[ 
https://issues.apache.org/jira/browse/PDFBOX-6330?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Martin Mirchev updated PDFBOX-6330:
-----------------------------------
    Attachment: ahx-ok.pdf

> ASCIIHexDecode and ASCII85Decode don't ignore all PDF white-space, so valid 
> streams decode wrongly
> --------------------------------------------------------------------------------------------------
>
>                 Key: PDFBOX-6330
>                 URL: https://issues.apache.org/jira/browse/PDFBOX-6330
>             Project: PDFBox
>          Issue Type: Bug
>          Components: Parsing
>    Affects Versions: 3.0.8 PDFBox, 4.0.0
>            Reporter: Martin Mirchev
>            Priority: Major
>         Attachments: a85-ok.pdf, a85-tabs.pdf, ahx-ok.pdf, ahx-wrapped-odd.pdf
>
>
> ISO 32000-1 (7.4.2 and 7.4.3) says that white-space characters are ignored in 
> ASCIIHexDecode and ASCII85Decode data, and PDF white-space is NUL, HT, LF, 
> FF, CR and SP (Table 1). Two cases go wrong on trunk:
> * {{ASCIIHexFilter.decode()}} skips white-space only before the first digit 
> of a pair. If white-space falls between the two digits, e.g. hex data wrapped 
> at an odd column, the white-space character is used as the second digit 
> ({{REVERSE_HEX}} gives -1), an error is logged and a wrong byte is written: 
> {{4\n142>}} decodes to {{3F 14 20}} instead of {{41 42}}.
> * {{ASCII85InputStream}} skips only SP, CR and LF. A tab, form feed or NUL 
> between groups makes decoding fail with "Invalid data in Ascii85 stream".
> Attached are two pages whose content stream ({{BT /F1 24 Tf 72 700 Td (Hello 
> PDFBox) Tj ET}}) is encoded this way: {{ahx-wrapped-odd.pdf}} (ASCIIHex in 
> 31-character lines) and {{a85-tabs.pdf}} (ASCII85 with a tab every 10 
> characters). {{PDFTextStripper.getText()}} returns an empty string for both, 
> and the pages render blank; {{ahx-ok.pdf}} and {{a85-ok.pdf}}, the same 
> content without the extra white-space, extract "Hello PDFBox". Each result 
> was the same in 3 runs.
> A possible change; with it, both files extract "Hello PDFBox" and the two 
> control files are unchanged:
> {code}
> --- a/pdfbox/src/main/java/org/apache/pdfbox/filter/ASCIIHexFilter.java
> +++ b/pdfbox/src/main/java/org/apache/pdfbox/filter/ASCIIHexFilter.java
> @@ -86,6 +86,10 @@
>              }
>              int value = REVERSE_HEX[firstByte] * 16;
>              int secondByte = encoded.read();
> +            while (isWhitespace(secondByte))
> +            {
> +                secondByte = encoded.read();
> +            }
>         
>              if (secondByte == -1 || isEOD(secondByte)) 
>              {
> --- a/pdfbox/src/main/java/org/apache/pdfbox/filter/ASCII85InputStream.java
> +++ b/pdfbox/src/main/java/org/apache/pdfbox/filter/ASCII85InputStream.java
> @@ -43,6 +43,12 @@
>      private static final char PADDING_U = 'u';
>      private static final char Z = 'z';
>  
> +    // PDF white-space characters: NUL, HT, LF, FF, CR and SP
> +    private static boolean isWhitespace(byte c)
> +    {
> +        return c == 0 || c == '\t' || c == NEWLINE || c == '\f' || c == 
> RETURN || c == SPACE;
> +    }
> +
>      /**
>       * Constructor.
>       *
> @@ -86,7 +92,7 @@
>                      return -1;
>                  }
>                  z = (byte) zz;
> -            } while (z == NEWLINE || z == RETURN || z == SPACE);
> +            } while (isWhitespace(z));
>  
>              if (z == TERMINATOR)
>              {
> @@ -114,7 +120,7 @@
>                              return -1;
>                          }
>                          z = (byte) zz;
> -                    } while (z == NEWLINE || z == RETURN || z == SPACE);
> +                    } while (isWhitespace(z));
>                      ascii[k] = z;
>                      if (z == TERMINATOR)
>                      {
> {code}
> This issue was found by an LLM-assisted tool and verified manually.



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to