From c60fa06a6a40c632e9e639eb2ae8f701174d1ca9 Mon Sep 17 00:00:00 2001 From: Nicholas Longo Date: Thu, 12 May 2011 21:46:18 +0000 Subject: [PATCH] This updates improves the MathML support of texvc. New new tags have been added, but the parser now recognizes numbers containing decimal points as being renderable using . The delimiters (,[,{ now render using tags, this allows for equations such as "y=f(x)" to be rendered to MathML. Finally for standard functions such as \sin, now render as \sin⁡, etc so standard function names are supported. --- math/html.ml | 2 ++ math/lexer.mll | 39 +++++++++++++++++++++++++-------------- math/mathml.ml | 6 +++++- math/parser.mly | 2 +- math/render_info.mli | 3 +++ math/texutil.ml | 1 + 6 files changed, 37 insertions(+), 16 deletions(-) diff --git a/math/html.ml b/math/html.ml index e880f07..1b0e8cb 100644 --- a/math/html.ml +++ b/math/html.ml @@ -28,6 +28,7 @@ let rec html_render_flat ctx = function TEX_LITERAL (HTMLABLE (ft,_,sh))::r -> (html_liberal (); (font_render sh (ctx,ft))^html_render_flat ctx r) | TEX_LITERAL (HTMLABLEC(ft,_,sh))::r -> (font_render sh (ctx,ft))^html_render_flat ctx r | TEX_LITERAL (MHTMLABLEC(ft,_,sh,_,_))::r -> (font_render sh (ctx,ft))^html_render_flat ctx r + | TEX_LITERAL (MHTMLABLEFC(ft,_,sh,_,_,_))::r -> (font_render sh (ctx,ft))^html_render_flat ctx r | TEX_LITERAL (HTMLABLEM(ft,_,sh))::r -> (html_moderate(); (font_render sh (ctx,ft))^html_render_flat ctx r) | TEX_LITERAL (HTMLABLE_BIG (_,sh))::r -> (html_liberal (); sh^html_render_flat ctx r) | TEX_FUN1hl (_,(f1,f2),a)::r -> f1^(html_render_flat ctx [a])^f2^html_render_flat ctx r @@ -75,6 +76,7 @@ let rec html_render_deep ctx = function | TEX_LITERAL (HTMLABLEM(ft,_,sh))::r -> (html_moderate(); ("",(font_render sh (ctx,ft)),"")::html_render_deep ctx r) | TEX_LITERAL (HTMLABLEC(ft,_,sh))::r -> ("",(font_render sh (ctx,ft)),"")::html_render_deep ctx r | TEX_LITERAL (MHTMLABLEC(ft,_,sh,_,_))::r -> ("",(font_render sh (ctx,ft)),"")::html_render_deep ctx r + | TEX_LITERAL (MHTMLABLEFC(ft,_,sh,_,_,_))::r -> ("",(font_render sh (ctx,ft)),"")::html_render_deep ctx r | TEX_LITERAL (HTMLABLE_BIG (_,sh))::r -> (html_liberal (); ("",sh,"")::html_render_deep ctx r) | TEX_FUN2h (_,f,a,b)::r -> (html_liberal (); (f a b)::html_render_deep ctx r) | TEX_INFIXh (_,f,a,b)::r -> (html_liberal (); (f a b)::html_render_deep ctx r) diff --git a/math/lexer.mll b/math/lexer.mll index 8a667b2..c53a43f 100644 --- a/math/lexer.mll +++ b/math/lexer.mll @@ -6,9 +6,9 @@ let space = [' ' '\t' '\n' '\r'] let alpha = ['a'-'z' 'A'-'Z'] let literal_id = ['a'-'z' 'A'-'Z'] -let literal_mn = ['0'-'9'] +let literal_mn = ['0'-'9' '.'] let literal_uf_lt = [',' ':' ';' '?' '!' '\''] -let delimiter_uf_lt = ['(' ')' '.'] +let delimiter_uf_lt = ['(' ')'] let literal_uf_op = ['+' '-' '*' '='] let delimiter_uf_op = ['/' '|'] let boxchars = ['0'-'9' 'a'-'z' 'A'-'Z' '+' '-' '*' ',' '=' '(' ')' ':' '/' ';' '?' '.' '!' '\'' '`' ' ' '\128'-'\255'] @@ -52,7 +52,7 @@ rule token = parse | literal_id { let str = Lexing.lexeme lexbuf in LITERAL (MHTMLABLEC (FONT_IT, str,str,MI,str)) } | literal_mn { let str = Lexing.lexeme lexbuf in LITERAL (MHTMLABLEC (FONT_RM, str,str,MN,str)) } | literal_uf_lt { let str = Lexing.lexeme lexbuf in LITERAL (HTMLABLEC (FONT_UFH, str,str)) } - | delimiter_uf_lt { let str = Lexing.lexeme lexbuf in DELIMITER (HTMLABLEC (FONT_UFH, str,str)) } + | delimiter_uf_lt { let str = Lexing.lexeme lexbuf in DELIMITER (MHTMLABLEC (FONT_UFH, str,str,MO,str)) } | "-" { let str = Lexing.lexeme lexbuf in LITERAL (MHTMLABLEC (FONT_UFH,"-"," − ",MO,str))} | literal_uf_op { let str = Lexing.lexeme lexbuf in LITERAL (MHTMLABLEC (FONT_UFH, str," "^str^" ",MO,str)) } | delimiter_uf_op { let str = Lexing.lexeme lexbuf in DELIMITER (MHTMLABLEC (FONT_UFH, str," "^str^" ",MO,str)) } @@ -60,20 +60,31 @@ rule token = parse | "\\sqrt" space * "[" { FUN_AR1opt "\\sqrt" } | "\\xleftarrow" space * "[" { Texutil.tex_use_ams(); FUN_AR1opt "\\xleftarrow" } | "\\xrightarrow" space * "[" { Texutil.tex_use_ams(); FUN_AR1opt "\\xrightarrow" } - | "\\" (latex_function_names as name) space * "(" { LITERAL (HTMLABLEC(FONT_UFH,"\\" ^ name ^ "(", name ^ "(")) } - | "\\" (latex_function_names as name) space * "[" { LITERAL (HTMLABLEC(FONT_UFH,"\\" ^ name ^ "[", name ^ "[")) } - | "\\" (latex_function_names as name) space * "\\{" { LITERAL (HTMLABLEC(FONT_UFH,"\\" ^ name ^ "\\{", name ^ "{")) } - | "\\" (latex_function_names as name) space * { LITERAL (HTMLABLEC(FONT_UFH,"\\" ^ name ^ " ", name ^ " ")) } - | "\\" (mediawiki_function_names as name) space * "(" { (Texutil.tex_use_ams(); LITERAL (HTMLABLEC(FONT_UFH,"\\operatorname{" ^ name ^ "}(", name^ "("))) } - | "\\" (mediawiki_function_names as name) space * "[" { (Texutil.tex_use_ams(); LITERAL (HTMLABLEC(FONT_UFH,"\\operatorname{" ^ name ^ "}[", name^ "["))) } - | "\\" (mediawiki_function_names as name) space * "\\{" { (Texutil.tex_use_ams(); LITERAL (HTMLABLEC(FONT_UFH,"\\operatorname{" ^ name ^ "}\\{", name^ "{"))) } - | "\\" (mediawiki_function_names as name) space * { (Texutil.tex_use_ams(); LITERAL (HTMLABLEC(FONT_UFH,"\\operatorname{" ^ name ^ "}", name ^ " "))) } + | "\\" (latex_function_names as name) space * "(" + { LITERAL (MHTMLABLEFC(FONT_UFH,"\\" ^ name ^ "(", name ^ "(", MF, name, "(")) } + | "\\" (latex_function_names as name) space * "[" + { LITERAL (MHTMLABLEFC(FONT_UFH,"\\" ^ name ^ "[", name ^ "[", MF, name, "[")) } + | "\\" (latex_function_names as name) space * "\\{" + { LITERAL (MHTMLABLEFC(FONT_UFH, "\\" ^ name ^ "\\{", name ^ "{", MF, name, "{")) } + | "\\" (latex_function_names as name) space * + { LITERAL (MHTMLABLEC(FONT_UFH,"\\" ^ name, name ^ " ", MF, name)) } + | "\\" (mediawiki_function_names as name) space * "(" + { (Texutil.tex_use_ams(); LITERAL (MHTMLABLEFC(FONT_UFH, + "\\operatorname{" ^ name ^ "}(", name ^ "(", MF, name, "("))) } + | "\\" (mediawiki_function_names as name) space * "[" + { (Texutil.tex_use_ams(); LITERAL (MHTMLABLEFC(FONT_UFH, + "\\operatorname{" ^ name ^ "}[", name ^ "[", MF, name, "["))) } + | "\\" (mediawiki_function_names as name) space * "\\{" + { (Texutil.tex_use_ams(); LITERAL (MHTMLABLEFC(FONT_UFH, + "\\operatorname{" ^ name ^ "}\\{", name ^ "{", MF, name, "{"))) } + | "\\" (mediawiki_function_names as name) space * + { (Texutil.tex_use_ams(); LITERAL (MHTMLABLEC(FONT_UFH,"\\operatorname{" ^ name ^ "}", name ^ " ", MF, name))) } | "\\," { LITERAL (HTMLABLE (FONT_UF, "\\,"," ")) } | "\\ " { LITERAL (HTMLABLE (FONT_UF, "\\ "," ")) } | "\\;" { LITERAL (HTMLABLE (FONT_UF, "\\;"," ")) } | "\\!" { LITERAL (TEX_ONLY "\\!") } - | "\\{" { DELIMITER (HTMLABLEC(FONT_UFH,"\\{","{")) } - | "\\}" { DELIMITER (HTMLABLEC(FONT_UFH,"\\}","}")) } + | "\\{" { DELIMITER (MHTMLABLEC(FONT_UFH,"\\{","{",MO,"{")) } + | "\\}" { DELIMITER (MHTMLABLEC(FONT_UFH,"\\}","}",MO,"}")) } | "\\|" { DELIMITER (HTMLABLE (FONT_UFH,"\\|","||")) } | "\\_" { LITERAL (HTMLABLEC(FONT_UFH,"\\_","_")) } | "\\#" { LITERAL (HTMLABLE (FONT_UFH,"\\#","#")) } @@ -109,7 +120,7 @@ rule token = parse | '%' { LITERAL (HTMLABLEC(FONT_UFH,"\\%","%")) } | '$' { LITERAL (HTMLABLEC(FONT_UFH,"\\$","$")) } | '~' { LITERAL (HTMLABLE (FONT_UF, "~"," ")) } - | '[' { DELIMITER (HTMLABLEC(FONT_UFH,"[","[")) } + | '[' { DELIMITER (MHTMLABLEC(FONT_UFH,"[","[",MO,"[")) } | ']' { SQ_CLOSE } | '{' { CURLY_OPEN } | '}' { CURLY_CLOSE } diff --git a/math/mathml.ml b/math/mathml.ml index b6c76af..d01414c 100644 --- a/math/mathml.ml +++ b/math/mathml.ml @@ -1,13 +1,15 @@ open Tex open Render_info -type t = TREE_MN of string | TREE_MO of string | TREE_MI of string +type t = TREE_MN of string | TREE_MO of string | TREE_MI of string | TREE_MF of string | TREE_MFB of string * string let rec make_mathml_tree = function TREE_MN a::otr,TEX_LITERAL(MHTMLABLEC(_,_,_,MN,b))::itr -> make_mathml_tree(TREE_MN (a^b)::otr,itr) | otr,TEX_LITERAL(MHTMLABLEC(_,_,_,MN,a))::itr -> make_mathml_tree(TREE_MN a::otr,itr) | otr,TEX_LITERAL(MHTMLABLEC(_,_,_,MO,a))::itr -> make_mathml_tree(TREE_MO a::otr,itr) | otr,TEX_LITERAL(MHTMLABLEC(_,_,_,MI,a))::itr -> make_mathml_tree(TREE_MI a::otr,itr) + | otr,TEX_LITERAL(MHTMLABLEC(_,_,_,MF,a))::itr -> make_mathml_tree(TREE_MF a::otr,itr) + | otr,TEX_LITERAL(MHTMLABLEFC(_,_,_,MF,a,b))::itr -> make_mathml_tree(TREE_MFB (a,b)::otr,itr) | otr,TEX_CURLY(crl)::itr -> make_mathml_tree(otr,crl@itr) | otr,[] -> List.rev otr | _ -> failwith "failed to render mathml" @@ -16,5 +18,7 @@ let render_mathml_tree = function TREE_MN s -> ""^s^"" | TREE_MI s -> ""^s^"" | TREE_MO s -> ""^s^"" + | TREE_MF s -> ""^s^"⁡" + | TREE_MFB (s,b) -> ""^s^"⁡"^""^b^"" let render tree = try Some (Util.mapjoin render_mathml_tree (make_mathml_tree ([],tree))) with _ -> None diff --git a/math/parser.mly b/math/parser.mly index 29882fb..29200df 100644 --- a/math/parser.mly +++ b/math/parser.mly @@ -2,7 +2,7 @@ open Tex open Render_info - let sq_close_ri = HTMLABLEC(FONT_UFH,"]", "]") + let sq_close_ri = MHTMLABLEC(FONT_UFH,"]", "]",MO,"]") %} %token LITERAL DELIMITER %token FUN_AR2 FUN_INFIX FUN_AR1 DECL FUN_AR1opt BIG FUN_AR2nb diff --git a/math/render_info.mli b/math/render_info.mli index d5e7fde..ee1c8b1 100644 --- a/math/render_info.mli +++ b/math/render_info.mli @@ -11,10 +11,13 @@ type math_class = MN | MI | MO + | MF + type t = HTMLABLEC of font_class * string * string | HTMLABLEM of font_class * string * string | HTMLABLE of font_class * string * string | MHTMLABLEC of font_class * string * string * math_class * string + | MHTMLABLEFC of font_class * string * string * math_class * string * string | HTMLABLE_BIG of string * string | TEX_ONLY of string diff --git a/math/texutil.ml b/math/texutil.ml index 61c27de..efbf6a3 100644 --- a/math/texutil.ml +++ b/math/texutil.ml @@ -9,6 +9,7 @@ let tex_part = function | HTMLABLEM (_,t,_) -> t | HTMLABLEC (_,t,_) -> t | MHTMLABLEC (_,t,_,_,_) -> t + | MHTMLABLEFC (_,t,_,_,_,_) -> t | HTMLABLE_BIG (t,_) -> t | TEX_ONLY t -> t