json.py 2.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
  1. import hashlib
  2. import json
  3. import os
  4. import re
  5. import requests
  6. from embedchain.loaders.base_loader import BaseLoader
  7. from embedchain.utils import clean_string
  8. VALID_URL_PATTERN = "^https:\/\/[0-9A-z.]+.[0-9A-z.]+.[a-z]+\/.*\.json$"
  9. class JSONLoader(BaseLoader):
  10. @staticmethod
  11. def load_data(content):
  12. """Load a json file. Each data point is a key value pair."""
  13. try:
  14. from llama_hub.jsondata.base import \
  15. JSONDataReader as LLHBUBJSONLoader
  16. except ImportError:
  17. raise Exception(
  18. f"Couldn't import the required packages to load {content}, \
  19. Do `pip install --upgrade 'embedchain[json]`"
  20. )
  21. loader = LLHBUBJSONLoader()
  22. if not isinstance(content, str):
  23. print(f"Invaid content input. Provide the correct path to the json file saved locally in {content}")
  24. data = []
  25. data_content = []
  26. # Load json data from various sources. TODO: add support for dictionary
  27. if os.path.isfile(content):
  28. with open(content, "r", encoding="utf-8") as json_file:
  29. json_data = json.load(json_file)
  30. elif re.match(VALID_URL_PATTERN, content):
  31. response = requests.get(content)
  32. if response.status_code == 200:
  33. json_data = response.json()
  34. else:
  35. raise ValueError(
  36. f"Loading data from the given url: {content} failed. \
  37. Make sure the url is working."
  38. )
  39. else:
  40. raise ValueError(f"Invalid content to load json data from: {content}")
  41. docs = loader.load_data(json_data)
  42. for doc in docs:
  43. doc_content = clean_string(doc.text)
  44. data.append({"content": doc_content, "meta_data": {"url": content}})
  45. data_content.append(doc_content)
  46. doc_id = hashlib.sha256((content + ", ".join(data_content)).encode()).hexdigest()
  47. return {"doc_id": doc_id, "data": data}