sgml.java

一个自然语言处理的Java开源工具包。LingPipe目前已有很丰富的功能

        SGML_MAP.put("boxHd",'\u2564'); // BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE
        SGML_MAP.put("boxhD",'\u2565'); // BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE
        SGML_MAP.put("boxHD",'\u2566'); // BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL
        SGML_MAP.put("boxhu",'\u2534'); // BOX DRAWINGS LIGHT UP AND HORIZONTAL
        SGML_MAP.put("boxHu",'\u2567'); // BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE
        SGML_MAP.put("boxhU",'\u2568'); // BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE
        SGML_MAP.put("boxHU",'\u2569'); // BOX DRAWINGS DOUBLE UP AND HORIZONTAL
        SGML_MAP.put("boxul",'\u2518'); // BOX DRAWINGS LIGHT UP AND LEFT
        SGML_MAP.put("boxuL",'\u255B'); // BOX DRAWINGS UP SINGLE AND LEFT DOUBLE
        SGML_MAP.put("boxUl",'\u255C'); // BOX DRAWINGS UP DOUBLE AND LEFT SINGLE
        SGML_MAP.put("boxUL",'\u255D'); // BOX DRAWINGS DOUBLE UP AND LEFT
        SGML_MAP.put("boxur",'\u2514'); // BOX DRAWINGS LIGHT UP AND RIGHT
        SGML_MAP.put("boxuR",'\u2558'); // BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE
        SGML_MAP.put("boxUr",'\u2559'); // BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE
        SGML_MAP.put("boxUR",'\u255A'); // BOX DRAWINGS DOUBLE UP AND RIGHT
        SGML_MAP.put("boxv",'\u2502'); // BOX DRAWINGS LIGHT VERTICAL
        SGML_MAP.put("boxV",'\u2551'); // BOX DRAWINGS DOUBLE VERTICAL
        SGML_MAP.put("boxvh",'\u253C'); // BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL
        SGML_MAP.put("boxvH",'\u256A'); // BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE
        SGML_MAP.put("boxVh",'\u256B'); // BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE
        SGML_MAP.put("boxVH",'\u256C'); // BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
        SGML_MAP.put("boxvl",'\u2524'); // BOX DRAWINGS LIGHT VERTICAL AND LEFT
        SGML_MAP.put("boxvL",'\u2561'); // BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE
        SGML_MAP.put("boxVl",'\u2562'); // BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE
        SGML_MAP.put("boxVL",'\u2563'); // BOX DRAWINGS DOUBLE VERTICAL AND LEFT
        SGML_MAP.put("boxvr",'\u251C'); // BOX DRAWINGS LIGHT VERTICAL AND RIGHT
        SGML_MAP.put("boxvR",'\u255E'); // BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE
        SGML_MAP.put("boxVr",'\u255F'); // BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE
        SGML_MAP.put("boxVR",'\u2560'); // BOX DRAWINGS DOUBLE VERTICAL AND RIGHT
        SGML_MAP.put("b.Phi",'\u03A6'); // GREEK CAPITAL LETTER PHI
        SGML_MAP.put("b.phis",'\u03C6'); // GREEK SMALL LETTER PHI
        SGML_MAP.put("b.phiv",'\u03D5'); // GREEK PHI SYMBOL
        SGML_MAP.put("b.Pi",'\u03A0'); // GREEK CAPITAL LETTER PI
        SGML_MAP.put("b.pi",'\u03C0'); // GREEK SMALL LETTER PI
        SGML_MAP.put("b.piv",'\u03D6'); // GREEK PI SYMBOL
        SGML_MAP.put("bprime",'\u2035'); // REVERSED PRIME
        SGML_MAP.put("b.Psi",'\u03A8'); // GREEK CAPITAL LETTER PSI
        SGML_MAP.put("b.psi",'\u03C8'); // GREEK SMALL LETTER PSI
        SGML_MAP.put("breve",'\u02D8'); // BREVE
        SGML_MAP.put("b.rho",'\u03C1'); // GREEK SMALL LETTER RHO
        SGML_MAP.put("b.rhov",'\u03F1'); // GREEK RHO SYMBOL
        SGML_MAP.put("brvbar",'\u00A6'); // BROKEN BAR
        SGML_MAP.put("b.Sigma",'\u03A3'); // GREEK CAPITAL LETTER SIGMA
        SGML_MAP.put("b.sigma",'\u03C3'); // GREEK SMALL LETTER SIGMA
        SGML_MAP.put("b.sigmav",'\u03C2'); // GREEK SMALL LETTER FINAL SIGMA
        SGML_MAP.put("bsim",'\u223D'); // REVERSED TILDE
        SGML_MAP.put("bsime",'\u22CD'); // REVERSED TILDE EQUALS
        // SGML_MAP.put("bsol","005C"); // REVERSE SOLIDUS
        SGML_MAP.put("b.tau",'\u03C4'); // GREEK SMALL LETTER TAU
        SGML_MAP.put("b.Theta",'\u0398'); // GREEK CAPITAL LETTER THETA
        SGML_MAP.put("b.thetas",'\u03B8'); // GREEK SMALL LETTER THETA
        SGML_MAP.put("b.thetav",'\u03D1'); // GREEK THETA SYMBOL
        SGML_MAP.put("bull",'\u2022'); // BULLET
        SGML_MAP.put("bump",'\u224E'); // GEOMETRICALLY EQUIVALENT TO
        SGML_MAP.put("bumpe",'\u224F'); // DIFFERENCE BETWEEN
        SGML_MAP.put("b.Upsi",'\u03A5'); // GREEK CAPITAL LETTER UPSILON
        SGML_MAP.put("b.upsi",'\u03C5'); // GREEK SMALL LETTER UPSILON
        SGML_MAP.put("b.Xi",'\u039E'); // GREEK CAPITAL LETTER XI
        SGML_MAP.put("b.xi",'\u03BE'); // GREEK SMALL LETTER XI
        SGML_MAP.put("b.zeta",'\u03B6'); // GREEK SMALL LETTER ZETA
        SGML_MAP.put("Cacute",'\u0106'); // LATIN CAPITAL LETTER C WITH ACUTE
        SGML_MAP.put("cacute",'\u0107'); // LATIN SMALL LETTER C WITH ACUTE
        SGML_MAP.put("Cap",'\u22D2'); // DOUBLE INTERSECTION
        SGML_MAP.put("cap",'\u2229'); // INTERSECTION
        SGML_MAP.put("caret",'\u2041'); // CARET INSERTION POINT
        SGML_MAP.put("caron",'\u02C7'); // CARON
        SGML_MAP.put("Ccaron",'\u010C'); // LATIN CAPITAL LETTER C WITH CARON
        SGML_MAP.put("ccaron",'\u010D'); // LATIN SMALL LETTER C WITH CARON
        SGML_MAP.put("Ccedil",'\u00C7'); // LATIN CAPITAL LETTER C WITH CEDILLA
        SGML_MAP.put("ccedil",'\u00E7'); // LATIN SMALL LETTER C WITH CEDILLA
        SGML_MAP.put("Ccirc",'\u0108'); // LATIN CAPITAL LETTER C WITH CIRCUMFLEX
        SGML_MAP.put("ccirc",'\u0109'); // LATIN SMALL LETTER C WITH CIRCUMFLEX
        SGML_MAP.put("Cdot",'\u010A'); // LATIN CAPITAL LETTER C WITH DOT ABOVE
        SGML_MAP.put("cdot",'\u010B'); // LATIN SMALL LETTER C WITH DOT ABOVE
        SGML_MAP.put("cedil",'\u00B8'); // CEDILLA
        SGML_MAP.put("cent",'\u00A2'); // CENT SIGN
        SGML_MAP.put("CHcy",'\u0427'); // CYRILLIC CAPITAL LETTER CHE
        SGML_MAP.put("chcy",'\u0447'); // CYRILLIC SMALL LETTER CHE
        SGML_MAP.put("check",'\u2713'); // CHECK MARK
        SGML_MAP.put("Chi",'\u03A7'); // GREEK CAPITAL LETTER CHI
        SGML_MAP.put("chi",'\u03C7'); // GREEK SMALL LETTER CHI
        SGML_MAP.put("cir",'\u25CB'); // WHITE CIRCLE
        SGML_MAP.put("circ",'\u02C6'); // MODIFIER LETTER CIRCUMFLEX ACCENT
        SGML_MAP.put("cire",'\u2257'); // RING EQUAL TO
        SGML_MAP.put("clubs",'\u2663'); // BLACK CLUB SUIT
        SGML_MAP.put("colon",'\u003A'); // COLON
        SGML_MAP.put("colone",'\u2254'); // COLON EQUALS
        SGML_MAP.put("comma",'\u002C'); // COMMA
        SGML_MAP.put("commat",'\u0040'); // COMMERCIAL AT
        SGML_MAP.put("comp",'\u2201'); // COMPLEMENT
        SGML_MAP.put("compfn",'\u2218'); // RING OPERATOR
        SGML_MAP.put("cong",'\u2245'); // APPROXIMATELY EQUAL TO
        SGML_MAP.put("conint",'\u222E'); // CONTOUR INTEGRAL
        SGML_MAP.put("coprod",'\u2210'); // N-ARY COPRODUCT
        SGML_MAP.put("copy",'\u00A9'); // COPYRIGHT SIGN
        SGML_MAP.put("copysr",'\u2117'); // SOUND RECORDING COPYRIGHT
        SGML_MAP.put("crarr",'\u21B5'); // DOWNWARDS ARROW WITH CORNER LEFTWARDS
        SGML_MAP.put("cross",'\u2717'); // BALLOT X
        SGML_MAP.put("cuepr",'\u22DE'); // EQUAL TO OR PRECEDES
        SGML_MAP.put("cuesc",'\u22DF'); // EQUAL TO OR SUCCEEDS
        SGML_MAP.put("cularr",'\u21B6'); // ANTICLOCKWISE TOP SEMICIRCLE ARROW
        SGML_MAP.put("Cup",'\u22D3'); // DOUBLE UNION
        SGML_MAP.put("cup",'\u222A'); // UNION
        SGML_MAP.put("cupre",'\u227C'); // PRECEDES OR EQUAL TO
        SGML_MAP.put("curarr",'\u21B7'); // CLOCKWISE TOP SEMICIRCLE ARROW
        SGML_MAP.put("curren",'\u00A4'); // CURRENCY SIGN
        SGML_MAP.put("cuvee",'\u22CE'); // CURLY LOGICAL OR
        SGML_MAP.put("cuwed",'\u22CF'); // CURLY LOGICAL AND
        SGML_MAP.put("dagger",'\u2020'); // DAGGER
        SGML_MAP.put("Dagger",'\u2021'); // DOUBLE DAGGER
        SGML_MAP.put("daleth",'\u2138'); // DALET SYMBOL
        SGML_MAP.put("dArr",'\u21D3'); // DOWNWARDS DOUBLE ARROW
        SGML_MAP.put("darr",'\u2193'); // DOWNWARDS ARROW
        SGML_MAP.put("darr2",'\u21CA'); // DOWNWARDS PAIRED ARROWS
        SGML_MAP.put("dash",'\u2010'); // HYPHEN
        SGML_MAP.put("dashv",'\u22A3'); // LEFT TACK
        SGML_MAP.put("dblac",'\u02DD'); // DOUBLE ACUTE ACCENT
        SGML_MAP.put("Dcaron",'\u010E'); // LATIN CAPITAL LETTER D WITH CARON
        SGML_MAP.put("dcaron",'\u010F'); // LATIN SMALL LETTER D WITH CARON
        SGML_MAP.put("Dcy",'\u0414'); // CYRILLIC CAPITAL LETTER DE
        SGML_MAP.put("dcy",'\u0434'); // CYRILLIC SMALL LETTER DE
        SGML_MAP.put("deg",'\u00B0'); // DEGREE SIGN
        SGML_MAP.put("Delta",'\u0394'); // GREEK CAPITAL LETTER DELTA
        SGML_MAP.put("delta",'\u03B4'); // GREEK SMALL LETTER DELTA
        SGML_MAP.put("Dgr",'\u0394'); // GREEK CAPITAL LETTER DELTA
        SGML_MAP.put("dgr",'\u03B4'); // GREEK SMALL LETTER DELTA
        SGML_MAP.put("dharl",'\u21C3'); // DOWNWARDS HARPOON WITH BARB LEFTWARDS
        SGML_MAP.put("dharr",'\u21C2'); // DOWNWARDS HARPOON WITH BARB RIGHTWARDS
        SGML_MAP.put("diam",'\u22C4'); // DIAMOND OPERATOR
        SGML_MAP.put("diams",'\u2666'); // BLACK DIAMOND SUIT
        SGML_MAP.put("die",'\u00A8'); // DIAERESIS
        SGML_MAP.put("divide",'\u00F7'); // DIVISION SIGN
        SGML_MAP.put("divonx",'\u22C7'); // DIVISION TIMES
        SGML_MAP.put("DJcy",'\u0402'); // CYRILLIC CAPITAL LETTER DJE
        SGML_MAP.put("djcy",'\u0452'); // CYRILLIC SMALL LETTER DJE
        SGML_MAP.put("dlarr",'\u2199'); // SOUTH WEST ARROW
        SGML_MAP.put("dlcorn",'\u231E'); // BOTTOM LEFT CORNER
        SGML_MAP.put("dlcrop",'\u230D'); // BOTTOM LEFT CROP
        SGML_MAP.put("dollar",'\u0024'); // DOLLAR SIGN
        SGML_MAP.put("dot",'\u02D9'); // DOT ABOVE
        SGML_MAP.put("Dot",'\u00A8'); // DIAERESIS
        SGML_MAP.put("DotDot",'\u20DC'); // COMBINING FOUR DOTS ABOVE
        SGML_MAP.put("drarr",'\u2198'); // SOUTH EAST ARROW
        SGML_MAP.put("drcorn",'\u231F'); // BOTTOM RIGHT CORNER
        SGML_MAP.put("drcrop",'\u230C'); // BOTTOM RIGHT CROP
        SGML_MAP.put("DScy",'\u0405'); // CYRILLIC CAPITAL LETTER DZE
        SGML_MAP.put("dscy",'\u0455'); // CYRILLIC SMALL LETTER DZE
        SGML_MAP.put("Dstrok",'\u0110'); // LATIN CAPITAL LETTER D WITH STROKE
        SGML_MAP.put("dstrok",'\u0111'); // LATIN SMALL LETTER D WITH STROKE
        SGML_MAP.put("dtri",'\u25BF'); // WHITE DOWN-POINTING SMALL TRIANGLE
        SGML_MAP.put("dtrif",'\u25BE'); // BLACK DOWN-POINTING SMALL TRIANGLE
        SGML_MAP.put("DZcy",'\u040F'); // CYRILLIC CAPITAL LETTER DZHE
        SGML_MAP.put("dzcy",'\u045F'); // CYRILLIC SMALL LETTER DZHE
        SGML_MAP.put("Eacgr",'\u0388'); // GREEK CAPITAL LETTER EPSILON WITH TONOS
        SGML_MAP.put("eacgr",'\u03AD'); // GREEK SMALL LETTER EPSILON WITH TONOS
        SGML_MAP.put("Eacute",'\u00C9'); // LATIN CAPITAL LETTER E WITH ACUTE
        SGML_MAP.put("eacute",'\u00E9'); // LATIN SMALL LETTER E WITH ACUTE
        SGML_MAP.put("Ecaron",'\u011A'); // LATIN CAPITAL LETTER E WITH CARON
        SGML_MAP.put("ecaron",'\u011B'); // LATIN SMALL LETTER E WITH CARON
        SGML_MAP.put("ecir",'\u2256'); // RING IN EQUAL TO
        SGML_MAP.put("Ecirc",'\u00CA'); // LATIN CAPITAL LETTER E WITH CIRCUMFLEX
        SGML_MAP.put("ecirc",'\u00EA'); // LATIN SMALL LETTER E WITH CIRCUMFLEX
        SGML_MAP.put("ecolon",'\u2255'); // EQUALS COLON
        SGML_MAP.put("Ecy",'\u042D'); // CYRILLIC CAPITAL LETTER E
        SGML_MAP.put("ecy",'\u044D'); // CYRILLIC SMALL LETTER E
        SGML_MAP.put("eDot",'\u2251'); // GEOMETRICALLY EQUAL TO
        SGML_MAP.put("Edot",'\u0116'); // LATIN CAPITAL LETTER E WITH DOT ABOVE
        SGML_MAP.put("edot",'\u0117'); // LATIN SMALL LETTER E WITH DOT ABOVE
        SGML_MAP.put("EEacgr",'\u0389'); // GREEK CAPITAL LETTER ETA WITH TONOS
        SGML_MAP.put("eeacgr",'\u03AE'); // GREEK SMALL LETTER ETA WITH TONOS
        SGML_MAP.put("EEgr",'\u0397'); // GREEK CAPITAL LETTER ETA
        SGML_MAP.put("eegr",'\u03B7'); // GREEK SMALL LETTER ETA
        SGML_MAP.put("efDot",'\u2252'); // APPROXIMATELY EQUAL TO OR THE IMAGE OF
        SGML_MAP.put("Egr",'\u0395'); // GREEK CAPITAL LETTER EPSILON
        SGML_MAP.put("egr",'\u03B5'); // GREEK SMALL LETTER EPSILON
        SGML_MAP.put("Egrave",'\u00C8'); // LATIN CAPITAL LETTER E WITH GRAVE
        SGML_MAP.put("egrave",'\u00E8'); // LATIN SMALL LETTER E WITH GRAVE
        SGML_MAP.put("egs",'\u22DD'); // EQUAL TO OR GREATER-THAN
        SGML_MAP.put("ell",'\u2113'); // SCRIPT SMALL L
        SGML_MAP.put("els",'\u22DC'); // EQUAL TO OR LESS-THAN
        SGML_MAP.put("Emacr",'\u0112'); // LATIN CAPITAL LETTER E WITH MACRON
        SGML_MAP.put("emacr",'\u0113'); // LATIN SMALL LETTER E WITH MACRON
        SGML_MAP.put("empty",'\u2205'); // EMPTY SET
        SGML_MAP.put("emsp",'\u2003'); // EM SPACE
        SGML_MAP.put("emsp13",'\u2004'); // THREE-PER-EM SPACE
        SGML_MAP.put("emsp14",'\u2005'); // FOUR-PER-EM SPACE
        SGML_MAP.put("ENG",'\u014A'); // LATIN CAPITAL LETTER ENG
        SGML_MAP.put("eng",'\u014B'); // LATIN SMALL LETTER ENG
        SGML_MAP.put("ensp",'\u2002'); // EN SPACE
        SGML_MAP.put("Eogon",'\u0118'); // LATIN CAPITAL LETTER E WITH OGONEK
        SGML_MAP.put("eogon",'\u0119'); // LATIN SMALL LETTER E WITH OGONEK
        SGML_MAP.put("epsi",'\u03B5'); // GREEK SMALL LETTER EPSILON
        SGML_MAP.put("Epsilon",'\u0395'); // GREEK CAPITAL LETTER EPSILON
        SGML_MAP.put("epsilon",'\u03B5'); // GREEK SMALL LETTER EPSILON
        SGML_MAP.put("epsis",'\u220A'); // SMALL ELEMENT OF
        // SGML_MAP.put("epsiv",'?'); // variant epsilon
        SGML_MAP.put("equals",'\u003D'); // EQUALS SIGN
        SGML_MAP.put("equiv",'\u2261'); // IDENTICAL TO
        SGML_MAP.put("erDot",'\u2253'); // IMAGE OF OR APPROXIMATELY EQUAL TO
        SGML_MAP.put("esdot",'\u2250'); // APPROACHES THE LIMIT
        SGML_MAP.put("Eta",'\u0397'); // GREEK CAPITAL LETTER ETA
        SGML_MAP.put("eta",'\u03B7'); // GREEK SMALL LETTER ETA
        SGML_MAP.put("ETH",'\u00D0'); // LATIN CAPITAL LETTER ETH
        SGML_MAP.put("eth",'\u00F0'); // LATIN SMALL LETTER ETH