From: Nick Bowler <nbowler@draconx.ca>
Date: Sat, 18 Jun 2022 01:29:28 +0000 (-0400)
Subject: gen-tree.awk: Add options to tweak the strtab output.
X-Git-Url: https://git.draconx.ca/gitweb/dxcommon.git/commitdiff_plain/cd475ab947ddc0338bd54406234f9cbfa5abe2fd

gen-tree.awk: Add options to tweak the strtab output.

Add a @nostrtab option to disable the string table output in tree
generation, and make it possible to omit the node identifiers from
the initializers.
---

diff --git a/scripts/gen-tree.awk b/scripts/gen-tree.awk
index 53557f1..d04378a 100755
--- a/scripts/gen-tree.awk
+++ b/scripts/gen-tree.awk
@@ -1,14 +1,18 @@
 #!/bin/awk -f
 #
-# Copyright © 2021 Nick Bowler
+# Copyright © 2021-2022 Nick Bowler
 #
 # Generate one or more C array initializers to encode simple tree structures
 # in a compact format.
 #
-# Each line of the input file defines a tree node.  The first field must be a
-# C identifier, which may be optionally followed by a comma.  The identifiers
-# used on non-leaf nodes must be unique with respect to other non-leaf nodes,
-# but leaf nodes may share the same identifier as other nodes.
+# Each nonempty line of the input file is either an option specification
+# or # a tree specification.  An option specification (described later)
+# begins with an @ character.  Other lines specify tree nodes.
+#
+# The first field of a tree specification must be a valid C identifier,
+# optionally followed by a comma.  The identifiers used on non-leaf nodes
+# must be unique with respect to other non-leaf nodes, but leaf nodes may
+# share the same identifier as other nodes.
 #
 # The tree structure is expressed through indentation.  Lines with no leading
 # whitespace designate root nodes.  When the number of white-space characters
@@ -30,11 +34,20 @@
 #
 # For each root node X, the object-like macro X_INITIALIZER is defined.  This
 # can be used to initialize an array with static storage duration.  Each line
-# of the input is surrounded by braces to construct the initializer for one
-# array element.  These are then ordered topologically beginning with the root,
-# but the root itself is not included.  Each subtree is terminated by an
-# "empty" element (with an initializer of {0}).  For example, the above input
-# will produce this (or an equivalent topological ordering) initializer:
+# of the input is defines the initializer for one array element.  These are
+# then ordered topologically beginning with the root, but the root itself is
+# not included.  Each subtree is terminated by an "empty" element (with an
+# initializer of {0}).
+#
+# If there is a comma after the identifier which begins a tree element, or
+# if the identifier is the only text on the line, then the element initializer
+# is constructed by surrounding the entire line by braces.  Otherwise, the
+# identifier is removed and the remainder of the line is surrounded by braces
+# to construct the initializer.  This allows the exact form of the initializer
+# to be customized.
+#
+# For example, the above input will produce this (or an equivalent topological
+# ordering) initializer:
 #
 #    OUTPUT: #define root_INITIALIZER \
 #                 {a}, {e}, {0}, {b}, {d}, {0}, {f}, {g}, {0}, {c}, {0}
@@ -55,6 +68,13 @@
 # constants will normally be in scope when using the root_INITIALIZER macro,
 # which means the resulting array initializers will use them.
 #
+# Options may be used to alter the normal behaviour.  They may be placed
+# anywhere in the input file.  The following options are defined:
+#
+#   @nostrtab
+#     Do not output a definition for tree_strtab or its associated
+#     enumeration constants.
+#
 # License WTFPL2: Do What The Fuck You Want To Public License, version 2.
 # This is free software: you are free to do what the fuck you want to.
 # There is NO WARRANTY, to the extent permitted by law.
@@ -71,6 +91,8 @@ END {
 }
 
 BEGIN {
+  opts["strtab"] = 1;
+
   depth = max_depth = 0;
   num_entries = 0;
 
@@ -80,6 +102,22 @@ BEGIN {
   indent_stack[0] = 0;
 }
 
+# Options
+sub(/^@/, "", $0) {
+  if (NF == 1) {
+    orig=$1
+    gsub(/-/, "_", $1);
+    val = !sub(/^no_?/, "", $1);
+    if ($1 in opts) {
+      opts[$1] = val;
+    } else {
+      print "error: unrecognized option: @" orig | "cat 1>&2"
+      exit 1
+    }
+  }
+  next
+}
+
 NF == 0 { next }
 { indent = index($0, $1) - 1 }
 
@@ -101,7 +139,19 @@ indent > 0 {
   entry_name = $1; sub(/,$/, "", entry_name);
   all_items[num_entries++] = entry_name;
 
-  $1 = ", \\\n\t{ " $1; $NF = $NF " }";
+  # Construct the element initializer for this tree node.
+  $1 = " " $1;
+  if (NF > 1) {
+    # Check if entry name is followed by a comma.  If it is not, the entry
+    # name is excluded from the initializer.
+    check_str = $1$2;
+    gsub(/[ \t]/, "", check_str);
+    if (index(check_str, entry_name ",") == 0) {
+      $1 = "";
+    }
+  }
+  $1 = ", \\\n\t{" $1; $NF = $NF " }";
+
   tree_items[depth] = tree_items[depth] $0;
   level_count[depth]++;
 }
@@ -117,32 +167,39 @@ END {
 indent == 0 { tree_identifier = $1 }
 
 END {
-  entry_strtab = "\1";
-  bucketsort(sorted_items, all_items);
-  for (i = 0; i < num_entries; i++) {
-    s = sorted_items[i];
-    if ((n = index(entry_strtab, s "\1")) > 0) {
-      entry_offsets[s] = n-1;
-    } else {
-      entry_offsets[s] = length(entry_strtab);
-      entry_strtab = entry_strtab s "\1";
+  prefix = "\nenum {\n";
+
+  if (opts["strtab"]) {
+    entry_strtab = "\1";
+    bucketsort(sorted_items, all_items);
+    for (i = 0; i < num_entries; i++) {
+      s = sorted_items[i];
+      if ((n = index(entry_strtab, s "\1")) > 0) {
+        entry_offsets[s] = n-1;
+      } else {
+        entry_offsets[s] = length(entry_strtab);
+        entry_strtab = entry_strtab s "\1";
+      }
     }
-  }
 
-  gsub(/\1/, "\"\n\t\"\\0\" \"", entry_strtab);
-  sub(/^"/, "", entry_strtab);
-  sub(/\n[^\n]*$/, ";", entry_strtab);
-  print "\nstatic const char tree_strtab[] =" entry_strtab
+    gsub(/\1/, "\"\n\t\"\\0\" \"", entry_strtab);
+    sub(/^"/, "", entry_strtab);
+    sub(/\n[^\n]*$/, ";", entry_strtab);
+    print "\nstatic const char tree_strtab[] =" entry_strtab
 
-  prefix = "\nenum {\n";
-  for (item in entry_offsets) {
-    printf "%s\t%s = %d", prefix, item, entry_offsets[item];
-    prefix = ",\n";
+    for (item in entry_offsets) {
+      printf "%s\t%s = %d", prefix, item, entry_offsets[item];
+      prefix = ",\n";
+    }
   }
+
   for (i in subtree_offsets) {
-    printf ",\n\t%s_OFFSET = %d", i, subtree_offsets[i];
+    printf "%s\t%s_OFFSET = %d", prefix, i, subtree_offsets[i];
+    prefix = ",\n";
+  }
+  if (!index(prefix, "enum")) {
+    print "\n};";
   }
-  print "\n};";
 }
 
 END {
diff --git a/tests/scripts.at b/tests/scripts.at
index 4a83807..dcbcce5 100644
--- a/tests/scripts.at
+++ b/tests/scripts.at
@@ -7,6 +7,7 @@ dnl There is NO WARRANTY, to the extent permitted by law.
 AT_BANNER([Script tests])
 
 AT_SETUP([gen-options.awk])
+AT_KEYWORDS([gen-options awk script scripts])
 
 AT_DATA([options.def],
 [[--option-only
@@ -220,6 +221,7 @@ p
 AT_CLEANUP
 
 AT_SETUP([gen-strtab.awk])
+AT_KEYWORDS([gen-strtab awk script scripts])
 
 AT_DATA([test.def],
 [[
@@ -301,6 +303,8 @@ oneline
 AT_CLEANUP
 
 AT_SETUP([gen-tree.awk])
+AT_KEYWORDS([gen-tree awk script scripts])
+
 AT_DATA([tree.def],
 [[ROOT0
   r0a, r0a_OFFSET
@@ -359,6 +363,7 @@ int main(void)
   print_subtree(tree0, 0, 1);
   printf("ROOT1\n");
   print_subtree(tree1, 0, 1);
+  return 0;
 }
 ]])
 cp tree.def expout
@@ -366,7 +371,76 @@ AT_CHECK([$CC -o test0$EXEEXT test0.c && ./test0$EXEEXT], [0], [expout])
 
 AT_CLEANUP
 
+# Test the gen-tree features that avoid creating string labels for nodes.
+AT_SETUP([gen-tree.awk @nostrtab option])
+AT_KEYWORDS([gen-tree awk script scripts])
+
+AT_DATA([tree.def],
+[[@nostrtab
+ROOT
+ a 1, a_OFFSET
+  b 1
+  c 2
+ d 2, d_OFFSET
+  e 1
+  f 2
+]])
+AT_CHECK([$AWK -f "$builddir/scripts/gen-tree.awk" <tree.def >tree.h])
+
+AT_DATA([test0.c],
+[[float tree_strtab = 0;
+#define a []
+#define b []
+#define c []
+#define e []
+#define f []
+#include "tree.h"
+#include <stdio.h>
+
+static struct { int num, offset; } root[] = { ROOT_INITIALIZER };
+
+int main(void)
+{
+  unsigned i;
+  for (i = 0; i < sizeof root / sizeof root[0]; i++) {
+    printf("%d, %d\n", root[i].num, root[i].offset);
+  }
+}
+]])
+
+AT_CHECK([$CC -o test0$EXEEXT test0.c && ./test0$EXEEXT], [0],
+[[1, 3
+2, 6
+0, 0
+1, 0
+2, 0
+0, 0
+1, 0
+2, 0
+0, 0
+]])
+
+AT_DATA([flat.def],
+[[FLAT
+ a 1
+ b 2
+ c 3
+@nostrtab
+]])
+AT_CHECK([$AWK -f "$builddir/scripts/gen-tree.awk" <flat.def >flat.h])
+
+sed -e 's/tree\.h/flat.h/' -e 's/ROOT/FLAT/' test0.c >test1.c
+AT_CHECK([$CC -o test1$EXEEXT test1.c && ./test1$EXEEXT], [0],
+[[1, 0
+2, 0
+3, 0
+0, 0
+]])
+
+AT_CLEANUP
+
 AT_SETUP([join.awk])
+AT_KEYWORDS([join awk script scripts])
 
 JOIN="$AWK -f $builddir/scripts/join.awk --"