Fix character encoding for OCR

The addition of the OCR feature revealed character encoding bugs that might have been invisible previously. This commit fixes the character encoding in multiple places to ensure the correct displaying of special characters in the web frontend.
os-autoinst · Sep 14, 2023 · 3a4e3e1 · 3a4e3e1
1 parent c8103e1
commit 3a4e3e1
Show file tree

Hide file tree

Showing 4 changed files with 8 additions and 7 deletions.
diff --git a/lib/OpenQA/Schema/Result/Jobs.pm b/lib/OpenQA/Schema/Result/Jobs.pm
@@ -1029,7 +1029,7 @@ sub append_log ($self, $log, $file_name) {
     my $path = $self->worker->get_property('WORKER_TMPDIR');
     return unless -d $path;    # we can't help
     $path .= "/$file_name";
-    if (open(my $fd, '>>', $path)) {
+    if (open(my $fd, '>>:utf8', $path)) {
         print $fd $log->{data};
         close($fd);
     }

diff --git a/lib/OpenQA/WebAPI/Controller/Running.pm b/lib/OpenQA/WebAPI/Controller/Running.pm
@@ -93,7 +93,7 @@ sub streamtext ($self, $file_name, $start_hook = undef, $close_hook = undef) {
     # if the open fails, continue, well check later
     my $log;
     my ($ino, $size);
-    if (open($log, '<', $logfile)) {
+    if (open($log, '<:utf8', $logfile)) {
         # Send the last 10KB of data from the logfile, so that
         # the client sees some data immediately
         $ino = (stat $logfile)[1];
@@ -122,7 +122,7 @@ sub streamtext ($self, $file_name, $start_hook = undef, $close_hook = undef) {
         TEXT_STREAMING_INTERVAL() => sub {
             if (!$ino) {
                 # log file was not yet opened
-                return unless open($log, '<', $logfile);
+                return unless open($log, '<:utf8', $logfile);
                 $ino = (stat $logfile)[1];
                 $size = -s $logfile;
             }

diff --git a/lib/OpenQA/Worker/Job.pm b/lib/OpenQA/Worker/Job.pm
@@ -25,6 +25,7 @@ use Try::Tiny;
 use Scalar::Util 'looks_like_number';
 use File::Map 'map_file';
 use List::Util 'max';
+use Encode 'decode';
 
 # define attributes for public properties
 has 'worker';
@@ -1238,7 +1239,7 @@ sub _log_snippet {
     sysseek($fd, $offset, Fcntl::SEEK_SET);    # FIXME: handle error?
     if (defined sysread($fd, my $buf = '', 100000)) {
         $ret{offset} = $offset;
-        $ret{data} = $buf;
+        $ret{data} = decode('utf-8', $buf);
     }
     if (my $new_offset = sysseek($fd, 0, Fcntl::SEEK_CUR)) {
         $self->{"_$offset_name"} = $new_offset;

diff --git a/templates/webapi/step/viewimg.html.ep b/templates/webapi/step/viewimg.html.ep
@@ -1,4 +1,4 @@
-% use Mojo::JSON 'encode_json';
+% use Mojo::JSON 'to_json';
 <div class="aligncenter" data-image="<%= url_for('test_img', filename => $screenshot) %>" id="step_view">
     <span class="candidate_needles">
         Candidate needles and tags:
@@ -37,8 +37,8 @@
                                 <tbody>
                                     % for my $needle (@{$needles_by_tag->{$tag}}) {
                                         <tr data-image="<%= $needle->{image} %>"
-                                            data-areas="<%= encode_json($needle->{areas})%>"
-                                            data-matches="<%= encode_json($needle->{matches}) %>"
+                                            data-areas="<%= to_json($needle->{areas}) %>"
+                                            data-matches="<%= to_json($needle->{matches}) %>"
                                             data-label="<%= $needle->{label} %>"
                                         % if (!$has_selection && $needle->{selected}) {
                                             % $has_selection = 1;