Files
meetus/Makefile
Mariano Gabriel ca8b3a784d doocus first ver
2026-07-05 10:08:42 -03:00

64 lines
2.6 KiB
Makefile

# meetus — convenience wrapper around ctrl/batch.sh (adhoc tool, nothing
# standardized here; this just saves typing the batch flags).
#
# make batch IN="/mnt/win/trainings" # outputs next to sources
# make batch IN="..." AUDIO_LANG=es # set audio language
# make batch IN="..." OUT=output/run1 # collect outputs elsewhere
# make batch IN="..." EXTRA="--skip-cache-whisper" # one-off extra flags
# make dry IN="..." # list what would run
# make batch IN="..." EXT="mp4 mov" # limit extensions
# make batch IN="..." FLAGS="--embed-images" # replace the base flags
#
# IN is required. OUT defaults to IN (write in place); otherwise the input
# folder structure is mirrored under OUT.
IN ?=
# Default: write outputs next to the source files (run folders land in the same
# dir as each video/audio). Pass OUT=... to collect them elsewhere instead.
OUT ?= $(IN)
# The base combo used on every run. LANG/EXTRA below add to this without
# needing to retype it; override FLAGS only to change the base itself.
FLAGS ?= --embed-images --scene-detection --scene-threshold 10 --diarize --transcript-formats srt
# Per-run knobs (appended after FLAGS): language, plus any one-off extra flags
# (e.g. --skip-cache-whisper, --transcript-formats srt).
# NB: named AUDIO_LANG, not LANG — LANG is the shell locale env var.
# Defaults to English; override (e.g. AUDIO_LANG=es) for other languages.
AUDIO_LANG ?= en
EXTRA ?=
# Optional: restrict scanned extensions / pick the python interpreter.
EXT ?=
PYTHON ?=
export PYTHON
# doocus (document extraction) knobs. DOCS_OUT defaults to docs-output; the input
# tree is mirrored under it. DOC_EXTRA forwards one-off flags to process_doc.py
# (e.g. --render for pdf page thumbnails, --ocr for images/scanned pdfs).
DOCS_OUT ?= docs-output
DOC_EXTRA ?=
.PHONY: batch dry docs docs-dry help
batch:
@ctrl/batch.sh -i "$(IN)" -o "$(OUT)" $(if $(EXT),-e "$(EXT)") -- \
$(FLAGS) $(if $(AUDIO_LANG),--language $(AUDIO_LANG)) $(EXTRA)
dry:
@ctrl/batch.sh -i "$(IN)" -o "$(OUT)" $(if $(EXT),-e "$(EXT)") -n
# Document extraction (doocus), mirrors the input tree into DOCS_OUT.
# make docs IN="/mnt/win/drive" # extract all docs
# make docs IN="..." DOC_EXTRA="--render --ocr" # thumbnails + OCR
# make docs-dry IN="..." # list what would run
docs:
@ctrl/batch_docs.sh -i "$(IN)" -o "$(DOCS_OUT)" $(if $(EXT),-e "$(EXT)") -- \
$(DOC_EXTRA)
docs-dry:
@ctrl/batch_docs.sh -i "$(IN)" -o "$(DOCS_OUT)" $(if $(EXT),-e "$(EXT)") -n
help:
@sed -n '3,14p' Makefile