From fda8fc728e0696277780e63a32e2429047fb24f7 Mon Sep 17 00:00:00 2001
From: Benjamin Loison <Benjamin_Loison@users.noreply.gitea.lemnoslife.com>
Date: Tue, 7 Feb 2023 20:15:36 +0100
Subject: [PATCH] #31: Add zip files search

---
 website/index.php     | 47 ++++++++++++++++++++++++++++++++++++++++++-
 website/search.py     | 33 ++++++++++++++++++++++--------
 website/websocket.php |  2 ++
 3 files changed, 73 insertions(+), 9 deletions(-)

diff --git a/website/index.php b/website/index.php
index 3a5dba1..742b474 100644
--- a/website/index.php
+++ b/website/index.php
@@ -14,9 +14,50 @@ See <?php echoUrl('https://gitea.lemnoslife.com/Benjamin_Loison/YouTube_captions
     <input type="submit" value="Search">
 </form>
 
+<ul id="channels">
+</ul>
+
 <script>
     var firstRun = true;
     var conn;
+    // Could parse DOM instead of using following variable.
+    var channels = [];
+
+    function createA(text, href) {
+        var a = document.createElement('a');
+        var text = document.createTextNode(text);
+        a.appendChild(text);
+        a.href = href;
+        return a;
+    }
+
+    function treatLine(line) {
+        console.log(line);
+        if (line.startsWith('alert:')) {
+            alert(line.replace('alert:', ''));
+        } else {
+            var channelsDom = document.getElementById('channels');
+            const channelFileParts = line.split('/');
+            const channel = channelFileParts[0];
+            const channelFile = channelFileParts[1];
+            const channelHref = `channels/${channel}`;
+            if (!channels.includes(channel)) {
+                channels.push(channel);
+                channelDom = document.createElement('li');
+                var a = createA(channel, channelHref);
+                channelDom.appendChild(a);
+                var channelFilesDom = document.createElement('ul');
+                channelDom.appendChild(channelFilesDom);
+                channelsDom.appendChild(channelDom);
+            }
+            var channelDom = channelsDom.lastChild;
+            var channelFilesDom = channelDom.lastChild;
+            var channelFileDom = document.createElement('li');
+            var a = createA(channelFile, `${channelHref}/${channelFile}`);
+            channelFileDom.appendChild(a);
+            channelFilesDom.appendChild(channelFileDom);
+        }
+    }
 
     function search(event) {
         // We don't want to refresh the webpage which is the default behavior.
@@ -25,11 +66,15 @@ See <?php echoUrl('https://gitea.lemnoslife.com/Benjamin_Loison/YouTube_captions
         if (firstRun) {
             firstRun = false;
             conn = new WebSocket('wss://crawler.yt.lemnoslife.com/websocket');
-            conn.onmessage = function(e) { console.log(e.data); };
+            conn.onmessage = function(e) {
+                e.data.split('\n').forEach(treatLine);
+            };
             // We can't directly proceed with `conn.send`, as the connection may not be already established.
             conn.onopen = function(e) { conn.send(query); };
         } else {
             // We assume at this point that the connection is established.
+            channels = [];
+            document.getElementById('channels').innerHTML = '';
             conn.send(query);
         }
     }
diff --git a/website/search.py b/website/search.py
index beddc38..c8f0bda 100755
--- a/website/search.py
+++ b/website/search.py
@@ -1,6 +1,8 @@
 #!/usr/bin/python3
 
-import sys, time, fcntl, os
+import sys, time, fcntl, os, zipfile
+
+path = '/mnt/HDD0/YouTube_captions_search_engine/channels/'
 
 clientId = sys.argv[1]
 message = sys.argv[2]
@@ -8,21 +10,35 @@ message = sys.argv[2]
 clientFilePath = f'users/{clientId}.txt'
 
 def write(s):
-    f = open(clientFilePath, 'w+')
+    f = open(clientFilePath, 'r+')
     try:
         fcntl.flock(f, fcntl.LOCK_EX)
         # If the output file is empty, then it means that `websocket.php` read it. Anyway we don't wait it and we append what we want to output.
         read = f.read()
-        f.write(f"{read}\n{s}")
+        # We are appening content, as we moved in-file cursor.
+        if read != '':
+            f.write("\n")
+        f.write(s)
+        f.flush()
+        fcntl.flock(f, fcntl.LOCK_UN)
+        f.close()
     except Exception as e:
         sys.exit(e)
-    f.close()
 
-for i in range(10):
-    write(f'{i}: {message}')
-    time.sleep(2)
+# Unclear if `os.listdir` takes a lot of time, as it's a generator.
+# As `zipgrep` doesn't support arguments to stop on first match for each file, we proceed manually to keep a good theoretical complexity.
+for file in os.listdir(path):
+    if file.endswith('.zip'):
+        zip = zipfile.ZipFile(path + file)
+        for fileInZip in zip.namelist():
+            f = zip.open(fileInZip)
+            for line in f.readlines():
+                if message in str(line):
+                    write(f'{file}/{fileInZip}')
+                    break
+            f.close()
 
-f = open(clientFilePath, 'r')
+f = open(clientFilePath)
 while True:
     try:
         fcntl.flock(f, fcntl.LOCK_EX)
@@ -30,6 +46,7 @@ while True:
             os.remove(clientFilePath)
             break
         else:
+            fcntl.flock(f, fcntl.LOCK_UN)
             time.sleep(1)
     except Exception as e:
         sys.exit(e)
diff --git a/website/websocket.php b/website/websocket.php
index ea20dca..6bf4ace 100644
--- a/website/websocket.php
+++ b/website/websocket.php
@@ -96,6 +96,7 @@ class MyProcess implements MessageComponentInterface
         if (preg_match("/^[a-zA-Z0-9-_ ]+$/", $msg) !== 1) {
             return;
         }
+        $from->send('alert:Started searching...');
         $client = $this->clients->offsetGet($from);
         // If a previous request was received, we execute the new one with another client for simplicity otherwise with current file deletion approach, we can't tell the worker `search.py` that we don't care about its execution anymore.
         if ($client->pid !== null) {
@@ -137,6 +138,7 @@ class MyProcess implements MessageComponentInterface
             } else {
                 // We don't need the periodic timer anymore, as the worker finished its work and acknowledged that `websocket.php` completely read its output.
                 $this->loop->cancelTimer($client->timer);
+                $from->send('alert:Search finished!');
             }
         });
     }