so really in French?
"Qu'est ce que c'est que cette chose là ?"

I use on some of our projects a very convenient created by me over several years of work on other projects, a flexible and fast defender of the site from requests of intruders and unauthorized parsing. The script class is written with minimal dependency on other modules, just in connection, does not require installation and has a flexible configuration. The essence of the defender's work: every IP will be checked (with caching) from reliable several (alternate, for reliability) sources. In this case, reliability is also possible on other grounds. There is a list (automatically extensible) of white and black IP (v6 or v4).
The ip list of some search engines is automatically added to the white list, ip is automatically added to the black when certain conditions are met. The number of requests per day, hour, the total number of requests, requests for a specific URL, requests with certain parameters, GET POST HEAD, etc., cookies, https protocols, etc. After installation, the daily load on the server decreased by more than 10-20 % - there were more than 50 useless search engines indexing despite the fact that they are prohibited in the robots file. Successful attempts at parsing sites are also reflected. In this case, almost all ordinary users have access to sites. The final decision to block ip can be taken instantly through transitions to the appropriate links to the mail. The essence of suspicious requests to the site is also understood.

For several years of work on many sites and services I used the same script with the increase of functional. Over time, this script became almost perfect. It has minimal connectivity and dependency on other modules, which ensures fast integration with any framework or CMS, it proportionally imposes a logo on the right place. This script can resize the original image with the specified size or add. Parameters. Also, he can take images above the root of the site, that is, unavailable originals from the outside, he can cache. It can log or notify emails about "called" missing images. He can check the referrer and if there is an insert from another site, he gives a message instead of an image, and if the picture is open to the whole screen (without the site), then redirects to the article the original ... In general, a very cool module turned out!

Several years ago, during the use of search on the created sites, I thought about why the search is not given due attention, because sometimes it will be convenient for the user to find through a convenient search on the site, and not through navigation, especially if the user is the first time on the site. Otherwise, the user simply leaves the site and searches for the search that he needs, but not through the search, but through a third-party service of the Google type, and this is forever lost potential customers (visitors) of the site.
In his spare time, he decided to complete a full-fledged search, it turned out that this is quite a labor-consuming activity, but the game is worth the candle. Because there is no advertising on competitors like using third-party embedded search engines, you can also accentuate and control search output absolutely as the service owner or marketers conceived, personal (and statistical) logging of all operations of a particular user is performed, and this increases the information content About the preference of visitors, which can then be used, for example, in mailings or in offers of the product (service) of interest.

To implement all this, I carried out the following: the lematization and definition of the parts of speech of all the words that are included in all the articles.
I built 2 indexes to implement the reverse index,
Experimentally, he selected a metric influencing the relevance of the issue. I also implemented queries and methods for the sample itself.
Then, noticed that the data for computer (automatic) correction of errors in the request is enough, and decided to apply one of the methods of correcting errors in the search query. (It turned out that quite often the user enters a query with errors in the search field). It has become very pleasant to use the search, which itself offers options for a valid search query.
And at the end I tried different methods of coloring the search results. In general, I am satisfied with the search engine I implemented on php + Laravel. I also want to try to add to the system a synonym and account for the distance between words when ranking)). By the way, on the basis of numerous solved subtasks, it was possible to quickly solve the task of calculating and displaying the time of reading the article ...