use new cnf-to-paths.awk script instead of the inline sed.

Also, change the generated paths.h so that all non-SELFAUTO paths
become a constant string "/nonesuch".  That way, if the definitions
change, the binaries don't all uselessly change with them.


git-svn-id: svn://tug.org/texlive/trunk@22582 c570f23f-e606-0410-a88d-b1316a301751

4 files changed, 64 insertions, 60 deletions

diff --git a/Build/source/texk/kpathsea/ChangeLog b/Build/source/texk/kpathsea/ChangeLog
index 0bdfe69cbec..f9121033957 100644
--- a/Build/source/texk/kpathsea/ChangeLog
+++ b/Build/source/texk/kpathsea/ChangeLog
@@ -1,3 +1,11 @@
+2011-05-23  Karl Berry  <karl@tug.org>
+
+	* cnf-to-paths.awk: new awk script.
+	* Makefile.am (stamp-paths): use it instead of the inline sed.
+	Also, change the generated paths.h so that all non-SELFAUTO paths
+	become a constant string "/nonesuch".  That way, if the definitions
+	change, the binaries don't all uselessly change with them.
+
 2011-05-22  Karl Berry  <karl@tug.org>
 
 	* texmf.cnf: doc fixes.
diff --git a/Build/source/texk/kpathsea/Makefile.am b/Build/source/texk/kpathsea/Makefile.am
index 823cc3726bf..52406abec48 100644
--- a/Build/source/texk/kpathsea/Makefile.am
+++ b/Build/source/texk/kpathsea/Makefile.am
@@ -157,40 +157,15 @@ kpseinclude_HEADERS = \
 	xopendir.h \
 	xstat.h
 
-## First null out comments and leading/trailing whitespace, then remove
-## lines that define invalid C identifiers, then remove blank lines and
-## lines that define lowercase values (those are never path values).
-## Each line remaining looks like
-## <name> = <value>
-## (but = and surrounding spaces are optional, hence can't remove all spaces).
-## Change this to #ifndef DEFAULT_name@#define DEFAULT_name "value"@#endif,
-## then change the @'s to newlines (sed isn't good at multiline replacements).
-##
-## No backslash-newline escapes in the long sed replacement because that
-## will turn into a space in the output.
-##
-## We replace the $TEXMF* references with nonexisting directories.  These
-## are the compile-time defaults and are not useful in the native TL
-## builds; we rely completely on the self-location.  (Previously, the
-## random build-time paths would get included in the binaries, to
-## consequent confusion.)
-##
-## We preserve the $SELFAUTO* references, since the selfauto-using
-## compile-time path for cnf files is how texmf.cnf gets found, which
-## determines everything else.
-##
-## The definition of DEFAULT_TEXMF (and other variables)
-## that winds up in the final paths.h will not be used.
-##
+## Handle backslash continuations, then null out comments and
+## leading/trailing whitespace.  An awk script does the rest.
+## 
 ## We don't want to rewrite paths.h when we have only changed comments
 ## in texmf.cnf that have no effect on paths.h, since that would cause
 ## almost everything to be rebuilt.
 ##
-## We also replace all semicolons with colons in the paths.
-## (The path-splitting code should be changed to understand both.)
-##
 paths.h: stamp-paths
-stamp-paths: texmf.cnf bsnl.awk
+stamp-paths: texmf.cnf bsnl.awk cnf-to-paths.awk
 # ensure grep is not completely broken, e.g., from GREP_OPTIONS.
 	echo a | grep -v b >/dev/null || { echo "*** grep broken, goodbye."; exit 1; }
 #
@@ -198,21 +173,7 @@ stamp-paths: texmf.cnf bsnl.awk
 	echo "/* paths.h: Generated from texmf.cnf. */" >paths.tmp
 	awk -f $(srcdir)/bsnl.awk $(srcdir)/texmf.cnf \
 	| sed -e 's/%.*//' -e 's/^[ 	]*//' -e 's/[ 	]*$$//' \
-	| grep '^[ 	]*[A-Z0-9_]*[ 	=]' \
-	| sed '/^$$/d' \
-	| sed 's/^\([^ 	=]*\)[ 	]*=*[ 	]*\(.*\)/#ifndef DEFAULT_\1%#define DEFAULT_\1 "\2"%#endif/' \
-	| tr '%' '\012' \
-	| sed -e 's%\$$TEXMFCONFIG%/nonesuch/texmfconfig%g' \
-	      -e 's%\$$TEXMFDIST%/nonesuch/texmfdist%g' \
-	      -e 's%\$$TEXMFDOCDIR%/nonesuch/texmfdocdir%g' \
-	      -e 's%\$$TEXMFLOCAL%/nonesuch/texmflocal%g' \
-	      -e 's%\$$TEXMFMAIN%/nonesuch/texmfmain%g' \
-	      -e 's%\$$TEXMFSYSCONFIG%/nonesuch/texmfsysconfig%g' \
-	      -e 's%\$$TEXMFSYSVAR%/nonesuch/texmfsysvar%g' \
-	      -e 's%\$$TEXMFVAR%/nonesuch/texmfvar%g' \
-	      -e 's%\$$TEXMF%/nonesuch/texmf%g' \
-	      -e 's%\$$VARTEXFONTS%/nonesuch/vartexfonts%g' \
-	| sed -e 's%;%:%g' \
+	| awk -f $(srcdir)/cnf-to-paths.awk \
 	>>paths.tmp
 	@if cmp -s paths.h paths.tmp 2>/dev/null; then \
 	  echo "paths.h is unchanged"; \
diff --git a/Build/source/texk/kpathsea/Makefile.in b/Build/source/texk/kpathsea/Makefile.in
index 436bf93db4c..0586ad196e3 100644
--- a/Build/source/texk/kpathsea/Makefile.in
+++ b/Build/source/texk/kpathsea/Makefile.in
@@ -1750,7 +1750,7 @@ rebuild: all
 $(libkpathsea_la_OBJECTS): paths.h kpathsea.h
 
 paths.h: stamp-paths
-stamp-paths: texmf.cnf bsnl.awk
+stamp-paths: texmf.cnf bsnl.awk cnf-to-paths.awk
 # ensure grep is not completely broken, e.g., from GREP_OPTIONS.
 	echo a | grep -v b >/dev/null || { echo "*** grep broken, goodbye."; exit 1; }
 #
@@ -1758,21 +1758,7 @@ stamp-paths: texmf.cnf bsnl.awk
 	echo "/* paths.h: Generated from texmf.cnf. */" >paths.tmp
 	awk -f $(srcdir)/bsnl.awk $(srcdir)/texmf.cnf \
 	| sed -e 's/%.*//' -e 's/^[ 	]*//' -e 's/[ 	]*$$//' \
-	| grep '^[ 	]*[A-Z0-9_]*[ 	=]' \
-	| sed '/^$$/d' \
-	| sed 's/^\([^ 	=]*\)[ 	]*=*[ 	]*\(.*\)/#ifndef DEFAULT_\1%#define DEFAULT_\1 "\2"%#endif/' \
-	| tr '%' '\012' \
-	| sed -e 's%\$$TEXMFCONFIG%/nonesuch/texmfconfig%g' \
-	      -e 's%\$$TEXMFDIST%/nonesuch/texmfdist%g' \
-	      -e 's%\$$TEXMFDOCDIR%/nonesuch/texmfdocdir%g' \
-	      -e 's%\$$TEXMFLOCAL%/nonesuch/texmflocal%g' \
-	      -e 's%\$$TEXMFMAIN%/nonesuch/texmfmain%g' \
-	      -e 's%\$$TEXMFSYSCONFIG%/nonesuch/texmfsysconfig%g' \
-	      -e 's%\$$TEXMFSYSVAR%/nonesuch/texmfsysvar%g' \
-	      -e 's%\$$TEXMFVAR%/nonesuch/texmfvar%g' \
-	      -e 's%\$$TEXMF%/nonesuch/texmf%g' \
-	      -e 's%\$$VARTEXFONTS%/nonesuch/vartexfonts%g' \
-	| sed -e 's%;%:%g' \
+	| awk -f $(srcdir)/cnf-to-paths.awk \
 	>>paths.tmp
 	@if cmp -s paths.h paths.tmp 2>/dev/null; then \
 	  echo "paths.h is unchanged"; \
diff --git a/Build/source/texk/kpathsea/cnf-to-paths.awk b/Build/source/texk/kpathsea/cnf-to-paths.awk
new file mode 100644
index 00000000000..755a5c30d14
--- /dev/null
+++ b/Build/source/texk/kpathsea/cnf-to-paths.awk
@@ -0,0 +1,49 @@
+# $Id$
+# cnf-to-paths.awk - convert texmf.cnf assignments to paths.h #define's.
+# Public domain.  Originally written 2011, Karl Berry.
+
+# We assume comments have been stripped.
+# 
+# we only care about definitions with a valid C identifier (e.g.,
+# TEXINPUTS, not TEXINPUTS.latex), that is, lines that look like this:
+#   <name> = <value>
+# (whitespace around the = is optional)
+#
+/^[ \t]*[A-Z0-9_]+[ \t]*=/ {
+  # On these lines, there are two cases:
+  # 
+  # 1) definitions referring to SELFAUTO*, which we want to keep.  In
+  # particular, this is how the compile-time TEXMFCNF gets defined and
+  # thus how texmf.cnf gets found.
+  # 
+  # 2) all the others, which we want to convert to a constant
+  # /nonesuch.  That way, the binaries don't get change just because we
+  # change definitions in texmf.cnf.
+  # 
+  # The definition of DEFAULT_TEXMF (and other variables)
+  # that winds up in the final paths.h will not be used.
+  
+  # Let's extract the identifier and the value from the line.  Since
+  # gawk's subexpression matching is an extension, do it with copies.
+  ident = $0;
+  sub(/^[ \t]*/, "", ident);
+  sub(/[ \t]*=.*/, "", ident);
+  
+  val = $0;
+  sub(/^.*=[ \t]*/, "", val);
+  sub(/[ \t]*$/, "", val);
+  #print "got ident", ident, "and val", val >"/dev/stderr"; 
+  
+  if (val ~ /\$SELFAUTO/) {
+    # Replace all semicolons with colons in the SELFAUTO paths we're keeping.
+    # (The path-splitting code should be changed to understand both.)
+    gsub(/;/, ":", val);
+  } else {
+    val = "/nonesuch";
+  }
+  
+  print "#ifndef DEFAULT_" ident;
+  print "#define DEFAULT_" ident " \"" val "\"";
+  print "#endif";
+  print "";
+}