@@ -0,0 +1,258 @@ |
| | 1 | +module ferp_mmap |
| | 2 | + !> Memory-mapped file I/O for FERP |
| | 3 | + !> Uses POSIX mmap for efficient file reading |
| | 4 | + use ferp_kinds |
| | 5 | + use, intrinsic :: iso_c_binding |
| | 6 | + implicit none |
| | 7 | + private |
| | 8 | + |
| | 9 | + public :: mmap_file_t |
| | 10 | + public :: mmap_open, mmap_close, mmap_get_line |
| | 11 | + |
| | 12 | + ! POSIX constants |
| | 13 | + integer(c_int), parameter :: PROT_READ = 1 |
| | 14 | + integer(c_int), parameter :: MAP_PRIVATE = 2 |
| | 15 | + integer(c_int), parameter :: MAP_FAILED = -1 |
| | 16 | + |
| | 17 | + ! C interfaces |
| | 18 | + interface |
| | 19 | + function c_open(pathname, flags) bind(C, name="open") |
| | 20 | + import :: c_char, c_int |
| | 21 | + character(kind=c_char), intent(in) :: pathname(*) |
| | 22 | + integer(c_int), value :: flags |
| | 23 | + integer(c_int) :: c_open |
| | 24 | + end function c_open |
| | 25 | + |
| | 26 | + function c_close(fd) bind(C, name="close") |
| | 27 | + import :: c_int |
| | 28 | + integer(c_int), value :: fd |
| | 29 | + integer(c_int) :: c_close |
| | 30 | + end function c_close |
| | 31 | + |
| | 32 | + function c_mmap(addr, length, prot, flags, fd, offset) bind(C, name="mmap") |
| | 33 | + import :: c_ptr, c_size_t, c_int, c_long |
| | 34 | + type(c_ptr), value :: addr |
| | 35 | + integer(c_size_t), value :: length |
| | 36 | + integer(c_int), value :: prot |
| | 37 | + integer(c_int), value :: flags |
| | 38 | + integer(c_int), value :: fd |
| | 39 | + integer(c_long), value :: offset |
| | 40 | + type(c_ptr) :: c_mmap |
| | 41 | + end function c_mmap |
| | 42 | + |
| | 43 | + function c_munmap(addr, length) bind(C, name="munmap") |
| | 44 | + import :: c_ptr, c_size_t, c_int |
| | 45 | + type(c_ptr), value :: addr |
| | 46 | + integer(c_size_t), value :: length |
| | 47 | + integer(c_int) :: c_munmap |
| | 48 | + end function c_munmap |
| | 49 | + |
| | 50 | + function c_fstat(fd, statbuf) bind(C, name="fstat") |
| | 51 | + import :: c_int, c_ptr |
| | 52 | + integer(c_int), value :: fd |
| | 53 | + type(c_ptr), value :: statbuf |
| | 54 | + integer(c_int) :: c_fstat |
| | 55 | + end function c_fstat |
| | 56 | + |
| | 57 | + function c_lseek(fd, offset, whence) bind(C, name="lseek") |
| | 58 | + import :: c_int, c_long |
| | 59 | + integer(c_int), value :: fd |
| | 60 | + integer(c_long), value :: offset |
| | 61 | + integer(c_int), value :: whence |
| | 62 | + integer(c_long) :: c_lseek |
| | 63 | + end function c_lseek |
| | 64 | + end interface |
| | 65 | + |
| | 66 | + ! lseek whence values |
| | 67 | + integer(c_int), parameter :: SEEK_SET = 0 |
| | 68 | + integer(c_int), parameter :: SEEK_END = 2 |
| | 69 | + |
| | 70 | + !> Memory-mapped file type |
| | 71 | + type :: mmap_file_t |
| | 72 | + type(c_ptr) :: data = c_null_ptr |
| | 73 | + integer(c_size_t) :: size = 0 |
| | 74 | + integer(c_size_t) :: pos = 0 ! Current position in file |
| | 75 | + integer :: line_number = 0 |
| | 76 | + integer(i64) :: byte_offset = 0 |
| | 77 | + logical :: is_open = .false. |
| | 78 | + character(len=max_path_len) :: filename = '' |
| | 79 | + contains |
| | 80 | + procedure :: open => mmap_open_method |
| | 81 | + procedure :: close => mmap_close_method |
| | 82 | + procedure :: read_line => mmap_read_line |
| | 83 | + procedure :: reset => mmap_reset |
| | 84 | + end type mmap_file_t |
| | 85 | + |
| | 86 | +contains |
| | 87 | + |
| | 88 | + function mmap_open(filename, mfile) result(success) |
| | 89 | + !> Open a file with memory mapping |
| | 90 | + character(len=*), intent(in) :: filename |
| | 91 | + type(mmap_file_t), intent(out) :: mfile |
| | 92 | + logical :: success |
| | 93 | + |
| | 94 | + integer(c_int) :: fd, istat |
| | 95 | + integer(c_int), parameter :: O_RDONLY = 0 |
| | 96 | + integer(c_size_t) :: file_size |
| | 97 | + integer(c_long) :: size_long |
| | 98 | + |
| | 99 | + success = .false. |
| | 100 | + mfile%is_open = .false. |
| | 101 | + mfile%filename = filename |
| | 102 | + |
| | 103 | + ! Open file |
| | 104 | + fd = c_open(trim(filename) // c_null_char, O_RDONLY) |
| | 105 | + if (fd < 0) return |
| | 106 | + |
| | 107 | + ! Get file size via lseek to end |
| | 108 | + size_long = c_lseek(fd, 0_c_long, SEEK_END) |
| | 109 | + if (size_long < 0) then |
| | 110 | + istat = c_close(fd) |
| | 111 | + return |
| | 112 | + end if |
| | 113 | + file_size = int(size_long, c_size_t) |
| | 114 | + |
| | 115 | + ! Seek back to beginning |
| | 116 | + size_long = c_lseek(fd, 0_c_long, SEEK_SET) |
| | 117 | + |
| | 118 | + if (file_size == 0) then |
| | 119 | + istat = c_close(fd) |
| | 120 | + mfile%size = 0 |
| | 121 | + mfile%is_open = .true. |
| | 122 | + success = .true. |
| | 123 | + return |
| | 124 | + end if |
| | 125 | + |
| | 126 | + ! Memory map the file |
| | 127 | + mfile%data = c_mmap(c_null_ptr, file_size, PROT_READ, MAP_PRIVATE, fd, 0_c_long) |
| | 128 | + istat = c_close(fd) ! Can close fd after mmap |
| | 129 | + |
| | 130 | + if (.not. c_associated(mfile%data)) return |
| | 131 | + |
| | 132 | + mfile%size = file_size |
| | 133 | + mfile%pos = 0 |
| | 134 | + mfile%line_number = 0 |
| | 135 | + mfile%byte_offset = 0 |
| | 136 | + mfile%is_open = .true. |
| | 137 | + success = .true. |
| | 138 | + |
| | 139 | + end function mmap_open |
| | 140 | + |
| | 141 | + function mmap_open_method(this, filename) result(success) |
| | 142 | + class(mmap_file_t), intent(inout) :: this |
| | 143 | + character(len=*), intent(in) :: filename |
| | 144 | + logical :: success |
| | 145 | + success = mmap_open(filename, this) |
| | 146 | + end function mmap_open_method |
| | 147 | + |
| | 148 | + subroutine mmap_close(mfile) |
| | 149 | + !> Close memory-mapped file |
| | 150 | + type(mmap_file_t), intent(inout) :: mfile |
| | 151 | + |
| | 152 | + integer(c_int) :: istat |
| | 153 | + |
| | 154 | + if (mfile%is_open .and. c_associated(mfile%data)) then |
| | 155 | + istat = c_munmap(mfile%data, mfile%size) |
| | 156 | + end if |
| | 157 | + |
| | 158 | + mfile%data = c_null_ptr |
| | 159 | + mfile%size = 0 |
| | 160 | + mfile%pos = 0 |
| | 161 | + mfile%is_open = .false. |
| | 162 | + |
| | 163 | + end subroutine mmap_close |
| | 164 | + |
| | 165 | + subroutine mmap_close_method(this) |
| | 166 | + class(mmap_file_t), intent(inout) :: this |
| | 167 | + call mmap_close(this) |
| | 168 | + end subroutine mmap_close_method |
| | 169 | + |
| | 170 | + subroutine mmap_reset(this) |
| | 171 | + !> Reset to beginning of file |
| | 172 | + class(mmap_file_t), intent(inout) :: this |
| | 173 | + this%pos = 0 |
| | 174 | + this%line_number = 0 |
| | 175 | + this%byte_offset = 0 |
| | 176 | + end subroutine mmap_reset |
| | 177 | + |
| | 178 | + function mmap_get_line(mfile, line, line_num, byte_off) result(success) |
| | 179 | + !> Get next line from memory-mapped file |
| | 180 | + type(mmap_file_t), intent(inout) :: mfile |
| | 181 | + character(len=:), allocatable, intent(out) :: line |
| | 182 | + integer, intent(out) :: line_num |
| | 183 | + integer(i64), intent(out) :: byte_off |
| | 184 | + logical :: success |
| | 185 | + |
| | 186 | + success = mfile%read_line(line, line_num, byte_off) |
| | 187 | + end function mmap_get_line |
| | 188 | + |
| | 189 | + function mmap_read_line(this, line, line_num, byte_off) result(success) |
| | 190 | + !> Read next line from memory-mapped file |
| | 191 | + class(mmap_file_t), intent(inout) :: this |
| | 192 | + character(len=:), allocatable, intent(out) :: line |
| | 193 | + integer, intent(out) :: line_num |
| | 194 | + integer(i64), intent(out) :: byte_off |
| | 195 | + logical :: success |
| | 196 | + |
| | 197 | + character(len=1, kind=c_char), pointer :: file_data(:) |
| | 198 | + integer(c_size_t) :: start_pos, end_pos, line_len |
| | 199 | + integer :: i |
| | 200 | + |
| | 201 | + success = .false. |
| | 202 | + line_num = 0 |
| | 203 | + byte_off = 0 |
| | 204 | + if (allocated(line)) deallocate(line) |
| | 205 | + |
| | 206 | + if (.not. this%is_open) return |
| | 207 | + if (this%pos >= this%size) return |
| | 208 | + |
| | 209 | + ! Map the C pointer to a Fortran character array |
| | 210 | + call c_f_pointer(this%data, file_data, [this%size]) |
| | 211 | + |
| | 212 | + ! Find start and end of line |
| | 213 | + start_pos = this%pos + 1 ! 1-based for Fortran |
| | 214 | + end_pos = start_pos |
| | 215 | + |
| | 216 | + ! Scan for newline |
| | 217 | + do while (end_pos <= this%size) |
| | 218 | + if (file_data(end_pos) == char(10)) exit ! LF |
| | 219 | + end_pos = end_pos + 1 |
| | 220 | + end do |
| | 221 | + |
| | 222 | + ! Calculate line length (excluding newline) |
| | 223 | + line_len = end_pos - start_pos |
| | 224 | + if (line_len > 0 .and. end_pos > start_pos) then |
| | 225 | + ! Check for CR before LF (Windows line ending) |
| | 226 | + if (file_data(end_pos - 1) == char(13)) then |
| | 227 | + line_len = line_len - 1 |
| | 228 | + end if |
| | 229 | + end if |
| | 230 | + |
| | 231 | + ! Allocate and copy line |
| | 232 | + if (line_len > 0) then |
| | 233 | + allocate(character(len=line_len) :: line) |
| | 234 | + do i = 1, int(line_len) |
| | 235 | + line(i:i) = file_data(start_pos + i - 1) |
| | 236 | + end do |
| | 237 | + else |
| | 238 | + line = '' |
| | 239 | + end if |
| | 240 | + |
| | 241 | + ! Update state |
| | 242 | + this%line_number = this%line_number + 1 |
| | 243 | + line_num = this%line_number |
| | 244 | + byte_off = int(this%pos, i64) |
| | 245 | + |
| | 246 | + ! Move past the newline |
| | 247 | + if (end_pos <= this%size) then |
| | 248 | + this%pos = end_pos ! Position after newline (0-based) |
| | 249 | + else |
| | 250 | + this%pos = this%size |
| | 251 | + end if |
| | 252 | + this%byte_offset = int(this%pos, i64) |
| | 253 | + |
| | 254 | + success = .true. |
| | 255 | + |
| | 256 | + end function mmap_read_line |
| | 257 | + |
| | 258 | +end module ferp_mmap |
