function _get_word_list($projectid)
{
$messages = array();
// get the latest project text of all pages up to last possible round
$last_possible_round = get_Round_for_round_number(MAX_NUM_PAGE_EDITING_ROUNDS);
$pages_res = page_info_query($projectid, $last_possible_round->id, 'LE');
$all_words_w_freq = get_distinct_words_in_text(get_page_texts($pages_res));
// load site word lists for project languages
$site_possible_bad_words = load_site_possible_bad_words_given_project($projectid);
// now, remove any words that are already on the project's bad word list
$site_possible_bad_words = array_diff($site_possible_bad_words, load_project_bad_words($projectid));
// $site_possible_bad_words doesn't have frequency info,
// so start with the info in $all_words_w_freq,
// and extract the items where the key matches a key in $bad_words.
$bad_words_w_freq = array_intersect_key($all_words_w_freq, array_flip($site_possible_bad_words));
// multisort screws up all-numeric words so we need to preprocess first
prep_numeric_keys_for_multisort($bad_words_w_freq);
// sort the list by frequency, then by word
array_multisort(array_values($bad_words_w_freq), SORT_DESC, array_map('strtolower', array_keys($bad_words_w_freq)), SORT_ASC, $bad_words_w_freq);
return array($bad_words_w_freq, $messages);
}
function show_visible_controls()
{
$goodWordData = html_safe($this->good_words);
$badWordData = html_safe($this->bad_words);
$fields = array("projectid" => _("Project ID"), "nameofwork" => _("Name of Work"), "authorsname" => _("Author's Name"), "projectmanager" => _("Project Manager"), "checkedoutby" => _("Post-Processor"), "language" => _("Language"));
foreach ($fields as $field => $label) {
echo "<tr>";
echo "<td class='label'>{$label}</td>";
echo "<td>" . $this->{$field} . "</td>";
echo "</tr>";
}
$exist_OCR_pages = $this->number_of_pages_in_round(null) > 0;
$exist_pages_in_P1_or_later = $this->number_of_pages_in_round(get_Round_for_round_number(1)) > 0;
// due to some special circumstances, not all projects may have pages in P1
// so we'll check to see if the project is in a state after P1 and count
// that as just as good
if (!$exist_pages_in_P1_or_later) {
$current_project_round = get_Round_for_project_state($this->state);
if ($current_project_round && $current_project_round->round_number > 1) {
$exist_pages_in_P1_or_later = true;
}
}
// if the project doesn't have any OCR pages and the project
// has no P1 or later pages, report a message. The second criteria
// is important for type-in projects that may have no OCR pages but
// will have P1 or later pages
if (!$exist_OCR_pages && !$exist_pages_in_P1_or_later) {
echo "<tr>";
echo "<td colspan='2'>";
echo "<p class='error' style='text-align: center;'>";
echo _("There are no pages associated with this project.");
echo "</p>";
echo "</td>";
echo "</tr>";
} else {
echo "<tr>";
echo "<td class='label' style='text-align: center;' colspan='2'>";
echo _("WordCheck Tools and Reports");
echo "</td>";
echo "</tr>";
echo "<tr>";
echo "<td class='label'>" . _("Ad Hoc Word Details") . "</td>";
echo "<td>" . new_window_link("show_adhoc_word_details.php?projectid={$this->projectid}", _("Show details for ad hoc words")) . "</td>";
echo "</tr>";
echo "<tr>";
echo "<td class='label'>" . _("WordCheck Statistics") . "</td>";
echo "<td>" . new_window_link("show_project_wordcheck_stats.php?projectid={$this->projectid}", _("Show WordCheck flagged word statistics")) . "</td>";
echo "</tr>";
if ($exist_pages_in_P1_or_later) {
echo "<tr>";
echo "<td class='label'>" . _("WordCheck Usage") . "</td>";
echo "<td>" . new_window_link("show_project_wordcheck_usage.php?projectid={$this->projectid}", _("Show WordCheck interface usage")) . "</td>";
echo "</tr>";
}
echo "<tr>";
echo "<td class='label' style='text-align: center;' colspan='2'>";
echo _("Word List Suggestion Tools");
echo "</td>";
echo "</tr>";
echo "<tr>";
echo "<td colspan='2'>";
echo "<table width='100%'>";
echo "<tr>";
echo "<td style='width: 50%; text-align: center;' valign='top'>";
echo "<p>";
echo "<b>" . _("Words that WordCheck would currently flag:") . "</b><br>";
echo new_window_link("show_current_flagged_words.php?projectid={$this->projectid}", _("Display"));
echo " | ";
echo "<a href='show_current_flagged_words.php?projectid={$this->projectid}&format=file'>" . _("Download") . "</a>";
echo "</p>";
$suggestions = load_project_good_word_suggestions($this->projectid);
if (count($suggestions)) {
echo "<p>";
echo "<b>" . _("Suggestions from proofreaders:") . "</b><br>";
echo new_window_link("show_good_word_suggestions.php?projectid={$this->projectid}", _("Display"));
echo " | ";
echo "<a href='show_good_word_suggestions.php?projectid={$this->projectid}&timeCutoff=0&format=file'>" . _("Download") . "</a>";
echo "</p>";
}
echo "</td>";
echo "<td style='width: 50%; text-align: center;' valign='top'>";
// see if the site has Possible Bad Word files
$possible_bad_words = load_site_possible_bad_words_given_project($this->projectid);
if (count($possible_bad_words)) {
echo "<p>";
echo "<b>" . _("Words in the Site's Possible bad words file:") . "</b><br>";
echo new_window_link("show_project_possible_bad_words.php?projectid={$this->projectid}", _("Display"));
echo " | ";
echo "<a href='show_project_possible_bad_words.php?projectid={$this->projectid}&format=file'>" . _("Download") . "</a>";
echo "</p>";
}
// see if there are P1 (or later) and OCR pages before showing the link.
// type-in projects may have P1 (or later) pages but no OCR pages
// and the current show_project_stealth_scannos.php page only works
// with projects that have OCR text
if ($exist_pages_in_P1_or_later && $exist_OCR_pages) {
echo "<p>";
echo "<b>" . _("Suggestions from diff analysis:") . "</b><br>";
echo new_window_link("show_project_stealth_scannos.php?projectid={$this->projectid}", _("Display"));
echo " | ";
echo "<a href='show_project_stealth_scannos.php?projectid={$this->projectid}&format=file'>" . _("Download") . "</a>";
echo "</p>";
}
echo "</td>";
echo "</tr>";
echo "</table>";
echo "</td>";
echo "</tr>";
}
echo "<tr>";
echo "<td class='label' style='text-align: center;' colspan='2'>";
echo _("Project Dictionary - Word Lists");
echo "</td>";
echo "</tr>";
echo "<tr>";
echo "<td colspan='2'>";
echo "<table width='100%'>";
echo "<tr>";
echo "<td class='label' style='text-align: center;'>" . _("Good Words") . "</td>";
echo "<td class='label' style='text-align: center;'>" . _("Bad Words") . "</td>";
echo "</tr>";
echo "<tr>";
echo "<td style='width: 50%;'>";
echo "<textarea class='mono' name='good_words' cols='40' rows='20'>{$goodWordData}</textarea>";
echo "</td>";
echo "<td style='width: 50%;'>";
echo "<textarea class='mono' name='bad_words' cols='40' rows='20'>{$badWordData}</textarea>";
echo "</td>";
echo "</tr>";
echo "</table>";
echo "</td>";
echo "</tr>";
echo "<tr>";
echo "<td colspan='2' style='text-align: center;'>";
echo sprintf(_("See the %s for more information on word lists."), new_window_link("../../faq/wordcheck-faq.php", _("WordCheck FAQ")));
echo "</td>";
echo "</tr>";
}
