OpenSystemsDevelopment / qpdf

Commit 9f444ffef3c11201d0a460b14b6234d3319ce861

Authored by Jay Berkenbilt
1 parent 359999a5

add QPDF::processMemoryFile and API additions to support it 

git-svn-id: svn+q:///qpdf/trunk@1034 71b93d88-0707-0410-a8cf-f5a4172ac649
Inline Side-by-side
Showing 7 changed files with 121 additions and 57 deletions
ChangeLog
  View file @9f444ff
  1 +2010-10-01 Jay Berkenbilt <ejb@ql.org>
  2 +
  3 + * include/qpdf/QPDF.hh: Add processMemoryFile method for
  4 + processing a PDF file from a memory buffer instead of a file.
  5 +
1 6 2010-09-24 Jay Berkenbilt <ejb@ql.org>
2 7  
3 8 * libqpdf/QPDF.cc: change private "file" method to be a
... ...
1   -2.2.1
2   -=====
3   -
4   -Add interface for working with in-memory PDF files. Here's some code
5   -to work with.
6   -
7   - Pl_Buffer b("b");
8   - FILE* f = QUtil::fopen_wrapper(std::string("open ") + filename,
9   - fopen(filename, "rb"));
10   - unsigned char buf[1024];
11   - size_t bytes_read = 0;
12   - while (true)
13   - {
14   - size_t len = fread(buf, 1, sizeof(buf), f);
15   - if (len == 0)
16   - {
17   - if (ferror(f))
18   - {
19   - throw QPDFExc(qpdf_e_system,
20   - filename, "",
21   - 0,
22   - "read");
23   - }
24   - else
25   - {
26   - b.finish();
27   - break;
28   - }
29   - }
30   - else
31   - {
32   - b.write(buf, len);
33   - bytes_read += len;
34   - }
35   - }
36   - fclose(f);
37   - this->file = new BufferInputSource(filename, b.getBuffer());
38   -
39 1 General
40 2 =======
41 3  
... ...
include/qpdf/Buffer.hh
  View file @9f444ff
... ... @@ -15,8 +15,17 @@ class Buffer
15 15 public:
16 16 QPDF_DLL
17 17 Buffer();
  18 +
  19 + // Create a Buffer object whose memory is owned by the class and
  20 + // will be freed when the Buffer object is destroyed.
18 21 QPDF_DLL
19 22 Buffer(unsigned long size);
  23 +
  24 + // Create a Buffer object whose memory is owned by the caller and
  25 + // will not be freed when the Buffer is destroyed.
  26 + QPDF_DLL
  27 + Buffer(unsigned char* buf, unsigned long size);
  28 +
20 29 QPDF_DLL
21 30 Buffer(Buffer const&);
22 31 QPDF_DLL
... ... @@ -31,10 +40,11 @@ class Buffer
31 40 unsigned char* getBuffer();
32 41  
33 42 private:
34   - void init(unsigned long size);
  43 + void init(unsigned long size, unsigned char* buf, bool own_memory);
35 44 void copy(Buffer const&);
36 45 void destroy();
37 46  
  47 + bool own_memory;
38 48 unsigned long size;
39 49 unsigned char* buf;
40 50 };
... ...
include/qpdf/QPDF.hh
  View file @9f444ff
... ... @@ -50,6 +50,15 @@ class QPDF
50 50 QPDF_DLL
51 51 void processFile(char const* filename, char const* password = 0);
52 52  
  53 + // Parse a PDF file loaded into a memory buffer. This works
  54 + // exactly like processFile except that the PDF file is in memory
  55 + // instead of on disk. The description appears in any warning or
  56 + // error message in place of the file name.
  57 + QPDF_DLL
  58 + void processMemoryFile(char const* description,
  59 + char const* buf, size_t length,
  60 + char const* password = 0);
  61 +
53 62 // Parameter settings
54 63  
55 64 // If true, ignore any cross-reference streams in a hybrid file
... ... @@ -362,7 +371,8 @@ class QPDF
362 371 class BufferInputSource: public InputSource
363 372 {
364 373 public:
365   - BufferInputSource(std::string const& description, Buffer* buf);
  374 + BufferInputSource(std::string const& description, Buffer* buf,
  375 + bool own_memory = false);
366 376 virtual ~BufferInputSource();
367 377 virtual std::string const& getName() const;
368 378 virtual off_t tell();
... ... @@ -372,6 +382,7 @@ class QPDF
372 382 virtual void unreadCh(char ch);
373 383  
374 384 private:
  385 + bool own_memory;
375 386 std::string description;
376 387 Buffer* buf;
377 388 off_t cur_offset;
... ... @@ -410,7 +421,7 @@ class QPDF
410 421 off_t end_after_space;
411 422 };
412 423  
413   - void parse();
  424 + void parse(char const* password);
414 425 void warn(QPDFExc const& e);
415 426 void setTrailer(QPDFObjectHandle obj);
416 427 void read_xref(off_t offset);
... ...
libqpdf/Buffer.cc
  View file @9f444ff
... ... @@ -4,17 +4,22 @@
4 4  
5 5 Buffer::Buffer()
6 6 {
7   - init(0);
  7 + init(0, 0, true);
8 8 }
9 9  
10 10 Buffer::Buffer(unsigned long size)
11 11 {
12   - init(size);
  12 + init(size, 0, true);
  13 +}
  14 +
  15 +Buffer::Buffer(unsigned char* buf, unsigned long size)
  16 +{
  17 + init(size, buf, false);
13 18 }
14 19  
15 20 Buffer::Buffer(Buffer const& rhs)
16 21 {
17   - init(0);
  22 + init(0, 0, true);
18 23 copy(rhs);
19 24 }
20 25  
... ... @@ -31,10 +36,18 @@ Buffer::~Buffer()
31 36 }
32 37  
33 38 void
34   -Buffer::init(unsigned long size)
  39 +Buffer::init(unsigned long size, unsigned char* buf, bool own_memory)
35 40 {
  41 + this->own_memory = own_memory;
36 42 this->size = size;
37   - this->buf = (size ? new unsigned char[size] : 0);
  43 + if (own_memory)
  44 + {
  45 + this->buf = (size ? new unsigned char[size] : 0);
  46 + }
  47 + else
  48 + {
  49 + this->buf = buf;
  50 + }
38 51 }
39 52  
40 53 void
... ... @@ -43,7 +56,7 @@ Buffer::copy(Buffer const&amp; rhs)
43 56 if (this != &rhs)
44 57 {
45 58 this->destroy();
46   - this->init(rhs.size);
  59 + this->init(rhs.size, 0, true);
47 60 if (this->size)
48 61 {
49 62 memcpy(this->buf, rhs.buf, this->size);
... ... @@ -54,7 +67,10 @@ Buffer::copy(Buffer const&amp; rhs)
54 67 void
55 68 Buffer::destroy()
56 69 {
57   - delete [] this->buf;
  70 + if (this->own_memory)
  71 + {
  72 + delete [] this->buf;
  73 + }
58 74 this->size = 0;
59 75 this->buf = 0;
60 76 }
... ...
libqpdf/QPDF.cc
  View file @9f444ff
... ... @@ -159,7 +159,8 @@ QPDF::FileInputSource::unreadCh(char ch)
159 159 }
160 160  
161 161 QPDF::BufferInputSource::BufferInputSource(std::string const& description,
162   - Buffer* buf) :
  162 + Buffer* buf, bool own_memory) :
  163 + own_memory(own_memory),
163 164 description(description),
164 165 buf(buf),
165 166 cur_offset(0)
... ... @@ -168,6 +169,10 @@ QPDF::BufferInputSource::BufferInputSource(std::string const&amp; description,
168 169  
169 170 QPDF::BufferInputSource::~BufferInputSource()
170 171 {
  172 + if (own_memory)
  173 + {
  174 + delete this->buf;
  175 + }
171 176 }
172 177  
173 178 std::string const&
... ... @@ -192,7 +197,7 @@ QPDF::BufferInputSource::seek(off_t offset, int whence)
192 197 break;
193 198  
194 199 case SEEK_END:
195   - this->cur_offset = this->buf->getSize() - offset;
  200 + this->cur_offset = this->buf->getSize() + offset;
196 201 break;
197 202  
198 203 case SEEK_CUR:
... ... @@ -306,11 +311,19 @@ QPDF::processFile(char const* filename, char const* password)
306 311 FileInputSource* fi = new FileInputSource();
307 312 this->file = fi;
308 313 fi->setFilename(filename);
309   - if (password)
310   - {
311   - this->provided_password = password;
312   - }
313   - parse();
  314 + parse(password);
  315 +}
  316 +
  317 +void
  318 +QPDF::processMemoryFile(char const* description,
  319 + char const* buf, size_t length,
  320 + char const* password)
  321 +{
  322 + this->file =
  323 + new BufferInputSource(description,
  324 + new Buffer((unsigned char*)buf, length),
  325 + true);
  326 + parse(password);
314 327 }
315 328  
316 329 void
... ... @@ -340,11 +353,16 @@ QPDF::getWarnings()
340 353 }
341 354  
342 355 void
343   -QPDF::parse()
  356 +QPDF::parse(char const* password)
344 357 {
345 358 static PCRE header_re("^%PDF-(1.\\d+)\\b");
346 359 static PCRE eof_re("(?s:startxref\\s+(\\d+)\\s+%%EOF\\b)");
347 360  
  361 + if (password)
  362 + {
  363 + this->provided_password = password;
  364 + }
  365 +
348 366 std::string line = this->file->readLine();
349 367 PCRE::Match m1 = header_re.match(line.c_str());
350 368 if (m1)
... ...
qpdf/test_driver.cc
  View file @9f444ff
... ... @@ -58,11 +58,53 @@ class Provider: public QPDFObjectHandle::StreamDataProvider
58 58 void runtest(int n, char const* filename)
59 59 {
60 60 QPDF pdf;
  61 + PointerHolder<char> file_buf;
61 62 if (n == 0)
62 63 {
63 64 pdf.setAttemptRecovery(false);
64 65 }
65   - pdf.processFile(filename);
  66 + if (n % 2 == 0)
  67 + {
  68 + pdf.processFile(filename);
  69 + }
  70 + else
  71 + {
  72 + // Exercise processMemoryFile
  73 + FILE* f = QUtil::fopen_wrapper(std::string("open ") + filename,
  74 + fopen(filename, "rb"));
  75 + fseek(f, 0, SEEK_END);
  76 + size_t size = (size_t) ftell(f);
  77 + fseek(f, 0, SEEK_SET);
  78 + file_buf = new char[size];
  79 + char* buf_p = file_buf.getPointer();
  80 + size_t bytes_read = 0;
  81 + size_t len = 0;
  82 + while ((len = fread(buf_p + bytes_read, 1, size - bytes_read, f)) > 0)
  83 + {
  84 + bytes_read += len;
  85 + }
  86 + if (bytes_read != size)
  87 + {
  88 + if (ferror(f))
  89 + {
  90 + throw std::runtime_error(
  91 + std::string("failure reading file ") + filename +
  92 + " into memory: read " +
  93 + QUtil::int_to_string(bytes_read) + "; wanted " +
  94 + QUtil::int_to_string(size));
  95 + }
  96 + else
  97 + {
  98 + throw std::logic_error(
  99 + std::string("premature eof reading file ") + filename +
  100 + " into memory: read " +
  101 + QUtil::int_to_string(bytes_read) + "; wanted " +
  102 + QUtil::int_to_string(size));
  103 + }
  104 + }
  105 + fclose(f);
  106 + pdf.processMemoryFile(filename, buf_p, size);
  107 + }
66 108  
67 109 if ((n == 0) || (n == 1))
68 110 {
... ...