pcbrf.c

harvest是一个下载html网页得机器人

struct parse pcbcone = {"CONE", lexcnm, conetab, 0, 0, 0, 0};
#undef ET2
#undef ER2
#undef CR2
#undef RS2
#undef ME2
#undef ME3
#undef ES2
#undef EE2
#undef NE2
#undef MD2
#undef MC2
#undef SC2
#undef SP2
#undef SR2
#undef TB2
/* PCBCONR: State and action table for content parse of replaceable character
            data.  Initial state assumes a start-tag was just processed.
            Only entity references and character references are recognized.
*/
/* Symbols for state names (end with a number). */
#define ET4     0   /* Markup found or buffer flushed; no data. */
#define DA4     2   /* Data in buffer. */
#define ER4     4   /* ERO found; start lookahead buffer. */
#define CR4     6   /* CRO found (ER2, RNI). */
#define ES4     8   /* TAGO found; start lookahead buffer. */
#define EE4    10   /* End-tag start (TAGO,ETI); move to lookahead buffer. */
#define NE4    12   /* End-tag start (TAGO,NET); process NET if not end-tag. */

static UNCH
/*      free nu   nmc  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
        nmre com  eti  net  mdo  msc  mso  pero pio  rni  tagc tago  */
et4 []={DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,DA4 ,DA4 ,ET4 ,ER4 ,
        ET4 ,DA4 ,DA4 ,ET4 ,DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,ES4 },/*et4*/
et4a[]={DAS_,DAS_,DAS_,DAS_,DAS_,NON_,EE_ ,GET_,RS_ ,REF_,DAS_,DAS_,NSC_,LAS_,
        REF_,DAS_,DAS_,NED_,DAS_,DAS_,DAS_,DAS_,DAS_,DAS_,DAS_,LAS_},

da4 []={DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,DA4 ,DA4 ,ET4 ,ET4 ,
        ET4 ,DA4 ,DA4 ,ET4 ,DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,DA4 ,ET4 },/*da4*/
da4a[]={NOP_,NOP_,NOP_,NOP_,NOP_,DAF_,DAF_,DAF_,DAF_,DAF_,NOP_,NOP_,DAF_,DAF_,
        DAF_,NOP_,NOP_,DAF_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,DAF_},

er4 []={ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ER4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,
        ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,CR4 ,ET4 ,ET4 },/*er4*/
er4a[]={LAF_,LAF_,LAF_,ERX_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAM_,LAF_,LAF_},

cr4 []={ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,CR4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,
        ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 },/*cr4*/
cr4a[]={LAF_,CRN_,LAF_,CRA_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_},

/*      free nu   nmc  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
        nmre com  eti  net  mdo  msc  mso  pero pio  rni  tagc tago  */
es4 []={ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ES4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,
        ET4 ,ET4 ,EE4 ,NE4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 },/*es4*/
es4a[]={LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAF_,LAM_,LAM_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_},

ee4 []={ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,EE4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,
        ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 },/*ee4*/
ee4a[]={LAF_,LAF_,LAF_,ETC_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,NET_,LAF_},

ne4 []={EE4 ,EE4 ,EE4 ,ET4 ,EE4 ,EE4 ,EE4 ,NE4 ,EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,
        EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,EE4 ,ET4 ,EE4 },/*ne4*/
ne4a[]={RC2_,RC2_,RC2_,ETC_,RC2_,RC2_,RC2_,GET_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,
        RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,NET_,RC2_},

*conrtab[] = {et4, et4a, da4, da4a, er4, er4a, cr4, cr4a,
              es4, es4a, ee4, ee4a, ne4, ne4a};
struct parse pcbconr = {"CONR", lexcon, conrtab, 0, 0, 0, 0};
#undef ET4
#undef DA4
#undef ER4
#undef CR4
#undef ES4
#undef EE4
#undef NE4
/* PCBCONC: State and action table for content parse of character data.
            Initial state assumes a start-tag was just processed.
*/
/* Symbols for state names (end with a number). */
#define ET6     0   /* Markup found or buffer flushed; no data. */
#define DA6     2   /* Data in buffer. */
#define ES6     4   /* TAGO found; start lookahead buffer. */
#define EE6     6   /* End-tag start (TAGO,ETI); move to lookahead buffer. */
#define NE6     8   /* End-tag start (TAGO,NET); process NET if not end-tag. */

static UNCH
/*      free nu   nmc  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
        nmre com  eti  net  mdo  msc  mso  pero pio  rni  tagc tago  */
et6 []={DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,DA6 ,DA6 ,ET6 ,DA6 ,
        ET6 ,DA6 ,DA6 ,ET6 ,DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,ES6 },/*et6*/
et6a[]={DAS_,DAS_,DAS_,DAS_,DAS_,NON_,EOF_,GET_,RS_ ,REF_,DAS_,DAS_,NSC_,DAS_,
        REF_,DAS_,DAS_,NED_,DAS_,DAS_,DAS_,DAS_,DAS_,DAS_,DAS_,LAS_},

da6 []={DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,DA6 ,DA6 ,ET6 ,ET6 ,
        ET6 ,DA6 ,DA6 ,ET6 ,DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,DA6 ,ET6 },/*da6*/
da6a[]={NOP_,NOP_,NOP_,NOP_,NOP_,DAF_,DAF_,DAF_,DAF_,DAF_,NOP_,NOP_,DAF_,DAF_,
        DAF_,NOP_,NOP_,DAF_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,DAF_},

/*      free nu   nmc  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
        nmre com  eti  net  mdo  msc  mso  pero pio  rni  tagc tago  */
es6 []={ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ES6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,
        ET6 ,ET6 ,EE6 ,NE6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 },/*es6*/
es6a[]={LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAF_,LAM_,LAM_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_},

ee6 []={ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,EE6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,
        ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 ,ET6 },/*ee6*/
ee6a[]={LAF_,LAF_,LAF_,ETC_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,NET_,LAF_},

ne6 []={EE6 ,EE6 ,EE6 ,ET6 ,EE6 ,EE6 ,EE6 ,NE6 ,EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,
        EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,EE6 ,ET6 ,EE6 },/*ne6*/
ne6a[]={RC2_,RC2_,RC2_,ETC_,RC2_,RC2_,RC2_,GET_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,
        RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,RC2_,NET_,RC2_},

*conctab[] = {et6, et6a, da6, da6a, es6, es6a, ee6, ee6a, ne6, ne6a};
struct parse pcbconc = {"CONC", lexcon, conctab, 0, 0, 0, 0};
#undef ET6
#undef DA6
#undef ES6
#undef EE6
#undef NE6
/* PCBPRO: State and action table for prolog parse.
           Initial state assumes document just began.
*/
/* Symbols for state names (end with a number). */
#define ET7     0   /* Markup found. */
#define ES7     2   /* TAGO found; start lookahead buffer. */
#define MD7     4   /* MDO found (TAGO, MDO[2]). */
#define MC7     6   /* MDO, COM found. */
#define EE7     8   /* TAGO, ETI found */

static UNCH
/*      free nu   nmc  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
        nmre com  eti  net  mdo  msc  mso  pero pio  rni  tagc tago */
et7 []={ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,
        ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ES7 },/*et7*/
et7a[]={DCE_,DCE_,DCE_,DCE_,NOP_,DCE_,EE_ ,GET_,RS_ ,NOP_,NOP_,DCE_,DCE_,DCE_,
        DCE_,DCE_,DCE_,DCE_,DCE_,DCE_,DCE_,DCE_,DCE_,DCE_,DCE_,LAS_},

es7 []={ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ES7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,
        ET7 ,ET7 ,EE7 ,ET7 ,MD7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 },/*es7*/
es7a[]={PEP_,PEP_,PEP_,STE_,PEP_,PEP_,PEP_,GET_,PEP_,PEP_,PEP_,PEP_,PEP_,PEP_,
        PEP_,PEP_,LAM_,PEP_,LAM_,PEP_,PEP_,PEP_,PIS_,PEP_,STE_,PEP_},

md7 []={ET7, ET7, ET7, ET7, ET7 ,ET7, ET7, MD7, ET7 ,ET7 ,ET7 ,ET7, ET7, ET7,
        ET7, MC7, ET7, ET7, ET7, ET7 ,ET7, ET7, ET7, ET7 ,ET7, ET7 },/*md7*/
md7a[]={LAF_,LAF_,LAF_,DTD_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAM_,LAF_,LAF_,LAF_,LAF_,MSP_,LAF_,LAF_,LAF_,NOP_,LAF_},

mc7 []={ET7, ET7, ET7, ET7, ET7, ET7 ,ET7, MC7, ET7 ,ET7, ET7 ,ET7, ET7, ET7,
        ET7, ET7, ET7, ET7, ET7, ET7 ,ET7 ,ET7, ET7 ,ET7 ,ET7, ET7 },/*mc7*/
mc7a[]={LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,MDC_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_},

ee7 []={ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,EE7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,
        ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 ,ET7 },/*ee7*/
ee7a[]={LAF_,LAF_,LAF_,ETE_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
        LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,ETE_,LAF_},

*protab[] = {et7, et7a, es7, es7a, md7, md7a, mc7, mc7a, ee7, ee7a};
struct parse pcbpro = {"PRO", lexcon, protab, 0, 0, 0, 0};
#undef ET7
#undef ES7
#undef MD7
#undef MC7
#undef EE7
/* PCBMDS: State and action table for parse of markup declaration subset.
           Initial state assumes subset just began (MSO found).
*/
/* Symbols for state names (end with a number). */
#define ET8     0   /* Markup found. */
#define ER8     2   /* PERO found; start lookahead buffer. */
#define ME8     4   /* MSC found. */
#define ME9     6   /* MSC, MSC found. */
#define ES8     8   /* TAGO found; start lookahead buffer. */
#define MD8    10   /* MDO found (TAGO, MDO[2]). */
#define MC8    12   /* MDO, CD found. */
#define DC8    14   /* Data characters found (erroneously). */

static UNCH
/*      free nu   nmc  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
        nmre com  eti  net  mdo  msc  mso  pero pio  rni  tagc tago */
et8 []={DC8 ,DC8 ,DC8 ,DC8 ,ET8 ,DC8 ,ET8 ,ET8 ,ET8 ,ET8 ,ET8 ,DC8 ,DC8 ,DC8 ,
        DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,ME8 ,DC8 ,ER8 ,DC8 ,DC8 ,DC8 ,ES8 },/*et8*/
et8a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,GET_,GET_,RS_ ,NOP_,NOP_,NOP_,SYS_,NOP_,
        NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

er8 []={DC8 ,DC8 ,DC8 ,ET8 ,DC8 ,DC8 ,DC8 ,ER8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,
        DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 },/*er8*/
er8a[]={NOP_,NOP_,NOP_,PER_,NOP_,SYS_,NOP_,GET_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,
        NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

me8 []={ET8, ET8, ET8, ET8, ET8 ,ET8, ET8, ME8, ET8 ,ET8 ,ET8 ,ET8, ET8, ET8,
        ET8 ,ET8, ET8 ,ET8, ET8, ME9 ,ET8, ET8, ET8, ET8 ,ET8, ET8 },/*me8*/
me8a[]={DTE_,DTE_,DTE_,DTE_,DTE_,DTE_,DTE_,GET_,DTE_,DTE_,DTE_,DTE_,DTE_,DTE_,
        DTE_,DTE_,DTE_,DTE_,DTE_,NOP_,DTE_,DTE_,DTE_,DTE_,DTE_,DTE_},

me9 []={DC8, DC8, DC8, DC8, DC8 ,DC8, DC8, ME9, DC8 ,DC8 ,DC8 ,DC8, DC8, DC8,
        DC8 ,DC8, DC8 ,DC8, DC8, DC8 ,DC8, DC8, DC8, DC8 ,ET8, DC8 },/*me9*/
me9a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,GET_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,
        NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,MSE_,NOP_},

/*      free nu   nmc  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
        nmre com  eti  net  mdo  msc  mso  pero pio  rni  tagc tago */
es8 []={DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,ES8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,
        DC8 ,DC8 ,DC8 ,DC8 ,MD8 ,DC8 ,DC8 ,DC8 ,ET8 ,DC8 ,DC8 ,DC8 },/*es8*/
es8a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,GET_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,
        NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,PIS_,NOP_,NOP_,NOP_},

md8 []={DC8, DC8, DC8, ET8, DC8 ,DC8, DC8, MD8, DC8 ,DC8 ,DC8 ,DC8, DC8, DC8,
        DC8 ,MC8, DC8 ,DC8, DC8, DC8 ,ET8, DC8, DC8, DC8 ,ET8, DC8 },/*md8*/
md8a[]={NOP_,NOP_,NOP_,MD_ ,NOP_,SYS_,NOP_,GET_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,
        NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,MSS_,NOP_,NOP_,NOP_,NOP_,NOP_},

mc8 []={DC8, DC8, DC8, DC8, DC8, DC8 ,DC8, MC8, DC8 ,DC8, DC8 ,DC8, DC8, DC8,
        DC8 ,ET8, DC8 ,DC8, DC8, DC8 ,DC8 ,DC8, DC8 ,DC8 ,DC8, DC8 },/*mc8*/
mc8a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,GET_,NOP_,NOP_,NOP_,NOP_,SYS_,NOP_,
        NOP_,MDC_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

dc8 []={DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,
        DC8 ,DC8 ,DC8 ,DC8 ,DC8 ,ET8 ,DC8 ,ET8 ,DC8 ,DC8 ,DC8 ,ET8 },/*dc8*/
dc8a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,GET_,GET_,RS_ ,NOP_,NOP_,NOP_,SYS_,NOP_,
        NOP_,NOP_,NOP_,NOP_,NOP_,CIR_,NOP_,CIR_,NOP_,NOP_,NOP_,CIR_},

*mdstab[] = {et8, et8a, er8, er8a, me8, me8a, me9, me9a,
             es8, es8a, md8, md8a, mc8, mc8a, dc8, dc8a};
struct parse pcbmds = {"MDS", lexcon, mdstab, 0, 0, 0, 0};
#undef ET8
#undef ER8
#undef ME8
#undef ME9
#undef ES8
#undef MD8
#undef MC8
#undef DC8
/* PCBGRCM: State and action table for content model group.
            Groups can nest.  Reserved names are allowed.
            Data tag token groups are allowed.
            A non-reserved name or model group can have a suffix.
            Columns are based on LEXGRP.C.
*/
/* Symbols for state names (end with a number). */
#define TK1     0   /* Token expected: name, #CHARS, data tag grp, model. */
#define CO1     2   /* Connector between tokens expected. */
#define ER1     4   /* PERO found when token was expected. */
#define SP1     6   /* Name or model: suffix or connector expected. */
#define RN1     8   /* RNI found; possible #PCDATA. */
#define DG1    10   /* Data tag: group begun; name expected. */
#define DN1    12   /* Data tag: name found; SEQ connector expected. */
#define DT1    14   /* Data tag: ignore template and pattern; MSC expected. */
#define DR1    16   /* PERO found when data tag name was expected. */
#define LI1    18   /* Literal in data tag group; search for LIT. */
#define LA1    20   /* Literal in data tag group; search for LITA. */