doocus first ver

pyproject.toml

@@ -0,0 +1,51 @@
+[project]
+name = "meetus"
+version = "0.1.0"
+description = "Local, offline toolkit: meetus (meeting recordings → enhanced transcripts) + doocus (documents → textified content + metadata)."
+requires-python = ">=3.10"
+# Base install pulls nothing — pick the feature groups you need (see below).
+dependencies = []
+
+# Feature groups (PEP 735), installed with uv. Examples:
+#   uv sync --group meetus                  meeting pipeline
+#   uv sync --group doocus                  document extraction
+#   uv sync --group doocus --group ocr      + OCR for images/scanned pdfs
+#   uv run process_doc.py file.pdf          run inside the synced venv
+[dependency-groups]
+# meetus: frame extraction from recordings. whisper/whisperx are installed
+# separately (heavy; see `transcribe` group / README) and called as CLIs.
+meetus = [
+  "opencv-python>=4.8.0",
+  "ffmpeg-python>=0.2.0",
+]
+
+# doocus: local, offline document extraction (office/pdf/web/text/image).
+doocus = [
+  "Pillow>=10.0.0",
+  "PyYAML>=6.0",
+  "python-docx>=1.1.0",
+  "python-pptx>=0.6.23",
+  "openpyxl>=3.1.0",
+  "pypdf>=4.0.0",
+  "beautifulsoup4>=4.12.0",
+  "lxml>=5.0.0",
+]
+
+# OCR for doocus images / scanned pdfs (--ocr). Needs system `tesseract`.
+ocr = ["pytesseract>=0.3.10"]
+
+# doocus pdf page-image thumbnails (--render). Needs system `poppler`.
+pdf-render = ["pdf2image>=1.17.0"]
+
+# NB: whisper / whisperx are NOT here. meetus invokes them as external CLIs
+# (via subprocess), like ffmpeg — install them your own way (e.g. a separate
+# GPU env). They're heavy and don't belong in the managed deps.
+
+# Old OCR/vision path in meetus/deprecated/ — unwired, reference only. Kept out
+# of every default install; this is the ONLY place `ollama` is used.
+deprecated = ["ollama>=0.1.0", "pytesseract>=0.3.10"]
+
+[tool.uv]
+# This repo is a collection of scripts (process_meeting.py, process_doc.py) plus
+# the meetus/ and doocus/ packages, not a distributable — don't build it.
+package = false