alex/yomichan-import
1
Fork 0
Code Releases Activity

cleanup

Browse Source
This commit is contained in:
Alex Yatskov 2016-12-13 18:15:29 -08:00
parent 26c55ee91d
commit cf7eecb8d3
1 changed files with 9 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
daijirin.go
View File

@ -29,7 +29,7 @@ import (
func makeDaijirinExtractor() epwingExtractor { func makeDaijirinExtractor() epwingExtractor {
return &daijirinExtractor{ return &daijirinExtractor{
partsExp: regexp.MustCompile(`(?P<reading>[^（【〖]+)(?:【(?P<expression>.*)】)?(?:〖(?P<native>.*)〗)?(?:（(?P<tag>.*)）)?`), partsExp: regexp.MustCompile(`([^（【〖]+)(?:【(.*)】)?(?:〖(.*)〗)?(?:（(.*)）)?`),
phonExp: regexp.MustCompile(`[-・]+`), phonExp: regexp.MustCompile(`[-・]+`),
variantExp: regexp.MustCompile(`\((.*)\)`), variantExp: regexp.MustCompile(`\((.*)\)`),
annotExp: regexp.MustCompile(`（(.*)）`), annotExp: regexp.MustCompile(`（(.*)）`),
@ -40,15 +40,9 @@ func (e *daijirinExtractor) extractTerms(entry epwingEntry) []dbTerm {
var expressions, readings, glossary, tags []string var expressions, readings, glossary, tags []string
matches := e.partsExp.FindStringSubmatch(entry.Heading) matches := e.partsExp.FindStringSubmatch(entry.Heading)
for i, name := range e.partsExp.SubexpNames() { if matches != nil {
value := matches[i] if expression := matches[2]; len(expression) > 0 {
if i == 0 || len(value) == 0 { expression = e.annotExp.ReplaceAllLiteralString(expression, "")
continue
}
switch name {
case "expression":
expression := e.annotExp.ReplaceAllLiteralString(value, "")
for _, split := range strings.Split(expression, `・`) { for _, split := range strings.Split(expression, `・`) {
splitInc := e.variantExp.ReplaceAllString(split, "$1") splitInc := e.variantExp.ReplaceAllString(split, "$1")
expressions = append(expressions, splitInc) expressions = append(expressions, splitInc)
@ -57,16 +51,17 @@ func (e *daijirinExtractor) extractTerms(entry epwingEntry) []dbTerm {
expressions = append(expressions, splitExc) expressions = append(expressions, splitExc)
} }
} }
case "reading": }
reading := e.phonExp.ReplaceAllLiteralString(value, "")
if reading := matches[1]; len(reading) > 0 {
reading = e.phonExp.ReplaceAllLiteralString(reading, "")
readings = append(readings, reading) readings = append(readings, reading)
} }
} }
for i, split := range strings.Split(entry.Text, "\n") { for i, split := range strings.Split(entry.Text, "\n") {
if i == 0 { if i == 0 {
matches := e.annotExp.FindStringSubmatch(split) if matches := e.annotExp.FindStringSubmatch(split); matches != nil {
if len(matches) >= 1 {
tags = append(tags, strings.Split(matches[1], `・`)...) tags = append(tags, strings.Split(matches[1], `・`)...)
} }
} }