vera-ingest examples

These examples use the public vera_ingest API directly. For equivalent shell workflows, use vera convert.

Convert one PDF

from vera_ingest import convert

output = convert(
    "manual.pdf",
    "manual.vera",
    model="hashing",
    store_original=True,
    pipeline_options={
        "chunk_size": 500,
        "overlap": 75,
        "ocr_mode": "auto",
    },
)
print(output)

The function returns the output path after validating and atomically publishing the archive. Legacy kwargs such as chunk_size= and ocr_mode= remain compatibility aliases; explicit pipeline_options win for matching keys.

Force OCR

from vera_ingest import convert

convert(
    "scanned-manual.pdf",
    "scanned-manual.vera",
    pipeline_options={
        "ocr_mode": "force",
        "ocr_language": "eng",
        "ocr_dpi": 300,
    },
)

Use forced OCR only when automatic detection misses scanned content.

Convert a directory

from vera_ingest import batch_convert

report = batch_convert(
    "./proposals",
    recursive=True,
    model="hashing",
    ocr_mode="auto",
)

print("converted:", report["converted"])
print("skipped existing:", report["skipped_existing"])
print("failed:", report["failed"])
print("malformed existing:", report["malformed_existing"])

Batch conversion continues after per-file failures. skipped_existing lists valid archives whose stored source_file_hash still matches the current source file. Same-stem sources that would write the same .vera path appear in errors instead of overwriting each other. Check both failed and malformed_existing before treating the batch as successful.

See Convert documents for every supported option and its filesystem behavior.