- much stricter synthesis parsing. fail on first error

- add test using a buggy synthesis that occured on kenobi

4 files changed, 31 insertions, 15 deletions

diff --git a/URPM.xs b/URPM.xs
index 6ea6dca..3704b40 100644
--- a/URPM.xs
+++ b/URPM.xs
@@ -979,14 +979,16 @@ call_package_callback(SV *urpm, SV *sv_pkg, SV *callback) {
   return sv_pkg != NULL;
 }
 
-static void
+static int
 parse_line(AV *depslist, HV *provides, URPM__Package pkg, char *buff, SV *urpm, SV *callback) {
   SV *sv_pkg;
   URPM__Package _pkg;
   char *tag, *data;
   int data_len;
 
-  if ((tag = strchr(buff, '@')) != NULL && (data = strchr(tag+1, '@')) != NULL) {
+  if (buff[0] == 0) {
+    return 1;
+  } else if ((tag = buff)[0] == '@' && (data = strchr(tag+1, '@')) != NULL) {
     *tag++ = *data++ = 0;
     data_len = 1+strlen(data);
     if (!strcmp(tag, "info")) {
@@ -1011,6 +1013,10 @@ parse_line(AV *depslist, HV *provides, URPM__Package pkg, char *buff, SV *urpm,
     } else if (!strcmp(tag, "summary")) {
       free(pkg->summary); pkg->summary = memcpy(malloc(data_len), data, data_len);
     }
+    return 1;
+  } else {
+    fprintf(stderr, "bad line <%s>\n", buff);
+    return 0;
   }
 }
 
@@ -3031,6 +3037,7 @@ Urpm_parse_synthesis__XS(urpm, filename, ...)
 	memset(&pkg, 0, sizeof(struct s_Package));
 	buff[sizeof(buff)-1] = 0;
 	p = buff;
+	int ok = 1;
 	while ((buff_len = gzread(f, p, sizeof(buff)-1-(p-buff))) >= 0 &&
 	       (buff_len += p-buff)) {
 	  buff[buff_len] = 0;
@@ -3038,15 +3045,17 @@ Urpm_parse_synthesis__XS(urpm, filename, ...)
 	  if ((eol = strchr(p, '\n')) != NULL) {
 	    do {
 	      *eol++ = 0;
-	      parse_line(depslist, provides, &pkg, p, urpm, callback);
+	      if (!parse_line(depslist, provides, &pkg, p, urpm, callback)) { ok = 0; break; }
 	      p = eol;
 	    } while ((eol = strchr(p, '\n')) != NULL);
 	  } else {
 	    /* a line larger than sizeof(buff) has been encountered, bad file problably */
+	    fprintf(stderr, "invalid line <%s>\n", p);
+	    ok = 0;
 	    break;
 	  }
 	  if (gzeof(f)) {
-	    parse_line(depslist, provides, &pkg, p, urpm, callback);
+	    if (!parse_line(depslist, provides, &pkg, p, urpm, callback)) ok = 0;
 	    break;
 	  } else {
 	    /* move the remaining non-complete-line at beginning */
@@ -3055,7 +3064,7 @@ Urpm_parse_synthesis__XS(urpm, filename, ...)
 	    p = &buff[buff_len-(p-buff)];
 	  }
 	}
-	int ok = gzclose(f) == 0;
+	if (gzclose(f) != 0) ok = 0;
 	SPAGAIN;
 	if (ok) {
 	  XPUSHs(sv_2mortal(newSViv(start_id)));
diff --git a/t/buggy_synthesis.cz b/t/buggy_synthesis.cz
new file mode 100644
index 0000000..fa1f68d
--- /dev/null
+++ b/t/buggy_synthesis.cz
diff --git a/t/empty_synthesis.cz b/t/empty_synthesis.cz
new file mode 100644
index 0000000..8aee682
--- /dev/null
+++ b/t/empty_synthesis.cz
diff --git a/t/synthesis.t b/t/synthesis.t
index 4050493..f44e769 100644
--- a/t/synthesis.t
+++ b/t/synthesis.t
@@ -2,19 +2,19 @@
 
 use strict ;
 use warnings ;
-use Test::More tests => 86;
+use Test::More tests => 88;
 use URPM;
 
 my $file1 = 'synthesis.sample.cz';
 
 open my $f, "| gzip -9 >$file1";
-print $f q{
-glibc-devel@provides@glibc-devel == 6:2.2.4-25mdk
-glibc-devel@requires@/sbin/install-info@glibc == 2.2.4@kernel-headers@kernel-headers >= 2.2.1@/bin/sh@/bin/sh@/bin/sh@rpmlib(PayloadFilesHavePrefix) <= 4.0-1@rpmlib(CompressedFileNames) <= 3.0.4-1
-glibc-devel@conflicts@texinfo < 3.11@gcc < 2.96-0.50mdk
-glibc-devel@obsoletes@libc-debug@libc-headers@libc-devel@linuxthreads-devel@glibc-debug
-glibc-devel@info@glibc-devel-2.2.4-25mdk.i586@6@45692097@Development/C
-};
+print $f <<'EOF';
+@provides@glibc-devel == 6:2.2.4-25mdk
+@requires@/sbin/install-info@glibc == 2.2.4@kernel-headers@kernel-headers >= 2.2.1@/bin/sh@/bin/sh@/bin/sh@rpmlib(PayloadFilesHavePrefix) <= 4.0-1@rpmlib(CompressedFileNames) <= 3.0.4-1
+@conflicts@texinfo < 3.11@gcc < 2.96-0.50mdk
+@obsoletes@libc-debug@libc-headers@libc-devel@linuxthreads-devel@glibc-debug
+@info@glibc-devel-2.2.4-25mdk.i586@6@45692097@Development/C
+EOF
 close $f;
 
 END { unlink $file1 }
@@ -22,9 +22,16 @@ END { unlink $file1 }
 my $a = new URPM;
 ok($a);
 
-my ($first, $end) = $a->parse_synthesis($file1);
+my ($first, $end);
+
+($first, $end) = URPM->new->parse_synthesis('t/empty_synthesis.cz');
+is("$first $end", "0 -1", 'parse empty synthesis');
+
+is(URPM->new->parse_synthesis('t/buggy_synthesis.cz'), undef, 'parse buggy synthesis');
+
+($first, $end) = $a->parse_synthesis($file1);
 ok($first == 0 && $end == 0);
-ok(@{$a->{depslist}} == 1);
+is(int @{$a->{depslist}}, 1);
 ok(keys(%{$a->{provides}}) == 3);
 ok(defined $a->{provides}{'glibc-devel'});
 ok(exists $a->{provides}{'/bin/sh'});