changelog

harvest是一个下载html网页得机器人

Mon Apr 25 14:08:16 1994  James Clark  ([e-mail])

	* sgmls.c (unescape): Work around Borland C++ 4.0 bug.

	* dos.cfg (sys_nerr, sys_errlist): Define for Borland C++ 4.0.

	* main.c (main): Rename -C to -m.
	(usage): Mention -m option.

	* Makefile (dist): Don't include version number in directory.

	* appl.h: Separate declarations of xmalloc and xrealloc in ...
	* alloc.h: New file.
	* catalog.c, main.c, portproc.c, process.c: Include alloc.h.
	* Makefile (HDRS): Add alloc.h.

	* sgmls.c: Don't use obsolete GNU C volatile extension.

Wed Apr 20 08:24:52 1994  James Clark  ([e-mail])

	* unix.cfg: Declare const as empty if __STDC__ is not defined.
	* tools.h, xfprintf.c, sgmls.c, catalog.c: Don't do it here.

	* Add catalog support.
	* catalog.c, catalog.h: New files.
	* entity.h: Include catalog.h.  Add catalog member to struct switches.
	* entgen.c (entgen): Use catalog.
	(catsysidgen, basename, entginit): New functions.
	* tools.h: Declare const as empty if __STDC__ is not defined.
	* sgml1.c (sgmlset): Call entginit.
	(getsubst): New function.
	* sgmlaux.h: Declare entginit, getsubst.
	* main.c (main): Create catalog. Handle -C option.
	(swinit): Initialize catalog.
	(do_catalog_error): New function.
	* unix.cfg (FILE_IS_RELATIVE, DIR_BASE_SEP, CATALOG_FILES_ENV_VAR,
	DEFAULT_CATALOG_FILES): New macros.

Fri Mar 25 11:15:00 1994  James Clark  ([e-mail])

	* traceset.c (tracepcd, tracetkn): Use %lu for RCNT argument.

	* sgmlxtrn.c (lastetd): Don't declare extern.

	* std.h: Always declare strerror, in case systems have it in the
	library but don't declare it.

Fri Mar 11 10:35:26 1994  James Clark  ([e-mail])

	* Makefile.vms: Install new version from John Lavagnino.

Sat Mar  5 09:35:58 1994  James Clark  ([e-mail])

	* sgmldecl.c (sdcapacity): Check that value of capacity does not
	exceed TOTALCAP.
	* msg.h: New message 235.

	* sgmldecl.c (sdcharset): Don't give an error if some characters
	in the range 0 - 256 are not described.

	* pars1.c (parsepro): For EOD_, check the DTD if we've had one.

Mon Feb 28 09:05:55 1994  James Clark  ([e-mail])

	* md2.c (mdsrmuse): Give error if after document element.
	* pars2.c (parse): For case CRA_, give error if after document
	element.
	* pars1.c (destack): If destacking document element, set docelsw.
	Also give error if current entity not document entity.
	* sgml2.c (entopen): Give error if after document element.
	* sgmlxtrn.c (docelsw): New variable.
	* sgmlxtrn.h: Declare it.
	* msg.h: New message 231, 232, 233, 234.

	* msg.h: Change message 191 to show that this is an unsupported
	feature.

	* pcbrf.c: Rework handling of separators in pcbpro and pcbsd.
	* md1.c, md2.c: Get rid of reinitialization of pcbmd.newstate to
	0, except at beginning of each declaration.
	* pars2.c (parsemd): Set pcbmd.newstate to 0, if NUM token was
	actually attribute value.

	* sgmldecl.c (sdcsdesc): Use sdsaverr.
	(sdsaverr): New function.
	(sgmldecl): Print or free saved errors.
	* sgml2.c (savmderr): New function.
	* msg.h: Make 189 and 190 errors. Change `should' to `must' in
	wording of 190.

	* pars1.c (checkdtd): Check that entities in short reference maps
	are declared.
	(shortref): Don't check this here.
	* msg.h: Change message 93 accordingly.

	* sgmlxtrn.c (lastetd): New variable.
	* sgmlxtrn.h: Declare it.
	* pars1.c (nstetd): When OMITTAG is NO, use lastetd.
	(destack): Set lastetd.

	* pcbrf.c (pcbpro): Recognize end tags.
	* pars1.c (parsepro): Handle ETE_.
	* action.h: Declare ETE_.

	* md2.c (parsefpi): Registered or unregistered owner identifier
	starts with +// or -//.  Unavailable text indicator is -//.
	Language cannot be empty.

	* pars2.c (parsemod): Check for entity nesting level when parsing
	suffix.

Fri Feb  4 17:51:41 1994  James Clark  ([e-mail])

	* lineout.c (output_attribute_token): Don't use print_string.
	(handle_single_token): Use print_string for CDATA attributes.

Thu Feb  3 11:22:15 1994  James Clark  ([e-mail])

	* sgmls.c (add_attribute): Don't sort attributes here.
	* rast.c (output_attribute_list, compare_attributes,
	init_sort_code): New functions.
	(output_element_start, output_external_entity_info): Use
	output_attribute_list.

	* Make Tstart local to group.
	* etype.h (struct mpos): Add tstart member.
	* context.h (Tstart): New macro.
	* context.c (traceend): When ending group, don't reset Tstart.
	* sgmlxtrn.h (Tstart): Deleted.
	* sgmlxtrn.c (Tstart): Deleted.
	* traceset.c (tracegi, traceend): Remove Tstart argument.
	* trace.h: Change declarations and macros accordingly.
	* context.c: Remove Tstart argument from calls to tracing macros.

Fri Jan  7 11:33:40 1994  James Clark  ([e-mail])

	* sgmldecl.c (sdshunchar): Use E_SHUNCHAR rather than E_XNUM.
	* msg.h: Use message 159 for this.

Thu Jan  6 20:11:29 1994  James Clark  ([e-mail])

	* sgmldecl.c (sgmlwrsd): Handle case where there are no shunned
	characters.

Tue Nov 30 11:35:49 1993  James Clark  ([e-mail])

	* sgml2.c (entopen): Ignore reference to empty CDATA entity.

Mon Nov 22 12:12:07 1993  James Clark  ([e-mail])

	* sgmldecl.c (setlexical): Handle lexmin.
	* lextaba.c (lexmin): New lexical table for minimum data.
	* synxtrn.h (lexmin): Declare it.
	(lexcode): Add minlitc.
	* pars2.c (parselit): Use lex.l.minlitc if appropriate.
	* pcbrf.c (pcblitv): Use lexmin.
	* lexrf.c (lextabs): Add lexmin.
	* genlex.c (lextabnames): Add lexmin.

	* lextaba.c (iso646charset, iso646C0charset, iso646G0charset,
	iso8859_1charset, iso6429C1charset): New variables.
	* sgmldecl.c (charset_map): Redo.
	* sgmldecl.h: Declare character set arrays.
	* genlex.c (main): Generate charsets.

	* entity.h (struct switches): Add swcap member.
	* sgml1.c (sgmlend): Only report capacity errors if swcap is set.
	* main.c (swinit): Initialize swcap to 0.
	(main): Set swcap to 1 if -c option is given.

Wed Nov  3 14:03:17 1993  James Clark  ([e-mail])

	* md1.c (etdadl): Don't complain about end-tag minimization of -
	for element with conref attribute.

	* pars1.c (checkdtd): Always check short reference maps.

Tue Nov  2 13:20:56 1993  James Clark  ([e-mail])

	* pars2.c (tokdata): Don't give special treatment to
	RS/RE/SEPCHAR. Only do tokenization.

	* msg.h: Make 152 an error.  Improve message.
	Delete message 159.
	* pars1.c (stack): Don't give an error here for undefined short
	reference map.

Mon Nov  1 12:55:51 1993  James Clark  ([e-mail])

	* pars1.c (stack): Don't call exclude.
	(stag): Put back in old exclusion checking, but change error for
	RCHITMEX to 216.
	* msg.h: Delete message 217. Reuse message 216
	* context.c (context): Save position at beginning if there are
	exclusions.  Restore position before returning RCMEX.
	* exclude.c: Deleted.
	* Makefile: Remove exclude.o.
	* Makefile.bcc: Likewise.

	* sgmlmsg.c (elttrace): Cast argument to char * to avoid gcc
	warning.

Mon Oct 25 13:11:27 1993  James Clark  ([e-mail])

	* pcbrf.c (pcbstag): Make actions for com like nmc.

Fri Sep 24 20:27:48 1993  James Clark  ([e-mail])

	* sgmldecl.c (sdfixstandard): Add additional argument that says
	not to give an error.
	(sdversion): Change caller.
	(sdpubcapacity, sdpubsyntax): Change callers so as not to give error.

Wed Sep 15 18:08:28 1993  James Clark  ([e-mail])

	* sgml2.c (usedef): If external entity cannot be find, give a
	warning but return a non-NULL ecb.
	(entref): Simplify now that usedef does not return 0.
	* msg.h: Make message 150 a warning.

Mon Aug 23 15:08:14 1993  James Clark  ([e-mail])

	* msg.h: Fix typo in message 204.

Wed Aug 18 23:32:29 1993  James Clark  ([e-mail])

	* md2.c (mdextid): If FORMAL is NO, parse the public identifier but
	don't give an error if it's not a valid formal public identifier.

Sun Jul 11 20:34:04 1993  James Clark  ([e-mail])

	* pars1.c (parsepro): For case DTE_, issue an error if marked
	section open.
	* msg.h: New message number 230.

Sat Jul  3 13:35:59 1993  James Clark  ([e-mail])

	* pcbrf.c (et2a): Change action for mso to SR25 and for fce to FCE_.

Fri May  7 11:28:52 1993  James Clark  (jjc at jclark.com)

	* sgml1.c (sgmltab): In state ST2 change state/action for ref_ to
	NR2, LOP_.

Mon Apr 26 19:51:52 1993  James Clark  (jjc at jclark.com)

	* msgcat.c (message_compare): Fix syntax_error.
	(print_text): Don't use UNCH.
	(ISASCII): Define.

Tue Mar 30 12:13:56 1993  James Clark  (jjc at jclark)

	* pcbrf.c (dt01): Change next state for dtgc to SP1.

	* pars2.c (parselit): Allow length of literal to grow to parmlen +
	1, until closing delimiter is seen, since final space in minimum
	literal is ignored.

	* sgmlxtrn.c (tages): New variable.
	* sgmlxtrn.h (tages): Declare it.
	* pars1.c (parsecon): Initialize tages.
	* sgml2.c (entget): Only increment tagctr if es is equal to tages.
	(filepend): Likewise.

Sun Mar 28 16:27:09 1993  James Clark  (jjc at jclark)

	* md1.c (mdelem): Allow ( immediately after end tag minimization.

Wed Mar  3 20:09:45 1993  James Clark  (jjc at jclark)

	* pcbrf.c (da0, da0a, da1, da1a): Change state/action for RNI to
	ET0/DAF_.

Mon Mar  1 16:24:36 1993  James Clark  (jjc at jclark)

	* traceset.c (traceadl): Trace NOTATION attributes using tracedcn.

Mon Feb 22 09:59:16 1993  James Clark  (jjc at jclark)

	* Version 1.1 released.

	* md1.c (fixdatt): New function.
	(mdnadl): For attlist #notation that occurs after an entity with
	the specified notation has been declared, call fixdatt rather
	that giving an error.
	* sgmlfnsm.h: Declare fixdatt.
	* md2.c (storedatt): Remove call to adlval.
	(mdextid): Precede calls to storedatt with call to adlval.
	* msg.h: Change message 218.

	* sgmls.c (read_line): Cast UINT_MAX to unsigned.

	* Makefile.bcc: Add optimization flags for Borland C++ 3.1.

	* stklen.c: Increase to 14k.

Sat Feb  6 19:29:01 1993  James Clark  (jjc at jclark)

	* etype.h (EMM): New etdmin flag.
	* md1.c (mdelem): Set EMM flag if minus was explicitly specified
	for end tag minimization.
	(etdadl): Give an error if there's a CONREF element and the EMM
	flag was set, instead of checking OMITTAG and the EMO flag.
	(mdelem): Likewise for declared content of EMPTY.

Wed Feb  3 11:08:02 1993  James Clark  (jjc at jclark)

	* unix.cfg: Add LIMITS_H_MISSING define.
	* configure: Handle LIMITS_H_MISSING.
	* std.h: Don't include <limits.h> is LIMITS_H_MISSING is defined.
	In this case guess UINT_MAX, based on sizeof(unsigned int).

Mon Feb  1 21:49:52 1993  James Clark  (jjc at jclark)

	* sgmls.c (read_line): Don't allow buffer length to exceed
	UINT_MAX.
	(E_LINELENGTH): New error.
	* std.h: Include <limits.h>

Sun Jan 31 23:58:55 1993  James Clark  (jjc at jclark)

	* pars1.c (destack): Don't give error message if conrefsw is set.

Sat Jan 30 21:36:36 1993  James Clark  (jjc at jclark)

	* lineout.c (process_document): Add argument that says whether
	document is subdocument.  Don't output appinfo if document is
	subdocument.
	* appl.h: Change declaration.
	* main.c (main): Change caller.

Fri Jan 15 04:30:17 1993  James Clark  (jjc at jclark)

	* Makefile: Add formatted man pages to distribution.  Include
	rules to build them.

	* sgmls.c (RECHAR): Delete definition.

	* appl.h: Delete comma after last enumerator.

Mon Jan 11 09:51:43 1993  James Clark  (jjc at jclark)

	* sgmldecl.c (sdnames): Size of temp buffer should be
	REFNAMELEN+1.

Sun Jan 10 22:37:20 1993  James Clark  (jjc at jclark)

	* rast.c (do_file): Add default case to switch.

Fri Jan  8 09:57:07 1993  James Clark  (jjc at jclark)

	* synrf.c: Delete NONSGML. Add POSTLINK.
	* synxtrn.h: Corresponding changes.

Thu Jan  7 20:00:07 1993  James Clark  (jjc at jclark)

	* rast.c (do_file): Free attributes when skipping subdocument.

	* sgmls.c (sgmls_free): Free entity.u.external.filenames.

	* latin1.c, ebcdic.h (SHIFTNON, UNSHIFTNON): Use simpler
	definitions.

Wed Jan  6 17:36:50 1993  James Clark  (jjc at jclark)

	* pcbrf.c (et0a): Change action for spcr to DAS_.

	* pars2.c (parse): In case ERX_ catch references to entities that
	are not allowed in replaceable character data.
	(parselit): Delete case PIE_. In case DEF_ don't catch references
	to external data entities.

	* md1.c (mdelem): Test for length of rank stem and rank suffix was
	off by 2.

	* pars2.c (parsemd): For case NUM, use NU not NUM for lexical class.

	* md1.c (mdattdef): Uncomment check for ID attribute with default
	value.

Sun Jan  3 20:30:34 1993  James Clark  (jjc at jclark)

	* sgmldecl.h: Delete declaration of appinfo[].

	* sgmldecl.c (sdnames): Return if keyword unrecognized.

	* synxtrn.h: Delete KDTD. Renumber K*.
	* synrf.c: Delete "DTD".

	* sgmldecl.c: Move definition of REFLITLEN to...
	* entity.h:
	* md2.c (mdextid): Length of minimum literal must not exceed
	REFLITLEN.