Support pdf files larger than 2Gig.

If FZ_LARGEFILE is defined when building, MuPDF uses 64bit offsets
for files; this allows us to open streams larger than 2Gig.

The downsides to this are that:

 * The xref entries are larger.

 * All PDF ints are held as 64bit things rather than 32bit things
   (to cope with /Prev entries, hint stream offsets etc).

 * All file positions are stored as 64bits rather than 32.

The implementation works by detecting FZ_LARGEFILE. Some #ifdeffery
in fitz/system.h sets fz_off_t to either int or int64_t as appropriate,
and sets defines for fz_fopen, fz_fseek, fz_ftell etc as required.

These call the fseeko64 etc functions on linux (and so define
_LARGEFILE64_SOURCE) and the explicit 64bit functions on windows.

9 files changed, 63 insertions, 22 deletions

diff --git a/include/mupdf/fitz/filter.h b/include/mupdf/fitz/filter.h
index f200d8c1..07bea7b4 100644
--- a/include/mupdf/fitz/filter.h
+++ b/include/mupdf/fitz/filter.h
@@ -10,7 +10,7 @@
 typedef struct fz_jbig2_globals_s fz_jbig2_globals;
 
 fz_stream *fz_open_copy(fz_context *ctx, fz_stream *chain);
-fz_stream *fz_open_null(fz_context *ctx, fz_stream *chain, int len, int offset);
+fz_stream *fz_open_null(fz_context *ctx, fz_stream *chain, int len, fz_off_t offset);
 fz_stream *fz_open_concat(fz_context *ctx, int max, int pad);
 void fz_concat_push(fz_context *ctx, fz_stream *concat, fz_stream *chain); /* Ownership of chain is passed in */
 fz_stream *fz_open_arc4(fz_context *ctx, fz_stream *chain, unsigned char *key, unsigned keylen);
diff --git a/include/mupdf/fitz/math.h b/include/mupdf/fitz/math.h
index 54c8f0c8..9e413b7c 100644
--- a/include/mupdf/fitz/math.h
+++ b/include/mupdf/fitz/math.h
@@ -33,6 +33,8 @@ float fz_atof(const char *s);
 /* atoi that copes with NULL */
 int fz_atoi(const char *s);
 
+fz_off_t fz_atoo(const char *s);
+
 /*
 	Some standard math functions, done as static inlines for speed.
 	People with compilers that do not adequately implement inlines may
@@ -68,6 +70,11 @@ static inline int fz_maxi(int a, int b)
 	return (a > b ? a : b);
 }
 
+static inline fz_off_t fz_maxo(fz_off_t a, fz_off_t b)
+{
+	return (a > b ? a : b);
+}
+
 static inline float fz_clamp(float f, float min, float max)
 {
 	return (f > min ? (f < max ? f : max) : min);
diff --git a/include/mupdf/fitz/stream.h b/include/mupdf/fitz/stream.h
index 003e1bbe..b07e7ce3 100644
--- a/include/mupdf/fitz/stream.h
+++ b/include/mupdf/fitz/stream.h
@@ -104,7 +104,7 @@ void fz_drop_stream(fz_context *ctx, fz_stream *stm);
 /*
 	fz_tell: return the current reading position within a stream
 */
-int fz_tell(fz_context *ctx, fz_stream *stm);
+fz_off_t fz_tell(fz_context *ctx, fz_stream *stm);
 
 /*
 	fz_seek: Seek within a stream.
@@ -115,7 +115,7 @@ int fz_tell(fz_context *ctx, fz_stream *stm);
 
 	whence: From where the offset is measured (see fseek).
 */
-void fz_seek(fz_context *ctx, fz_stream *stm, int offset, int whence);
+void fz_seek(fz_context *ctx, fz_stream *stm, fz_off_t offset, int whence);
 
 /*
 	fz_read: Read from a stream into a given data block.
@@ -157,7 +157,7 @@ int fz_stream_meta(fz_context *ctx, fz_stream *stm, int key, int size, void *ptr
 
 typedef int (fz_stream_next_fn)(fz_context *ctx, fz_stream *stm, int max);
 typedef void (fz_stream_close_fn)(fz_context *ctx, void *state);
-typedef void (fz_stream_seek_fn)(fz_context *ctx, fz_stream *stm, int offset, int whence);
+typedef void (fz_stream_seek_fn)(fz_context *ctx, fz_stream *stm, fz_off_t offset, int whence);
 typedef int (fz_stream_meta_fn)(fz_context *ctx, fz_stream *stm, int key, int size, void *ptr);
 
 struct fz_stream_s
@@ -165,7 +165,7 @@ struct fz_stream_s
 	int refs;
 	int error;
 	int eof;
-	int pos;
+	fz_off_t pos;
 	int avail;
 	int bits;
 	unsigned char *rp, *wp;
diff --git a/include/mupdf/fitz/system.h b/include/mupdf/fitz/system.h
index 1ae66eea..0088bd64 100644
--- a/include/mupdf/fitz/system.h
+++ b/include/mupdf/fitz/system.h
@@ -1,6 +1,15 @@
 #ifndef MUPDF_FITZ_SYSTEM_H
 #define MUPDF_FITZ_SYSTEM_H
 
+/* The very first decision we need to make is, are we using the 64bit
+ * file pointers code. This must happen before the stdio.h include. */
+#ifdef FZ_LARGEFILE
+/* Set _LARGEFILE64_SOURCE so that we know fopen64 et al will be declared. */
+#ifndef _LARGEFILE64_SOURCE
+#define _LARGEFILE64_SOURCE
+#endif
+#endif
+
 /*
 	Include the standard libc headers.
 */
@@ -100,7 +109,7 @@ int gettimeofday(struct timeval *tv, struct timezone *tz);
 
 FILE *fz_fopen_utf8(const char *name, const char *mode);
 
-#define fopen fz_fopen_utf8
+#define fz_fopen fz_fopen_utf8
 
 char *fz_utf8_from_wchar(const wchar_t *s);
 wchar_t *fz_wchar_from_utf8(const char *s);
@@ -109,6 +118,10 @@ FILE *fz_fopen_utf8(const char *name, const char *mode);
 char **fz_argv_from_wargv(int argc, wchar_t **wargv);
 void fz_free_argv(int argc, char **argv);
 
+#define fseeko64 _fseeki64
+#define ftello64 _ftelli64
+#define atoll _atoi64
+
 #else /* Unix or close enough */
 
 #include <stdint.h>
@@ -122,6 +135,25 @@ void fz_free_argv(int argc, char **argv);
 
 #endif
 
+#ifdef FZ_LARGEFILE
+#ifndef fz_fopen
+#define fz_fopen fopen64
+#endif
+typedef int64_t fz_off_t;
+#define fz_fseek fseeko64
+#define fz_ftell ftello64
+#define fz_atoo_imp atoll
+#else
+#ifndef fz_fopen
+#define fz_fopen fopen
+#endif
+#define fz_fseek fseek
+#define fz_ftell ftell
+typedef int fz_off_t;
+#define fz_atoo_imp atoi
+#endif
+
+
 #ifdef __ANDROID__
 #include <android/log.h>
 #define LOG_TAG "libmupdf"
diff --git a/include/mupdf/pdf/document.h b/include/mupdf/pdf/document.h
index 41ee6b07..8f2c6ef6 100644
--- a/include/mupdf/pdf/document.h
+++ b/include/mupdf/pdf/document.h
@@ -24,7 +24,7 @@ struct pdf_lexbuf_s
 	int size;
 	int base_size;
 	int len;
-	int i;
+	fz_off_t i;
 	float f;
 	char *scratch;
 	char buffer[PDF_LEXBUF_SMALL];
@@ -170,8 +170,8 @@ struct pdf_document_s
 	fz_stream *file;
 
 	int version;
-	int startxref;
-	int file_size;
+	fz_off_t startxref;
+	fz_off_t file_size;
 	pdf_crypt *crypt;
 	pdf_ocg_descriptor *ocg;
 	pdf_hotspot hotspot;
@@ -190,14 +190,14 @@ struct pdf_document_s
 
 	/* State indicating which file parsing method we are using */
 	int file_reading_linearly;
-	int file_length;
+	fz_off_t file_length;
 
 	pdf_obj *linear_obj; /* Linearized object (if used) */
 	pdf_obj **linear_page_refs; /* Page objects for linear loading */
 	int linear_page1_obj_num;
 
 	/* The state for the pdf_progressive_advance parser */
-	int linear_pos;
+	fz_off_t linear_pos;
 	int linear_page_num;
 
 	int hint_object_offset;
@@ -221,17 +221,17 @@ struct pdf_document_s
 	struct
 	{
 		int number; /* Page object number */
-		int offset; /* Offset of page object */
-		int index; /* Index into shared hint_shared_ref */
+		fz_off_t offset; /* Offset of page object */
+		fz_off_t index; /* Index into shared hint_shared_ref */
 	} *hint_page;
 	int *hint_shared_ref;
 	struct
 	{
 		int number; /* Object number of first object */
-		int offset; /* Offset of first object */
+		fz_off_t offset; /* Offset of first object */
 	} *hint_shared;
 	int hint_obj_offsets_max;
-	int *hint_obj_offsets;
+	fz_off_t *hint_obj_offsets;
 
 	int resources_localised;
 
diff --git a/include/mupdf/pdf/object.h b/include/mupdf/pdf/object.h
index 2db0e57b..433419a3 100644
--- a/include/mupdf/pdf/object.h
+++ b/include/mupdf/pdf/object.h
@@ -14,6 +14,7 @@ typedef struct pdf_obj_s pdf_obj;
 pdf_obj *pdf_new_null(fz_context *ctx, pdf_document *doc);
 pdf_obj *pdf_new_bool(fz_context *ctx, pdf_document *doc, int b);
 pdf_obj *pdf_new_int(fz_context *ctx, pdf_document *doc, int i);
+pdf_obj *pdf_new_int_offset(fz_context *ctx, pdf_document *doc, fz_off_t off);
 pdf_obj *pdf_new_real(fz_context *ctx, pdf_document *doc, float f);
 pdf_obj *pdf_new_name(fz_context *ctx, pdf_document *doc, const char *str);
 pdf_obj *pdf_new_string(fz_context *ctx, pdf_document *doc, const char *str, int len);
@@ -74,6 +75,7 @@ void pdf_clean_obj(fz_context *ctx, pdf_obj *obj);
 /* safe, silent failure, no error reporting on type mismatches */
 int pdf_to_bool(fz_context *ctx, pdf_obj *obj);
 int pdf_to_int(fz_context *ctx, pdf_obj *obj);
+fz_off_t pdf_to_offset(fz_context *ctx, pdf_obj *obj);
 float pdf_to_real(fz_context *ctx, pdf_obj *obj);
 char *pdf_to_name(fz_context *ctx, pdf_obj *obj);
 char *pdf_to_str_buf(fz_context *ctx, pdf_obj *obj);
diff --git a/include/mupdf/pdf/parse.h b/include/mupdf/pdf/parse.h
index 23584029..ea715326 100644
--- a/include/mupdf/pdf/parse.h
+++ b/include/mupdf/pdf/parse.h
@@ -29,7 +29,7 @@ pdf_token pdf_lex_no_string(fz_context *ctx, fz_stream *f, pdf_lexbuf *lexbuf);
 pdf_obj *pdf_parse_array(fz_context *ctx, pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
 pdf_obj *pdf_parse_dict(fz_context *ctx, pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
 pdf_obj *pdf_parse_stm_obj(fz_context *ctx, pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
-pdf_obj *pdf_parse_ind_obj(fz_context *ctx, pdf_document *doc, fz_stream *f, pdf_lexbuf *buf, int *num, int *gen, int *stm_ofs, int *try_repair);
+pdf_obj *pdf_parse_ind_obj(fz_context *ctx, pdf_document *doc, fz_stream *f, pdf_lexbuf *buf, int *num, int *gen, fz_off_t *stm_ofs, int *try_repair);
 
 /*
 	pdf_print_token: print a lexed token to a buffer, growing if necessary
diff --git a/include/mupdf/pdf/xref.h b/include/mupdf/pdf/xref.h
index aa54605f..845c936e 100644
--- a/include/mupdf/pdf/xref.h
+++ b/include/mupdf/pdf/xref.h
@@ -36,8 +36,8 @@ struct pdf_xref_entry_s
 	char type;	/* 0=unset (f)ree i(n)use (o)bjstm */
 	unsigned char flags; /* bit 0 = marked */
 	unsigned short gen;	/* generation / objstm index */
-	int ofs;	/* file offset / objstm object number */
-	int stm_ofs;	/* on-disk stream */
+	fz_off_t ofs;	/* file offset / objstm object number */
+	fz_off_t stm_ofs;	/* on-disk stream */
 	fz_buffer *stm_buf; /* in-memory stream (for updated objects) */
 	pdf_obj *obj;	/* stored/cached object */
 };
@@ -53,7 +53,7 @@ struct pdf_xref_subsec_s
 {
 	pdf_xref_subsec *next;
 	int len;
-	int start;
+	fz_off_t start;
 	pdf_xref_entry *table;
 };
 
@@ -79,7 +79,7 @@ fz_stream *pdf_open_stream(fz_context *ctx, pdf_document *doc, int num, int gen)
 fz_stream *pdf_open_inline_stream(fz_context *ctx, pdf_document *doc, pdf_obj *stmobj, int length, fz_stream *chain, fz_compression_params *params);
 fz_compressed_buffer *pdf_load_compressed_stream(fz_context *ctx, pdf_document *doc, int num, int gen);
 void pdf_load_compressed_inline_image(fz_context *ctx, pdf_document *doc, pdf_obj *dict, int length, fz_stream *cstm, int indexed, fz_image *image);
-fz_stream *pdf_open_stream_with_offset(fz_context *ctx, pdf_document *doc, int num, int gen, pdf_obj *dict, int stm_ofs);
+fz_stream *pdf_open_stream_with_offset(fz_context *ctx, pdf_document *doc, int num, int gen, pdf_obj *dict, fz_off_t stm_ofs);
 fz_stream *pdf_open_compressed_stream(fz_context *ctx, fz_compressed_buffer *);
 fz_stream *pdf_open_contents_stream(fz_context *ctx, pdf_document *doc, pdf_obj *obj);
 fz_buffer *pdf_load_raw_renumbered_stream(fz_context *ctx, pdf_document *doc, int num, int gen, int orig_num, int orig_gen);
@@ -103,7 +103,7 @@ void pdf_mark_xref(fz_context *ctx, pdf_document *doc);
 void pdf_clear_xref(fz_context *ctx, pdf_document *doc);
 void pdf_clear_xref_to_mark(fz_context *ctx, pdf_document *doc);
 
-int pdf_repair_obj(fz_context *ctx, pdf_document *doc, pdf_lexbuf *buf, int *stmofsp, int *stmlenp, pdf_obj **encrypt, pdf_obj **id, pdf_obj **page, int *tmpofs);
+int pdf_repair_obj(fz_context *ctx, pdf_document *doc, pdf_lexbuf *buf, fz_off_t *stmofsp, int *stmlenp, pdf_obj **encrypt, pdf_obj **id, pdf_obj **page, fz_off_t *tmpofs);
 
 pdf_obj *pdf_progressive_advance(fz_context *ctx, pdf_document *doc, int pagenum);
 
diff --git a/include/mupdf/xps.h b/include/mupdf/xps.h
index 13c8eea0..fd21fe7c 100644
--- a/include/mupdf/xps.h
+++ b/include/mupdf/xps.h
@@ -216,7 +216,7 @@ typedef struct xps_entry_s xps_entry;
 struct xps_entry_s
 {
 	char *name;
-	int offset;
+	fz_off_t offset;
 	int csize;
 	int usize;
 };