[w3c_log] handle an invalid log correctly

Fixes #988

aminclude_static.am

@@ -1,6 +1,6 @@
 
 # aminclude_static.am generated automatically by Autoconf
-# from AX_AM_MACROS_STATIC on Sat Jun  4 10:09:23 PDT 2022
+# from AX_AM_MACROS_STATIC on Sat Jun  4 14:05:13 PDT 2022
 
 
 # Code coverage

configure.ac

@@ -102,7 +102,7 @@ AC_SEARCH_LIBS(BZ2_bzopen, bz2,
 AC_SUBST(BZIP2_SUPPORT)
 AC_SEARCH_LIBS(dlopen, dl)
 AC_SEARCH_LIBS(backtrace, execinfo)
-LIBCURL_CHECK_CONFIG([], [7.23.0], [], [AC_MSG_ERROR([libz required to build])], [test x"${enable_static}" = x"yes"])
+LIBCURL_CHECK_CONFIG([], [7.23.0], [], [AC_MSG_ERROR([libcurl required to build])], [test x"${enable_static}" = x"yes"])
 
 # Sometimes, curses depends on these libraries being linked in...
 AC_ARG_ENABLE([tinfo],

m4/libcurl.m4

@@ -128,7 +128,14 @@ AC_DEFUN([LIBCURL_CHECK_CONFIG],
            fi
            if test x"$LIBCURL" = "x" ; then
               if $5; then
-                 LIBCURL=`$_libcurl_config --static-libs`
+                 case "$host_os" in
+                 darwin*)
+                    LIBCURL=`$_libcurl_config --libs`
+                    ;;
+                 *)
+                    LIBCURL=`$_libcurl_config --static-libs`
+                    ;;
+                 esac
               else
                  LIBCURL=`$_libcurl_config --libs`
               fi
@@ -143,6 +150,23 @@ AC_DEFUN([LIBCURL_CHECK_CONFIG],
               esac
            fi
 
+           dnl If we are on OS X and we haven't picked up libcurl static or
+           dnl otherwise, then let's just go ahead and use the one present on
+           dnl the system. Since this compiled binary will only run on OS X
+           dnl which almost always has cURL installed, it's OK to add the
+           dnl static dependency.
+           AS_IF([test "x${LIBCURL}" = "x"],
+                [AS_CASE(["$host_os"],
+                    [darwin*],
+                    [AS_IF([test "x$_libcurl_config" != "x"],
+                        AS_VAR_SET(LIBCURL, $($_libcurl_config --libs)),
+                        []
+                    )],
+                    []
+                )],
+                []
+            )
+
            # All curl-config scripts support --feature
            _libcurl_features=`$_libcurl_config --feature`
 

src/log_format_impls.cc

@@ -236,7 +236,10 @@ class generic_log_format : public log_format {
 
     std::shared_ptr<log_format> specialized(int fmt_lock) override
     {
-        return std::make_shared<generic_log_format>(*this);
+        auto retval = std::make_shared<generic_log_format>(*this);
+
+        retval->lf_specialized = true;
+        return retval;
     }
 };
 
@@ -505,11 +508,15 @@ public:
         }
 
         if (found_ts) {
+            if (!this->lf_specialized) {
+                for (auto& ll : dst) {
+                    ll.set_ignore(true);
+                }
+            }
             dst.emplace_back(li.li_file_range.fr_offset, tv, level, 0, opid);
             return SCAN_MATCH;
-        } else {
-            return SCAN_NO_MATCH;
         }
+        return SCAN_NO_MATCH;
     }
 
     scan_result_t scan(logfile& lf,
@@ -728,7 +735,10 @@ public:
 
     std::shared_ptr<log_format> specialized(int fmt_lock = -1) override
     {
-        return std::make_shared<bro_log_format>(*this);
+        auto retval = std::make_shared<bro_log_format>(*this);
+
+        retval->lf_specialized = true;
+        return retval;
     }
 
     class bro_log_table : public log_format_vtab_impl {
@@ -1111,11 +1121,16 @@ public:
             tv.tv_sec = tm2sec(&tm.et_tm);
             tv.tv_usec = tm.et_nsec / 1000;
 
+            if (!this->lf_specialized) {
+                for (auto& ll : dst) {
+                    ll.set_ignore(true);
+                }
+            }
             dst.emplace_back(li.li_file_range.fr_offset, tv, level, 0);
             return SCAN_MATCH;
-        } else {
-            return SCAN_NO_MATCH;
         }
+
+        return SCAN_NO_MATCH;
     }
 
     scan_result_t scan(logfile& lf,
@@ -1249,7 +1264,6 @@ public:
         }
 
         if (!this->wlf_format_name.empty() && !this->wlf_field_defs.empty()) {
-            dst.clear();
             return this->scan_int(dst, li, sbr);
         }
 
@@ -1322,7 +1336,10 @@ public:
 
     std::shared_ptr<log_format> specialized(int fmt_lock = -1) override
     {
-        return std::make_shared<w3c_log_format>(*this);
+        auto retval = std::make_shared<w3c_log_format>(*this);
+
+        retval->lf_specialized = true;
+        return retval;
     }
 
     class w3c_log_table : public log_format_vtab_impl {
@@ -1771,7 +1788,10 @@ public:
 
     std::shared_ptr<log_format> specialized(int fmt_lock) override
     {
-        return std::make_shared<logfmt_format>(*this);
+        auto retval = std::make_shared<logfmt_format>(*this);
+
+        retval->lf_specialized = true;
+        return retval;
     }
 };
 

src/logfile.cc

@@ -260,6 +260,8 @@ logfile::process_prefix(shared_buffer_ref& sbr, const line_info& li)
         }
     }
 
+    log_debug("match found %d %d", found, li.li_file_range.fr_offset);
+
     switch (found) {
         case log_format::SCAN_MATCH:
             if (!this->lf_index.empty()) {
@@ -328,6 +330,8 @@ logfile::process_prefix(shared_buffer_ref& sbr, const line_info& li)
             break;
     }
 
+    log_debug("index size %d", this->lf_index.size());
+
     return retval;
 }
 

test/drive_logfile.cc

@@ -137,9 +137,6 @@ main(int argc, char* argv[])
         lf->rebuild_index();
         assert(!lf->is_closed());
         assert(lf->get_activity().la_polls == 3);
-        if (lf->size() > 1) {
-            assert(lf->get_activity().la_reads == 2);
-        }
         if (expected_format.empty()) {
             assert(lf->get_format() == nullptr);
         } else {

test/expected/expected.am

@@ -264,6 +264,8 @@ EXPECTED_FILES = \
     $(srcdir)/%reldir%/test_logfile.sh_3fc6bfd8a6160817211f3e14fde957af75b9dbe7.out \
     $(srcdir)/%reldir%/test_logfile.sh_4a2a907fcb069b8d6e65961a7b2e796d6c3a87b1.err \
     $(srcdir)/%reldir%/test_logfile.sh_4a2a907fcb069b8d6e65961a7b2e796d6c3a87b1.out \
+    $(srcdir)/%reldir%/test_logfile.sh_7c2e11488bccc59458b5775db4b90de964858259.err \
+    $(srcdir)/%reldir%/test_logfile.sh_7c2e11488bccc59458b5775db4b90de964858259.out \
     $(srcdir)/%reldir%/test_logfile.sh_a7037efd0c4bbf51940137a44e57d94e9307e83e.err \
     $(srcdir)/%reldir%/test_logfile.sh_a7037efd0c4bbf51940137a44e57d94e9307e83e.out \
     $(srcdir)/%reldir%/test_logfile.sh_c18e14a26d8261c9f72747118a469266121d5459.err \

test/expected/test_logfile.sh_290a3c49e53c2229a7400c107338fa0bb38375e2.out

@@ -1 +1,2 @@
+#Fields: ?	)
 0

test/expected/test_logfile.sh_4a2a907fcb069b8d6e65961a7b2e796d6c3a87b1.out

@@ -1,2 +1,4 @@
+#Fields: 0	cs-bytes
+#Fields: 0
 	0 #
 0

test/expected/test_logfile.sh_7c2e11488bccc59458b5775db4b90de964858259.out

@@ -0,0 +1,6 @@
+000
+000
+#Fields: 0
+0
+#Fields: 0
+0

test/test_logfile.sh

@@ -6,6 +6,9 @@ echo ${top_builddir}
 printf '#Date:\t20\x800-2-02\n0\n' | run_cap_test \
     env TEST_COMMENT="short timestamp" ${lnav_test} -n
 
+printf '000\n000\n#Fields: 0\n0\n#Fields: 0\n0' | run_cap_test \
+    env TEST_COMMENT="invalid w3c log" ${lnav_test} -n
+
 cat > rollover_in.0 <<EOF
 2600/2 0 00:00:00 0:
 00:2 0 00:00:00 0: