Crate otf_pdf_syntax 

PDF syntax layer: byte sources, the file header, the lexer and object parser, the cross-reference table, stream filters, decryption and repair.

Every input is untrusted (docs/PLAN.md §2): nothing here panics on any bytes, every resource is bounded by Limits, and all reads go through the injected Source trait rather than touching the filesystem.

Re-exports

pub use crypt::Entropy;

pub use crypt::Permissions;

Modules

crypt

The standard security handler, read side (ISO 32000-2 §7.6).

filters

Stream filters (ISO 32000-2 §7.4).

lexer

Tokenizer for PDF syntax (ISO 32000-2 §7.2), also used for content streams.

parser

Object parser on top of the Lexer.

serialize

Writing objects back as PDF syntax.

xref

Cross-reference sections: classic tables and xref streams (§7.5.4, §7.5.8).

Structs

Diagnostic

One repair, with where it happened.

Diagnostics

A shared, append-only list of diagnostics, capped so a hostile file cannot grow it without bound.

Dict

A dictionary. Keys are kept sorted, so iteration (and anything written from it) is deterministic. A repeated key keeps its last value.

File

A PDF file opened for reading. Objects are parsed on first access and cached; the file itself is never modified.

Header

A located file header.

Limits

Limits applied while reading. Every one is enforced; exceeding one is an error (crate::Error::Limit), never a panic or unbounded work.

Name

A name object, without its leading / and with #xx escapes resolved.

ObjId

An indirect object’s identity: object number and generation.

OpenOptions

How to open a file.

PdfString

A string object’s bytes, and whether it was written in hex.

SourceError

A Source failed to serve a read.

Stream

A stream: its dictionary and its bytes exactly as stored in the file, still encoded and (if the file is encrypted) still encrypted.

Version

A PDF version as written in the header, e.g. 1.7.

Enums

DiagnosticKind

What was repaired or tolerated.

Error

Everything that can go wrong in the syntax layer.

HeaderError

Why no header could be read.

LimitKind

Which limit was exceeded.

Object

Any PDF object.

Strictness

Whether malformed input may be repaired.

Constants

HEADER_SEARCH_WINDOW

How far into a file the header is looked for.

Traits

Budget

A CPU budget hook: work is charged in abstract units (roughly, bytes or objects processed), and the hook stops processing by returning false.

Sink

Where a saved document is written. Like crate::Source, it keeps the core crates free of ambient I/O: the host decides whether bytes go to memory, a file or a socket.

Source

Random-access bytes a document is read from.

Functions

find_header

Finds the %PDF-M.m header in the first HEADER_SEARCH_WINDOW bytes.

Type Aliases

Result

The syntax layer’s result type.

SharedSource

A source that can be shared across threads.