Adjust a couple of remaining C comment patterns.

Print context all the way to the end in awk and python.
Internal testsuite now passes in awk and sed.
The internal testsuite triggers exponential behavior in python.

git-svn-id: http://www.fsfla.org/svn/fsfla/software/linux-libre/scripts@5860 559672b5-ba27-0410-b829-e8f1faed8b1b

diff --git a/deblob-check-awk b/deblob-check-awk
index d2372c7fc21d2111ba60f14370438eb08e7d457c..3ce0e8732566598d152d46a978aa4d68ec659842 100755 (executable)
--- a/deblob-check-awk
+++ b/deblob-check-awk
@@ -1061,7 +1061,7 @@ set_except () {
     blobna 'for[ ][(][^\n]*ARRAY_SIZE[(]\(sr\|twinax\)_edc[)][^\n]*[)][\n][^;]*mdio_write[^;]*[;]' drivers/net/cxgb3/ael1002.c
     blobname '\(cxgb3[/]\)\?t3\(fw\|\(%c\|.\)_p\(rotocol_\)\?sram\)-\(%d\|[0-9]*\)\.\(%d\|[0-9]*\)\.\(%d\|[0-9]*\)\.bin' drivers/net/cxgb3/cxgb3_main.c
 
-    blob '\([/][*][*]*[*][/][\n]*\)*\([/][*]\([*]*[^*/][^*/]*[/]*\)*[*]*Micro[ ]code\([*]*[^*/][^*/]*[/]*\)*[*]*8086:\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\([\n]*[/][*]\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\)*\|#define[ ][ ]*D10\(1M\(_B\)\?\|1S\|2_E\)_\(CPUSAVER_\(TIMER\|BUNDLE\|MIN_SIZE\)_DWORD\|RCVBUNDLE_UCODE\)[   ]\(\\[\n]\|[^\n]\)*\)\([\n]*[/][*]\([^/]*\|[^*/][/]*\|[*][/][\n]*[/][*]\)*Micro[ ]code\([*]*[^*/][^*/]*[/]*\)*[*]*8086:\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\([\n]*[/][*]\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\)*\|[\n][\n]*#define[ ][ ]*D10\(1M\(_B\)\?\|1S\|2_E\)_\(CPUSAVER_\(TIMER\|BUNDLE\|MIN_SIZE\)_DWORD\|RCVBUNDLE_UCODE\)[         ]\(\\[\n]\|[^\n]\)*\)*' drivers/net/e100.c
+    blob '\([/][*][*]*[*][/][\n]*\)*\([/][*]\([*]*[^*/][^*/]*[/]*\)*[*]*Micro[ ]code\([*]*[^*/][^*/]*[/]*\)*[*]*8086:\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\([\n]*[/][*]\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\)*\|#define[ ][ ]*D10\(1M\(_B\)\?\|1S\|2_E\)_\(CPUSAVER_\(TIMER\|BUNDLE\|MIN_SIZE\)_DWORD\|RCVBUNDLE_UCODE\)[   ]\(\\[\n]\|[^\n]\)*\)\([\n]*[/][*]\([*]*[^*/][^*/]*[/]*\|[*]*[*][/][\n]*[/][*]\)*[*]*Micro[ ]code\([*]*[^*/][^*/]*[/]*\)*[*]*8086:\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\([\n]*[/][*]\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/]\)*\|[\n][\n]*#define[ ][ ]*D10\(1M\(_B\)\?\|1S\|2_E\)_\(CPUSAVER_\(TIMER\|BUNDLE\|MIN_SIZE\)_DWORD\|RCVBUNDLE_UCODE\)[      ]\(\\[\n]\|[^\n]\)*\)*' drivers/net/e100.c
     blobna '\([/][*]\([*]*[^*/][^*/]*[/]*\)*[*]*[*][/][\n]*[   ][      ]\)\(ucode\[opts->\(timer\|bundle\|min_size\)_dword\][ ].=[ ][^;]*[;][\n][\n]*[ ][      ]\)*[^}]*UCODE_SIZE[^}]*cb_ucode[^}]*return[;][\n][     ][}]' drivers/net/e100.c
 
     blob 'static[ ]unsigned[ ]char[ ]__devinitdata[ ]lanai4_\(code\|data\)\[[0-9]*\][ ]=[ ][{][^;]*[}][;]' drivers/net/myri_code.h
@@ -2575,7 +2575,7 @@ bol="\\(^\\|[\\n]\\)"
 eol="\\([\\n]\\|\$\\)"
 
 # Regular expression that matches a C-style comment.
-comment="\\([/][*]\\([^/]*\\|[^*/][/]*\\)*[*][/]\\|[/][/][^\\n]*[\\n]\\)"
+comment="\\([/][*]\\([*]*[^*/][^*/]*[/]*\\)*[*]*[*][/]\\|[/][/][^\\n]*[\\n]\\)"
 
 # Regular expression that matches comments typically used in assembly.
 asmcomment="\\($comment\\|[;#][^\\n]*[\\n]\\)"
@@ -3326,7 +3326,7 @@ for line in sys.stdin:
     if s[-1] == '\n':
         s = s[:-1]
 
-    pp = p = 0
+    pp = p = pend = 0
     match = rxc.search (s, p)
     while match != None:
         firstmatch = match
@@ -3337,6 +3337,7 @@ for line in sys.stdin:
             what = match.lastgroup
 
             if what == 'cblob':
+               pend = match.end()
                 p = match.start() + 1
                 match = rxc.search (s, p)
                 continue
@@ -3362,6 +3363,9 @@ for line in sys.stdin:
             sys.stdout.write (s[pp:firstmatch.start()])
         pp = firstmatch.start()
 
+       if (p < pend):
+           p = pend
+
         p = s.find ('\n', p)
         if p == -1:
             p = len(s)
@@ -3504,12 +3508,12 @@ BEGIN {
 
     s = substr (s, 1, length (s) - 1)
 
-    pp = p = 1;
+    pp = p = pend = 1;
     while (match (substr (s, p),
                  /^($falsepos)|^($cblob)|[^\n]*($blob)/)) {
        blobs = falses = 0;
        firstmatchstart = RSTART + p - 1;
-       do {
+       for (;;) {
            matchstart = RSTART + p - 1;
            matchlen = RLENGTH;
            if (verbose)
@@ -3518,7 +3522,10 @@ BEGIN {
                blob_p = 0;
                falses++;
            } else if (match (substr (s, matchstart, matchlen), /^($cblob)$/) == 1) {
+               pend = p + matchlen;
                p = matchstart + 1;
+               if (verbose) print "search after cblob starting at", p;
+               match (substr (s, p, matchlen), /^($falsepos)|^($cblob)|[^\n]*($blob)/);
                continue;
            } else {
                # Match again, to skip 
@@ -3536,18 +3543,27 @@ BEGIN {
                p = matchstart + length (replacement);
            } else
                p = matchstart + matchlen;
-       } while (match (substr (s, p), /^($falsepos)|^($cblob)|[^\n]*($blob)/) == 1);
+           if (!match (substr (s, p),
+                       /^($falsepos)|^($cblob)|[^\n]*($blob)/) \\
+               || (RSTART != 1 && p + RSTART >= pend))
+               break;
+       }
 
        if (print_nomatch)
            printf "%s", substr (s, pp, firstmatchstart - pp);
        pp = firstmatchstart;
 
+       if (p < pend)
+           p = pend;
+
        i = index (substr (s, p), "\n");
        if (i)
            p += i;
        else
            p = length (s);
 
+       if (verbose) print "match set range:", pp, p
+
        if ((print_blob && blobs) || (print_falsepos && falses)) {
            if (!print_nomatch)
                for (i = nfilenames; i--;)