oed.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

#include <u.h>
#include <libc.h>
#include <bio.h>
#include "dict.h"

enum {
	Buflen=1000,
	Maxaux=5,
};

/* Possible tags */
enum {
	A,		/* author in quote (small caps) */
	B,		/* bold */
	Ba,		/* author inside bib */
	Bch,		/* builtup chem component */
	Bib,		/* surrounds word 'in' for bibliographic ref */
	Bl,		/* bold */
	Bo,		/* bond over */
	Bu,		/* bond under */
	Cb,		/* ? block of stuff (indent) */
	Cf,		/* cross ref to another entry (italics) */
	Chem,		/* chemistry formula */
	Co,		/* over (preceding sum, integral, etc.) */
	Col,		/* column of table (aux just may be r) */
	Cu,		/* under (preceding sum, integral, etc.) */
	Dat,		/* date */
	Db,		/* def block? indent */
	Dn,		/* denominator of fraction */
	E,		/* main entry */
	Ed,		/* editor's comments (in [...]) */
	Etym,		/* etymology (in [...]) */
	Fq,		/* frequency count (superscript) */
	Form,		/* formula */
	Fr,		/* fraction (contains <nu>, then <dn>) */
	Gk,		/* greek (transliteration) */
	Gr,		/* grammar? (e.g., around 'pa.' in 'pa. pple.') */
	Hg,		/* headword group */
	Hm,		/* homonym (superscript) */
	Hw,		/* headword (bold) */
	I,		/* italics */
	Il,		/* italic list? */
	In,		/* inferior (subscript) */
	L,		/* row of col of table */
	La,		/* status or usage label (italic) */
	Lc,		/* chapter/verse sort of thing for works */
	N,		/* note (smaller type) */
	Nu,		/* numerator of fraction */
	Ov,		/* needs overline */
	P,		/* paragraph (indent) */
	Ph,		/* pronunciation (transliteration) */
	Pi,		/* pile (frac without line) */
	Pqp,		/* subblock of quote */
	Pr,		/* pronunciation (in (...)) */
	Ps,		/* position (e.g., adv.) (italic) */
	Pt,		/* part (in lc) */
	Q,		/* quote in quote block */
	Qd,		/* quote date (bold) */
	Qig,		/* quote number (greek) */
	Qla,		/* status or usage label in quote (italic) */
	Qp,		/* quote block (small type, indent) */
	Qsn,		/* quote number */
	Qt,		/* quote words */
	R,		/* roman type style */
	Rx,		/* relative cross reference (e.g., next) */
	S,		/* another form? (italic) */
	S0,		/* sense (sometimes surrounds several sx's) */
	S1,		/* sense (aux num: indented bold letter) */
	S2,		/* sense (aux num: indented bold capital rom num) */
	S3,		/* sense (aux num: indented number of asterisks) */
	S4,		/* sense (aux num: indented bold number) */
	S5,		/* sense (aux num: indented number of asterisks) */
	S6,		/* subsense (aux num: bold letter) */
	S7a,		/* subsense (aux num: letter) */
	S7n,		/* subsense (aux num: roman numeral) */
	Sc,		/* small caps */
	Sgk,		/* subsense (aux num: transliterated greek) */
	Sn,		/* sense of subdefinition (aux num: roman letter) */
	Ss,		/* sans serif */
	Ssb,		/* sans serif bold */
	Ssi,		/* sans serif italic */
	Su,		/* superior (superscript) */
	Sub,		/* subdefinition */
	Table,		/* table (aux cols=number of columns) */
	Tt,		/* title? (italics) */
	Vd,		/* numeric label for variant form */
	Ve,		/* variant entry */
	Vf,		/* variant form (light bold) */
	Vfl,		/* list of vf's (starts with Also or Forms) */
	W,		/* work (e.g., Beowulf) (italics) */
	X,		/* cross reference to main word (small caps) */
	Xd,		/* cross reference to quotation by date */
	Xi,		/* internal cross reference ? (italic) */
	Xid,		/* cross reference identifer, in quote ? */
	Xs,		/* cross reference sense (lower number) */
	Xr,		/* list of x's */
	Ntag		/* end of tags */
};

/* Assoc tables must be sorted on first field */

static Assoc tagtab[] = {
	{"a",		A},
	{"b",		B},
	{"ba",		Ba},
	{"bch",		Bch},
	{"bib",		Bib},
	{"bl",		Bl},
	{"bo",		Bo},
	{"bu",		Bu},
	{"cb",		Cb},
	{"cf",		Cf},
	{"chem",	Chem},
	{"co",		Co},
	{"col",		Col},
	{"cu",		Cu},
	{"dat",		Dat},
	{"db",		Db},
	{"dn",		Dn},
	{"e",		E},
	{"ed",		Ed},
	{"et",		Etym},
	{"etym",	Etym},
	{"form",	Form},
	{"fq",		Fq},
	{"fr",		Fr},
	{"frac",	Fr},
	{"gk",		Gk},
	{"gr",		Gr},
	{"hg",		Hg},
	{"hm",		Hm},
	{"hw",		Hw},
	{"i",		I},
	{"il",		Il},
	{"in",		In},
	{"l",		L},
	{"la",		La},
	{"lc",		Lc},
	{"n",		N},
	{"nu",		Nu},
	{"ov",		Ov},
	{"p",		P},
	{"ph",		Ph},
	{"pi",		Pi},
	{"pqp",		Pqp},
	{"pr",		Pr},
	{"ps",		Ps},
	{"pt",		Pt},
	{"q",		Q},
	{"qd",		Qd},
	{"qig",		Qig},
	{"qla",		Qla},
	{"qp",		Qp},
	{"qsn",		Qsn},
	{"qt",		Qt},
	{"r",		R},
	{"rx",		Rx},
	{"s",		S},
	{"s0",		S0},
	{"s1",		S1},
	{"s2",		S2},
	{"s3",		S3},
	{"s4",		S4},
	{"s5",		S5},
	{"s6",		S6},
	{"s7a",		S7a},
	{"s7n",		S7n},
	{"sc",		Sc},
	{"sgk",		Sgk},
	{"sn",		Sn},
	{"ss",		Ss,},
	{"ssb",		Ssb},
	{"ssi",		Ssi},
	{"su",		Su},
	{"sub",		Sub},
	{"table",	Table},
	{"tt",		Tt},
	{"vd",		Vd},
	{"ve",		Ve},
	{"vf",		Vf},
	{"vfl",		Vfl},
	{"w",		W},
	{"x",		X},
	{"xd",		Xd},
	{"xi",		Xi},
	{"xid",		Xid},
	{"xr",		Xr},
	{"xs",		Xs},
};

/* Possible tag auxilliary info */
enum {
	Cols,		/* number of columns in a table */
	Num,		/* letter or number, for a sense */
	St,		/* status (e.g., obs) */
	Naux
};

static Assoc auxtab[] = {
	{"cols",	Cols},
	{"num",		Num},
	{"st",		St}
};

static Assoc spectab[] = {
	{"3on4",	L'¾'},
	{"Aacu",	L'Á'},
	{"Aang",	L'Å'},
	{"Abarab",	L'Ā'},
	{"Acirc",	L'Â'},
	{"Ae",		L'Æ'},
	{"Agrave",	L'À'},
	{"Alpha",	L'Α'},
	{"Amac",	L'Ā'},
	{"Asg",		L'Ʒ'},		/* Unicyle. Cf "Sake" */
	{"Auml",	L'Ä'},
	{"Beta",	L'Β'},
	{"Cced",	L'Ç'},
	{"Chacek",	L'Č'},
	{"Chi",		L'Χ'},
	{"Chirho",	L'☧'},		/* Chi Rho U+2627 */
	{"Csigma",	L'Ϛ'},
	{"Delta",	L'Δ'},
	{"Eacu",	L'É'},
	{"Ecirc",	L'Ê'},
	{"Edh",		L'Ð'},
	{"Epsilon",	L'Ε'},
	{"Eta",		L'Η'},
	{"Gamma",	L'Γ'},
	{"Iacu",	L'Í'},
	{"Icirc",	L'Î'},
	{"Imac",	L'Ī'},
	{"Integ",	L'∫'},
	{"Iota",	L'Ι'},
	{"Kappa",	L'Κ'},
	{"Koppa",	L'Ϟ'},
	{"Lambda",	L'Λ'},
	{"Lbar",	L'Ł'},
	{"Mu",		L'Μ'},
	{"Naira",	L'N'},		/* should have bar through */
	{"Nplus",	L'N'},		/* should have plus above */
	{"Ntilde",	L'Ñ'},
	{"Nu",		L'Ν'},
	{"Oacu",	L'Ó'},
	{"Obar",	L'Ø'},
	{"Ocirc",	L'Ô'},
	{"Oe",		L'Œ'},
	{"Omega",	L'Ω'},
	{"Omicron",	L'Ο'},
	{"Ouml",	L'Ö'},
	{"Phi",		L'Φ'},
	{"Pi",		L'Π'},
	{"Psi",		L'Ψ'},
	{"Rho",		L'Ρ'},
	{"Sacu",	L'Ś'},
	{"Sigma",	L'Σ'},
	{"Summ",	L'∑'},
	{"Tau",		L'Τ'},
	{"Th",		L'Þ'},
	{"Theta",	L'Θ'},
	{"Tse",		L'Ц'},
	{"Uacu",	L'Ú'},
	{"Ucirc",	L'Û'},
	{"Upsilon",	L'Υ'},
	{"Uuml",	L'Ü'},
	{"Wyn",		L'ƿ'},		/* wynn U+01BF */
	{"Xi",		L'Ξ'},
	{"Ygh",		L'Ʒ'},		/* Yogh	U+01B7 */
	{"Zeta",	L'Ζ'},
	{"Zh",		L'Ʒ'},		/* looks like Yogh. Cf "Sake" */
	{"a",		L'a'},		/* ante */
	{"aacu",	L'á'},
	{"aang",	L'å'},
	{"aasper",	MAAS},
	{"abreve",	L'ă'},
	{"acirc",	L'â'},
	{"acu",		LACU},
	{"ae",		L'æ'},
	{"agrave",	L'à'},
	{"ahook",	L'ą'},
	{"alenis",	MALN},
	{"alpha",	L'α'},
	{"amac",	L'ā'},
	{"amp",		L'&'},
	{"and",		MAND},
	{"ang",		LRNG},
	{"angle",	L'∠'},
	{"ankh",	L'☥'},		/* ankh U+2625 */
	{"ante",	L'a'},		/* before (year) */
	{"aonq",	MAOQ},
	{"appreq",	L'≃'},
	{"aquar",	L'♒'},
	{"arDadfull",	L'ض'},		/* Dad U+0636 */
	{"arHa",	L'ح'},		/* haa U+062D */
	{"arTa",	L'ت'},		/* taa U+062A */
	{"arain",	L'ع'},		/* ain U+0639 */
	{"arainfull",	L'ع'},		/* ain U+0639 */
	{"aralif",	L'ا'},		/* alef U+0627 */
	{"arba",	L'ب'},		/* baa U+0628 */
	{"arha",	L'ه'},		/* ha U+0647 */
	{"aries",	L'♈'},
	{"arnun",	L'ن'},		/* noon U+0646 */
	{"arnunfull",	L'ن'},		/* noon U+0646 */
	{"arpa",	L'ه'},		/* ha U+0647 */
	{"arqoph",	L'ق'},		/* qaf U+0642 */
	{"arshinfull",	L'ش'},		/* sheen U+0634 */
	{"arta",	L'ت'},		/* taa U+062A */
	{"artafull",	L'ت'},		/* taa U+062A */
	{"artha",	L'ث'},		/* thaa U+062B */
	{"arwaw",	L'و'},		/* waw U+0648 */
	{"arya",	L'ي'},		/* ya U+064A */
	{"aryafull",	L'ي'},		/* ya U+064A */
	{"arzero",	L'٠'},		/* indic zero U+0660 */
	{"asg",		L'ʒ'},		/* unicycle character. Cf "hallow" */
	{"asper",	LASP},
	{"assert",	L'⊢'},
	{"astm",	L'⁂'},		/* asterism: should be upside down */
	{"at",		L'@'},
	{"atilde",	L'ã'},
	{"auml",	L'ä'},
	{"ayin",	L'ع'},		/* arabic ain U+0639 */
	{"b1",		L'-'},		/* single bond */
	{"b2",		L'='},		/* double bond */
	{"b3",		L'≡'},		/* triple bond */
	{"bbar",	L'ƀ'},		/* b with bar U+0180 */
	{"beta",	L'β'},
	{"bigobl",	L'/'},
	{"blC",		L'C'},		/* should be black letter */
	{"blJ",		L'J'},		/* should be black letter */
	{"blU",		L'U'},		/* should be black letter */
	{"blb",		L'b'},		/* should be black letter */
	{"blozenge",	L'◊'},		/* U+25CA; should be black */
	{"bly",		L'y'},		/* should be black letter */
	{"bra",		MBRA},
	{"brbl",	LBRB},
	{"breve",	LBRV},
	{"bslash",	L'\\'},
	{"bsquare",	L'■'},		/* black square U+25A0 */
	{"btril",	L'◀'},		/* U+25C0 */
	{"btrir",	L'▶'},		/* U+25B6 */
	{"c",		L'c'},		/* circa */
	{"cab",		L'〉'},
	{"cacu",	L'ć'},
	{"canc",	L'♋'},
	{"capr",	L'♑'},
	{"caret",	L'^'},
	{"cb",		L'}'},
	{"cbigb",	L'}'},
	{"cbigpren",	L')'},
	{"cbigsb",	L']'},
	{"cced",	L'ç'},
	{"cdil",	LCED},
	{"cdsb",	L'〛'},		/* ]] U+301b */
	{"cent",	L'¢'},
	{"chacek",	L'č'},
	{"chi",		L'χ'},
	{"circ",	LRNG},
	{"circa",	L'c'},		/* about (year) */
	{"circbl",	L'̥'},		/* ring below accent U+0325 */
	{"circle",	L'○'},		/* U+25CB */
	{"circledot",	L'⊙'},
	{"click",	L'ʖ'},
	{"club",	L'♣'},
	{"comtime",	L'C'},
	{"conj",	L'☌'},
	{"cprt",	L'©'},
	{"cq",		L'\''},
	{"cqq",		L'”'},
	{"cross",	L'✠'},		/* maltese cross U+2720 */
	{"crotchet",	L'♩'},
	{"csb",		L']'},
	{"ctilde",	L'c'},		/* +tilde */
	{"ctlig",	MLCT},
	{"cyra",	L'а'},
	{"cyre",	L'е'},
	{"cyrhard",	L'ъ'},
	{"cyrjat",	L'ѣ'},
	{"cyrm",	L'м'},
	{"cyrn",	L'н'},
	{"cyrr",	L'р'},
	{"cyrsoft",	L'ь'},
	{"cyrt",	L'т'},
	{"cyry",	L'ы'},
	{"dag",		L'†'},
	{"dbar",	L'đ'},
	{"dblar",	L'⇋'},
	{"dblgt",	L'≫'},
	{"dbllt",	L'≪'},
	{"dced",	L'd'},		/* +cedilla */
	{"dd",		MDD},
	{"ddag",	L'‡'},
	{"ddd",		MDDD},
	{"decr",	L'↓'},
	{"deg",		L'°'},
	{"dele",	L'd'},		/* should be dele */
	{"delta",	L'δ'},
	{"descnode",	L'☋'},		/* descending node U+260B */
	{"diamond",	L'♢'},
	{"digamma",	L'ϝ'},
	{"div",		L'÷'},
	{"dlessi",	L'ı'},
	{"dlessj1",	L'j'},		/* should be dotless */
	{"dlessj2",	L'j'},		/* should be dotless */
	{"dlessj3",	L'j'},		/* should be dotless */
	{"dollar",	L'$'},
	{"dotab",	LDOT},
	{"dotbl",	LDTB},
	{"drachm",	L'ʒ'},
	{"dubh",	L'-'},
	{"eacu",	L'é'},
	{"earth",	L'♁'},
	{"easper",	MEAS},
	{"ebreve",	L'ĕ'},
	{"ecirc",	L'ê'},
	{"edh",		L'ð'},
	{"egrave",	L'è'},
	{"ehacek",	L'ě'},
	{"ehook",	L'ę'},
	{"elem",	L'∊'},
	{"elenis",	MELN},
	{"em",		L'—'},
	{"emac",	L'ē'},
	{"emem",	MEMM},
	{"en",		L'–'},
	{"epsilon",	L'ε'},
	{"equil",	L'⇋'},
	{"ergo",	L'∴'},
	{"es",		MES},
	{"eszett",	L'ß'},
	{"eta",		L'η'},
	{"eth",		L'ð'},
	{"euml",	L'ë'},
	{"expon",	L'↑'},
	{"fact",	L'!'},
	{"fata",	L'ɑ'},
	{"fatpara",	L'¶'},		/* should have fatter, filled in bowl */
	{"female",	L'♀'},
	{"ffilig",	MLFFI},
	{"fflig",	MLFF},
	{"ffllig",	MLFFL},
	{"filig",	MLFI},
	{"flat",	L'♭'},
	{"fllig",	MLFL},
	{"frE",		L'E'},		/* should be curly */
	{"frL",		L'L'},		/* should be curly */
	{"frR",		L'R'},		/* should be curly */
	{"frakB",	L'B'},		/* should have fraktur style */
	{"frakG",	L'G'},
	{"frakH",	L'H'},
	{"frakI",	L'I'},
	{"frakM",	L'M'},
	{"frakU",	L'U'},
	{"frakX",	L'X'},
	{"frakY",	L'Y'},
	{"frakh",	L'h'},
	{"frbl",	LFRB},
	{"frown",	LFRN},
	{"fs",		L' '},
	{"fsigma",	L'ς'},
	{"gAacu",	L'Á'},		/* should be Α+acute */
	{"gaacu",	L'α'},		/* +acute */
	{"gabreve",	L'α'},		/* +breve */
	{"gafrown",	L'α'},		/* +frown */
	{"gagrave",	L'α'},		/* +grave */
	{"gamac",	L'α'},		/* +macron */
	{"gamma",	L'γ'},
	{"gauml",	L'α'},		/* +umlaut */
	{"ge",		L'≧'},
	{"geacu",	L'ε'},		/* +acute */
	{"gegrave",	L'ε'},		/* +grave */
	{"ghacu",	L'η'},		/* +acute */
	{"ghfrown",	L'η'},		/* +frown */
	{"ghgrave",	L'η'},		/* +grave */
	{"ghmac",	L'η'},		/* +macron */
	{"giacu",	L'ι'},		/* +acute */
	{"gibreve",	L'ι'},		/* +breve */