public class pdfAnalysis {
/**
* @throws IOException
* @param download PDF from the Internet, intercept PDF string,
*/
public static void main(String[] args) throws IOException {
// The downloaded link is the downloaded name and the downloaded path
// pdfAnalysis.downLoadByUrl("", "KK.pdf", "F:/");
// read the file
pdfAnalysis pdf = new pdfAnalysis();
// read the file
String pdfName = "F:\\CC.pdf";
// Parse the value in the PDF and store it in the variable pdf_Body
String pdf_Body = pdf.readFileOfPDF(pdfName);
//System.out.println(pdf_Body);
String str = (pdf_Body.substring(pdf_Body.indexOf("Passenger Details"), pdf_Body.indexOf("Meal Details")));
System.out.println("I am the value of str: " + str);
int a = 0;
if (str.contains("先生") && str.contains("女士")) {
a = appearNumber(str, "先生");
System.out.println("Number of occurrences of Mr." + a);
a = appearNumber(str, "女士");
System.out.println("Number of times the lady appeared" + a);
a += a;
System.out.println("A total of " + a + " times");
} else if (str.contains("先生")) {
a = appearNumber(str, "先生");
System.out.println("Number of occurrences of Mr." + a);
} else if (str.contains("女士")) {
a = appearNumber(str, "女士");
System.out.println("Number of times the lady appeared" + a);
}
String m2=null,wm1=null;
for(int i = 0;i<a ;i++){
if(str.contains("先生")||str.contains("女士")){
String m = str.substring(str.indexOf("先生"));
m2 = m.substring(m.indexOf("先生") + 2, m.indexOf("-")).trim();
String wm = str.substring(str.indexOf("女士") + 2);
wm1 = (wm.substring(0, wm.indexOf("-"))).trim();
}else if(str.contains("女士")){
String wm = str.substring(str.indexOf("女士") + 2);
wm1 = (wm.substring(0, wm.indexOf("-"))).trim();
}else if(str.contains("先生")){
String m = str.substring(str.indexOf("先生"));
m2 = m.substring(m.indexOf("先生") + 2, m.indexOf("-")).trim();
}
}
System.out.println(m2+":"+wm1);
/* List<String> list = new ArrayList<>();
if (str.contains("先生") && str.contains("女士")) {
String m2 = null;
String wm1 = null;
for (int i = 0; i < a; i++) {
String m = str.substring(str.indexOf("先生"));
m2 = m.substring(m.indexOf("先生") + 2, m.indexOf("-")).trim();
String wm = str.substring(str.indexOf("女士") + 2);
wm1 = (wm.substring(0, wm.indexOf("-"))).trim();
}
list.add(m2);
list.add(wm1);
} else if (str.contains("先生")) {
for (int i = 0; i < a; i++) {
String m = str.substring(str.indexOf("先生") + 2);
String m2 = (m.substring(0, m.indexOf("-"))).trim();
list.add(m2);
}
} else if (str.contains("女士")) {
for (int i = 0; i < a; i++) {
String wm = str.substring(str.indexOf("女士") + 2);
String wm1 = (wm.substring(0, wm.indexOf("-"))).trim();
list.add(wm1);
}
}
for (String s : list) {
System.out.println("人名" + s);
}*/
/* String str = pdf_Body.substring(pdf_Body.indexOf("Arrival"), pdf_Body.indexOf("Payment Details"));
String str1 = str.substring(str.indexOf("H ("));
String [] pp ={"Monday","Tuesday","Wednesday","Thursday","Friday","Saturday" ,"Sunday" };
for(String sto:pp){
if(str1.contains(sto)){
String result = str1.substring(str1.indexOf(sto));
//System.out.println(result);
//System.out.println(result.length());
String result2 = result.substring(0,result.indexOf(","));
String result3 = result2.trim();
System.out.println("Hello everyone, I am Yan Laowu. The time I want: "+result3+"I am "+pdfName+"file");
}
}*/
/*if(str1.contains("Monday")||str1.contains("Tuesday")||
str1.contains("Wednesday")||str1.contains("Thursday")||
str1.contains("Friday")||str1.contains("Saturday")||str1.contains("Sunday")){
}*/
// System.out.println(str1);
/* // Get the value in the Department (leave the ground)
String depart_Temp = pdf_Body.substring(pdf_Body.indexOf("Depart"), pdf_Body.indexOf("Arrive"));
String depart_Temp2 = depart_Temp.substring(depart_Temp.indexOf("Depart"));
// System.out.println("Total value of Department taken out:" + depart_Temp2);
String depart_Temp3 = depart_Temp2.substring(depart_Temp2.indexOf("("), depart_Temp2.indexOf(")"));
// replace parentheses with empty strings and remove empty strings
String depart = depart_Temp3.replace("(", "").trim();
System.out.println("Depart:" + depart);
// Get the value of Arrive (arrival)
String arrive_Temp = pdf_Body.substring(pdf_Body.indexOf("Arrive:"), pdf_Body.indexOf("passenger details"));
String arrive_Temp1 = arrive_Temp.substring(arrive_Temp.indexOf("("), arrive_Temp.indexOf(")"));
// replace parentheses with empty strings and remove empty strings
String arrive = arrive_Temp1.replace("(", "").trim();
System.out.println("Arrive：" + arrive);
// get the money value
String money = pdf_Body.substring(pdf_Body.indexOf("AUD ") + 4, pdf_Body.indexOf("GST"));
System.out.println("Total amount:" + money);
// get the name value
String name_Temp = pdf_Body.substring(pdf_Body.indexOf("Arrive"), pdf_Body.indexOf("passenger details"));
// System.out.println(str);
String name_Temp1 = null;
String result_name = null;
List<String> list_Name = new ArrayList<>();
for (int i = 1; i < name_Temp.length(); i++) {
if (name_Temp.contains(i + ".")) {
name_Temp1 = name_Temp.substring(name_Temp.indexOf(i + "."));
result_name = name_Temp1.substring(name_Temp1.indexOf(i + ".") + 3,
name_Temp1.indexOf("Seat Number Services"));
list_Name.add(result_name);
}
// System.out.println(add);
// System.out.println(str2);
if (name_Temp1.equals("null")) {
continue;
}
}
for (String i : list_Name) {
System.out.println("All names: " + i);
}*/
if (pdfAnalysis.infile != null) {
pdfAnalysis.infile.close();
System.out.println("I'm about to close the PDF document");
}
}
public static int appearNumber(String srcText, String temp) {
int count = 0;
Pattern p = Pattern.compile(temp);
Matches m = p.matches (srcText);
while (m.find()) {
count++;
}
return count;
}
public static FileInputStream infile = null;
public String readFileOfPDF(String pdfName) throws IOException {
String context = null;
File file = new File(pdfName);//Create a file object
try {
infile = new FileInputStream(pdfName);// Create a file input stream
// Create a new PDF parser object
PDFParser parser = new PDFParser(infile);
// Parse the PDF file
parser.parse();
// Get the PDF document object obtained after parsing
PDDocument pdfdocument = parser.getPDDocument();
// Create a new PDF text stripper
PDFTextStripper stripper = new PDFTextStripper();
// Strip text from the PDF document object
context = stripper.getText(pdfdocument);
System.out.println("PDF file" + file.getAbsolutePath() + "The text content is as follows: ");
// System.out.println(context);
} catch (Exception e) {
System.out.println("Read PDF file" + file.getAbsolutePath() + "Failed!" + e.getMessage());
} finally {
if (infile != null) {
try {
infile.close();
} catch (IOException e1) {
}
}
}
return context;
}