quantum-espresso/upflib/dom.f90

989 lines
30 KiB
Fortran

!
! Copyright (C) 2021-2022 Quantum ESPRESSO Foundation
! This file is distributed under the terms of the
! GNU General Public License. See the file `License'
! in the root directory of the present distribution,
! or http://www.gnu.org/copyleft/gpl.txt .
!
#undef __debug
!! define __debug to print information on opened and closed tags
module dom
!!
!! Poor-man FoX_dom replacement - Paolo Giannozzi, 2022-2024
!!
implicit none
!
integer, parameter :: dp = selected_real_kind(14,200)
! Type 'nodelist' is a "linked list" of pointers to type 'node'
type :: nodelist
type(node), pointer :: node
type(nodelist), pointer :: nextlist
end type nodelist
! Type 'node' contains: tag name, attributes, data, pointer to parent tag,
! a linked list of pointers to tags contained in this node
type :: node
character(:), allocatable :: tag
character(:), allocatable :: attr
character(:), allocatable :: data
type (node), pointer :: prev => null()
type (nodelist), pointer :: linklist => null()
end type node
! Used for check: parsing and cleaning should end at level -1
integer :: nlevel = -1
! The DOM is stored here
type(node), target, save :: root
! Used to simulate FoX behavior
type :: domexception
integer :: code
end type domexception
! The following machinery is used only to ensure that linked lists
! produced by "getelementsbytagname" are properly deallocated when
! "destroy" is called. Not sure this is the smartest way to do that.
! "meta-list" ml is a linked list to linked lists (!)
type :: metalist
type(nodelist), pointer :: linklist => null()
type(metalist), pointer :: prevmeta => null()
end type metalist
type(metalist), pointer :: ml => null()
!
private
! Callable routines or interfaces
public :: node, nodelist
public :: parsefile, item, getelementsbytagname, getlength, destroy
public :: gettagname, hasattribute, extractdataattribute, extractdatacontent
public :: getfirstchild, domexception, getexceptioncode
public :: parsestring
!
interface extractdataattribute
module procedure extractdataattribute_c, &
extractdataattribute_i, &
extractdataattribute_iv,&
extractdataattribute_l, &
extractdataattribute_r
end interface extractdataattribute
!
interface extractdatacontent
module procedure extractdatacontent_c, &
extractdatacontent_cv,&
extractdatacontent_l, &
extractdatacontent_i, &
extractdatacontent_iv,&
extractdatacontent_r, &
extractdatacontent_rv,&
extractdatacontent_rm
end interface extractdatacontent
!
CONTAINS
!
function parsestring( string, ex )
!
character(len=*), intent(in) :: string
type(domexception), intent (out), optional :: ex
type(node), pointer :: parsestring
!
parsestring => parse( strbuf=string, ex=ex )
!
end function parsestring
!
function parsefile ( filename, ex )
!
character(len=*), intent (in) :: filename
type(domexception), intent (out), optional :: ex
type(node), pointer :: parsefile
integer :: iunit, ierr
!
open(newunit=iunit, file=filename, form='formatted', status='old', &
iostat=ierr)
!
if ( ierr /= 0 ) then
if ( present(ex) ) then
ex%code = ierr
else
print *,'error opening file: ierr=',ierr
stop
end if
else
!
parsefile => parse( iun=iunit, ex=ex )
!
end if
!
end function parsefile
!
function parse ( iun, strbuf, ex )
! This is where the action is: parse either from unit "iun"
! or from a buffer "strbuf", returns a pointer to the node "root"
! and optionally an error code in "ex"
character(len=*), intent (in), optional :: strbuf
integer, intent(in), optional :: iun
type(domexception), intent (inout), optional :: ex
type(node), pointer :: parse
!
integer, parameter :: maxline=1024, maxdim=maxline+16
character(len=maxdim) :: line
!
integer :: ierr, i0, i1
logical :: firstline
!
nlevel = -1
ierr = 0
i0 = 1
firstline = .true.
!
readline: do
!
if ( present(iun) .and. .not. present(strbuf) ) then
! read from file
read(iun,'(a)',end=10) line
else if ( present(strbuf) .and. .not. present(iun) ) then
! read from buffer
if ( i0 > len(strbuf) ) go to 10
! locate newline (ascii n.10)
i1= index( strbuf(i0:), char(10))
if ( i1 > 1 ) then
! skip LF and go to next line
line = strbuf(i0:i0+i1-2)
i0=i0+i1
else if ( i1 == 1 ) then
! empty line
line = strbuf(i0:i0)
i0=i0+i1
else
! i1=0: last line (if no LF at the end)
line = strbuf(i0:)
i0 = len(strbuf)+1
end if
else
if ( .not.present(ex) ) then
print *, 'error: both unit and string, or none, in input'
else
ierr = 1001
end if
exit readline
end if
if ( len_trim(line) > maxline ) then
if ( .not.present(ex) ) &
print *, 'error: line exceeds ', maxline, ' characters'
ierr = 1
exit readline
end if
! print *,'debug:',trim(line)
!
ierr = parseline ( firstline, line, ex )
if ( ierr /= 0 ) exit readline
!
end do readline
!
10 continue
if (present(iun) ) close(iun)
if ( present(ex) ) ex%code = ierr
!
if ( ierr == 0 .and. nlevel /= -1) then
if ( present(ex) ) then
ex%code = nlevel
else
print *, 'error: parsing ended with ',nlevel+1,' level(s) open'
end if
else if ( ierr > 0 .and. .not. present(ex) ) then
print *,'error in parsing: ierr=',ierr
stop
end if
parse => root
!
end function parse
!
function parseline ( firstline, line, ex )
! result is returned in pointer "root"
! input line to be parsed
logical, intent(inout) :: firstline
character(len=*), intent(in) :: line
! optional exception code - if present, error messages are not printed
type(domexception), intent (in), optional :: ex
! error code: 0 = success, otherwise parsing failed
integer :: parseline
!
integer, parameter :: maxlength=80
! for checks, not really needed:
! integer, parameter :: maxlevel=9
! character(len=maxlength), dimension(0:maxlevel) :: open_tags
!
! variables keeping track of parsing status, must be conserved between calls
! (tag is actually a local variable, except maybe when used for debugging)
logical, save :: in_comment
logical, save :: in_attribute
logical, save :: in_data
character(len=maxlength), save :: tag
! pointer to current node
type(node), pointer, save :: curr
! next node: the "root" pointer points to the memory allocated in "next"
type(node), pointer :: next
! local variables
type(node), pointer :: prev
integer :: n, nl, n1, n2, n3, m
logical :: is_found
!
! Initialization
if ( firstline ) then
firstline = .false.
curr => null()
in_comment = .false.
in_attribute = .false.
in_data = .false.
tag = ' '
end if
n = 1
nl= len_trim(line)
!
parseline = 0
scanline: do while ( n < nl+1 )
!print *, 'debug: n=',n, line(n:n)
if ( in_comment ) then
! trick to avoid trespassing the EOL
n1 = min (n+1,nl)
n2 = min (n+2,nl)
if ( line(n:n2) == ']]>' ) then
in_comment = .false.
n = n+3
#if defined ( __debug )
print *, 'debug: cdata ends'
#endif
else if ( line(n:n2) == '-->' ) then
in_comment = .false.
n = n+3
#if defined ( __debug )
print *, 'debug: comment ends'
#endif
else if ( line(n:n1) == '?>' ) then
in_comment = .false.
n = n+2
#if defined ( __debug )
print *, 'debug: process ends'
#endif
else
n = n+1
end if
else
if ( line(n:n) == '<' ) then
! trick to avoid trespassing the EOL
n1 = min (n+1,nl)
n2 = min (n+3,nl)
n3 = min (n+8,nl)
if ( line(n1:n2) == '!--' ) then
n = n+4
in_comment = .true.
#if defined ( __debug )
print *, 'debug: comment begins'
#endif
else if ( line(n1:n3) == '![CDATA[' ) then
n = n+9
in_comment = .true.
#if defined ( __debug )
print *, 'debug: cdata begins'
#endif
else if ( line(n1:n1) == '?' ) then
n = n+2
in_comment = .true.
#if defined ( __debug )
print *, 'debug: process begins'
#endif
else if ( line(n1:n1) == '/' ) then
! tag = trim( open_tags(nlevel) )
tag = curr%tag
n = n+2
m = min(n+len_trim(tag)+1,nl)
if ( line(n:m) == trim(tag)//'>' ) then
#if defined ( __debug )
print *, 'debug: closing tag </',trim(tag),'> found'
#endif
prev => curr%prev
curr => prev
in_data = .false.
else
if ( .not.present(ex) ) then
print *, n,m,nlevel,tag
print *, 'error: unexpected closing tag </',line(n:nl),'> found'
end if
parseline = 2
return
end if
nlevel = nlevel - 1
n = m+1
else
scantag: do m = n+1, nl
is_found = .false.
if ( line(m:m) == '>' ) then
if ( m == n+1 ) then
if ( .not.present(ex) ) &
print *, 'error: empty tag <>'
parseline = 3
return
end if
is_found = .true.
in_data = .true.
in_attribute = .false.
else if ( line(m:m) == ' ' .or. line(m:m) == '/' &
.or. m == nl ) then
! case '/' may occur for empty tags like "<tag/>"
if ( m == n+1 ) then
if ( .not.present(ex) ) &
print *, 'error: space after <'
parseline = 4
return
end if
is_found = .true.
in_data = .false.
in_attribute = .true.
end if
if ( is_found ) then
tag = line(n+1:m-1)
#if defined ( __debug )
if ( in_attribute ) then
print *, 'debug: tag with attributes ',trim(tag),'...'
else
print *, 'debug: tag <',trim(tag),'> found'
endif
#endif
nlevel = nlevel + 1
! open_tags(nlevel) = trim(tag)
allocate(next)
next%tag = trim(tag)
if ( in_attribute ) next%attr=' '
if ( in_data ) next%data=' '
if ( associated(curr) ) then
next%prev => curr
call add_to_list(curr%linklist,next)
curr => next
else
if ( allocated(root%tag) ) then
if ( .not.present(ex) ) &
print *, 'error: more than one root tag'
parseline = 5
return
end if
curr => root
root = next
deallocate(next)
end if
!
n = m+1
exit scantag
end if
end do scantag
if ( m > nl) then
tag = ' '
n = nl+1
exit scanline
end if
end if
else if ( line(n:n) == '>' ) then
if ( in_attribute ) then
if ( line(n-1:n-1) == '/' ) then
#if defined ( __debug )
print *, 'info short tag ',trim(tag),' found'
#endif
! remove slash from attribute
curr%attr(len(curr%attr):len(curr%attr)) = ' '
prev => curr%prev
curr => prev
nlevel = nlevel - 1
else
#if defined ( __debug )
print *, 'debug: tag with attributes ',trim(tag),' found'
#endif
in_data = .true.
end if
in_attribute = .false.
else
if ( .not.present(ex) ) &
print *, 'error: closed tag that was not open'
parseline = 6
return
end if
n = n+1
else
if ( in_attribute ) then
if ( .not. allocated(curr%attr) ) curr%attr = ' '
curr%attr = curr%attr // line(n:n)
end if
if ( in_data ) then
if ( .not. allocated(curr%data) ) curr%data = ' '
curr%data = curr%data // line(n:n)
end if
n = n+1
end if
end if
end do scanline
! if data extends over more than one line, add space between lines
if ( in_data .and. associated(curr) ) then
if ( allocated(curr%data) ) curr%data = curr%data // ' '
end if
end function parseline
integer function getexceptioncode(ex)
type(domexception), intent(in):: ex
getexceptioncode = ex%code
end function getexceptioncode
!
subroutine add_to_list(linklist, next)
type(node), pointer :: next
type(nodelist), pointer :: linklist
type(nodelist), pointer :: nextlist
type(nodelist), pointer :: currlist
!
if ( .not. associated(linklist) ) then
allocate(linklist)
linklist%node => next
linklist%nextlist => null()
else
currlist => linklist
do while ( associated(currlist%nextlist) )
currlist => currlist%nextlist
end do
allocate(nextlist)
nextlist%node => next
nextlist%nextlist => null()
currlist%nextlist => nextlist
end if
!
end subroutine add_to_list
!
recursive subroutine destroy ( curr, iun )
!
! This (obscure) code goes down recursively into the "curr" tree,
! then deallocates (hopefully) everything. If "iun" is present,
! the tree is reprinted to unit "iun". Useful for debugging: the
! reprinted tree should have the same structure as the original file
!
type(node), pointer :: curr, next
type(nodelist), pointer :: linklist, nextlist
integer, intent(in), optional :: iun
!
nlevel = nlevel + 1
! print *, nlevel, '<', curr%tag,'>, ',curr%attr
! print *, curr%data(1:min(80,len(curr%data)))
if ( present(iun ) ) then
if ( allocated(curr%attr) ) then
write(iun,'("<",A," ",A,">")') trim(curr%tag),trim(curr%attr)
else
write(iun,'("<",A,">")') trim(curr%tag)
end if
if ( allocated(curr%data) ) write(iun,'(A)') trim(curr%data)
end if
! Go down recursively on the tree (note the call to itself below)
linklist => curr%linklist
do while ( associated(linklist) )
call destroy(linklist%node, iun)
nextlist => linklist%nextlist
deallocate (linklist)
! The linked list must be explicitly deallocated to avoid memory leaks
linklist => nextlist
end do
!
if ( present(iun ) ) write(iun,'("</",A,">")') trim(curr%tag)
nlevel = nlevel - 1
! now deallocate all memory
if ( allocated(curr%tag) ) deallocate (curr%tag)
if ( allocated(curr%data) ) deallocate (curr%data)
if ( allocated(curr%attr) ) deallocate (curr%attr)
!
if ( associated(curr%prev) ) then
! go down one level and deallocate
next => curr%prev
deallocate(curr)
curr => next
else
call destroyml ( )
if ( nlevel /= -1 ) print *, 'destroy: did not reach root level?'
end if
!
end subroutine destroy
!
subroutine destroyll (linklist)
type(nodelist), pointer :: linklist
type(nodelist), pointer :: nextlist
do while ( associated(linklist) )
nextlist => linklist%nextlist
deallocate(linklist)
! if ( .not.associated(nextlist) ) exit
linklist =>nextlist
end do
end subroutine destroyll
!
subroutine destroyml ( )
type(metalist), pointer :: prevml
! deallocate all linked lists by going back into "ml"
do while ( associated(ml) )
call destroyll(ml%linklist)
prevml => ml%prevmeta
deallocate (ml)
ml => prevml
end do
end subroutine destroyml
!
function getelementsbytagname(root,tag)
!
type(node), pointer, intent(in) :: root
character(len=*), intent(in) :: tag
type(nodelist), pointer :: getelementsbytagname
!
type(nodelist), pointer :: linklist, outlist, newlist
type(metalist), pointer :: nextml
integer :: n
!
n = -1
getelementsbytagname => null()
if ( associated( root%linklist ) ) then
linklist => root%linklist
lista: do
if ( trim(adjustl(tag)) == linklist%node%tag ) then
n = n+1
! print *, 'info: tag: ',tag,' found: n=',n
allocate(newlist)
newlist%node => linklist%node
newlist%nextlist => null()
if ( n == 0 ) then
getelementsbytagname => newlist
outlist => getelementsbytagname
else
outlist%nextlist => newlist
outlist => newlist
end if
end if
if ( .not. associated( linklist%nextlist ) ) exit lista
linklist => linklist%nextlist
end do lista
!
! Store linked list at the end of "meta-list" ml,
! keeping track of the previous one, for later deallocation
!
if ( .not.associated(ml) ) then
allocate(ml)
else
allocate(nextml)
nextml%prevmeta => ml
ml => nextml
end if
ml%linklist => getelementsbytagname
!
end if
! if ( n < 0 ) print *, ' tag: ',tag,' not found'
!
end function getelementsbytagname
!
function getfirstchild(root, ex)
!
type(node), pointer, intent(in) :: root
type(domexception), intent (out), optional :: ex
type(node), pointer :: getfirstchild
!
if ( associated( root ) ) then
getfirstchild => root
if ( present(ex) ) ex%code = 0
else
getfirstchild => null()
if ( present(ex) ) ex%code = 1
endif
!
end function getfirstchild
!
function gettagname(root, ex)
!
type(node), pointer, intent(in) :: root
type(domexception), intent (out), optional :: ex
character(len=:), allocatable :: gettagname
!
gettagname = root%tag
! ignored
if ( present(ex) ) ex%code = 0
!
end function gettagname
!
integer function getlength (llist) result(n)
!
type(nodelist), pointer, intent(in) :: llist
type(nodelist), pointer :: mylist
!
n = 0
if ( .not.associated(llist) ) return
n = 1
mylist => llist
lista: do while( associated(mylist%nextlist) )
n = n + 1
mylist => mylist%nextlist
end do lista
!
end function getlength
!
function item (llist,n)
!
type(nodelist), pointer, intent(in) :: llist
integer, intent(in) :: n
type(node), pointer :: item
!
type(nodelist), pointer :: mylist
integer :: i
!
item => null()
if ( .not.associated(llist) ) return
mylist => llist
lista: do i=0,n-1
mylist => mylist%nextlist
end do lista
item => mylist%node
!
end function item
!
logical function hasattribute(root,attr,val) result(found)
!
type(node), pointer, intent(in) :: root
character(len=*), intent(in) :: attr
character(len=*), intent(out), optional :: val
!
integer :: la, l0, i1, i2, i
logical :: in_attrval
character(len=1) :: delimiter
!
if(allocated(root%attr)) then
la = len_trim(root%attr)
l0 = len_trim(adjustl(root%attr))
else
la = 0
l0 = 0
endif
in_attrval=.false.
found = .false.
i1 = 0
scan: do i=la-l0+1,la
if ( .not. in_attrval) then
if (root%attr(i:i) == '"' .or. root%attr(i:i) == "'") then
in_attrval=.true.
delimiter = root%attr(i:i)
! write(*,'("attr:",a,", ")',advance='no') root%attr(i1:i2)
found = ( attr == root%attr(i1:i2) )
i1 = i+1
i2 = 0
else if (i1 == 0 .and. root%attr(i:i) /= ' ') then
i1 = i
else if (i1 > 0 .and. root%attr(i:i) /= ' ' .and. root%attr(i:i) /= '=' ) then
i2 = i
end if
else
if (root%attr(i:i) == delimiter ) then
in_attrval=.false.
i2 = i-1
! write(*,'("value:",a,".")') root%attr(i1:i2)
if ( present(val) ) val = root%attr(i1:i2)
if ( found ) return
i1 = 0
end if
end if
end do scan
!
end function hasattribute
!
subroutine extractdataattribute_c(root, attr, cval, iostat)
type(node), pointer, intent(in) :: root
character(len=*), intent(in) :: attr
character(len=*), intent(out) :: cval
integer, intent(out), optional:: iostat
!
if (present(iostat)) iostat=0
if ( hasattribute(root, attr, cval) ) return
if (present(iostat)) iostat=1
cval = ' '
!
end subroutine extractdataattribute_c
!
subroutine extractdataattribute_l(root, attr, lval, iostat)
type(node), pointer, intent(in) :: root
character(len=*), intent(in) :: attr
logical, intent(out) :: lval
character(len=80) :: val
integer, intent(out), optional:: iostat
!
if (present(iostat)) iostat=0
if ( hasattribute(root, attr, val) ) then
read(val,*, end=10,err=10) lval
return
end if
! not found or not readable
10 lval = .false.
if (present(iostat)) iostat=1
!
end subroutine extractdataattribute_l
!
subroutine extractdataattribute_i(root, attr, ival, iostat)
type(node), pointer, intent(in) :: root
character(len=*), intent(in) :: attr
integer, intent(out) :: ival
character(len=80) :: val
integer, intent(out), optional:: iostat
!
if (present(iostat)) iostat=0
if ( hasattribute(root, attr, val) ) then
read(val,*, end=10,err=10) ival
return
end if
! not found or not readable
10 ival = 0
if (present(iostat)) iostat=1
!
end subroutine extractdataattribute_i
!
subroutine extractdataattribute_iv(root, attr, ivec, iostat)
type(node), pointer, intent(in) :: root
character(len=*), intent(in) :: attr
integer, intent(out) :: ivec(:)
character(len=80) :: val
integer, intent(out), optional:: iostat
!
if (present(iostat)) iostat=0
if ( hasattribute(root, attr, val) ) then
read(val,*, end=10,err=10) ivec
return
end if
! not found or not readable
10 ivec = 0
if (present(iostat)) iostat=1
!
end subroutine extractdataattribute_iv
!
subroutine extractdataattribute_r(root, attr, rval, iostat)
type(node), pointer, intent(in) :: root
character(len=*), intent(in) :: attr
real(dp), intent(out) :: rval
character(len=80) :: val
integer, intent(out), optional:: iostat
!
if (present(iostat)) iostat=0
if ( hasattribute(root, attr, val) ) then
read(val,*, end=10,err=10) rval
return
end if
! not found or not readable
10 rval = 0
if (present(iostat)) iostat=1
!
end subroutine extractdataattribute_r
!
subroutine extractdatacontent_c(root, cval, iostat)
type(node), pointer, intent(in) :: root
character(len=*), intent(out) :: cval
integer, intent(out), optional :: iostat
integer :: ios
!
cval=' '
ios = 0
if ( allocated(root%data) ) then
if ( len_trim(root%data) > 0 ) read(root%data,*,iostat=ios) cval
else
ios = 1
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_c
!
subroutine extractdatacontent_cv(root, cvec, iostat)
type(node), pointer, intent(in) :: root
character(len=*), intent(inout), pointer :: cvec(:)
integer, intent(out), optional :: iostat
integer :: ibeg, iend, n, ios
!
cvec(:) = ' '
ios = 0
if ( allocated(root%data) ) then
if ( len_trim(root%data) > 0 ) then
iend = 0
do n=1,size(cvec)
ios = find_token( root%data, ibeg, iend)
if ( ios == 0 ) then
cvec(n) = root%data(ibeg:iend)
else
cvec(n) = ' '
end if
end do
end if
else
ios = 1
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_cv
!
subroutine extractdatacontent_l(root, lval, iostat)
type(node), pointer, intent(in) :: root
logical, intent(out) :: lval
integer, intent(out), optional :: iostat
integer :: ios
if ( allocated(root%data) ) then
read(root%data,*,iostat=ios) lval
else
lval=.false.
ios = 1
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_l
!
subroutine extractdatacontent_i(root, ival, iostat)
type(node), pointer, intent(in) :: root
integer, intent(out) :: ival
integer, intent(out), optional :: iostat
integer :: ios
!
if ( allocated(root%data) ) then
read(root%data,*,iostat=ios) ival
else
ival=0
ios = 1
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_i
!
subroutine extractdatacontent_iv(root, ivec, iostat)
type(node), pointer, intent(in) :: root
integer, intent(out) :: ivec(:)
integer, intent(out), optional :: iostat
integer :: ios, n, iend, ibeg
!
ios = 1
if ( allocated(root%data) ) then
! the simple solution fails if root%data > 1024 characters:
! read(root%data,*,iostat=ios) ivec
iend = 0
do n=1,size(ivec)
ios = find_token( root%data, ibeg, iend)
if ( ios == 0 ) then
read(root%data(ibeg:iend),*,iostat=ios) ivec(n)
else
ivec(n) = 0
end if
end do
else
ivec(:) = 0
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_iv
!
subroutine extractdatacontent_r(root, rval, iostat)
type(node), pointer, intent(in) :: root
real(dp), intent(out) :: rval
integer, intent(out), optional :: iostat
integer :: ios
!
if ( allocated(root%data) ) then
read(root%data,*,iostat=ios) rval
else
rval=0.0_dp
ios = 1
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_r
!
subroutine extractdatacontent_rv(root, rvec, iostat)
type(node), pointer, intent(in) :: root
real(dp), intent(out) :: rvec(:)
integer, intent(out), optional :: iostat
integer :: ios, n, iend, ibeg
!
ios = 1
if ( allocated(root%data) ) then
! the simple solution fails if root%data > 1024 characters:
! read(root%data,*,iostat=ios) rvec
iend = 0
do n=1,size(rvec)
ios = find_token( root%data, ibeg, iend)
if ( ios == 0 ) then
read(root%data(ibeg:iend),*,iostat=ios) rvec(n)
else
rvec(n) = 0.0_dp
end if
end do
else
rvec(:) = 0.0_dp
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_rv
!
subroutine extractdatacontent_rm(root, rmat, iostat)
type(node), pointer, intent(in) :: root
real(dp), intent(out) :: rmat(:,:)
integer, intent(out), optional :: iostat
integer :: ios, n, m, iend, ibeg
!
ios = 1
if ( allocated(root%data) ) then
! the simple solution fails if root%data > 1024 characters:
! read(root%data,*,iostat=ios) rmat
iend = 0
do m=1,size(rmat,2)
do n=1,size(rmat,1)
ios = find_token( root%data, ibeg, iend)
if ( ios == 0 ) then
read(root%data(ibeg:iend),*,iostat=ios) rmat(n,m)
else
rmat(n,m) = 0.0_dp
end if
end do
end do
else
rmat = 0.0_dp
end if
if ( present(iostat) ) iostat=ios
!
end subroutine extractdatacontent_rm
!
integer function find_token ( data, ibeg, iend )
!
! Locate tokens (numbers, fields) in a string
! Tokens are assumed to be separated by space or commas
!
! on input:
! data string containing tokens
! iend 0 on first run, end position of previous token otherwise
! On output:
! find_token 0 if token found, 1 otherwise
! ibeg, iend if find_token, data(ibeg:iend) contains a token
! if not, ibeg=iend, iend unchanged
!
! Beware: will not work if empty tokens and multiple commas and present,
! e.g.: "field1, ,field3" where field2 is empty
!
character(len=:), allocatable, intent(in) :: data
integer, intent(out) :: ibeg
integer, intent(inout):: iend
integer:: lt
!
lt = len_trim(data)
find_token = 1
do ibeg = iend+1, lt
if ( data(ibeg:ibeg) == ' ' .or. data(ibeg:ibeg) == ',' ) then
cycle
else
find_token = 0
exit
end if
end do
if ( find_token == 0 ) then
do iend = ibeg, lt
if ( data(iend:iend) /= ' ' .and. data(iend:iend) /= ',' ) then
cycle
else
exit
end if
end do
iend = min(iend, lt)
else
ibeg = iend
end if
!
end function find_token
!
end module dom