function build_indexes($opts, $pipe, $cmd = __FUNCTION__, $msg_suffix = null)
{
# set prefix
$prefix = 'build_indexes';
# merge opts
if ($pipe !== null) {
$opts = merge_opts($opts, $pipe, 'indexes');
}
# get fields opt
$fields = get_opt($prefix, $opts, 'fields');
if (!check_opt_set_type($cmd, $fields, 'fields', 'array_of_strings')) {
return false;
}
# get data opt
$data =& get_opt($prefix, $opts, 'data');
if (!check_opt_set_type($cmd, $data, 'data', 'array')) {
return false;
}
# get indexes opt
$indexes =& get_opt($prefix, $opts, 'indexes');
if (!check_opt_set_type($cmd, $indexes, 'indexes', 'array_of_strings')) {
return false;
}
# set up indexes
$res_indexes_assoc = array();
$res_indexes_list = array();
foreach ($indexes as $index_name) {
$index = array();
$res_indexes_assoc[$index_name] =& $index;
$res_indexes_list[] =& $index;
}
# set up field indexes
$field_indexes = build_field_indexes($fields, $indexes, $cmd, $msg_suffix);
if ($field_indexes === false) {
return false;
}
$index_count = count($indexes);
# loop over data to build indexes
for ($i = 0; $i < count($data); $i++) {
unset($line);
$line =& $data[$i];
for ($j = 0; $j < $index_count; $j++) {
$idx = $field_indexes[$j];
$key = $line[$idx];
$res_indexes_list[$j][$key] =& $line;
}
}
# sort the indexes
foreach ($res_indexes_assoc as $name => $index) {
ksort($index);
}
# save the indxes
$opts['indexes'] = $res_indexes_assoc;
return $opts;
}
function load_data_from_doc_file($opts, $pipe, $cmd = __FUNCTION__)
{
# set prefix
$prefix = $cmd;
# merge opts
$opts = merge_opts($opts, $pipe, 'doc_type');
# get file opt
$file = get_opt($prefix, $opts, 'file');
if (!check_opt_set_type($cmd, $file, 'file', 'string')) {
return false;
}
# get defined fields opt
$defined_fields = get_opt($prefix, $opts, 'defined_fields');
if (!check_opt_if_set_type($cmd, $defined_fields, 'defined_fields', 'array_of_strings')) {
return false;
}
# get save fields
$save_fields = get_opt($prefix, $opts, 'save_fields');
if (!check_opt_if_set_type($cmd, $save_fields, 'save_fields', 'array_of_strings')) {
return false;
}
# get doc type opt
$doc_type = get_opt($prefix, $opts, 'doc_type', 'json');
if (!check_opt_set_type($cmd, $doc_type, 'doc_type', 'document_file_type')) {
return false;
}
$doc_type_upper = strtoupper($doc_type);
# get data structure opt
$data_structure = get_opt($prefix, $opts, 'data_structure', 'list_of_objects');
if (!check_opt_set_type($cmd, $data_structure, 'data_structure', 'data_structure_type')) {
return false;
}
# get limit opt
$limit = get_opt($prefix, $opts, 'limit', 0);
if (!check_opt_set_type($cmd, $limit, 'limit', 'positive_integer')) {
return false;
}
# get offset opt
$offset = get_opt($prefix, $opts, 'offset', 0);
if (!check_opt_set_type($cmd, $offset, 'offset', 'positive_integer')) {
return false;
}
# check the file exists
if (!file_exists($file)) {
return error($cmd, "file does not exist : {$file}");
}
# display generic info about the data
$data_structure_str = str_replace('_', ' ', $data_structure);
debug_echo($cmd, "creating data from {$doc_type_upper} file : {$file}");
debug_echo($cmd, "source data of type '{$data_structure_str}'");
# read the file into memory
$data_str = @file_get_contents($file);
if (!is_string($data_str)) {
return error($cmd, "could not read file : {$file}");
}
# load data depending on type
switch ($doc_type) {
case 'json':
$data = @json_decode($data_str, true);
break;
case 'yaml':
$data = @yaml_decode($data_str);
break;
}
if (!$data) {
return error($cmd, "invalid {$doc_type_upper} in file : {$file}");
}
$data_count = count($data);
# set up end point
if ($limit == 0) {
$end = $data_count;
} else {
$end = $offset + $limit;
}
# set up the fields and build data
$res_data = array();
switch ($data_structure) {
case 'list_of_columns':
# set up fields
if ($defined_fields) {
$defined_fields_source = '(from the config)';
} else {
$defined_fields = array_shift($data);
$defined_fields_source = '(from the source file)';
}
# set up response field indexes
if ($save_fields) {
$res_fields = $save_fields;
$res_field_count = count($res_fields);
$res_field_indexes = build_field_indexes($defined_fields, $res_fields, $cmd);
if ($res_field_indexes === false) {
return;
}
} else {
$res_fields = $defined_fields;
}
# display info about the fields
debug_echo($cmd, "defined fields {$defined_fields_source} : ");
debug_dump_list($defined_fields, true);
if ($save_fields) {
debug_echo($cmd, "saved fields :");
debug_dump_list($save_fields, true);
}
# gather the data
# TODO: change for having offset / limit
$res_data = array();
if ($save_fields) {
$res_data = array();
for ($i = $offset; $i < $end; $i++) {
$line = $data[$i];
$new_line = array();
for ($j = 0; $j < $res_field_count; $j++) {
$new_line[] = $line[$res_field_indexes[$j]];
}
$res_data[] = $new_line;
}
} elseif ($offset == 0 && $limit == 0) {
$res_data = $data;
} else {
$res_data = array();
for ($i = $offset; $i < $end; $i++) {
$res_data[] = $data[$i];
}
}
break;
case 'list_of_objects':
$res_data = array();
if ($save_fields) {
# set up result fields
$res_fields = $save_fields;
$res_fields_count = count($res_fields);
# display info about what fields will be saved
debug_echo($cmd, "saved fields :");
debug_dump_list($res_fields, true);
# build the data
for ($i = $offset; $i < $end; $i++) {
$obj = $data[$i];
$new_line = array();
for ($j = 0; $j < $res_fields_count; $j++) {
$new_line[] = @$obj[$res_fields[$j]];
}
$res_data[] = $new_line;
}
$res_fields = $save_fields;
} else {
# display info about saved files
debug_echo($cmd, "saving all fields - they will be listed as they are added");
# build the data
$res_fields = array();
$res_fields_count = 0;
for ($i = $offset; $i < $end; $i++) {
$obj = $data[$i];
$new_line = array();
# save all the existing data in order
for ($j = 0; $j < $res_fields_count; $j++) {
$new_line[] = @$obj[$res_fields[$j]];
}
# index and save any new fields
$res_fields_added = array();
foreach ($obj as $key => $value) {
if (in_array($key, $res_fields)) {
continue;
}
$res_fields_added[] = $key;
$res_fields[] = $key;
$res_fields_count++;
$new_line[] = $value;
}
$res_data[$i] = $new_line;
# display list of fields added (if any)
if ($res_fields_added) {
$line_no = $i + 1;
debug_echo($cmd, "the following fields were added on row {$line_no} :");
debug_dump_list($res_fields_added, true);
}
}
# add back in any empty fields
for ($i = 0; $i < count($res_data); $i++) {
$line =& $res_data[$i];
$line_fields_count = count($line);
if ($line_fields_count == $res_fields_count) {
break;
}
for ($j = $line_fields_count; $j < $res_fields_count; $j++) {
$line[] = '';
}
}
}
break;
}
# detail results
$line_count = count($res_data);
debug_echo($cmd, "creation of data from JSON file complete ({$line_count} lines processed)");
# create result
return build_result_data($res_fields, $res_data);
}
function map_data_fields(&$data, &$fields, $map_fields, $cmd)
{
# sort fields to map / not
$new_map_fields = array();
$ignore_fields = array();
foreach ($map_fields as $orig => $new) {
if ($new) {
$new_map_fields[$orig] = $new;
} else {
$ignore_fileds[] = $orig;
}
}
$map_fields = $new_map_fields;
# set up new field indexes
$index_fields = array_keys($map_fields);
$field_indexes = build_field_indexes($fields, $index_fields, $cmd);
if ($field_indexes === false) {
return false;
}
# display info about the map
debug_echo($cmd, "mapping data fields using the following map :");
debug_dump_yaml($map_fields, true);
if ($ignore_fields) {
debug_echo($cmd, "ignoring the following empty result fields in the map :");
debug_dump_yaml($ignore_fields, true);
}
# set up new fields
$fields = array_values($map_fields);
$field_count = count($field_indexes);
# build the data
$data_count = count($data);
for ($i = 0; $i < $data_count; $i++) {
$line = $data[$i];
$new_line = array();
for ($j = 0; $j < $field_count; $j++) {
$new_line[] = $line[$field_indexes[$j]];
}
$data[$i] = $new_line;
}
return true;
}
