Added docker support

.dockerignore

@@ -0,0 +1,12 @@
+.git
+.venv
+__pycache__/
+*.pyc
+runtime/
+debug/
+images/
+input.*
+audio.*
+video.*
+output.*
+*.json

Dockerfile

@@ -0,0 +1,36 @@
+# syntax=docker/dockerfile:1
+
+ARG GPU=nvidia
+
+FROM nvidia/cuda:13.0.2-cudnn-runtime-ubuntu24.04 AS base-nvidia
+FROM rocm/dev-ubuntu-24.04:7.2.1-complete AS base-amd
+
+FROM base-${GPU} AS app
+
+ENV DEBIAN_FRONTEND=noninteractive \
+    PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    XDG_CACHE_HOME=/tmp/sumka-cache
+
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        ca-certificates curl ffmpeg fonts-dejavu-core \
+        python3 python3-pip python3-venv \
+    && curl -fsSL -o /tmp/chrome.deb \
+        https://dl.google.com/linux/direct/google-chrome-stable_current_amd64.deb \
+    && apt-get install -y /tmp/chrome.deb \
+    && rm -rf /var/lib/apt/lists/* /tmp/chrome.deb
+
+WORKDIR /app
+COPY requirements.txt requirements-nvidia.txt requirements-amd.txt ./
+ARG GPU
+RUN python3 -m venv /opt/venv \
+    && /opt/venv/bin/pip install --no-cache-dir --upgrade pip \
+    && /opt/venv/bin/pip install --no-cache-dir -r requirements-${GPU}.txt
+
+ENV PATH=/opt/venv/bin:$PATH
+
+COPY . ./
+WORKDIR /work
+
+ENTRYPOINT ["python3", "/app/main.py"]

README.md

@@ -15,7 +15,36 @@
 
 ## Установка (Docker)
 
-TODO
+Образы собираются из одного Dockerfile:
+
+```bash
+docker build --build-arg GPU=nvidia -t sumka:nvidia .
+docker build --build-arg GPU=amd -t sumka:amd .
+```
+
+Для NVIDIA нужен настроенный NVIDIA Container Toolkit. Для AMD на хосте должен
+быть установлен совместимый ROCm. Рабочая директория монтируется в `/work`:
+
+```bash
+WORK_DIR=$(mktemp -d)
+cp input.mkv "$WORK_DIR/"
+
+# nvidia
+docker run --rm --gpus all -e AI_API_KEY \
+  --mount type=volume,src=sumka-whisper-cache,dst=/root/.cache/whisper \
+  --mount type=bind,src="$WORK_DIR",dst=/work \
+  --tmpfs /tmp:rw,size=8g sumka:nvidia
+
+# amd
+docker run --rm --device=/dev/kfd --device=/dev/dri --group-add video \
+  --security-opt seccomp=unconfined -e AI_API_KEY \
+  --mount type=volume,src=sumka-whisper-cache,dst=/root/.cache/whisper \
+  --mount type=bind,src="$WORK_DIR",dst=/work \
+  --tmpfs /tmp:rw,size=8g sumka:amd
+```
+
+Все входные, промежуточные и итоговые файлы находятся в `WORK_DIR`. Для дампов
+запросов заранее создайте `"$WORK_DIR/debug"`; промпты остаются внутри образа.
 
 ## Установка (Python)
 
@@ -28,8 +57,8 @@ cd 2026-linux-sumka
 python3 -m venv .venv
 # Активировать виртуальное окружение
 . .venv/bin/activate
-# Установить зависимости
+# Установить зависимости под свою видеокарту
-pip install -r requirements.txt
+pip install -r requirements-nvidia.txt  # либо requirements-amd.txt
 
 # Также должны быть установлены ffmpeg и Chromium/Google Chrome
 

asr_eventizer.py

@@ -1,13 +1,12 @@
 from typing import Any
 import traceback
-import shutil
-import os
 import json
 
 from pydantic import BaseModel
 from agent import Agent, AgentMessage
 from windowizer import Windowizer, Window
 from asr_filter import AsrFilterResult, AsrFilterSegment
+from paths import debug_path, prompt_path
 from utils import Timeline, Event
 
 class _PreEvent(BaseModel):
@@ -41,21 +40,19 @@ class AsrEventizer:
                 role="user"
             )
         ]
-        debug_dir = None
+        debug_dir = debug_path("AsrEventizer", AsrEventizer.DEBUG_ID)
-        if os.path.isdir("debug"):
-            debug_dir = f"debug/AsrEventizer/{AsrEventizer.DEBUG_ID}"
-            AsrEventizer.DEBUG_ID += 1
-            os.makedirs(debug_dir, exist_ok=True)
         if debug_dir:
-            with open(f"{debug_dir}/request.txt", "w") as f:
+            AsrEventizer.DEBUG_ID += 1
-                f.write(messages[1].content)
+            (debug_dir / "request.txt").write_text(messages[1].content, encoding="utf-8")
         retries_left = 5
         while retries_left > 0:
             retries_left -= 1
             response = self._agent.completion(messages=messages, response_format=_EventizeResult)
             if debug_dir:
-                with open(f"{debug_dir}/{retries_left}-retries-left.txt", "w") as f:
+                (debug_dir / f"{retries_left}-retries-left.txt").write_text(
-                    f.write(response)
+                    response,
+                    encoding="utf-8",
+                )
             # validate data
             try:
                 response = json.loads(response)
@@ -85,10 +82,9 @@ class AsrEventizer:
         """
         self._agent = agent
         self._windowizer = windowizer
-        with open("prompts/asr_eventizer.md", "r") as f:
         self._system_prompt = AgentMessage(
-                content=f.read(),
+            content=prompt_path("asr_eventizer.md").read_text(encoding="utf-8"),
-                role="system"
+            role="system",
         )
         
     def eventize(self, asr_filter_result: AsrFilterResult) -> Timeline:

paths.py

@@ -0,0 +1,24 @@
+"""Paths for bundled resources and per-run artifacts."""
+
+from pathlib import Path
+
+
+SOURCE_DIR = Path(__file__).resolve().parent
+
+
+def prompt_path(name: str) -> Path:
+    """Return a prompt bundled with the application source code."""
+    path = SOURCE_DIR / "prompts" / name
+    if not path.is_file():
+        raise FileNotFoundError(f"Prompt not found: {path}")
+    return path
+
+
+def debug_path(component: str, index: int) -> Path | None:
+    """Create a debug directory in the current working directory when enabled."""
+    root = Path.cwd() / "debug"
+    if not root.is_dir():
+        return None
+    path = root / component / str(index)
+    path.mkdir(parents=True, exist_ok=True)
+    return path

pdf_builder.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import html
+import os
 import re
 import shutil
 import subprocess
@@ -277,6 +278,8 @@ class PdfBuilder:
                 f"--print-to-pdf={destination}",
                 html_path.as_uri(),
             ]
+            if os.geteuid() == 0:
+                command.insert(1, "--no-sandbox")
             try:
                 result = subprocess.run(
                     command,

reference_resolver.py

@@ -13,6 +13,7 @@ from typing import Literal
 from pydantic import BaseModel, ConfigDict, ValidationError, model_validator
 
 from agent import Agent, AgentMessage
+from paths import debug_path, prompt_path
 from utils import Event, Timeline
 from video_references import UnresolvedReference, UnresolvedReferences
 
@@ -94,10 +95,14 @@ class ReferenceResolver:
         self._image_dir = Path(image_dir)
         self._frame_interval = frame_interval
         self._max_frames = max_frames
-        with open("prompts/reference_resolver.md", "r", encoding="utf-8") as f:
+        self._system_prompt = AgentMessage(
-            self._system_prompt = AgentMessage(content=f.read(), role="system")
+            content=prompt_path("reference_resolver.md").read_text(encoding="utf-8"),
-        with open("prompts/reference_resolver_ocr.md", "r", encoding="utf-8") as f:
+            role="system",
-            self._ocr_prompt = AgentMessage(content=f.read(), role="system")
+        )
+        self._ocr_prompt = AgentMessage(
+            content=prompt_path("reference_resolver_ocr.md").read_text(encoding="utf-8"),
+            role="system",
+        )
 
     def _sampling_timestamps(self, start: float, end: float) -> list[float]:
         start = max(0.0, start)
@@ -217,13 +222,10 @@ class ReferenceResolver:
             ),
         ]
 
-        debug_dir = None
+        debug_dir = debug_path("ReferenceResolver", ReferenceResolver.DEBUG_ID)
-        if os.path.isdir("debug"):
+        if debug_dir:
-            debug_dir = f"debug/ReferenceResolver/{ReferenceResolver.DEBUG_ID}"
             ReferenceResolver.DEBUG_ID += 1
-            os.makedirs(debug_dir, exist_ok=True)
+            (debug_dir / "request.txt").write_text(messages[1].content, encoding="utf-8")
-            with open(f"{debug_dir}/request.txt", "w", encoding="utf-8") as f:
-                f.write(messages[1].content)
 
         retries_left = self.MAX_RETRIES
         while retries_left > 0:
@@ -235,12 +237,10 @@ class ReferenceResolver:
                 response_format=_FrameResult,
             )
             if debug_dir:
-                with open(
+                (debug_dir / f"{retries_left}-retries-left.txt").write_text(
-                    f"{debug_dir}/{retries_left}-retries-left.txt",
+                    response,
-                    "w",
                     encoding="utf-8",
-                ) as f:
+                )
-                    f.write(response)
 
             try:
                 result = _FrameResult.model_validate_json(response)
@@ -312,14 +312,11 @@ class ReferenceResolver:
             ),
         ]
 
-        debug_dir = None
+        debug_dir = debug_path("ReferenceResolverOCR", ReferenceResolver.OCR_DEBUG_ID)
-        if os.path.isdir("debug"):
+        if debug_dir:
-            debug_dir = f"debug/ReferenceResolverOCR/{ReferenceResolver.OCR_DEBUG_ID}"
             ReferenceResolver.OCR_DEBUG_ID += 1
-            os.makedirs(debug_dir, exist_ok=True)
+            (debug_dir / "request.txt").write_text(messages[1].content, encoding="utf-8")
-            with open(f"{debug_dir}/request.txt", "w", encoding="utf-8") as f:
+            shutil.copyfile(crop_path, debug_dir / "crop.jpg")
-                f.write(messages[1].content)
-            shutil.copyfile(crop_path, f"{debug_dir}/crop.jpg")
 
         retries_left = self.MAX_RETRIES
         while retries_left > 0:
@@ -331,12 +328,10 @@ class ReferenceResolver:
                 response_format=_OCRResult,
             )
             if debug_dir:
-                with open(
+                (debug_dir / f"{retries_left}-retries-left.txt").write_text(
-                    f"{debug_dir}/{retries_left}-retries-left.txt",
+                    response,
-                    "w",
                     encoding="utf-8",
-                ) as f:
+                )
-                    f.write(response)
 
             try:
                 result = _OCRResult.model_validate_json(response)

requirements-amd.txt

@@ -0,0 +1,3 @@
+-r requirements.txt
+--extra-index-url https://download.pytorch.org/whl/rocm7.2
+torch==2.14.0+rocm7.2

requirements-nvidia.txt

@@ -0,0 +1,3 @@
+-r requirements.txt
+--extra-index-url https://download.pytorch.org/whl/cu130
+torch==2.14.0+cu130

requirements.txt

@@ -1,6 +1,5 @@
-httpx2
+httpx2==2.13.0
-latex2mathml
+latex2mathml==3.81.1
-openai
+openai==3.14.0
-openai-whisper
+openai-whisper==20250625
-pydantic>=2
+pydantic==2.13.5
-torch

structure_builder.py

@@ -1,11 +1,11 @@
 import json
-import os
 import traceback
 from typing import Annotated, Literal
 
 from pydantic import BaseModel, ConfigDict, Field, ValidationError, model_validator
 
 from agent import Agent, AgentMessage
+from paths import debug_path, prompt_path
 from utils import Event, Timeline
 from windowizer import Window, Windowizer
 
@@ -146,8 +146,10 @@ class StructureBuilder:
         """
         self._agent = agent
         self._windowizer = windowizer
-        with open("prompts/structure_builder.md", "r", encoding="utf-8") as f:
+        self._system_prompt = AgentMessage(
-            self._system_prompt = AgentMessage(content=f.read(), role="system")
+            content=prompt_path("structure_builder.md").read_text(encoding="utf-8"),
+            role="system",
+        )
 
     @staticmethod
     def _validate_result(result: _BuildResult, window: Window[Event]) -> None:
@@ -192,25 +194,20 @@ class StructureBuilder:
             ),
         ]
 
-        debug_dir = None
+        debug_dir = debug_path("StructureBuilder", StructureBuilder.DEBUG_ID)
-        if os.path.isdir("debug"):
+        if debug_dir:
-            debug_dir = f"debug/StructureBuilder/{StructureBuilder.DEBUG_ID}"
             StructureBuilder.DEBUG_ID += 1
-            os.makedirs(debug_dir, exist_ok=True)
+            (debug_dir / "request.txt").write_text(messages[1].content, encoding="utf-8")
-            with open(f"{debug_dir}/request.txt", "w", encoding="utf-8") as f:
-                f.write(messages[1].content)
 
         retries_left = self.MAX_RETRIES
         while retries_left > 0:
             retries_left -= 1
             response = self._agent.completion(messages=messages, response_format=_BuildResult)
             if debug_dir:
-                with open(
+                (debug_dir / f"{retries_left}-retries-left.txt").write_text(
-                    f"{debug_dir}/{retries_left}-retries-left.txt",
+                    response,
-                    "w",
                     encoding="utf-8",
-                ) as f:
+                )
-                    f.write(response)
 
             try:
                 result = _BuildResult.model_validate_json(response)

structure_refiner.py

@@ -1,10 +1,10 @@
 import json
-import os
 import traceback
 
 from pydantic import BaseModel, ConfigDict, Field, ValidationError, model_validator
 
 from agent import Agent, AgentMessage
+from paths import debug_path, prompt_path
 from structure_builder import ImageElement, Structure, StructureElement
 from windowizer import Window, Windowizer
 
@@ -39,8 +39,10 @@ class StructureRefiner:
     def __init__(self, agent: Agent, windowizer: Windowizer[StructureElement]) -> None:
         self._agent = agent
         self._windowizer = windowizer
-        with open("prompts/structure_refiner.md", "r", encoding="utf-8") as f:
+        self._system_prompt = AgentMessage(
-            self._system_prompt = AgentMessage(content=f.read(), role="system")
+            content=prompt_path("structure_refiner.md").read_text(encoding="utf-8"),
+            role="system",
+        )
 
     @staticmethod
     def _validate_result(
@@ -69,13 +71,10 @@ class StructureRefiner:
             ),
         ]
 
-        debug_dir = None
+        debug_dir = debug_path("StructureRefiner", StructureRefiner.DEBUG_ID)
-        if os.path.isdir("debug"):
+        if debug_dir:
-            debug_dir = f"debug/StructureRefiner/{StructureRefiner.DEBUG_ID}"
             StructureRefiner.DEBUG_ID += 1
-            os.makedirs(debug_dir, exist_ok=True)
+            (debug_dir / "request.txt").write_text(messages[1].content, encoding="utf-8")
-            with open(f"{debug_dir}/request.txt", "w", encoding="utf-8") as f:
-                f.write(messages[1].content)
 
         retries_left = self.MAX_RETRIES
         while retries_left > 0:
@@ -85,12 +84,10 @@ class StructureRefiner:
                 response_format=_RefineResult,
             )
             if debug_dir:
-                with open(
+                (debug_dir / f"{retries_left}-retries-left.txt").write_text(
-                    f"{debug_dir}/{retries_left}-retries-left.txt",
+                    response,
-                    "w",
                     encoding="utf-8",
-                ) as f:
+                )
-                    f.write(response)
 
             try:
                 result = _RefineResult.model_validate_json(response)

video_references.py

@@ -1,11 +1,11 @@
 import json
-import os
 import traceback
 from typing import Literal
 
 from pydantic import BaseModel, ConfigDict, Field, ValidationError, model_validator
 
 from agent import Agent, AgentMessage
+from paths import debug_path, prompt_path
 from utils import Timeline
 from windowizer import Window, Windowizer
 
@@ -71,8 +71,10 @@ class VideoReferenceBuilder:
     ) -> None:
         self._agent = agent
         self._windowizer = windowizer
-        with open("prompts/video_references.md", "r", encoding="utf-8") as f:
+        self._system_prompt = AgentMessage(
-            self._system_prompt = AgentMessage(content=f.read(), role="system")
+            content=prompt_path("video_references.md").read_text(encoding="utf-8"),
+            role="system",
+        )
 
     @staticmethod
     def _validate_result(
@@ -113,13 +115,10 @@ class VideoReferenceBuilder:
             ),
         ]
 
-        debug_dir = None
+        debug_dir = debug_path("VideoReferenceBuilder", self.DEBUG_ID)
-        if os.path.isdir("debug"):
+        if debug_dir:
-            debug_dir = f"debug/VideoReferenceBuilder/{self.DEBUG_ID}"
             VideoReferenceBuilder.DEBUG_ID += 1
-            os.makedirs(debug_dir, exist_ok=True)
+            (debug_dir / "request.txt").write_text(messages[1].content, encoding="utf-8")
-            with open(f"{debug_dir}/request.txt", "w", encoding="utf-8") as f:
-                f.write(messages[1].content)
 
         retries_left = self.MAX_RETRIES
         while retries_left > 0:
@@ -129,12 +128,10 @@ class VideoReferenceBuilder:
                 response_format=_WindowResult,
             )
             if debug_dir:
-                with open(
+                (debug_dir / f"{retries_left}-retries-left.txt").write_text(
-                    f"{debug_dir}/{retries_left}-retries-left.txt",
+                    response,
-                    "w",
                     encoding="utf-8",
-                ) as f:
+                )
-                    f.write(response)
 
             try:
                 result = _WindowResult.model_validate_json(response)