Spaces:
Sleeping
Sleeping
Update pipelines/anonymize.py
Browse files- pipelines/anonymize.py +27 -43
pipelines/anonymize.py
CHANGED
|
@@ -2,63 +2,47 @@ import re
|
|
| 2 |
from typing import Tuple, Dict
|
| 3 |
from reportlab.lib.pagesizes import A4
|
| 4 |
from reportlab.pdfgen import canvas
|
| 5 |
-
from
|
| 6 |
-
import io
|
| 7 |
|
| 8 |
-
#
|
| 9 |
EMAIL_RE = re.compile(r"[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}")
|
| 10 |
PHONE_RE = re.compile(r"(?:\+?\d{1,3}[ -]?)?(?:\(\d{2,4}\)[ -]?)?\d{2,4}[ -]?\d{2,4}[ -]?\d{3,4}")
|
| 11 |
-
|
| 12 |
|
| 13 |
def anonymize_text(text: str) -> Tuple[str, Dict[str, str]]:
|
| 14 |
mapping = {}
|
| 15 |
-
# メール
|
| 16 |
def _mask_email(m):
|
| 17 |
-
|
| 18 |
-
mapping[
|
| 19 |
-
return key + "@masked"
|
| 20 |
-
text = EMAIL_RE.sub(_mask_email, text)
|
| 21 |
-
|
| 22 |
-
# 電話
|
| 23 |
def _mask_phone(m):
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
if len(re.sub(r"\D", "", ph)) < 7:
|
| 27 |
-
return ph
|
| 28 |
-
key = f"TEL_{len(mapping)+1}"
|
| 29 |
-
mapping[ph] = key
|
| 30 |
-
return key
|
| 31 |
-
text = PHONE_RE.sub(_mask_phone, text)
|
| 32 |
-
|
| 33 |
-
# 氏名行
|
| 34 |
-
def _mask_name(m):
|
| 35 |
-
val = m.group(1).strip()
|
| 36 |
-
key = f"NAME_{len(mapping)+1}"
|
| 37 |
-
mapping[val] = key
|
| 38 |
-
return m.group(0).replace(val, key)
|
| 39 |
-
text = NAME_LINE_RE.sub(_mask_name, text)
|
| 40 |
|
| 41 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
|
| 43 |
def render_anonymized_pdf(text: str) -> bytes:
|
| 44 |
-
buf =
|
| 45 |
c = canvas.Canvas(buf, pagesize=A4)
|
| 46 |
width, height = A4
|
| 47 |
-
|
| 48 |
-
# 素朴なテキスト描画(自動改頁)
|
| 49 |
-
margin_x = 15 * mm
|
| 50 |
-
margin_y = 15 * mm
|
| 51 |
-
y = height - margin_y
|
| 52 |
-
c.setFont("Helvetica", 10)
|
| 53 |
-
|
| 54 |
for line in text.splitlines():
|
| 55 |
-
if y <
|
| 56 |
c.showPage()
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
y -= 12
|
| 61 |
-
|
| 62 |
-
c.showPage()
|
| 63 |
c.save()
|
| 64 |
return buf.getvalue()
|
|
|
|
| 2 |
from typing import Tuple, Dict
|
| 3 |
from reportlab.lib.pagesizes import A4
|
| 4 |
from reportlab.pdfgen import canvas
|
| 5 |
+
from io import BytesIO
|
|
|
|
| 6 |
|
| 7 |
+
# ごく簡易な匿名化(メール/電話/氏名候補っぽい行)
|
| 8 |
EMAIL_RE = re.compile(r"[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}")
|
| 9 |
PHONE_RE = re.compile(r"(?:\+?\d{1,3}[ -]?)?(?:\(\d{2,4}\)[ -]?)?\d{2,4}[ -]?\d{2,4}[ -]?\d{3,4}")
|
| 10 |
+
NAME_HINT = re.compile(r"^\s*(氏名|Name)\s*[::]?\s*(.+)$")
|
| 11 |
|
| 12 |
def anonymize_text(text: str) -> Tuple[str, Dict[str, str]]:
|
| 13 |
mapping = {}
|
|
|
|
| 14 |
def _mask_email(m):
|
| 15 |
+
val = m.group(0); k = f"EMAIL_{len(mapping)+1}"
|
| 16 |
+
mapping[val] = k; return k + "@masked"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
def _mask_phone(m):
|
| 18 |
+
val = m.group(0); k = f"TEL_{len(mapping)+1}"
|
| 19 |
+
mapping[val] = k; return k
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
+
# line単位で氏名候補をマスク
|
| 22 |
+
lines = []
|
| 23 |
+
for line in text.splitlines():
|
| 24 |
+
nm = NAME_HINT.search(line)
|
| 25 |
+
if nm:
|
| 26 |
+
full = nm.group(0)
|
| 27 |
+
mapping[full] = "NAME_MASKED"
|
| 28 |
+
lines.append("氏名: NAME_MASKED")
|
| 29 |
+
else:
|
| 30 |
+
lines.append(line)
|
| 31 |
+
masked = "\n".join(lines)
|
| 32 |
+
masked = EMAIL_RE.sub(_mask_email, masked)
|
| 33 |
+
masked = PHONE_RE.sub(_mask_phone, masked)
|
| 34 |
+
return masked, mapping
|
| 35 |
|
| 36 |
def render_anonymized_pdf(text: str) -> bytes:
|
| 37 |
+
buf = BytesIO()
|
| 38 |
c = canvas.Canvas(buf, pagesize=A4)
|
| 39 |
width, height = A4
|
| 40 |
+
x, y = 40, height - 40
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
for line in text.splitlines():
|
| 42 |
+
if y < 40:
|
| 43 |
c.showPage()
|
| 44 |
+
y = height - 40
|
| 45 |
+
c.drawString(x, y, line[:1000]) # 1行制限(超長行ガード)
|
| 46 |
+
y -= 14
|
|
|
|
|
|
|
|
|
|
| 47 |
c.save()
|
| 48 |
return buf.getvalue()
|