lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-09-23 12:43:38 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

#24 add different field types and code to read them

Browse Source
This commit is contained in:
Eliot Jones
2019-01-02 22:28:50 +00:00
parent ea9ac795ff
commit 21aa964e0b
4 changed files with 567 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
src/UglyToad.PdfPig/AcroForms/AcroForm.cs
View File

@@ -1,6 +1,8 @@
namespace UglyToad.PdfPig.AcroForms
{
using System;
using System.Collections.Generic;
using Fields;
using Tokens;
using Util.JetBrains.Annotations;
@@ -29,14 +31,21 @@
/// </summary>
public bool NeedAppearances { get; }
/// <summary>
/// All root fields in this form with their corresponding references.
/// </summary>
public IReadOnlyDictionary<IndirectReference, AcroFieldBase> Fields { get; }
/// <summary>
/// Create a new <see cref="AcroForm"/>.
/// </summary>
public AcroForm(DictionaryToken dictionary, SignatureFlags signatureFlags, bool needAppearances)
public AcroForm(DictionaryToken dictionary, SignatureFlags signatureFlags, bool needAppearances,
IReadOnlyDictionary<IndirectReference, AcroFieldBase> fields)
{
Dictionary = dictionary ?? throw new ArgumentNullException(nameof(dictionary));
SignatureFlags = signatureFlags;
NeedAppearances = needAppearances;
Fields = fields ?? throw new ArgumentNullException(nameof(fields));
}
/// <inheritdoc />

196
src/UglyToad.PdfPig/AcroForms/AcroFormFactory.cs
View File

@@ -1,8 +1,10 @@
namespace UglyToad.PdfPig.AcroForms
{
using System;
using System.Collections.Generic;
using Content;
using Exceptions;
using Fields;
using Filters;
using Parser.Parts;
using Tokenization.Scanner;
@@ -77,14 +79,206 @@
throw new PdfDocumentFormatException($"Could not retrieve the fields array for an AcroForm: {acroDictionary}.");
}
var fields = new Dictionary<IndirectReference, AcroFieldBase>(fieldsArray.Length);
foreach (var fieldToken in fieldsArray.Data)
{
if (!(fieldToken is IndirectReferenceToken fieldReferenceToken))
{
throw new PdfDocumentFormatException($"The fields array should only contain indirect references, instead got: {fieldToken}.");
}
var fieldDictionary = DirectObjectFinder.Get<DictionaryToken>(fieldToken, tokenScanner);
var field = GetAcroField(fieldDictionary);
fields[fieldReferenceToken.Data] = field;
}
return new AcroForm(acroDictionary, signatureFlags, needAppearances);
return new AcroForm(acroDictionary, signatureFlags, needAppearances, fields);
}
private AcroFieldBase GetAcroField(DictionaryToken fieldDictionary)
{
fieldDictionary.TryGet(NameToken.Ft, out NameToken fieldType);
fieldDictionary.TryGet(NameToken.Ff, out NumericToken fieldFlagsToken);
var kids = new List<DictionaryToken>();
if (fieldDictionary.TryGetOptionalTokenDirect(NameToken.Kids, tokenScanner, out ArrayToken kidsToken))
{
foreach (var kid in kidsToken.Data)
{
if (!(kid is IndirectReferenceToken kidReferenceToken))
{
throw new PdfDocumentFormatException($"AcroForm kids should only contain indirect reference, instead got: {kid}.");
}
var kidObject = tokenScanner.Get(kidReferenceToken.Data);
if (kidObject.Data is DictionaryToken kidDictionaryToken)
{
kids.Add(kidDictionaryToken);
}
else
{
throw new PdfDocumentFormatException($"Unexpected type of kid in AcroForm field. Expected dictionary but got: {kidObject.Data}.");
}
}
}
fieldDictionary.TryGetOptionalStringDirect(NameToken.T, tokenScanner, out var partialFieldName);
fieldDictionary.TryGetOptionalStringDirect(NameToken.Tu, tokenScanner, out var alternateFieldName);
fieldDictionary.TryGetOptionalStringDirect(NameToken.Tm, tokenScanner, out var mappingName);
fieldDictionary.TryGet(NameToken.Parent, out IndirectReferenceToken parentReferenceToken);
var information = new AcroFieldCommonInformation(parentReferenceToken?.Data, partialFieldName, alternateFieldName, mappingName);
var fieldFlags = (uint) (fieldFlagsToken?.Long ?? 0);
AcroFieldBase result;
if (fieldType == null)
{
var children = new List<AcroFieldBase>();
foreach (var kid in kids)
{
var kidField = GetAcroField(kid);
children.Add(kidField);
}
result = new NonTerminalAcroField(fieldDictionary, "Non-Terminal Field", fieldFlags, information, children);
}
else if (fieldType == NameToken.Btn)
{
var buttonFlags = (AcroButtonFieldFlags)fieldFlags;
if (buttonFlags.HasFlag(AcroButtonFieldFlags.Radio))
{
var field = new AcroRadioButtonsField(fieldDictionary, fieldType, buttonFlags, information);
result = field;
}
else if (buttonFlags.HasFlag(AcroButtonFieldFlags.PushButton))
{
var field = new AcroPushButtonField(fieldDictionary, fieldType, buttonFlags, information);
result = field;
}
else
{
if (!fieldDictionary.TryGetOptionalTokenDirect(NameToken.V, tokenScanner, out NameToken valueToken))
{
valueToken = NameToken.Off;
}
var field = new AcroCheckboxField(fieldDictionary, fieldType, buttonFlags, information, valueToken);
result = field;
}
}
else if (fieldType == NameToken.Tx)
{
var textFlags = (AcroTextFieldFlags)fieldFlags;
var textValue = default(string);
if (fieldDictionary.TryGet(NameToken.V, out var textValueToken))
{
if (DirectObjectFinder.TryGet(textValueToken, tokenScanner, out StringToken valueStringToken))
{
textValue = valueStringToken.Data;
}
else if (DirectObjectFinder.TryGet(textValueToken, tokenScanner, out HexToken valueHexToken))
{
textValue = valueHexToken.Data;
}
else if (DirectObjectFinder.TryGet(textValueToken, tokenScanner, out StreamToken valueStreamToken))
{
}
}
var field = new AcroTextField(fieldDictionary, fieldType, textFlags, information, textValue);
result = field;
}
else if (fieldType == NameToken.Ch)
{
var options = new List<AcroChoiceOption>();
if (fieldDictionary.TryGetOptionalTokenDirect(NameToken.Opt, tokenScanner, out ArrayToken optionsArrayToken))
{
for (var i = 0; i < optionsArrayToken.Data.Count; i++)
{
var optionToken = optionsArrayToken.Data[i];
if (DirectObjectFinder.TryGet(optionToken, tokenScanner, out StringToken optionStringToken))
{
options.Add(new AcroChoiceOption(i, optionStringToken.Data));
}
else if (DirectObjectFinder.TryGet(optionToken, tokenScanner, out HexToken optionHexToken))
{
options.Add(new AcroChoiceOption(i, optionHexToken.Data));
}
else if (DirectObjectFinder.TryGet(optionToken, tokenScanner, out ArrayToken optionArrayToken))
{
if (optionArrayToken.Length != 2)
{
throw new PdfDocumentFormatException($"An option array containing array elements should contain 2 strings, instead got: {optionArrayToken}.");
}
if (DirectObjectFinder.TryGet(optionArrayToken.Data[0], tokenScanner, out StringToken exportValueStringToken))
{
}
else if (DirectObjectFinder.TryGet(optionArrayToken.Data[0], tokenScanner, out HexToken exportValueHexToken))
{
}
else
{
throw new PdfDocumentFormatException($"An option array array element's first value should be the export value string, instead got: {optionArrayToken.Data[0]}.");
}
if (DirectObjectFinder.TryGet(optionArrayToken.Data[1], tokenScanner, out StringToken nameStringToken))
{
}
else if (DirectObjectFinder.TryGet(optionArrayToken.Data[1], tokenScanner, out HexToken nameHexToken))
{
}
else
{
throw new PdfDocumentFormatException($"An option array array element's second value should be the option name string, instead got: {optionArrayToken.Data[1]}.");
}
}
else
{
throw new PdfDocumentFormatException($"An option array should contain either strings or 2 element arrays, instead got: {optionToken}.");
}
}
}
var choiceFlags = (AcroChoiceFieldFlags)fieldFlags;
if (choiceFlags.HasFlag(AcroChoiceFieldFlags.Combo))
{
var field = new AcroComboBoxField(fieldDictionary, fieldType, choiceFlags, information);
result = field;
}
else
{
var field = new AcroListBoxField(fieldDictionary, fieldType, choiceFlags, information);
result = field;
}
}
else if (fieldType == NameToken.Sig)
{
var field = new AcroSignatureField(fieldDictionary, fieldType, fieldFlags, information);
result = field;
}
else
{
throw new PdfDocumentFormatException($"Unexpected type for field in AcroForm: {fieldType}.");
}
return result;
}
}
}

344
src/UglyToad.PdfPig/AcroForms/Fields/AcroFieldBase.cs Normal file
View File

@@ -0,0 +1,344 @@
namespace UglyToad.PdfPig.AcroForms.Fields
{
using System;
using System.Collections.Generic;
using Tokens;
using Util.JetBrains.Annotations;
/// <summary>
/// A field in an interactive <see cref="AcroForm"/>.
/// </summary>
internal abstract class AcroFieldBase
{
/// <summary>
/// The raw PDF dictionary for this field.
/// </summary>
public DictionaryToken Dictionary { get; }
/// <summary>
/// The <see cref="string"/> representing the type of this field.
/// </summary>
public string FieldType { get; }
/// <summary>
/// Specifies various characteristics of the field.
/// </summary>
public uint FieldFlags { get; }
public AcroFieldCommonInformation Information { get; }
protected AcroFieldBase(DictionaryToken dictionary, string fieldType, uint fieldFlags, AcroFieldCommonInformation information)
{
Dictionary = dictionary ?? throw new ArgumentNullException(nameof(dictionary));
FieldType = fieldType ?? throw new ArgumentNullException(nameof(fieldType));
FieldFlags = fieldFlags;
Information = information;
}
}
/// <summary>
/// A non-leaf field in the form's structure.
/// </summary>
internal class NonTerminalAcroField : AcroFieldBase
{
public IReadOnlyList<AcroFieldBase> Children { get; }
public NonTerminalAcroField(DictionaryToken dictionary, string fieldType, uint fieldFlags, AcroFieldCommonInformation information,
IReadOnlyList<AcroFieldBase> children) :
base(dictionary, fieldType, fieldFlags, information)
{
Children = children ?? throw new ArgumentNullException(nameof(children));
}
}
internal class AcroRadioButtonsField : AcroFieldBase
{
public AcroButtonFieldFlags Flags { get; }
public AcroRadioButtonsField(DictionaryToken dictionary, string fieldType, AcroButtonFieldFlags fieldFlags,
AcroFieldCommonInformation information) :
base(dictionary, fieldType, (uint)fieldFlags, information)
{
Flags = fieldFlags;
}
}
internal class AcroPushButtonField : AcroFieldBase
{
public AcroButtonFieldFlags Flags { get; }
public AcroPushButtonField(DictionaryToken dictionary, string fieldType, AcroButtonFieldFlags fieldFlags,
AcroFieldCommonInformation information) :
base(dictionary, fieldType, (uint)fieldFlags, information)
{
Flags = fieldFlags;
}
}
internal class AcroCheckboxField : AcroFieldBase
{
public AcroButtonFieldFlags Flags { get; }
public NameToken CurrentValue { get; }
public AcroCheckboxField(DictionaryToken dictionary, string fieldType, AcroButtonFieldFlags fieldFlags,
AcroFieldCommonInformation information, NameToken currentValue) :
base(dictionary, fieldType, (uint)fieldFlags, information)
{
Flags = fieldFlags;
CurrentValue = currentValue;
}
}
internal class AcroTextField : AcroFieldBase
{
public AcroTextFieldFlags Flags { get; }
public AcroTextField(DictionaryToken dictionary, string fieldType, AcroTextFieldFlags fieldFlags,
AcroFieldCommonInformation information, string textValue) :
base(dictionary, fieldType, (uint)fieldFlags, information)
{
Flags = fieldFlags;
}
}
internal class AcroSignatureField : AcroFieldBase
{
public AcroSignatureField(DictionaryToken dictionary, string fieldType, uint fieldFlags, AcroFieldCommonInformation information) :
base(dictionary, fieldType, fieldFlags, information)
{
}
}
internal class AcroListBoxField : AcroFieldBase
{
public AcroChoiceFieldFlags Flags { get; }
public AcroListBoxField(DictionaryToken dictionary, string fieldType, AcroChoiceFieldFlags fieldFlags,
AcroFieldCommonInformation information) :
base(dictionary, fieldType, (uint)fieldFlags, information)
{
Flags = fieldFlags;
}
}
internal class AcroComboBoxField : AcroFieldBase
{
public AcroChoiceFieldFlags Flags { get; }
public AcroComboBoxField(DictionaryToken dictionary, string fieldType, AcroChoiceFieldFlags fieldFlags,
AcroFieldCommonInformation information) :
base(dictionary, fieldType, (uint)fieldFlags, information)
{
Flags = fieldFlags;
}
}
/// <summary>
/// An option in a choice field, either <see cref="AcroComboBoxField"/> or <see cref="AcroListBoxField"/>.
/// </summary>
internal class AcroChoiceOption
{
/// <summary>
/// The index of this option in the array.
/// </summary>
public int Index { get; }
/// <summary>
/// The text of the option.
/// </summary>
public string Name { get; }
/// <summary>
/// The value of the option when the form is exported.
/// </summary>
public string ExportValue { get; }
/// <summary>
/// Create a new <see cref="AcroChoiceOption"/>.
/// </summary>
public AcroChoiceOption(int index, string name, string exportValue = null)
{
Index = index;
Name = name;
ExportValue = exportValue;
}
}
/// <summary>
/// Information from the field dictionary which is common across all field types.
/// </summary>
internal class AcroFieldCommonInformation
{
/// <summary>
/// The reference to the field which is the parent of this one, if applicable.
/// </summary>
public IndirectReference? Parent { get; set; }
/// <summary>
/// The partial field name for this field. The fully qualified field name is the
/// period '.' joined name of all parents' partial names and this field's partial name.
/// </summary>
[CanBeNull]
public string PartialName { get; }
/// <summary>
/// The alternate field name to be used instead of the fully qualified field name where
/// the field is being identified on the user interface or by screen readers.
/// </summary>
[CanBeNull]
public string AlternateName { get; }
/// <summary>
/// The mapping name used when exporting form field data from the document.
/// </summary>
[CanBeNull]
public string MappingName { get; }
/// <summary>
/// Create a new <see cref="AcroFieldCommonInformation"/>.
/// </summary>
public AcroFieldCommonInformation(IndirectReference? parent, string partialName, string alternateName, string mappingName)
{
Parent = parent;
PartialName = partialName;
AlternateName = alternateName;
MappingName = mappingName;
}
/// <inheritdoc />
public override string ToString()
{
return $"Parent: {Parent}. Partial: {PartialName}. Alternate: {AlternateName}. Mapping: {MappingName}.";
}
}
/// <summary>
/// Flags specifying various characteristics of a button type field in an <see cref="AcroFieldBase"/>.
/// </summary>
[Flags]
internal enum AcroButtonFieldFlags : uint
{
/// <summary>
/// The user may not change the value of the field.
/// </summary>
ReadOnly = 1 << 0,
/// <summary>
/// The field must have a value before the form can be submitted.
/// </summary>
Required = 1 << 1,
/// <summary>
/// Must not be exported by the submit form action.
/// </summary>
NoExport = 1 << 2,
/// <summary>
/// For radio buttons, one radio button must be set at all times.
/// </summary>
NoToggleToOff = 1 << 14,
/// <summary>
/// The field is a set of radio buttons.
/// </summary>
Radio = 1 << 15,
/// <summary>
/// The field is a push button.
/// </summary>
PushButton = 1 << 16,
/// <summary>
/// For radio buttons a group of radio buttons will toggle on/off at the same time based on their initial value.
/// </summary>
RadiosInUnison = 1 << 25
}
/// <summary>
/// Flags specifying various characteristics of a text type field in an <see cref="AcroFieldBase"/>.
/// </summary>
[Flags]
internal enum AcroTextFieldFlags : uint
{
/// <summary>
/// The user may not change the value of the field.
/// </summary>
ReadOnly = 1 << 0,
/// <summary>
/// The field must have a value before the form can be submitted.
/// </summary>
Required = 1 << 1,
/// <summary>
/// Must not be exported by the submit form action.
/// </summary>
NoExport = 1 << 2,
/// <summary>
/// The field can contain multiple lines of text.
/// </summary>
Multiline = 1 << 12,
/// <summary>
/// The field is for a password and should not be displayed as text and should not be stored to file.
/// </summary>
Password = 1 << 13,
/// <summary>
/// The field represents a file path selection.
/// </summary>
FileSelect = 1 << 20,
/// <summary>
/// The text entered is not spell checked.
/// </summary>
DoNotSpellCheck = 1 << 22,
/// <summary>
/// The field does not scroll if the text exceeds the bounds of the field.
/// </summary>
DoNotScroll = 1 << 23,
/// <summary>
/// For a text field which is not a <see cref="Password"/>, <see cref="Multiline"/> or <see cref="FileSelect"/>
/// the field text is evenly spaced by splitting into 'combs' based on the MaxLen entry in the field dictionary.
/// </summary>
Comb = 1 << 24,
/// <summary>
/// The value of the field is a rich text string.
/// </summary>
RichText = 1 << 25
}
/// <summary>
/// Flags specifying various characteristics of a choice type field in an <see cref="AcroFieldBase"/>.
/// </summary>
[Flags]
internal enum AcroChoiceFieldFlags : uint
{
/// <summary>
/// The user may not change the value of the field.
/// </summary>
ReadOnly = 1 << 0,
/// <summary>
/// The field must have a value before the form can be submitted.
/// </summary>
Required = 1 << 1,
/// <summary>
/// Must not be exported by the submit form action.
/// </summary>
NoExport = 1 << 2,
/// <summary>
/// The field is a combo box.
/// </summary>
Combo = 1 << 17,
/// <summary>
/// The combo box includes an editable text box. <see cref="Combo"/> must be set.
/// </summary>
Edit = 1 << 18,
/// <summary>
/// The options should be sorted alphabetically, this should be ignored by viewer applications.
/// </summary>
Sort = 1 << 19,
/// <summary>
/// The field allows multiple options to be selected.
/// </summary>
MultiSelect = 1 << 21,
/// <summary>
/// The text entered in the field is not spell checked. <see cref="Combo"/> and <see cref="Edit"/> must be set.
/// </summary>
DoNotSpellCheck = 1 << 22,
/// <summary>
/// Any associated field action is fired when the selection is changed rather than on losing focus.
/// </summary>
CommitOnSelectionChange = 1 << 26
}
}

18
src/UglyToad.PdfPig/Util/DictionaryTokenExtensions.cs
View File

@@ -82,6 +82,24 @@
return false;
}
public static bool TryGetOptionalStringDirect(this DictionaryToken token, NameToken name, IPdfTokenScanner scanner, out string result)
{
result = default(string);
if (token.TryGetOptionalTokenDirect(name, scanner, out StringToken stringToken))
{
result = stringToken.Data;
return true;
}
if (token.TryGetOptionalTokenDirect(name, scanner, out HexToken hexToken))
{
result = hexToken.Data;
return true;
}
return false;
}
}
internal static class ArrayTokenExtensions