lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-09-20 20:07:57 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

continue migrating code to tokenizer

Browse Source
This commit is contained in:
Eliot Jones
2018-01-20 18:42:29 +00:00
parent 3d2a66cbf9
commit 7d90f4858a
30 changed files with 542 additions and 368 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
src/UglyToad.PdfPig/Tokenization/Scanner/IObjectLocationProvider.cs
View File

@@ -1,40 +1,65 @@
namespace UglyToad.PdfPig.Tokenization.Scanner
{
using System;
using System.Collections.Generic;
using ContentStream;
using Cos;
using Parser.Parts;
using Tokens;
internal interface IObjectLocationProvider
{
bool TryGetOffset(IndirectReference reference, out long offset);
void UpdateOffset(IndirectReference reference, long offset);
bool TryGetCached(IndirectReference reference, out ObjectToken objectToken);
void Cache(ObjectToken objectToken);
}
internal class ObjectLocationProvider : IObjectLocationProvider
{
private readonly CrossReferenceTable crossReferenceTable;
private readonly Dictionary<IndirectReference, ObjectToken> cache = new Dictionary<IndirectReference, ObjectToken>();
/// <summary>
/// Since we want to scan objects while reading the cross reference table we lazily load it when it's ready.
/// </summary>
private readonly Func<CrossReferenceTable> crossReferenceTable;
private readonly CosObjectPool pool;
private readonly BruteForceSearcher searcher;
/// <summary>
/// Indicates whether we now have a cross reference table.
/// </summary>
private bool loadedFromTable;
private readonly Dictionary<IndirectReference, long> offsets = new Dictionary<IndirectReference, long>();
public ObjectLocationProvider(CrossReferenceTable crossReferenceTable, CosObjectPool pool, BruteForceSearcher searcher)
public ObjectLocationProvider(Func<CrossReferenceTable> crossReferenceTable, CosObjectPool pool, BruteForceSearcher searcher)
{
this.crossReferenceTable = crossReferenceTable;
foreach (var offset in crossReferenceTable.ObjectOffsets)
{
offsets[offset.Key] = offset.Value;
}
this.pool = pool;
this.searcher = searcher;
}
public bool TryGetOffset(IndirectReference reference, out long offset)
{
if (!loadedFromTable)
{
var table = crossReferenceTable.Invoke();
if (table != null)
{
foreach (var objectOffset in table.ObjectOffsets)
{
offsets[objectOffset.Key] = objectOffset.Value;
}
loadedFromTable = true;
}
}
if (offsets.TryGetValue(reference, out offset))
{
return true;
@@ -54,5 +79,20 @@
{
offsets[reference] = offset;
}
public bool TryGetCached(IndirectReference reference, out ObjectToken objectToken)
{
return cache.TryGetValue(reference, out objectToken);
}
public void Cache(ObjectToken objectToken)
{
if (objectToken == null)
{
throw new ArgumentNullException();
}
cache[objectToken.Number] = objectToken;
}
}
}

117
src/UglyToad.PdfPig/Tokenization/Scanner/PdfTokenScanner.cs
View File

@@ -6,19 +6,22 @@
using System.IO;
using ContentStream;
using Exceptions;
using Filters;
using IO;
using Parser.Parts;
using Tokens;
using Util;
internal interface IPdfObjectScanner : ISeekableTokenScanner
internal interface IPdfTokenScanner : ISeekableTokenScanner
{
ObjectToken Get(IndirectReference reference);
}
internal class PdfTokenScanner : IPdfObjectScanner
internal class PdfTokenScanner : IPdfTokenScanner
{
private readonly IInputBytes inputBytes;
private readonly IObjectLocationProvider objectLocationProvider;
private readonly IFilterProvider filterProvider;
private readonly CoreTokenScanner coreTokenScanner;
/// <summary>
@@ -35,10 +38,11 @@
public long CurrentPosition => coreTokenScanner.CurrentPosition;
public PdfTokenScanner(IInputBytes inputBytes, IObjectLocationProvider objectLocationProvider)
public PdfTokenScanner(IInputBytes inputBytes, IObjectLocationProvider objectLocationProvider, IFilterProvider filterProvider)
{
this.inputBytes = inputBytes;
this.objectLocationProvider = objectLocationProvider;
this.filterProvider = filterProvider;
coreTokenScanner = new CoreTokenScanner(inputBytes);
}
@@ -46,7 +50,7 @@
{
// Read until we find object-number generation obj, e.g. "69 420 obj".
int tokensRead = 0;
while (coreTokenScanner.MoveNext() && coreTokenScanner.CurrentToken != OperatorToken.StartObject)
while (coreTokenScanner.MoveNext() && !Equals(coreTokenScanner.CurrentToken, OperatorToken.StartObject))
{
if (coreTokenScanner.CurrentToken is CommentToken)
{
@@ -79,21 +83,21 @@
}
// Read all tokens between obj and endobj.
while (coreTokenScanner.MoveNext() && coreTokenScanner.CurrentToken != OperatorToken.EndObject)
while (coreTokenScanner.MoveNext() && !Equals(coreTokenScanner.CurrentToken, OperatorToken.EndObject))
{
if (coreTokenScanner.CurrentToken is CommentToken)
{
continue;
}
if (coreTokenScanner.CurrentToken == OperatorToken.StartObject)
if (ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.StartObject))
{
// This should never happen.
Debug.Assert(false, "Encountered a start object 'obj' operator before the end of the previous object.");
return false;
}
if (coreTokenScanner.CurrentToken == OperatorToken.StartStream)
if (ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.StartStream))
{
// Read stream: special case.
if (TryReadStream(coreTokenScanner.CurrentTokenStart, out var stream))
@@ -114,7 +118,7 @@
previousTokenPositions[1] = coreTokenScanner.CurrentPosition;
}
if (coreTokenScanner.CurrentToken != OperatorToken.EndObject)
if (!ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.EndObject))
{
readTokens.Clear();
return false;
@@ -125,7 +129,7 @@
IToken token;
if (readTokens.Count == 3 && readTokens[0] is NumericToken objNum
&& readTokens[1] is NumericToken genNum
&& readTokens[2] == OperatorToken.R)
&& ReferenceEquals(readTokens[2], OperatorToken.R))
{
// I have no idea if this can ever happen.
token = new IndirectReferenceToken(new IndirectReference(objNum.Long, genNum.Int));
@@ -196,10 +200,7 @@
// Track any 'endobj' or 'endstream' operators we see.
var observedEndLocations = new List<PossibleStreamEndLocation>();
// Keep track of the previous byte.
byte previousByte = 0;
// Begin reading the stream.
using (var memoryStream = new MemoryStream())
using (var binaryWrite = new BinaryWriter(memoryStream))
@@ -287,8 +288,7 @@
endObjPosition = 0;
commonPartPosition = 0;
}
previousByte = inputBytes.CurrentByte;
binaryWrite.Write(inputBytes.CurrentByte);
read++;
@@ -451,11 +451,24 @@
public ObjectToken Get(IndirectReference reference)
{
if (objectLocationProvider.TryGetCached(reference, out var objectToken))
{
return objectToken;
}
if (!objectLocationProvider.TryGetOffset(reference, out var offset))
{
throw new InvalidOperationException($"Could not find the object with reference: {reference}.");
}
// Negative offsets refer to a stream with that number.
if (offset < 0)
{
var result = GetObjectFromStream(reference, offset);
return result;
}
Seek(offset);
if (!MoveNext())
@@ -465,5 +478,79 @@
return (ObjectToken)CurrentToken;
}
private ObjectToken GetObjectFromStream(IndirectReference reference, long offset)
{
var streamObjectNumber = offset * -1;
var streamObject = Get(new IndirectReference(streamObjectNumber, 0));
if (!(streamObject.Data is StreamToken stream))
{
throw new PdfDocumentFormatException("Requested a stream object by reference but the requested stream object " +
$"was not a stream: {reference}, {streamObject.Data}.");
}
var objects = ParseObjectStream(stream, offset);
foreach (var o in objects)
{
objectLocationProvider.Cache(o);
}
if (!objectLocationProvider.TryGetCached(reference, out var result))
{
throw new PdfDocumentFormatException($"Could not find the object {reference} in the stream {streamObjectNumber}.");
}
return result;
}
private IReadOnlyList<ObjectToken> ParseObjectStream(StreamToken stream, long offset)
{
if (!stream.StreamDictionary.TryGet(NameToken.N, out var numberToken)
|| !(numberToken is NumericToken numberOfObjects))
{
throw new PdfDocumentFormatException($"Object stream dictionary did not provide number of objects {stream.StreamDictionary}.");
}
if (!stream.StreamDictionary.TryGet(NameToken.First, out var firstToken)
|| !(firstToken is NumericToken))
{
throw new PdfDocumentFormatException($"Object stream dictionary did not provide first object offset {stream.StreamDictionary}.");
}
// Read the N integers
var bytes = new ByteArrayInputBytes(stream.Decode(filterProvider));
var scanner = new CoreTokenScanner(bytes);
var objects = new List<Tuple<long, long>>();
for (var i = 0; i < numberOfObjects.Int; i++)
{
scanner.MoveNext();
var objectNumber = (NumericToken) scanner.CurrentToken;
scanner.MoveNext();
var byteOffset = (NumericToken) scanner.CurrentToken;
objects.Add(Tuple.Create(objectNumber.Long, byteOffset.Long));
}
var results = new List<ObjectToken>();
for (var i = 0; i < objects.Count; i++)
{
var obj = objects[i];
scanner.MoveNext();
var token = scanner.CurrentToken;
results.Add(new ObjectToken(offset, new IndirectReference(obj.Item1, 0), token));
}
return results;
}
}
}