OLD | NEW |
1 /* | 1 /* |
2 * Copyright 2013 Google Inc. | 2 * Copyright 2013 Google Inc. |
3 * | 3 * |
4 * Use of this source code is governed by a BSD-style license that can be | 4 * Use of this source code is governed by a BSD-style license that can be |
5 * found in the LICENSE file. | 5 * found in the LICENSE file. |
6 */ | 6 */ |
7 | 7 |
8 #ifndef SkPdfNativeTokenizer_DEFINED | 8 #ifndef SkPdfNativeTokenizer_DEFINED |
9 #define SkPdfNativeTokenizer_DEFINED | 9 #define SkPdfNativeTokenizer_DEFINED |
10 | 10 |
11 #include "SkPdfConfig.h" | |
12 | |
13 #include "SkTDArray.h" | |
14 #include "SkTDict.h" | |
15 #include <math.h> | 11 #include <math.h> |
16 #include <string.h> | 12 #include <string.h> |
17 | 13 |
| 14 #include "SkPdfConfig.h" |
| 15 #include "SkTDArray.h" |
| 16 #include "SkTDict.h" |
| 17 |
| 18 // All these constants are defined by the PDF 1.4 Spec. |
| 19 |
18 class SkPdfDictionary; | 20 class SkPdfDictionary; |
19 class SkPdfImageDictionary; | 21 class SkPdfImageDictionary; |
| 22 class SkPdfNativeDoc; |
| 23 class SkPdfNativeObject; |
| 24 |
20 | 25 |
21 // White Spaces | 26 // White Spaces |
22 #define kNUL_PdfWhiteSpace '\x00' | 27 #define kNUL_PdfWhiteSpace '\x00' |
23 #define kHT_PdfWhiteSpace '\x09' | 28 #define kHT_PdfWhiteSpace '\x09' |
24 #define kLF_PdfWhiteSpace '\x0A' | 29 #define kLF_PdfWhiteSpace '\x0A' |
25 #define kFF_PdfWhiteSpace '\x0C' | 30 #define kFF_PdfWhiteSpace '\x0C' |
26 #define kCR_PdfWhiteSpace '\x0D' | 31 #define kCR_PdfWhiteSpace '\x0D' |
27 #define kSP_PdfWhiteSpace '\x20' | 32 #define kSP_PdfWhiteSpace '\x20' |
28 | 33 |
29 // PdfDelimiters | 34 // PdfDelimiters |
(...skipping 12 matching lines...) Expand all Loading... |
42 #define kBackspace_PdfSpecial '\x08' | 47 #define kBackspace_PdfSpecial '\x08' |
43 | 48 |
44 // TODO(edisonn): what is the faster way for compiler/machine type to evaluate t
his expressions? | 49 // TODO(edisonn): what is the faster way for compiler/machine type to evaluate t
his expressions? |
45 // we should evaluate all options. might be even different from one machine to a
nother | 50 // we should evaluate all options. might be even different from one machine to a
nother |
46 // 1) expand expression, let compiler optimize it | 51 // 1) expand expression, let compiler optimize it |
47 // 2) binary search | 52 // 2) binary search |
48 // 3) linear search in array | 53 // 3) linear search in array |
49 // 4) vector (e.f. T type[256] .. return type[ch] ... | 54 // 4) vector (e.f. T type[256] .. return type[ch] ... |
50 // 5) manually build the expression with least number of operators, e.g. for con
secutive | 55 // 5) manually build the expression with least number of operators, e.g. for con
secutive |
51 // chars, we can use an binary equal ignoring last bit | 56 // chars, we can use an binary equal ignoring last bit |
52 #define isPdfWhiteSpace(ch) (((ch)==kNUL_PdfWhiteSpace)||((ch)==kHT_PdfWhiteSpac
e)||((ch)==kLF_PdfWhiteSpace)||((ch)==kFF_PdfWhiteSpace)||((ch)==kCR_PdfWhiteSpa
ce)||((ch)==kSP_PdfWhiteSpace)) | 57 #define isPdfWhiteSpace(ch) (((ch)==kNUL_PdfWhiteSpace)|| \ |
| 58 ((ch)==kHT_PdfWhiteSpace)|| \ |
| 59 ((ch)==kLF_PdfWhiteSpace)|| \ |
| 60 ((ch)==kFF_PdfWhiteSpace)|| \ |
| 61 ((ch)==kCR_PdfWhiteSpace)|| \ |
| 62 ((ch)==kSP_PdfWhiteSpace)) |
53 | 63 |
54 #define isPdfEOL(ch) (((ch)==kLF_PdfWhiteSpace)||((ch)==kCR_PdfWhiteSpace)) | 64 #define isPdfEOL(ch) (((ch)==kLF_PdfWhiteSpace)||((ch)==kCR_PdfWhiteSpace)) |
55 | 65 |
56 | 66 |
57 #define isPdfDelimiter(ch) (((ch)==kOpenedRoundBracket_PdfDelimiter)||\ | 67 #define isPdfDelimiter(ch) (((ch)==kOpenedRoundBracket_PdfDelimiter)||\ |
58 ((ch)==kClosedRoundBracket_PdfDelimiter)||\ | 68 ((ch)==kClosedRoundBracket_PdfDelimiter)||\ |
59 ((ch)==kOpenedInequityBracket_PdfDelimiter)||\ | 69 ((ch)==kOpenedInequityBracket_PdfDelimiter)||\ |
60 ((ch)==kClosedInequityBracket_PdfDelimiter)||\ | 70 ((ch)==kClosedInequityBracket_PdfDelimiter)||\ |
61 ((ch)==kOpenedSquareBracket_PdfDelimiter)||\ | 71 ((ch)==kOpenedSquareBracket_PdfDelimiter)||\ |
62 ((ch)==kClosedSquareBracket_PdfDelimiter)||\ | 72 ((ch)==kClosedSquareBracket_PdfDelimiter)||\ |
63 ((ch)==kOpenedCurlyBracket_PdfDelimiter)||\ | 73 ((ch)==kOpenedCurlyBracket_PdfDelimiter)||\ |
64 ((ch)==kClosedCurlyBracket_PdfDelimiter)||\ | 74 ((ch)==kClosedCurlyBracket_PdfDelimiter)||\ |
65 ((ch)==kNamed_PdfDelimiter)||\ | 75 ((ch)==kNamed_PdfDelimiter)||\ |
66 ((ch)==kComment_PdfDelimiter)) | 76 ((ch)==kComment_PdfDelimiter)) |
67 | 77 |
68 #define isPdfWhiteSpaceOrPdfDelimiter(ch) (isPdfWhiteSpace(ch)||isPdfDelimiter(c
h)) | 78 #define isPdfWhiteSpaceOrPdfDelimiter(ch) (isPdfWhiteSpace(ch)||isPdfDelimiter(c
h)) |
69 | 79 |
70 #define isPdfDigit(ch) ((ch)>='0'&&(ch)<='9') | 80 #define isPdfDigit(ch) ((ch)>='0'&&(ch)<='9') |
71 #define isPdfNumeric(ch) (isPdfDigit(ch)||(ch)=='+'||(ch)=='-'||(ch)=='.') | 81 #define isPdfNumeric(ch) (isPdfDigit(ch)||(ch)=='+'||(ch)=='-'||(ch)=='.') |
72 | 82 |
73 const unsigned char* skipPdfWhiteSpaces(const unsigned char* buffer, const unsig
ned char* end); | 83 const unsigned char* skipPdfWhiteSpaces(const unsigned char* buffer, const unsig
ned char* end); |
74 const unsigned char* endOfPdfToken(const unsigned char* start, const unsigned ch
ar* end); | 84 const unsigned char* endOfPdfToken(const unsigned char* start, const unsigned ch
ar* end); |
75 | 85 |
76 // TODO(edisonn): typedef read and integer tyepes? make less readable... | 86 #define BUFFER_SIZE 1024 |
77 //typedef double SkPdfReal; | |
78 //typedef int64_t SkPdfInteger; | |
79 | 87 |
80 // an allocator only allocates memory, and it deletes it all when the allocator
is destroyed | 88 /** \class SkPdfAllocator |
81 // this would allow us not to do any garbage collection while we parse or draw a
pdf, and defere it | 89 * |
82 // while the user is looking at the image | 90 * An allocator only allocates memory, and it deletes it all when the allocato
r is destroyed. |
83 | 91 * This strategy would allow us not to do any garbage collection while we pars
e and/or render |
84 class SkPdfNativeObject; | 92 * a pdf. |
85 | 93 * |
| 94 */ |
86 class SkPdfAllocator { | 95 class SkPdfAllocator { |
87 #define BUFFER_SIZE 1024 | |
88 SkTDArray<SkPdfNativeObject*> fHistory; | |
89 SkTDArray<void*> fHandles; | |
90 SkPdfNativeObject* fCurrent; | |
91 int fCurrentUsed; | |
92 | |
93 SkPdfNativeObject* allocBlock(); | |
94 size_t fSizeInBytes; | |
95 | |
96 public: | 96 public: |
97 SkPdfAllocator() { | 97 SkPdfAllocator() { |
98 fSizeInBytes = sizeof(*this); | 98 fSizeInBytes = sizeof(*this); |
99 fCurrent = allocBlock(); | 99 fCurrent = allocBlock(); |
100 fCurrentUsed = 0; | 100 fCurrentUsed = 0; |
101 } | 101 } |
102 | 102 |
103 ~SkPdfAllocator(); | 103 ~SkPdfAllocator(); |
104 | 104 |
| 105 // Allocates an object. It will be reset automatically when ~SkPdfAllocator(
) is called. |
105 SkPdfNativeObject* allocObject(); | 106 SkPdfNativeObject* allocObject(); |
106 | 107 |
107 // TODO(edisonn): free this memory in destructor, track the usage? | 108 // Allocates a buffer. It will be freed automatically when ~SkPdfAllocator()
is called. |
108 void* alloc(size_t bytes) { | 109 void* alloc(size_t bytes) { |
109 void* data = malloc(bytes); | 110 void* data = malloc(bytes); |
110 fHandles.push(data); | 111 fHandles.push(data); |
111 fSizeInBytes += bytes; | 112 fSizeInBytes += bytes; |
112 return data; | 113 return data; |
113 } | 114 } |
114 | 115 |
| 116 // Returns the number of bytes used in this allocator. |
115 size_t bytesUsed() const { | 117 size_t bytesUsed() const { |
116 return fSizeInBytes; | 118 return fSizeInBytes; |
117 } | 119 } |
| 120 |
| 121 private: |
| 122 SkTDArray<SkPdfNativeObject*> fHistory; |
| 123 SkTDArray<void*> fHandles; |
| 124 SkPdfNativeObject* fCurrent; |
| 125 int fCurrentUsed; |
| 126 |
| 127 SkPdfNativeObject* allocBlock(); |
| 128 size_t fSizeInBytes; |
118 }; | 129 }; |
119 | 130 |
120 class SkPdfNativeDoc; | 131 // Type of a parsed token. |
121 const unsigned char* nextObject(const unsigned char* start, const unsigned char*
end, SkPdfNativeObject* token, SkPdfAllocator* allocator, SkPdfNativeDoc* doc); | |
122 | |
123 enum SkPdfTokenType { | 132 enum SkPdfTokenType { |
124 kKeyword_TokenType, | 133 kKeyword_TokenType, |
125 kObject_TokenType, | 134 kObject_TokenType, |
126 }; | 135 }; |
127 | 136 |
| 137 |
| 138 /** \struct PdfToken |
| 139 * |
| 140 * Stores the result of the parsing - a keyword or an object. |
| 141 * |
| 142 */ |
128 struct PdfToken { | 143 struct PdfToken { |
129 const char* fKeyword; | 144 const char* fKeyword; |
130 size_t fKeywordLength; | 145 size_t fKeywordLength; |
131 SkPdfNativeObject* fObject; | 146 SkPdfNativeObject* fObject; |
132 SkPdfTokenType fType; | 147 SkPdfTokenType fType; |
133 | 148 |
134 PdfToken() : fKeyword(NULL), fKeywordLength(0), fObject(NULL) {} | 149 PdfToken() : fKeyword(NULL), fKeywordLength(0), fObject(NULL) {} |
135 }; | 150 }; |
136 | 151 |
| 152 /** \class SkPdfNativeTokenizer |
| 153 * |
| 154 * Responsible to tokenize a stream in small tokens, eityh a keyword or an obj
ect. |
| 155 * A renderer can feed on the tokens and render a pdf. |
| 156 * |
| 157 */ |
137 class SkPdfNativeTokenizer { | 158 class SkPdfNativeTokenizer { |
138 public: | 159 public: |
139 SkPdfNativeTokenizer(SkPdfNativeObject* objWithStream, SkPdfAllocator* alloc
ator, SkPdfNativeDoc* doc); | 160 SkPdfNativeTokenizer(SkPdfNativeObject* objWithStream, |
140 SkPdfNativeTokenizer(const unsigned char* buffer, int len, SkPdfAllocator* a
llocator, SkPdfNativeDoc* doc); | 161 SkPdfAllocator* allocator, SkPdfNativeDoc* doc); |
| 162 SkPdfNativeTokenizer(const unsigned char* buffer, int len, |
| 163 SkPdfAllocator* allocator, SkPdfNativeDoc* doc); |
141 | 164 |
142 virtual ~SkPdfNativeTokenizer(); | 165 virtual ~SkPdfNativeTokenizer(); |
143 | 166 |
| 167 // Reads one token. Returns false if there are no more tokens. |
144 bool readToken(PdfToken* token); | 168 bool readToken(PdfToken* token); |
145 bool readTokenCore(PdfToken* token); | 169 |
| 170 // Put back a token to be read in the nextToken read. Only one token is allo
wed to be put |
| 171 // back. Must not necesaarely be the last token read. |
146 void PutBack(PdfToken token); | 172 void PutBack(PdfToken token); |
| 173 |
| 174 // Reads the inline image that is present in the stream. At this point we ju
st consumed the ID |
| 175 // token already. |
147 SkPdfImageDictionary* readInlineImage(); | 176 SkPdfImageDictionary* readInlineImage(); |
148 | 177 |
149 private: | 178 private: |
| 179 bool readTokenCore(PdfToken* token); |
| 180 |
150 SkPdfNativeDoc* fDoc; | 181 SkPdfNativeDoc* fDoc; |
151 SkPdfAllocator* fAllocator; | 182 SkPdfAllocator* fAllocator; |
152 | 183 |
153 const unsigned char* fUncompressedStreamStart; | 184 const unsigned char* fUncompressedStreamStart; |
154 const unsigned char* fUncompressedStream; | 185 const unsigned char* fUncompressedStream; |
155 const unsigned char* fUncompressedStreamEnd; | 186 const unsigned char* fUncompressedStreamEnd; |
156 | 187 |
157 bool fEmpty; | 188 bool fEmpty; |
158 bool fHasPutBack; | 189 bool fHasPutBack; |
159 PdfToken fPutBack; | 190 PdfToken fPutBack; |
160 }; | 191 }; |
161 | 192 |
| 193 const unsigned char* nextObject(const unsigned char* start, const unsigned char*
end, |
| 194 SkPdfNativeObject* token, |
| 195 SkPdfAllocator* allocator, |
| 196 SkPdfNativeDoc* doc); |
| 197 |
162 #endif // SkPdfNativeTokenizer_DEFINED | 198 #endif // SkPdfNativeTokenizer_DEFINED |
OLD | NEW |