changelog

harvest是一个下载html网页得机器人

Sat Jan  2 17:03:11 1993  James Clark  (jjc at jclark)

	* pars1.c (checkdtd): Check for undefined short reference maps if
	-u flag is used.

	* md2.c (mdsrmdef): Don't free the map on error if it was
	previously used in a USEMAP declaration.

Fri Jan  1 19:37:38 1993  James Clark  (jjc at jclark)

	* entgen.c (sysidgen): Return 0 if no non-empty filenames were found.

Thu Dec 31 02:52:38 1992  James Clark  (jjc at jclark)

	* pars2.c (charrefn): Split last part off into...
	(datachar): New function.
	(charrefa): Split part off into...
	(setcurchar): New function.
	(parse): In case NON_, call datachar().
	* sgmlfnsm.h: Declare new functions.

	* sgmls.h: Use separate length for pi.
	* sgmls.c (sgmls_next): Handle this.
	* rast.c (output_processing_instruction): Add length argument.
	(do_file): Change caller.
	* sgmlsasp.c (output_pi): Add length argument.
	(do_file): Change caller.

	* pars2.c (parse): In case NSC_, give an error.
	* pcbrf.c (pcblitc, pcbmds, pcblitv): Use SYS_ action for NSC.
	(pcbmsi): Get rid of NS4 state.  Handle NSC like NON.
	* pars2.c (charrefn): Only potentially confusing characters need
	to be prefixed with DELNONCH.
	* msg.h: Change message 4 so as not to say `in markup'.
	* pars2.c (parse): When handling SYS_, if non-SGML character was
	DELNONCH skip another character.

Wed Dec 30 11:00:15 1992  James Clark  (jjc at jclark)

	* lextoke.h: Add SP class for SPACE.  SEP is now just SEPCHAR.
	* pcbrf.c (pcbval): Change accordingly.  Don't allow RS, RE or SEPCHAR
	during tokenization.
	* lextaba.c (lextoke): Change code for space to SP.
	* pars2.c (parseseq): Test that it's not a space in addition to
	testing that its code is not SEP.
	* sgmldecl.c (sdnaming): Check that class is SP or SEP.

	* sgmlmsg.c: Define HDREE string.
	(printit): Use HDREE string when the current character is EOS.  

	* pars2.c (charrefa): Rewrite. If reading from an internal entity,
	need to copy the entity before changing it.
	* source.h (struct source): Add copied member.
	(COPIEDSW): New macro.
	* sgml2.c (entopen): Initialize COPIEDSW.
	(entget): If internal entity was copied, free copy before popping
	SCB stack.

	* entity.h (SHIFTNON, UNSHIFTNON): Move definitions to latin1.h
	* ebcdic.h (SHIFTNON, UNSHIFTNON): Provide appropriate
	definitions.

	* appl.h: Move DELNONCH, DELCDATA, DELSDATA to latin1.h and ebcdic.h.
	* ebcdic.c: Use these.
	* entity.h (struct switches): Remove delnonch, delcdata and
	delsdata members.
	* main.c (swinit): Don't initialize delnonch.
	* sgmlio.c (ioinit): Don't save swp->delnonch.
	(ioread): Use DELNONCH constant.
	* sgmlmsg.c (swdelcdata, swdelsdata, swdelnonch): Deleted.
	(msginit): Don't initialize these.
	(printit): Use DELNONCH constant.  Can now use switch statement.
	(transparm): Use constants for delimiters.
	* sgmldecl.c (sdinit): Use DEL* constants.
	* sgmldecl.c (sdnaming): Use DELNONCH constant.
	* pars2.c (tokdata): Likewise.
	(parselit): Likewise.
	* sgml1.c (sgmlset): Don't initialize nonchbuf.  Don't change
	lexical codes for sw.delnonch.
	* sgmlxtrn.c (nonchbuf): Initialize first character.
	* lextaba.c (lexcnm, lexcon, lexlms): Make code for DELNONCH be
	NSC.
	* synxtrn.h (struct lexcode): Delete lmnsc member.
	* lexrf.c (lex): Change initialization accordingly.
	* lexcode.h: Move definition of NSC to lextaba.c.

	* Get rid of BADCHAR hack by having entity escape magic characters
	with delnonch just as happens in internal entities.
	* sgmldecl.c (sdinit): Don't make BADCHAR a magic character.
	* pars2.c (parse): When handling NON_ don't check if character is
	BADCHAR.  When handling NSC_, give an error if at file level.
	* latin.h, ebcdic.h (BADCHAR): Deleted.
	* ebcdic.c: Don't use BADCHAR.
	* sgmlio.c (ioinit): Don't make BADCHAR bad. Remember
	swp->delnonch in delnonch.
	(iocont): Count a bad character as 2 characters.
	(ioread): Prefix bad character with delnonch.  Make sure there's
	always room for two characters plus end of buffer.
	* source.h (struct source): Add nextchar member.
	(NEXTC): New macro.
	* sgml2.c (scbset): Update NEXTC.
	* sgmlaux.h (struct location): Add nextchar member.
	* sgml1.c (get_location): Update nextchar.
	* sgmlmsg.c (IS_MAGIC): Deleted.
	(printit): Print the real next character if curchar is delnonch.
	* sgmlio.c (isbad): Rename to ismagic.
	* msg.h: Delete message 152.

	* appl.h (ADDNONCH): Move to entity.h.
	* entity.h (struct switches): Delete addnonch member.
	(UNSHIFTNON, SHIFTNON): New macro.
	* main.c (swinit): Don't initialize addnonch.
	* lineout.c (print_string): Use UNSHIFTNON.
	* sgmlmsg.c (transparm): Likewise.
	* sgmldecl.c (sdnaming): Likewise.
	* pars2.c (shiftnon): Deleted.
	* sgmlfnsm.h: Delete prototype.
	* pars2.c (parse, charrefn): Use SHIFTNON macro instead of
	shiftnon function.
	* action.h (NSC_): New action.
	* pars2.c (parse): Handle NSC_.
	* pcbrf.c: Undo change of Jun 10, except to pcbmsi.
	For parse modes which use NON_ action, handle NSC class by using
	NSC_ action and transforming state as for NON class.

Tue Dec 15 10:31:43 1992  James Clark  (jjc at jclark)

	* lineout.c (process_document); Don't define subdoc if suppsw is set.

	* Makefile (OBJS3): Add getopt.o.

	* sgmlio.c (ioclose): Don't call closefile if fp is 0.

Thu Dec  3 09:46:41 1992  James Clark  (jjc at jclark)

	* md1.c (mdattdef): Check that normalized length of CDATA default
	value does not exceed LITLEN.

	* entity.h (struct switches): Add die member.
	* main.c (die): New function.
	(swinit): Initialize die.
	* sgmlmsg.c (msginit): Initialize die.
	(msgprint): Call die for EXITERR errors.

	* main.c (OPTIONS): Rename to options.  Rewrite as an array of
	char so as to avoid duplication.

Tue Dec  1 07:19:47 1992  James Clark  (jjc at jclark)

	* Makefile.bcc: rast needs getopt.obj.

Mon Nov 30 15:52:03 1992  James Clark  (jjc at jclark)

	* sgmldecl.c (charset_map, capset_map, sdfeatures): Fully bracket
	initializer.
	* sgml1.c (sgmltab): Likewise.
	* sgmlxtrn.c (dctab, deftab, dvtab, enttab, exttab, extettab,
	funtab, mstab, pubcltab): Likewise.

Thu Nov 26 11:44:40 1992  James Clark  (jjc at jclark)

	* md2.c (mdentity): Check for duplicate definition and earlier
	usage of default entity after parsing complete definition.

Fri Nov 20 05:19:07 1992  James Clark  (jjc at jclark)

	* sgmlsasp.c: Delete definitions of RECHAR and RSCHAR.  Use
	definitions from config.h.

	* rast.c (output_char): Use definitions of RECHAR, RSCHAR and
	TABCHAR from config.h.

	* lexrf.c (lex): Use GENRECHAR rather than '\b' for genre field.

	* sgmldecl.h (CANON_NONSGML, CANON_DATACHAR): Move definitions
	into ascii.h and ebcdic.h.
	* genlex.c (main): Characters that aren't special in some way need
	to be treated as NONSGML or DATACHAR depending on whether they are
	the number of a shunned character.

Thu Nov 19 06:32:37 1992  James Clark  (jjc at jclark)

	* ecbdic.h: Use a better formal public identifier.

	* cms.cfg (DEFAULT_PATH, PATH_FILE_SEP, SYSID_FILE_SEP,
	MIN_DAT_SUBS_FROM, MIN_DAT_SUBS_TO): Fix definitions.

	* ebcdic.c: Include ebcdic.h and use character definitions that it
	provides.

Tue Nov 17 05:46:34 1992  James Clark  (jjc at jclark)

	* cms.cfg: New file.
	* ebcdic.h: New file.
	* Makefile.cms: New file.
	* genlex.c: New file.
	* ebcdic.c: New file.
	* Makefile (lextabe.c, genlex): New targets.
	* Makefile: Add new files to distribution.

	* lexrf.c: Separate out character set dependent parts into...
	* lextaba.c: New file.
	* lexcode.h: New file. Move definitions needed by lextaba.c and
	lexrf.c here.
	* sgmldecl.c (asciicharset): Move definition into lextaba.c.
	Move definitions of UNKNOWN etc constants into sgmldecl.h.
	* sgmldecl.h: Declare asciicharset.

	* entity.h: Separate character set dependent parts into...
	* latin1.h: New file.
	* *.cfg: Include latin1.h.

	* lexrf.c (lex): Rewrite short reference delimiter strings that use
	RS, RE or TAB so as to use RSCHAR, RECHAR and TABCHAR definitions.

	* lexrf.c (lextran): Don't use character constants.

	* lexrf.c (lextran): Fix entry for character number 172.

Mon Nov 16 12:39:41 1992  James Clark  (jjc at jclark)

	* arc*.c: Rename to *.c.
	* mod*.c: Rename to *.c.
	* Makefile, Makefile.vms, Makefile.bcc: Corresponding changes.

	* msg.h: Make message 58 a warning.  Rewrite message.

Sun Nov 15 11:41:04 1992  James Clark  (jjc at jclark)

	* main.c (munge_program_name): Use isupper() and tolower().

	* main.c (main): Don't allow -a option.

Sat Nov 14 10:34:47 1992  James Clark  (jjc at jclark)

	* lineout.c (attribute_type_string): Don't distinguish ID and
	IDREF attributes from tokenized attributes.
	* sgmls.h (sgmls_attribute_type): Delete SGMLS_ATTR_ID and
	SGMLS_ATTR_IDREF.
	* sgmls.c (parse_attribute, sgmls_free_attribute): Don't handle ID
	and IDREF attribute types.
	* sgmlsasp.c (output_attribute): Don't handle SGMLS_ATTR_ID and
	SGMLS_ATTR_IDREF.
	* rast.c (output_attribute): Likewise.

	* main.c (store_sgmldecl): Handle possibility that SIGTERM is not
	defined.

Thu Nov  5 11:01:35 1992  James Clark  (jjc at jclark)

	* msg.h: In message 3, change `NAMELEN' to  `NAMELEN or LITLEN'.

	* modmd1.c (mdattdef): Handle CDR action when parsing default
	value.

Thu Oct 29 17:19:13 1992  James Clark  (jjc at jclark)

	* lineout.c (output_conforming): Only generate output if !suppsw.

Wed Oct 28 10:29:28 1992  James Clark  (jjc at jclark)

	* rast.c (main): Add -o option to specify output file.  Avoid use of
	temporary file when -o specified.
	(usage): New function.

Sun Oct 25 10:30:18 1992  James Clark  (jjc at jclark)

	* modmd1.c (adlfree): New function.
	(mdadl, mdnadl, mdattdef): Fix some leaks.
	* sgmlfnsm.h: Declare adlfree.

	* main.c (OPTIONS): Correct definitions that apply when
	CANT_REDIRECT_STDERR is defined.

	* Makefile (dist): Fix up the compilation options in
	the Makefile.

Sat Oct 24 14:02:35 1992  James Clark  (jjc at jclark)

	* entity.h: Define STDINNAME.
	* main.c (main): Don't require an argument.
	(make_docent): Treat no arguments as equivalent to `-'.
	* entgen.c (entgen): Don't search for `-'.
	* modsgml2.c (filepend): Pass buffer as argument to
	iopend().
	* sgmlaux.h (iopend): Fix declaration.
	* sgmlio.c (openfile, closefile, isreg): New functions.
	(ioopen): Use openfile. Initialize pendbuf.
	(ioclose): Use closefile.
	(iopend, iocont): Handle unseekable files.
	(ioread): Return the contents of pendbuf if pendbuf is
	non-NULL.  Handle the case where fgetpos fails.
	Use openfile and closefile.
	* unix.cfg, dos.cfg: Add HAVE_SYS_STAT_H configuration option.
	* configure: Handle HAVE_SYS_STAT_H.

Fri Oct 23 11:38:06 1992  James Clark  (jjc at jclark)

	* Makefile.bcc: Handle rast.

	* Makefile (install): Rewrite.

	* sgmls/sgmls.c (unescape1, parse_data): Cast value of escape
	sequence.

	* lineout.h (CONFORMING_CODE): New output command.
	* lineout.c (output_conforming): New function.
	* appl.h: Declare it.
	* main.c (main): Call output_conforming() if appropriate.
	* sgmls.h: New event SGMLS_EVENT_CONFORMING.
	* sgmls.c (sgmls_next): Handle CONFORMING_CODE.
	* sgmlsasp.c (do_file): Handle SGMLS_EVENT_CONFORMING.

	* CHANGES: Rename to
	* NEWS.
	* Makefile: Change definition of MISC accordingly.

	* rast.c: New file.
	* rast.man: New file.
	* Makefile: Handle rast.
	(install.man): Rewrite.

	* main.c (print_caps): Deleted.
	(make_argv): Use -c rather than -z.  Always pass -c to
	subprocesses.
	(main): Delete -c option. Rename -z to -c.
	(OPTIONS, usage): Change accordingly.
	(capsw): Deleted.

Wed Oct 21 20:13:02 1992  James Clark  (jjc at jclark)

	* modpars2.c (attval): Decrement already tokenized value before
	comparing it to a FIXED value.

Tue Oct 20 16:45:34 1992  James Clark  (jjc at jclark)

	* modsgml2.c (filepend): Don't suspend the file if it's already
	suspended.

Sun Oct 18 15:24:13 1992  James Clark  (jjc at jclark)

	* sgmls.pl: Handle f, p, s commands.  Fix handling of S, E, and N
	commands.
	(esc, esc_data): Don't handle \s. Ignore \012.

Sat Oct 17 12:48:56 1992  James Clark  (jjc at jclark)

	* sgmls.c (parse_data): Translate \n into '\015'.
	(unescape_file): New function.
	(sgmls_next, parse_location): Use unescape_file.
	* sgmlsasp.c (output_data_char): New function.
	(output_data, output_pi): Use output_data_char.

	* sgmls.c (parse_data): Don't support \s.
	(unescape1): Likewise.

	* lineout.c (print_string): Don't strip RSs.

	* sgmls.c (add_attribute): New function.
	(sgmls_next): Use add_attribute.

	* lineout.c (process_subdoc): Handle the possibility that `id' is
	0.
	* appl.h (E_SUBDOC): Declare it.
	* main.c (errlist): Add error message for E_SUBDOC.
	* modsgml2.c (entopen): Handle the possibility that etx.x is 0 for
	external entities.
	* modmd1.c (mddtds): If entgen fails, behave as if there was no
	external DTD.
	* modmd2.c (mdentity): Don't give up if entgen fails.
	* msg.h: New message 149, 229.  Make errors 147 and 148 warnings.

	* entgen.c (field): Fix typo in handling of field 'D' for subdocs.

Fri Oct 16 09:55:40 1992  James Clark  (jjc at jclark)

	* sgmls.h (struct sgmls_event): Delete `subdoc' from union.
	(struct sgmls_external_entity): Add pubid and sysid members.
	* sgmls.c (parse_filenames): Deleted.
	(parse_external_entity, parse_subdoc_entity): Don't call
	parse_filenames.
	(parse_notation): Don't parse pubid and sysid.
	(define_external_entity): Fill in pubid, sysid and filenames
	fields from stored values in sp.
	(define_notation): Fill in pubid and sysid fields from stored
	values in sp.
	(sgmls_next): Handle SYSID_CODE, PUBID_CODE and FILE_CODE.
	For {START,END}_SUBDOC_CODE look up the subdoc entity.
	(sgmls_free): Free the pubid, sysid and files members of sp.
	Free the pubid and sysid fields of external entities.
	(struct sgmls): Add files, nfiles, sysid and pubid members.
	(sgmls_create): Initialize them.