Add support for -d $'\n' (cut by line!) and posix -nb (wraps to start of -c)

2 files changed, 59 insertions, 34 deletions

diff --git a/tests/cut.test b/tests/cut.test
index 8d8c4ba1..889fc186 100755
--- a/tests/cut.test
+++ b/tests/cut.test
@@ -13,29 +13,27 @@ echo "one:two:three:four:five:six:seven
 alpha:beta:gamma:delta:epsilon:zeta:eta:theta:iota:kappa:lambda:mu
 the quick brown fox jumps over the lazy dog" >abc.txt
 
-testing "-b a,a,a" "cut -b 3,3,3 abc.txt" "e\np\ne\n" "" ""
-testing "-b overlaps" "cut -b 1-3,2-5,7-9,9-10 abc.txt" \
+testcmd "-b a,a,a" "-b 3,3,3 abc.txt" "e\np\ne\n" "" ""
+testcmd "-b overlaps" "-b 1-3,2-5,7-9,9-10 abc.txt" \
   "one:to:th\nalphabeta\nthe qick \n" "" ""
-testing "-b encapsulated" "cut -b 3-8,4-6 abc.txt" "e:two:\npha:be\ne quic\n" \
+testcmd "-b encapsulated" "-b 3-8,4-6 abc.txt" "e:two:\npha:be\ne quic\n" \
   "" ""
-testing "-bO overlaps" \
-  "cut --output-delimiter ' ' -b 1-3,2-5,7-9,9-10 abc.txt" \
+testcmd "-bO overlaps" "--output-delimiter ' ' -b 1-3,2-5,7-9,9-10 abc.txt" \
   "one:t o:th\nalpha beta\nthe q ick \n" "" ""
-testing "high-low error" "cut -b 8-3 abc.txt 2>/dev/null || echo err" "err\n" \
+testcmd "high-low error" "-b 8-3 abc.txt 2>/dev/null || echo err" "err\n" \
   "" ""
 
-testing "-c a-b" "cut -c 4-10 abc.txt" ":two:th\nha:beta\n quick \n" "" ""
-testing "-c a-" "cut -c 41- abc.txt" "\ntheta:iota:kappa:lambda:mu\ndog\n" "" ""
-testing "-c -b" "cut -c -39 abc.txt" \
+testcmd "-c a-b" "-c 4-10 abc.txt" ":two:th\nha:beta\n quick \n" "" ""
+testcmd "-c a-" "-c 41- abc.txt" "\ntheta:iota:kappa:lambda:mu\ndog\n" "" ""
+testcmd "-c -b" "-c -39 abc.txt" \
   "one:two:three:four:five:six:seven\nalpha:beta:gamma:delta:epsilon:zeta:eta\nthe quick brown fox jumps over the lazy\n" \
   "" ""
-testing "-c a" "cut -c 40 abc.txt" "\n:\n \n" "" ""
-testing "-c a,b-c,d" "cut -c 3,5-7,10 abc.txt" "etwoh\npa:ba\nequi \n" "" ""
-toyonly testing "-c japan.txt" 'cut -c 3-6,9-12 "$FILES/utf8/japan.txt"' \
+testcmd "-c a" "-c 40 abc.txt" "\n:\n \n" "" ""
+testcmd "-c a,b-c,d" "-c 3,5-7,10 abc.txt" "etwoh\npa:ba\nequi \n" "" ""
+toyonly testcmd "-c japan.txt" '-c 3-6,9-12 "$FILES/utf8/japan.txt"' \
   "ガラスをられます\n" "" ""
 
-toyonly testing "-C test1.txt" 'cut -C -1 "$FILES/utf8/test1.txt"' \
-  "l̴̗̞̠\n" "" ""
+toyonly testcmd "-C test1.txt" '-C -1 "$FILES/utf8/test1.txt"' "l̴̗̞̠\n" "" ""
 
 # substitute for awk
 toyonly testcmd "-DF" "-DF 2,7,5" \
@@ -47,24 +45,26 @@ Weather forecast for tonight : dark.
 Apple: you can buy better, but you can't pay more.
 Subcalifragilisticexpialidocious.
 Auntie Em: Hate you, hate Kansas. Took the dog. Dorothy."
+toyonly testcmd "-DF 2" "-DF 7,1,3-6,2-5" \
+  "seven one three four five six two three four five\n" "" \
+  "one two three four five six seven eight nine\n"
 
 testcmd "empty field" "-d ':' -f 1-3" "a::b\n" "" "a::b\n"
 testcmd "empty field 2" "-d ':' -f 3-5" "b::c\n" "" "a::b::c:d\n"
 
-testing "-f a-" "cut -d ':' -f 5- abc.txt" "five:six:seven\nepsilon:zeta:eta:theta:iota:kappa:lambda:mu\nthe quick brown fox jumps over the lazy dog\n" "" ""
+testcmd "-f a-" "-d ':' -f 5- abc.txt" "five:six:seven\nepsilon:zeta:eta:theta:iota:kappa:lambda:mu\nthe quick brown fox jumps over the lazy dog\n" "" ""
 
-testing "show whole line with no delim" "cut -d ' ' -f 3 abc.txt" \
+testcmd "show whole line with no delim" "-d ' ' -f 3 abc.txt" \
 	"one:two:three:four:five:six:seven\nalpha:beta:gamma:delta:epsilon:zeta:eta:theta:iota:kappa:lambda:mu\nbrown\n" "" ""
 
-testing "with echo, -c (a-b)" "echo 'ref_categorie=test' | cut -c 1-15 " "ref_categorie=t\n" "" ""
-testing "with echo, -c (a)" "echo 'ref_categorie=test' | cut -c 14" "=\n" "" ""
+testcmd "-c (a-b)" "-c 1-15 " "ref_categorie=t\n" "" "ref_categorie=test\n"
+testcmd "-c (a)" "-c 14" "=\n" "" "ref_categorie=test\n"
 
 # Modifying abc.txt data as per new testcase
 echo "abcdefghijklmnopqrstuvwxyz" >abc.txt
 
-testing "with -c (a,b,c)" "cut -c 4,5,20 abc.txt" "det\n" "" ""
-
-testing "with -b (a,b,c)" "cut -b 4,5,20 abc.txt" "det\n" "" ""
+testcmd "-c (a,b,c)" "-c 4,5,20 abc.txt" "det\n" "" ""
+testcmd "-b (a,b,c)" "-b 4,5,20 abc.txt" "det\n" "" ""
 
 # Modifying abc.txt data as per testcase
 echo "406378:Sales:Itorre:Jan
@@ -72,13 +72,17 @@ echo "406378:Sales:Itorre:Jan
 636496:Research:Ancholie:Mel
 396082:Sales:Jucacion:Ed" >abc.txt
 
-testing "with -d -f(:) -s" "cut -d: -f3 -s abc.txt" "Itorre\nNasium\nAncholie\nJucacion\n" "" ""
-
-testing "with -d -f( ) -s" "cut -d' ' -f3 -s abc.txt && echo yes" "yes\n" "" ""
+testcmd "-d -f(:) -s" "-d: -f3 -s abc.txt" "Itorre\nNasium\nAncholie\nJucacion\n" "" ""
+testcmd "-d -f( ) -s" "-d' ' -f3 -s abc.txt && echo yes" "yes\n" "" ""
+testcmd "-d -f(a) -s" "-da -f3 -s abc.txt" "n\nsium:Jim\n\ncion:Ed\n" "" ""
+testcmd "-d -f(a) -s -n" "-da -f3 -s -n abc.txt" "n\nsium:Jim\n\ncion:Ed\n" "" ""
 
-testing "with -d -f(a) -s" "cut -da -f3 -s abc.txt" "n\nsium:Jim\n\ncion:Ed\n" "" ""
+# Feature posix documents but nobody bothers to implement
+toyonly testcmd "-nb" '-nb 8-17 "$FILES/utf8/japan.txt"' "ガラス\n" "" ""
 
-testing "with -d -f(a) -s -n" "cut -da -f3 -s -n abc.txt" "n\nsium:Jim\n\ncion:Ed\n" "" ""
+# Feature that is, as far as I can tell, totally undocumented?
+testcmd "-d newline" "-d \$'\n' -f 2-3,5" "two\nthree\nfive\n" "" \
+  'one\ntwo\nthree\nfour\nfive\nsix\seven\n'
 
 # Removing abc.txt file for cleanup purpose
 rm abc.txt
diff --git a/toys/posix/cut.c b/toys/posix/cut.c
index 5072949d..4e9f56c1 100644
--- a/toys/posix/cut.c
+++ b/toys/posix/cut.c
@@ -42,26 +42,36 @@ GLOBALS(
   char *d, *O;
   struct arg_list *select[5]; // we treat them the same, so loop through
 
+  unsigned line;
   int pairs;
   regex_t reg;
 )
 
-
 // Apply selections to an input line, producing output
 static void cut_line(char **pline, long len)
 {
-  unsigned *pairs = (void *)toybuf;
+  unsigned *pairs = (void *)toybuf, wc;
   char *line;
-  int i, j;
+  int i, j, k;
 
   if (!pline) return;
   line = *pline;
   if (len && line[len-1]=='\n') line[--len] = 0;
+  TT.line++;
 
   // Loop through selections
   for (i=0; i<TT.pairs; i++) {
     unsigned start = pairs[2*i], end = pairs[(2*i)+1], count;
-    char *s = line, *ss;
+    char *s = line, *ss, *sss;
+
+    // when the delimiter is \n output lines.
+    if (*TT.d == '\n') {
+      if (TT.line<start || TT.line>end) {
+        if (i+1 == TT.pairs) return;
+        continue;
+      }
+      goto write_line;
+    }
 
     // input: start/end position, count=difference between them
     // output: s = start of string, len = bytes to output
@@ -72,8 +82,20 @@ static void cut_line(char **pline, long len)
     count = end-start;
 
     // Find start and end of output string for the relevant selection type
-    if (FLAG(b)) s += start;
-    else if (FLAG(C)) {
+    if (FLAG(b)) {
+      if (!FLAG(n)) s += start;
+      else {
+        if (end>len) end = len;
+        for (sss = ss = s; (k = (ss-line))<end;) {
+          if (0>(j = utf8towc(&wc, ss, len))) ss++;
+          else {
+            if (((ss += j)-line)<=end) sss = ss;
+            if ((ss-line)<=start) s = ss;
+          }
+        }
+        if (!(count = sss-s)) continue;
+      }
+    } else if (FLAG(C)) {
       // crunch_str() currently assumes that combining characters get
       // escaped, to provide an unambiguous visual representation.
       // This assumes the input string is null terminated.
@@ -85,8 +107,6 @@ static void cut_line(char **pline, long len)
       count = ss-s;
 
     } else if (FLAG(c)) {
-      unsigned wc;
-      char *sss;
 
       // Find start
       ss = line+len;
@@ -132,6 +152,7 @@ static void cut_line(char **pline, long len)
       if (!j && end == start) {
         if (FLAG(D)) break;
         if (FLAG(s)) return;
+write_line:
         fwrite(line, len, 1, stdout);
         break;
       } else if (!*s) continue;