Avoid skipping blobs that precede blobs in the same line (awk and python).

git-svn-id: http://www.fsfla.org/svn/fsfla/software/linux-libre/scripts@5871 559672b5-ba27-0410-b829-e8f1faed8b1b

diff --git a/deblob-check-awk b/deblob-check-awk
index 74b3fee31643227cb5503f13a6310be3469e048a..75002ecf534b1078033a10a3099330a9f0a846e2 100755 (executable)
--- a/deblob-check-awk
+++ b/deblob-check-awk
@@ -3283,7 +3283,9 @@ s,^\\(.*\\)\$,blob = r'(?P<blob>\\1)',;\
 if no_falsepos or falsepos is None:
     falsepos = r'(?!)'
 
-rx = '^%s|[^\n]*%s' % (falsepos, blob)
+# Use non-greedy operator to skip partial line before blob, to avoid
+# skipping blobs when more than one appears in the same line.
+rx = '^%s|[^\n]*?%s' % (falsepos, blob)
 
 if with_context:
     rx += '|^' + cblob
@@ -3521,7 +3523,7 @@ BEGIN {
     pp = p = pend = 1;
     if (verbose > 1) print "searching starting at", substr (s, p, 10)
     while (match (substr (s, p),
-                 /(^|[\n])($falsepos)|(^|[\n])($cblob)|[^\n]*($blob)/)) {
+                 /(^|[\n])($falsepos)|(^|[\n])($cblob)|$blob/)) {
        blobs = falses = 0;
        firstmatchstart = RSTART + p - 1;
        if (substr (s, firstmatchstart, 1) == "\n")
@@ -3550,10 +3552,6 @@ BEGIN {
                continue;
            } else {
                if (verbose) print "match is a blob";
-               # Match again, to skip the leading non-blob characters.
-               match (substr (s, matchstart), /$blob/);
-               matchstart = RSTART + matchstart - 1;
-               matchlen = RLENGTH;
                blob_p = 1;
                blobs++;
            }
@@ -3565,14 +3563,23 @@ BEGIN {
                p = matchstart + length (replacement);
            } else
                p = matchstart + matchlen;
+           i = index (substr (s, p), "\n");
+           if (!i)
+               i = length (s)
+           p--;
            if (!match (substr (s, p),
-                       /^($falsepos)|^($cblob)|[^\n]*($blob)/) \\
-               || (RSTART != 1 && p + RSTART >= pend))
+                       /[\n]($falsepos)|[\n]($cblob)|$blob/) \\
+               || (RSTART > i && p + RSTART >= pend))
                break;
        }
 
        if (print_nomatch)
            printf "%s", substr (s, pp, firstmatchstart - pp);
+        else if (print_blob || print_falsepos) {
+           lastline = substr (s, pp, firstmatchstart - pp);
+           sub (/.*[\n]/, "", lastline);
+           firstmatchstart -= length (lastline);
+        }
        pp = firstmatchstart;
 
        if (p < pend)