Datasets:
Tasks:
Document Question Answering
Size:
n<1K
Tags:
benchmark
document-ai
information-extraction
structured-extraction
key-information-extraction
ocr
License:
File size: 13,685 Bytes
8143f06 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 | """run a single document through Anthropic Claude with the source document and schema.
Writes the benchmark result envelope:
python3 scripts/run_claude.py <document_path> <schemas/doc_id.json> <output.json>
Set ANTHROPIC_API_KEY. Override ANTHROPIC_MODEL to change the model; the default is
claude-sonnet-4-6.
"""
from __future__ import annotations
import base64
import json
import mimetypes
import os
import re
import sys
import time
import zipfile
from io import BytesIO
from pathlib import Path
from typing import Any
from xml.etree import ElementTree
import requests
sys.path.insert(0, str(Path(__file__).resolve().parent))
from run_gpt import (
IMAGE_EXTENSIONS,
TIFF_EXTENSIONS,
ExtractionFailure,
content_type,
load_env_file,
load_json,
load_prompt_template,
normalize_output_schema,
tiff_png_parts,
validate_value,
write_json,
)
ANTHROPIC_API_BASE = os.environ.get("ANTHROPIC_API_BASE", "https://api.anthropic.com/v1")
DEFAULT_ANTHROPIC_MODEL = "claude-sonnet-4-6"
ANTHROPIC_VERSION = "2023-06-01"
DEFAULT_MAX_TOKENS = int(os.environ.get("ANTHROPIC_MAX_TOKENS", "20000"))
SCHEMA_MODE = "anthropic_output_config_nullable_v1"
TEXT_EXTENSIONS = {".txt", ".csv", ".xml", ".html", ".htm", ".md", ".json", ".tsv", ".yaml", ".yml"}
def api_key() -> str:
key = os.environ.get("ANTHROPIC_API_KEY")
if not key:
raise RuntimeError("ANTHROPIC_API_KEY not set")
return key
def anthropic_model() -> str:
return os.environ.get("ANTHROPIC_MODEL", DEFAULT_ANTHROPIC_MODEL)
def headers() -> dict[str, str]:
return {
"x-api-key": api_key(),
"anthropic-version": ANTHROPIC_VERSION,
"Content-Type": "application/json",
}
def b64(data: bytes) -> str:
return base64.b64encode(data).decode("ascii")
def read_text(path: Path) -> str:
try:
return path.read_text(encoding="utf-8")
except UnicodeDecodeError:
return path.read_text(encoding="latin-1")
def docx_text(path: Path) -> str:
with zipfile.ZipFile(path) as zf:
xml = zf.read("word/document.xml")
root = ElementTree.fromstring(xml)
ns = {"w": "http://schemas.openxmlformats.org/wordprocessingml/2006/main"}
lines: list[str] = []
for paragraph in root.findall(".//w:p", ns):
text = "".join(node.text or "" for node in paragraph.findall(".//w:t", ns))
if text.strip():
lines.append(text)
return "\n".join(lines)
def xlsx_text(path: Path) -> str:
with zipfile.ZipFile(path) as zf:
shared_strings: list[str] = []
if "xl/sharedStrings.xml" in zf.namelist():
root = ElementTree.fromstring(zf.read("xl/sharedStrings.xml"))
for si in root:
shared_strings.append("".join(node.text or "" for node in si.iter() if node.tag.endswith("}t")))
lines: list[str] = []
for name in sorted(n for n in zf.namelist() if re.match(r"xl/worksheets/sheet\d+\.xml$", n)):
root = ElementTree.fromstring(zf.read(name))
lines.append(f"# {name}")
for row in root.iter():
if not row.tag.endswith("}row"):
continue
values: list[str] = []
for cell in row:
if not cell.tag.endswith("}c"):
continue
cell_type = cell.attrib.get("t")
value_node = next((child for child in cell if child.tag.endswith("}v")), None)
if value_node is None or value_node.text is None:
values.append("")
elif cell_type == "s":
idx = int(value_node.text)
values.append(shared_strings[idx] if idx < len(shared_strings) else value_node.text)
else:
values.append(value_node.text)
if any(v.strip() for v in values):
lines.append("\t".join(values))
return "\n".join(lines)
def extracted_text_for_file(path: Path) -> tuple[str, dict[str, Any]]:
suffix = path.suffix.lower()
if suffix in TEXT_EXTENSIONS:
return read_text(path), {"input_mode": "text"}
if suffix == ".docx":
return docx_text(path), {"input_mode": "docx_text"}
if suffix == ".xlsx":
return xlsx_text(path), {"input_mode": "xlsx_text"}
raise ExtractionFailure("unsupported_input", f"Claude runner cannot send {suffix} as native input")
def claude_document_parts(path: Path) -> tuple[list[dict[str, Any]], dict[str, Any]]:
suffix = path.suffix.lower()
if suffix == ".pdf":
return [
{
"type": "document",
"source": {
"type": "base64",
"media_type": "application/pdf",
"data": b64(path.read_bytes()),
},
}
], {"input_mode": "pdf_document"}
if suffix in IMAGE_EXTENSIONS:
return [
{
"type": "image",
"source": {
"type": "base64",
"media_type": content_type(path),
"data": b64(path.read_bytes()),
},
}
], {"input_mode": "image"}
if suffix in TIFF_EXTENSIONS:
openai_parts, meta = tiff_png_parts(path)
parts: list[dict[str, Any]] = []
page = 0
for item in openai_parts:
if item["type"] == "input_text":
page += 1
parts.append({"type": "text", "text": f"TIFF page {page}, converted to PNG:"})
else:
encoded = item["image_url"].split(",", 1)[1]
parts.append({
"type": "image",
"source": {"type": "base64", "media_type": "image/png", "data": encoded},
})
return parts, meta
text, meta = extracted_text_for_file(path)
return [{"type": "text", "text": f"Document text extracted from {path.name}:\n\n{text}"}], meta
def build_prompt(doc_id: str, schema: dict[str, Any] | None = None) -> str:
# canonical prompt (prompts/extraction_prompt.txt) plus one Claude-specific sentence,
# since Claude returns conversational text unless told to emit only the JSON object.
prompt = load_prompt_template().format(doc_id=doc_id) + " Return only the JSON object, with no prose or markdown."
if schema is not None:
prompt += "\n\nJSON schema:\n" + json.dumps(schema, ensure_ascii=False)
return prompt
def create_message(doc_id: str, document_parts: list[dict[str, Any]], schema: dict[str, Any], *, strict_output: bool = True) -> dict[str, Any]:
payload = {
"model": anthropic_model(),
"max_tokens": DEFAULT_MAX_TOKENS,
"messages": [
{
"role": "user",
"content": [
*document_parts,
{"type": "text", "text": build_prompt(doc_id, None if strict_output else schema)},
],
}
],
}
if strict_output:
payload["output_config"] = {
"format": {
"type": "json_schema",
"schema": schema,
}
}
resp = requests.post(f"{ANTHROPIC_API_BASE}/messages", headers=headers(), json=payload, timeout=900)
if resp.status_code >= 300:
try:
body = resp.json()
except ValueError:
body = {"error": {"message": resp.text[:800]}}
err = body.get("error") if isinstance(body, dict) else None
message = err.get("message") if isinstance(err, dict) else str(body)[:800]
raise ExtractionFailure("api_error", f"message creation failed {resp.status_code}: {message}", {"http_status": resp.status_code})
return resp.json()
def should_retry_with_prompt_schema(exc: ExtractionFailure) -> bool:
if exc.kind != "api_error":
return False
message = exc.message.lower()
return any(
marker in message
for marker in [
"compiled grammar",
"output_config",
"invalid schema",
"schema contains",
"schemas contains",
"union",
]
)
def extract_output_text(message: dict[str, Any]) -> str:
chunks: list[str] = []
for item in message.get("content") or []:
if item.get("type") == "text" and item.get("text") is not None:
chunks.append(item["text"])
elif item.get("type") == "json" and "json" in item:
return json.dumps(item["json"], ensure_ascii=False)
return "".join(chunks).strip()
def normalize_json_text(text: str) -> str:
text = text.strip()
text = re.sub(r"^```(?:json)?\s*|\s*```$", "", text, flags=re.IGNORECASE | re.DOTALL).strip()
if not text.startswith("{"):
start = text.find("{")
end = text.rfind("}")
if start != -1 and end != -1 and end > start:
text = text[start:end + 1]
return text
def parse_response_data(message: dict[str, Any], output_schema: dict[str, Any]) -> dict[str, Any]:
if message.get("stop_reason") in {"max_tokens", "model_context_window_exceeded"}:
raise ExtractionFailure("incomplete_response", f"stop_reason was {message.get('stop_reason')}")
text = extract_output_text(message)
if not text:
raise ExtractionFailure("empty_response", "response contained no output text")
text = normalize_json_text(text)
try:
data = json.loads(text)
except json.JSONDecodeError as exc:
raise ExtractionFailure("invalid_json", f"response was not valid JSON: {exc}") from exc
errors = validate_value(data, output_schema)
if errors:
raise ExtractionFailure("schema_mismatch", "; ".join(errors[:20]), {"validation_errors": errors[:200]})
return data
def estimate_cost(usage: dict[str, Any]) -> float | None:
try:
input_rate = float(os.environ["ANTHROPIC_INPUT_USD_PER_1M"])
output_rate = float(os.environ["ANTHROPIC_OUTPUT_USD_PER_1M"])
except (KeyError, ValueError):
return None
input_tokens = usage.get("input_tokens", 0) or 0
output_tokens = usage.get("output_tokens", 0) or 0
return (input_tokens * input_rate + output_tokens * output_rate) / 1_000_000
def failure_result(kind: str, message: str, *, started_at: float, doc_id: str, response: dict[str, Any] | None = None, extra_meta: dict[str, Any] | None = None) -> dict[str, Any]:
usage = (response or {}).get("usage") or {}
meta = {
"provider": "anthropic",
"model": anthropic_model(),
"doc_id": doc_id,
"response_id": (response or {}).get("id"),
"usage": usage,
"schema_mode": SCHEMA_MODE,
}
if extra_meta:
meta.update(extra_meta)
return {
"status": "failed",
"error": {"type": kind, "message": message},
"cost": estimate_cost(usage),
"time_sec": time.time() - started_at,
"data": {},
"meta": meta,
}
def run(doc_id: str, file_path: Path, json_schema: dict[str, Any]) -> dict[str, Any]:
started_at = time.time()
response: dict[str, Any] | None = None
input_meta: dict[str, Any] = {}
output_schema = normalize_output_schema(json_schema)
try:
document_parts, input_meta = claude_document_parts(file_path)
try:
response = create_message(doc_id, document_parts, output_schema)
except ExtractionFailure as exc:
if not should_retry_with_prompt_schema(exc):
raise
input_meta["schema_fallback"] = "prompt_json_schema"
response = create_message(doc_id, document_parts, output_schema, strict_output=False)
data = parse_response_data(response, output_schema)
except ExtractionFailure as exc:
return failure_result(exc.kind, exc.message, started_at=started_at, doc_id=doc_id, response=response, extra_meta={**input_meta, **exc.meta})
except requests.RequestException as exc:
return failure_result("request_error", str(exc), started_at=started_at, doc_id=doc_id, response=response, extra_meta=input_meta)
except Exception as exc:
return failure_result(exc.__class__.__name__, str(exc), started_at=started_at, doc_id=doc_id, response=response, extra_meta=input_meta)
usage = response.get("usage") or {}
return {
"status": "ok",
"cost": estimate_cost(usage),
"time_sec": time.time() - started_at,
"data": data,
"meta": {
"provider": "anthropic",
"model": anthropic_model(),
"doc_id": doc_id,
"response_id": response.get("id"),
"usage": usage,
"schema_mode": SCHEMA_MODE,
**input_meta,
},
}
def main() -> int:
if len(sys.argv) != 4:
print("usage: python3 scripts/run_claude.py <document_path> <schemas/doc_id.json> <output.json>")
return 1
file_path = Path(sys.argv[1])
schema_path = Path(sys.argv[2])
output_path = Path(sys.argv[3])
repo_root = Path(__file__).resolve().parent.parent
load_env_file(repo_root / ".env")
load_env_file(Path.cwd() / ".env")
result = run(doc_id=schema_path.stem, file_path=file_path, json_schema=load_json(schema_path))
write_json(output_path, result)
if result.get("status") == "ok":
print(f"wrote {output_path}")
else:
error = result.get("error") or {}
print(f"wrote failed result {output_path}: {error.get('type')}: {error.get('message')}")
return 0
if __name__ == "__main__":
raise SystemExit(main())
|