C editing a file

[PhilSorum]

Hi all.
I've to edit a txt file, deleting a given word.

My approach is to open the file and read it char by char, copying that to a second file.
I use a buffer to identify the given word, so if I recognize the pattern I don't pass it to the new file. Finally I copy the new file to the former.

Is there a better way to carry out my task, maybe editing "in loco" (without using a support file)?
Before opening the file I know in which row/rows the word is located.
For example: cat, row 3, row 5 twice, row 13. Is it someway useful to speed up the process?

[kreagan]

Hi all.
I've to edit a txt file, deleting a given word.

My approach is to open the file and read it char by char, copying that to a second file.
I use a buffer to identify the given word, so if I recognize the pattern I don't pass it to the new file. Finally I copy the new file to the former.

Is there a better way to carry out my task, maybe editing "in loco" (without using a support file)?
Before opening the file I know in which row/rows the word is located.
For example: cat, row 3, row 5 twice, row 13. Is it someway useful to speed up the process?

If you have the row/column, you can target the word quickly.

if you know nothing about the word, (assuming you must maintain order or the file), I would search the file. If you are worried about efficiency, maybe you can compair letter by letter instead of whole string compairs.

If there is some order to the file, you can possibly make a cleaver algorithm. For example, if your file is sorted, you can perform different searches to enhance efficiency.

[PhilSorum]

If you have the row/column, you can target the word quickly.

if you know nothing about the word, (assuming you must maintain order or the file), I would search the file. If you are worried about efficiency, maybe you can compair letter by letter instead of whole string compairs.

If there is some order to the file, you can possibly make a cleaver algorithm. For example, if your file is sorted, you can perform different searches to enhance efficiency.

Hello, thanks for the reply.
My file is not sorted, it's a random txt file, for example a program readme.txt or a forum FAQ, so words can occur in any order.
I know the row/rows in which the word occurs, if any, but I haven't the column.
As in the example above I know:

word: cat, row 3, row 5 twice, row8.
word: dog, not present.

I'm in search for a good efficiency!

[kreagan]

Hello, thanks for the reply.
My file is not sorted, it's a random txt file, for example a program readme.txt or a forum FAQ, so words can occur in any order.
I know the row/rows in which the word occurs, if any, but I haven't the column.
As in the example above I know:

word: cat, row 3, row 5 twice, row8.
word: dog, not present.

I'm in search for a good efficiency!

I really don't know what you have done or your limitations, so my ideas are unbounded.

instead of parsing the file word for word, I would suggest using strstr function to point to every instance of that string.
strstr

[PhilSorum]

I really don't know what you have done or your limitations, so my ideas are unbounded.

instead of parsing the file word for word, I would suggest using strstr function to point to every instance of that string.
strstr

I have no limitations, my program can be simplified to this recipe:
take a random txt file
take a random word
a magician tells you how many occurrence of the given word are located in the file and in what rows they occur
you have to edit the file, removing them
is the magician hint useful?
is it possible to carry out the task without using a second file?

[StarvingLinguist]

the way I can think of to do the job with the fewest resources is to set an fseek write point at the beginning of the discovered word, and write the letters which follow the word into the space the word used to be in. continuing to the end of the file skipping (advancing the read point by one word size) each point where the word is again recognized. If those are disk access commands it might be a little slower than doing the same thing with a large array of all the characters and rewriting the file back to disk.

I hope your programming tasks eventually become more stimulating.

[RRick]

You're going to have to be careful and not delete words that have your search word contained in them. For example, if you need to delete cat in the following sentence, "Did you catch the cat?", you probably don't want "Did you ch the cat?"

Since your search data is centered on lines, then the simplest thing to do is read the complete file into an array of lines. Now you can get to the line you want very easily. Once there, parse by word instead of character and now the comparison is easy.

You'll have to watch out for delimiters because when you read the last word in the sentence, you'll get "cat?", not just cat.

[PhilSorum]

You're going to have to be careful and not delete words that have your search word contained in them. For example, if you need to delete cat in the following sentence, "Did you catch the cat?", you probably don't want "Did you ch the cat?"

Since your search data is centered on lines, then the simplest thing to do is read the complete file into an array of lines. Now you can get to the line you want very easily. Once there, parse by word instead of character and now the comparison is easy.

Ok, I like this approach.
Once I've built an array of lines with every line edited (removing the word), I imagine I've to copy again all the lines in the source file. If a line contains only the word I've to cancel, I can directly skip that line and go on with next line or some adjustment on the file size/allocation are required?

[RRick]

You are going to have to write out the file again with this approach. You've got two choices, write to a new file or overwrite the original file. I prefer option 1 because you can compare the two files.

Be careful with lines that contain only your search word. Does your app want you to delete the line or make it an empty line. These are two different things. An empty line only has a '\n' but that is still something.

[PhilSorum]

When I have only the given word in a row, I have to remove the whole line, so decreasing the total number of rows in the file.
I have a question: if I have to use a backup file where copying data to, is it better to create an array of lines from the original file, as just said, or I can simply scan the file and copy in the dest file char by char?