The core of search engine technology revealed_PHP tutorial-PHP Tutorial-php.cn

Editor's note: This is a wonderful programming teaching article. It not only analyzes the principles of search engines in detail, but also provides some of the author's own ideas on using PHP to compile search engines. The whole article explains it in simple terms. I believe that both experts and novices can get a lot of inspiration from it.

When talking about web search engines, most people think of Yahoo. Indeed, Yahoo created an era of Internet search. However, the technology Yahoo currently uses to search the web was not originally developed by the company. In August 2000, Yahoo adopted technology from Google (www.google.com), a venture company founded by Stanford University students. The reason is very simple. Google's search engine can search for the required information faster and more accurately than the technology previously used by Yahoo.

It is probably impossible for us to design and develop a powerful and efficient search engine and database ourselves in a short period of time in terms of technology and funds. However, since Yahoo is using other people’s technology, then we are Can't we also use other people's ready-made search engine websites?

Analysis of programming ideas

We can imagine this: simulate a query, issue a search command in a corresponding format to a search engine website, then return the search results, and analyze the HTML code of the results , strip off redundant characters and codes, and finally display it in our own website page in the required format.

In this way, the key to the problem is that we need to select a search information that is accurate (so that our search will be more meaningful) and fast (because we need extra time to analyze the search results and display them) , a search website with concise search results (convenient for HTML source code analysis and stripping). Due to various excellent features of the new generation search engine Google, here we choose it as an example to see how to use PHP to implement background processing of Google (www. google.com) search and front-end personalized display process.

Let’s first take a look at the composition of Google’s query commands. Enter the www.google.com website, enter "abcd" in the query bar, and click the query button. We can find that the browser's address bar changes to: "http://www.google.com/search?q=abcd&btnG=Google %CB%D1%CB%F7&hl=zh-CN&lr=", it can be seen that Google passes the query parameters and submits the query command through the get method of the form. We can use the file() function in PHP to simulate this query process.

Understand the File() function

Syntax: array file(string filename);

The return value is an array, and all files are read into the array variable. The file here can be local or remote. The remote file must indicate the protocol used. For example: result=file(“http://www.google.com/search?q=abcd&btnG=Google%CB%D1%CB%F7&hl=zh-CN&lr=”), this statement will simulate our query of words on Google "abcd" process and transfer the search results back to the array variable result in the form of elements per row. Because the file read here is remote, the protocol name "http://" cannot be missing.

If we want users to enter search characters for any search, we can make an input text box and submit button, and replace the searched character "abcd" above with a variable:
echo '';

if (isset( keywords)) //After submission, PHP will generate the variable kwywords, which requires the following program to run after submission
{
urlencode( keywords); //URL encode user input content
result=file("http://www.google.com/search?q=". keywords."&btnG=Google%CB%D1%CB%F7&hl=zh- CN&lr=");
//Replace variables in the query statement and save the query results in the array variable result
result_string=join(" ", result); //Merge the array $result into a string. Paste each array element with spaces
... //Further processing
}
?>

The above program can already query based on user input content, and Synthesize the returned result into a string variable $result_string. Please note that you must use the urlencode() function to URL-encode the user input so that the input Chinese characters, spaces, and other special characters can be queried normally. This also simulates Google's query commands as realistically as possible to ensure the search results are accurate. Correctness.

Analysis of Google

For ease of understanding, let’s assume that what we really need is: the title of the search results. URL and introduction, etc. This is a simple and typical requirement. In this way, all we have to do is: remove the header and footer of Google search results, including a Google logo, input box for re-search, search result description, etc., and strip the original HTML from the remaining search result items. Format tag, replace it with the format we want.

To do this, we must carefully analyze the HTML source code of Google search results and find the patterns. It is not difficult to find that the text of Google search results is always included between the first

tag and the penultimate

tag of the source code, and the penultimate

tag is followed by the table character , and this combination "

All the following procedures are continued in the "further processing" section of the above procedure.

　result_string = strstr( result_string, "

"); //Get the string after result_string starting from the first

to remove the Google header
position= strpos( result_string, "The position of the

table symbol
result_string= substr(result_string,0, position);//Intercept the string before the first

table symbol to remove the footnote

Application and Implementation

OK, now that we have obtained the useful HTML source code backbone, the remaining problem is how to display this content independently. Let’s analyze these search result entries and find that there are also very regular patterns between each entry. are separated by
, that is, each is a paragraph. According to this feature, we use the explode() function to cut each entry:

Syntax: explode(string separator, string string);

Returns an array, and each small string split by separator is saved in the array.

So:
result_array=explode("

", result_string); //Use words. String "

" cuts the result into

We get an array result_array, where each element is a search result entry. All we have to do is study each entry and its HTML display format code. , and then replace as required. Use a loop to process each entry in result_array
for( i=0; i {
... //Process each entry
}
.
For each entry, we can easily find some characteristics: each entry is composed of title, abstract, introduction, category, URL, etc. Each part is line-wrapped, that is, it contains the
tag, so it is divided again: (The following handlers are placed in the loop above)
every_item=explode("
", result_array[ i]);

　 In this way we get an array every_item, where every_item[0] is Title, every_item[1] and every_item[2] are two lines of summary, if the headers of every_item[3], every_item[4], etc. contain "Introduction:", "< font size=-1 color=#6f6f6f>Category:< /font>" characters, it is an introduction or category (because some result entries do not have this item). If the header contains "< font color=green>", it must be the URL. We often use regular expressions (omitted) for this comparison and judgment. It is also very convenient if you want to replace it. For example, $every_item[0] containing the title is itself linked. We hope to modify the link attribute so that it can be displayed in a new window. Open the link:
echo eregi_replace(' {
... //Process each item in each entry except the first item (the first item is the title, which has been displayed)
... // More format modifications
}

This modifies the link attributes, and many other display format modifications, stripping, and replacements can be completed using regular replacement eregi_replace().

At this point we have obtained each item of each search item, and can modify the format of each item at will, and even put a beautiful table on it. However, a good program should be able to adapt to various operating environments, and here is no exception. In fact, we have only discussed a framework method for HTML stripping of search results. To really do it perfectly, there are many things to consider, such as It displays the total number of search results, how many pages it is divided into, etc. It can even remove the "category", "introduction" and other codes related to Google, so that customers cannot see the original website at all. However, we can all extract these contents and requirements by analyzing HTML. Now everyone can do it themselves and build a highly personalized search engine.

Excerpted from: http://tech.163.com/tm/010228/010228_15747.html
Author: maxid