File size: 8,220 Bytes
5f7dc7e | 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 | import json
import csv
import io
import re
DEFAULT_CHUNK_SIZE = 2000
DEFAULT_OVERLAP = 200
def _estimate_tokens(text: str) -> int:
return len(text) // 4
def _recursive_split(
text: str, chunk_size: int = DEFAULT_CHUNK_SIZE, overlap: int = DEFAULT_OVERLAP
) -> list[str]:
separators = ["\n\n", "\n", ". ", " "]
chunks = []
start = 0
while start < len(text):
end = min(start + chunk_size * 4, len(text))
if end < len(text):
best_sep = -1
for sep in separators:
pos = text.rfind(sep, start, end)
if pos > best_sep:
best_sep = pos
if best_sep > start:
end = best_sep + len(sep) if best_sep >= 0 else end
chunk = text[start:end].strip()
if chunk:
chunks.append(chunk)
start = end - overlap * 4 if end < len(text) else len(text)
return chunks if chunks else [text.strip()]
def chunk_markdown(
content: str, filename: str, chunk_size: int = DEFAULT_CHUNK_SIZE
) -> list[dict]:
lines = content.split("\n")
sections = []
current_header = "Introduction"
current_body = []
current_level = 0
for line in lines:
header_match = re.match(r"^(#{1,6})\s+(.+)$", line)
if header_match:
if current_body:
sections.append((current_header, "\n".join(current_body).strip()))
current_level = len(header_match.group(1))
current_header = header_match.group(2).strip()
current_body = []
else:
current_body.append(line)
if current_body:
sections.append((current_header, "\n".join(current_body).strip()))
chunks = []
for i, (header, body) in enumerate(sections):
if not body:
continue
text = f"[{header}] {body}"
if _estimate_tokens(text) > chunk_size:
sub_chunks = _recursive_split(body, chunk_size)
for j, sub in enumerate(sub_chunks):
chunks.append(
{
"text": f"[{header}] {sub}",
"source_file": filename,
"chunk_index": i * 1000 + j,
"doc_type": "markdown",
"section_header": header,
}
)
else:
chunks.append(
{
"text": text,
"source_file": filename,
"chunk_index": i,
"doc_type": "markdown",
"section_header": header,
}
)
return chunks
def chunk_json_array(
content: str, filename: str, chunk_size: int = DEFAULT_CHUNK_SIZE
) -> list[dict]:
try:
data = json.loads(content)
except json.JSONDecodeError:
return [
{
"text": content,
"source_file": filename,
"chunk_index": 0,
"doc_type": "json_array",
}
]
if not isinstance(data, list):
text = json.dumps(data, indent=2)
return [
{
"text": text,
"source_file": filename,
"chunk_index": 0,
"doc_type": "json_object",
}
]
chunks = []
for i, item in enumerate(data):
if isinstance(item, dict):
parts = []
for key in (
"text",
"message",
"content",
"subject",
"description",
"resolution",
"body",
):
if item.get(key):
parts.append(f"{key}: {item[key]}")
for key in (
"user",
"author",
"channel",
"priority",
"customer_plan",
"status",
):
if item.get(key):
parts.append(f"{key}: {item[key]}")
text = " | ".join(parts)
if not text:
text = json.dumps(item)
elif isinstance(item, str):
text = item
else:
text = json.dumps(item)
if text:
chunks.append(
{
"text": text,
"source_file": filename,
"chunk_index": i,
"doc_type": "json_array",
}
)
return chunks
def chunk_csv(
content: str, filename: str, chunk_size: int = DEFAULT_CHUNK_SIZE
) -> list[dict]:
reader = csv.DictReader(io.StringIO(content))
if reader.fieldnames is None:
return [
{
"text": content,
"source_file": filename,
"chunk_index": 0,
"doc_type": "csv",
}
]
headers = reader.fieldnames
rows = list(reader)
if not rows:
return []
chunks = []
batch = []
batch_text = ""
for i, row in enumerate(rows):
row_parts = [f"{k}: {v}" for k, v in row.items() if v]
row_str = " | ".join(row_parts)
if _estimate_tokens(batch_text + "\n" + row_str) > chunk_size and batch:
chunks.append(
{
"text": batch_text,
"source_file": filename,
"chunk_index": len(chunks),
"doc_type": "csv",
}
)
batch = [row]
batch_text = row_str
else:
if batch_text:
batch_text += "\n"
batch_text += row_str
batch.append(row)
if batch:
chunks.append(
{
"text": batch_text,
"source_file": filename,
"chunk_index": len(chunks),
"doc_type": "csv",
}
)
return chunks
def chunk_html(
content: str, filename: str, chunk_size: int = DEFAULT_CHUNK_SIZE
) -> list[dict]:
text = re.sub(r"<style[^>]*>.*?</style>", "", content, flags=re.DOTALL)
text = re.sub(r"<script[^>]*>.*?</script>", "", text, flags=re.DOTALL)
text = re.sub(r"<[^>]+>", " ", text)
text = re.sub(r"\s+", " ", text).strip()
sections = re.split(r"\n\s*(?=(?:##|###|####|h[1-6]))", text)
chunks = []
for i, section in enumerate(sections):
section = section.strip()
if not section:
continue
if _estimate_tokens(section) > chunk_size:
subs = _recursive_split(section, chunk_size)
for j, sub in enumerate(subs):
chunks.append(
{
"text": sub,
"source_file": filename,
"chunk_index": i * 1000 + j,
"doc_type": "html",
}
)
else:
chunks.append(
{
"text": section,
"source_file": filename,
"chunk_index": i,
"doc_type": "html",
}
)
return (
chunks
if chunks
else [
{
"text": text[: chunk_size * 4],
"source_file": filename,
"chunk_index": 0,
"doc_type": "html",
}
]
)
def chunk_plain_text(
content: str,
filename: str,
chunk_size: int = DEFAULT_CHUNK_SIZE,
overlap: int = DEFAULT_OVERLAP,
) -> list[dict]:
parts = _recursive_split(content, chunk_size, overlap)
return [
{
"text": part,
"source_file": filename,
"chunk_index": i,
"doc_type": "plain_text",
}
for i, part in enumerate(parts)
]
CHUNKERS = {
"markdown": chunk_markdown,
"json_array": chunk_json_array,
"json_object": chunk_json_array,
"csv": chunk_csv,
"html": chunk_html,
"plain_text": chunk_plain_text,
}
def get_chunker(doc_type: str):
return CHUNKERS.get(doc_type, chunk_plain_text)
|