\ProvidesExplPackage {tagpdf-struct-code} {2019/07/02} {0.61} {part of tagpdf - code related to storing structure} % I will use a latex counter for the structure count % to have a chance to avoid double structures in align etc \newcounter { g__uftag_struct_abs_int } \int_gzero:N \c@g__uftag_struct_abs_int \zref@newprop {tagstruct} [0] { \int_use:N \c@g__uftag_struct_abs_int } \zref@newlist {uftagzrlstruct} \zref@addprop {uftagzrlstruct}{tagstruct} % a sequence stores structnum -> the obj numbers % to allow easy mapping over the structures \__uftag_seq_new:N \g__uftag_struct_objR_seq % a sequence for the structure stack. When a sequence is opened it's number is put on the stack. \seq_new:N \g__uftag_struct_stack_seq \seq_gpush:Nn \g__uftag_struct_stack_seq {0} %this variables will hold the top entry and the parent on the stack \tl_new:N \l__uftag_struct_stack_parent_tmp_tl \tl_new:N \g__uftag_struct_stack_current_tl % I need at least one structure: the StructTreeRoot % normally it should have only one kid, e.g. the document element. % The data of the StructTreeRoot and the StructElem are in properties: % \g__uftag_struct_0_prop for the root % \g__uftag_struct_N_prop, N >=1 % they have all the keys % objnum - number, % Type - StructTreeRoot or StructElem % num - number (identical to the num in the name, or 0 for the root) % and the keys from the two following lists % (the root has a special set of properties). % the values of the prop should be already escaped properly % when the entries are created (title,lange,alt,E,actualtext) \seq_const_from_clist:Nn \c__uftag_struct_StructTreeRoot_entries_seq {%p. 857/858 Type, % always /StructTreeRoot K, % kid, dictionary or array of dictionaries IDTree, % currently unused ParentTree, % required,obj ref to the parent tree ParentTreeNextKey, %optional RoleMap, ClassMap } \seq_const_from_clist:Nn \c__uftag_struct_StructElem_entries_seq {%p 858 f Type, %always /StructElem S, %tag/type P, %parent ID, %optional Pg, %obj num of starting page, optional K, %kids A, %attributes, probably unused C, %class "" %R, T, %title, value in () or <> Lang, %language Alt, % value in () or <> E, %abreviation ActualText } % I need an output handler for each prop, to get expandable output % see https://tex.stackexchange.com/questions/424208/expandable-version-of-a-expl3-command/424213#424213 \cs_new:Nn \__uftag_struct_output_prop_aux:nn %#1 num, #2 key { \prop_if_in:cnT { g__uftag_struct_#1_prop } { #2 } { \c_space_tl/#2~ \prop_item:cn{ g__uftag_struct_#1_prop } { #2 } } } \cs_new:Nn \__uftag_new_output_prop_handler:n { \cs_new:cn { __uftag_struct_output_prop_#1:n } { \__uftag_struct_output_prop_aux:nn {#1}{##1} } } % the first one, the StructTreeRoot is special, so % created manually: \__uftag_prop_new:c { g__uftag_struct_0_prop } \__uftag_new_output_prop_handler:n {0} \tl_gset:Nn \g__uftag_struct_stack_current_tl {0} \__uftag_seq_new:c { g__uftag_struct_kids_0_seq} \__uftag_prop_gput:cno { g__uftag_struct_0_prop } { objnum} { \c_uftag_tree_obj_structtreeroot_tl } \__uftag_prop_gput:cno { g__uftag_struct_0_prop } { Type } { /StructTreeRoot } % the constants are defined in the tree code. \__uftag_prop_gput:cnx { g__uftag_struct_0_prop } { ParentTree } { \c__uftag_tree_obj_parenttree_tl\c_space_tl 0\c_space_tl R } \__uftag_prop_gput:cnx { g__uftag_struct_0_prop } { RoleMap } { \c__uftag_tree_obj_rolemap_tl\c_space_tl 0\c_space_tl R } \__uftag_prop_gput:cno { g__uftag_struct_0_prop } { entries } { StructTreeRoot } \__uftag_prop_gput:cno { g__uftag_struct_0_prop } { num } { 0 } % commands to store the kids % I don't compare the page objects number yet, but always add the /Pg key, perhaps later \cs_new:Nn \__uftag_struct_kid_mc_gput_right:nn %#1 structure num, #2 MCID absnum% { \__uftag_store_pdfpageref:Nn \l_tmpa_tl { \zref@extractdefault{mcid-#2}{tagabspage}{1} } \__uftag_seq_gput_right:cx { g__uftag_struct_kids_#1_seq } { << /Type\c_space_tl/MCR\c_space_tl /Pg\c_space_tl\l_tmpa_tl\c_space_tl0\c_space_tl R\c_space_tl /MCID\c_space_tl\zref@extractdefault{mcid-#2}{tagmcid}{1} >> } } \cs_new:Nn\__uftag_struct_kid_struct_gput_right:nn %#1 num of parent struct, #2 kid struct { \__uftag_seq_gput_right:cx { g__uftag_struct_kids_#1_seq } { \prop_item:cn { g__uftag_struct_#2_prop } { objnum } \c_space_tl 0 \c_space_tl R } } \cs_new:Nn\__uftag_struct_kid_link_gput_right:nn %#1 num of parent struct, #2 obj number of link { \__uftag_pdfobj:Nn \l__uftag_tmpa_tl { << /Type \c_space_tl /OBJR \c_space_tl /Obj \c_space_tl #2 \c_space_tl 0 \c_space_tl R >> } \__uftag_seq_gput_right:cx { g__uftag_struct_kids_#1_seq } { \l__uftag_tmpa_tl \c_space_tl 0 \c_space_tl R } } \cs_generate_variant:Nn\__uftag_struct_kid_link_gput_right:nn { nx} \cs_new:Nn\__uftag_struct_exchange_kid_command:N %N= seq { \seq_gpop_left:NN #1 \l_tmpa_tl \regex_replace_once:nnN {\c{\__uftag_mc_insert_mcid_kids:n}} {\c{\__uftag_mc_insert_mcid_single_kids:n}} \l_tmpa_tl \seq_gput_left:NV #1 \l_tmpa_tl } \cs_generate_variant:Nn\__uftag_struct_exchange_kid_command:N { c } \cs_new:Nn \__uftag_struct_fill_kid_key:n %#1 is the struct num { \int_case:nnF { \seq_count:c { g__uftag_struct_kids_\prop_item:cn{ g__uftag_struct_#1_prop }{num}_seq } } { { 0 } { } %no kids, do nothing { 1 } % 1 kid, insert { % in this case we in luamode need a special command in luamode to get the array right. See issue #13 \bool_if:NT\g__uftag_mode_lua_bool { \__uftag_struct_exchange_kid_command:c {g__uftag_struct_kids_\prop_item:cn{ g__uftag_struct_#1_prop }{num}_seq} } \__uftag_prop_gput:cnx { g__uftag_struct_#1_prop } {K} { \seq_item:cn { g__uftag_struct_kids_\prop_item:cn{ g__uftag_struct_#1_prop }{num}_seq }{1} } } % } { %many kids, use an array \__uftag_prop_gput:cnx { g__uftag_struct_#1_prop } {K} { [ \seq_use:cn { g__uftag_struct_kids_\prop_item:cn{ g__uftag_struct_#1_prop }{num}_seq } { \c_space_tl } ] } } } % this command can be used for roots and structure elements % #1 is a num \tl_new:N \l_uftag_struct_dict_content_tl \cs_new:Nn \__uftag_struct_get_dict_content:n { \tl_set:Nn \l_uftag_struct_dict_content_tl {<<} \seq_map_inline:cn { c__uftag_struct_\prop_item:cn{ g__uftag_struct_#1_prop }{entries}_entries_seq } { \tl_put_right:Nx \l_uftag_struct_dict_content_tl { \prop_if_in:cnT { g__uftag_struct_#1_prop } { ##1 } { \c_space_tl/##1~\prop_item:cn{ g__uftag_struct_#1_prop } { ##1 } } } } \tl_put_right:Nn \l_uftag_struct_dict_content_tl { >> } } % #1 is the struct num \cs_new:Nn \__uftag_struct_write_obj:n { \prop_if_in:cnTF { g__uftag_struct_#1_prop } { objnum } { \__uftag_struct_fill_kid_key:n { #1 } %\prop_show:c { g__uftag_struct_#1_prop } \__uftag_struct_get_dict_content:n { #1 } \__uftag_pdfuseobjnum:xx { \prop_item:cn { g__uftag_struct_#1_prop } {objnum} } { \l_uftag_struct_dict_content_tl } } { \msg_error:nnn {uftag} { struct-no-objnum } { #1} } } % keys for the user commands % should I pass the values (e.g. the tag) through an escape command? \keys_define:nn { uftag / struct } { label .tl_set:N = \l__uftag_struct_key_label_tl, stash .bool_set:N = \l__uftag_struct_elem_stash_bool, tag .code:n = % S property {%%????????? \pdfescapename?? \tl_set:Nx \l__uftag_tmpa_tl { #1 } \bool_if:NT \g__uftag_check_tags_bool { \__uftag_check_structure_tag:N \l__uftag_tmpa_tl } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { S } { /\exp_not:V\l__uftag_tmpa_tl } }, title .code:n = % T property { \str_set_convert:Nnon \l__uftag_tmpa_str { #1 } { \g__uftag_inputencoding_tl } { utf16/hex } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { T } { <\l__uftag_tmpa_str> } }, title-o .code:n = % T property { \str_set_convert:Nnon \l__uftag_tmpa_str { #1 } { \g__uftag_inputencoding_tl } { utf16/hex } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { T } { <\l__uftag_tmpa_str> } }, alttext .code:n = % Alt property { \str_set_convert:Nnon \l__uftag_tmpa_str { #1 } { \g__uftag_inputencoding_tl } { utf16/hex } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { Alt } { <\l__uftag_tmpa_str> } }, alttext-o .code:n = % Alt property { \str_set_convert:Noon \l__uftag_tmpa_str { #1 } { \g__uftag_inputencoding_tl } { utf16/hex } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { Alt } { <\l__uftag_tmpa_str> } }, actualtext .code:n = % ActualText property { \str_set_convert:Nnon \l__uftag_tmpa_str { #1 } { \g__uftag_inputencoding_tl } { utf16/hex } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { ActualText } { <\l__uftag_tmpa_str>} }, actualtext-o .code:n = % ActualText property { \str_set_convert:Noon \l__uftag_tmpa_str { #1 } { \g__uftag_inputencoding_tl } { utf16/hex } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { ActualText } { <\l__uftag_tmpa_str>} }, } \cs_new_protected:Nn \uftag_struct_begin:n { \group_begin: \int_gincr:N \c@g__uftag_struct_abs_int \__uftag_prop_new:c { g__uftag_struct_\int_eval:n { \c@g__uftag_struct_abs_int }_prop } \__uftag_new_output_prop_handler:n {\int_eval:n { \c@g__uftag_struct_abs_int }} \__uftag_seq_new:c { g__uftag_struct_kids_\int_eval:n { \c@g__uftag_struct_abs_int }_seq} \__uftag_pdfreserveobjnum:N \l_tmpa_tl \__uftag_prop_gput:cno { g__uftag_struct_\int_eval:n { \c@g__uftag_struct_abs_int }_prop } { objnum} { \l_tmpa_tl } \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n { \c@g__uftag_struct_abs_int }_prop } { num} { \int_eval:n { \c@g__uftag_struct_abs_int } } \__uftag_prop_gput:cno { g__uftag_struct_\int_eval:n { \c@g__uftag_struct_abs_int }_prop } { Type } { /StructElem } \__uftag_prop_gput:cno { g__uftag_struct_\int_eval:n { \c@g__uftag_struct_abs_int }_prop } { entries } { StructElem } \keys_set:nn {uftag / struct} { #1 } \__uftag_check_structure_has_tag:n { \int_eval:n {\c@g__uftag_struct_abs_int} } \tl_if_empty:NF {\l__uftag_struct_key_label_tl} { \zref@labelbylist{tagpdfstruct-\l__uftag_struct_key_label_tl}{uftagzrlstruct} } %get the potential parent from the stack: \seq_get:NNF \g__uftag_struct_stack_seq \l__uftag_struct_stack_parent_tmp_tl { \msg_error:nn { uftag } { struct-faulty-nesting } } \seq_gpush:NV \g__uftag_struct_stack_seq \c@g__uftag_struct_abs_int \tl_gset:NV \g__uftag_struct_stack_current_tl \c@g__uftag_struct_abs_int %\seq_show:N \g__uftag_struct_stack_seq \bool_if:NF \l__uftag_struct_elem_stash_bool {%set the parent \__uftag_prop_gput:cnx { g__uftag_struct_\int_eval:n {\c@g__uftag_struct_abs_int}_prop } { P } { \prop_item:cn { g__uftag_struct_\l__uftag_struct_stack_parent_tmp_tl _prop} { objnum }~0~R } %record this structure as kid: %\tl_show:N \g__uftag_struct_stack_current_tl %\tl_show:N \l__uftag_struct_stack_parent_tmp_tl \__uftag_struct_kid_struct_gput_right:nn { \l__uftag_struct_stack_parent_tmp_tl } { \g__uftag_struct_stack_current_tl } %\prop_show:c { g__uftag_struct_\g__uftag_struct_stack_current_tl _prop } %\seq_show:c {g__uftag_struct_kids_\l__uftag_struct_stack_parent_tmp_tl _seq} } %\prop_show:c { g__uftag_struct_\g__uftag_struct_stack_current_tl _prop } %\seq_show:c {g__uftag_struct_kids_\l__uftag_struct_stack_parent_tmp_tl _seq} \group_end: } \cs_new_protected:Nn \uftag_struct_end: {%take the current structure num from the stack: %the objects are written later, lua mode hasn't all needed info yet %\seq_show:N \g__uftag_struct_stack_seq \seq_gpop:NNTF \g__uftag_struct_stack_seq \l_tmpa_tl { \int_compare:nNnT {\l__uftag_loglevel_int} > { 0 } { \__uftag_check_info_closing_struct:o { \g__uftag_struct_stack_current_tl } } } { \__uftag_check_no_open_struck: } % get the previous one, shouldn't be empty as the root should be there \seq_get:NNTF \g__uftag_struct_stack_seq \l_tmpa_tl { \tl_gset:NV \g__uftag_struct_stack_current_tl \l_tmpa_tl } { \__uftag_check_no_open_struck: } } \cs_new_protected:Nn \uftag_struct_use:n %#1 is the label { \prop_if_exist:cTF { g__uftag_struct_\zref@extractdefault{tagpdfstruct-#1}{tagstruct}{unknown}_prop } %?????????? { \__uftag_check_struct_used:n {#1} %add the label structure as kid to the current structure (can be the root) \__uftag_struct_kid_struct_gput_right:nn { \g__uftag_struct_stack_current_tl } { \zref@extractdefault{tagpdfstruct-#1}{tagstruct}{0} } %add the current structure to the labeled one as parents \__uftag_prop_gput:cnx { g__uftag_struct_\zref@extractdefault{tagpdfstruct-#1}{tagstruct}{0}_prop } { P } { \prop_item:cn { g__uftag_struct_\g__uftag_struct_stack_current_tl _prop} { objnum }~0~R } } {\msg_warning:nnn{uftag}{struct-label-unknown}{#1}} } %%%% Code to tag links %%%% this works for url, see exp-link.pdf. %%%% it must be checked for other links \cs_new_protected:Nn \__uftag_struct_finish_link: { \bool_if:NT \g_uftag_active_struct_bool { %get the number of the parent link structure: \seq_get:NNF \g__uftag_struct_stack_seq \l__uftag_struct_stack_parent_tmp_tl { \msg_error:nn { uftag } { struct-faulty-nesting } } %put the obj number of link annot in the kid entry: \__uftag_struct_kid_link_gput_right:nx { \l__uftag_struct_stack_parent_tmp_tl } { \int_use:N\__uftag_pdf_lastlink: } % add the parent obj number to the parent tree: \__uftag_parenttree_add_objr:nn { \int_use:N\c@g__uftag_parenttree_obj_int } { \prop_item:cn{ g__uftag_struct_\l__uftag_struct_stack_parent_tmp_tl _prop } { objnum } } % increase the int: \stepcounter{ g__uftag_parenttree_obj_int } } } \endinput