pcbrf.c

harvest是一个下载html网页得机器人

ic23 []={IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC3 ,NC1 ,IC3 ,IC2 ,NC1 ,IC2 ,IC2 ,IC2 ,IC2 ,
         IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 },/*ic3*/
ic23a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

/*       bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */
li21 []={LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,NC1 ,LI1 ,
         LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 ,LI1 },/*li1*/
li21a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

la21 []={LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,NC1 ,
         LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 ,LA1 },/*la1*/
la21a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

*mditab[] = {nc21, nc21a, ic21, ic21a, ic22, ic22a,
             ic23, ic23a, li21, li21a, la21, la21a};
struct parse pcbmdi = {"MDI", lexmark, mditab, 0, 0, 0, 0};
#undef NC1
#undef IC1
#undef IC2
#undef IC3
#undef LI1
#undef LA1
/* PCBMSRC: State and action table for marked section in RCDATA mode.
            Nested marked sections are not recognized; the first MSE ends it.
            Initial state assumes an MS declaration was processed.
            Columns are based on LEXLMS.C but LITC column needn't exist.
*/
/* Symbols for state names (end with a number). */
#define ET0     0   /* MSS processed or buffer flushed; no data. */
#define DA0     2   /* Data in buffer. */
#define ER0     4   /* ERO found; start lookahead buffer. */
#define CR0     6   /* CRO found (ER0, RNI). */
#define ME0     8   /* MSC found. */
#define ME1    10   /* MSC, MSC found. */

static UNCH
/*       free nu   min  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
         mdo  msc  mso  pero rni  tagc tago litc */
et30 []={DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,DA0 ,DA0 ,ET0 ,ER0 ,
         DA0 ,ME0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 },/*et0*/
et30a[]={DAS_,DAS_,DAS_,DAS_,DAS_,NON_,EE_ ,GET_,RS_ ,REF_,DAS_,DAS_,NSC_,LAS_,
         DAS_,LAS_,DAS_,DAS_,DAS_,DAS_,DAS_,DAS_},

da30 []={DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,DA0 ,DA0 ,ET0 ,ET0 ,
         DA0 ,ET0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 },/*da0*/
da30a[]={NOP_,NOP_,NOP_,NOP_,NOP_,DAF_,DAF_,DAF_,DAF_,DAF_,NOP_,NOP_,DAF_,DAF_,
         NOP_,DAF_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

er30 []={ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ER0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,
         ET0 ,ET0 ,ET0 ,ET0 ,CR0 ,ET0 ,ET0 ,ET0 },/*er0*/
er30a[]={LAF_,LAF_,LAF_,ERX_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
         LAF_,LAF_,LAF_,LAF_,LAM_,LAF_,LAF_,LAF_},

/*       free nu   min  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
         mdo  msc  mso  pero rni  tagc tago litc */
cr30 []={ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,CR0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,
         ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 ,ET0 },/*cr0*/
cr30a[]={LAF_,CRN_,LAF_,CRA_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
         LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_},

me30 []={ET0, ET0, ET0, ET0, ET0 ,ET0, ET0, ME0, ET0 ,ET0 ,ET0 ,ET0, ET0 ,ET0 ,
         ET0, ME1, ET0 ,ET0, ET0 ,ET0, ET0 ,ET0 },/*me0*/
me30a[]={LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
         LAF_,LAM_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_},

me31 []={ET0, ET0, ET0, ET0, ET0 ,ET0, ET0, ME1, ET0 ,ET0 ,ET0 ,ET0, ET0 ,ET0 ,
         ET0, ET0, ET0 ,ET0, ET0 ,ET0, ET0 ,ET0,},/*me1*/
me31a[]={LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
         LAF_,LAF_,LAF_,LAF_,LAF_,MSE_,LAF_,LAF_},

*msrctab[]={et30, et30a, da30, da30a, er30, er30a, cr30, cr30a,
	    me30, me30a, me31, me31a};
struct parse pcbmsrc = {"MSRCDATA", lexlms, msrctab, 0, 0, 0, 0};
#undef ET0
#undef DA0
#undef ER0
#undef CR0
#undef ME0
#undef ME1
/* PCBMSC: State and action table for marked section in CDATA mode.
           Nested marked sections are not recognized; the first MSE ends it.
           Initial state assumes an MS declaration was processed.
*/
/* Symbols for state names (end with a number). */
#define ET2     0   /* MSS processed or buffer flushed; no data. */
#define DA2     2   /* Data in buffer. */
#define ME2     4   /* MSC found. */
#define ME3     6   /* MSC, MSC found. */

static UNCH
/*       free nu   min  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
         mdo  msc  mso  pero rni  tagc tago litc */
et32 []={DA2 ,DA2 ,DA2 ,DA2 ,DA2 ,ET2 ,ET2 ,ET2 ,ET2 ,ET2 ,DA2 ,DA2 ,ET2 ,DA2 ,
         DA2 ,ME2 ,DA2 ,DA2 ,DA2 ,DA2 ,DA2 ,DA2 },/*et2*/
et32a[]={DAS_,DAS_,DAS_,DAS_,DAS_,NON_,EOF_,GET_,RS_ ,REF_,DAS_,DAS_,NSC_,DAS_,
         DAS_,LAS_,DAS_,DAS_,DAS_,DAS_,DAS_,DAS_},

da32 []={DA2 ,DA2 ,DA2 ,DA2 ,DA2 ,ET2 ,ET2 ,ET2 ,ET2 ,ET2 ,DA2 ,DA2 ,ET2 ,DA2 ,
         DA2 ,ET2 ,DA2 ,DA2 ,DA2 ,DA2 ,DA2 ,DA2 },/*da2*/
da32a[]={NOP_,NOP_,NOP_,NOP_,NOP_,DAF_,DAF_,DAF_,DAF_,DAF_,NOP_,NOP_,DAF_,NOP_,
         NOP_,DAF_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

me32 []={ET2, ET2, ET2, ET2, ET2 ,ET2, ET2, ME2, ET2 ,ET2 ,ET2 ,ET2, ET2 ,ET2 ,
         ET2, ME3, ET2 ,ET2, ET2 ,ET2, ET2, ET2,},/*me2*/
me32a[]={LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
         LAF_,LAM_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_},

me33 []={ET2, ET2, ET2, ET2, ET2 ,ET2, ET2, ME3, ET2 ,ET2 ,ET2 ,ET2, ET2 ,ET2 ,
         ET2, ET2, ET2 ,ET2, ET2 ,ET2, ET2, ET2,},/*me3*/
me33a[]={LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,GET_,LAF_,LAF_,LAF_,LAF_,LAF_,LAF_,
         LAF_,LAF_,LAF_,LAF_,LAF_,MSE_,LAF_,LAF_},

*msctab[]={et32, et32a, da32, da32a, me32, me32a, me33, me33a};
struct parse pcbmsc = {"MSCDATA", lexlms, msctab, 0, 0, 0, 0};
#undef ET2
#undef DA2
#undef ME2
#undef ME3
/* PCBMSI: State and action table for marked section in IGNORE mode.
           Nested marked sections are recognized; the matching MSE ends it.
           Initial state assumes an MS declaration, MSS, or MSE was processed.
*/
/* Symbols for state names (end with a number). */
#define ET4     0   /* Markup found or buffer flushed; no data. */
#define ME4     2   /* MSC found. */
#define ME5     4   /* MSC, MSC found. */
#define ES4     6   /* TAGO found. */
#define MD4     8   /* MDO found (TAGO, MDO[2]). */

static UNCH
/*       free nu   min  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
         mdo  msc  mso  pero rni  tagc tago litc refc */
et34 []={ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,
         ET4 ,ME4 ,ET4 ,ET4 ,ET4 ,ET4 ,ES4 ,ET4 ,ET4 },/*et4*/
et34a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,SYS_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

me34 []={ET4, ET4, ET4, ET4, ET4 ,ET4, ET4, ME4, ET4 ,ET4 ,ET4 ,ET4, ET4, ET4 ,
         ET4, ME5 ,ET4, ET4, ET4 ,ET4, ET4, ET4, ET4,},/*me4*/
me34a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,SYS_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

me35 []={ET4, ET4, ET4, ET4, ET4 ,ET4, ET4, ME5, ET4 ,ET4 ,ET4 ,ET4, ET4, ET4 ,
         ET4, ET4 ,ET4, ET4, ET4 ,ET4, ET4, ET4, ET4,},/*me5*/
me35a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,SYS_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,MSE_,NOP_,NOP_,NOP_},

/*       free nu   min  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
         mdo  msc  mso  pero rni  tagc tago litc */
es34 []={ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ES4 ,ET4 ,ES4 ,ET4 ,ET4 ,ET4 ,ET4 ,ES4 ,ET4 ,
         MD4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 ,ET4 },/*es4*/
es34a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,SYS_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

md34 []={ET4, ET4, ET4, ET4, ET4 ,MD4, ET4, MD4, ET4 ,ET4 ,ET4 ,ET4, ET4, ET4 ,
         ET4, ET4 ,ET4, ET4, ET4 ,ET4, ET4, ET4,},/*md4*/
md34a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,SYS_,NOP_,
         NOP_,NOP_,MSS_,NOP_,NOP_,NOP_,NOP_,NOP_},

*msitab[]={et34, et34a, me34, me34a, me35, me35a, es34, es34a, md34, md34a};
struct parse pcbmsi = {"MSIGNORE", lexlms, msitab, 0, 0, 0, 0};
#undef ET4
#undef ME4
#undef ME5
#undef ES4
#undef MD4
#undef NS4
/* PCBSTAG: State and action table for start-tag parse.
            Columns are based on LEXMARK.C.
*/
/* Symbols for state names (end with a number). */
#define SP1     0   /* Separator before name expected. */
#define AN1     2   /* Attribute name expected. */
#define SP2     4   /* Separator or value indicator expected. */
#define VI1     6   /* Value indicator expected. */
#define AV1     8   /* Attribute value expected. */

int pcbstan = AN1;            /* PCBSTAG: attribute name expected. */

static UNCH
/*       bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */
sp41 []={SP1 ,SP1 ,SP1 ,SP1 ,AN1 ,SP1 ,SP1 ,SP1 ,AN1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,
         SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 },
sp41a[]={INV_,LEN_,LEN_,LEN_,NOP_,SYS_,EOF_,GET_,RS_ ,LEN_,ETIC,INV_,INV_,INV_,
         INV_,DSC ,INV_,INV_,INV_,INV_,TAGC,TAGO,INV_},

an41 []={SP1 ,SP1 ,SP1 ,SP2 ,AN1 ,AN1 ,AN1 ,AN1 ,AN1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,
         SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 },
an41a[]={INV_,NTV ,NTV ,NVS ,NOP_,SYS_,EOF_,GET_,RS_ ,NTV ,ETIC,INV_,INV_,INV_,
         INV_,DSC ,INV_,INV_,INV_,INV_,TAGC,TAGO,INV_},

sp42 []={SP1 ,SP1 ,SP1 ,SP1 ,VI1 ,SP2 ,SP2 ,SP2 ,VI1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,
         SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,AV1 },
sp42a[]={INV_,LEN_,LEN_,LEN_,NOP_,SYS_,EOF_,GET_,RS_ ,LEN_,NASV,INV_,INV_,INV_,
         INV_,NASV,INV_,INV_,INV_,INV_,NASV,NASV,NOP_},

/*       bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */
vi41 []={SP1 ,AN1 ,AN1 ,AN1 ,VI1 ,VI1 ,VI1 ,VI1 ,VI1 ,AN1 ,SP1 ,SP1 ,SP1 ,SP1 ,
         SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,AV1 },
vi41a[]={INV_,NASV,NASV,NASV,NOP_,SYS_,EOF_,GET_,RS_ ,NASV,NASV,INV_,INV_,INV_,
         INV_,NASV,INV_,INV_,INV_,INV_,NASV,NASV,NOP_},

av41 []={SP1 ,SP1 ,SP1 ,SP1 ,AV1 ,AV1 ,AV1 ,AV1 ,AV1 ,SP1 ,SP1 ,SP1 ,AN1 ,AN1 ,
         SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 },
av41a[]={INV_,AVU ,AVU ,AVU ,NOP_,SYS_,EOF_,GET_,RS_ ,AVU ,INV_,INV_,AVD ,AVDA,
         INV_,INV_,INV_,INV_,INV_,INV_,INV_,INV_,INV_},

*stagtab[] = {sp41, sp41a, an41, an41a, sp42, sp42a, vi41, vi41a, av41, av41a};
struct parse pcbstag = {"STAG", lexmark, stagtab, 0, 0, 0, 0};
#undef SP1
#undef AN1
#undef SP2
#undef VI1
#undef AV1
/* PCBETAG: State and action table for end-tag parse.
*/
#define TC1     0   /* Tag close expected (no attributes allowed). */

static UNCH
/*       bit  nmc  nu   nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */
tc41 []={TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,
         TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 ,TC1 },/*tc1*/
tc41a[]={INV_,INV_,INV_,INV_,NOP_,SYS_,EOF_,GET_,RS_ ,INV_,INV_,INV_,INV_,INV_,
         INV_,INV_,INV_,INV_,INV_,INV_,TAGC,TAGO,INV_},

*etagtab[] = {tc41, tc41a};
struct parse pcbetag = {"ETAG", lexmark, etagtab, 0, 0, 0, 0};
#undef TC1
/* PCBVAL: State and action table for tokenizing attribute values.
           Columns are based on lextoke (but EOB cannot occur).
*/
/* Symbols for state names (end with a number). */
#define TK1     0   /* Token expected. */
#define SP1     2   /* Separator before token expected. */

static UNCH
/*       inv  rec  sep  sp   nmc  nms  nu   eob   */
tk51 []={TK1 ,TK1 ,TK1 ,TK1 ,SP1 ,SP1 ,SP1 },/*tk1*/
tk51a[]={INVA,INVA,INVA,NOPA,NMTA,NASA,NUMA},

sp51 []={TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 },/*sp1*/
sp51a[]={INVA,INVA,INVA,NOPA,LENA,LENA,LENA},

*valtab[] = {tk51, tk51a, sp51, sp51a};
struct parse pcbval = {"VAL", lextoke, valtab, 0, 0, 0, 0};
#undef TK1
#undef SP1
/* PCBEAL: State and action table for end of attribute specification list.
           If delimiter occurs, process it.  Otherwise, put invalid character
           back for the next parse.
*/
/*