changelog

harvest是一个下载html网页得机器人

Sun May  3 09:41:05 1992  James Clark  (jjc at jclark)

	* sgmlmsg.c (printit): Don't rely on return value of fprintf.

Wed Apr 29 00:03:42 1992  James Clark  (jjc at jclark)

	* modpars1.c (destack): Undo Apr 26 change.
	* exclude.c: New file.
	* msg.h: New message 216.
	* Makefile: Add exclude.c.
	* modpars1.c (stack): Call exclude().
	* sgmlfnsm.h: Declare exclude().
	* modpars1.c (stag): Don't give errors here for invalid exclusions.

	* ambig.c (listambig): Clear marks afterwards.

	* synxtrn.h: Don't use an enum for the key indices.

Tue Apr 28 09:22:43 1992  James Clark  (jjc at jclark)

	* vms.cfg: New file.
	* Makefile (MISC): Include vms.cfg.

	* portproc.c (run_process): Free the right thing.

	* Makefile (OBJS2): Add getopt.o.

	* modsgml1.c (sgmltab): Undo change of April 13. When pending RE
	is followed by another RE that is ignored, output the first RE
	immediately.

Sun Apr 26 11:36:49 1992  James Clark  (jjc at jclark)

	* modpars1.c (destack): When element ends without required element,
	check whether required element has been excluded.

Tue Apr 21 12:39:42 1992  James Clark  (jjc at jclark)

	* entity.h (struct switches): Add swundef member.
	* main.c (swinit): Initialize swundef.
	(OPTIONS): Add u.
	(main): Support -u.
	(usage): Mention -u.
	* modpars1.c (checkdtd): New function.
	(parsepro): Call checkdtd.
	* msg.h: New messages 142 and 190.

Mon Apr 20 15:09:40 1992  James Clark  (jjc at jclark)

	* modmd1.c (mdadl): Only call etdadl if the element has already
	been defined.
	(etdadl): No need to check that p->etdmod is non-NULL.

Thu Apr 16 12:56:15 1992  James Clark  (jjc at jclark)

	* main.c (main): Implement -i.
	(swinit): Initialize includes.
	* entity.h (struct switches): New includes member.
	* modpars1.c (pentname, doincludes): New functions.
	(startdtd): Call doincludes.
	(endprolog): Call doincludes if there was no DTD.
	* msg.h: New message 138.

Wed Apr 15 15:58:53 1992  James Clark  (jjc at jclark)

	* traceset.c (tracemap): Deleted.
	* trace.h (TRACEMAP): Deleted.
	* modserv.c (mapsrch): Don't use TRACEMAP.
	* modsgml2.c (ptrsrch): Likewise.

	* trace.h, traceset.c *.cfg: Include tracing is TRACE is defined,
	rather than if FINAL is not defined.
	* entity.h: Include trace and ptrace in struct switched only if
	TRACE is defined.
	* main.c (main, swinit, build_subargv): Support -x and -y only if
	TRACE is defined.
	(xrealloc): New function.
	* appl.h: Declare xrealloc.

	* traceset.c (tracemod): Calculate size of pg correctly.

Tue Apr 14 13:47:21 1992  James Clark  (jjc at jclark)

	* action.h: Get rid ETGP_. Renumber accordingly.
	* modsgml1.c (sgmltab): Get rid of column for ETGP_.

	* modpars2.c (parseseq): Rewrite.

	* modpars2.c (s2valnm): Cosmetic changes.

	* modpars2.c (parsetkn): Rewrite.  Don't store length ahead of
	name.
	(parseatt): Don't call dellen.
	(mdattdef): Likewise.
	(charrefn): Number doesn't have length byte.
	* modmd1.c (mdelem): ranksuff doesn't have length byte.
	* modserv.c (dellen): Deleted.
	* sgmlfnsm.h (dellen): Delete declaration.
	* sgmldecl.c: Numbers no longer have length bytes.

	* modpars2.c (parse): Use NAMELEN rather than 8 when parsing
	number in numeric character reference.

Mon Apr 13 23:21:45 1992  James Clark  (jjc at jclark)

	* modsgml1.c (sgmltab): Add two new states to handle the case
	where the pending RE has been followed by an RE that was ignored.

Wed Apr  8 10:59:48 1992  James Clark  (jjc at jclark)

	* modsgml1.c (sgmltab): In state NR1, hold on to REF_ as in state
	DA1.

	* Makefile (install, install.man): New targets.

	* modsgml1.c (sgmlact): When popping the pcbsgml stack, change the
	state of the new top of stack pcb as if there had been a
	processing instruction or markup declaration.

Tue Apr  7 13:55:17 1992  James Clark  (jjc at jclark)

	* configure: New file.
	* Makefile: Add configure to distribution.
	(config.h): Build using configure.
	* unix.cfg: Some minor cleanup to work better with configure.
	Move things that can't be auto-configured to top.

Sun Apr  5 20:53:42 1992  James Clark  (jjc at jclark)

	* main.c (main): Call MAIN_HOOK if it's defined.

	* Makefile: Add thinkc.cfg to MISC.

	* getopt.c: Include getopt.h.

	* sgmlio.c (nextstr): Make declaration a prototype.

Wed Apr  1 09:30:33 1992  James Clark  (jjc at jclark)

	* main.c (usage): Mention -p.

Sun Mar 29 15:47:44 1992  James Clark  (jjc at jclark)

	* sgmldecl.c (sgmldecl): Use separate buffer to parse closing >.
	Return non-zero if there was some appinfo.
	(sdappinfo): Don't save appinfo.  Delete appinfo array.
	(sgmlappinfo): Deleted.
	(sgmlwrsd): Always write out APPINFO NONE.	
	* sgmlmain.g (sgmlappinfo): Delete declaration.
	* sgmlfnsm.h (sgmldecl): Change return type to int.
	* action.h (APP_): New action.
	* modpars1.c (parsepro): Return APP_ after parsing SGML
	declaration if there was some appinfo.
	* sgmlmain.h (SGMLAPP): New event.
	(ADATA, ADATALEN): New macros.
	* modsgml1.c (sgmlnext): Return SGMLAPP for APP_.
	* lineout.h: Define APPINFO_CODE.
	* lineout.c (process_document): Handle SGMLAPP.
	(output_appinfo): New function.
	* sgmls.c (sgmls_next): Handle APPINFO_CODE.
	* sgmls.h (sgmls_event_type): Add SGMLS_EVENT_APPINFO.
	(sgmls_event): Add appinfo field to union.
	* sgmlsasp.c (do_file): Ignore SGMLS_APPINFO.

Fri Mar 27 11:18:06 1992  James Clark  (jjc at jclark)

	* Version 0.7 released.

	* modsgml2.c (scbset): Check that es >= 0.

	* sgmldecl.c (sgmlappinfo): New function.
	* sgmlmain.h: Declare it.

	* lineout.c (output_internal_entity): Distinguish CDATA and SDATA
	entities with a keyword rather than with \|.
	* sgmls.c (parse_internal_entity): Grok this.
	Rename E_ENTITYTYPE to E_BADEXTERNAL.
	Change error message for E_BADINTERNAL.
	(unescape): Split off part into
	(unescape1): New function.
	* sgmls.pl: Grok new I format.

	* sgmlio.c (nifopen): New function.
	(FOPENR): Define as nifopen if HAVE_O_NOINHERIT is defined.
	* dos.cfg: Define HAVE_NO_INHERIT.

	* stklen.c: New file.
	* Makefile.bcc: Compile stklen.c.
	* Makefile: Include stklen.c in distribution.

	* main.c (main): New -p option.

	* modpars1.c (parsepro): Give an error for R/CDATA marked section
	in the DTD subset.
	* msg.h: Use message 135 for this.

	* modpars1.c (common): Rename to commbufs. Change callers.
	* sgmlfnsm.h: Change declaration.

	* sgmlsasp.c (main): New -n option.
	(fold_general_names): New variable.
	* sgmlsasp.h (fold_general_names): Declare it.
	* replace.c (scan_name): Fold name to upper case only if
	fold_general_names is non-zero.

Thu Mar 26 12:39:24 1992  James Clark  (jjc at jclark)

	* modpars1.c (setdtype): Don't give message about the document
	type element's not being declared if we already gave an error
	message about there not being a doctype declaration.
	(parsepro): In onlypro mode, call setdtype.

Wed Mar 25 11:13:18 1992  James Clark  (jjc at jclark)

	* sgmldecl.c (sdquantity): Eliminate unnecessary inner
	redeclaration of i.

	* sgmldecl.c (sdnaming): Eliminate unnecessary assignment to i.

	* sgmls.c (parse_data): Eliminate unnecessary assignment to start.

	* main.c (make_argv): Eliminate unnecessary increment of i.

	* entgen.c (entgen): Eliminate unnecessary assignment to buf.

	* modpars2.c (antvget): Make advn local to block in which it is
	used. Eliminate unnecessary assignment.

	* modpars2.c (alperm): Eliminate unnecessary assignment to alperm.

	* modpars1.c (parsecon): Eliminate unnecessary assignment to pcb.

	* context.c (newtoken): Make Psave and rc local to block in which
	it is used. Eliminate unnecessary assignments.

	* Cast some arguments of xrealloc, memcpy, MEMZERO to UNIV.

	* tools.c (VP): New macro.
	* sgmlsasp.h (VP): New macro.
	* appl.h (appl_error): 
	* main.c (fatal):
	* replace.c (parse_error):
	* sgmlmsg.c (xfprintf): Use VP for prototypes of varargs functions.

Mon Mar 23 13:57:55 1992  James Clark  (jjc at jclark)

	* sgmlxtrn.c (savedpos): New variable.
	* sgmlxtrn.h: Declare it.
	* modpars1.c (endprolog): Separate allocation of struct mpos's
	into...
	(newmpos): New function.
	(endprolog): Allocate savedpos.
	* context.c (context): If there's an applicable inclusion, save
	the current position in savedpos.  Don't return RCPEX unless the
	element is valid in this model only as an inclusion; in this case,
	restore the saved position.

Sun Mar 22 10:34:54 1992  James Clark  (jjc at jclark)

	* entity.h (struct switches): Add onlypro member.
	* main.c (swinit): Initialize onlypro.
	* modpars1.c (parsepro): If onlypro switch is set, don't give an
	error for EOD after DTD subset is finished, and also return EOD at
	end of prolog.

Sat Mar 21 19:27:59 1992  James Clark  (jjc at jclark)

	* etype.h (TXOPT, TXREP, TXOREP): New ttype flags to support dump.c.
	* modpars2.c (parsemod, parsegcm): Set TX* flags when occurrence
	indicator is specified explicitly.

	* modmd2.c (mdentity): Build the entity text for ESS and ESE
	types. Handle ESS and ESE like other internal entity types.
	(tagtext, entfix): Deleted.
	* sgmlxtrn.c (tagtext, entfix): Delete declarations.
	* modpars1.c (shortref): Don't handle ENTEGI, ENTSGI. Delete
	stagrc variable.
	* modpars2.c (parse): Likewise. Delete stagrc variable.
	* modsgml2.c (entopen): Handle ESS and ESE normally.
	* traceset.c (traceecb): Likewise.
	* source.h (ENTEGI, ENTSGI): Deleted.
	(ENTDATA, ENTPI): Renumber.

Mon Feb 24 16:12:59 1992  James Clark  (jjc at jclark)

	* context.c (testend): Insert brackets around && within || to keep
	gcc 2 quiet.

	* traceset.c: Add some casts to keep gcc 2's -Wformat happy.

Fri Feb 21 12:21:05 1992  James Clark  (jjc at jclark)

	* lineout.c (process_subdoc): Forget the current filename after
	processing subdoc.

	* main.c (build_subargv): Pass on -l.

Thu Feb 20 11:48:06 1992  James Clark  (jjc at jclark)

	* action.h (MSP_): New action.
	* arcpcbrf.c (md7a): Change action for MSO from LAF_ to MSP_.
	* modpars1.c (parsepro): Handle MSP_.  For MSS_ return DCE_.
	* modsgml1.c (sgmlnext): Change handling of MSS_ in the prolog.

	* modserv.c (hout): Use a struct hash **.  Had bug when removing
	first element in chain.

Tue Feb 18 11:11:35 1992  James Clark  (jjc at jclark)

	* sgmlxtrn.c: Don't declare array sizes unnecessarily. Insert
	newlines in license.

	* msg.h: Give errors about unsupported features a `U' type code.
	* sgmlsmsg.c: Use a different header for `U' type messages.

Fri Feb 14 11:41:51 1992  James Clark  (jjc at jclark)

	* sgmls.c (parse_data): Don't use strchr.

	* replace.c (hash): Change hashing function. Do modulus in hash().
	Make return type int.  Change callers.

	* sgmls.h, sgmls.c, sgmlsasp.c: Redo error handling so as not to use
	setjmp.
	* std.h: Don't include <setjmp.h>.	

	* source.h (ENTHASH):
	* etype.h (ETDHASH):
	* adl.h (IDHASH): Bump up values.

	* modserv.c (hash): Use better hashing function.

	* modserv.c (idget): Deleted.
	* sgmlfnsm.h: Delete declaration.

	* momd11.c (mdelem, mdadl):
	* modmd2.c (mdsrmuse): Don't allocate nmgrp
	* modmd1.c (mdnadl, mdelem): Don't allocate nnmgrp.
	* modpars2.c (parsemod): Don't allocate gbuf.
	* modmd2.c (mdelem): Don't allocate nmbuf
	* modmd2.c (mdextid): Don't allocate sysibuf, pubibuf.
	* sgmldecl.c (bufsalloc, bufsrealloc): Handle only scbs and tbuf.
	* modpars1.c (parsepro): Call startdtd() before mddtds.
	(startdtd): New function. Allocate nmgrp, nnmgrp, gbuf, nmbuf,
	pubibuf, sysibuf.
	(common): New function.  Allocate al, lbuf, entbuf.
	(endprolog): Free buffers allocated in startdtd().  Call common()
	if it hasn't already been called.

Thu Feb 13 10:42:26 1992  James Clark  (jjc at jclark)

	* modsgml2.c (entref): Handle the case where the entity has been
	used in a shortref mapping declaration, but has not been defined.

	* Redo tracing code.

	* traceset.c (traceecb): Return if p is NULL.
	* modsgml2.c (entfind): Call traceecb even if p is NULL.

	* sgmlincl.h: Always include trace.h.

	* traceset.c (estype1, estype2): Make them local static variables
	in traceecb.
	* trace.h (estype1, estype2): Declarations deleted.

	* sgmlxtrn.[ch] (pifldlen): Deleted.
	* modmd2.c (pubfield): Additional argument that gets length of
	field.
	(parsefpi): Change callers.  Don't use pifldlen.

	* modpars2.c (antvget): Pass pointer which receives value.
	(parseatt): Pass pointer to antvget() to receive value.
	* sgmlxtrn.[ch]: Delete antvptr.

	* sgmlxtrn.[ch] (pvalptr): Deleted.
	* modpars2.c (parseval): Don't set pvalptr.
	(attval):
	* modmd1.c (mdattdef):
	* modmd2.c (tagtext): Use last argument passed to parseval
	instead of pvalptr.

Tue Feb 11 13:23:03 1992  James Clark  (jjc at jclark)

	* modpars1.c (endprolog): tag::tpos needs to be GRPLVL+2 long.

	* entity.h (MAXNAMELEN): Move into sgmldecl.c.

	* modmd1.c (mdelem): Use lbuf for rank suffix. Use tbuf for
	comgibuf.

	* modmd2.c (mdsrmuse): Store short reference name in lbuf.
	Simplify accordingly.