summaryrefslogtreecommitdiff
path: root/cmd/dehyphenate
diff options
context:
space:
mode:
authorNick White <git@njw.name>2020-08-25 16:03:42 +0100
committerNick White <git@njw.name>2020-08-25 16:03:42 +0100
commit090c9986c9b93f45de70f21f649877536839dc29 (patch)
tree0f54d8e393ffe836bc2adfc1502c89fdb07b33c6 /cmd/dehyphenate
parent496fe0af1cb212a6a4af932b6c188bf082d15383 (diff)
Move dehyphenate string code into its own function
Diffstat (limited to 'cmd/dehyphenate')
-rw-r--r--cmd/dehyphenate/main.go47
1 files changed, 27 insertions, 20 deletions
diff --git a/cmd/dehyphenate/main.go b/cmd/dehyphenate/main.go
index afd1aac..0974368 100644
--- a/cmd/dehyphenate/main.go
+++ b/cmd/dehyphenate/main.go
@@ -24,6 +24,32 @@ import (
// - loses any formatting; doesn't need to be identical, but e.g. linebreaks after elements would be handy
// - need to handle OcrChar
+// dehyphenateString replaces hyphens at the end of a line
+// with the first word from the following line, and removes
+// that word from its line.
+func dehyphenateString(in string) string {
+ var newlines []string
+ lines := strings.Split(in, "\n")
+ for i, line := range lines {
+ words := strings.Split(line, " ")
+ last := words[len(words)-1]
+ // the - 2 here is to account for a trailing newline and counting from zero
+ if len(last) > 0 && last[len(last) - 1] == '-' && i < len(lines) - 2 {
+ nextwords := strings.Split(lines[i+1], " ")
+ if len(nextwords) > 0 {
+ line = line[0:len(line)-1] + nextwords[0]
+ }
+ if len(nextwords) > 1 {
+ lines[i+1] = strings.Join(nextwords[1:], " ")
+ } else {
+ lines[i+1] = ""
+ }
+ }
+ newlines = append(newlines, line)
+ }
+ return strings.Join(newlines, "\n")
+}
+
func main() {
flag.Usage = func() {
fmt.Fprintf(os.Stderr, "Usage: dehyphenate [-hocr] in out\n")
@@ -63,26 +89,7 @@ func main() {
}
}
} else {
- var newlines []string
- lines := strings.Split(string(in), "\n")
- for i, line := range lines {
- words := strings.Split(line, " ")
- last := words[len(words)-1]
- // the - 2 here is to account for a trailing newline and counting from zero
- if len(last) > 0 && last[len(last) - 1] == '-' && i < len(lines) - 2 {
- nextwords := strings.Split(lines[i+1], " ")
- if len(nextwords) > 0 {
- line = line[0:len(line)-1] + nextwords[0]
- }
- if len(nextwords) > 1 {
- lines[i+1] = strings.Join(nextwords[1:], " ")
- } else {
- lines[i+1] = ""
- }
- }
- newlines = append(newlines, line)
- }
- finaltxt = strings.Join(newlines, "\n")
+ finaltxt = dehyphenateString(string(in))
}
f, err := os.Create(flag.Arg(1))