[RESOLVED] Extracting time stamps from pdf

Hi, I am trying to automatically extract all time stamps in a pdf file. These are typically in a line like:

when="2010-07-30T15:20:30+04:00"

For this I was thinking of using CStdioFile and the ReadString function. Somehow this doesn't work. My example code is below. Is this because pdf is not a true text file, because strings read can be longer than some max,...? What is my mistake or anyone have another quick way of reading the file and extracting the desired text between the brackets?

Code:

---

CStdioFile InputFile;

if (InputFile.Open(FileName,CFile::modeRead))
{
 CString Line; CString ToFind; ToFind.Format("when");
       
 while (InputFile.ReadString(Line))
 {
  if (Line.Find(ToFind)!=-1)
  {
  CString Item; AfxExtractSubString(Item,Line,0,'"');
  AfxMessageBox(Item); // to be replaced with further processing
  }
 }

 InputFile.Close();
}

---

Well, open your .pdf file in the notepad and search for "when" text. Will it be found?

Quote:

---

Originally Posted by VictorN

Well, open your .pdf file in the notepad and search for "when" text. Will it be found?

---

Already did that. What if there are a gazillion instances? I want to automate it. I thought the ten lines above could do it. Obviously I was wrong.

Quote:

---

Originally Posted by Simon666

Already did that. What if there are a gazillion instances? I want to automate it. I thought the ten lines above could do it. Obviously I was wrong.

---

Did what? Did you open your (it means the file ttat for sure contains the text you are looking for).pdf file in the notepad?
If you did it then what about the search result?

I don't use CStdioFile, but in general, if you open a file up
in text mode and read while looping, you can come across
EOF early. Try opening in binary mode.

InputFile.Open(FileName,CFile::modeRead|CFile::typeBinary))

Let's me read a lot further but it still exits way before the actual end of the file.

Some things to consider:

1) I do not think that CString::Find() will work if their are embedded NULLS in the
string before the time stamp.

2) one "line" could contain multiple time stamps

I've looked at several .pdf files and none of them have when in them. The only time related stuff that I could find was /CreationDate eg. /CreationDate (D:20060208110100)

Quote:

---

Originally Posted by 2kaud

I've looked at several .pdf files and none of them have when in them. The only time related stuff that I could find was /CreationDate eg. /CreationDate (D:20060208110100)

---

Yeah, It was the first thing I've done before posting my first answer:

Quote:

---

Originally Posted by VictorN

Well, open your .pdf file in the notepad and search for "when" text. Will it be found?

---

Just because I coudn't find "when" (either as ANSI or UNICODE) in some of my .pdf files!
However, OP seems to ignore my opnion... :eek:

Quote:

---

Originally Posted by Philip Nicoletti

Some things to consider:

1) I do not think that CString::Find() will work if their are embedded NULLS in the
string before the time stamp.

2) one "line" could contain multiple time stamps

---

1) I checked that doesn't occur.
2) Same.

Quote:

---

Originally Posted by 2kaud

I've looked at several .pdf files and none of them have when in them. The only time related stuff that I could find was /CreationDate eg. /CreationDate (D:20060208110100)

---

It is metadata of pictures from Adobe Photoshop.

Quote:

---

Originally Posted by VictorN

Just because I coudn't find "when" (either as ANSI or UNICODE) in some of my .pdf files!
However, OP seems to ignore my opnion... :eek:

---

Victor, I did not ignore you. I addressed that issue specifically. There are about 80 entries per pdf and 10 pdf's. I didn't want to do roughly all 800 of them manually when roughly 10 lines of code could do but by now the time spent will be about equal.

Quote:

---

Originally Posted by Simon666

Victor, I did not ignore you. I addressed that issue specifically. There are about 80 entries per pdf and 10 pdf's. I didn't want to do roughly all 800 of them manually when roughly 10 lines of code could do but by now the time spent will be about equal.

---

So could you find in notepad at least one of these "80 entries per pdf" or not?

Yes I could, why?

What "why"?
Are they found as ANSI or UNICODE?
Is your buils ANSI or UNICODE?

Anyway, I got an idea, I might use GetPosition and check if it is anywhere near the end of the file, if not just call ReadString again in a while loop. I'll first try that.

Quote:

---

Originally Posted by Simon666

Anyway, I got an idea, I might use GetPosition and check if it is anywhere near the end of the file, if not just call ReadString again in a while loop. I'll first try that.

---

Or you could respond to Victor's questions.

Quote:

---

Originally Posted by Simon666

Victor, I did not ignore you. I addressed that issue specifically. There are about 80 entries per pdf and 10 pdf's. I didn't want to do roughly all 800 of them manually when roughly 10 lines of code could do but by now the time spent will be about equal.

---

What is the significance of the file being PDF?

All you're really asking is "how to find text in a quick way in a file". It doesn't matter if the file is PDF or not. The only thing that would need to be known is whether the file can contain control characters or not. Then you need to open the file in binary mode if it contains NULLs or control characters.

Regards,

Paul McKenzie

Code:

---

ToFind.Format("when");

---

I don't have access to VS compiler right now ... but would
that compile if using a UNICODE build ?

Quote:

---

Originally Posted by VictorN

What "why"?
Are they found as ANSI or UNICODE?
Is your buils ANSI or UNICODE?

---

I use Multi Byte character set in my project configuration. The pdf I have no clue how to find out.

I just opened a couple of PDF files in a binary editor. They appear to be a mixture of binary and text data, lots of embedded nulls. The binary data means that you can't effectively use CString and CStdioFile to manipulate them.

Can you run the following console code and see if the results
are what you expect ? Change the name of the input file
and check the results.txt file when done.

Code:

---

#include <fstream>
#include <string>
#include <iostream>

int main()
{
    using namespace std;

    ifstream in("your pdf file",ios::binary);

    if (!in)
    {
        cout << "could not open file\n";
        cin.ignore(100,'\n');

        return 0;
    }

    ofstream out("results.txt");

    string line;

    int count = 0;

    while (getline(in,line))
    {
        size_t start = 0;
        size_t pos;

        while ( (pos=line.find("when",start)) != string::npos )
        {
            size_t pos1 = line.find('"',pos);
            size_t pos2 = line.find('"',pos1+1);

            if (pos1 != string::npos && pos2!=string::npos)
            {
                ++count;
                start = pos2 + 1;
                out << line.substr(pos1+1,pos2-pos1-1) << "\n";
            }
            else
            {
                // ill formed ... skip
                start = pos + 1;
            }
        }
    }

    out << "number of occurrences = " << count << "\n";
   
    return 0;
}

---

Anyway, this did the trick:

Code:

---

        CStdioFile InputFile;

        if (InputFile.Open(FileName,CFile::modeRead|CFile::typeBinary))
        {
        CString Line; CString ToFind; ToFind.Format("when");

        ULONGLONG End = InputFile.SeekToEnd();

        InputFile.SeekToBegin();
       
        while ( (InputFile.ReadString(Line)) || (InputFile.GetPosition()!=End) )
        {
          if (Line.Find(ToFind)!=-1)
          {
          //Further processing
          }
        }

        InputFile.Close();
        }

---

Quote:

---

Originally Posted by Philip Nicoletti

Can you run the following console code and see if the results
are what you expect ? Change the name of the input file
and check the results.txt file when done.

Code:

---

#include <fstream>
#include <string>
#include <iostream>

int main()
{
    using namespace std;

    ifstream in("your pdf file",ios::binary);

    if (!in)
    {
        cout << "could not open file\n";
        cin.ignore(100,'\n');

        return 0;
    }

    ofstream out("results.txt");

    string line;

    int count = 0;

    while (getline(in,line))
    {
        size_t start = 0;
        size_t pos;

        while ( (pos=line.find("when",start)) != string::npos )
        {
            size_t pos1 = line.find('"',pos);
            size_t pos2 = line.find('"',pos1+1);

            if (pos1 != string::npos && pos2!=string::npos)
            {
                ++count;
                start = pos2 + 1;
                out << line.substr(pos1+1,pos2-pos1-1) << "\n";
            }
            else
            {
                // ill formed ... skip
                start = pos + 1;
            }
        }
    }

    out << "number of occurrences = " << count << "\n";
   
    return 0;
}

---

---

Wow, that works even better. Thanks a lot. There were 400+ occurrences in one file alone, I had seriously underestimated the number of occurrences and the manual work it would have taken. There are further time stamp entries of slightly different formatting but I can handle things now from here on. Thanks to everyone who weighed in. :thumb: