
|
|
56
|
+
guard_bindings: nil)
|
|
57
|
+
payload, reason = decode_state(
|
|
58
|
+
state_path,
|
|
59
|
+
directory_binding: directory_binding,
|
|
60
|
+
guard_bindings: guard_bindings
|
|
61
|
+
)
|
|
62
|
+
return PublicationVerification.new(verified: false, generation_id: nil, reason: reason.freeze) unless payload
|
|
63
|
+
|
|
64
|
+
mismatch = publication_mismatch_reason(
|
|
65
|
+
payload,
|
|
66
|
+
source_snapshot: source_snapshot,
|
|
67
|
+
output_paths: output_paths,
|
|
68
|
+
asr_contract_key: asr_contract_key,
|
|
69
|
+
render_contract_key: render_contract_key,
|
|
70
|
+
directory_binding: directory_binding,
|
|
71
|
+
guard_bindings: guard_bindings
|
|
72
|
+
)
|
|
73
|
+
return PublicationVerification.new(verified: false, generation_id: nil, reason: mismatch.freeze) if mismatch
|
|
74
|
+
|
|
75
|
+
PublicationVerification.new(
|
|
76
|
+
verified: true,
|
|
77
|
+
generation_id: payload.fetch("generation_id").dup.freeze,
|
|
78
|
+
reason: nil
|
|
79
|
+
)
|
|
80
|
+
rescue EncodingError, SystemCallError, TypeError, ArgumentError => e
|
|
81
|
+
PublicationVerification.new(
|
|
82
|
+
verified: false,
|
|
83
|
+
generation_id: nil,
|
|
84
|
+
reason: "cannot verify output generation (#{e.class}: #{e.message})".freeze
|
|
85
|
+
)
|
|
86
|
+
end
|
|
87
|
+
|
|
88
|
+
def publication_mismatch_reason(payload, source_snapshot:, output_paths:, asr_contract_key:,
|
|
89
|
+
render_contract_key:, directory_binding: nil,
|
|
90
|
+
guard_bindings: nil)
|
|
91
|
+
return "state is #{payload["kind"].inspect}, not published" unless payload["kind"] == "published"
|
|
92
|
+
return "state marker ASR contract does not match" unless payload["asr_contract_key"] == asr_contract_key
|
|
93
|
+
return "state marker render contract does not match" unless payload["render_contract_key"] == render_contract_key
|
|
94
|
+
return "state marker source snapshot does not match" unless payload["source"] == source_snapshot.payload
|
|
95
|
+
|
|
96
|
+
generation_id = payload["generation_id"]
|
|
97
|
+
return "state marker generation ID is invalid" unless generation_id.is_a?(String) && !generation_id.empty?
|
|
98
|
+
|
|
99
|
+
records = payload["outputs"]
|
|
100
|
+
return "state marker output formats do not match" unless records.is_a?(Hash) && records.keys.sort == output_paths.keys.sort
|
|
101
|
+
|
|
102
|
+
output_paths.each do |format, path_value|
|
|
103
|
+
path = Pathname(path_value)
|
|
104
|
+
record = records[format]
|
|
105
|
+
return "state marker path for #{format} does not match" unless record.is_a?(Hash) && record["name"] == path.basename.to_s
|
|
106
|
+
|
|
107
|
+
begin
|
|
108
|
+
size, sha256 = if directory_binding
|
|
109
|
+
bound_output_record(
|
|
110
|
+
path,
|
|
111
|
+
directory_binding: directory_binding,
|
|
112
|
+
guard_bindings: guard_bindings
|
|
113
|
+
)
|
|
114
|
+
else
|
|
115
|
+
stat = path.lstat
|
|
116
|
+
return "#{format} output is missing or not regular" unless stat.file? && !stat.symlink?
|
|
117
|
+
|
|
118
|
+
[stat.size, Digest::SHA256.file(path).hexdigest]
|
|
119
|
+
end
|
|
120
|
+
return "#{format} output does not match its state marker" if record["size"] != size || record["sha256"] != sha256
|
|
121
|
+
rescue Errno::ENOENT
|
|
122
|
+
return "#{format} output is missing or not regular"
|
|
123
|
+
end
|
|
124
|
+
end
|
|
125
|
+
nil
|
|
126
|
+
end
|
|
127
|
+
|
|
128
|
+
def bound_output_record(path, directory_binding:, guard_bindings:)
|
|
129
|
+
with_bound_parent(
|
|
130
|
+
path,
|
|
131
|
+
directory_binding: directory_binding,
|
|
132
|
+
guard_bindings: guard_bindings
|
|
133
|
+
) do |bound, basename|
|
|
134
|
+
bound.verify!
|
|
135
|
+
handle = nil
|
|
136
|
+
Thread.handle_interrupt(DEFERRED_PUBLICATION_EXCEPTIONS) do
|
|
137
|
+
handle = bound.open_regular(basename)
|
|
138
|
+
end
|
|
139
|
+
opened = handle.stat
|
|
140
|
+
digest = Digest::SHA256.new
|
|
141
|
+
digest << handle.read(1024 * 1024) until handle.eof?
|
|
142
|
+
unless bound.same_regular_entry?(basename, opened)
|
|
143
|
+
raise TranscriptionRuntimeError, "Output changed while it was being verified: #{path}"
|
|
144
|
+
end
|
|
145
|
+
|
|
146
|
+
bound.verify!
|
|
147
|
+
[opened.size, digest.hexdigest]
|
|
148
|
+
ensure
|
|
149
|
+
handle&.close
|
|
150
|
+
end
|
|
151
|
+
end
|
|
152
|
+
private_class_method :bound_output_record
|
|
153
|
+
end
|
|
154
|
+
end
|
|
155
|
+
end
|