*** ir/irtfiles.c.orig	Wed Apr  7 13:39:40 1993
--- ir/irtfiles.c	Fri Apr 16 19:13:56 1993
***************
*** 743,749 ****
  boolean check_for_file_already_indexed;
  boolean word_position, word_pairs;
  {
!   /* Addes words to the index for a given file.
     * The function arguments can be NULL which means it would
     *  always answer NULL.
     * separator_function is called on every line to see if it
--- 743,752 ----
  boolean check_for_file_already_indexed;
  boolean word_position, word_pairs;
  {
!   /* Adds words to the index for a given file.
!    * "Words" are extracted as strings of alphanumeric chars, whose
!    *  length is >=3 but <toobig.  Long lines are handled in chunks
!    *  so that binary files with embedded text are correctly processed.
     * The function arguments can be NULL which means it would
     *  always answer NULL.
     * separator_function is called on every line to see if it
***************
*** 779,784 ****
--- 782,790 ----
    long file_position_before_document = 0;
    long date;

+   int charsleftover;
+   char leftovers[MAX_LINE_LENGTH];
+
  
 
     if(filename != NULL &&
***************
*** 887,892 ****
--- 893,899 ----
    the_document_table_entry.number_of_lines = 0;
    the_document_table_entry.date = 0;

+   charsleftover = 0;
    while(TRUE){
      long line_length;
      boolean newline_terminated;
***************
*** 894,909 ****
      char header[MAX_LINE_LENGTH];
      char* read_line_result;
      boolean eof;

!     /* printf("ftell: %ld\n", ftell(input_stream)); */
!     /* read a line */
!     read_line_result  = fgets(line, MAX_LINE_LENGTH, input_stream);
      beFriendly();
!  	
!     /* eof = feof(input_stream); */ /* zero means not eof */
!     eof	= !read_line_result;
    
!     the_document_table_entry.number_of_lines++;

      header[0] = '\0';		/* set it to the empty string */

--- 901,941 ----
      char header[MAX_LINE_LENGTH];
      char* read_line_result;
      boolean eof;
+     int i, cut, charsread;

!     /*
!      *  Read a block of up to MAX_LINE_LENGTH chars from a single "line"
!      *	and extract a whole number of words from it.  Save partial words
!      *	lying at edge of block in the leftovers[] array for the next loop.
!      *	This accomodates binary files formats with embedded text, which
!      *	often have very long "lines".   [burchard@geom.umn.edu 4/16/93]
!      */
!   
!     /* prefill line with leftovers before reading in new data */
      beFriendly();
!     for(i=0; i<charsleftover; i++) line[i] = leftovers[i];
!     line[charsleftover] = '\0';
!     read_line_result = fgets(line + charsleftover,
!       MAX_LINE_LENGTH - charsleftover, input_stream);
!
!     /* increment line count if we have finished reading a "line" */
!     charsread = strlen(line);
!     if(charsread>charsleftover && line[charsread-1]=='\n')
!       the_document_table_entry.number_of_lines++;
!
!     /* save word frag at end as leftovers (unless it fills whole block) */
!     for(cut=charsread-1; cut>=0; cut--)
!       if(!isascii(line[cut]) || !isalnum(line[cut])) break;
!     if(++cut <= 0) charsleftover = 0;
!     else {
!       for(charsleftover=0, i=cut; i<charsread; charsleftover++, i++)
!         leftovers[charsleftover] = line[i];
!       charsread -= charsleftover;
!     }
!     line[charsread] = '\0';
    
!     /* don't say EOF yet if we still have leftovers to process */
!     eof = (!read_line_result && !charsread && !charsleftover);

      header[0] = '\0';		/* set it to the empty string */

***************
*** 935,941 ****
  	sprintf(header, "%s   %s", pathname_name(full_path),
  		pathname_directory(full_path, directory));
        }
!       the_document_table_entry.number_of_lines--; /* dont count separator */
        /* finish off the last */
        finish_document( dataops->extraheaderweight,
  		      header, line, document_id,
--- 967,974 ----
  	sprintf(header, "%s   %s", pathname_name(full_path),
  		pathname_directory(full_path, directory));
        }
!       if(the_document_table_entry.number_of_lines > 0)
!         the_document_table_entry.number_of_lines--; /* dont count separator */
        /* finish off the last */
        finish_document( dataops->extraheaderweight,
  		      header, line, document_id,
*** ir/ircfiles.c.orig	Wed Apr  7 13:37:19 1993
--- ir/ircfiles.c	Sun Apr 18 06:34:42 1993
***************
*** 374,448 ****

  #define pir_data_tab	16

! /* pir Flat-file format:
! ENTRY           CCHU       #Type Protein
  12345678901234567890123456789012345678901234567890123456789012345678901234567890
  .........1.........2.........3.........4.........5.........6.........7.........8
  all data starts at tab=17 or further

! ENTRY       blah	<< Start entry, index it
! TITLE	    blah   	<< Index def line == HEADER line
! ACCESSION   blah	<< Index acc line
! KEYWORDS    blah	<< index keywords
! SOURCE      blah	<< index source
! REFERENCE   blah	<< Index
! SUPERFAMILY blah	<< Index
!             blah	<< Index
! ANYOTHERS	jazz	<< skipit
! any word starting w/ "#", skipit
! ///  				<< end of entry == entry separator
! ENTRY           CCHU       #Type Protein
! TITLE           Cytochrome c - Human
! DATE            #Sequence 30-Sep-1991 #Text 30-Sep-1991
! PLACEMENT          1.0    1.0    1.0    1.0    1.0
! SOURCE          Homo sapiens #Common-name man
! ACCESSION       A31764\ A05676\ A00001
! REFERENCE
!    #Authors     Evans M.J., Scarpulla R.C.
!    #Journal     Proc. Natl. Acad. Sci. U.S.A. (1988) 85:9625-9629
!    #Title       The human somatic cytochrome c gene: two classes of
!                   processed pseudogenes demarcate a period of rapid
!                   molecular evolution.
!    #Reference-number A31764
!    #Accession   A31764
!    #Molecule-type DNA
!    #Residues    1-105 <EVA>
!    #Cross-reference GB:M22877
! REFERENCE
!    #Authors     Matsubara H., Smith E.L.
!    #Journal     J. Biol. Chem. (1963) 238:2732-2753
!    #Reference-number A05676
!    #Accession   A05676
!    #Molecule-type protein
!    #Residues    2-28;29-46;47-100;101-105 <MATS>
! REFERENCE
!    #Authors     Matsubara H., Smith E.L.
!    #Journal     J. Biol. Chem. (1962) 237:3575-3576
!    #Reference-number A00001
!    #Comment     66-Leu is found in 10% of the molecules in pooled
!                   protein.
! GENETIC
!    #Introns        57/1
! SUPERFAMILY     #Name cytochrome c
! KEYWORDS        acetylation\ electron transport\ heme\
!                   mitochondrion\ oxidative phosphorylation\
!                   polymorphism\ respiratory chain
! FEATURE
!    2-105                   #Protein cytochrome c (experimental)
!                              <MAT>\
!    2                       #Modified-site acetylated amino end
!                              (experimental)\
!    15,18                   #Binding-site heme (covalent)\
!    19,81                   #Binding-site heme iron (axial ligands)
! SUMMARY       #Molecular-weight 11749  #Length 105  #Checksum  3247
! SEQUENCE
!                 5        10        15        20        25        30
!       1 M G D V E K G K K I F I M K C S Q C H T V E K G G K H K T G
!      31 P N L H G L F G R K T G Q A P G Y S Y T A A N K N K G I I W
!      61 G E D T L M E Y L E N P K K Y I P G T K M I F V G I K K K E
!      91 E R A D L I A Y L K K A T N E
! ///
!
  *****/


--- 374,448 ----

  #define pir_data_tab	16

! /* pir Flat-file format (lines are prepended with '|' to keep cpp happy):
! |ENTRY           CCHU       #Type Protein
  12345678901234567890123456789012345678901234567890123456789012345678901234567890
  .........1.........2.........3.........4.........5.........6.........7.........8
  all data starts at tab=17 or further

! |ENTRY       blah	<< Start entry, index it
! |TITLE	    blah   	<< Index def line == HEADER line
! |ACCESSION   blah	<< Index acc line
! |KEYWORDS    blah	<< index keywords
! |SOURCE      blah	<< index source
! |REFERENCE   blah	<< Index
! |SUPERFAMILY blah	<< Index
! |            blah	<< Index
! |ANYOTHERS	jazz	<< skipit
! |any word starting w/ "#", skipit
! |///  				<< end of entry == entry separator
! |ENTRY           CCHU       #Type Protein
! |TITLE           Cytochrome c - Human
! |DATE            #Sequence 30-Sep-1991 #Text 30-Sep-1991
! |PLACEMENT          1.0    1.0    1.0    1.0    1.0
! |SOURCE          Homo sapiens #Common-name man
! |ACCESSION       A31764\ A05676\ A00001
! |REFERENCE
! |   #Authors     Evans M.J., Scarpulla R.C.
! |   #Journal     Proc. Natl. Acad. Sci. U.S.A. (1988) 85:9625-9629
! |   #Title       The human somatic cytochrome c gene: two classes of
! |                  processed pseudogenes demarcate a period of rapid
! |                  molecular evolution.
! |   #Reference-number A31764
! |   #Accession   A31764
! |   #Molecule-type DNA
! |   #Residues    1-105 <EVA>
! |   #Cross-reference GB:M22877
! |REFERENCE
! |   #Authors     Matsubara H., Smith E.L.
! |   #Journal     J. Biol. Chem. (1963) 238:2732-2753
! |   #Reference-number A05676
! |   #Accession   A05676
! |   #Molecule-type protein
! |   #Residues    2-28;29-46;47-100;101-105 <MATS>
! |REFERENCE
! |   #Authors     Matsubara H., Smith E.L.
! |   #Journal     J. Biol. Chem. (1962) 237:3575-3576
! |   #Reference-number A00001
! |   #Comment     66-Leu is found in 10% of the molecules in pooled
! |                  protein.
! |GENETIC
! |   #Introns        57/1
! |SUPERFAMILY     #Name cytochrome c
! |KEYWORDS        acetylation\ electron transport\ heme\
! |                  mitochondrion\ oxidative phosphorylation\
! |                  polymorphism\ respiratory chain
! |FEATURE
! |   2-105                   #Protein cytochrome c (experimental)
! |                             <MAT>\
! |   2                       #Modified-site acetylated amino end
! |                             (experimental)\
! |   15,18                   #Binding-site heme (covalent)\
! |   19,81                   #Binding-site heme iron (axial ligands)
! |SUMMARY       #Molecular-weight 11749  #Length 105  #Checksum  3247
! |SEQUENCE
! |                5        10        15        20        25        30
! |      1 M G D V E K G K K I F I M K C S Q C H T V E K G G K H K T G
! |     31 P N L H G L F G R K T G Q A P G Y S Y T A A N K N K G I I W
! |     61 G E D T L M E Y L E N P K K Y I P G T K M I F V G I K K K E
! |     91 E R A D L I A Y L K K A T N E
! |///
! |
  *****/


***************
*** 1054,1073 ****
   * ==========================================
   */

! /*------ example
! #Abnormal:  see A
! #abnormal abdomen:  see a( )
! #Abnormal abdomen:  see A
! # abnormal eye:  see mit15
! #abnormal oocytes:  see abo
! #abnormal tergites:  see abt
! #abnormal wings:  see abw
! #abo:  abnormal oocyte
!  location:  2-44.0 (mapped with respect to J, 2-41).
!  origin:  Naturally occurring allele recovered near Rome,
!    Italy.
!  references:  Sandler, Lindsley, Nicoletti, and Trippa,
!  ...
  ----*/


--- 1054,1073 ----
   * ==========================================
   */

! /*------ example (lines are prepended with '|' to keep cpp happy):
! |#Abnormal:  see A
! |#abnormal abdomen:  see a( )
! |#Abnormal abdomen:  see A
! |# abnormal eye:  see mit15
! |#abnormal oocytes:  see abo
! |#abnormal tergites:  see abt
! |#abnormal wings:  see abw
! |#abo:  abnormal oocyte
! | location:  2-44.0 (mapped with respect to J, 2-41).
! | origin:  Naturally occurring allele recovered near Rome,
! |   Italy.
! | references:  Sandler, Lindsley, Nicoletti, and Trippa,
! | ...
  ----*/


***************
*** 2537,2544 ****
   */


- #ifdef NeXT /* only do this if it is on a NeXT */
-
  /*----------------------- FSA -------------------*/
  #define fsa_max_edges 4
  #define fsa_error_state (-1)
--- 2537,2542 ----
***************
*** 2802,2808 ****
      return;
  }

- #endif /* def NeXT */


  /* ==============================
--- 2800,2805 ----
*** ir/irbuild.c.orig	Fri Apr  9 14:37:01 1993
--- ir/irbuild.c	Sun Apr 18 06:50:56 1993
***************
*** 188,196 ****
    fprintf(stderr,"           | refer /* refer format */\n");
    fprintf(stderr,"           | rn /* netnews saved by the [rt]?rn newsreader */\n");
    fprintf(stderr,"           | server /* server structures for the dir of servers */\n");
- #ifdef NeXT
    fprintf(stderr,"           | objc /* objective-C .h and .m files */\n");
- #endif /* def NeXT */
    fprintf(stderr,"           | tiff /* tiff files, only indexes the filename */\n");
  #ifdef BIO
    fprintf(stderr,"           | genbank  /* GenBank flatfile format */\n");
--- 188,194 ----
*** ir/sersrch.c.orig	Fri Apr  9 10:02:37 1993
--- ir/sersrch.c	Sun Apr 18 07:17:49 1993
***************
*** 52,58 ****

  #include <math.h>

- #define MAXINT (unsigned long)2^(sizeof(long)*8-1)
  #define VALUE 1000000L
  /* francois */
  #include "stemmer.h" 
--- 52,57 ----
***************
*** 760,766 ****
                txt_stream = s_fopen(txt_filename, "rb");
                strcpy(prev_txt_filename, txt_filename);
                }
!             else if (0 != strcmp(txt_stream, prev_txt_filename)) {
                s_fclose(txt_stream);
                txt_stream = s_fopen(txt_filename, "rb");
                strcpy(prev_txt_filename, txt_filename);		/* 2/92 GS TLG */
--- 759,765 ----
                txt_stream = s_fopen(txt_filename, "rb");
                strcpy(prev_txt_filename, txt_filename);
                }
!             else if (0 != strcmp(txt_filename, prev_txt_filename)) {
                s_fclose(txt_stream);
                txt_stream = s_fopen(txt_filename, "rb");
                strcpy(prev_txt_filename, txt_filename);		/* 2/92 GS TLG */
*** ir/Makefile.orig	Fri Apr  9 11:29:22 1993
--- ir/Makefile	Sun Apr 18 07:33:58 1993
***************
*** 195,200 ****
--- 195,202 ----
  # for newer BSD that needs to use <sys/dir.h>, add -DBSD43
  # for System V add -DSYSV
  # for XENIX add -M3e -Zi
+ # for NeXTSTEP add -DNeXT
+ #
  # -DSECURE_SERVER for waisserver to setuid to uucp after startup.
  #
  # -- dgg additions for biology data
*** lib/scandir.c.orig	Wed Apr  7 14:17:09 1993
--- lib/scandir.c	Sun Apr 18 06:54:02 1993
***************
*** 23,29 ****

  int
  scandir(name, list, selector, sorter)
!     char		  *name;
      struct dirent		***list;
      int			 (*selector)();
      int			 (*sorter)();
--- 23,29 ----

  int
  scandir(name, list, selector, sorter)
!     const char		  *name;
      struct dirent		***list;
      int			 (*selector)();
      int			 (*sorter)();
*** Makefile.orig	Fri Apr  9 15:18:42 1993
--- Makefile	Sun Apr 18 07:40:04 1993
***************
*** 48,56 ****
  # for other UNIX-like OS's set this to this directory.
  #TOP:sh = pwd
  # or fill in the blank for other OS's
! #TOP = ?
! comment-me:
! 	@echo You must set "\$$(TOP)" to point to the freeWAIS src directory

  SUPDIR = $(TOP)/ir

--- 48,56 ----
  # for other UNIX-like OS's set this to this directory.
  #TOP:sh = pwd
  # or fill in the blank for other OS's
! TOP = /u/burchard/Pending/freeWAIS
! #comment-me:
! #	@echo You must set "\$$(TOP)" to point to the freeWAIS src directory

  SUPDIR = $(TOP)/ir

***************
*** 62,67 ****
--- 62,68 ----
  # for XENIX add -M3e -Zi
  # USG for Unix Dirent in lib
  # for SGIs running IRIX 4.0.1, add -cckr
+ # for NeXTSTEP add -DNeXT
  #
  # For a little better security in the server, add -DSECURE_SERVER
  # this sets the server user id to -u argument after startup.
***************
*** 93,99 ****
  # -DLIST_STEMS to show stemmed words in server log and indexer output
  #

! CFLAGS = -g -I$(SUPDIR) -DSECURE_SERVER -DRELEVANCE_FEEDBACK -DUSG -DBOOLEANS -DPARTIALWORD -DLITERAL
  MFLAGS = -k

  MAKE = make $(MFLAGS)
--- 94,100 ----
  # -DLIST_STEMS to show stemmed words in server log and indexer output
  #

! CFLAGS = -g -I$(SUPDIR) -DNeXT -DSECURE_SERVER -DRELEVANCE_FEEDBACK -DUSG -DBOOLEANS -DPARTIALWORD -DLITERAL
  MFLAGS = -k

  MAKE = make $(MFLAGS)
***************
*** 122,128 ****
  # you may have to edit the CFLAGS in the Makefile yourself.

  x::
! 	cd x; (env TOP=$(TOP) CC=$(CC) CFLAGS="$(CFLAGS)" MAKE="$(MAKE)" ./makex.sh)

  bin::
  	cd bin; $(MAKE) CC=$(CC) CFLAGS="$(CFLAGS)" TOP=$(TOP)
--- 123,131 ----
  # you may have to edit the CFLAGS in the Makefile yourself.

  x::
! #	cd x; (env TOP=$(TOP) CC=$(CC) CFLAGS="$(CFLAGS)" MAKE="$(MAKE)" ./makex.sh)
! 	cd x; (export TOP CC CFLAGS MAKE; \
! 	TOP=$(TOP) CC=$(CC) CFLAGS="$(CFLAGS)" MAKE="$(MAKE)"; ./makex.sh)

  bin::
  	cd bin; $(MAKE) CC=$(CC) CFLAGS="$(CFLAGS)" TOP=$(TOP)
*** ir/ircfiles.c.orig	Wed Apr  7 13:37:19 1993
--- ir/ircfiles.c	Tue Apr 27 20:38:10 1993
***************
*** 2710,2716 ****
  		if((wobjc_header_end - wobjc_header)<(MAX_HEADER_LEN-5)
  		    && !(strchr(WOBJC_BLANK, *(wobjc_header_end-1))
  			&& strchr(WOBJC_BLANK, input)))
! 		    { *wobjc_header_end+= input; *wobjc_header_end = 0; }
  	    }
  	  

--- 2708,2714 ----
  		if((wobjc_header_end - wobjc_header)<(MAX_HEADER_LEN-5)
  		    && !(strchr(WOBJC_BLANK, *(wobjc_header_end-1))
  			&& strchr(WOBJC_BLANK, input)))
! 		    { *wobjc_header_end++ = input; *wobjc_header_end = 0; }
  	    }
  	  

***************
*** 2725,2734 ****
  		    || (wobjc_context==14 && !wobjc_header_end))
  		    wobjc_header_end = wobjc_header;
  		if((wobjc_class_end - wobjc_class_end)<(MAX_HEADER_LEN/2))
! 		    { *wobjc_class_end+= input; *wobjc_class_end = 0; }
  		if((wobjc_context==13 || wobjc_context==14)
  		    && (wobjc_header_end-wobjc_header_end)<(MAX_HEADER_LEN/2))
! 		    { *wobjc_header_end+= input; *wobjc_header_end = 0; }
  	    }
  	}
  	
--- 2723,2732 ----
  		    || (wobjc_context==14 && !wobjc_header_end))
  		    wobjc_header_end = wobjc_header;
  		if((wobjc_class_end - wobjc_class_end)<(MAX_HEADER_LEN/2))
! 		    { *wobjc_class_end++ = input; *wobjc_class_end = 0; }
  		if((wobjc_context==13 || wobjc_context==14)
  		    && (wobjc_header_end-wobjc_header_end)<(MAX_HEADER_LEN/2))
! 		    { *wobjc_header_end++ = input; *wobjc_header_end = 0; }
  	    }
  	}
  	
***************
*** 2787,2797 ****
  {
      char *p;
    
-
      /* Flush terminal blanks and balance opening '[' if any. */
      for(p=wobjc_header+strlen(wobjc_header);
      	p>wobjc_header && strchr(WOBJC_BLANK, *(p-1)); p--);
!     if(wobjc_header[0]=='+' || wobjc_header[0]=='-') *p+= ']';
      *p = 0;
    

--- 2785,2794 ----
  {
      char *p;
    
      /* Flush terminal blanks and balance opening '[' if any. */
      for(p=wobjc_header+strlen(wobjc_header);
      	p>wobjc_header && strchr(WOBJC_BLANK, *(p-1)); p--);
!     if(wobjc_header[0]=='+' || wobjc_header[0]=='-') *p++ = ']';
      *p = 0;
    

*** ir/ircfiles.h.orig	Wed Apr  7 13:49:54 1993
--- ir/ircfiles.h	Tue Apr 27 21:09:28 1993
***************
*** 187,198 ****
  void nhyp_header_function _AP((char *line));
  void nhyp_finish_header_function _AP((char *header));

! #ifdef NeXT
! /* objective-C code */
  boolean wobjc_separator_function _AP((char *line));
  void wobjc_header_function _AP((char *line));
  void wobjc_finish_header_function _AP((char *header));
- #endif /* def NeXT */

  /* Ziff files */
  boolean ziff_separator_function _AP((char *line));
--- 187,196 ----
  void nhyp_header_function _AP((char *line));
  void nhyp_finish_header_function _AP((char *header));

! /* Objective-C code */
  boolean wobjc_separator_function _AP((char *line));
  void wobjc_header_function _AP((char *line));
  void wobjc_finish_header_function _AP((char *header));

  /* Ziff files */
  boolean ziff_separator_function _AP((char *line));