|
|
@ -51,7 +51,7 @@ class AnalyzeByJSoup { |
|
|
|
val textS = getStringList(ruleStr) |
|
|
|
val textS = getStringList(ruleStr) |
|
|
|
return if (textS.isEmpty()) { |
|
|
|
return if (textS.isEmpty()) { |
|
|
|
null |
|
|
|
null |
|
|
|
} else join("\n", textS).trim { it <= ' ' } |
|
|
|
} else join(",", textS).trim { it <= ' ' } |
|
|
|
} |
|
|
|
} |
|
|
|
|
|
|
|
|
|
|
|
/** |
|
|
|
/** |
|
|
@ -75,7 +75,7 @@ class AnalyzeByJSoup { |
|
|
|
//拆分规则 |
|
|
|
//拆分规则 |
|
|
|
val sourceRule = SourceRule(ruleStr) |
|
|
|
val sourceRule = SourceRule(ruleStr) |
|
|
|
if (isEmpty(sourceRule.elementsRule)) { |
|
|
|
if (isEmpty(sourceRule.elementsRule)) { |
|
|
|
textS.add(element!!.data()) |
|
|
|
textS.add(element?.data() ?: "") |
|
|
|
} else { |
|
|
|
} else { |
|
|
|
val elementsType: String |
|
|
|
val elementsType: String |
|
|
|
val ruleStrS: Array<String> |
|
|
|
val ruleStrS: Array<String> |
|
|
@ -105,7 +105,7 @@ class AnalyzeByJSoup { |
|
|
|
} else { |
|
|
|
} else { |
|
|
|
getResultList(ruleStrX) |
|
|
|
getResultList(ruleStrX) |
|
|
|
} |
|
|
|
} |
|
|
|
if (temp != null && temp.isNotEmpty()) { |
|
|
|
if (!temp.isNullOrEmpty()) { |
|
|
|
results.add(temp) |
|
|
|
results.add(temp) |
|
|
|
if (results.isNotEmpty() && elementsType == "|") { |
|
|
|
if (results.isNotEmpty() && elementsType == "|") { |
|
|
|
break |
|
|
|
break |
|
|
@ -360,13 +360,15 @@ class AnalyzeByJSoup { |
|
|
|
textS.add(text) |
|
|
|
textS.add(text) |
|
|
|
} |
|
|
|
} |
|
|
|
"textNodes" -> for (element in elements) { |
|
|
|
"textNodes" -> for (element in elements) { |
|
|
|
|
|
|
|
val tn = arrayListOf<String>() |
|
|
|
val contentEs = element.textNodes() |
|
|
|
val contentEs = element.textNodes() |
|
|
|
for (i in contentEs.indices) { |
|
|
|
for (item in contentEs) { |
|
|
|
val temp = contentEs[i].text().trim { it <= ' ' } |
|
|
|
val temp = item.text().trim { it <= ' ' } |
|
|
|
if (!isEmpty(temp)) { |
|
|
|
if (!isEmpty(temp)) { |
|
|
|
textS.add(temp) |
|
|
|
tn.add(temp) |
|
|
|
} |
|
|
|
} |
|
|
|
} |
|
|
|
} |
|
|
|
|
|
|
|
textS.add(join("\n", tn)) |
|
|
|
} |
|
|
|
} |
|
|
|
"ownText", "html" -> { |
|
|
|
"ownText", "html" -> { |
|
|
|
elements.select("script").remove() |
|
|
|
elements.select("script").remove() |
|
|
|