Lint convert.py
Browse files- src/axolotl/convert.py +32 -3
src/axolotl/convert.py
CHANGED
|
@@ -1,47 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import json
|
| 2 |
import sys
|
| 3 |
|
| 4 |
|
| 5 |
class FileReader:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
def read(self, file_path):
|
| 7 |
-
with open(file_path, "
|
| 8 |
return file.read()
|
| 9 |
|
| 10 |
|
| 11 |
class FileWriter:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
def __init__(self, file_path):
|
| 13 |
self.file_path = file_path
|
| 14 |
|
| 15 |
def write(self, content):
|
| 16 |
-
with open(self.file_path, "w") as file:
|
| 17 |
file.write(content)
|
| 18 |
|
| 19 |
|
| 20 |
class StdoutWriter:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
def write(self, content):
|
| 22 |
sys.stdout.write(content)
|
| 23 |
sys.stdout.write("\n")
|
| 24 |
|
| 25 |
|
| 26 |
class JsonParser:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
def parse(self, content):
|
| 28 |
return json.loads(content)
|
| 29 |
|
| 30 |
|
| 31 |
class JsonlSerializer:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
def serialize(self, data):
|
| 33 |
lines = [json.dumps(item) for item in data]
|
| 34 |
return "\n".join(lines)
|
| 35 |
|
| 36 |
|
| 37 |
class JsonToJsonlConverter:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
def __init__(self, file_reader, file_writer, json_parser, jsonl_serializer):
|
| 39 |
self.file_reader = file_reader
|
| 40 |
self.file_writer = file_writer
|
| 41 |
self.json_parser = json_parser
|
| 42 |
self.jsonl_serializer = jsonl_serializer
|
| 43 |
|
| 44 |
-
def convert(
|
|
|
|
|
|
|
| 45 |
content = self.file_reader.read(input_file_path)
|
| 46 |
data = self.json_parser.parse(content)
|
| 47 |
# data = [r for r in data if r["conversations"]] # vicuna cleaned has rows with empty conversations
|
|
|
|
| 1 |
+
"""Module containing File Reader, File Writer, Json Parser, and Jsonl Serializer classes"""
|
| 2 |
+
|
| 3 |
+
|
| 4 |
import json
|
| 5 |
import sys
|
| 6 |
|
| 7 |
|
| 8 |
class FileReader:
|
| 9 |
+
"""
|
| 10 |
+
Reads a file and returns its contents as a string
|
| 11 |
+
"""
|
| 12 |
+
|
| 13 |
def read(self, file_path):
|
| 14 |
+
with open(file_path, encoding="utf-8") as file:
|
| 15 |
return file.read()
|
| 16 |
|
| 17 |
|
| 18 |
class FileWriter:
|
| 19 |
+
"""
|
| 20 |
+
Writes a string to a file
|
| 21 |
+
"""
|
| 22 |
+
|
| 23 |
def __init__(self, file_path):
|
| 24 |
self.file_path = file_path
|
| 25 |
|
| 26 |
def write(self, content):
|
| 27 |
+
with open(self.file_path, "w", encoding="utf-8") as file:
|
| 28 |
file.write(content)
|
| 29 |
|
| 30 |
|
| 31 |
class StdoutWriter:
|
| 32 |
+
"""
|
| 33 |
+
Writes a string to stdout
|
| 34 |
+
"""
|
| 35 |
+
|
| 36 |
def write(self, content):
|
| 37 |
sys.stdout.write(content)
|
| 38 |
sys.stdout.write("\n")
|
| 39 |
|
| 40 |
|
| 41 |
class JsonParser:
|
| 42 |
+
"""
|
| 43 |
+
Parses a string as JSON and returns the result
|
| 44 |
+
"""
|
| 45 |
+
|
| 46 |
def parse(self, content):
|
| 47 |
return json.loads(content)
|
| 48 |
|
| 49 |
|
| 50 |
class JsonlSerializer:
|
| 51 |
+
"""
|
| 52 |
+
Serializes a list of JSON objects into a JSONL string
|
| 53 |
+
"""
|
| 54 |
+
|
| 55 |
def serialize(self, data):
|
| 56 |
lines = [json.dumps(item) for item in data]
|
| 57 |
return "\n".join(lines)
|
| 58 |
|
| 59 |
|
| 60 |
class JsonToJsonlConverter:
|
| 61 |
+
"""
|
| 62 |
+
Converts a JSON file to JSONL
|
| 63 |
+
"""
|
| 64 |
+
|
| 65 |
def __init__(self, file_reader, file_writer, json_parser, jsonl_serializer):
|
| 66 |
self.file_reader = file_reader
|
| 67 |
self.file_writer = file_writer
|
| 68 |
self.json_parser = json_parser
|
| 69 |
self.jsonl_serializer = jsonl_serializer
|
| 70 |
|
| 71 |
+
def convert(
|
| 72 |
+
self, input_file_path, output_file_path
|
| 73 |
+
): # pylint: disable=unused-argument
|
| 74 |
content = self.file_reader.read(input_file_path)
|
| 75 |
data = self.json_parser.parse(content)
|
| 76 |
# data = [r for r in data if r["conversations"]] # vicuna cleaned has rows with empty conversations
|