Merge pull request #3619 from Kelimion/csv_iterator

Add iterator_next(&r) to CSV.
This commit is contained in:
Jeroen van Rijn
2024-05-24 14:06:42 +02:00
committed by GitHub
2 changed files with 111 additions and 2 deletions

View File

@@ -0,0 +1,88 @@
//+build ignore
package encoding_csv
import "core:fmt"
import "core:encoding/csv"
import "core:os"
// Requires keeping the entire CSV file in memory at once
iterate_csv_from_string :: proc(filename: string) {
r: csv.Reader
r.trim_leading_space = true
r.reuse_record = true // Without it you have to delete(record)
r.reuse_record_buffer = true // Without it you have to each of the fields within it
defer csv.reader_destroy(&r)
if csv_data, ok := os.read_entire_file(filename); ok {
csv.reader_init_with_string(&r, string(csv_data))
defer delete(csv_data)
} else {
fmt.printfln("Unable to open file: %v", filename)
return
}
for r, i, err in csv.iterator_next(&r) {
if err != nil { /* Do something with error */ }
for f, j in r {
fmt.printfln("Record %v, field %v: %q", i, j, f)
}
}
}
// Reads the CSV as it's processed (with a small buffer)
iterate_csv_from_stream :: proc(filename: string) {
fmt.printfln("Hellope from %v", filename)
r: csv.Reader
r.trim_leading_space = true
r.reuse_record = true // Without it you have to delete(record)
r.reuse_record_buffer = true // Without it you have to each of the fields within it
defer csv.reader_destroy(&r)
handle, errno := os.open(filename)
if errno != os.ERROR_NONE {
fmt.printfln("Error opening file: %v", filename)
return
}
defer os.close(handle)
csv.reader_init(&r, os.stream_from_handle(handle))
for r, i in csv.iterator_next(&r) {
for f, j in r {
fmt.printfln("Record %v, field %v: %q", i, j, f)
}
}
fmt.printfln("Error: %v", csv.iterator_last_error(r))
}
// Read all records at once
read_csv_from_string :: proc(filename: string) {
r: csv.Reader
r.trim_leading_space = true
r.reuse_record = true // Without it you have to delete(record)
r.reuse_record_buffer = true // Without it you have to each of the fields within it
defer csv.reader_destroy(&r)
if csv_data, ok := os.read_entire_file(filename); ok {
csv.reader_init_with_string(&r, string(csv_data))
defer delete(csv_data)
} else {
fmt.printfln("Unable to open file: %v", filename)
return
}
records, err := csv.read_all(&r)
if err != nil { /* Do something with CSV parse error */ }
defer {
for rec in records {
delete(rec)
}
delete(records)
}
for r, i in records {
for f, j in r {
fmt.printfln("Record %v, field %v: %q", i, j, f)
}
}
}

View File

@@ -57,6 +57,9 @@ Reader :: struct {
field_indices: [dynamic]int,
last_record: [dynamic]string,
sr: strings.Reader, // used by reader_init_with_string
// Set and used by the iterator. Query using `iterator_last_error`
last_iterator_error: Error,
}
@@ -121,6 +124,25 @@ reader_destroy :: proc(r: ^Reader) {
bufio.reader_destroy(&r.r)
}
/*
Returns a record at a time.
for record, row_idx in csv.iterator_next(&r) { ... }
TIP: If you process the results within the loop and don't need to own the results,
you can set the Reader's `reuse_record` and `reuse_record_reuse_record_buffer` to true;
you won't need to delete the record or its fields.
*/
iterator_next :: proc(r: ^Reader) -> (record: []string, idx: int, err: Error, more: bool) {
record, r.last_iterator_error = read(r)
return record, r.line_count - 1, r.last_iterator_error, r.last_iterator_error == nil
}
// Get last error if we the iterator
iterator_last_error :: proc(r: Reader) -> (err: Error) {
return r.last_iterator_error
}
// read reads a single record (a slice of fields) from r
//
// All \r\n sequences are normalized to \n, including multi-line field
@@ -460,5 +482,4 @@ _read_record :: proc(r: ^Reader, dst: ^[dynamic]string, allocator := context.all
r.fields_per_record = len(dst)
}
return dst[:], err
}
}