Merge pull request #5 from splitbrain/codesearch

Rebase the plugin on grok

.gitattributes

@@ -0,0 +1,2 @@
+/.* export-ignore
+/_test export-ignore

.github/workflows/phpTestLinux.yml

@@ -0,0 +1,52 @@
+name: PHP Tests on Linux
+
+on: [push, pull_request]
+
+jobs:
+    testLinux:
+        name: PHP ${{ matrix.php-versions }} DokuWiki ${{ matrix.dokuwiki-branch }}
+        runs-on: ubuntu-latest
+        if: github.event_name == 'push' || github.event.pull_request.head.repo.full_name != github.repository
+
+        strategy:
+            matrix:
+                php-versions: ['7.2', '7.3', '7.4', '8.0']
+                dokuwiki-branch: [ 'master', 'stable']
+                exclude:
+                    -   dokuwiki-branch: 'stable'
+                        php-versions: '8.0'
+            fail-fast: false
+
+        steps:
+            - name: Checkout
+              uses: actions/checkout@v2
+
+            - name: Setup PHP
+              uses: shivammathur/setup-php@v2
+              with:
+                  php-version: ${{ matrix.php-versions }}
+                  extensions: mbstring, intl, PDO, pdo_sqlite, bz2
+
+            - name: Setup problem matchers
+              run: |
+                echo ::add-matcher::${{ runner.tool_cache }}/php.json
+                echo ::add-matcher::${{ runner.tool_cache }}/phpunit.json
+
+            - name: Download DokuWiki Test-setup
+              run: wget https://raw.github.com/splitbrain/dokuwiki-travis/master/travis.sh
+
+            - name: Run DokuWiki Test-setup
+              env:
+                  CI_SERVER: 1
+                  DOKUWIKI: ${{ matrix.dokuwiki-branch }}
+              run: sh travis.sh
+
+            - name: Setup PHPUnit
+              run: |
+                  php _test/fetchphpunit.php
+                  cd _test
+
+            - name: Run PHPUnit
+              run: |
+                cd _test
+                php phpunit.phar --verbose --stderr --group plugin_xref

Grok.php

@@ -0,0 +1,76 @@
+<?php
+
+namespace dokuwiki\plugin\xref;
+
+use dokuwiki\HTTP\DokuHTTPClient;
+
+class Grok
+{
+
+    protected $baseUrl;
+    protected $def;
+    protected $path;
+
+    public function __construct($reference, $baseUrl = 'https://codesearch.dokuwiki.org')
+    {
+        $heuristic = new Heuristics($reference);
+        $this->def = $heuristic->getDef();
+        $this->path = $heuristic->getPath();
+        $this->baseUrl = rtrim($baseUrl, '/');
+    }
+
+    /**
+     * Return the URL that leads to the search interface
+     *
+     * @return string
+     */
+    public function getSearchUrl()
+    {
+        if ($this->def === '' && $this->path === '') return $this->baseUrl;
+
+        $url = $this->baseUrl . '/search?';
+        $param = [
+            'project' => 'dokuwiki',
+            'defs' => $this->def,
+            'path' => $this->path,
+        ];
+        $url .= buildURLparams($param, '&');
+        return $url;
+    }
+
+    /**
+     * Return the URL that allows to query the API
+     *
+     * @return string
+     */
+    public function getAPIUrl()
+    {
+        $url = $this->baseUrl . '/api/v1/search?';
+        $param = [
+            'projects' => 'dokuwiki',
+            'def' => $this->def,
+            'path' => $this->path,
+        ];
+        $url .= buildURLparams($param, '&');
+        return $url;
+    }
+
+    /**
+     * Return the number of results to expect
+     *
+     * @return false|int false on errors
+     */
+    public function getResultCount()
+    {
+        if ($this->def === '' && $this->path === '') return 0;
+
+        $http = new DokuHTTPClient();
+        $http->timeout = 5;
+        $json = $http->get($this->getAPIUrl());
+        if (!$json) return false;
+        $data = json_decode($json, true);
+        if (!$data) return false;
+        if (!isset($data['resultCount'])) return false;
+        return $data['resultCount'];
+    }
+}

Heuristics.php

@@ -0,0 +1,169 @@
+<?php
+
+namespace dokuwiki\plugin\xref;
+
+/**
+ * Figure out what to send to Grok to hopefully show the right, single hit
+ */
+class Heuristics
+{
+
+    /** @var string the definition to search */
+    protected $def = '';
+    /** @var string the path to use */
+    protected $path = '';
+
+    /**
+     * Try to gues what the given reference means and how to best search for it
+     *
+     * @param string $reference
+     */
+    public function __construct($reference)
+    {
+        if ($reference !== '') $reference = $this->checkHash($reference);
+        if ($reference !== '') $reference = $this->checkFilename($reference);
+        if ($reference !== '') $reference = $this->checkNamespace($reference);
+        if ($reference !== '') $reference = $this->checkClassPrefix($reference);
+        if ($reference !== '') $reference = $this->checkVariable($reference);
+        if ($reference !== '') $reference = $this->checkFunction($reference);
+        if ($reference !== '') $reference = $this->checkPSRClass($reference);
+        if ($reference !== '') $this->def = $reference;
+    }
+
+    /**
+     * @return string
+     */
+    public function getDef()
+    {
+        return trim(preg_replace('/[^\w]+/', '', $this->def));
+    }
+
+    /**
+     * @return string
+     */
+    public function getPath()
+    {
+        return trim(preg_replace('/[^\w.]+/', ' ', $this->path));
+    }
+
+    /**
+     * Handle things in the form path#symbol
+     *
+     * @param string $reference
+     * @return string
+     */
+    protected function checkHash($reference)
+    {
+        if (strpos($reference, '#') === false) return $reference;
+        list($this->path, $this->def) = explode('#', $reference, 2);
+        return '';
+    }
+
+    /**
+     * Known file extension?
+     *
+     * @param string $reference
+     * @return mixed|string
+     */
+    protected function checkFilename($reference)
+    {
+        if (preg_match('/\.(php|js|css|html)$/', $reference)) {
+            $this->def = '';
+            $this->path = $reference;
+            return '';
+        }
+        return $reference;
+    }
+
+    /**
+     * Namespaces are paths
+     *
+     * @param string $reference
+     * @return string
+     */
+    protected function checkNamespace($reference)
+    {
+        if (strpos($reference, '\\') === false) return $reference;
+
+        $parts = explode('\\', $reference);
+        $parts = array_filter($parts);
+        $reference = array_pop($parts); // last part may be more than a class
+
+        // our classes are in inc
+        if ($parts[0] == 'dokuwiki') $parts[0] = 'inc';
+
+        $this->path = join(' ', $parts);
+
+        return $reference;
+    }
+
+    /**
+     * Is there something called on a class?
+     */
+    protected function checkClassPrefix($reference)
+    {
+        if (
+            strpos($reference, '::') === false &&
+            strpos($reference, '->') === false
+        ) {
+            return $reference;
+        }
+        list($class, $reference) = preg_split('/(::|->)/', $reference, 2);
+
+        $this->path .= ' ' . $class;
+        $this->def = $reference;
+        return '';
+    }
+
+    /**
+     * Clearly a variable
+     *
+     * @param string $reference
+     * @return string
+     */
+    protected function checkVariable($reference)
+    {
+        if ($reference[0] == '$') {
+            $this->def = $reference;
+            return '';
+        }
+        return $reference;
+    }
+
+    /**
+     * It's a function
+     *
+     * @param string $reference
+     * @return string
+     */
+    protected function checkFunction($reference)
+    {
+        if (substr($reference, -2) == '()') {
+            $this->def = $reference;
+            return '';
+        }
+        if (preg_match('/\(.+?\)$/', $reference)) {
+            [$reference, /* $arguments */] = explode('(', $reference, 2);
+            $this->def = $reference;
+            return '';
+        }
+        return $reference;
+    }
+
+    /**
+     * Upercase followed by lowercase letter, must be a class
+     *
+     * Those are in their own files, so add it to the path
+     * @param $reference
+     * @return mixed|string
+     */
+    protected function checkPSRClass($reference)
+    {
+        if (preg_match('/^[A-Z][a-z]/', $reference)) {
+            $this->def = $reference;
+            $this->path .= ' ' . $reference;
+            return '';
+        }
+        return $reference;
+    }
+}