changelog

harvest是一个下载html网页得机器人

Thu Oct 15 10:53:33 1992  James Clark  (jjc at jclark)

	* adl.h (struct ne): Add nepubid and nesysid members.
	(NEPUBID, NESYSID): New macros.
	(NEDCNPUBID, NEDCNSYSID, NEDCNDEFINED): New macros.
	* msg.h: Deleted message 149.
	* entity.h (struct dcncb): Delete dcnid member.  Add sysid and
	pubid members.
	* modmd2.c (mdentity): Fill in nepubid and nesysid.
	(mdnot): Don't call entgen. Fill in sysid, pubid and defined
	members of dcb.
	* entgen.c (nentgen): Deleted.
	(entgen): Don't call nentgen.
	(field): When checking for Y substitution storage class must be <=
	* lineout.h (PUBID_CODE, SYSID_CODE, FILE_CODE): New codes.
	* lineout.c (process_document): Output a definition for a
	subdocument entity before processing it.
	(print_id): Rewrite.
	(output_external_entity, output_subdoc): Add two additional
	arguments.  Change printing of external identifiers.
	(output_notation): Delete `id' argument. Add two additional
	arguments.  Change printing of external identifiers.
	(define_external_entity, handle_single_token): Change callers.
	(print_filename): Don't use \s.
	* modpars2.c (aenttst): Use NEDCNDEFINED rather than NEDCNID.
	* modsgml2.c (entopen): Likewise.
	* modpars1.c (checkdtd): Use `defined' member of dcncb in place of
	`dcnid'.
	* traceset.c (tracedcn): Don't use `dcnid' member.

	* modsgml1.c (sgmlend): Merge loops. Divide maxsubcap values by
	cappoints values.
	* main.c (write_caps): Use RACT format.
	(get_subcaps): Read new format.

Wed Oct 14 09:54:19 1992  James Clark  (jjc at jclark)

	* modmd2.c (mdentity): Free pne after errors 147 or 148.

	* modmd2.c (mdnot): Free ntx on error.

Mon Sep 28 13:05:35 1992  James Clark  (jjc at jclark)

	* Version 1.0 released.

Fri Sep 25 11:17:34 1992  James Clark  (jjc at jclark)

	* Makefile.vms: New file.

	* modpars1.c (parsepro): Set rc for STE_ case.

	* modmd2.c (mdentity): For the default entity copy the system and
	public identifiers.

	* modmd2.c (mdentity): Free pne if estore != ESN.

Thu Sep 24 13:39:01 1992  James Clark  (jjc at jclark)

	* source.h (struct entity): Add dflt member.
	* modsgml2.c (usedef): Set ecb->dflt.
	* modmd2.c (mdentity): Check whether earlier existing definition
	of entity was defaulted. If so, give a warning and remove this
	definition.
	* msg.h: New message 228.

Wed Sep 23 12:14:44 1992  James Clark  (jjc at jclark)

	* modpars2.c (attval): Don't pass attribute value as argument to
	error 67.
	* msg.h: Don't use attribute value in error message.

Fri Sep 11 15:32:30 1992  James Clark  (jjc at jclark)

	* sgmlio.c (ioread): Set f->fp to 0 after closing it if there is
	no next file.
	(iopend): Don't call fclose if f->fp is 0.

Mon Aug 31 11:06:57 1992  James Clark  (jjc at jclark)

	* Makefile (config.h): Don't use `export'.

Fri Aug 21 15:12:48 1992  James Clark  (jjc at jclark)

	* modsgml2.c (entopen): For ESX and ESC return ENTDATA even if
	entity is empty.
	* lineout.c (print_data): If n == 0, don't ignore it if it's sdata.

Tue Aug 18 11:30:38 1992  James Clark  (jjc at jclark)

	* modsgml2.c (entref): Use error 35 if it's a parameter entity.

Mon Aug 17 12:53:48 1992  James Clark  (jjc at jclark)

	* source.h (union etext): Delete etd member.

	* Makefile: Include sgml-mode.el.

Thu Aug 13 15:04:47 1992  James Clark  (jjc at jclark)

	* sgmldecl.c (sdcsdesc): Give a warning for an unknown designating
	escape sequence.
	(sgmlwrsd): Use the system character set as the document character
	set.
	(DUMMY_CHARSET): Deleted.
	* msg.h: New message 227.

	* test: Put out and err files in separate directories.

	* main.c (main): Ignore -a.
	(usage): Don't mention -a.
	(swinit): Set swambig to 1.
	(build_subargv): Don't pass -a.

Wed Aug 12 22:02:24 1992  James Clark  (jjc at jclark)

	* action.h (STE_): New action.
	* arcpcbrf.c (es7a): Change action for nms and tagc from PEP_ to STE_.
	* modpars1.c (parsepro): Handle STE_.
	* modsgml1.c (sgmlnext): Start the document element for DCE_ or MSS_.

Thu Aug  6 16:42:37 1992  James Clark  (jjc at jclark)

	* modpars2.c (parseval): If a number token is terminated by a name
	character, give error 17.

Wed Jul 29 01:49:20 1992  James Clark  (jjc at jclark)

	* modpars1.c (startdtd): Allocate an extra character for nmbuf.
	* modmd2.c (mdentity): Check that length of parameter entity name
	does not exceed NAMELEN - 1.  Pass NAMELEN to parsemd when
	parsing parameter entity name.

	* msg.h: New message 225.
	* modserv.c (sandwich): Move to...
	* modmd2.c (sandwich): Return 0 if the result is too long.
	* modmd2.c (mdentity): Give an error if the result was too long
	after adding bracketing delimiters.

	* msg.h: New message 224.
	* modmd1.c (mdattdef): It is an error if the value returned by
	vallen exceeds LITLEN.
	* modpars2.c (attval): Likewise.

	* sgmls.c (unescape1): If s does not contain a backslash, return
	the length of s.

Tue Jul 28 18:35:54 1992  James Clark  (jjc at jclark)

	* action.h (PENR): New action.
	* modpars2.c (parsemd): Handle PENR.
	* arcpcbrf.c (pr21a): Change actions for ee, com, pero from PCI_
	to PENR.
	(sp21, tk21): After ee, state is TK1.

	* modpars2.c (parsegrp, parsengrp): Detect duplicates.

	* modpars1.c (parsepro): Handle EE_.

Thu Jul 23 12:50:34 1992  James Clark  (jjc at jclark)

	* modmd1.c (mdattdef): If anmtgrp returns an error and adlim <
	GRPCNT, report that ATTCNT not GRPCNT was exceeded.

Mon Jun 29 00:46:09 1992  James Clark  (jjc at jclark)

	* sgmldecl.c (systemcharset): New charset.
	(charset_map): By default assume the system character set is ISO
	Latin-1.
	(nlextoke, nlextran): New variables.
	(setnonsgml): Rename to...
	(setlexical): Handle characters made significant by NAMING
	section.  Use nlextoke and nlextran to munge lexical tables.
	(sgmldecl): Call setlexical.
	(sdpubsyntax): Delete call to setnonsgml().
	(sdshunchar): Delete calls to setnonsgml().
	(sdnaming): Redo parsing of naming classes.
	(sgmlwrsd): Generate NAMING section from lextoke and lextran.
	Use system character set as syntax reference character set.

	* msg.h: Remove trailing period from message 218.

Sun Jun 28 15:59:25 1992  James Clark  (jjc at jclark)

	* modpars2.c (tokdata): Insert missing parentheses around
	`parmlen -= 2'.

Sat Jun 20 10:09:39 1992  James Clark  (jjc at jclark)

	* Makefile (depend): Use sed { command.

	* Makefile: Pass PREFIX to configure.
	* configure: Edit DEFAULT_PATH using PREFIX.

Mon Jun 15 19:41:55 1992  James Clark  (jjc at jclark)

	* configure: More elaborate test for HAVE_EXTENDED_PRINTF.

Fri Jun 12 13:20:49 1992  James Clark  (jjc at jclark)

	* context.c (testend): Set andoptsw to 1 after ending a group.

Wed Jun 10 16:49:50 1992  James Clark  (jjc at jclark)

	* arcpcbrf.c (pcbconm): New state NS0.
	(et0, da0, da1): After NSC state changes to NS0.
	(pcbmsrc): New state NS0.
	(et30, da30): After NSC state changes to NS0.
	(pcbmsc): New state NS2.
	(et32, da32): After NSC state changes to NS2.
	(pcbmsi): New state NS4.
	(et34, me34, me35, es34, md34): After NSC state changes to NS4.

	* lineout.c (print_string): Don't recognize character preceded by
	DELNONCH as DELCDATA or DELSDATA.

Tue Jun  9 21:17:15 1992  James Clark  (jjc at jclark)

	* modpars1.c (parsepro): Call setdtype() when EOD_ encountered.
	Don't call setdtype() or checkdtd() if we gave an error message
	for the EOD_.

Sun Jun  7 09:57:32 1992  James Clark  (jjc at jclark)

	* configure: When determining void_ret, use a typedef in order to
	catch compilers that can't handle this typedef.

	* replace.c (hash): Cast enumeration type to int.
	* sgmls.c (error): Likewise.
	* sgmldecl.h: Use #define rather than enum to define *CAP and Q*
	constants.
	* sgmlmsg.c: Likewise for HDR*.

Fri May 22 13:31:58 1992  James Clark  (jjc at jclark)

	* sgmldecl.c (bufsalloc, bufsrealloc): Allocate entbuf here.
	* modpars1.c (commbufs): Don't allocate entbuf here.

Wed May 20 11:55:42 1992  James Clark  (jjc at jclark)

	* Makefile (config.h): Depends on configure and unix.cfg.

	* configure: Pipe yes into man.

	* arclexrf.c: Don't include stdio.h.  Use 0 instead of NULL.

	* unixproc.c: Include standard header files before std.h.

	* std.h: Reorder definitions so that all standard headers are
	included before conditional definitions of SEEK_SET etc.

Mon May 18 15:38:35 1992  James Clark  (jjc at jclark)

	* modpars2.c (charrefn): Return DAF_ instead of DAS_.

Mon May 18 08:37:23 1992  James Clark  (jjc at jclark)

	* configure: Catch signals when testing for extended printf.

Sun May 17 09:49:14 1992  James Clark  (jjc at jclark)

	* modmd1.c (mdadl, mdelem): Don't dereference nmgrp[0] if it is NULL.

	* replace.c (parse_error): Declare as NO_RETURN.

	* context.c (copypos): Cast unsigned char to int in comparison.
	* modmd1.c (mdelem): Likewise.
	* modmd2.c (storedatt): Likewise.
	* modpars2.c (parsenm): Likewise.
	(s2valnm): Likewise.
	* sgmlxtrn.[ch]: Make Tstart an int.
	* context.c: Don't cast Tstart to int.
	* modpars2.c (shiftnon): Cast unsigned char to unsigned in modulus
	expression.
	* context.c (HITSET, HITOFF): Cast `n' to unsigned in shift
	expression.

	* arclexrf.c: Avoid use within comments of character sequences
	that will be interpreted as trigraphs.

	* std.h: Don't include getopt.h.
	* configure: Don't handle HAVE_GETOPT_H.
	* *.cfg: Delete HAVE_GETOPT_H.
	* getopt.c, main.c, sgmlsasp.c: Include getopt.h.
	* getopt.h: Don't use prototypes.

	* msgcat.h: Don't use prototypes.

Sat May 16 12:15:43 1992  James Clark  (jjc at jclark)

	* entgen.c (nentgen): Treat empty system/public identifiers as
	missing.

Fri May 15 23:51:10 1992  James Clark  (jjc at jclark)

	* std.h: Declare errno in case <errno.h> doesn't.

Tue May 12 18:58:54 1992  James Clark  (jjc at jclark)

	* *.cfg (HAVE_GETOPT_H): New define.
	* configure: Handle HAVE_GETOPT_H.
	* std.h: Include getopt.h.
	* getopt.c, main.c, sgmlsasp.c: Don't include getopt.h.

	* configure: Fixes for Ultrix.

	* entgen.c (entgen): Cast f->fpisysis to char * in argument to
	strchr.

	* sgmlsasp.c (error): Define as NO_RETURN.

Mon May 11 08:54:06 1992  James Clark  (jjc at jclark)

	* Version 0.8 released.

	* Makefile.bcc (OBJS1): Add exclude.obj.

Sat May  9 10:26:16 1992  James Clark  (jjc at jclark)

	* dos.cfg: Add CANT_REDIRECT_STDERR.
	* main.c: Add -f option if CANT_REDIRECT_STDERR is defined.

	* adl.h (NEDCNADL): Change dcnadl to adl.

	* entity.h (struct dcncb): Add entsw member.
	* modmd1.c (mdnadl): Check that an entity with this notation has
	not already been declared.
	* modmd2.c (mdextid): Set the entsw field of the notation.
	* msg.h: New message 218.

	* msg.h: Improve message 192.

	* msg.h: Improve message 78.

	* modpars1.c (checkdtd): Check for undefined notations even if
	sw.undef not specified.
	* msg.h: Change type of 192 to 'W'.

	* modmd2.c (mdextid): Separate copying of attribute list into
	storedatt.  Call this also if there was no data attribute
	specification list but this notation does have an attribute list.
	(storedatt): New function.  Don't increase ds.attgcnt.
	* modpars2.c (parseatt): Separate initialization of ad into
	initatt.  No need to copy al: this is done in storedatt.
	(initatt): New function.
	* sgmlfnsm.h: Declare storedatt and initatt.

Fri May  8 13:00:56 1992  James Clark  (jjc at jclark)

	* sgmldecl.c: Use ISO 8879:1986 instead of ISO 8879-1986.
	Also recognize the latter but give an error message.
	(sdfixcolon): Rename to sdfixstandard.
	* msg.h: Change message 163 accordingly.

	* modpars2.c (adlval): Don't dereference newetd if it's NULL.

	* entgen.c (entgen): Initialize sysidsrch.  Don't always call
	sysidgen() if there's a system identifier.
	(field): Use %E for escape sequence. %S gives system identifier.

Wed May  6 21:40:25 1992  James Clark  (jjc at jclark)

	* sgmldecl.c (sdtranschar): Cast arguments to ltous to long.

	* msgcat.c: Conditionalize use of prototypes on USE_PROTOTYPES not
	__STDC__.

Tue May  5 00:24:39 1992  James Clark  (jjc at jclark)

	* lineout.c (process_subdoc): Only call get_subcaps() if
	subprocess exited successfully.

	* msg.h: New message 217.
	* exclude.c (excktok): Use message 217 instead of 41. Second
	parameter is current gi not gi that does excluding.

	* modpars1.c (iopend): Save offset into line.  Don't seek to
	middle of line.
	(iocont): Seek to beginning of line, then skip to middle of line
	using saved offset.
	(struct iofcb): Delete penderr and sverrno.  New pendoff member.

	* modpars1.c (parsecon): An RE that would be ignored can force a
	start tag.

Mon May  4 23:26:21 1992  James Clark  (jjc at jclark)

	* unix.cfg (FPOS_MISSING): New define.
	* configure: Handle FPOS_MISSING.
	* std.h: Provide fpos_t, fgetpos, fsetpos if FPOS_MISSING.
	* sgmlio.c (iopend, iocont, ioread): Use fsetpos/fgetpos rather
	than fseek/ftell. Use penderr to indicate that iopend() failed.
	(struct iofcb): Make off an fpos_t. Add penderr member.