spectre_pdf 1.0.0

Native Rust PDF extraction engine: text, markdown for RAG, AcroForm widgets, image decoding, and encrypted PDFs. Lazy parser, persistent Document handle, no C dependencies.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

plaintext baseline: 37559 chars
==============================================================================================================
variant                cipher                      pmu ms   spec ms   speedup  text match
--------------------------------------------------------------------------------------------------------------
V=1 R=2                RC4 40-bit                   12.80      1.66     7.71x          OK
V=2 R=3                RC4 128-bit                  11.33      1.73     6.54x          OK
V=4 R=4                AES-128 CBC                  12.73      1.63     7.79x          OK
V=4 R=4                AES-128, empty user          13.08      1.63     8.03x          OK
V=5 R=6                AES-256 (Alg 2.B)            29.01      2.24    12.93x          OK
V=5 R=6                AES-256, empty user          35.70      2.21    16.14x          OK
--------------------------------------------------------------------------------------------------------------
TOTAL  speedup 10.32x (114.6ms / 11.1ms)
  spectre baseline-text match:  6/6
  pymupdf baseline-text match:  0/6

Wrong-password test (expect both libraries to reject):
  v4_aes_128.pdf: spectre rejected (ValueError)
  v4_aes_128.pdf: pymupdf rc=0
  v5_aes_256.pdf: spectre rejected (ValueError)
  v5_aes_256.pdf: pymupdf rc=0