lee/heritrix3
1
0
Fork 0
mirror of https://github.com/internetarchive/heritrix3.git synced 2026-09-23 06:05:46 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #122 from nlevitt/form-login-checked-boxes

Browse Source 
only submit checkbox and radio button form fields if they are on by d…
This commit is contained in:
Hunter
2016-01-14 08:00:49 -08:00
parent bf2a887688 3c4d09f6c0
commit eeeea6f0c4
2 changed files with 34 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/src/main/java/org/archive/modules/forms/ExtractorHTMLForms.java
+8 -1
View File
@@ -152,7 +152,14 @@ public class ExtractorHTMLForms extends Extractor {
String type = findAttributeValueGroup("(?i)^[^>]*\\stype\\s*=\\s*([^>\\s]+)[^>]*>",1,input);
String name = findAttributeValueGroup("(?i)^[^>]*\\sname\\s*=\\s*([^>\\s]+)[^>]*>",1,input);
String value = findAttributeValueGroup("(?i)^[^>]*\\svalue\\s*=\\s*([^>\\s]+)[^>]*>",1,input);
form.addField(type,name,value);
Matcher m = TextUtils.getMatcher("(?i)^[^>]*\\schecked\\s*[^>]*>", input);
boolean checked = false;
try {
checked = m.find();
} finally {
TextUtils.recycleMatcher(m);
}
form.addField(type, name, value, checked);
}
if (form.seemsLoginForm() || getExtractAllForms()) {
curi.getDataList(A_HTML_FORM_OBJECTS).add(form);
modules/src/main/java/org/archive/modules/forms/HTMLForm.java
+26 -4
View File
@@ -37,9 +37,14 @@ public class HTMLForm {
public String type;
public String name;
public String value;
public boolean checked = false;
@Override
public String toString() {
return type+" "+name+" "+value;
String str = "input[@type='" + type+"'][@name='" + name + "'][@value='" + value + "']";
if (checked) {
str = str + "[@checked]";
}
return str;
}
}
@@ -56,8 +61,9 @@ public class HTMLForm {
* @param type
* @param name
* @param value
* @param checked true if "checked" attribute is present (for radio buttons and checkboxes)
*/
public void addField(String type, String name, String value) {
public void addField(String type, String name, String value, boolean checked) {
FormInput input = new FormInput();
input.type = type;
// default input type is text per html standard
@@ -72,6 +78,18 @@ public class HTMLForm {
} else if ("password".equalsIgnoreCase(type)) {
candidatePasswordInputs.add(input);
}
input.checked = checked;
}
/**
* Add a discovered INPUT, tracking it as potential
* username/password receiver.
* @param type
* @param name
* @param value
*/
public void addField(String type, String name, String value) {
addField(type, name, value, false);
}
public void setMethod(String method) {
@@ -131,8 +149,12 @@ public class HTMLForm {
nameVals.add(TextUtils.urlEscape(input.name) + "=" + TextUtils.urlEscape(username));
} else if(input == candidatePasswordInputs.get(0)) {
nameVals.add(TextUtils.urlEscape(input.name) + "=" + TextUtils.urlEscape(password));
} else if (StringUtils.isNotEmpty(input.name) && StringUtils.isNotEmpty(input.value)) {
nameVals.add(TextUtils.urlEscape(input.name) + "=" + TextUtils.urlEscape(input.value));
} else if (StringUtils.isNotEmpty(input.name)
&& StringUtils.isNotEmpty(input.value)
&& (!"radio".equalsIgnoreCase(input.type)
&& !"checkbox".equals(input.type) || input.checked)) {
nameVals.add(TextUtils.urlEscape(input.name) + "="
+ TextUtils.urlEscape(input.value));
}
}