Spaces:
Running
Running
File size: 10,045 Bytes
aacd162 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 | from __future__ import annotations
import argparse
from pathlib import Path
import uuid
import sys
import os
from tqdm import tqdm
from .storage import LocalStorageAdapter
from .extractors import extract_text_from_txt, extract_text_from_url, extract_text_from_pdf, extract_text_from_pptx
from .chunker import chunk_text
from .embeddings import EmbeddingAdapter
from .vectorstore import ChromaAdapter
_EXTRACTORS = {
".txt": extract_text_from_txt,
".pdf": extract_text_from_pdf,
".pptx": extract_text_from_pptx,
}
def handle_upload(args: argparse.Namespace, adapter: LocalStorageAdapter):
"""Upload, extract, and optionally ingest a file."""
try:
path = Path(args.path)
if not path.exists():
raise FileNotFoundError(f"File not found: {path}")
source_id = args.source_id or str(uuid.uuid4())
print(f"[*] Uploading {path.name} (source_id={source_id})...")
dest = adapter.save_raw_file(args.user, args.notebook, source_id, path)
print(f"[β] Saved raw file to: {dest}")
# Extract based on extension
ext = path.suffix.lower()
if ext not in _EXTRACTORS:
print(f"[!] No extractor for {ext}. Raw file saved.")
return
print(f"[*] Extracting text from {ext}...")
extractor = _EXTRACTORS[ext]
if ext == ".pdf":
use_ocr = args.ocr if hasattr(args, "ocr") else False
result = extractor(path, use_ocr=use_ocr)
else:
result = extractor(path)
text = result.get("text", "")
if not text.strip():
print(f"[!] No text extracted from {path.name}.")
return
adapter.save_extracted_text(args.user, args.notebook, source_id, "content", text)
print(f"[β] Extracted text saved ({len(text)} chars) for source {source_id}")
# Auto-ingest if requested
if hasattr(args, "auto_ingest") and args.auto_ingest:
print(f"[*] Auto-ingesting into Chroma...")
_do_ingest(args.user, args.notebook, source_id, adapter, args)
else:
print(f"[>] To ingest: python -m src.ingestion.cli ingest --user {args.user} --notebook {args.notebook} --source-id {source_id}")
except Exception as e:
print(f"[ERROR] Upload failed: {e}", file=sys.stderr)
raise SystemExit(1)
def handle_url(args: argparse.Namespace, adapter: LocalStorageAdapter):
"""Fetch, extract, and optionally ingest from a URL."""
try:
source_id = args.source_id or str(uuid.uuid4())
print(f"[*] Fetching and extracting from {args.url}...")
result = extract_text_from_url(args.url)
text = result.get("text", "")
if not text.strip():
print(f"[!] No text extracted from {args.url}.")
return
nb = adapter.ensure_notebook(args.user, args.notebook)
raw_dir = nb / "files_raw" / source_id
raw_dir.mkdir(parents=True, exist_ok=True)
raw_path = raw_dir / "page.html"
raw_path.write_text(result.get("html", ""), encoding="utf-8")
print(f"[β] Saved raw HTML to: {raw_path}")
adapter.save_extracted_text(args.user, args.notebook, source_id, "content", text)
print(f"[β] Extracted text saved ({len(text)} chars) for source {source_id}")
# Auto-ingest if requested
if hasattr(args, "auto_ingest") and args.auto_ingest:
print(f"[*] Auto-ingesting into Chroma...")
_do_ingest(args.user, args.notebook, source_id, adapter, args)
else:
print(f"[>] To ingest: python -m src.ingestion.cli ingest --user {args.user} --notebook {args.notebook} --source-id {source_id}")
except Exception as e:
print(f"[ERROR] URL extraction failed: {e}", file=sys.stderr)
raise SystemExit(1)
def _do_ingest(user: str, notebook: str, source_id: str, adapter: LocalStorageAdapter, args: argparse.Namespace):
"""Internal helper: chunk, embed, and ingest into Chroma."""
try:
nb = adapter.ensure_notebook(user, notebook)
extracted_path = nb / "files_extracted" / source_id / "content.txt"
if not extracted_path.exists():
raise FileNotFoundError(f"Extracted content not found: {extracted_path}")
print(f"[*] Loading extracted text from {source_id}...")
text = extracted_path.read_text(encoding="utf-8")
text_len = len(text)
if not text.strip():
raise ValueError(f"Source {source_id} has no content.")
print(f"[*] Chunking text ({text_len} chars)...")
chunk_model = getattr(args, "chunk_model", None) or "sentence-transformers/all-MiniLM-L6-v2"
chunks = chunk_text(text, model_name=chunk_model)
for c in chunks:
c["source_id"] = source_id
c["page"] = None
print(f"[β] Created {len(chunks)} chunks")
# Initialize embedder with provider switching
provider = getattr(args, "embedding_provider", None) or os.getenv("EMBEDDING_PROVIDER", "local")
model_name = getattr(args, "embedding_model", None) or os.getenv("EMBEDDING_MODEL", "all-MiniLM-L6-v2")
print(f"[*] Computing embeddings (provider={provider}, model={model_name})...")
embedder = EmbeddingAdapter(model_name=model_name, provider=provider)
texts = [c["text"] for c in chunks]
embeddings = []
batch_size = 32
for i in tqdm(range(0, len(texts), batch_size), desc="Embedding", unit="batch"):
batch = texts[i : i + batch_size]
embeddings.extend(embedder.embed_texts(batch, batch_size=len(batch)))
print(f"[β] Computed {len(embeddings)} embeddings")
print(f"[*] Upserting to Chroma...")
chroma_dir = str((nb / "chroma").resolve())
store = ChromaAdapter(persist_directory=chroma_dir)
store.upsert_chunks(user, notebook, chunks, embeddings)
print(f"[β] Ingested {len(chunks)} chunks into Chroma collection '{user}_{notebook}'")
except Exception as e:
print(f"[ERROR] Ingestion failed: {e}", file=sys.stderr)
raise SystemExit(1)
def handle_ingest(args: argparse.Namespace, adapter: LocalStorageAdapter):
"""Chunk, embed, and ingest into Chroma."""
_do_ingest(args.user, args.notebook, args.source_id, adapter, args)
def main():
p = argparse.ArgumentParser(
description="NotebookLM-style ingestion CLI: upload, extract, chunk, embed, and store.",
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog="""
Examples:
# Upload and extract (manual 2-step workflow):
python -m src.ingestion.cli upload --user alice --notebook nb1 --path docs/notes.txt
python -m src.ingestion.cli ingest --user alice --notebook nb1 --source-id <id>
# Upload and auto-ingest (one-shot workflow):
python -m src.ingestion.cli upload --user alice --notebook nb1 --path docs/notes.txt --auto-ingest
# URL extraction and auto-ingest:
python -m src.ingestion.cli url --user alice --notebook nb1 --url https://example.com --auto-ingest
# Ingest with custom embedding provider:
python -m src.ingestion.cli ingest --user alice --notebook nb1 --source-id <id> \\
--embedding-provider openai --embedding-model text-embedding-3-large
""",
)
sub = p.add_subparsers(dest="cmd", required=True)
# Upload command
up = sub.add_parser("upload", help="Upload and extract a file")
up.add_argument("--user", required=True, help="User ID")
up.add_argument("--notebook", required=True, help="Notebook ID")
up.add_argument("--path", required=True, help="Path to file (*.txt, *.pdf, *.pptx)")
up.add_argument("--source-id", required=False, help="Source ID (auto-generated if omitted)")
up.add_argument("--ocr", action="store_true", help="(PDF only) Enable OCR on images")
up.add_argument("--auto-ingest", action="store_true", help="Automatically chunk, embed, and ingest into Chroma")
# URL command
urlp = sub.add_parser("url", help="Extract text from a URL")
urlp.add_argument("--user", required=True, help="User ID")
urlp.add_argument("--notebook", required=True, help="Notebook ID")
urlp.add_argument("--url", required=True, help="URL to fetch")
urlp.add_argument("--source-id", required=False, help="Source ID (auto-generated if omitted)")
urlp.add_argument("--auto-ingest", action="store_true", help="Automatically chunk, embed, and ingest into Chroma")
# Ingest command
ingp = sub.add_parser("ingest", help="Chunk, embed, and ingest into Chroma")
ingp.add_argument("--user", required=True, help="User ID")
ingp.add_argument("--notebook", required=True, help="Notebook ID")
ingp.add_argument("--source-id", required=True, help="Source ID (from upload/url)")
ingp.add_argument(
"--embedding-provider",
choices=["local", "openai", "huggingface"],
default="local",
help="Embedding provider (default: local). Set API keys via env vars.",
)
ingp.add_argument(
"--embedding-model",
default="all-MiniLM-L6-v2",
help="Embedding model name (default: all-MiniLM-L6-v2)",
)
ingp.add_argument(
"--chunk-model",
default="sentence-transformers/all-MiniLM-L6-v2",
help="Tokenizer model for chunking (default: all-MiniLM-L6-v2)",
)
args = p.parse_args()
try:
adapter = LocalStorageAdapter()
if args.cmd == "upload":
handle_upload(args, adapter)
elif args.cmd == "url":
handle_url(args, adapter)
elif args.cmd == "ingest":
handle_ingest(args, adapter)
except KeyboardInterrupt:
print("\n[!] Cancelled by user.", file=sys.stderr)
raise SystemExit(130)
if __name__ == "__main__":
main()
|