Merge pull request #3 from gsu-library/develop

Code Reorganization

CHANGELOG.md

@@ -4,6 +4,9 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.2.0] - 2023-02-01
+- Reorganize code.
+
 ## [1.1.0] - 2022-09-28
 - Update .htaccess to use mod_authz_core directives.
 - Change config.ini to config.php.

README.md

@@ -3,7 +3,7 @@ Code Repository: https://github.com/gsu-library/datacite-bulk-doi-creator-webapp
 Author: Matt Brooks <[email protected]>  
 Date Created: 2022-06-29  
 License: [GPL3](LICENSE)  
-Version: 1.1.0
+Version: 1.2.0
 
 ## Description
 A PHP WebApp that bulk creates DataCite DOIs from a provided CSV file. DOIs are created in the findable state. If you are looking for the python version of this WebApp see [DataCite Bulk DOI Creator](https://github.com/gsu-library/datacite-bulk-doi-creator).

config/config.sample.php

@@ -1,5 +1,9 @@
 <?php
-
+/**
+ * Configuration.
+ *
+ * @var array  [url, doiPrefix, username, password, maxSubmittedFiles, maxReportedFiles, maxUploadSize]
+ */
 const CONFIG = [
    'url' => 'https://api.datacite.org/dois',
    'doiPrefix' => '',

includes/functions.php

@@ -1,18 +1,36 @@
 <?php
-// Go back to the index page.
+/**
+ * Redirects browser to the index page.
+ *
+ * // TODO: check to see if this needs to be loaded on any pages or just on includes.
+ *
+ * @return void
+ */
 function go_home() {
    header('location: .');
    exit;
 }
 
 
-// Loads the configuration file.
+/**
+ * Loads the configuration file.
+ *
+ * Configuration will be in a CONFIG variable.
+ *
+ * @return void
+ */
 function load_config_file() {
    require_once('config'.DIRECTORY_SEPARATOR.'config.php');
 }
 
 
-// Lists out files in either the reports or uploads directories.
+/**
+ * Lists out files in either the reports or uploads directories.
+ *
+ * @param string  $type    Folder to parse.
+ * @param integer $amount  Number of results to return.
+ * @return void
+ */
 function list_files($type, $amount) {
    $files = glob($type.DIRECTORY_SEPARATOR.'*.csv');
 
@@ -38,7 +56,12 @@ function list_files($type, $amount) {
 }
 
 
-// Prints the navigation, marking the current page as active.
+/**
+ * Prints the navigation, marking the current page as active.
+ *
+ * @param string $currentPage Filename of the current page.
+ * @return void
+ */
 function print_header($currentPage) {
    echo '
    <header class="container my-3">
@@ -89,7 +112,11 @@ function print_header($currentPage) {
 }
 
 
-// Prints the footer.
+/**
+ * Prints the footer.
+ *
+ * @return void
+ */
 function print_footer() {
    echo '
    <footer class="container my-3">

includes/index_functions.php

@@ -0,0 +1,39 @@
+<?php
+/**
+ * Prints the $output session variable.
+ *
+ * @return void
+ */
+function print_output() {
+   if(isset($_SESSION['output'])) {
+      echo implode('<br>', $_SESSION['output']);
+   }
+
+   $_SESSION['output'] = [];
+}
+
+
+/**
+ * Prints a link to the current report found in session variable $reportPath.
+ *
+ * @return void
+ */
+function print_report_link() {
+   if(isset($_SESSION['reportPath']) && $_SESSION['reportPath']) {
+      echo '<a href="'.htmlspecialchars($_SESSION['reportPath'], ENT_QUOTES).'" download>Download Report</a>';
+   }
+
+   unset($_SESSION['reportPath']);
+}
+
+
+/**
+ * Sets the CSRF token in session ($csrfToken).
+ *
+ * @return void
+ */
+function set_csrf_token() {
+   if(!isset($_SESSION['csrfToken'])) {
+      $_SESSION['csrfToken'] = bin2hex(random_bytes(32));
+   }
+}

includes/submit_functions.php

@@ -0,0 +1,219 @@
+<?php
+/**
+ * Check for PHP cURL and that both the reports and uploads directories are writable.
+ *
+ * @return void
+ */
+function check_capabilities() {
+   $go_home = false;
+
+   if(!function_exists('curl_init')) {
+      array_push($_SESSION['output'], 'Please install/enable the PHP cURL library.');
+      $go_home = true;
+   }
+
+   if(!is_writable('uploads')) {
+      array_push($_SESSION['output'], 'The uploads directory is not writable.');
+      $go_home = true;
+   }
+
+   if(!is_writable('reports')) {
+      array_push($_SESSION['output'], 'The reports directory is not writable.');
+      $go_home = true;
+   }
+
+   if($go_home) {
+      go_home();
+   }
+}
+
+
+/**
+ * Validates CSRF token.
+ *
+ * @return void
+ */
+function validate_csrf_token() {
+   if(!isset($_POST['csrfToken']) || !isset($_SESSION['csrfToken'])) {
+      array_push($_SESSION['output'], 'CSRF token not found.');
+      go_home();
+   }
+   else {
+      if($_POST['csrfToken'] !== $_SESSION['csrfToken']) {
+         array_push($_SESSION['output'], 'The CSRF token is invalid.');
+         go_home();
+      }
+      else {
+         unset($_SESSION['csrfToken']);
+      }
+   }
+}
+
+
+/**
+ * Check if all needles exist in haystack.
+ *
+ * @param   array $needles    Array of values to search for.
+ * @param   array $haystack   Array of values to search in.
+ * @return  bool
+ */
+function in_array_all($needles, $haystack) {
+   return empty(array_diff($needles, $haystack));
+}
+
+
+/**
+ * Remove oldest files from directory.
+ *
+ * @param   string   $filePattern   File name pattern to search for.
+ * @param   integer  $maxFileCount  Max number of files to keep.
+ * @return  void
+ */
+function remove_old_files($filePattern, $maxFileCount) {
+   if($maxFileCount < 1) {
+      $maxFileCount = 1;
+   }
+
+   $files = glob($filePattern);
+   $extraFiles = count($files) - $maxFileCount;
+
+   if($extraFiles > 0) {
+      // Sort by last modified ascending.
+      usort($files, function($x, $y) {
+         return filemtime($x) > filemtime($y);
+      });
+
+      for($i = 0; $i < $extraFiles; $i++) {
+         unlink($files[$i]);
+      }
+   }
+}
+
+
+/**
+ * Return a valid file name.
+ *
+ * @param   string      $fileName Desired file name.
+ * @return  string|null A valid file name.
+ */
+function find_file_name($fileName) {
+   if(!file_exists($fileName)) {
+      return $fileName;
+   }
+
+   $fileParts = pathinfo($fileName);
+   $fileCount = count(glob($fileParts['dirname'] . DIRECTORY_SEPARATOR . "*"));
+
+   for($i = 1; $i <= $fileCount; $i++) {
+      $tempName = $fileParts['dirname'] . DIRECTORY_SEPARATOR . $fileParts['filename'] . " ($i)." . $fileParts['extension'];
+
+      if(!file_exists($tempName)) {
+         return $tempName;
+      }
+   }
+
+   return null;
+}
+
+
+/**
+ * Process uploaded file. Checks for upload file, file size, and file name.
+ *
+ * @return string Full file path of the uploaded file.
+ */
+function process_uploaded_file() {
+   // Check upload and move it to folder.
+   if($fileName = $_FILES['fileUpload']['name'] ?? null) {
+      if($_FILES['fileUpload']['size'] > CONFIG['maxUploadSize']){
+         array_push($_SESSION['output'], 'The uploaded file is too large.');
+         go_home();
+      }
+
+      if(!($filePath = find_file_name('uploads'.DIRECTORY_SEPARATOR.$fileName))) {
+         array_push($_SESSION['output'], 'There was an error saving the uploaded file.');
+         go_home();
+      }
+
+      move_uploaded_file($_FILES['fileUpload']['tmp_name'], $filePath);
+      remove_old_files('uploads'.DIRECTORY_SEPARATOR.'*.csv', CONFIG['maxSubmittedFiles']);
+
+      return $filePath;
+   }
+   else {
+      array_push($_SESSION['output'], 'Could not find uploaded file.');
+      go_home();
+   }
+}
+
+
+/**
+ * Processes the headers in the uploaded file. Checks to make sure all required headers are present.
+ *
+ * @param   resource $uploadFp File pointer to upload file.
+ * @return  array    Array of normalized headers from the upload file.
+ */
+function process_upload_headers($uploadFp) {
+   // Headers required to process the upload file.
+   $requiredHeaders = [
+      'doi_suffix',
+      'title',
+      'year',
+      'type',
+      'description',
+      'publisher',
+      'source_url',
+      'creator1',
+      'creator1_type', // TODO: will depend on orchid id
+                     // don't require and assume personal?
+      'creator1_given',
+      'creator1_family',
+   ];
+
+   // Save file headers.
+   if(($headers = fgetcsv($uploadFp)) === false) {
+      array_push($_SESSION['output'], 'No data was found in the uploaded file.');
+      go_home();
+   }
+
+   // Trim and lowercase headers in CSV file.
+   $headers = array_map(function($header) {
+      return strtolower(trim($header));
+   }, $headers);
+
+   // Make sure CSV file has all required headers.
+   // TODO: specify missing headers
+   if(!in_array_all($requiredHeaders, $headers)) {
+      array_push($_SESSION['output'], 'The uploaded CSV file is missing required headers.');
+      go_home();
+   }
+
+   return $headers;
+}
+
+
+/**
+ * Opens up a file to create a report. The file name is based on the submitted file.
+ *
+ * @param   string   $uploadFullPath Full path of the uploaded file.
+ * @return  resource File pointer of opened report file.
+ */
+function open_report_file($uploadFullPath) {
+   $fileParts = pathinfo($uploadFullPath);
+   $fileName = $fileParts['filename'];
+
+   // Open a file for the upload report.
+   if(!($reportFullPath = find_file_name('reports'.DIRECTORY_SEPARATOR.basename($fileName).' report.csv'))) {
+      array_push($_SESSION['output'], 'There was an error saving the report file.');
+      go_home();
+   }
+
+   if(!$reportFp = fopen($reportFullPath, 'w')) {
+      array_push($_SESSION['output'], 'Cannot write to the reports folder.');
+      go_home();
+   }
+
+   // Save report path to add link on index page.
+   $_SESSION['reportPath'] = $reportFullPath;
+
+   return $reportFp;
+}

index.php

@@ -1,31 +1,9 @@
 <?php
 session_start();
 require_once('includes'.DIRECTORY_SEPARATOR.'functions.php');
+require_once('includes'.DIRECTORY_SEPARATOR.'index_functions.php');
 load_config_file();
-
-// Print output session variable.
-function print_output() {
-   if(isset($_SESSION['output'])) {
-      echo implode('<br>', $_SESSION['output']);
-   }
-
-   $_SESSION['output'] = [];
-}
-
-
-// Prints link to current report.
-function print_report_link() {
-   if(isset($_SESSION['reportPath']) && $_SESSION['reportPath']) {
-      echo '<a href="'.htmlspecialchars($_SESSION['reportPath'], ENT_QUOTES).'" download>Download Report</a>';
-   }
-
-   unset($_SESSION['reportPath']);
-}
-
-
-if(!isset($_SESSION['csrfToken'])) {
-   $_SESSION['csrfToken'] = bin2hex(random_bytes(32));
-}
+set_csrf_token();
 ?>
 <!DOCTYPE html>
 <html lang="en">