/**
* \brief Combine copyright holders by name \n
* Input records contain: content and type \n
* Output records: copyright_count, content, type, hash \n
* where content has been simplified from
* the raw records and hash is the md5 of this
* new content.
* \return If $hash non zero, only rows with that hash will
* be returned.
*/
function GroupHolders(&$rows, $hash)
{
/* Step 1: Clean up content, and add hash
*/
$NumRows = count($rows);
for ($RowIdx = 0; $RowIdx < $NumRows; $RowIdx++) {
if (MassageContent($rows[$RowIdx], $hash)) {
unset($rows[$RowIdx]);
}
/* debug to compare original with new content
else
{
echo "<br>row $RowIdx: ".htmlentities($rows[$RowIdx]['content']) . "<br>";
echo "row $RowIdx: ".htmlentities($rows[$RowIdx]['original']) . "<br>";
}
*/
}
/* Step 2: sort the array by the new content */
usort($rows, 'hist_rowcmp');
/* Step 3: group content (remove dups, add counts) */
$NumRows = count($rows);
for ($RowIdx = 1; $RowIdx < $NumRows; $RowIdx++) {
if ($rows[$RowIdx]['content'] == $rows[$RowIdx - 1]['content']) {
$rows[$RowIdx]['copyright_count'] = $rows[$RowIdx - 1]['copyright_count'] + 1;
unset($rows[$RowIdx - 1]);
}
}
/** sorting */
$ordercount = '-1';
$ordercopyright = '-1';
if (isset($_GET['orderc'])) {
$ordercount = $_GET['orderc'];
}
if (isset($_GET['ordercp'])) {
$ordercopyright = $_GET['ordercp'];
}
// sort by count
if (1 == $ordercount) {
usort($rows, 'hist_rowcmp_count_desc');
} else {
if (0 == $ordercount) {
usort($rows, 'hist_rowcmp_count_asc');
} else {
if (1 == $ordercopyright) {
usort($rows, 'hist_rowcmp_desc');
} else {
if (0 == $ordercopyright) {
usort($rows, 'hist_rowcmp');
} else {
usort($rows, 'hist_rowcmp_count_desc');
}
}
}
}
// default as sorting by count desc
/* note $rows indexes may not be contiguous due to unset in step 3 */
return $rows;
}
/**
* \brief Remove unwanted rows by hash and type and
* exclusions and filter
* \param $NumRows - the number of instances.
* \return new array and $NumRows
*/
function GetRequestedRows($rows, $hash, $type, $excl, &$NumRows, $filter)
{
global $PG_CONN;
$NumRows = count($rows);
$prev = 0;
$ExclArray = explode(":", $excl);
/* filter will need to know the rf_pk of "No_license_found" or "Void" */
if (!empty($filter)) {
$NoLicStr = "No_license_found";
$VoidLicStr = "Void";
$rf_clause = "";
$sql = "select rf_pk from license_ref where rf_shortname IN ('{$NoLicStr}', '{$VoidLicStr}')";
$result = pg_query($PG_CONN, $sql);
DBCheckResult($result, $sql, __FILE__, __LINE__);
if (pg_num_rows($result) > 0) {
$rf_rows = pg_fetch_all($result);
foreach ($rf_rows as $row) {
if (!empty($rf_clause)) {
$rf_clause .= " or ";
}
$rf_clause .= " rf_fk={$row['rf_pk']}";
}
}
pg_free_result($result);
}
for ($RowIdx = 0; $RowIdx < $NumRows; $RowIdx++) {
$row = $rows[$RowIdx];
/* Remove undesired types */
if ($rows[$RowIdx]['type'] != $type) {
unset($rows[$RowIdx]);
continue;
}
/* remove excluded files */
if ($excl) {
$FileExt = GetFileExt($rows[$RowIdx]['ufile_name']);
if (in_array($FileExt, $ExclArray)) {
unset($rows[$RowIdx]);
continue;
}
}
/* apply filters */
if ($filter == "nolics" and $rf_clause) {
/* discard file unless it has no license */
$sql = "select rf_fk from license_file where ({$rf_clause}) and pfile_fk={$row['pf']}";
$result = pg_query($PG_CONN, $sql);
DBCheckResult($result, $sql, __FILE__, __LINE__);
$FoundRows = pg_num_rows($result);
pg_free_result($result);
if ($FoundRows == 0) {
unset($rows[$RowIdx]);
continue;
}
}
/* rewrite content for easier human assimilation */
if (MassageContent($rows[$RowIdx], $hash)) {
unset($rows[$RowIdx]);
}
}
/* reset array keys, keep order (uploadtree_pk) */
$rows2 = array();
foreach ($rows as $row) {
$rows2[] = $row;
}
unset($rows);
/* remove duplicate files */
$NumRows = count($rows2);
$prev = 0;
for ($RowIdx = 0; $RowIdx < $NumRows; $RowIdx++) {
if ($RowIdx > 0) {
/* Since rows are ordered by uploadtree_pk,
* remove duplicate uploadtree_pk's. This can happen if there
* are multiple same copyrights in one file.
*/
if ($rows2[$RowIdx - 1]['uploadtree_pk'] == $rows2[$RowIdx]['uploadtree_pk']) {
unset($rows2[$RowIdx - 1]);
}
}
}
/* sort by name so output has some order */
usort($rows2, 'copyright_namecmp');
return $rows2;
}
