7
autocad reference manual
$dimtol 70 dimension tolerances generated if nonzero.
$dimtp 40 plus tolerance.
$dimtsz 40 dimensioning tick size: 0=no ticks.
$dimtvp 40 text vertical position.
$dimtxt 40 dimensioning text height.
$dimzin 70 zero suppression for "feet & inch" dimensions.
$dragmode 70 0=off, 1=on, 2=auto.
$elevation 40 current elevation set by elev command.
$extmax 10,20,30 xy drawing extents upper right corner (in wcs).
$extmin 10,20,30 xy drawing extents lower left corner (in wcs).
$filletrad 40 fillet radius.
$fillmode 70 fill mode on if nonzero.
$flatland 70 force compatibility with older versions if nonzero.
$handling 70 handles enabled if nonzero.
$handseed 5 next available handle.
$insbase 10,20,30 insertion base set by base command (in wcs).
$limcheck 70 nonzero if limits checking is on.
$limmax 10,20 xy drawing limits upper right corner (in wcs).
$limmin 10,20 xy drawing limits lower left corner (in wcs).
$ltscale 40 global linetype scale.
$lunits 70 units format for coordinates and distances.
$luprec 70 units precision for coordinates and distances.
$menu 1 name of menu file.
$mirrtext 70 mirror text if nonzero.
$orthomode 70 ortho mode on if nonzero.
$osmode 70 running object snap modes.
$pdmode 70 point display mode.
$pdsize 40 point display size.
$plinewid 40 default polyline width.
$qtextmode 70 quick text mode on if nonzero.
$regenmode 70 regenauto mode on if nonzero.
$sketchinc 40 sketch record increment.
$skpoly 70 0=sketch lines, 1=sketch polylines.
$splframe 70 spline control polygon display, 1 = on, 0 = off.
$splinesegs 70 number of line segments per spline patch.
$splinetype 70 spline curve type for "pedit spline"
(see appendix a).
$surftab1 70 number of mesh tabulations in first direction.
$surftab2 70 number of mesh tabulations in second direction.
$surftype 70 surface type for "pedit smooth" (see appendix a).
$surfu 70 surface density (for "pedit smooth") in m direction.
$surfv 70 surface density (for "pedit smooth") in n direction.
$tdcreate 40 date/time of drawing creation.
$tdindwg 40 cumulative editing time for this drawing.
$tdupdate 40 date/time of last drawing update.
$tdusrtimer 40 user elapsed timer.
$textsize 40 default text height.
$textstyle 7 current text style name.
$thickness 40 current thickness set by elev command.
$tracewid 40 default trace width.
$ucsname 1 name of current ucs.
$ucsorg 10,20,30 origin of current ucs (in wcs).
8
(c) drawing interchange and file formats $ucsxdir 10,20,30 direction of current ucs's x axis
(in world coordinates).
$ucsydir 10,20,30 direction of current ucs's y axis
(in world coordinates).
$useri1 - 5 70 five integer variables intended for use by
third-party developers.
$userr1 - 5 40 five real variables intended for use by
third-party developers.
$usrtimer 70 0=timer off, 1=timer on.
$worldview 70 1=set ucs to wcs during dview/vpoint,
0=don't change ucsthe header variables listed below existed prior to autocad release 10 but
now have independent settings for each active viewport. they are not
output by dxfout unless system variable flatland is set to 1. dxfin honors
these variables when read from dxf files, but if a vport symbol table with
"*active" entries is present (as is true for any dxf file produced by
release 10 or higher), the values in the vport table entries will override
the values of these header variables.
$fastzoom 70 fast zoom enabled if nonzero.
$gridmode 70 grid mode on if nonzero.
$gridunit 10,20 grid x and y spacing.
$snapang 50 snap grid rotation angle.
$snapbase 10,20 snap/grid base point (in ucs).
$snapisopair 70 isometric plane: 0=left, 1=top, 2=right.
$snapmode 70 snap mode on if nonzero.
$snapstyle 70 snap style: 0=standard, 1=isometric.
$snapunit 10,20 snap grid x and y spacing.
$viewctr 10,20 xy center of current view on screen.
$viewdir 10,20,30 viewing direction (direction from target, in wcs).
$viewsize 40 height of view.
the date/time variables ($tdcreate and $tdupdate) are output as real num-
bers in the format:
.
the elapsed time variables ($tdindwg and $tdusrtimer) have a similar
format:
.
c.1.5.2 tables section
the tables section contains several tables, each of which in turn contains
a variable number of table entries. the order of the tables may change, but
the ltype table will always precede the layer table. each table is intro-
duced with a 0 group with the label "table". this is followed by a 2 group
identifying the particular table (vport, ltype, layer, style, view, ucs, or
dwgmgr) and a 70 group that specifies the maximum number of table entries
9
autocad reference manual
that may follow. the tables in a drawing may contain deleted items, but
these are not written to the dxf file. thus, fewer table entries may
follow the table header than are indicated by the 70 group, so don't use
the count in the 70 group as an index to read in the table. it is provided
so that your program to read dxf files can allocate an array in advance
large enough to hold all the table entries that follow.
following this header for each table are the table entries. each table
item consists of a 0 group identifying the item type (same as table name,
e.g., "ltype" or "layer"), a 2 group giving the name of the table entry, a
70 group specifying flags relevant to the table entry (defined for each
table below), and additional groups that give the value of the table entry.
the end of each table is indicated by a 0 group with the value "endtab".
if any table entry has bit value 64 set in its group 70 flags, the table
entry was referenced by at least one entity in the drawing the last time
the drawing editor was entered to edit this drawing. this "referenced"
flag is for the benefit of the purge command; it can be ignored by most
programs that read dxf files, and need not be set by programs that write
dxf files.
the following are the groups used for each type of table item. all groups
are present for each table item.
ltype 3 (descriptive text for linetype), 72 (alignment code), 73
(number of dash length items), 40 (total pattern length), 49
(dash length 1), 49 (dash length 2), . . . layer 62 (color number, negative if layer is off), 6 (linetype
name). the 1 bit is set in the 70 group flags if the layer is
frozen. style 40 (fixed text height; 0 if not fixed), 41 (width factor), 50
(obliquing angle), 71 (text generation flags), 42 (last height
used), 3 (primary font file name), 4 ("bigfont" file name;
blank if none). if the third bit (4) is set in the 70 group
flags, this is a vertically-oriented text style. a style table item is used to record shape file load requests
also. in this case the first bit (1) is set in the 70 group
flags and only the 3 group (shape file name) is meaningful
(all the other groups are output, however). the "text generation flags" are a bit-coded field with the
following bit meanings: flag bit value meaning
2 text is backwards (mirrored in x)
4 text is upside down (mirrored in y)
10
(c) drawing interchange and file formats view 40 and 41 (view height and width), 10 and 20 (view center
point), 11, 21, 31 (view direction from target, in wcs), 12,
22, 32 (target point, in wcs), 42 (lens length), 43 and 44
(front and back clipping planes-offsets from target point), 50
(twist angle), 71 view mode (see viewmode system variable
appendix a). ucs 10, 20, 30 (origin), 11, 21, 31 (x axis direction), 12, 22, 32
(y axis direction). all in world coordinates. vport 10 and 20 (lower left corner of viewport; 0.0 to 1.0), 11 and
21 (upper right corner), 12 and 22 (view center point), 13 and
23 (snap base point), 14 and 24 (snap spacing, x and y), 15
and 25 (grid spacing, x and y), 16, 26, 36 (view direction
from target point), 17, 27, 37 (view target point), 40 (view
height), 41 (viewport aspect ratio), 42 (lens length), 43 and
44 (front and back clipping planes; offsets from target
point), 50 (snap rotation angle), 51 (view twist angle), 71
(view mode; see viewmode system variable in appendix a), 72
(circle zoom percent), 73 (fast zoom setting), 74 (ucsicon
setting), 75 (snap on/off), 76 (grid on/off), 77 (snap style),
78 (snap isopair). the vport table is unique in that it may contain several
entries with the same name (indicating a multiple-viewport
configuration). the entries corresponding to the active view-
port configuration all have the name "*active". the first
such entry describes the current viewport. dwgmgr for future use. fields not yet defined.
c.1.5.3 blocks section
the blocks section of the dxf file contains all the block definitions.
this section contains the entities that make up the blocks used in the
drawing, including "anonymous" blocks generated by the hatch command and by
associative dimensioning. the format of the entities in this section is
identical to those in the entities section described below, so refer to
that section for details. all entities in the blocks section appear
between block and endblk entities. block and endblk entities appear only
in the blocks section. block definitions are never nested (that is, no
block or endblk entity ever appears within another block-endblk pair).
c.1.5.4 entities section
entity items appear in both the block and entities sections of the dxf
file. the appearance of entities in the two sections is identical, with
the exception that entities in the block section never have handles. the
following gives the format of each entity as it appears in the file. some
groups that define an entity always appear, and some are optional and
appear only if they differ from their default values. in the following
11
autocad reference manual
discussion, groups that always occur are given by their group number and
function, while optional groups are indicated by "-optional n" following
the group description. "n" is the default value if the group is omitted.
programs that read dxf files should not assume that the groups describing
an entity occur in the order given here. the end of the groups that make
up an entity is indicated by the next 0 group, beginning the next entity or
indicating the end of the section.
remember that a dxf file is a complete representation of the drawing data-
base, and that as autocad is further enhanced, new groups will be added to
entities to accommodate additional features. writing your dxf processing
program in a table-driven way, making no assumptions about the order of
groups in an entity, and ignoring any groups not presently defined, will
make it much easier to accommodate dxf files from future releases of
autocad.
each entity begins with a 0 group identifying the entity type. the names
used for the entities are given in the table that follows. every entity
contains an 8 group that gives the name of the layer on which the entity
resides. each entity may have elevation, thickness, linetype, or color
information associated with it. if handles are enabled, every entity has a
5 group containing its handle (as a string representing a hexadecimal
number). the following groups are included only if the entity has nonde-
fault values for these properties.
group code meaning
6 linetype name (if not "bylayer"). the special name "byblock"
indicates a floating linetype. 38 elevation (if nonzero). output only if system variable
flatland is 1. otherwise, z coordinates are supplied as
3x-groups as part of each of the entity's defining points. 39 thickness (if nonzero).
62 color number (if not "bylayer"). zero indicates the
"byblock" (floating) color. 210, these groups are included for each line, point, circle, shape,
220, text, arc, trace, solid, block reference, polyline, dimension,
230 attribute, and attribute definition entity if its extrusion
direction is not parallel to the world z axis. the indicate
the x, y, and z components of the entity's extrusion direction.the rest of the groups that make up an entity item are described below.
many of the entities include "flag" groups. these are integer codes (6x or
7x groups) that encode various pieces of information regarding the entity,
and are specific to the particular entity type. in the following descrip-
tions, the term "bit-coded" means that the flag contains various true/false
values coded as the sum of the bit values given. any bits not defined in
the following section should be ignored in these fields and set to zero
when constructing a dxf file.
12
(c) drawing interchange and file formats line 10, 20, 30 (start point), 11, 21, 31 (end point).
point 10, 20, 30 (point), 50 (angle of x axis for the ucs in effect
when the point was drawn -optional 0, for use when pdmode is
nonzero). circle 10, 20, 30 (center), 40 (radius).
arc 10, 20, 30 (center), 40 (radius), 50 (start angle), 51 (end
angle). trace four points defining the corners of the trace: (10, 20, 30),
(11, 21, 31), (12, 22, 32), and (13, 23, 33). solid four points defining the corners of the solid: (10, 20, 30),
(11, 21, 31), (12, 22, 32), and (13, 23, 33). if only three
points were entered (forming a triangular solid), the third
and fourth points will be the same. text 10, 20, 30 (insertion point), 40 (height), 1 (text value), 50
(rotation angle -optional 0), 41 (relative x scale factor
-optional 1), 51 (obliquing angle -optional 0), 7 (text style
name -optional "standard"), 71 (text generation flags
-optional 0), 72 (justification type -optional 0), 11, 21, 31
(alignment point -optional, appears only if 72 group is
present and nonzero). the "text generation flags" are a bit-coded field with mean-
ings as follows: flag bit value meaning
2 text is backwards (mirrored in x)
4 text is upside down (mirrored in y) the "justification type" value (not bit-coded) indicates the
text justification style used on this entity, as shown in the
following table. value meaning
0 text is left justified
1 text is centered along its baseline
2 text is right justified
3 text is aligned between two points (height varies)
4 text is "middle" (fully) centered
5 text is fit between two points (width varies) if the justification is anything other than 0 (left justi-
fied), 11, 21, and 31 groups will also appear in the entity
to specify the alignment point of the text (center, right-
most, or second alignment point). dxfout handles ascii control characters in text strings by
expanding the character into a "^" (caret) followed by the 13
autocad reference manual
appropriate letter. for example, an ascii control-g (bel,
decimal code 7) is output as "^g". if the text itself con-
tains a caret character, it is expanded to "^ " (caret,
space). dxfin performs the complementary conversion. shape 10, 20, 30 (insertion point), 40 (size), 2 (shape name), 50
(rotation angle -optional 0), 41 (relative x scale factor
-optional 1), 51 (obliquing angle -optional 0). block 2 (block name), 70 (block type flags), 10, 20, 30 (block base
point). appears only in blocks section. the "block type
flags" are bit-coded, with the following bit meanings: flag bit value meaning
1 this is an "anonymous" block generated by
hatching, associative dimensioning, or
other internal operations.
2 this block has attributes. endblk no groups. appears only in blocks section.
insert 66 ("attributes follow" flag -optional 0), 2 (block name),
10, 20, 30 (insertion point), 41 (x scale factor -optional
1), 42 (y scale factor -optional 1), 43 (z scale factor
-optional 1), 50 (rotation angle -optional 0), 70 and 71
(column and row counts -optional 1), 44 and 45 (column and
row spacing -optional 0). if the value of the "attributes follow" flag is 1, a series
of attribute (attrib) entities is expected to follow the
insert, terminated by a sequence end (seqend) entity. attdef 10, 20, 30 (text start), 40 (text height), 1 (default value,
see text above for handling of ascii control characters), 3
(prompt string), 2 (tag string), 70 (attribute flags), 73
(field length -optional 0), 50 (text rotation -optional 0),
41 (relative x scale factor -optional 1), 51 (obliquing angle
-optional 0), 7 (text style name -optional "standard"), 71
(text generation flags -optional 0, see text above), 72 (text
justification type -optional 0, see text above)), 11, 21, 31
(alignment point -optional, appears only if 72 group is
present and nonzero). the "attribute flags" are a bit-coded field in which the bits
have the following meanings: flag bit value meaning
1 attribute is invisible (does not display)
2 this is a constant attribute
4 verification is required on input of this
attribute.
8 attribute is preset (no prompt during
insertion)14
(c) drawing interchange and file formats attrib 10, 20, 30 (text start), 40 (text height), 1 (value, see text
above for handling of ascii control characters), 2 (attribute
tag), 70 (attribute flags; see attdef above), 73 (field
length -optional 0), 50 (text rotation -optional 0), 41 (rel-
ative x scale factor -optional 1), 51 (obliquing angle
-optional 0), 7 (text style name -optional "standard"), 71
(text generation flags -optional 0, see text above), 72 (text
justification type -optional 0, see text above), 11, 21, 31
(alignment point -optional, appears only if 72 group is
present and nonzero). polyline 66 ("vertices follow flag"), 70 (polyline flags), 40 (default
starting width), 41 (default ending width), 71 and 72 (poly-
gon mesh m and n vertex counts -optional 0), 73 and 74
(smooth surface m and n densities -optional 0), 75 (smooth
surface type -optional 0). the default widths apply to any
vertex that doesn't supply widths (see below). the "vertices follow" flag is always 1, indicating that a
series of vertex entities is expected to follow the polyline,
terminated by a sequence end (seqend) entity. the "polyline
flags" group is a bit-coded field with bits defined as fol-
lows: flag bit value meaning
1 this is a closed polyline (or a polygon
mesh closed in the m direction)
2 curve-fit vertices have been added
4 spline-fit vertices have been added
8 this is a 3d polyline
16 this is a 3d polygon mesh. group 75 indi-
cates the smooth surface type, as follows: 0 = no smooth surface fitted
5 = quadratic b-spline surface
6 = cubic b-spline surface
8 = bezier surface 32 the polygon mesh is closed in the n direc-
tion vertex 10, 20, 30 (location), 40 (starting width -optional, see
above), 41 (ending width -optional, see above), 42 (bulge),
70 (vertex flags), 50 (curve fit tangent direction
-optional). the bulge is the tangent of 1/4 the included
angle for an arc segment, made negative if the arc goes
clockwise from the start point to the end point; a bulge of 0
indicates a straight segment, and a bulge of 1 is a semicir-
cle. the meanings of the bit-coded "vertex flags" are shown
in the following table.
15
autocad reference manual
flag bit value meaning
1 extra vertex created by curve fitting
2 curve fit tangent defined for this vertex.
a curve fit tangent direction of 0 may be
omitted from the dxf output, but is signif-
icant if this bit is set.
4 unused (never set in dxf files)
8 spline vertex created by spline fitting
16 spline frame control point
32 3d polyline vertex
64 3d polygon mesh vertex seqend no fields. this entity marks the end of vertices (vertex
type name) for a polyline, or the end of attribute entities
(attrib type name) for an insert entity that has attributes
(indicated by 66 group present and nonzero in insert entity). 3dline 10, 20, 30 (start point), 11, 21, 31 (end point).
3dface four points defining the corners of the face: (10, 20, 30),
(11, 21, 31), (12, 22, 32), and (13, 23, 33). 70 (invisible
edge flags -optional 0). if only three points were entered
(forming a triangular face), the third and fourth points will
be the same. the meanings of the bit-coded "invisible edge
flags" are shown in the following table. flag bit value meaning
1 first edge is invisible
2 second edge is invisible
4 third edge is invisible
8 fourth edge is invisible dimension 2 (name of pseudo-block containing the current dimension pic-
ture), 10, 20, 30 (definition point for all dimension types),
11, 21, 31 (middle point of dimension text), 12, 22, 32
(insertion point for clones of a dimension (for baseline and
continue), 70 (dimension type; 0=rotated, horizontal, or ver-
tical; 1=aligned; 2=angular; 3=diameter; 4=radius - the value
128 is added to this field if the dimension text has been
positioned at a user-defined location rather than at the
default location), 1 (dimension text explicitly entered by
the user. if null, the dimension measurement is drawn as the
text. otherwise, this text is drawn (but if it includes the
sequence "<>", the dimension measurement is drawn in place of
the "<>")), 13, 23, 33 (definition point for linear and angu-
lar dimensions), 14, 24, 34 (definition point for linear and
angular dimensions), 15, 25, 35 (definition point for diame-
ter, radius, and angular dimensions), 16, 26, 36 (point
defining dimension arc for angular dimensions), 40 (leader
length for radius and diameter dimensions), 50 (angle of
rotated, horizontal, or vertical linear dimensions).
16
(c) drawing interchange and file formats in addition, all dimension types have an optional group (code
51) that indicates the "horizontal" direction for the dimen-
sion entity. this determines the orientation of dimension
text and dimension lines for horizontal, vertical and rotated
linear dimensions. the group value is the negative of the
ecs angle of the ucs x axis in effect when the dimension was
drawn. in other words, the x axis of the ucs in effect when
the dimension was drawn is always parallel to the xy plane
for the dimension's ecs, and the angle between the ucs x axis
and the ecs x axis is a single 2d angle. the value in group
51 is the angle from "horizontal" (the effective x axis) to
the ecs x axis. entity coordinate systems (ecs) are
described later in this section. for all dimension types, the following groups represent 3d
wcs points, regardless of the flatland setting. 10, 20, 30
13, 23, 33
14, 24, 34
15, 25, 35 for all dimension types, the following groups represent ecs
points, and are 2d or 3d depending on the flatland setting. 11, 21(, 31)
12, 22(, 32)
16, 26(, 36) linear (13,23,33) the point used to specify the first extension line.
(14,24,34) the point used to specify the second extension line.
(10,20,30) the point used to specify the dimension line. angular (13,23,33) and (14,24,34) the endpoints of the first line
(10,20,30) and (15,25,35) the endpoints of the second line
(16,26,36) the point used to specify the dimen-
sion line arc diameter (15,25,35) the point used to pick the circle/arc to dimension
(10,20,30) the point on that circle directly across from the
pick point. radius (15,25,35) the point used to pick the circle/arc to dimension
(10,20,30) the center of that circle.
17
autocad reference manual
entity coordinate systems (ecs)
to save space in the drawing database (and in the dxf file), the points
associated with each entity are expressed in terms of its own entity coor-
dinate system (ecs). the entity coordinate system allows autocad to use a
much more compact means of representation for entities. with ecs, the only
additional information needed to describe its position in 3d space is the
3d vector describing the z axis of the ecs, and the elevation value.
for a given z axis (or extrusion) direction, there is an infinite number of
coordinate systems, defined by translating the origin in 3d space and by
rotating the x and y axes around the z axis. however, for the same z axis
direction, there is only one entity coordinate system. it has the follow-
ing properties:
o its origin coincides with the wcs origin.
o the orientation of the x and y axes within the xy plane are calcu-
lated in an arbitrary, but consistent manner. autocad performs
this calculation using the "arbitrary axis" algorithm described
below.for some entities, the ecs is equivalent to the world coordinate system and
all points (dxf groups 10-37) are expressed in world coordinates. see the
following table.
entities notes
line, point, 3dface, 3d these entities do not lie in
polyline, 3d vertex, 3d a particular plane. all
mesh, 3d mesh vertex points are expressed in
world coordinates. of these
entities, only lines and
points can be extruded;
their extrusion direction can
differ from the world z axis. circle, arc, solid, trace, these entities are planar in
text, attrib, attdef, shape, nature. all points are
insert, 2d polyline, 2d expressed in entity coordi-
vertex nates. all these entities
can be extruded; their
extrusion direction can
differ from the world z axis. dimension some of a dimension's points are
expressed in wcs, and some in ecs. others the remaining entities have
no point data and their
coordinate systems are
therefore irrelevant.once autocad has established the ecs for a given entity, here's how it
works:
18
(c) drawing interchange and file formats o the elevation value stored with an entity indicates how far along
the z axis to shift the xy plane from the wcs origin to make it
coincide with the plane that the entity is in. how much of this
is the user-defined elevation is unimportant.
o any 2d points describing the entity that were entered through the
ucs are transformed into the corresponding 2d points in the ecs,
which (more often than not) is shifted and rotated with respect to
the ucs.a few ramifications of this process are:
o you can not reliably find out what ucs was in effect when an
entity was acquired. you can only find out where the entity is in
the current ucs if the current ucs has the same z axis direction
as the original ucs (i.e., they both reduce to the same ecs).
o when you enter the xy coordinates of an entity in a given ucs and
then do a dxfout, you probably won't recognize those xy coordi-
nates in the dxf file. you'll have to know the method by which
autocad calculates the x and y axes in order to work with these
values.
o the elevation value stored with an entity and output in dxf files
will be a sum of the z coordinate difference between the ucs xy
plane and the ecs xy plane, and the elevation value that the user
specified at the time the entity was drawn.
arbitrary axis algorithm
the arbitrary axis algorithm is used by autocad internally to implement the
"arbitrary but consistent" generation of entity coordinate systems for all
entities except lines, points, 3d faces, and 3d polylines, which contain
points in world coordinates.
given a unit-length vector to be used as the z axis of a coordinate system,
the arbitrary axis algorithm generates a corresponding x axis for the coor-
dinate system. the y axis follows by application of the right hand rule.
the method is to examine the given z axis (also called the normal vector)
and see if it is close to the positive or negative world z axis. if it is,
cross the world y axis with the given z axis to arrive at the arbitrary x
axis. if not, cross the world z axis with the given z axis to arrive at
the arbitrary x axis. the boundary at which the decision is made was
chosen to be both inexpensive to calculate and completely portable across
machines. this is achieved by having a sort of "square" polar cap, the
bounds of which is 1/64, which is precisely specifiable in 6 decimal frac-
tion digits and in 6 binary fraction bits.
in mathematical terms, the algorithm does the following (all "vectors" are
assumed to be in 3d space, specified in the world coordinate system).
let the given normal vector be called n.
let the world y axis be called wy, which is always (0,1,0).
let the world z axis be called wz, which is always (0,0,1). 19
autocad reference manual
we are looking for the arbitrary x and y axes to go with the normal n.
they'll be called ax and ay. n could also be called az (the arbitrary z
axis).
if (nx < 1/64) and (ny < 1/64) then
ax = wy * n (where "*" is the cross-product operator).
otherwise,
ax = wz * n.
scale ax to unit length.the method of getting the ay vector would be:
ay = n * ax.
scale ay to unit length.
c.1.6 writing dxf interface programs
writing a program that communicates with autocad via the dxf mechanism
often appears far more difficult than it really is. the dxf file contains
a seemingly overwhelming amount of information, and examining a dxf file
manually may lead to the conclusion that the task is hopeless.
however, the dxf file has been designed to be easy to process by program,
not manually. the format was constructed with the deliberate intention of
making it easy to ignore information you don't care about while easily
reading the information you need. just remember to handle the groups in
any order and ignore any group you don't care about, and you'll be home
free.
as an example, the following is a microsoft basic program that reads a dxf
file and extracts all the line entities from the drawing (ignoring lines
that appear inside blocks). it prints the endpoints of these lines on the
screen. as an exercise you might try entering this program into your com-
puter, running it on a dxf file from one of your drawings, then enhancing
it to print the center point and radius of any circles it encounters. this
program is not put forward as an example of clean programming technique nor
the way a general dxf processor should be written; it is presented as an
example of just how simple a dxf-reading program can be.
1000 rem
1010 rem extract lines from dxf file
1020 rem
1030 g1% = 0
1040 line input "dxf file name: "; a$
1050 open "i", 1, a$ + ".dxf"
1060 rem
1070 rem ignore until section start encountered
1080 rem
1090 gosub 2000
1100 if g% <> 0 then 1090
1110 if s$ <> "section" then 1090
20
(c) drawing interchange and file formats 1120 gosub 2000
1130 rem
1140 rem skip unless entities section
1150 rem
1160 if s$ <> "entities" then 1090
1170 rem
1180 rem scan until end of section, processing lines
1190 rem
1200 gosub 2000
1210 if g% = 0 and s$ = "endsec" then 2200
1220 if g% = 0 and s$ = "line" then gosub 1400 : goto 1210
1230 goto 1200
1400 rem
1410 rem accumulate line entity groups
1420 rem
1430 gosub 2000
1440 if g% = 10 then x1 = x : y1 = y : z1 = z
1450 if g% = 11 then x2 = x : y2 = y : z2 = z
1460 if g% = 0 then print "line from (";x1;",";y1;",";z1;") to (";x2;
",";y2;",";z2;")
1470 goto 1430
2000 rem
2010 rem read group code and following value
2020 rem for x coordinates, read y and possibly z also
2030 rem
2040 if g1% < 0 then g% = -g1% : g1% = 0 else input #1, g%
2050 if g% < 10 or g% = 999 then line input #1, s$ : return
2060 if g% >= 38 and g% <= 49 then input #1, v : return
2080 if g% >= 50 and g% <= 59 then input #1, a : return
2090 if g% >= 60 and g% <= 69 then input #1, p% : return
2100 if g% >= 70 and g% <= 79 then input #1, f% : return
2110 if g% >= 210 and g% <= 219 then 2130
2120 if g% >= 20 then print "invalid group code";g% : stop
2130 input #1, x
2140 input #1, g1%
2150 if g1% <> (g%+10) then print "invalid y coord code";g1% : stop
2160 input #1, y
2170 input #1, g1%
2180 if g1% <> (g%+20) then g1% = -g1% else input #1, z
2190 return
2200 close 1writing a program that constructs a dxf file is more difficult, because you
must maintain consistency within the drawing in order for autocad to find
it acceptable. autocad allows you to omit many items in a dxf file and
still obtain a usable drawing. the entire header section can be omitted if
you don't need to set any header variables. any of the tables in the
tables section can be omitted if you don't need to make any entries, and in
fact the entire tables section can be dropped if nothing in it is required.
if you define any linetypes in the ltype table, this table must appear
before the layer table. if no block definitions are used in the drawing,
the blocks section can be omitted. if present, however, it must appear
before the entities section. within the entities section, you can refer-
ence layer names even though you haven't defined them in the layer table.
21
autocad reference manual
such layers will be automatically created with color 7 and the continuous
linetype. the eof item must be present at the end of file.
the following microsoft basic program constructs a dxf file representing a
polygon with a specified number of sides, leftmost origin point, and side
length. this program supplies only the entities section of the dxf file,
and places all entities generated on the default layer "0". this may be
taken as an example of a minimum dxf generation program. since this pro-
gram doesn't create the drawing header, the drawing limits, extents, and
current view will be invalid after performing a dxfin on the drawing gener-
ated by this program. you can do a "zoom e" to fill the screen with the
drawing generated. then adjust the limits manually.
1000 rem
1010 rem polygon generator
1020 rem
1030 line input "drawing (dxf) file name: "; a$
1040 open "o", 1, a$ + ".dxf"
1050 print #1, 0
1060 print #1, "section"
1070 print #1, 2
1080 print #1, "entities"
1090 pi = atn(1) * 4
1100 input "number of sides for polygon: "; s%
1110 input "starting point (x,y): "; x, y
1120 input "polygon side: "; d
1130 a1 = (2 * pi) / s%
1140 a = pi / 2
1150 for i% = 1 to s%
1160 print #1, 0
1170 print #1, "line"
1180 print #1, 8
1190 print #1, "0"
1200 print #1, 10
1210 print #1, x
1220 print #1, 20
1230 print #1, y
1240 print #1, 30
1250 print #1, 0.0
1260 nx = d * cos(a) + x
1270 ny = d * sin(a) + y
1280 print #1, 11
1290 print #1, nx
1300 print #1, 21
1310 print #1, ny
1320 print #1, 31
1330 print #1, 0.0
1340 x = nx
1350 y = ny
1360 a = a + a1
1370 next i%
1380 print #1, 0
1390 print #1, "endsec"22
(c) drawing interchange and file formats 1400 print #1, 0
1410 print #1, "eof"
1420 close 1the dxfin command is relatively forgiving with respect to the format of
data items. as long as a properly formatted item appears on the line on
which the data is expected, dxfin will accept it (of course, string items
should not have leading spaces unless these are intended to be part of the
string). the above program takes advantage of this flexibility in input
format, and does not go to great effort to generate a file appearing
exactly like one generated by autocad.
in the case of error loading a dxf file using dxfin, autocad reports the
error with a message indicating the nature of the error detected and the
last line processed in the dxf file before the error was detected. this
may not be the line on which the error occurred, especially in the case of
such errors as omission of required groups.
c.2 binary drawing interchange files
the ascii dxf file format described in the preceding sections of this
appendix is a complete representation of an autocad drawing in an ascii
text form easily processed by other programs. in addition, autocad can
produce or read a binary form of the full dxf file, and accepts limited
input in another binary file format. these binary files are described in
the following sections.
c.2.1 binary dxf files
the dxfout command provides a "binary" option that writes binary dxf files.
such a file contains all of the information present in an ascii dxf file,
but in a much more compact form that takes, typically, 25% less file space
and can be read and written more quickly (typically 5 times faster) by
autocad. unlike ascii dxf files, which entail a trade-off between size and
floating-point accuracy, binary dxf files preserve all of the accuracy in
the drawing database. autocad release 10 is the first version to support
this form of dxf file; it cannot be read by older versions.
a binary dxf file begins with a 22-byte sentinel consisting of:
"autocad binary dxf"
following the sentinel are (group,value) pairs as in an ascii dxf file, but
represented in binary form. the group code is a single-byte binary value,
and the value that follows is one of the following:
o a two-byte integer with the least significant byte first and the
most significant byte last,
23
autocad reference manual
o an eight-byte ieee double precision floating-point number stored
with the least significant byte first and the most significant
byte last, or o an ascii string terminated by a zero (nul) byte.
the type of the datum following a group is determined from the group code
according to the same rules used in decoding ascii dxf files. translation
of angles to degrees, and dates to fractional julian date representation,
is performed for binary files as well as for ascii dxf files. the comment
group, 999, is not used in binary dxf files.
dxfout writes binary dxf files with the same file type (".dxf") as for
ascii dxf files. the dxfin command automatically recognizes a binary file
(by means of its sentinel string) and loads it. there is no need for you
to identify it as a binary file.if dxfin encounters an error in a binary dxf file, it reports the byte
address within the file where the error was detected.
c.3 binary drawing interchange (dxb) files
the dxf file formats described earlier in this appendix are complete repre-
sentations of an autocad drawing that can be written and read by autocad
and other programs. however, autoshade(tm) and programs executed via the
"external commands" facility (appendix b) often have a need to supply
simple geometric input to autocad. for these purposes, another file format
even more compact than the binary dxf format is supported. this format,
called dxb (for "drawing interchange binary") is limited in the entities it
can represent. furthermore, autocad has a command to read such files, but
no direct method of writing them. (the adi plotter driver can plot to a
file in dxb format.)
c.3.1 dxbin command
to load a dxb file produced by a program such as autoshade, enter the dxbin
command:
command: dxbin
dxb file:enter the name of the file you wish to load. don't include a file type;
".dxb" is assumed.
24
(c) drawing interchange and file formatsc.3.2 dxb file format
this information is for experienced programmers, and is subject to change
without notice.
the format of a dxb file is as follows:
header: "autocad dxb 1.0" cr lf ^z nul (19 bytes)
data: . . . zero or more data records . . .
terminator: nul (1 byte)each data record begins with a single byte giving its type, followed by
data items. the data items have various forms of representation and encod-
ing. in the descriptions below, each data item is prefixed with a letter
and a hyphen. the meaning of the letter codes is as follows:
w- 16-bit integer, byte reversed in the standard 8086 style (least
significant byte first, most significant byte second). f- ieee 64-bit floating-point value stored with lsb first, msb last
(as stored by an 8087). l- 32-bit integer with the bytes reversed 8086-style.
n- number which may be either a 16-bit integer or a floating-point
number depending on the most recent setting of the "number mode"
data item. the number mode defaults to 0, signifying integers. if
set to 1, all n- items will be read as floating-point. u- item which is either a 32-bit integer or a floating-point number
depending on the most recent number mode setting. if a 32-bit
integer, the value is scaled by multiplying it by 65536 (2^16). if
a floating-point value, no scaling is applied. a- item representing an angle. if number mode is integer, this is a
32-bit integer representing an angle in units of millionths of a
degree (range 0 to 360,000,000). if a floating-point number, rep-
resents degrees.in the following table, the lengths include the item-type byte and assume
the number mode is set to zero (integer mode). if number mode is floating-
point, add 6 bytes to the length for each n- item present and 4 bytes for
each a-, or u- item present.
item type code data items length
(decimal) (bytes)
line 1 n-fromx n-fromy 9
n-tox n-toy
point 2 n-x n-y 5
circle 3 n-ctrx n-ctry n-rad 7
arc 8 n-ctrx n-ctry n-rad 19
a-starta a-enda
continued ... 25
autocad reference manual
item type code data items length
(decimal) (bytes)
trace 9 n-x1 n-y1 n-x2 n-y2 17
n-x3 n-y3 n-x4 n-y4
solid 11 n-x1 n-y1 n-x2 n-y2 17
n-x3 n-y3 n-x4 n-y4
seqend 17 (none) 1
polyline 19 w-closureflag 3
vertex 20 n-x n-y 5
3dline 21 n-fromx n-fromy n-fromz 13
n-tox n-toy n-toz
3dface 22 n-x1 n-y1 n-z1 25
n-x2 n-y2 n-x2
n-x3 n-y3 n-z3
n-x4 n-y4 n-z4
scale factor 128 f-scalefac 9
new layer 129 "layername" nul "layername"
length + 2
line extension 130 n-tox n-toy 5
trace extension 131 n-x3 n-y3 n-x4 n-y4 9
block base 132 n-bx n-by 5
bulge 133 u-2h/d 5
width 134 n-startw n-endw 5
number mode 135 w-mode 3
new color 136 w-colornum 3
3dline extension 137 n-tox n-toy n-toz 7the line extension item extends the last line or line extension from its
"to" point to a new "to point". the trace extension item similarly extends
the last trace solid, or trace extension from its x3,y3-x4,y4 ending line
to a new x3,y3-x4,y4 line.
the scale factor is a floating-point value by which all integer coordinates
are multiplied to obtain the floating-point coordinates used by the actual
entities. the initial scale factor when a file is read is 1.0. the new
layer item will create a layer if none exists, giving it the same defaults
as the "layer new" command, and will set that layer as the current layer
for subsequent entities. at the end of the dxb file load, the layer in
effect before the command will be restored.
the block base item specifies the base (origin) point of a block being cre-
ated. the block base must be defined before the first entity record is
encountered. if dxb is not defining a block, this specification will be
ignored.
a polyline consists of straight segments of fixed width connecting the ver-
tices, except as overridden by the bulge and width items described below.
the closure flag should be 0 or 1; if it is 1, then there is an implicit
segment from the last vertex (immediately before the seqend) to the first
vertex.
a bulge item, encountered between two vertex items (or after the last
vertex of a closed polyline), indicates that the two vertices are connected
26
(c) drawing interchange and file formatsby an arc rather than a straight segment. if the line segment connecting
the vertices would have length d, and the perpendicular distance from the
midpoint of that segment to the arc is h, then the magnitude of the bulge
is (2 * h / d). the sign is negative if the arc from the first vertex to
the second is clockwise. a semicircle thus has a bulge of 1 (or -1). if
the number mode is 0 (integer), bulge items are scaled by 216. if the
number mode has been set to floating-point, then the floating-point value
supplied is just 2*h/d (not scaled).
the width item indicates the starting and ending widths of the segment
(straight or curved) connecting two vertices. this width stays in effect
until the next width item or the seqend. if there is a width item between
the polyline item and the first vertex, it is stored as a default width for
the polyline; this will save considerable database space if the polyline
has several segments of this width.
the number mode item controls the mode of items with types given in the
table above as n-, a-, or u-. if the value supplied is zero, these values
will be integers, otherwise floating-point. the storage and implicit scal-
ing conventions for these values in both modes are described above.
lines and 3dlines share the same cells to remember the last to-point, so
you shouldn't mix extension groups for the two entities without an initial
group before the extension. there is no "extension" group for 3dfaces, as
there's no obvious edge to extend from.
the "new color" group specifies the color for subsequent entities in the
dxb file. the "w-colornum" word argument is in the range from 0 to 256. 0
means color by block, 1-255 are the standard autocad colors, and 256 means
color by layer. a color outside the range from 0 to 256 sets the color
back to the current entity color (you can do this deliberately, and it can
be quite handy). the initial entity color of material added by dxbin is
the current entity color.
all points specified in the dxb file are interpreted in terms of the cur-
rent ucs at the time the dxbin command is executed.
c.3.3 writing dxb files
there is no direct autocad command to write a dxb file, but the special
"adi" plotter driver can write such a file. if you want to create a dxb
file from an autocad drawing, configure the "adi" plotter and select its
dxb file output option.
27
autocad reference manual
c.4 initial graphics exchange standard (iges) files
using the commands described in this section, you can instruct autocad to
read and write iges format interchange files.
note: the format of iges files and the mapping performed to translate
between autocad drawing information and iges are described in the separate
autocad / iges interface specifications document (one of the items supplied
when you return your autocad license registration card).
c.4.1 igesout command
you can generate an initial graphics exchange standard (iges) interchange
file from an existing autocad drawing by means of the drawing editor's
igesout command. the command format is:
command: igesout file name: (name or return)
the default name for the output file is the same as that of the current
drawing, but with a file type of ".igs". if you specify an explicit file
name without including a file type, ".igs" is assumed. if a file with the
same name already exists, it is deleted.
c.4.2 igesin command
an iges interchange file can be converted into an autocad drawing by means
of the igesin command. first enter the drawing editor using the "create
new drawing" task from the main menu. then issue the igesin command.
command: igesin file name: (name)
enter the name of the iges file to be loaded.
if a serious error is encountered, the input process is halted and an error
message is displayed reporting where the error was found. the partial
drawing is not discarded.
28
(c) drawing interchange and file formatsc.5 slide file format
autocad slide files are screen images written by the mslide command and
read by the vslide command. this section describes the format of slide
files, for the benefit of developers who wish to incorporate support for
autocad slides into their programs.
this information is for experienced programmers, and is subject to change
without notice.
the general format of a slide file is:
1. header (31 bytes)
2. one or more data records (variable length)
all coordinates and sizes written to the slide file reflect the graphics
area of the display device from which the slide was created, with point
(0,0) located at the lower left corner of the graphics area. for autocad
release 9 and later, the slide file header consists of the following
fields:
field bytes description
id string 17 "autocad slide" cr lf ^z nul
type indicator 1 currently set to 86 (decimal).
level indicator 1 currently set to 2.
high x dot 2 width of the graphics area - 1, in pixels.
high y dot 2 height of the graphics area - 1, in pixels.
aspect ratio 4 aspect ratio (horizontal size / vertical size in
inches) of the graphics area, scaled by
10,000,000. this value is always written with
the least significant byte first. hardware fill 2 either 0 or 2 (value is unimportant).
test number 2 a number (1234 hex) used to determine whether
all 2-byte values in this slide file were writ-
ten with the high byte first (as by intel
8086-family cpus) or the low byte first (as by
motorola 68000-family cpus).data records follow the header. each data record begins with a 2-byte
field whose high-order byte is the record type. the remainder of the
record may be composed of 1-byte or 2-byte fields, as described in the fol-
lowing table. to determine whether the 2-byte fields are written with the
high byte first or the low byte first, examine the test number field of the
header, described above.
29
autocad reference manual
record length meaning description
type (hex) (bytes)
00 - 7f 8 vector the from-x coordinate for an
ordinary vector. from-y, to-x,
and to-y follow in that order, as
2-byte values. the from point is
saved as the last point.
80 - fa - undefined reserved for future use.
fb 5 offset vector the low-order byte and the fol-
lowing three bytes specify the
endpoints (from-x, from-y, to-x,
to-y) of a vector, in terms of
offsets (-128 to +127) from the
saved last point. the adjusted
from point is saved as the last
point for use by subsequent vec-
tors.
fc 2 end of file the low-order byte is 00.
fd 6 solid fill the low-order byte is always
zero. the following two 2-byte
values specify the x and y coor-
dinates of one vertex of a poly-
gon to be solid-filled. 3 to 10
such records occur in sequence.
a solid fill record with a nega-
tive y coordinate indicates the
start or end of such a flood
sequence. in the start record,
the x coordinate indicates the
number of vertex records to
follow.
fe 3 common this is a vector starting at the
endpoint last point. the low-order byte
vector and the following byte specify
to-x and to-y in terms of offsets
(-128 to +127) from the saved
last point. the adjusted to
point is saved as the last point
for use by subsequent vectors.
ff 2 new color subsequent vectors are to be
drawn using the color number
indicated by the low-order byte.if a slide contains any vectors at all, a new color record will be the
first data record. the order of the vectors in a slide, and the order of
the endpoints of those vectors, may vary.
for example, the following is an annotated hex dump of a simple slide file
created on an ibm pc/at with an ibm enhanced graphics adapter. the slide
consists of a white diagonal line from the lower left corner to the upper
right corner of the graphics area, a green vertical line near the lower
left corner, and a small red rectangle at the lower left corner.
30
(c) drawing interchange and file formats 41 75 74 6f 43 41 id string ("autocad slide" cr lf ^z nul)
44 20 53 6c 69 64
65 0d 0a 1a 00
56 type indicator (86)
02 level indicator (2)
3c 02 high x dot (572)
24 01 high y dot (292)
0b 80 df 00 aspect ratio (14,647,307 / 10,000,000 = 1.46)
02 00 hardware fill (2)
34 12 test number (1234 hex)
07 ff new color (7 = white)
3c 02 24 01 00 00 00 00 vector from 572,292 to 0,0. 572,292 becomes
"last" point
03 ff new color (3 = green)
0f 00 32 00 0f 00 13 00 vector from 15,50 to 15,19. 15,50 becomes
"last" point
01 ff new color (1 = red)
12 fb e7 12 ce offset vector from 15+18,50-25 (33,25) to
15+18,50-50 (33,0). 33,25 becomes "last" point
df fe 00 common-endpoint vector from 33,25 to
33-33,25+0 (0,25). 0,25 becomes "last" point
00 fe e7 common-endpoint vector from (0,25) to
0+0,25-25 (0,0). 0,0 becomes "last" point
21 fe 00 common-endpoint vector from (0,0) to
0+33,0+0 (33,0). 33,0 becomes "last" point
00 fc end of file
old slide header
the slide format described above is that produced by autocad release 9 and
later, and is portable among all computers running autocad release 9 or
later. previous versions of autocad (as well as autoshade 1.0 and
autosketch 1.02) produce slides with a somewhat different header, as shown
below.
field bytes description
id string 17 "autocad slide" cr lf ^z nul
type indicator 1 86 (decimal).
level indicator 1 1 (old format).
high x dot 2 width of the graphics area - 1, in pixels.
high y dot 2 height of the graphics area - 1, in pixels.
aspect ratio 8 aspect ratio (horizontal size / vertical size in
inches) of the graphics area, written as a
floating-point number.
hardware fill 2 either 0 or 2 (value is unimportant).
filler byte 1 unusednote that the old-format header does not contain a test number field. the
floating-point aspect ratio value and all two-byte integers are written in
the native format of the cpu used to create the file (for 8086-family cpus,
ieee double-precision and low byte first). old-format slide files are not
portable across machine types, but they can be read by any version of autocad
running on the same cpu type as the cpu with which the slide was created.
31
autocad reference manual
c.6 slide library file format
this section describes the format of autocad slide libraries (release 9 and
later), for the benefit of developers who wish to incorporate support for
slide libraries into their programs.
this information is for experienced programmers, and is subject to change
without notice.
the general format of a slide library is:
1. header (32 bytes)
"autocad slide library 1.0" cr lf ^z nul nul nul nul
2. one or more slide directory entries (36 bytes each)
3. one or more slides (variable length)slide directory entries have the following format:
1. slide name (nul terminated) (32 bytes)
2. address of slide within library file (4 bytes)
the slide address is always written with the low byte first. each slide to
which the directory points is a complete slide file as described in the
previous section. the end of the slide directory is signified by an entry
with a null slide name (first byte is nul). a slide library may contain a
mixture of old-format and new-format slides.
返回
of autocad running on the same cpu type as the cpu with which the slide was created. 31
autocad reference manual
c.6 slide library file format
this section describes the format of autocad slide libraries (release 9 and
later), for the benefit of developers who wish to incorporate support for
slide libraries into their programs.
this information is for experienced programmers, and is subject to change
without notice.
the general format of a slide library is:
1. header (32 bytes)
"autocad slide library 1.0" cr lf ^z nul nul nul nul
2. one or more slide directory entries (36 bytes each)
3. one or more slides (variable length)slide directory entries have the following format:
1. slide name (nul terminated) (32 bytes)
2. address of slide within library file (4 bytes)
the slide address is always written with the low byte first. each slide to
which the directory points is a complete slide file as described in the
previous section. the end of the slide directory is signified by an entry
with a null slide name (first byte is nul). a slide library may contain a
mixture of old-format and new-format slides.
