Reading a certain line

scross

How could you read a certain line from a file, so for example you have a file:

row1column1|row1column2|row1column3|row1column4|row1column5
row2column1|row2column2|row2column3|row2column4|row2column5
row3column1|row3column2|row3column3|row3column4|row3column5
row4column1|row4column2|row4column3|row4column4|row4column5

I know you can read through the lines with fgets() but I want to know whether there is a way to get only a certain line without having to load the whole file (or any other rows) into php (imagine a file with 1000+ rows).

EDIT: Note that each line does not have a fixed size.

bpat1434

You can't really read just one line into PHP, but you can get the lines like so:

$contents = file_get_contents('path/to/file.txt');
$lines = explode("\n", $contents);

Now, you can use a preg_match to find the one line you're looking for, or just get the line number you want, and subtract 1 to get the array ID.

scross

bpat1434 wrote:
You can't really read just one line into PHP, but you can get the lines like so:
$contents = file_get_contents('path/to/file.txt');
$lines = explode("\n", $contents);
Now, you can use a preg_match to find the one line you're looking for, or just get the line number you want, and subtract 1 to get the array ID.

Yes, but with a large file (eg 10M😎 it will most likely slow down.

bpat1434

Seems like you're stuck between a rock and a hard place then.... either way, the file has to be completely read into memory (or a string) for manipulation. You can't just pull an arbitrary line (unless it's the first/last line. In which case [man]fread/man would be your friend...

bradgrafelman

As bpat said, you either have to know where exactly in the file you want to start reading data from and use a function like [man]fseek/man or just [man]fread/man, or else you have to load the entire file and then use other PHP functions to find the line you want.

scross

Thanks for your help. I have an idea: I could place some sort of definition at the start of the file (or in another file) with a line name next to the position and then the script would look through that to find the position and then do a quick call to fseek. I'll try it in a sec...

Once again, thanks for your help 🙂

Weedpacket

fgets() will read line-by-line. (More accurately: it will read the number of bytes you specify or to the end of the current line, whichever comes first). You still have to read the whole file up to that line, so that you can find the linebreaks, but you do not have to read the whole file.

scross

Placing the positions of the lines at the beginning was effective and I seperated them by a colon. All I had to do then was read the first line containing the positions and use explode() on it. However, I found that seperating the positions by a newline and using fgets to search through the positions was very efficient since I only had to load a single position number each time rather than the whole set.

Weedpacket

Yup; what you can then do is put the list of positions in a separate "index" file (so that the main file would be kept purely for the table data itself); the file positions in the file wouldn't then be altered by changes in the list of positions at the start.

Then you'd be halfway to implementing your own DBMS. 😉

scross

Ok, but I have a problem when extracting the data. I basically pull each row out and see if it should be in the results. The problem is that I'm not sure what to do with each row that is included in the results, should I keep them all in an results array or should I refer to them by what line each result is on and then pull out each result as it is requested??

EDIT: Forgot to say that I might need to sort the results so would an array be quicker/easier??

my_name

Are you guys saying that if you had a 1000-line file, and the line you needed just so happened to be line 900, you'd have to read the first 899 before it?

scross

no, the idea is that the character position of the start of each line is saved so you can automatically move to that line and read it quickly. You can also read backwards from the end of the file but this can be quite slow. Otherwise you do have to read all the lines before it.