/src/php/document/get-document.php

<?php
/* Copyright 2012 Aditi Muralidharan. See the file "LICENSE" for the full license governing this code. */

/****************************************************************************
	get-document.php
	
	Functions for fetching a single document's sub-structure, along with
	the metadata associated with all sub-structures.
****************************************************************************/
include_once '../util.php';
include_once '../subsets/read.php';

$wordseer_instance = getGetParam('instance');
$path = '../../../instances/'.$wordseer_instance.'/config.php';
include_once $path;

//Query parameters
$gov = getGetParam('gov');
$govtype = getGetParam('govtype');
$dep = getGetParam('dep');
$deptype = getGetParam('deptype');
$relation = getGetParam('relation');
$collection = getGetParam('collection');
$statistics = getGetParam('statistics');
$phrasess = decodeGetJson('phrases');
$metadata = decodeGetJson('metadata');
$timing = getGetParam('timing');

$documentID = getGetParam('id');
$include_text = getGetParam('include_text') == 'true';

if ($documentID) {
	$documentUnit = getDocument($documentID, $include_text);
	echo json_encode($documentUnit);
}

/** Retrieves information about the document with the given ID. If 
$include_text is true, also includes the text of the document, otherwise only
retrieves the metadata. The fields returned are those expected by 
{@link WordSeer.model.DocumentModel}.
*/
function getDocument($document_id, $include_text) {
	$units = array();
	$chilren = array();
	$metadata = array();
	if (!$include_text) {
		// Assemble the metadata for each unit and send it back.
		$sql = "SELECT * from metadata, metadata_structure
		 		WHERE metadata.document_id = $document_id
				AND metadata.property_id = metadata_structure.property_id
				AND metadata.unit_id = $document_id
				AND metadata.unit_name = 'document';";
		$result = mysql_query($sql) or die (mysql_error(). " Getting metadata
			<br> on query 
			<br> $sql
			<at> documentviewer/get-document.php l. 75.");
		while ($row = mysql_fetch_assoc($result)) {
			array_push($metadata, $row);
		}
	} else {
		// Assemble the full text for each unit that matches the given filters
		// and send it back.

		//Query parameters
		$collection = getGetParam('collection');
		$phrasess = decodeGetJson('phrases');
		$metadata = decodeGetJson('metadata');
		include_once "../document/get-metadata.php";
		$sql = "SELECT * from document_structure 
			WHERE document_id = $document_id
			ORDER BY unit_number ASC;";
			$sentence_ids_for_filters = getSentenceIDsForFilters($metadata, 
			$collection, $phrases);
		if ($sentence_ids_for_filters != "all") {
			$unit_ids = getUnitsFromSentenceIDs($sentence_ids_for_filters);
			$unit_id_string = join(", ", $unit_ids);
			$sentence_id_string = join(", ", $sentence_ids_for_filters);
			$sql = " SELECT *
			FROM document_structure
			WHERE unit_id in ($unit_id_string)
			OR unit_id in ($sentence_id_string)
			ORDER BY unit_id, unit_number ASC;";
		}
		$result = mysql_query($sql) or die (mysql_error(). " Getting document units
			<br> on query 
			<br> $sql
			<at> documentviewer/get-document.php l. 75.");
		$units = array("document"=>array(
			$document_id=>array(
				"metadata"=>array(),
				"unit_id"=>$document_id,
				"unit_name"=>"document",
				)));
		$children = array();
		$parent_ids = array();
		$unit_ids = array();
		$word_set_memberships = getWordSetMemberships();
		while ($row = mysql_fetch_assoc($result)) {
			$unit_name = $row["unit_name"];
			$unit_id = $row["unit_id"];
			array_push($unit_ids, $unit_id);
			$unit_number = $row["unit_number"];
			$parent_id = $row["parent_id"];
			$parent_name = $row["parent_name"];
			if (!array_key_exists($unit_name, $units)) {
				$units[$unit_name] = array();
				$parent_ids[$unit_name] = array();
			}
			if (!array_key_exists($unit_id, $units[$unit_name])){
				$units[$unit_name][$unit_id] = $row;
				$units[$unit_name][$unit_id]["metadata"] = array();
				$parent_ids[$unit_name][$unit_id] = array($parent_id, $parent_name);
				if ($unit_name == "sentence") {
					$units[$unit_name][$unit_id]["words"] = array();
					$units[$unit_name][$unit_id]["sentence_id"] = $unit_id;
				} else {
					if (!array_key_exists($unit_name, $children)) {
						$children[$unit_name] = array();
					}
					if (!array_key_exists($unit_id, $children[$unit_name])) {
						$children[$unit_name][$unit_id] = array();
					}
				}
			}
			if (!array_key_exists($parent_name, $children)) {
				$children[$parent_name] = array();
			}
			if (!array_key_exists($parent_id, $children[$parent_name])) {
				$children[$parent_name][$parent_id] = array();
			}
			array_push($children[$parent_name][$parent_id],
				array("id"=>$unit_id, "name"=>$unit_name));
		}

		// Assemble the metadata for each unit under each unit type and id.
		$unit_id_string = join(", ", $unit_ids);
		$sql = "SELECT *, metadata.unit_name as unit_name from metadata, metadata_structure
		 		WHERE metadata.document_id = $document_id
				AND metadata.property_id = metadata_structure.property_id
				AND unit_id in ($unit_id_string);";
		if ($timing != 0) {
			echo "<br> $sql <br>";
		}
		$result = mysql_query($sql) or die (mysql_error(). " Getting metadata
			<br> on query 
			<br> $sql
			<at> documentviewer/get-document.php l. 75.");
		while ($row = mysql_fetch_assoc($result)) {
			array_push($units[$row["unit_name"]][$row["unit_id"]]["metadata"], 
				$row);
		}
		// The top-level metadata for the document is under the "document"
		// unit, so pull it out.
		$metadata = $units["document"][$document_id]["metadata"];

		// Get the words in each sentence in the document.
		$sql = "SELECT surface, space_after, word_id, sentence_id 
		from sentence_xref_word WHERE document_id = $document_id
		AND sentence_id in ($unit_id_string)
		ORDER BY position ASC;";
		$result = mysql_query($sql) or die (mysql_error(). " Getting words
			<br> on query 
			<br> $sql
			<at> documentviewer/get-document.php l. 75.");
		while ($row = mysql_fetch_assoc($result)) {
			$word = array(
				'word'=>replaceWeirdCharacters($row['surface']),
				'word_id'=>$row['word_id'],
				'space_after'=>str_replace("\n", "<br>", $row['space_after'])
			);
			if (array_key_exists($row['word_id'], $word_set_memberships)) {
				$word['word_set'] = join(" ", 
					$word_set_memberships[$row['word_id']]);
			}
			array_push($units["sentence"][$row["sentence_id"]]["words"], $word);
		}
	}
	$sql = "SELECT title from document WHERE id = $document_id;";
	$result = mysql_query($sql) or die(mysql_error()." getting document title:
										<br> $sql");
	$row = mysql_fetch_assoc($result);
	$title = $row['title'];
	$results = array(
		"has_text"=>$include_text,
		"units"=>$units, 
		"children"=>$children,
		"id"=>$document_id, 
		"title"=>$row["title"],
		"metadata"=>$metadata);
	foreach ($metadata as $metadata_information) {
		$results[$metadata_information["property_name"]] = 
			$metadata_information["value"];
	}
	array_merge($results, $results["metadata"]);
	return $results;
}


?>