Add tools-mode shell scripts (#34)

Author: digama0

mmhlpa.c

@@ -88,7 +88,8 @@ H("");
 
 g_printHelp = !strcmp(saveHelpCmd, "HELP ADD");
 H("This command adds a character string prefix and/or suffix to each");
-H("line in a file.");
+H("line in a file.  To add only a prefix, set <endstr> to the empty string,");
+H("and set <begstr> to the empty string to add only a suffix.");
 H("Syntax:  ADD <iofile> <begstr> <endstr>");
 
 g_printHelp = !strcmp(saveHelpCmd, "HELP TAG");
@@ -132,7 +133,7 @@ H("  Q - Do not alter characters in quotes (ignored by T and U)");
 H("  T - (Tab) Convert spaces to equivalent tabs");
 H("  U - (Untab) Convert tabs to equivalent spaces");
 H("Some other subcommands are also available:");
-H("  P - Trim parity (8th) bit from each character");
+H("  P - Clear parity (8th) bit from each character");
 H("  G - Discard garbage characters CR,FF,ESC,BS");
 H("  C - Convert to upper case");
 H("  L - Convert to lower case");
@@ -156,7 +157,8 @@ H("Note: The SUBSTITUTE command may be abbreviated by S.");
 
 g_printHelp = !strcmp(saveHelpCmd, "HELP SWAP");
 H("This command swaps the parts of each line before and after a");
-H("specified string.");
+H("specified string <middle>.");
+H("Syntax:  SWAP <iofile> <middle>");
 
 
 g_printHelp = !strcmp(saveHelpCmd, "HELP INSERT");

mmvstr.c

@@ -286,7 +286,7 @@ vstring edit(vstring sin,long control) {
   /* EDIT$ (from VMS BASIC manual)
        Syntax:  str-vbl = EDIT$(str-exp, int-exp)
        Values   Effect
-       1        Trim parity bits
+       1        Clear parity bits
        2        Discard all spaces and tabs
        4        Discard characters: CR, LF, FF, ESC, RUBOUT, and NULL
        8        Discard leading spaces and tabs
@@ -307,20 +307,20 @@ vstring edit(vstring sin,long control) {
   vstring sout;
   long i, j, k, m;
   int last_char_is_blank;
-  int trim_flag, discardctrl_flag, bracket_flag, quote_flag, case_flag;
+  int clear_parity_flag, discardctrl_flag, bracket_flag, quote_flag, uppercase_flag;
   int alldiscard_flag, leaddiscard_flag, traildiscard_flag,
       traildiscardLF_flag, reduce_flag;
   int processing_inside_quote=0;
   int lowercase_flag, tab_flag, untab_flag, screen_flag, discardcr_flag;
   unsigned char graphicsChar;
 
   /* Set up the flags */
-  trim_flag = control & 1;
+  clear_parity_flag = control & 1;
   alldiscard_flag = control & 2;
   discardctrl_flag = control & 4;
   leaddiscard_flag = control & 8;
   reduce_flag = control & 16;
-  case_flag = control & 32;
+  uppercase_flag = control & 32;
   bracket_flag = control & 64;
   traildiscard_flag = control & 128;
   traildiscardLF_flag = control & 16384;
@@ -361,8 +361,8 @@ vstring edit(vstring sin,long control) {
     if ((alldiscard_flag) && isblank_(sout[i]))
         sout[i] = 0;
 
-    /* Trim parity (eighth?) bit */
-    if (trim_flag)
+    /* Clear parity (eighth?) bit */
+    if (clear_parity_flag)
        sout[i] = sout[i] & 0x7F;
 
     /* Discard CR,LF,FF,ESC,BS */
@@ -382,13 +382,13 @@ vstring edit(vstring sin,long control) {
 
     /* Convert lowercase to uppercase */
     /*
-    if ((case_flag) && (islower(sout[i])))
+    if ((uppercase_flag) && (islower(sout[i])))
        sout[i] = toupper(sout[i]);
     */
     /* 13-Jun-2009 nm The upper/lower case C functions have odd behavior
        with characters > 127, at least in lcc.  So this was rewritten to
        not use them. */
-    if ((case_flag) && (sout[i] >= 'a' && sout[i] <= 'z'))
+    if ((uppercase_flag) && (sout[i] >= 'a' && sout[i] <= 'z'))
        sout[i] = (char)(sout[i] - ('a' - 'A'));
 
     /* Convert [] to () */

tools/add.sh

@@ -0,0 +1,19 @@
+#!/bin/sh
+# Shell script equivalent of Metamath ADD tool
+# Mario Carneiro 5-Jan-2022
+
+usage() {
+  cat << HELP
+ADD - Add a specified string to each line in a file
+
+This command adds a character string prefix and/or suffix to each
+line in a file. To add only a prefix, set ENDSTR to the empty string,
+and set BEGSTR to the empty string to add only a suffix.
+Syntax:  add.sh BEGSTR ENDSTR < INFILE > OUTFILE
+HELP
+}
+
+if [ $# -eq 1 ] && [ "$1" = "-h" -o "$1" = "--help" ]; then usage; exit; fi
+if [ $# -ne 2 ]; then usage; exit 1; fi
+
+while read line; do echo $1${line}$2; done

tools/beep.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+# Shell script equivalent of Metamath BEEP tool
+# Mario Carneiro 7-Jan-2022
+
+usage() {
+  cat << HELP
+BEEP - Make a beep sound
+
+This command will produce a beep.  By typing it ahead after a long-
+running command has started, it will alert you that the command is
+finished.
+Syntax: beep.sh
+HELP
+}
+
+if [ $# -eq 1 ] && [ "$1" = "-h" -o "$1" = "--help" ]; then usage; exit; fi
+if [ $# -ne 0 ]; then usage; exit 1; fi
+
+beep

tools/break.sh

@@ -0,0 +1,31 @@
+#!/bin/sh
+# Shell script equivalent of Metamath BREAK tool
+# Mario Carneiro 7-Jan-2022
+
+usage() {
+  cat << HELP
+BREAK - Break up (parse) a file into a list of tokens (one per line)
+
+This command breaks up a file into tokens, one per line, breaking at
+whitespace and any special characters you specify as delimiters.
+Use an explicit (quoted) space as SPECCHARS to avoid the default
+special characters - ()[],=:;{} - and break only on whitespace.
+Syntax:  break.sh SPECCHARS < INFILE > OUTFILE
+HELP
+}
+
+if [ $# -ne 1 ]; then usage; exit 1; fi
+
+if [ "$1" = "" ]; then specChars="()[],=:;{}"; else specChars=$1; fi
+
+awk -v specChars="$specChars" \
+'BEGIN {
+  gsub(/[\\\]\-\^]/, "\\\\&", specChars);
+  pattern = "[" specChars "]";
+} {
+  gsub(pattern, " & ");
+  gsub(/^[ \t]+/, "");
+  gsub(/[ \t]+$/, "");
+  gsub(/[ \t]+/, "\n");
+  if ($0 != "") print;
+}'

tools/build.sh

@@ -0,0 +1,32 @@
+#!/bin/sh
+# Shell script equivalent of Metamath BUILD tool
+# Mario Carneiro 7-Jan-2022
+
+usage() {
+  cat << HELP
+BUILD - Build a file with multiple tokens per line from a list
+
+This command combines a list of tokens into multiple tokens per line,
+as many as will fit per line, separating them with spaces.
+The line length is hard-coded to 73 characters long.
+Syntax:  build.sh < INFILE > OUTFILE
+HELP
+}
+
+if [ $# -eq 1 ] && [ "$1" = "-h" -o "$1" = "--help" ]; then usage; exit; fi
+if [ $# -ne 0 ]; then usage; exit 1; fi
+
+awk '{
+  if ($0 != "") {
+    if (out == "")
+      out = $0;
+    else if (length(out) + length($0) > 72) {
+      print out;
+      out = $0;
+    } else {
+      out = out " " $0;
+    }
+  }
+} END {
+  if (out != "") print out;
+}'

tools/clean.sh

@@ -0,0 +1,109 @@
+#!/bin/sh
+# Shell script equivalent of Metamath CLEAN tool
+# Mario Carneiro 5-Jan-2022
+
+usage() {
+  cat << HELP
+CLEAN - Trim spaces and tabs on each line in a file; convert characters
+
+This command processes spaces and tabs in each line of a file
+according to the following subcommands:
+  D - Delete all spaces and tabs
+  B - Delete spaces and tabs at the beginning of each line
+  E - Delete spaces and tabs at the end of each line
+  R - Reduce multiple spaces and tabs to one space
+  Q - Do not alter characters in quotes (ignored by T and U)
+  T - (Tab) Convert spaces to equivalent tabs (unsupported)
+  U - (Untab) Convert tabs to equivalent spaces (unsupported)
+Some other subcommands are also available:
+  P - Clear parity (8th) bit from each character
+  G - Discard garbage characters CR,FF,ESC,BS
+  C - Convert to upper case
+  L - Convert to lower case
+  V - Convert VT220 screen print frame graphics to -,|,+ characters
+Subcommands are separated by spaces, e.g., B E R Q
+Syntax:  clean.sh subcmd subcmd ... < INFILE > OUTFILE
+HELP
+}
+
+if [ $# -eq 1 ] && [ "$1" = "-h" -o "$1" = "--help" ]; then usage; exit; fi
+if [ $# -eq 0 ]; then usage; exit 1; fi
+
+for var in "$@"; do
+  case $var in
+    D | d) allDiscard=1;;
+    B | b) leadDiscard=1;;
+    E | e) trailDiscard=1;;
+    R | r) reduce=1;;
+    Q | q) quote=1;;
+    T | t) echo "T (tab) flag unsupported"; exit 2;;
+    U | u) echo "U (untab) flag unsupported"; exit 2;;
+    P | p) clearParity=1;;
+    G | g) discardCtrl=1;;
+    C | c) uppercase=1;;
+    L | l) lowercase=1;;
+    V | v) screen=1;;
+  esac
+done
+
+awk \
+  -v clearParity=$clearParity \
+  -v allDiscard=$allDiscard \
+  -v discardCtrl=$discardCtrl \
+  -v leadDiscard=$leadDiscard \
+  -v reduce=$reduce \
+  -v uppercase=$uppercase \
+  -v trailDiscard=$trailDiscard \
+  -v quote=$quote \
+  -v lowercase=$lowercase \
+  -v tab=$tab \
+  -v untab=$untab \
+  -v screen=$screen \
+'BEGIN {
+  FS="";
+  if (clearParity || uppercase || lowercase) {
+    for (n=0; n<256; n++) {
+      ch = sprintf("%c", n);
+      chr[n] = ch;
+      ord[ch] = n;
+    }
+  }
+  if (discardCtrl) {
+    discardArr["\r"] = 1;
+    discardArr["\n"] = 1;
+    discardArr["\f"] = 1;
+    discardArr["\x27"] = 1;
+    discardArr["\x08"] = 1;
+  }
+  if (screen) {
+    gfxArr["\xEA"] = gfxArr["\xEB"] = gfxArr["\xEC"] = gfxArr["\xED"] = "+";
+    gfxArr["\xDA"] = gfxArr["\xD9"] = gfxArr["\xBF"] = gfxArr["\xC0"] = "+";
+    gfxArr["\xF1"] = gfxArr["\xC4"] = "-";
+    gfxArr["\xF8"] = gfxArr["\xA6"] = gfxArr["\xB3"] = "|";
+  }
+} {
+  if (leadDiscard)
+    gsub(/^[ \t]+/, "");
+  inQuote = 0;
+  out = "";
+  for (i=1;i<=NF;i++) {
+    if (quote && ($i == "\"" || $i == "\x39"))
+      inQuote = !inQuote;
+    if (inQuote) {
+      out=out $i;
+      continue;
+    }
+    if (allDiscard && ($i == " " || $i == "\t")) continue;
+    if (clearParity) $i = chr[and(ord[$i], 0x7f)];
+    if (discardCtrl && discardArr[$i]) continue;
+    if (uppercase && ord[$i] < 128) $i = toupper($i);
+    if (lowercase && ord[$i] < 128) $i = tolower($i);
+    if (screen && gfxArr[$i]) $i = gfxArr[$i];
+    out=out $i;
+  }
+  if (trailDiscard)
+    gsub(/[ \t]+$/, "", out);
+  if (reduce)
+    gsub(/[ \t]+/, " ", out);
+  print out;
+}'

tools/copy.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+# Shell script equivalent of Metamath COPY tool
+# Mario Carneiro 7-Jan-2022
+
+usage() {
+  cat << HELP
+COPY - Similar to Unix "cat" but safe (same input & output name allowed)
+
+This command copies (concatenates) all input files in a space separated
+list to an output file.  The output file may have
+the same name as an input file.
+Unlike the metamath version, this script does not save a previous
+version of the output file with a ~1 extension. (TODO: implement?)
+Example: "copy.sh 1.tmp 1.tmp 1.tmp 2.tmp" followed by "unique.sh 1.tmp"
+will result in 1.tmp containing those lines of 2.tmp that didn't
+previously exist in 1.tmp.
+Syntax:  copy.sh OUTFILE INFILE INFILE ...
+HELP
+}
+
+if [ $# -eq 1 ] && [ "$1" = "-h" -o "$1" = "--help" ]; then usage; exit; fi
+
+outfile=$1; shift
+buf=$(cat $@)
+echo "$buf" > "$outfile"

tools/delete.sh

@@ -0,0 +1,36 @@
+#!/bin/sh
+# Shell script equivalent of Metamath DELETE tool
+# Mario Carneiro 5-Jan-2022
+
+usage() {
+  cat << HELP
+DELETE - Delete a section of each line in a file
+
+This command deletes the part of a line between (and including) the first
+occurrence of STARTSTR and the first occurrence of ENDSTR (when both
+exist) for all lines in a file.  If either string doesn't exist in a line,
+the line will be unchanged.  If STARTSTR is blank (''), the deletion
+will start from the beginning of the line.  If ENDSTR is blank, the
+deletion will end at the end of the line.
+Syntax:  delete.sh STARTSTR ENDSTR < INFILE > OUTFILE
+HELP
+}
+
+if [ $# -eq 1 ] && [ "$1" = "-h" -o "$1" = "--help" ]; then usage; exit; fi
+if [ $# -ne 2 ]; then usage; exit 1; fi
+
+awk -v startStr="$1" -v endStr="$2" '{
+  start = index($0, startStr);
+  if (start) {
+    if (endStr == "") {
+      $0 = substr($0, 1, start - 1);
+    } else {
+      start2 = index(substr($0, start), endStr);
+      if (start2) {
+        end = start - 1 + start2 + length(endStr);
+        $0 = substr($0, 1, start - 1) substr($0, end);
+      }
+    }
+  }
+  print;
+}'

tools/duplicate.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+# Shell script equivalent of Metamath DUPLICATE tool
+# Mario Carneiro 7-Jan-2022
+
+usage() {
+  cat << HELP
+DUPLICATE - Extract first occurrence of any line occurring more than
+      once in a file, discarding lines occurring exactly once
+
+This command finds all duplicate lines in a file and places them, in
+sorted order, into the output file.
+Syntax:  duplicate.sh < INFILE > OUTFILE
+HELP
+}
+
+if [ $# -eq 1 ] && [ "$1" = "-h" -o "$1" = "--help" ]; then usage; exit; fi
+if [ $# -ne 0 ]; then usage; exit 1; fi
+
+sort | uniq -d