OpenSystemsDevelopment / qpdf

Commit 05e52843bf8db42bd59ccba1553f61ebbd70c329

Authored by m-holger
1 parent a686cc5f

Refactor `Pl_Base64` to use `std::string` for output buffering and `std::string_… 

…view` for encoding/decoding, improving memory efficiency and simplifying buffer management
Inline Side-by-side
Showing 2 changed files with 41 additions and 21 deletions
libqpdf/Pl_Base64.cc
  View file @05e5284
  1 +#include <qpdf/assert_debug.h>
  2 +
1 3 #include <qpdf/Pl_Base64.hh>
2 4  
3 5 #include <qpdf/QIntC.hh>
... ... @@ -43,9 +45,12 @@ Pl_Base64::write(unsigned char const* data, size_t len)
43 45 }
44 46  
45 47 void
46   -Pl_Base64::decode(unsigned char const* data, size_t len)
  48 +Pl_Base64::decode(std::string_view data)
47 49 {
48   - unsigned char const* p = data;
  50 + auto len = data.size();
  51 + auto res = (len / 4u + 1u) * 3u;
  52 + out_buffer.reserve(res);
  53 + unsigned char const* p = reinterpret_cast<const unsigned char*>(data.data());
49 54 while (len > 0) {
50 55 if (!util::is_space(to_c(*p))) {
51 56 buf[pos++] = *p;
... ... @@ -56,12 +61,28 @@ Pl_Base64::decode(unsigned char const* data, size_t len)
56 61 ++p;
57 62 --len;
58 63 }
  64 + if (pos > 0) {
  65 + for (size_t i = pos; i < 4; ++i) {
  66 + buf[i] = '=';
  67 + }
  68 + flush_decode();
  69 + }
  70 + qpdf_assert_debug(out_buffer.size() <= res);
59 71 }
60 72  
61 73 void
62   -Pl_Base64::encode(unsigned char const* data, size_t len)
  74 +Pl_Base64::encode(std::string_view data)
63 75 {
64   - unsigned char const* p = data;
  76 + auto len = data.size();
  77 + static const size_t max_len = (std::string().max_size() / 4u - 1u) * 3u;
  78 + // Change to constexpr once AppImage is build with GCC >= 12
  79 + if (len > max_len) {
  80 + throw std::length_error(getIdentifier() + ": base64 decode: data exceeds maximum length");
  81 + }
  82 +
  83 + auto res = (len / 3u + 1u) * 4u;
  84 + out_buffer.reserve(res);
  85 + unsigned char const* p = reinterpret_cast<const unsigned char*>(data.data());
65 86 while (len > 0) {
66 87 buf[pos++] = *p;
67 88 if (pos == 3) {
... ... @@ -70,6 +91,10 @@ Pl_Base64::encode(unsigned char const* data, size_t len)
70 91 ++p;
71 92 --len;
72 93 }
  94 + if (pos > 0) {
  95 + flush_encode();
  96 + }
  97 + qpdf_assert_debug(out_buffer.size() <= res);
73 98 }
74 99  
75 100 void
... ... @@ -78,7 +103,7 @@ Pl_Base64::flush_decode()
78 103 if (end_of_data) {
79 104 throw std::runtime_error(getIdentifier() + ": base64 decode: data follows pad characters");
80 105 }
81   - int pad = 0;
  106 + size_t pad = 0;
82 107 int shift = 18;
83 108 int outval = 0;
84 109 for (size_t i = 0; i < 4; ++i) {
... ... @@ -110,7 +135,7 @@ Pl_Base64::flush_decode()
110 135 to_uc(0xff & outval),
111 136 };
112 137  
113   - next()->write(out, QIntC::to_size(3 - pad));
  138 + out_buffer.append(reinterpret_cast<const char*>(out), 3u - pad);
114 139 reset();
115 140 }
116 141  
... ... @@ -144,7 +169,7 @@ Pl_Base64::flush_encode()
144 169 for (size_t i = 0; i < 3 - pos; ++i) {
145 170 out[3 - i] = '=';
146 171 }
147   - next()->write(out, 4);
  172 + out_buffer.append(reinterpret_cast<const char*>(out), 4);
148 173 reset();
149 174 }
150 175  
... ... @@ -152,21 +177,16 @@ void
152 177 Pl_Base64::finish()
153 178 {
154 179 if (action == a_decode) {
155   - decode(reinterpret_cast<unsigned char const*>(in_buffer.data()), in_buffer.size());
156   - if (pos > 0) {
157   - for (size_t i = pos; i < 4; ++i) {
158   - buf[i] = '=';
159   - }
160   - flush_decode();
161   - }
  180 + decode(in_buffer);
  181 +
162 182 } else {
163   - encode(reinterpret_cast<unsigned char const*>(in_buffer.data()), in_buffer.size());
164   - if (pos > 0) {
165   - flush_encode();
166   - }
  183 + encode(in_buffer);
167 184 }
168 185 in_buffer.clear();
169 186 in_buffer.shrink_to_fit();
  187 + next()->write(reinterpret_cast<unsigned char const*>(out_buffer.data()), out_buffer.size());
  188 + out_buffer.clear();
  189 + out_buffer.shrink_to_fit();
170 190 next()->finish();
171 191 }
172 192  
... ...
libqpdf/qpdf/Pl_Base64.hh
  View file @05e5284
... ... @@ -13,8 +13,8 @@ class Pl_Base64 final: public Pipeline
13 13 void finish() final;
14 14  
15 15 private:
16   - void decode(unsigned char const* buf, size_t len);
17   - void encode(unsigned char const* buf, size_t len);
  16 + void decode(std::string_view data);
  17 + void encode(std::string_view data);
18 18 void flush_decode();
19 19 void flush_encode();
20 20 void reset();
... ... @@ -23,8 +23,8 @@ class Pl_Base64 final: public Pipeline
23 23 unsigned char buf[4]{0, 0, 0, 0};
24 24 size_t pos{0};
25 25 std::string in_buffer;
  26 + std::string out_buffer;
26 27 bool end_of_data{false};
27   - bool finished{false};
28 28 };
29 29  
30 30 #endif // PL_BASE64_HH
... ...