Implement pdfioFileOpen reading of xref table.

This commit is contained in:
Michael R Sweet 2021-05-01 11:41:43 -04:00
parent ab740207a5
commit d13364a4fd
No known key found for this signature in database
GPG Key ID: 999559A027815955
5 changed files with 241 additions and 6 deletions

View File

@ -105,6 +105,60 @@ _pdfioFileGetChar(pdfio_file_t *pdf) // I - PDF file
} }
//
// '_pdfioFileGets()' - Read a line from a PDF file.
//
bool // O - `true` on success, `false` on error
_pdfioFileGets(pdfio_file_t *pdf, // I - PDF file
char *buffer, // I - Line buffer
size_t bufsize) // I - Size of line buffer
{
bool eol = false; // End of line?
char *bufptr = buffer, // Pointer into buffer
*bufend = buffer + bufsize - 1; // Pointer to end of buffer
while (!eol)
{
// If there are characters ready in the buffer, use them...
while (pdf->bufptr < pdf->bufend && bufptr < bufend)
{
char ch = *(pdf->bufptr++); // Next character in buffer
if (ch == '\n' || ch == '\r')
{
// CR, LF, or CR + LF end a line...
eol = true;
if (ch == '\r')
{
// Check for a LF after CR
if (pdf->bufptr >= pdf->bufend)
fill_buffer(pdf);
if (pdf->bufptr < pdf->bufend && *(pdf->bufptr) == '\n')
pdf->bufptr ++;
}
}
else
*bufptr++ = ch;
}
// Fill the read buffer as needed...
if (!eol)
{
if (!fill_buffer(pdf))
break;
}
}
*bufptr = '\0';
return (eol);
}
// //
// '_pdfioFilePrintf()' - Write a formatted string to a PDF file. // '_pdfioFilePrintf()' - Write a formatted string to a PDF file.
// //

View File

@ -21,7 +21,8 @@
// Local functions... // Local functions...
// //
static bool write_trailer(pdfio_file_t *pdf); static pdfio_obj_t *add_object(pdfio_file_t *pdf);
static bool write_trailer(pdfio_file_t *pdf);
// //
@ -162,6 +163,9 @@ pdfioFileCreateObject(
return (NULL); return (NULL);
} }
if (pdf->mode != _PDFIO_MODE_WRITE)
return (NULL);
if (dict->pdf != pdf) if (dict->pdf != pdf)
dict = pdfioDictCopy(pdf, dict); // Copy dictionary to new PDF dict = pdfioDictCopy(pdf, dict); // Copy dictionary to new PDF
@ -303,6 +307,9 @@ pdfioFileOpen(
void *error_data) // I - Error callback data, if any void *error_data) // I - Error callback data, if any
{ {
pdfio_file_t *pdf; // PDF file pdfio_file_t *pdf; // PDF file
char line[1024], // Line from file
*ptr; // Pointer into line
off_t xref_offset; // Offset to xref table
// Range check input... // Range check input...
@ -341,8 +348,180 @@ pdfioFileOpen(
return (NULL); return (NULL);
} }
// TODO: read header, trailer, and xref table... // Read the header from the first line...
if (!_pdfioFileGets(pdf, line, sizeof(line)))
goto error;
if ((strncmp(line, "%PDF-1.", 6) && strncmp(line, "%PDF-2.", 6)) || !isdigit(line[6] & 255))
{
// Bad header
_pdfioFileError(pdf, "Bad header '%s'.", line);
goto error;
}
// Copy the version number...
pdf->version = strdup(line + 4);
// Grab the last 32 characters of the file to find the start of the xref table...
_pdfioFileSeek(pdf, 32, SEEK_END);
if (_pdfioFileRead(pdf, line, 32) < 32)
goto error;
line[32] = '\0';
if ((ptr = strstr(line, "startxref")) == NULL)
{
_pdfioFileError(pdf, "Unable to find start of xref table.");
goto error;
}
xref_offset = (off_t)strtol(ptr + 9, NULL, 10);
if (_pdfioFileSeek(pdf, xref_offset, SEEK_SET) != xref_offset)
{
_pdfioFileError(pdf, "Unable to seek to start of xref table.");
goto error;
}
if (!_pdfioFileGets(pdf, line, sizeof(line)))
{
_pdfioFileError(pdf, "Unable to read start of xref table.");
goto error;
}
if (strcmp(line, "xref"))
{
_pdfioFileError(pdf, "Bad xref table header '%s'.", line);
goto error;
}
// Read the xref tables
while (_pdfioFileGets(pdf, line, sizeof(line)))
{
intmax_t number, // Object number
num_objects; // Number of objects
if (!strcmp(line, "trailer"))
break;
if (sscanf(line, "%jd%jd", &number, &num_objects) != 2)
{
_pdfioFileError(pdf, "Malformed xref table section '%s'.", line);
goto error;
}
// Read this group of objects...
for (; num_objects > 0; num_objects --, number ++)
{
intmax_t offset; // Offset in file
int generation; // Generation number
pdfio_obj_t *obj; // Object
// Read a line from the file and validate it...
if (_pdfioFileRead(pdf, line, 20) != 20)
goto error;
line[20] = '\0';
if (strcmp(line + 18, "\r\n") && strcmp(line + 18, " \n") && strcmp(line + 18, " \r"))
{
_pdfioFileError(pdf, "Malformed xref table entry '%s'.", line);
goto error;
}
line[18] = '\0';
// Parse the line
if ((offset = strtoimax(line, &ptr, 10)) < 0)
{
_pdfioFileError(pdf, "Malformed xref table entry '%s'.", line);
goto error;
}
if ((generation = (int)strtol(ptr, &ptr, 10)) < 0 || generation > 65535)
{
_pdfioFileError(pdf, "Malformed xref table entry '%s'.", line);
goto error;
}
if (*ptr != ' ')
{
_pdfioFileError(pdf, "Malformed xref table entry '%s'.", line);
goto error;
}
ptr ++;
if (*ptr != 'f' && *ptr != 'n')
{
_pdfioFileError(pdf, "Malformed xref table entry '%s'.", line);
goto error;
}
if (*ptr == 'f')
continue; // Don't care about free objects...
// Create a placeholder for the object in memory...
if ((obj = add_object(pdf)) == NULL)
goto error;
obj->number = (size_t)number;
obj->generation = (unsigned short)generation;
obj->offset = offset;
}
}
if (strcmp(line, "trailer"))
{
_pdfioFileError(pdf, "Missing trailer.");
goto error;
}
// TODO: Read trailer dict...
return (pdf); return (pdf);
// If we get here we had a fatal read error...
error:
pdfioFileClose(pdf);
return (NULL);
}
//
// 'add_object()' - Add an object to a PDF file.
//
static pdfio_obj_t * // O - New object
add_object(pdfio_file_t *pdf) // I - PDF file
{
pdfio_obj_t *obj; // New object
// Allocate memory for the object...
if ((obj = (pdfio_obj_t *)calloc(1, sizeof(pdfio_obj_t))) == NULL)
{
_pdfioFileError(pdf, "Unable to allocate memory for object - %s", strerror(errno));
return (NULL);
}
// Expand the objects array as needed
if (pdf->num_objs >= pdf->alloc_objs)
{
pdfio_obj_t **temp = (pdfio_obj_t **)realloc(pdf->objs, (pdf->alloc_objs + 32) * sizeof(pdfio_obj_t *));
if (!temp)
{
_pdfioFileError(pdf, "Unable to allocate memory for object - %s", strerror(errno));
free(obj);
return (NULL);
}
pdf->objs = temp;
pdf->alloc_objs += 32;
}
pdf->objs[pdf->num_objs ++] = obj;
return (obj);
} }

View File

@ -78,7 +78,7 @@ pdfioObjGetDict(pdfio_obj_t *obj) // I - Object
// 'pdfioObjGetGeneration()' - Get the object's generation number. // 'pdfioObjGetGeneration()' - Get the object's generation number.
// //
size_t // O - Generation number (0 to 65535) unsigned short // O - Generation number (0 to 65535)
pdfioObjGetGeneration(pdfio_obj_t *obj) // I - Object pdfioObjGetGeneration(pdfio_obj_t *obj) // I - Object
{ {
return (obj ? obj->generation : 0); return (obj ? obj->generation : 0);

View File

@ -17,6 +17,7 @@
# include "pdfio.h" # include "pdfio.h"
# include <stdarg.h> # include <stdarg.h>
# include <errno.h> # include <errno.h>
# include <inttypes.h>
# include <fcntl.h> # include <fcntl.h>
# include <unistd.h> # include <unistd.h>
# include <string.h> # include <string.h>
@ -141,8 +142,8 @@ struct _pdfio_file_s // PDF file structure
struct _pdfio_obj_s // Object struct _pdfio_obj_s // Object
{ {
pdfio_file_t *pdf; // PDF file pdfio_file_t *pdf; // PDF file
size_t number, // Number size_t number; // Number
generation; // Generation unsigned short generation; // Generation
off_t offset, // Offset to object in file off_t offset, // Offset to object in file
length_offset, // Offset to /Length in object dict length_offset, // Offset to /Length in object dict
stream_offset; // Offset to start of stream in file stream_offset; // Offset to start of stream in file
@ -179,6 +180,7 @@ extern bool _pdfioFileDefaultError(pdfio_file_t *pdf, const char *message, void
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 _pdfioFileGets(pdfio_file_t *pdf, char *buffer, size_t bufsize) 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 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, char *buffer, size_t bytes) PDFIO_INTERNAL; extern ssize_t _pdfioFileRead(pdfio_file_t *pdf, char *buffer, size_t bytes) PDFIO_INTERNAL;

View File

@ -152,7 +152,7 @@ extern pdfio_file_t *pdfioFileOpen(const char *filename, pdfio_error_cb_t error_
extern bool pdfioObjClose(pdfio_obj_t *obj) PDFIO_PUBLIC; extern bool pdfioObjClose(pdfio_obj_t *obj) PDFIO_PUBLIC;
extern pdfio_stream_t *pdfioObjCreateStream(pdfio_obj_t *obj, pdfio_filter_t compression) PDFIO_PUBLIC; extern pdfio_stream_t *pdfioObjCreateStream(pdfio_obj_t *obj, pdfio_filter_t compression) PDFIO_PUBLIC;
extern pdfio_dict_t *pdfioObjGetDict(pdfio_obj_t *obj) PDFIO_PUBLIC; extern pdfio_dict_t *pdfioObjGetDict(pdfio_obj_t *obj) PDFIO_PUBLIC;
extern size_t pdfioObjGetGeneration(pdfio_obj_t *obj) PDFIO_PUBLIC; extern unsigned short pdfioObjGetGeneration(pdfio_obj_t *obj) PDFIO_PUBLIC;
extern size_t pdfioObjGetNumber(pdfio_obj_t *obj) PDFIO_PUBLIC; extern size_t pdfioObjGetNumber(pdfio_obj_t *obj) PDFIO_PUBLIC;
extern const char *pdfioObjGetType(pdfio_obj_t *obj) PDFIO_PUBLIC; extern const char *pdfioObjGetType(pdfio_obj_t *obj) PDFIO_PUBLIC;
extern pdfio_stream_t *pdfioObjOpenStream(pdfio_obj_t *obj) PDFIO_PUBLIC; extern pdfio_stream_t *pdfioObjOpenStream(pdfio_obj_t *obj) PDFIO_PUBLIC;