dany/pdfio
1
0
Fork 0
mirror of https://github.com/michaelrsweet/pdfio.git synced 2025-02-18 18:02:51 +01:00
Code Issues Packages Projects Releases Wiki Activity

Rework token reading to be separate from the PDF file so we can add support

Browse Source 
for compressed object streams.
This commit is contained in:
Michael R Sweet 2021-05-08 07:38:44 -04:00
parent 923409ceff
commit 9a919d478f
No known key found for this signature in database
GPG Key ID: 999559A027815955
9 changed files with 211 additions and 166 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
pdfio-array.c
View File

@ -517,7 +517,8 @@ _pdfioArrayGetValue(pdfio_array_t *a, // I - Array
// //
pdfio_array_t * // O - New array pdfio_array_t * // O - New array
_pdfioArrayRead(pdfio_file_t *pdf) // I - PDF file _pdfioArrayRead(pdfio_file_t *pdf, // I - PDF file
_pdfio_token_t *tb) // I - Token buffer/stack
{ {
pdfio_array_t *array; // New array pdfio_array_t *array; // New array
char token[8192]; // Token from file char token[8192]; // Token from file
@ -528,7 +529,7 @@ _pdfioArrayRead(pdfio_file_t *pdf) // I - PDF file
array = pdfioArrayCreate(pdf); array = pdfioArrayCreate(pdf);
// Read until we get "]" to end the array... // Read until we get "]" to end the array...
while (_pdfioFileGetToken(pdf, token, sizeof(token))) while (_pdfioTokenGet(tb, token, sizeof(token)))
{ {
if (!strcmp(token, "]")) if (!strcmp(token, "]"))
{ {
@ -537,8 +538,8 @@ _pdfioArrayRead(pdfio_file_t *pdf) // I - PDF file
} }
// Push the token and decode the value... // Push the token and decode the value...
_pdfioFilePushToken(pdf, token); _pdfioTokenPush(tb, token);
if (!_pdfioValueRead(pdf, &value)) if (!_pdfioValueRead(pdf, tb, &value))
break; break;
append_value(array, &value); append_value(array, &value);

63
pdfio-common.c
View File

@ -23,24 +23,6 @@ static ssize_t read_buffer(pdfio_file_t *pdf, char *buffer, size_t bytes);
static bool write_buffer(pdfio_file_t *pdf, const void *buffer, size_t bytes); static bool write_buffer(pdfio_file_t *pdf, const void *buffer, size_t bytes);
//
// '_pdfioFileClearTokens()' - Clear the token stack.
//
void
_pdfioFileClearTokens(pdfio_file_t *pdf)// I - PDF file
{
PDFIO_DEBUG("_pdfioFileClearTokens(pdf=%p)\n", pdf);
while (pdf->num_tokens > 0)
{
pdf->num_tokens --;
free(pdf->tokens[pdf->num_tokens]);
pdf->tokens[pdf->num_tokens] = NULL;
}
}
// //
// '_pdfioFileConsume()' - Consume bytes from the file. // '_pdfioFileConsume()' - Consume bytes from the file.
// //
@ -144,35 +126,6 @@ _pdfioFileGetChar(pdfio_file_t *pdf) // I - PDF file
} }
//
// '_pdfioFileGetToken()' - Get a token from a PDF file.
//
bool // O - `true` on success, `false` on failure
_pdfioFileGetToken(pdfio_file_t *pdf, // I - PDF file
char *buffer,// I - String buffer
size_t bufsize)// I - Size of string buffer
{
// See if we have a token waiting on the stack...
if (pdf->num_tokens > 0)
{
// Yes, return it...
pdf->num_tokens --;
strncpy(buffer, pdf->tokens[pdf->num_tokens], bufsize - 1);
buffer[bufsize - 1] = '\0';
PDFIO_DEBUG("_pdfioFileGetToken(pdf=%p, buffer=%p, bufsize=%u): Popping '%s' from stack.\n", pdf, buffer, (unsigned)bufsize, buffer);
free(pdf->tokens[pdf->num_tokens]);
pdf->tokens[pdf->num_tokens] = NULL;
return (true);
}
// No, read a new one...
return (_pdfioTokenRead(pdf, buffer, bufsize, (_pdfio_tpeek_cb_t)_pdfioFilePeek, (_pdfio_tconsume_cb_t)_pdfioFileConsume, pdf));
}
// //
// '_pdfioFileGets()' - Read a line from a PDF file. // '_pdfioFileGets()' - Read a line from a PDF file.
// //
@ -306,22 +259,6 @@ _pdfioFilePrintf(pdfio_file_t *pdf, // I - PDF file
} }
//
// '()' - Push a token on the token stack.
//
void
_pdfioFilePushToken(pdfio_file_t *pdf, // I - PDF file
const char *token)// I - Token
{
if (pdf->num_tokens < (sizeof(pdf->tokens) / sizeof(pdf->tokens[0])))
{
if ((pdf->tokens[pdf->num_tokens ++] = strdup(token)) == NULL)
pdf->num_tokens --;
}
}
// //
// '_pdfioFilePuts()' - Write a literal string to a PDF file. // '_pdfioFilePuts()' - Write a literal string to a PDF file.
// //

7
pdfio-dict.c
View File

@ -373,7 +373,8 @@ _pdfioDictGetValue(pdfio_dict_t *dict, // I - Dictionary
// //
pdfio_dict_t * // O - New dictionary pdfio_dict_t * // O - New dictionary
_pdfioDictRead(pdfio_file_t *pdf) // I - PDF file _pdfioDictRead(pdfio_file_t *pdf, // I - PDF file
_pdfio_token_t *tb) // I - Token buffer/stack
{ {
pdfio_dict_t *dict; // New dictionary pdfio_dict_t *dict; // New dictionary
char key[256]; // Dictionary key char key[256]; // Dictionary key
@ -385,7 +386,7 @@ _pdfioDictRead(pdfio_file_t *pdf) // I - PDF file
// Create a dictionary and start reading... // Create a dictionary and start reading...
dict = pdfioDictCreate(pdf); dict = pdfioDictCreate(pdf);
while (_pdfioFileGetToken(pdf, key, sizeof(key))) while (_pdfioTokenGet(tb, key, sizeof(key)))
{ {
// Get the next key or end-of-dictionary... // Get the next key or end-of-dictionary...
if (!strcmp(key, ">>")) if (!strcmp(key, ">>"))
@ -400,7 +401,7 @@ _pdfioDictRead(pdfio_file_t *pdf) // I - PDF file
} }
// Then get the next value... // Then get the next value...
if (!_pdfioValueRead(pdf, &value)) if (!_pdfioValueRead(pdf, tb, &value))
{ {
_pdfioFileError(pdf, "Missing value for dictionary key."); _pdfioFileError(pdf, "Missing value for dictionary key.");
break; break;

11
pdfio-file.c
View File

@ -521,6 +521,7 @@ load_xref(pdfio_file_t *pdf, // I - PDF file
num_objects, // Number of objects num_objects, // Number of objects
offset; // Offset in file offset; // Offset in file
int generation; // Generation number int generation; // Generation number
_pdfio_token_t tb; // Token buffer/stack
while (!done) while (!done)
@ -581,7 +582,9 @@ load_xref(pdfio_file_t *pdf, // I - PDF file
return (false); return (false);
} }
if (!_pdfioValueRead(pdf, &trailer)) _pdfioTokenInit(&tb, pdf, (_pdfio_tconsume_cb_t)_pdfioFileConsume, (_pdfio_tpeek_cb_t)_pdfioFilePeek, pdf);
if (!_pdfioValueRead(pdf, &tb, &trailer))
{ {
_pdfioFileError(pdf, "Unable to read cross-reference stream dictionary."); _pdfioFileError(pdf, "Unable to read cross-reference stream dictionary.");
return (false); return (false);
@ -594,7 +597,7 @@ load_xref(pdfio_file_t *pdf, // I - PDF file
obj->value = trailer; obj->value = trailer;
if (!_pdfioFileGetToken(pdf, line, sizeof(line)) || strcmp(line, "stream")) if (!_pdfioTokenGet(&tb, line, sizeof(line)) || strcmp(line, "stream"))
{ {
_pdfioFileError(pdf, "Unable to get stream after xref dictionary."); _pdfioFileError(pdf, "Unable to get stream after xref dictionary.");
return (false); return (false);
@ -748,7 +751,9 @@ load_xref(pdfio_file_t *pdf, // I - PDF file
return (false); return (false);
} }
if (!_pdfioValueRead(pdf, &trailer)) _pdfioTokenInit(&tb, pdf, (_pdfio_tconsume_cb_t)_pdfioFileConsume, (_pdfio_tpeek_cb_t)_pdfioFilePeek, pdf);
if (!_pdfioValueRead(pdf, &tb, &trailer))
{ {
_pdfioFileError(pdf, "Unable to read trailer dictionary."); _pdfioFileError(pdf, "Unable to read trailer dictionary.");
return (false); return (false);

5
pdfio-object.c
View File

@ -128,6 +128,7 @@ _pdfioObjLoad(pdfio_obj_t *obj) // I - Object
{ {
char line[1024], // Line from file char line[1024], // Line from file
*ptr; // Pointer into line *ptr; // Pointer into line
_pdfio_token_t tb; // Token buffer/stack
PDFIO_DEBUG("_pdfioObjLoad(obj=%p(%lu)), offset=%lu\n", obj, (unsigned long)obj->number, (unsigned long)obj->offset); PDFIO_DEBUG("_pdfioObjLoad(obj=%p(%lu)), offset=%lu\n", obj, (unsigned long)obj->number, (unsigned long)obj->offset);
@ -167,9 +168,9 @@ _pdfioObjLoad(pdfio_obj_t *obj) // I - Object
} }
// Then grab the object value... // Then grab the object value...
_pdfioFileClearTokens(obj->pdf); _pdfioTokenInit(&tb, obj->pdf, (_pdfio_tconsume_cb_t)_pdfioFileConsume, (_pdfio_tpeek_cb_t)_pdfioFilePeek, obj->pdf);
if (!_pdfioValueRead(obj->pdf, &obj->value)) if (!_pdfioValueRead(obj->pdf, &tb, &obj->value))
{ {
_pdfioFileError(obj->pdf, "Unable to read value for object %lu.", (unsigned long)obj->number); _pdfioFileError(obj->pdf, "Unable to read value for object %lu.", (unsigned long)obj->number);
return (false); return (false);

36
pdfio-private.h
View File

@ -80,6 +80,22 @@ typedef enum _pdfio_predictor_e // PNG predictor constants
_PDFIO_PREDICTOR_PNG_PAETH = 14 // PNG Paeth predictor _PDFIO_PREDICTOR_PNG_PAETH = 14 // PNG Paeth predictor
} _pdfio_predictor_t; } _pdfio_predictor_t;
typedef ssize_t (*_pdfio_tconsume_cb_t)(void *data, size_t bytes);
typedef ssize_t (*_pdfio_tpeek_cb_t)(void *data, void *buffer, size_t bytes);
typedef struct _pdfio_token_s // Token buffer/stack
{
pdfio_file_t *pdf; // PDF file
_pdfio_tconsume_cb_t consume_cb; // Consume callback
_pdfio_tpeek_cb_t peek_cb; // Peek callback
void *cb_data; // Callback data
unsigned char buffer[32], // Buffer
*bufptr, // Pointer into buffer
*bufend; // Last valid byte in buffer
size_t num_tokens; // Number of tokens in stack
char *tokens[4]; // Token stack
} _pdfio_token_t;
typedef struct _pdfio_value_s // Value structure typedef struct _pdfio_value_s // Value structure
{ {
pdfio_valtype_t type; // Type of value pdfio_valtype_t type; // Type of value
@ -163,8 +179,6 @@ struct _pdfio_file_s // PDF file structure
size_t num_strings, // Number of strings size_t num_strings, // Number of strings
alloc_strings; // Allocated strings alloc_strings; // Allocated strings
char **strings; // Nul-terminated strings char **strings; // Nul-terminated strings
size_t num_tokens; // Number of tokens in stack
char *tokens[4]; // Token stack
}; };
struct _pdfio_obj_s // Object struct _pdfio_obj_s // Object
@ -198,9 +212,6 @@ struct _pdfio_stream_s // Stream
*pbuffers[2]; // Predictor buffers, as needed *pbuffers[2]; // Predictor buffers, as needed
}; };
typedef ssize_t (*_pdfio_tconsume_cb_t)(void *data, size_t bytes);
typedef ssize_t (*_pdfio_tpeek_cb_t)(void *data, void *buffer, size_t bytes);
// //
// Functions... // Functions...
@ -211,7 +222,7 @@ extern void _pdfioArrayDebug(pdfio_array_t *a) PDFIO_INTERNAL;
# endif // DEBUG # endif // DEBUG
extern void _pdfioArrayDelete(pdfio_array_t *a) PDFIO_INTERNAL; extern void _pdfioArrayDelete(pdfio_array_t *a) PDFIO_INTERNAL;
extern _pdfio_value_t *_pdfioArrayGetValue(pdfio_array_t *a, size_t n) PDFIO_INTERNAL; extern _pdfio_value_t *_pdfioArrayGetValue(pdfio_array_t *a, size_t n) PDFIO_INTERNAL;
extern pdfio_array_t *_pdfioArrayRead(pdfio_file_t *pdf) PDFIO_INTERNAL; extern pdfio_array_t *_pdfioArrayRead(pdfio_file_t *pdf, _pdfio_token_t *ts) PDFIO_INTERNAL;
extern bool _pdfioArrayWrite(pdfio_array_t *a) PDFIO_INTERNAL; extern bool _pdfioArrayWrite(pdfio_array_t *a) PDFIO_INTERNAL;
# ifdef DEBUG # ifdef DEBUG
@ -219,21 +230,18 @@ extern void _pdfioDictDebug(pdfio_dict_t *dict) PDFIO_INTERNAL;
# endif // DEBUG # endif // DEBUG
extern void _pdfioDictDelete(pdfio_dict_t *dict) PDFIO_INTERNAL; extern void _pdfioDictDelete(pdfio_dict_t *dict) PDFIO_INTERNAL;
extern _pdfio_value_t *_pdfioDictGetValue(pdfio_dict_t *dict, const char *key) PDFIO_INTERNAL; extern _pdfio_value_t *_pdfioDictGetValue(pdfio_dict_t *dict, const char *key) PDFIO_INTERNAL;
extern pdfio_dict_t *_pdfioDictRead(pdfio_file_t *pdf) PDFIO_INTERNAL; extern pdfio_dict_t *_pdfioDictRead(pdfio_file_t *pdf, _pdfio_token_t *ts) PDFIO_INTERNAL;
extern bool _pdfioDictSetValue(pdfio_dict_t *dict, const char *key, _pdfio_value_t *value) PDFIO_INTERNAL; extern bool _pdfioDictSetValue(pdfio_dict_t *dict, const char *key, _pdfio_value_t *value) PDFIO_INTERNAL;
extern bool _pdfioDictWrite(pdfio_dict_t *dict, off_t *length) PDFIO_INTERNAL; extern bool _pdfioDictWrite(pdfio_dict_t *dict, off_t *length) PDFIO_INTERNAL;
extern void _pdfioFileClearTokens(pdfio_file_t *pdf) PDFIO_INTERNAL;
extern bool _pdfioFileConsume(pdfio_file_t *pdf, size_t bytes) PDFIO_INTERNAL; extern bool _pdfioFileConsume(pdfio_file_t *pdf, size_t bytes) PDFIO_INTERNAL;
extern bool _pdfioFileDefaultError(pdfio_file_t *pdf, const char *message, void *data) PDFIO_INTERNAL; extern bool _pdfioFileDefaultError(pdfio_file_t *pdf, const char *message, void *data) PDFIO_INTERNAL;
extern bool _pdfioFileError(pdfio_file_t *pdf, const char *format, ...) PDFIO_FORMAT(2,3) PDFIO_INTERNAL; extern bool _pdfioFileError(pdfio_file_t *pdf, const char *format, ...) PDFIO_FORMAT(2,3) PDFIO_INTERNAL;
extern bool _pdfioFileFlush(pdfio_file_t *pdf) PDFIO_INTERNAL; extern bool _pdfioFileFlush(pdfio_file_t *pdf) PDFIO_INTERNAL;
extern int _pdfioFileGetChar(pdfio_file_t *pdf) PDFIO_INTERNAL; extern int _pdfioFileGetChar(pdfio_file_t *pdf) PDFIO_INTERNAL;
extern bool _pdfioFileGetToken(pdfio_file_t *pdf, char *buffer, size_t bufsize) PDFIO_INTERNAL;
extern bool _pdfioFileGets(pdfio_file_t *pdf, char *buffer, size_t bufsize) PDFIO_INTERNAL; extern bool _pdfioFileGets(pdfio_file_t *pdf, char *buffer, size_t bufsize) PDFIO_INTERNAL;
extern ssize_t _pdfioFilePeek(pdfio_file_t *pdf, void *buffer, size_t bytes) PDFIO_INTERNAL; extern ssize_t _pdfioFilePeek(pdfio_file_t *pdf, void *buffer, size_t bytes) PDFIO_INTERNAL;
extern bool _pdfioFilePrintf(pdfio_file_t *pdf, const char *format, ...) PDFIO_FORMAT(2,3) PDFIO_INTERNAL; extern bool _pdfioFilePrintf(pdfio_file_t *pdf, const char *format, ...) PDFIO_FORMAT(2,3) PDFIO_INTERNAL;
extern void _pdfioFilePushToken(pdfio_file_t *pdf, const char *token) PDFIO_INTERNAL;
extern bool _pdfioFilePuts(pdfio_file_t *pdf, const char *s) PDFIO_INTERNAL; extern bool _pdfioFilePuts(pdfio_file_t *pdf, const char *s) PDFIO_INTERNAL;
extern ssize_t _pdfioFileRead(pdfio_file_t *pdf, void *buffer, size_t bytes) PDFIO_INTERNAL; extern ssize_t _pdfioFileRead(pdfio_file_t *pdf, void *buffer, size_t bytes) PDFIO_INTERNAL;
extern off_t _pdfioFileSeek(pdfio_file_t *pdf, off_t offset, int whence) PDFIO_INTERNAL; extern off_t _pdfioFileSeek(pdfio_file_t *pdf, off_t offset, int whence) PDFIO_INTERNAL;
@ -248,14 +256,18 @@ extern pdfio_stream_t *_pdfioStreamOpen(pdfio_obj_t *obj, bool decode) PDFIO_INT
extern bool _pdfioStringIsAllocated(pdfio_file_t *pdf, const char *s) PDFIO_INTERNAL; extern bool _pdfioStringIsAllocated(pdfio_file_t *pdf, const char *s) PDFIO_INTERNAL;
extern bool _pdfioTokenRead(pdfio_file_t *pdf, char *buffer, size_t bufsize, _pdfio_tpeek_cb_t peek_cb, _pdfio_tconsume_cb_t consume_cb, void *data); extern void _pdfioTokenClear(_pdfio_token_t *ts) PDFIO_INTERNAL;
extern bool _pdfioTokenGet(_pdfio_token_t *ts, char *buffer, size_t bufsize) PDFIO_INTERNAL;
extern void _pdfioTokenInit(_pdfio_token_t *ts, pdfio_file_t *pdf, _pdfio_tconsume_cb_t consume_cb, _pdfio_tpeek_cb_t peek_cb, void *cb_data);
extern void _pdfioTokenPush(_pdfio_token_t *ts, const char *token) PDFIO_INTERNAL;
extern bool _pdfioTokenRead(_pdfio_token_t *ts, char *buffer, size_t bufsize);
extern _pdfio_value_t *_pdfioValueCopy(pdfio_file_t *pdfdst, _pdfio_value_t *vdst, pdfio_file_t *pdfsrc, _pdfio_value_t *vsrc) PDFIO_INTERNAL; extern _pdfio_value_t *_pdfioValueCopy(pdfio_file_t *pdfdst, _pdfio_value_t *vdst, pdfio_file_t *pdfsrc, _pdfio_value_t *vsrc) PDFIO_INTERNAL;
# ifdef DEBUG # ifdef DEBUG
extern void _pdfioValueDebug(_pdfio_value_t *v) PDFIO_INTERNAL; extern void _pdfioValueDebug(_pdfio_value_t *v) PDFIO_INTERNAL;
# endif // DEBUG # endif // DEBUG
extern void _pdfioValueDelete(_pdfio_value_t *v) PDFIO_INTERNAL; extern void _pdfioValueDelete(_pdfio_value_t *v) PDFIO_INTERNAL;
extern _pdfio_value_t *_pdfioValueRead(pdfio_file_t *pdf, _pdfio_value_t *v) PDFIO_INTERNAL; extern _pdfio_value_t *_pdfioValueRead(pdfio_file_t *pdf, _pdfio_token_t *ts, _pdfio_value_t *v) PDFIO_INTERNAL;
extern bool _pdfioValueWrite(pdfio_file_t *pdf, _pdfio_value_t *v) PDFIO_INTERNAL; extern bool _pdfioValueWrite(pdfio_file_t *pdf, _pdfio_value_t *v) PDFIO_INTERNAL;
#endif // !PDFIO_PRIVATE_H #endif // !PDFIO_PRIVATE_H

7
pdfio-stream.c
View File

@ -121,12 +121,17 @@ pdfioStreamGetToken(
char *buffer, // I - String buffer char *buffer, // I - String buffer
size_t bufsize) // I - Size of string buffer size_t bufsize) // I - Size of string buffer
{ {
_pdfio_token_t tb; // Token buffer/stack
// Range check input... // Range check input...
if (!st || st->pdf->mode != _PDFIO_MODE_READ || !buffer || !bufsize) if (!st || st->pdf->mode != _PDFIO_MODE_READ || !buffer || !bufsize)
return (false); return (false);
// Read using the token engine... // Read using the token engine...
return (_pdfioTokenRead(st->pdf, buffer, bufsize, (_pdfio_tpeek_cb_t)pdfioStreamPeek, (_pdfio_tconsume_cb_t)pdfioStreamConsume, st)); _pdfioTokenInit(&tb, st->pdf, (_pdfio_tconsume_cb_t)pdfioStreamConsume, (_pdfio_tpeek_cb_t)pdfioStreamPeek, st);
return (_pdfioTokenRead(&tb, buffer, bufsize));
} }

220
pdfio-token.c
View File

@ -44,26 +44,99 @@
#define PDFIO_DELIM_CHARS "<>(){}[]/%" #define PDFIO_DELIM_CHARS "<>(){}[]/%"
//
// Types...
//
typedef struct _pdfio_tbuffer_s // Token reading buffer
{
unsigned char buffer[32], // Buffer
*bufptr, // Pointer into buffer
*bufend; // Last valid byte in buffer
_pdfio_tpeek_cb_t peek_cb; // Peek callback
_pdfio_tconsume_cb_t consume_cb; // Consume callback
void *data; // Callback data
} _pdfio_tbuffer_t;
// //
// Local functions... // Local functions...
// //
static int get_char(_pdfio_tbuffer_t *tb); static int get_char(_pdfio_token_t *tb);
//
// '_pdfioTokenClear()' - Clear the token stack.
//
void
_pdfioTokenClear(_pdfio_token_t *tb) // I - Token buffer/stack
{
PDFIO_DEBUG("_pdfioTokenClear(tb=%p)\n", tb);
while (tb->num_tokens > 0)
{
tb->num_tokens --;
free(tb->tokens[tb->num_tokens]);
tb->tokens[tb->num_tokens] = NULL;
}
}
//
// '_pdfioTokenGet()' - Get a token.
//
bool // O - `true` on success, `false` on failure
_pdfioTokenGet(_pdfio_token_t *tb, // I - Token buffer/stack
char *buffer, // I - String buffer
size_t bufsize) // I - Size of string buffer
{
// See if we have a token waiting on the stack...
if (tb->num_tokens > 0)
{
// Yes, return it...
tb->num_tokens --;
strncpy(buffer, tb->tokens[tb->num_tokens], bufsize - 1);
buffer[bufsize - 1] = '\0';
PDFIO_DEBUG("_pdfioTokenGet(tb=%p, buffer=%p, bufsize=%u): Popping '%s' from stack.\n", tb, buffer, (unsigned)bufsize, buffer);
free(tb->tokens[tb->num_tokens]);
tb->tokens[tb->num_tokens] = NULL;
return (true);
}
// No, read a new one...
return (_pdfioTokenRead(tb, buffer, bufsize));
}
//
// '_pdfioTokenInit()' - Initialize a token buffer/stack.
//
void
_pdfioTokenInit(
_pdfio_token_t *ts, // I - Token buffer/stack
pdfio_file_t *pdf, // I - PDF file
_pdfio_tconsume_cb_t consume_cb, // I - Consume callback
_pdfio_tpeek_cb_t peek_cb, // I - Peek callback
void *cb_data) // I - Callback data
{
// Zero everything out and then initialize key pointers...
memset(ts, 0, sizeof(_pdfio_token_t));
ts->pdf = pdf;
ts->consume_cb = consume_cb;
ts->peek_cb = peek_cb;
ts->cb_data = cb_data;
ts->bufptr = ts->buffer;
ts->bufend = ts->buffer;
}
//
// '_pdfioTokenPush()' - Push a token on the token stack.
//
void
_pdfioTokenPush(_pdfio_token_t *tb, // I - Token buffer/stack
const char *token) // I - Token to push
{
if (tb->num_tokens < (sizeof(tb->tokens) / sizeof(tb->tokens[0])))
{
if ((tb->tokens[tb->num_tokens ++] = strdup(token)) == NULL)
tb->num_tokens --;
}
}
// //
@ -71,19 +144,14 @@ static int get_char(_pdfio_tbuffer_t *tb);
// //
bool // O - `true` on success, `false` on failure bool // O - `true` on success, `false` on failure
_pdfioTokenRead( _pdfioTokenRead(_pdfio_token_t *tb, // I - Token buffer/stack
pdfio_file_t *pdf, // I - PDF file char *buffer, // I - String buffer
char *buffer, // I - String buffer size_t bufsize) // I - Size of string buffer
size_t bufsize, // I - Size of string buffer
_pdfio_tpeek_cb_t peek_cb, // I - "peek" callback
_pdfio_tconsume_cb_t consume_cb, // I - "consume" callback
void *data) // I - Callback data
{ {
_pdfio_tbuffer_t tb; // Token buffer int ch; // Character
int ch; // Character char *bufptr, // Pointer into buffer
char *bufptr, // Pointer into buffer *bufend, // End of buffer
*bufend, // End of buffer state = '\0'; // Current state
state = '\0'; // Current state
// //
@ -99,21 +167,16 @@ _pdfioTokenRead(
// - 'N' for number // - 'N' for number
// Read the next token, skipping any leading whitespace... // Read the next token, skipping any leading whitespace...
memset(&tb, 0, sizeof(tb));
tb.peek_cb = peek_cb;
tb.consume_cb = consume_cb;
tb.data = data;
bufptr = buffer; bufptr = buffer;
bufend = buffer + bufsize - 1; bufend = buffer + bufsize - 1;
// Skip leading whitespace... // Skip leading whitespace...
while ((ch = get_char(&tb)) != EOF) while ((ch = get_char(tb)) != EOF)
{ {
if (ch == '%') if (ch == '%')
{ {
// Skip comment // Skip comment
while ((ch = get_char(&tb)) != EOF) while ((ch = get_char(tb)) != EOF)
{ {
if (ch == '\n' || ch == '\r') if (ch == '\n' || ch == '\r')
break; break;
@ -147,14 +210,14 @@ _pdfioTokenRead(
switch (state) switch (state)
{ {
case '(' : // Literal string case '(' : // Literal string
while ((ch = get_char(&tb)) != EOF && ch != ')') while ((ch = get_char(tb)) != EOF && ch != ')')
{ {
if (ch == '\\') if (ch == '\\')
{ {
// Quoted character... // Quoted character...
int i; // Looping var int i; // Looping var
switch (ch = get_char(&tb)) switch (ch = get_char(tb))
{ {
case '0' : // Octal character escape case '0' : // Octal character escape
case '1' : case '1' :
@ -166,13 +229,13 @@ _pdfioTokenRead(
case '7' : case '7' :
for (ch -= '0', i = 0; i < 2; i ++) for (ch -= '0', i = 0; i < 2; i ++)
{ {
int tch = get_char(&tb); // Next char int tch = get_char(tb); // Next char
if (tch >= '0' && tch <= '7') if (tch >= '0' && tch <= '7')
ch = (char)((ch << 3) | (tch - '0')); ch = (char)((ch << 3) | (tch - '0'));
else else
{ {
tb.bufptr --; tb->bufptr --;
break; break;
} }
} }
@ -204,7 +267,7 @@ _pdfioTokenRead(
break; break;
default : default :
_pdfioFileError(pdf, "Unknown escape '\\%c' in literal string.", ch); _pdfioFileError(tb->pdf, "Unknown escape '\\%c' in literal string.", ch);
return (false); return (false);
} }
} }
@ -217,25 +280,25 @@ _pdfioTokenRead(
else else
{ {
// Out of space // Out of space
_pdfioFileError(pdf, "Token too large."); _pdfioFileError(tb->pdf, "Token too large.");
return (false); return (false);
} }
} }
if (ch != ')') if (ch != ')')
{ {
_pdfioFileError(pdf, "Unterminated string literal."); _pdfioFileError(tb->pdf, "Unterminated string literal.");
return (false); return (false);
} }
break; break;
case 'K' : // keyword case 'K' : // keyword
while ((ch = get_char(&tb)) != EOF && !isspace(ch)) while ((ch = get_char(tb)) != EOF && !isspace(ch))
{ {
if (strchr(PDFIO_DELIM_CHARS, ch) != NULL) if (strchr(PDFIO_DELIM_CHARS, ch) != NULL)
{ {
// End of keyword... // End of keyword...
tb.bufptr --; tb->bufptr --;
break; break;
} }
else if (bufptr < bufend) else if (bufptr < bufend)
@ -246,19 +309,19 @@ _pdfioTokenRead(
else else
{ {
// Out of space... // Out of space...
_pdfioFileError(pdf, "Token too large."); _pdfioFileError(tb->pdf, "Token too large.");
return (false); return (false);
} }
} }
break; break;
case 'N' : // number case 'N' : // number
while ((ch = get_char(&tb)) != EOF && !isspace(ch)) while ((ch = get_char(tb)) != EOF && !isspace(ch))
{ {
if (!isdigit(ch) && ch != '.') if (!isdigit(ch) && ch != '.')
{ {
// End of number... // End of number...
tb.bufptr --; tb->bufptr --;
break; break;
} }
else if (bufptr < bufend) else if (bufptr < bufend)
@ -269,19 +332,19 @@ _pdfioTokenRead(
else else
{ {
// Out of space... // Out of space...
_pdfioFileError(pdf, "Token too large."); _pdfioFileError(tb->pdf, "Token too large.");
return (false); return (false);
} }
} }
break; break;
case '/' : // "/name" case '/' : // "/name"
while ((ch = get_char(&tb)) != EOF && !isspace(ch)) while ((ch = get_char(tb)) != EOF && !isspace(ch))
{ {
if (strchr(PDFIO_DELIM_CHARS, ch) != NULL) if (strchr(PDFIO_DELIM_CHARS, ch) != NULL)
{ {
// End of keyword... // End of keyword...
tb.bufptr --; tb->bufptr --;
break; break;
} }
else if (ch == '#') else if (ch == '#')
@ -291,11 +354,11 @@ _pdfioTokenRead(
for (i = 0, ch = 0; i < 2; i ++) for (i = 0, ch = 0; i < 2; i ++)
{ {
int tch = get_char(&tb); int tch = get_char(tb);
if (!isxdigit(tch & 255)) if (!isxdigit(tch & 255))
{ {
_pdfioFileError(pdf, "Bad # escape in name."); _pdfioFileError(tb->pdf, "Bad # escape in name.");
return (false); return (false);
} }
else if (isdigit(tch)) else if (isdigit(tch))
@ -312,14 +375,14 @@ _pdfioTokenRead(
else else
{ {
// Out of space // Out of space
_pdfioFileError(pdf, "Token too large."); _pdfioFileError(tb->pdf, "Token too large.");
return (false); return (false);
} }
} }
break; break;
case '<' : // Potential hex string case '<' : // Potential hex string
if ((ch = get_char(&tb)) == '<') if ((ch = get_char(tb)) == '<')
{ {
// Dictionary delimiter // Dictionary delimiter
*bufptr++ = (char)ch; *bufptr++ = (char)ch;
@ -327,11 +390,11 @@ _pdfioTokenRead(
} }
else if (!isspace(ch & 255) && !isxdigit(ch & 255)) else if (!isspace(ch & 255) && !isxdigit(ch & 255))
{ {
_pdfioFileError(pdf, "Syntax error: '<%c'", ch); _pdfioFileError(tb->pdf, "Syntax error: '<%c'", ch);
return (false); return (false);
} }
while ((ch = get_char(&tb)) != EOF && ch != '>') while ((ch = get_char(tb)) != EOF && ch != '>')
{ {
if (isxdigit(ch)) if (isxdigit(ch))
{ {
@ -343,46 +406,65 @@ _pdfioTokenRead(
else else
{ {
// Too large // Too large
_pdfioFileError(pdf, "Token too large."); _pdfioFileError(tb->pdf, "Token too large.");
return (false); return (false);
} }
} }
else if (!isspace(ch)) else if (!isspace(ch))
{ {
_pdfioFileError(pdf, "Invalid hex string character '%c'.", ch); _pdfioFileError(tb->pdf, "Invalid hex string character '%c'.", ch);
return (false); return (false);
} }
} }
if (ch == EOF) if (ch == EOF)
{ {
_pdfioFileError(pdf, "Unterminated hex string."); _pdfioFileError(tb->pdf, "Unterminated hex string.");
return (false); return (false);
} }
break; break;
case '>' : // Dictionary case '>' : // Dictionary
if ((ch = get_char(&tb)) == '>') if ((ch = get_char(tb)) == '>')
{ {
*bufptr++ = '>'; *bufptr++ = '>';
} }
else else
{ {
_pdfioFileError(pdf, "Syntax error: '>%c'.", ch); _pdfioFileError(tb->pdf, "Syntax error: '>%c'.", ch);
return (false); return (false);
} }
break; break;
} }
while (tb.bufptr < tb.bufend && isspace(*(tb.bufptr))) while (tb->bufptr < tb->bufend && isspace(*(tb->bufptr)))
tb.bufptr ++; tb->bufptr ++;
if (tb.bufptr > tb.buffer) if (tb->bufptr > tb->buffer)
(consume_cb)(data, (size_t)(tb.bufptr - tb.buffer)); {
size_t remaining = (size_t)(tb->bufend - tb->bufptr);
// Remaining bytes in buffer
// Consume what we've used...
(tb->consume_cb)(tb->cb_data, (size_t)(tb->bufptr - tb->buffer));
if (remaining > 0)
{
// Shuffle remaining bytes for next call...
memmove(tb->buffer, tb->bufptr, remaining);
tb->bufptr = tb->buffer;
tb->bufend = tb->buffer + remaining;
}
else
{
// Nothing left, reset pointers...
tb->bufptr = tb->bufend = tb->buffer;
}
}
*bufptr = '\0'; *bufptr = '\0';
PDFIO_DEBUG("_pdfioTokenRead(pdf=%p, ...): Read '%s'.\n", pdf, buffer); PDFIO_DEBUG("_pdfioTokenRead: Read '%s'.\n", buffer);
return (bufptr > buffer); return (bufptr > buffer);
} }
@ -393,7 +475,7 @@ _pdfioTokenRead(
// //
static int // O - Character or `EOF` on end-of-file static int // O - Character or `EOF` on end-of-file
get_char(_pdfio_tbuffer_t *tb) // I - Token buffer get_char(_pdfio_token_t *tb) // I - Token buffer
{ {
ssize_t bytes; // Bytes peeked ssize_t bytes; // Bytes peeked
@ -403,10 +485,10 @@ get_char(_pdfio_tbuffer_t *tb) // I - Token buffer
{ {
// Consume previous bytes... // Consume previous bytes...
if (tb->bufend > tb->buffer) if (tb->bufend > tb->buffer)
(tb->consume_cb)(tb->data, (size_t)(tb->bufend - tb->buffer)); (tb->consume_cb)(tb->cb_data, (size_t)(tb->bufend - tb->buffer));
// Peek new bytes... // Peek new bytes...
if ((bytes = (tb->peek_cb)(tb->data, tb->buffer, sizeof(tb->buffer))) < 0) if ((bytes = (tb->peek_cb)(tb->cb_data, tb->buffer, sizeof(tb->buffer))) <= 0)
{ {
tb->bufptr = tb->bufend = tb->buffer; tb->bufptr = tb->bufend = tb->buffer;
return (EOF); return (EOF);

19
pdfio-value.c
View File

@ -157,6 +157,7 @@ _pdfioValueDelete(_pdfio_value_t *v) // I - Value
_pdfio_value_t * // O - Value or `NULL` on error/EOF _pdfio_value_t * // O - Value or `NULL` on error/EOF
_pdfioValueRead(pdfio_file_t *pdf, // I - PDF file _pdfioValueRead(pdfio_file_t *pdf, // I - PDF file
_pdfio_token_t *tb, // I - Token buffer/stack
_pdfio_value_t *v) // I - Value _pdfio_value_t *v) // I - Value
{ {
char token[8192]; // Token buffer char token[8192]; // Token buffer
@ -180,21 +181,21 @@ _pdfioValueRead(pdfio_file_t *pdf, // I - PDF file
PDFIO_DEBUG("_pdfioValueRead(pdf=%p, v=%p)\n", pdf, v); PDFIO_DEBUG("_pdfioValueRead(pdf=%p, v=%p)\n", pdf, v);
if (!_pdfioFileGetToken(pdf, token, sizeof(token))) if (!_pdfioTokenGet(tb, token, sizeof(token)))
return (NULL); return (NULL);
if (!strcmp(token, "[")) if (!strcmp(token, "["))
{ {
// Start of array // Start of array
v->type = PDFIO_VALTYPE_ARRAY; v->type = PDFIO_VALTYPE_ARRAY;
if ((v->value.array = _pdfioArrayRead(pdf)) == NULL) if ((v->value.array = _pdfioArrayRead(pdf, tb)) == NULL)
return (NULL); return (NULL);
} }
else if (!strcmp(token, "<<")) else if (!strcmp(token, "<<"))
{ {
// Start of dictionary // Start of dictionary
v->type = PDFIO_VALTYPE_DICT; v->type = PDFIO_VALTYPE_DICT;
if ((v->value.dict = _pdfioDictRead(pdf)) == NULL) if ((v->value.dict = _pdfioDictRead(pdf, tb)) == NULL)
return (NULL); return (NULL);
} }
else if (token[0] == '(') else if (token[0] == '(')
@ -259,7 +260,7 @@ _pdfioValueRead(pdfio_file_t *pdf, // I - PDF file
token3[8192], // Third token ("R") token3[8192], // Third token ("R")
*tokptr; // Pointer into token *tokptr; // Pointer into token
if (_pdfioFileGetToken(pdf, token2, sizeof(token2))) if (_pdfioTokenGet(tb, token2, sizeof(token2)))
{ {
// Got the second token, is it an integer? // Got the second token, is it an integer?
for (tokptr = token2; *tokptr; tokptr ++) for (tokptr = token2; *tokptr; tokptr ++)
@ -271,12 +272,12 @@ _pdfioValueRead(pdfio_file_t *pdf, // I - PDF file
if (*tokptr) if (*tokptr)
{ {
// Not an object reference, push this token for later use... // Not an object reference, push this token for later use...
_pdfioFilePushToken(pdf, token2); _pdfioTokenPush(tb, token2);
} }
else else
{ {
// A possible reference, get one more... // A possible reference, get one more...
if (_pdfioFileGetToken(pdf, token3, sizeof(token3))) if (_pdfioTokenGet(tb, token3, sizeof(token3)))
{ {
if (!strcmp(token3, "R")) if (!strcmp(token3, "R"))
{ {
@ -292,14 +293,14 @@ _pdfioValueRead(pdfio_file_t *pdf, // I - PDF file
else else
{ {
// Not a reference, push the tokens back... // Not a reference, push the tokens back...
_pdfioFilePushToken(pdf, token3); _pdfioTokenPush(tb, token3);
_pdfioFilePushToken(pdf, token2); _pdfioTokenPush(tb, token2);
} }
} }
else else
{ {
// Not a reference... // Not a reference...
_pdfioFilePushToken(pdf, token2); _pdfioTokenPush(tb, token2);
} }
} }
} }