added README example that makes replacements in multiple files #230
Conversation
|
Nice work @sam-mininberg! |
| @@ -220,6 +220,28 @@ script.Echo("a\nb\nc").FilterScan(func(line string, w io.Writer) { | |||
| // scanned line: "c" | |||
| ``` | |||
|
|
|||
| Alternatively, we can use a `Filter` function that returns a `string`. For example, let's fix typos we've made in the CSV files in our working directory: | |||
There was a problem hiding this comment.
I wonder if referencing CSV files is a red herring here. Readers might think this example is something to do with CSV parsing. Should we just talk about making arbitrary edits to all files matching a pattern?
There was a problem hiding this comment.
Sorry for disappearing for awhile! Yes I think that's better. I'll make that change.
| ```go | ||
| script.ListFiles("*.csv").FilterLine(func(file string) string { | ||
| search := "typ" | ||
| replace := "typo" |
There was a problem hiding this comment.
Maybe we should use completely different search and replace strings, as this particular search and replace would not be idempotent: if you keep running it, "typo" will become "typoo", and so on.
There was a problem hiding this comment.
Good point! I'll change it to "helo" --> "hello".
| script.ListFiles("*.csv").FilterLine(func(file string) string { | ||
| search := "typ" | ||
| replace := "typo" | ||
| s, err := File(file).String() |
There was a problem hiding this comment.
Reading the whole file into a single string is a bit limiting—and unnecessarily so when we only need to read it line by line. What about using script's own Replace method, if that's possible here?
There was a problem hiding this comment.
That's a good point! I originally chose to read in the whole file like this so that I could use strings.Count. But I agree with your other comment that the counts are not essential to this example. I'll update my snippet to use Replace instead.
There was a problem hiding this comment.
Quick update: I'm working on using Replace but am running into #137. I've yet to come up with an elegant example that works around the issue, but I'll keep at it.
There was a problem hiding this comment.
Yes, a temporary file is essential here, isn't it? That's what sed does (presumably).
| replace := "typo" | ||
| s, err := File(file).String() | ||
| if err != nil { | ||
| return fmt.Sprintf("%s %s", file, err) |
There was a problem hiding this comment.
Maybe it's better after all to use Filter, then, since that can return an error and stop the pipe if necessary; something doesn't feel right about just flattening it like this and carrying on. What do you think a sensible text-replacement program should do if it hits an error with some file? What would the equivalent sed command do?
There was a problem hiding this comment.
I created a test script to test what the equivalent sed would do:
$ bash test.sh
1.txt before sed: helo world!
3.txt before sed: helo there!
Running "sed -i 's/helo/hello/g' 1.txt 2.txt 3.txt"
sed: can't read 2.txt: No such file or directory
1.txt after sed: hello world!
3.txt after sed: hello there!
Even though sed doesn't behave this way, I do agree with you that it doesn't feel right to continue after encountering an error.
It seems like two approaches are:
- Use
Filterand stop the pipe at the first error - Use
FilterLineand put any error messages on the pipe
With all of this in mind, which do you prefer? Or perhaps there is a third approach I'm not considering?
There was a problem hiding this comment.
Well, you're the customer: as the person who wants to replace text in multiple files, which behaviour would be more convenient or natural to you?
There was a problem hiding this comment.
Stopping at the first error seems more natural to me. I'll update my implementation accordingly.
| if err != nil { | ||
| return fmt.Sprintf("%s %s", file, err) | ||
| } | ||
| count := strings.Count(s, search) |
There was a problem hiding this comment.
It's nice to have a count of the number of replacements, but not essential—and the key to a good example is eliminating all non-essential detail. I don't deny that we probably would want a real script text-replacement program to do this, but examples like this are more effective when they're as brief and focused as possible, wouldn't you say?
Addresses #229