146 lines
4.8 KiB
Python
146 lines
4.8 KiB
Python
# utils/boxapi_client.py
|
|
import os
|
|
import requests
|
|
from typing import Any, Dict, List, Tuple, Optional
|
|
from utils.boxapi_log import log_boxapi_response
|
|
|
|
BOXAPI_USERNAME = os.environ.get("BOXAPI_USERNAME", "")
|
|
BOXAPI_PASSWORD = os.environ.get("BOXAPI_PASSWORD", "")
|
|
# Default matches /get_media_by_username; override in .env if needed
|
|
BOXAPI_ENDPOINT = os.environ.get(
|
|
"BOXAPI_ENDPOINT",
|
|
"https://boxapi.ir/api/instagram/user/get_media_by_username",
|
|
)
|
|
|
|
# NEW: simple debug flag
|
|
BOXAPI_DEBUG = os.getenv("BOXAPI_DEBUG", "0") == "1"
|
|
|
|
class BoxAPIError(Exception):
|
|
pass
|
|
|
|
def _dprint(msg: str) -> None:
|
|
if BOXAPI_DEBUG:
|
|
print(f"[BOXAPI] {msg}")
|
|
|
|
|
|
def get_media_page(
|
|
username: str,
|
|
count: int = 10,
|
|
max_id: Optional[str] = None,
|
|
) -> Tuple[List[Dict[str, Any]], str, Optional[str]]:
|
|
"""
|
|
Perform a single BoxAPI request and return a page of results.
|
|
Returns: (items, log_path, next_max_id)
|
|
Notes:
|
|
- BoxAPI returns at most 12 items per request.
|
|
- Pass `max_id` to fetch older pages (use the previous page's next_max_id).
|
|
"""
|
|
if not BOXAPI_USERNAME or not BOXAPI_PASSWORD:
|
|
raise BoxAPIError("Missing BOXAPI_USERNAME/BOXAPI_PASSWORD in environment.")
|
|
|
|
# BoxAPI hard cap per request
|
|
per_req = max(1, min(int(count or 1), 12))
|
|
|
|
payload: Dict[str, Any] = {"username": username, "count": per_req}
|
|
if max_id:
|
|
payload["max_id"] = max_id
|
|
|
|
_dprint(f"→ POST {BOXAPI_ENDPOINT} payload={payload}")
|
|
|
|
resp_json: Optional[Dict[str, Any]] = None
|
|
resp_text: Optional[str] = None
|
|
status_code: Optional[int] = None
|
|
|
|
try:
|
|
resp = requests.post(
|
|
BOXAPI_ENDPOINT,
|
|
json=payload,
|
|
auth=(BOXAPI_USERNAME, BOXAPI_PASSWORD),
|
|
timeout=60,
|
|
)
|
|
status_code = resp.status_code
|
|
resp_text = resp.text
|
|
|
|
# Try to parse JSON; if it fails, log and raise
|
|
try:
|
|
resp_json = resp.json()
|
|
except Exception:
|
|
log_path = log_boxapi_response(
|
|
username, payload, status_code, response_text=resp_text,
|
|
note="Non-JSON response (error)"
|
|
)
|
|
raise BoxAPIError(f"BoxAPI non-JSON response. See log: {log_path}")
|
|
|
|
# Expected shape: status -> response -> status_code + body.items + body.next_max_id
|
|
top_status = resp_json.get("status")
|
|
inner = resp_json.get("response") or {}
|
|
inner_code = inner.get("status_code")
|
|
body = inner.get("body") or {}
|
|
items = body.get("items") or []
|
|
next_max_id = body.get("next_max_id")
|
|
|
|
log_path = log_boxapi_response(
|
|
username, payload, inner_code, response_json=resp_json, note="OK"
|
|
)
|
|
|
|
if top_status != "done" or inner_code != 200 or not isinstance(items, list):
|
|
snapshot = {
|
|
"top_status": top_status,
|
|
"inner_code": inner_code,
|
|
"has_body": isinstance(body, dict),
|
|
"items_type": str(type(items)),
|
|
"items_len": len(items) if isinstance(items, list) else None,
|
|
}
|
|
raise BoxAPIError(
|
|
f"Unexpected response format from BoxAPI ({snapshot}). See log: {log_path}"
|
|
)
|
|
|
|
_dprint(f"← items={len(items)} next_max_id={next_max_id!r}")
|
|
|
|
return items, log_path, next_max_id
|
|
|
|
except BoxAPIError:
|
|
raise
|
|
except Exception as e:
|
|
log_path = log_boxapi_response(
|
|
username, payload, status_code,
|
|
response_json=resp_json, response_text=resp_text,
|
|
note=f"HTTP/parse error (error): {e}"
|
|
)
|
|
raise BoxAPIError(f"BoxAPI request failed ({e}). See log: {log_path}")
|
|
|
|
|
|
def get_media(username: str, count: int = 10) -> Tuple[List[Dict[str, Any]], str]:
|
|
"""
|
|
Backward-compatible wrapper that fetches ONE page (latest) without pagination.
|
|
Returns: (items, log_path)
|
|
"""
|
|
items, log_path, _next = get_media_page(username=username, count=count, max_id=None)
|
|
return items, log_path
|
|
|
|
|
|
def fetch_boxapi_posts(username: str, count: int = 30) -> List[Dict[str, Any]]:
|
|
"""
|
|
Backward-compatible helper that PAGES until it collects `count` items (or exhausts).
|
|
Uses get_media_page(...) under the hood (12-per-request windows).
|
|
"""
|
|
try:
|
|
target = max(1, int(count))
|
|
collected: List[Dict[str, Any]] = []
|
|
next_id: Optional[str] = None
|
|
|
|
while len(collected) < target:
|
|
need = min(12, target - len(collected))
|
|
items, _log, next_id = get_media_page(username=username, count=need, max_id=next_id)
|
|
if not items:
|
|
break
|
|
collected.extend(items)
|
|
if not next_id: # no more pages
|
|
break
|
|
|
|
return collected[:target]
|
|
|
|
except BoxAPIError as e:
|
|
print(f"❌ Failed to fetch posts from BoxAPI: {e}")
|
|
return []
|