hayro_syntax/
pdf.rs

1//! The starting point for reading PDF files.
2
3use crate::PdfData;
4use crate::object::Object;
5use crate::page::Pages;
6use crate::page::cached::CachedPages;
7use crate::reader::Reader;
8use crate::xref::{XRef, XRefError, fallback, root_xref};
9use std::sync::Arc;
10
11pub use crate::crypto::DecryptionError;
12use crate::metadata::Metadata;
13
14/// A PDF file.
15pub struct Pdf {
16    xref: Arc<XRef>,
17    header_version: PdfVersion,
18    pages: CachedPages,
19    data: PdfData,
20}
21
22/// An error that occurred while loading a PDF file.
23#[derive(Debug, Copy, Clone, PartialEq, Eq)]
24pub enum LoadPdfError {
25    /// An error occurred while processing an encrypted document.
26    Decryption(DecryptionError),
27    /// The PDF was invalid or could not be parsed due to some other unknown reason.
28    Invalid,
29}
30
31#[allow(clippy::len_without_is_empty)]
32impl Pdf {
33    /// Try to read the given PDF file.
34    ///
35    /// Returns `Err` if it was unable to read it.
36    pub fn new(data: PdfData) -> Result<Self, LoadPdfError> {
37        Self::new_with_password(data, "")
38    }
39
40    /// Try to read the given PDF file with a password.
41    ///
42    /// Returns `Err` if it was unable to read it or if the password is incorrect.
43    pub fn new_with_password(data: PdfData, password: &str) -> Result<Self, LoadPdfError> {
44        let password = password.as_bytes();
45        let version = find_version(data.as_ref().as_ref()).unwrap_or(PdfVersion::Pdf10);
46        let xref = match root_xref(data.clone(), password) {
47            Ok(x) => x,
48            Err(e) => match e {
49                XRefError::Unknown => {
50                    fallback(data.clone(), password).ok_or(LoadPdfError::Invalid)?
51                }
52                XRefError::Encryption(e) => return Err(LoadPdfError::Decryption(e)),
53            },
54        };
55        let xref = Arc::new(xref);
56
57        let pages = CachedPages::new(xref.clone()).ok_or(LoadPdfError::Invalid)?;
58
59        Ok(Self {
60            xref,
61            header_version: version,
62            pages,
63            data,
64        })
65    }
66
67    /// Return the number of objects present in the PDF file.
68    pub fn len(&self) -> usize {
69        self.xref.len()
70    }
71
72    /// Return an iterator over all objects defined in the PDF file.
73    pub fn objects(&self) -> impl IntoIterator<Item = Object<'_>> {
74        self.xref.objects()
75    }
76
77    /// Return the version of the PDF file.
78    pub fn version(&self) -> PdfVersion {
79        self.xref
80            .trailer_data()
81            .version
82            .unwrap_or(self.header_version)
83    }
84
85    /// Return the underlying data of the PDF file.
86    pub fn data(&self) -> &PdfData {
87        &self.data
88    }
89
90    /// Return the pages of the PDF file.
91    pub fn pages(&self) -> &Pages<'_> {
92        self.pages.get()
93    }
94
95    /// Return the xref of the PDF file.
96    pub fn xref(&self) -> &XRef {
97        &self.xref
98    }
99
100    /// Return the metadata in the document information dictionary of the document.
101    pub fn metadata(&self) -> &Metadata {
102        self.xref.metadata()
103    }
104}
105
106fn find_version(data: &[u8]) -> Option<PdfVersion> {
107    let data = &data[..data.len().min(2000)];
108    let mut r = Reader::new(data);
109
110    while r.forward_tag(b"%PDF-").is_none() {
111        r.read_byte()?;
112    }
113
114    PdfVersion::from_bytes(r.tail()?)
115}
116
117/// The version of a PDF document.
118#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
119pub enum PdfVersion {
120    /// PDF 1.0.
121    Pdf10,
122    /// PDF 1.1.
123    Pdf11,
124    /// PDF 1.2.
125    Pdf12,
126    /// PDF 1.3.
127    Pdf13,
128    /// PDF 1.4.
129    Pdf14,
130    /// PDF 1.5.
131    Pdf15,
132    /// PDF 1.6.
133    Pdf16,
134    /// PDF 1.7.
135    Pdf17,
136    /// PDF 2.0.
137    Pdf20,
138}
139
140impl PdfVersion {
141    pub(crate) fn from_bytes(bytes: &[u8]) -> Option<Self> {
142        match bytes.get(..3)? {
143            b"1.0" => Some(Self::Pdf10),
144            b"1.1" => Some(Self::Pdf11),
145            b"1.2" => Some(Self::Pdf12),
146            b"1.3" => Some(Self::Pdf13),
147            b"1.4" => Some(Self::Pdf14),
148            b"1.5" => Some(Self::Pdf15),
149            b"1.6" => Some(Self::Pdf16),
150            b"1.7" => Some(Self::Pdf17),
151            b"2.0" => Some(Self::Pdf20),
152            _ => None,
153        }
154    }
155}
156
157#[cfg(test)]
158mod tests {
159    use crate::pdf::{Pdf, PdfVersion};
160    use std::sync::Arc;
161
162    #[test]
163    fn issue_49() {
164        let data = Arc::new([]);
165        let _ = Pdf::new(data);
166    }
167
168    #[test]
169    fn pdf_version_header() {
170        let data = std::fs::read("../hayro-tests/downloads/pdfjs/alphatrans.pdf").unwrap();
171        let pdf = Pdf::new(Arc::new(data)).unwrap();
172
173        assert_eq!(pdf.version(), PdfVersion::Pdf17);
174    }
175
176    #[test]
177    fn pdf_version_catalog() {
178        let data = std::fs::read("../hayro-tests/downloads/pdfbox/2163.pdf").unwrap();
179        let pdf = Pdf::new(Arc::new(data)).unwrap();
180
181        assert_eq!(pdf.version(), PdfVersion::Pdf14);
182    }
183}
hayro_syntax/pdf.rs

hayro_syntax/
pdf.rs
