Nelze vybrat více než 25 témat Téma musí začínat písmenem nebo číslem, může obsahovat pomlčky („-“) a může být dlouhé až 35 znaků.
 
 

225 řádky
9.5 KiB

  1. #!/usr/bin/env python
  2. import base64
  3. import datetime
  4. import json
  5. import os
  6. import pathlib
  7. import shutil
  8. import time
  9. import urllib.parse
  10. from typing import Optional
  11. import click
  12. import logging
  13. import requests
  14. import minio
  15. from progress import Progress
  16. logging.basicConfig(level=logging.INFO)
  17. BACKFEED_DELIM = "\n"
  18. @click.group()
  19. def sender():
  20. pass
  21. def watch_pass(input_directory: pathlib.Path, work_directory: pathlib.Path, ia_collection: str, ia_item_title: str,
  22. ia_item_prefix: str, ia_item_date: str, project: str, dispatcher: str, delete: bool, backfeed_key: str):
  23. logging.info("Checking for new items...")
  24. for original_directory in input_directory.iterdir():
  25. if original_directory.is_dir():
  26. original_name = original_directory.name
  27. new_directory = work_directory.joinpath(original_name)
  28. try:
  29. try:
  30. original_directory.rename(new_directory)
  31. except FileNotFoundError:
  32. logging.warning(f"Unable to move item {original_directory}")
  33. single_impl(new_directory, ia_collection, ia_item_title, ia_item_prefix, ia_item_date, project,
  34. dispatcher, delete, backfeed_key)
  35. return True
  36. finally:
  37. # If we exit somehow without deleting, move it back. Likely ctrl+c.
  38. if new_directory.exists():
  39. if len(list(new_directory.iterdir())) > 0:
  40. logging.warning("Stopped upload but files remain, moving back to queue...")
  41. try:
  42. new_directory.rename(original_directory)
  43. mj = original_directory.joinpath("__upload_meta.json")
  44. if mj.exists():
  45. os.remove(mj)
  46. except FileNotFoundError:
  47. logging.warning(f"Unable to move item {new_directory}")
  48. return False
  49. @sender.command()
  50. @click.option('--input-directory', envvar='UPLOAD_QUEUE_DIR', default="/data/upload-queue",
  51. type=click.Path(exists=True))
  52. @click.option('--work-directory', envvar='UPLOADER_WORKING_DIR', default="/data/uploader-work",
  53. type=click.Path(exists=True))
  54. @click.option('--ia-collection', envvar='IA_COLLECTION', required=True)
  55. @click.option('--ia-item-title', envvar='IA_ITEM_TITLE', required=True)
  56. @click.option('--ia-item-prefix', envvar='IA_ITEM_PREFIX', required=True)
  57. @click.option('--ia-item-date', envvar='IA_ITEM_DATE', required=False)
  58. @click.option('--project', envvar='PROJECT', required=True)
  59. @click.option('--dispatcher', envvar='DISPATCHER', required=True)
  60. @click.option('--delete/--no-delete', envvar='DELETE', default=False)
  61. @click.option('--backfeed-key', envvar='BACKFEED_KEY', required=True)
  62. def watch(input_directory: pathlib.Path, work_directory: pathlib.Path, ia_collection: str, ia_item_title: str,
  63. ia_item_prefix: str, ia_item_date: str, project: str, dispatcher: str, delete: bool, backfeed_key: str):
  64. if not isinstance(input_directory, pathlib.Path):
  65. input_directory = pathlib.Path(input_directory)
  66. if not isinstance(work_directory, pathlib.Path):
  67. work_directory = pathlib.Path(work_directory)
  68. while True:
  69. if not watch_pass(input_directory, work_directory, ia_collection, ia_item_title, ia_item_prefix, ia_item_date,
  70. project, dispatcher, delete, backfeed_key):
  71. logging.info("No item found, sleeping...")
  72. time.sleep(10)
  73. @sender.command()
  74. @click.option('--item-directory', type=click.Path(exists=True), required=True)
  75. @click.option('--ia-collection', envvar='IA_COLLECTION', required=True)
  76. @click.option('--ia-item-title', envvar='IA_ITEM_TITLE', required=True)
  77. @click.option('--ia-item-prefix', envvar='IA_ITEM_PREFIX', required=True)
  78. @click.option('--ia-item-date', envvar='IA_ITEM_DATE', required=False)
  79. @click.option('--project', envvar='PROJECT', required=True)
  80. @click.option('--dispatcher', envvar='DISPATCHER', required=True)
  81. @click.option('--delete/--no-delete', envvar='DELETE', default=False)
  82. @click.option('--backfeed-key', envvar='BACKFEED_KEY', required=True)
  83. def single(item_directory: pathlib.Path, ia_collection: str, ia_item_title: str, ia_item_prefix: str,
  84. ia_item_date: Optional[str], project: str, dispatcher: str, delete: bool, backfeed_key: str):
  85. single_impl(item_directory, ia_collection, ia_item_title, ia_item_prefix, ia_item_date, project, dispatcher, delete,
  86. backfeed_key)
  87. def single_impl(item_directory: pathlib.Path, ia_collection: str, ia_item_title: str, ia_item_prefix: str,
  88. ia_item_date: Optional[str], project: str, dispatcher: str, delete: bool, backfeed_key: str):
  89. if not isinstance(item_directory, pathlib.Path):
  90. item_directory = pathlib.Path(item_directory)
  91. logging.info(f"Processing item {item_directory}...")
  92. if ia_item_date is None:
  93. s = item_directory.name.split("_")
  94. if len(s) > 0:
  95. ds = s[0]
  96. try:
  97. d = datetime.datetime.strptime(ds, "%Y%m%d%H%M%S")
  98. ia_item_date = d.strftime("%Y-%m")
  99. except ValueError:
  100. pass
  101. meta_json_loc = item_directory.joinpath('__upload_meta.json')
  102. if meta_json_loc.exists():
  103. raise Exception("META JSON EXISTS WTF")
  104. meta_json = {
  105. "IA_COLLECTION": ia_collection,
  106. "IA_ITEM_TITLE": f"{ia_item_title} {item_directory.name}",
  107. "IA_ITEM_DATE": ia_item_date,
  108. "IA_ITEM_NAME": f"{ia_item_prefix}{item_directory.name}",
  109. "PROJECT": project,
  110. }
  111. with open(meta_json_loc, 'w') as f:
  112. f.write(json.dumps(meta_json))
  113. logging.info("Wrote metadata json.")
  114. total_size = 0
  115. files = list(item_directory.glob("**/*"))
  116. for item in files:
  117. total_size = total_size + os.path.getsize(item)
  118. logging.info(f"Item size is {total_size} bytes across {len(files)} files.")
  119. meta_json["SIZE_HINT"] = str(total_size)
  120. while True:
  121. try:
  122. r = requests.get(f"{dispatcher}/offload_target", params=meta_json, timeout=60)
  123. if r.status_code == 200:
  124. data = r.json()
  125. url = data["url"]
  126. break
  127. else:
  128. raise Exception(f"Invalid status code {r.status_code}: {r.text}")
  129. except Exception:
  130. logging.exception("Unable to fetch target")
  131. time.sleep(30)
  132. logging.info(f"Assigned target {url}")
  133. parsed_url = urllib.parse.urlparse(url)
  134. bf_item = None
  135. if parsed_url.scheme == "minio+http" or parsed_url.scheme == "minio+https":
  136. secure = (parsed_url.scheme == "minio+https")
  137. ep = parsed_url.hostname
  138. if parsed_url.port is not None:
  139. ep = f"{ep}:{parsed_url.port}"
  140. client = None
  141. while True:
  142. try:
  143. client = minio.Minio(endpoint=ep, access_key=parsed_url.username, secret_key=parsed_url.password, secure=secure)
  144. break
  145. except Exception:
  146. logging.exception("Failed to connect to minio")
  147. time.sleep(30)
  148. bucket_name = item_directory.name.replace("_", "-")
  149. logging.info("Making bucket...")
  150. while True:
  151. try:
  152. if client.bucket_exists(bucket_name=bucket_name):
  153. raise Exception("Bucket already exists!")
  154. client.make_bucket(bucket_name=bucket_name)
  155. break
  156. except Exception:
  157. logging.exception("Failed to make bucket")
  158. time.sleep(30)
  159. logging.info("Starting uploads...")
  160. for file in files:
  161. rel_file = file.relative_to(item_directory)
  162. while True:
  163. try:
  164. logging.info(f"Uploading file {rel_file}...")
  165. client.fput_object(bucket_name=bucket_name, object_name=str(rel_file), file_path=file,
  166. progress=Progress())
  167. break
  168. except minio.error.MinioException:
  169. logging.exception("Failed to upload")
  170. time.sleep(30)
  171. except Exception:
  172. logging.exception("Failed to upload")
  173. time.sleep(30)
  174. item_data = {"url": url, "item_name": item_directory.name, "bucket_name": bucket_name}
  175. bf_item_part = base64.urlsafe_b64encode(str(json.dumps(item_data)).encode("UTF-8")).decode("UTF-8")
  176. bf_item = f"{project}:{parsed_url.hostname}:{bf_item_part}"
  177. else:
  178. raise Exception("Unable to upload, don't understand url: {url}")
  179. if bf_item is None:
  180. raise Exception("Unable to create backfeed item")
  181. if backfeed_key == "SKIPBF":
  182. logging.warning(f"Skipping backfeed! Would have submitted: {bf_item}")
  183. else:
  184. while True:
  185. u = f"https://legacy-api.arpa.li/backfeed/legacy/{backfeed_key}"
  186. logging.info(f"Attempting to submit bf item {bf_item} to {u}...")
  187. resp = requests.post(u, params={"skipbloom": "1", "delimiter": BACKFEED_DELIM},
  188. data=f"{bf_item}{BACKFEED_DELIM}".encode("UTF-8"), timeout=60)
  189. if resp.status_code == 200:
  190. break
  191. logging.warning(f"Failed to submit to backfeed {resp.status_code}: {resp.text}")
  192. time.sleep(30)
  193. logging.info("Backfeed submit complete!")
  194. if delete:
  195. logging.info("Removing item...")
  196. shutil.rmtree(item_directory)
  197. logging.info("Upload complete!")
  198. if __name__ == '__main__':
  199. sender()