Oberon/Text

Confusingly for a novice, three forms of line ending can be found in text files in contemporary computer systems.

In Oberon and some other systems (most prominently Classic MacOS), a line of text ends with the carriage return character, denoted CR.
In Unix-like and some other systems, a line of text ends with the line feed character denoted LF.
In DOS, Microsoft Windows and some other systems, a line of text ends with the two characters, CR and LF.

This multiplicity poses no difficulty in Oberon; most Oberon systems allow convenient editing with any of these line endings.^[1]

Also confusing for a novice, an Oberon Text^[2] can appear at the first sight to be no more than a plain text comprising a sequence of ASCII characters. Nevertheless Text in Oberon is a type defined in the Text module. A Text of this type is a sequence of characters including non-printing characters. A character in a Text can have attributes including typeface, size and color. Furthermore a Text can include non-character objects; an image or a hyperlink for example. Consequently two displayed documents, one an Oberon Text and the other an HTML text, can have the same appearance and the same behavior of links.

Editing a Text

Each Oberon system has an Edit module. If Edit.Open Example.Text appears almost anywhere on the screen, a MM on Edit.Open will open a viewer. If a file named Example.Text exists, it's content will appear in the viewer. If no file with the name exists, the viewer will be empty. The caret can be set by ML and characters can be inserted using a keyboard. In addition to Edit, ETH Oberon has ET and the Gadgets subsystem. Commands ET.Open <aFile> and Desktops.OpenDoc <aFile> are available. ETH Oberon also has the Hex module providing a hexadecimal editor with command Hex.Open <aFile>.

Programmatical Access to a Text

A Text can be manipulated using procedures in the Texts module; for ETH Oberon and for V5. Also summarized for ETH Oberon in DEFINITION Texts.

To read a Text programmatically, a RECORD termed a Reader is opened on the Text at a specified offset. (A Reader is a RECORD; not a procedure.) With reader open, the Read(reader, ch) procedure retrieves a character each time it is executed as in this fragment from ETH Oberon.

 VAR
   T: Texts.Text;
   R: Texts.Reader;
   ch: CHAR;
 BEGIN
   NEW(T);
   Texts.Open(T, “Texts.Mod“);
   Texts.OpenReader(R, T, 0);
   Texts.Read(R, ch);
   WHILE ~R.eot DO
     Out.Char(ch);
     Texts.Read(R, ch)
   END
 END

The font of a character is referenced in the Objects.Library of the Reader. The Reader also has fields, col, the color of the character and voff, vertical offset of the character. After an execution of Read(reader, ch), the freshly read character is available in ch and attributes are in the components of reader. This assumes that the object read was a character as evidenced by (reader.lib IS Fonts.Font) being TRUE. Each execution of Read() advances reader through the Text by one character until the end is reached and reader.eot becomes TRUE.

How Text Works ...

These tables show the structure of records in memory representing a Text. When Texts.Store() records a Text in a file for a storage medium, the information in the record structure is serialized. In the inverse process, Texts.Load() deserializes the information to the record structure of the Text in memory.

Click on a hyperlink to see the module where a type is defined.

In V2

The Type Text
Type	Fields of TextDesc	Types of Fields	Notes
Texts.Text, a pointer to a Texts.TextDesc	len	LONGINT	Length of text, in bytes.
	changed	BOOLEAN	Flag indicating a revision.
	notify	Texts.Notifier	Pointer to a method to notify interested clients of state changes.
	trailer	Texts.Piece	Pointer to the Sentinel node in the list of pieces.
	pce	Texts.Piece	Last found piece.
	org	LONGINT	Offset in [0,len) of first character in last found piece.

Texts.Piece
Type	Fields of PieceDesc	Types of Fields	Notes
Texts.Piece, a pointer to a Texts.PieceDesc	f	Files.File	Pointer, to file.
	off	LONGINT	Integer offset in Text of first character in Piece.
	len	LONGINT	Number of bytes in Piece.
	fnt	Fonts.Font	Pointer to a font.
	col	INTEGER
	voff	INTEGER	Vertical offset of characters in pixels.
	prev	Texts.Piece	Pointer to previous piece of Text.
	next	Texts.Piece	Pointer to next piece of Text.

In ETH Oberon

TextDesc is an extension of Objects.ObjDesc. The fields added to make TextDesc are distinguished from the fields inherited from ObjDesc by a differing background color.

The Type Text
Type	Fields of TextDesc	Types of Fields	Notes
Texts.Text, a pointer to a Texts.TextDesc	stamp	LONGINT	Integer
	dlink	Objects.Object	Pointer
	slink	Objects.Object	Pointer
	lib	Objects.Library	Pointer
	ref	INTEGER
	handle	Objects.Handler	Pointer
	len	LONGINT	Length of text.
	obs	Objects.Library	Pointer
	trailer	Texts.Piece	Pointer to Sentinel node in list of pieces.
	org	LONGINT	Offset in [0,len) of first character in last found piece.
	pce	Texts.Piece	Last found piece.

Objects.Library in the table above
Type	Fields of LibDesc	Types of Fields	Notes
Objects.Library, a pointer to a LibDesc	next	Objects.Library	Pointer
	ind	Objects.Index	Pointer
	f	Files.File	Pointer
	R	Files.Rider	Pointer
	name	Objects.Name
	dict	Objects.Dictionary	Pointer
	maxref	INTEGER
	GName	POINTER

Texts.Piece
Type	Fields of PieceDesc	Types of Fields	Notes
Texts.Piece, a pointer to a Texts.PieceDesc	f	Files.File	Pointer
	off	LONGINT	Integer
	len	LONGINT
	obj	Objects.Object	Pointer
	lib	Objects.Library	Pointer
	ref	INTEGER
	col	SHORTINT
	voff	SHORTINT
	prev	Piece	Pointer
	next	Piece	Pointer

In the Oberon Subsystem of A2

TextDesc is an extension of Objects.ObjDesc. The fields added to make TextDesc are distinguished from the fields inherited from ObjDesc by differing background colors.

The Type Text
Type	Fields of TextDesc	Types of Fields	Notes
Texts.Text, a pointer to a Texts.TextDesc	stamp	LONGINT	Integer
	dlink	Objects.Object	Pointer
	slink	Objects.Object	Pointer
	lib	Objects.Library	Pointer
	ref	INTEGER
	handle	Objects.Handler	Pointer
	len	LONGINT	Length of text.
	obs	Objects.Library	Pointer
	trailer	Texts.Piece	Pointer to Sentinel node in list of pieces.
	org	LONGINT	Offset in [0,len) of first character in last found piece.
	pce	Texts.Piece	Last found piece.

Objects.Library in the table above
Type	Fields of LibDesc	Types of Fields	Notes
Objects.Library, a pointer to a LibDesc	next	Objects.Library	Pointer
	ind	Objects.Index	Pointer
	f	Files.File	Pointer
	R	Files.Rider	Pointer
	name	Objects.Name
	dict	Objects.Dictionary	Pointer
	maxref	INTEGER
	GName	POINTER

Texts.Piece
Type	Fields of PieceDesc	Types of Fields	Notes
Texts.Piece, a pointer to a Texts.PieceDesc	f	Files.File	Pointer
	off	LONGINT	Integer
	len	LONGINT
	obj	Objects.Object	Pointer
	lib	Objects.Library	Pointer
	ref	INTEGER
	col	SHORTINT
	voff	SHORTINT
	prev	Piece	Pointer
	next	Piece	Pointer

In V5

The Type Text
Type	Fields of TextDesc	Types of Fields	Notes
Texts.Text, a pointer to a Texts.TextDesc	len	INTEGER^[3]	Length of text, in bytes.
	changed	BOOLEAN	Flag indicating a revision.
	notify	Texts.Notifier	Pointer to a method to notify interested clients of state changes.
	trailer	Texts.Piece	Pointer to the Sentinel node in the list of pieces.
	pce	Texts.Piece	Last found piece.
	org	INTEGER	Offset in [0,len) of first character in last found piece.

Texts.Piece
Type	Fields of PieceDesc	Types of Fields	Notes
Texts.Piece, a pointer to a Texts.PieceDesc	f	Files.File	Pointer, to file.
	off	INTEGER	Integer offset in Text of first character in Piece.
	len	INTEGER	Number of bytes in Piece.
	fnt	Fonts.Font	Pointer to a font.
	col	INTEGER
	voff	INTEGER	Vertical offset of characters in pixels.
	prev	Texts.Piece	Pointer to previous piece of Text.
	next	Texts.Piece	Pointer to next piece of Text.

Texts.FindPiece and the cache

For a given Text, T, and offset pos in [0, T.len), procedure Texts.FindPiece has the task of locating the piece containing pos. At each execution, FindPiece could begin at offset 0 and add lengths of pieces until the piece containing pos is located. A cache based upon T.pce and T.org allows better efficiency. When FindPiece completes a search, the pointer to the found piece is recorded in T.pce; the offset of the first character of that piece is recorded in T.org. The next execution of FindPiece begins at that cached location. With a result from FindPiece often being near the preceeding result, this strategy avoids repeated summation of lengths from the beginning of the first piece.

References

↑ Each case is easily displayed explicitly. In an Oberon system with the hexadecimal editor, Hex, MM on Hex.Open <aFile>. In ETH Oberon a file with lines ending CR LF is edited by ET.OpenAscii <aFile>.
↑ With Text being a fundamental type in an Oberon system, "Oberon Text" is considered a proper name. Hence the capitalization "Text".
↑ In V5 the only integer type is INTEGER. Cf. LONGINT in V2.

[1] Each case is easily displayed explicitly. In an Oberon system with the hexadecimal editor, Hex, MM on Hex.Open <aFile>. In ETH Oberon a file with lines ending CR LF is edited by ET.OpenAscii <aFile>.

[2] With Text being a fundamental type in an Oberon system, "Oberon Text" is considered a proper name. Hence the capitalization "Text".

[3] In V5 the only integer type is INTEGER. Cf. LONGINT in V2.

[1]

[2]

[3]