Introduce the Buf type

benches/oneshot.rs

@@ -47,7 +47,7 @@ fn bench_content() -> Vec<u8> {
     c.save_state();
     c.set_flatness(10);
     c.restore_state();
-    c.finish()
+    c.finish().into_bytes()
 }
 
 fn bench_new() -> Pdf {

examples/limits.rs

@@ -0,0 +1,53 @@
+//! This example shows how you can track PDF limits of your chunks.
+
+use pdf_writer::{Chunk, Content, Limits, Name, Ref};
+
+fn main() {
+    let mut limits = Limits::new();
+
+    let mut content = Content::new();
+    content.transform([-3.4, 0.0, 0.0, 3.1, 100.0, 100.0]);
+    content.line_to(15.0, -26.1);
+    let buf = content.finish();
+    // This will have the limits:
+    // - Max real number: 26.1 (for negative values we use their absolute value)
+    // - Max int number 100 (even though above 100.0 is a float number, it will be coerced into an
+    //   integer, and thus counts towards the int limit)
+    limits.merge(buf.limits());
+
+    let mut chunk = Chunk::new();
+    chunk.stream(Ref::new(1), &buf.into_bytes());
+    chunk.type3_font(Ref::new(2)).name(Name(b"A_long_font_name"));
+    // This will update the limit for the maximum name and dictionary length.
+    limits.merge(chunk.limits());
+
+    // This is what the final PDF will look like.
+    assert_eq!(
+        chunk.as_bytes(),
+        b"1 0 obj
+<<
+  /Length 34
+>>
+stream
+-3.4 0 0 3.1 100 100 cm
+15 -26.1 l
+endstream
+endobj
+
+2 0 obj
+<<
+  /Type /Font
+  /Subtype /Type3
+  /Name /A_long_font_name
+>>
+endobj
+
+"
+    );
+
+    // And the limits should match, as well!
+    assert_eq!(limits.int(), 100);
+    assert_eq!(limits.real(), 26.1);
+    assert_eq!(limits.name_len(), 16);
+    assert_eq!(limits.dict_entries(), 3);
+}

src/buf.rs

@@ -1,29 +1,132 @@
 use super::Primitive;
 
-/// Additional methods for byte buffers.
-pub trait BufExt {
-    fn push_val<T: Primitive>(&mut self, value: T);
-    fn push_int(&mut self, value: i32);
-    fn push_float(&mut self, value: f32);
-    fn push_decimal(&mut self, value: f32);
-    fn push_hex(&mut self, value: u8);
-    fn push_hex_u16(&mut self, value: u16);
-    fn push_octal(&mut self, value: u8);
+use std::ops::Deref;
+
+/// Tracks the limits of data types used in a buffer.
+#[derive(Clone, PartialEq, Debug, Default)]
+pub struct Limits {
+    int: i32,
+    real: f32,
+    name_len: usize,
+    str_len: usize,
+    array_len: usize,
+    dict_entries: usize,
+}
+
+impl Limits {
+    /// Create a new `Limits` struct with all values initialized to zero.
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    /// Get the absolute value of the largest positive/negative integer number.
+    pub fn int(&self) -> i32 {
+        self.int
+    }
+
+    /// Get the absolute value of the largest positive/negative real number.
+    pub fn real(&self) -> f32 {
+        self.real
+    }
+
+    /// Get the maximum length of any used name.
+    pub fn name_len(&self) -> usize {
+        self.name_len
+    }
+
+    /// Get the maximum length of any used array.
+    pub fn array_len(&self) -> usize {
+        self.array_len
+    }
+
+    /// Get the maximum number of entries in any dictionary.
+    pub fn dict_entries(&self) -> usize {
+        self.dict_entries
+    }
+
+    /// Get the maximum length of any used string.
+    pub fn str_len(&self) -> usize {
+        self.str_len
+    }
+
+    pub(crate) fn register_int(&mut self, val: i32) {
+        self.int = self.int.max(val.abs());
+    }
+
+    pub(crate) fn register_real(&mut self, val: f32) {
+        self.real = self.real.max(val.abs());
+    }
+
+    pub(crate) fn register_name_len(&mut self, len: usize) {
+        self.name_len = self.name_len.max(len);
+    }
+
+    pub(crate) fn register_str_len(&mut self, len: usize) {
+        self.str_len = self.str_len.max(len);
+    }
+
+    pub(crate) fn register_array_len(&mut self, len: usize) {
+        self.array_len = self.array_len.max(len);
+    }
+
+    pub(crate) fn register_dict_entries(&mut self, len: usize) {
+        self.dict_entries = self.dict_entries.max(len);
+    }
+
+    /// Merge two `Limits` with each other, taking the maximum
+    /// of each field from both.
+    pub fn merge(&mut self, other: &Limits) {
+        self.register_int(other.int);
+        self.register_real(other.real);
+        self.register_name_len(other.name_len);
+        self.register_str_len(other.str_len);
+        self.register_array_len(other.array_len);
+        self.register_dict_entries(other.dict_entries);
+    }
+}
+
+/// A buffer of arbitrary PDF content.
+#[derive(Clone, PartialEq, Debug)]
+pub struct Buf {
+    pub(crate) inner: Vec<u8>,
+    pub(crate) limits: Limits,
 }
 
-impl BufExt for Vec<u8> {
+impl Buf {
+    pub(crate) fn new() -> Self {
+        Self { inner: Vec::new(), limits: Limits::new() }
+    }
+
+    pub(crate) fn with_capacity(capacity: usize) -> Self {
+        Self {
+            inner: Vec::with_capacity(capacity),
+            limits: Limits::new(),
+        }
+    }
+
+    /// Get the underlying bytes of the buffer.
+    pub fn into_bytes(self) -> Vec<u8> {
+        self.inner
+    }
+
+    /// Return the limits of the buffer.
+    pub fn limits(&self) -> &Limits {
+        &self.limits
+    }
+
     #[inline]
-    fn push_val<T: Primitive>(&mut self, value: T) {
+    pub(crate) fn push_val<T: Primitive>(&mut self, value: T) {
         value.write(self);
     }
 
     #[inline]
-    fn push_int(&mut self, value: i32) {
-        self.extend(itoa::Buffer::new().format(value).as_bytes());
+    pub(crate) fn push_int(&mut self, value: i32) {
+        self.limits.register_int(value);
+        self.extend_slice(itoa::Buffer::new().format(value).as_bytes());
     }
 
     #[inline]
-    fn push_float(&mut self, value: f32) {
+    pub(crate) fn push_float(&mut self, value: f32) {
         // Don't write the decimal point if we don't need it.
         // Also, integer formatting is way faster.
         if value as i32 as f32 == value {
@@ -35,22 +138,40 @@ impl BufExt for Vec<u8> {
 
     /// Like `push_float`, but forces the decimal point.
     #[inline]
-    fn push_decimal(&mut self, value: f32) {
+    pub(crate) fn push_decimal(&mut self, value: f32) {
+        self.limits.register_real(value);
+
         if value == 0.0 || (value.abs() > 1e-6 && value.abs() < 1e12) {
-            self.extend(ryu::Buffer::new().format(value).as_bytes());
+            self.extend_slice(ryu::Buffer::new().format(value).as_bytes());
         } else {
             #[inline(never)]
-            fn write_extreme(buf: &mut Vec<u8>, value: f32) {
+            fn write_extreme(buf: &mut Buf, value: f32) {
                 use std::io::Write;
-                write!(buf, "{}", value).unwrap();
+                write!(buf.inner, "{}", value).unwrap();
             }
 
             write_extreme(self, value);
         }
     }
 
     #[inline]
-    fn push_hex(&mut self, value: u8) {
+    pub(crate) fn extend_slice(&mut self, other: &[u8]) {
+        self.inner.extend(other);
+    }
+
+    #[inline]
+    pub(crate) fn extend(&mut self, other: &Buf) {
+        self.limits.merge(&other.limits);
+        self.inner.extend(&other.inner);
+    }
+
+    #[inline]
+    pub(crate) fn push(&mut self, b: u8) {
+        self.inner.push(b);
+    }
+
+    #[inline]
+    pub(crate) fn push_hex(&mut self, value: u8) {
         fn hex(b: u8) -> u8 {
             if b < 10 {
                 b'0' + b
@@ -64,13 +185,13 @@ impl BufExt for Vec<u8> {
     }
 
     #[inline]
-    fn push_hex_u16(&mut self, value: u16) {
+    pub(crate) fn push_hex_u16(&mut self, value: u16) {
         self.push_hex((value >> 8) as u8);
         self.push_hex(value as u8);
     }
 
     #[inline]
-    fn push_octal(&mut self, value: u8) {
+    pub(crate) fn push_octal(&mut self, value: u8) {
         fn octal(b: u8) -> u8 {
             b'0' + b
         }
@@ -79,4 +200,87 @@ impl BufExt for Vec<u8> {
         self.push(octal((value >> 3) & 7));
         self.push(octal(value & 7));
     }
+
+    #[inline]
+    pub(crate) fn reserve(&mut self, additional: usize) {
+        self.inner.reserve(additional)
+    }
+}
+
+impl Deref for Buf {
+    type Target = [u8];
+
+    fn deref(&self) -> &Self::Target {
+        &self.inner
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::{Chunk, Content, Finish, Name, Rect, Ref, Str, TextStr};
+
+    #[test]
+    fn test_content_limits() {
+        let mut limits = Limits::default();
+
+        let mut content = Content::new();
+        content.cubic_to(14.3, 16.2, 22.6, 30.9, 50.1, 40.0);
+        content.show(Str(b"Some text"));
+        content.set_font(Name(b"NotoSans"), 10.0);
+        let buf = content.finish();
+        limits.merge(buf.limits());
+
+        let mut content = Content::new();
+        content.line_to(55.0, -75.3);
+        content.set_font(Name(b"Noto"), 10.0);
+        content
+            .show_positioned()
+            .items()
+            .show(Str(b"A"))
+            .show(Str(b"B"))
+            .adjust(32.0);
+        content
+            .marked_content_point_with_properties(Name(b"Hi"))
+            .properties()
+            .actual_text(TextStr("text"));
+        let buf = content.finish();
+        limits.merge(buf.limits());
+
+        assert_eq!(
+            limits,
+            Limits {
+                int: 55,
+                real: 75.3,
+                name_len: 10,
+                str_len: 9,
+                array_len: 3,
+                dict_entries: 1,
+            }
+        )
+    }
+
+    #[test]
+    fn test_chunk_limits() {
+        let mut limits = Limits::default();
+
+        let mut chunk = Chunk::new();
+        let mut x_object = chunk.form_xobject(Ref::new(1), &[]);
+        x_object.bbox(Rect::new(4.0, 6.0, 22.1, 31.0));
+        x_object.finish();
+
+        limits.merge(chunk.limits());
+
+        assert_eq!(
+            limits,
+            Limits {
+                int: 31,
+                real: 22.1,
+                name_len: 7,
+                str_len: 0,
+                array_len: 4,
+                dict_entries: 4,
+            }
+        )
+    }
 }