# encoding: utf-8

import re

import sys

from unicodedata import normalize

from collections import namedtuple

from ..lib.cocoa import *

from plotdevice import DeviceError

from .typography import *

from .geometry import Transform, Region, Size, Point, Pair

from .colors import Color

from .bezier import Bezier

from .atoms import TransformMixin, ColorMixin, EffectsMixin, StyleMixin, FrameMixin, Grob

from ..util import _copy_attrs, trim_zeroes, numlike, ordered, XMLParser, Element, read

from ..lib import foundry

from . import _ns_context

_ctx = None

__all__ = ("Text", "LEFT", "RIGHT", "CENTER", "JUSTIFY",)

# text alignments

LEFT = "left"

RIGHT = "right"

CENTER = "center"

JUSTIFY = "justify"

_TEXT=dict(

left = NSLeftTextAlignment,

right = NSRightTextAlignment,

center = NSCenterTextAlignment,

justify = NSJustifiedTextAlignment

)

class Text(EffectsMixin, TransformMixin, FrameMixin, StyleMixin, Grob):

# from TransformMixin: transform transformmode translate() rotate() scale() skew() reset()

# from EffectsMixin: alpha blend shadow

# from FrameMixin: x y width height

# from StyleMixin: stylesheet fill _parse_style()

stateAttrs = ('_nodes', )

opts = ('str', 'xml', 'src')

def __init__(self, *args, **kwargs):

# assemble the NSMachinery

self._engine = NSLayoutManager.alloc().init()

self._engine.setUsesScreenFonts_(False)

self._engine.setUsesFontLeading_(False)

self._store = NSTextStorage.alloc().init()

self._store.addLayoutManager_(self._engine)

if args and isinstance(args[0], Text):

# create a parallel set of nstext objects when copying an existing Text

# then bail out immediately (ignoring any other args)

orig = args[0]

self.inherit(orig)

self._blocks = [TextBlock(self) for f in orig._blocks]

for src, dst in zip(orig._blocks, self._blocks):

dst.offset, dst.size = src.offset, src.size

self._store.appendAttributedString_(orig._store)

return

# let the various mixins have a crack at the kwargs

super(Text, self).__init__(**kwargs)

# create a text block to manage layout and glyph-drawing

self._blocks = [TextBlock(self)]

# maintain a lookup table of nodes within xml input

self._nodes = {}

# use the first positional arg if no xml/str/src kwarg was passed

if not any(o in kwargs for o in self.opts) and len(args):

kwargs['str'], args = str(args[0]), args[1:]

# merge in any numlike positional args to define bounds

if args:

try:

self._frame._parse(args)

except:

raise DeviceError("Expected an x/y point (and optional width/height) but got %r" % (args,))

# fontify the str/xml/src arg and store it in the TextBlock

self.append(**{k:v for k,v in kwargs.items() if k in self.opts})

def __repr__(self):

total = len(self.text)

displayed = sum(self.blocks[-1]._chars)

msg = 'chars=%i'%displayed

if displayed < total:

msg += ' overflow=%i'%(total-displayed)

if self.blocks[1:]:

msg += ' blocks=%i' % len(self.blocks)

if hasattr(self, 'folio'):

msg += ' idx=%i folio=%i'%(self.idx, self.folio)

return "Text(%s)" % msg

def append(self, txt=None, **kwargs):

"""Add a string to the end of the text run (with optional styling)

Usage:

txt.append(str, **kwargs) # add the string using included styling kwargs

txt.append(str="", **kwargs) # equivalent to first usage

txt.append(xml="", **kwargs) # parses xml for styling before rendering

txt.append(src="", **kwargs) # reads from the contents of a file or url

Keyword Args:

Accepts the same keyword arguments as the text() command. For any styling

parameters that are omitted the appended string will inherit the style of

the Text object it's being added to.

"""

is_xml = 'xml' in kwargs

txt = kwargs.pop('xml', kwargs.pop('str', txt))

src = kwargs.pop('src', None)

StyleMixin.validate(kwargs)

attrib_txt = None

if src is not None:

# fetch the url or file's contents as unicode

txt = read(src, format='txt')

is_xml = src.lower().endswith('.xml')

# try using the nsmagic parsing of HTML/RTF to build an attributed string

if not is_xml:

txt = normalize("NFC", txt)

txt_bytes = txt.encode('utf-8')

txt_opts = {'CharacterEncoding': NSUTF8StringEncoding}

txt, info, err = NSMutableAttributedString.alloc().initWithData_options_documentAttributes_error_(

NSData.dataWithBytes_length_(txt_bytes, len(txt_bytes)), txt_opts, None, None

)

# if the data got unpacked into anything more interesting than plain text,

# preserve its styling. otherwise fall through and style the txt val

if re.search(r'(html|rtf)$', info.get('UTI')):

attrib_txt = txt

if txt is not None and not attrib_txt:

# convert non-textual `str` args to strings

if not isinstance(txt, str) and not is_xml:

txt = repr(txt)

txt = normalize("NFC", txt)

# use the inherited baseline style but allow one-off overrides from kwargs

merged_style = self._font._spec

merged_style.update(fill=self._fillcolor)

merged_style.update(self._parse_style(**kwargs))

# if the text is xml, parse it an overlay any stylesheet entries that map to

# its tag names. otherwise apply the merged style to the entire string

if is_xml:

# find any tagged regions that need styling

parser = XMLParser(txt, offset=len(self.text))

# update our internal lookup table of nodes

for tag, elts in parser.nodes.items():

old_elts = self._nodes.get(tag, [])

self._nodes[tag] = old_elts + elts

# start building the display-string (with all the tags now removed)

attrib_txt = NSMutableAttributedString.alloc().initWithString_(parser.text)

# generate the proper `ns' font attrs for each unique cascade of xml tags

attrs = {seq:self._fontify(merged_style, *seq) for seq in sorted(parser.regions)}

# apply the attributes to the runs found by the parser

for cascade, runs in parser.regions.items():

style = attrs[cascade]

for rng in runs:

attrib_txt.setAttributes_range_(style, rng)

else:

# don't parse as xml, just apply the current font(), align(), and fill()

attrs = self._fontify(merged_style)

attrib_txt = NSMutableAttributedString.alloc().initWithString_attributes_(txt, attrs)

# ensure the very-first character of a Text is indented flush left. also watch for

# double-newlines at the edge of the existing string and the appended chars. grafs

# can suppress their indentation with a \b (a.k.a. \x08) at the beginning of the

# line and un-indented lead-grafs can force indentation by beginning with \t

pre_txt = self._store.string()

if not pre_txt or re.search(r'\n[\n\x08]$', pre_txt):

Text._dedent(attrib_txt)

elif pre_txt.endswith('\n'):

if re.match(r'\n[^\n]', attrib_txt.string()):

Text._dedent(attrib_txt, 1)

elif re.match(r'\x08', attrib_txt.string()):

Text._dedent(attrib_txt)

# ensure that any paragraph with more than one leading newline is indented flush-left

# (and let `\n\b` override auto-indentation)

for m in re.finditer(r'\n\x08|\n\n+[^\n]', attrib_txt.string()):

Text._dedent(attrib_txt, m.end()-1)

if attrib_txt:

# let the typesetter deal with the new substring

self._store.beginEditing()

self._store.appendAttributedString_(attrib_txt)

self._store.endEditing()

self._resized()

### NSAttributedString de/manglers ###

def _fontify(self, defaults, *styles):

"""Merge the named-styles and defaults in order and return nsattibutedstring attrs"""

# use the inherited context settings as a baseline spec

spec = dict(defaults)

# layer the styles to generate a final font and color

for tag in styles:

spec.update(self.stylesheet._styles.get(tag,{}))

# assign a font and color based on the coalesced spec

font = Font({k:v for k,v in spec.items() if k in Stylesheet.kwargs})

color = Color(spec.pop('fill')).copy()

# factor the relevant attrs into a paragraph style

graf = NSMutableParagraphStyle.alloc().init()

graf.setLineBreakMode_(NSLineBreakByWordWrapping)

graf.setAlignment_(_TEXT[spec['align']])

graf.setHyphenationFactor_(spec['hyphenate'])

eps = sys.float_info.epsilon

# force the typesetter to deal with real leading rather than `lineheight'

face_height = eps + font.size * (font._face.ascent - font._face.descent) / 1000.0

graf.setLineHeightMultiple_(spec['leading'] * font.size / face_height)

graf.setMaximumLineHeight_(font.size*spec['leading'] + eps)

# handle indentation, horizontal margins, and vertical graf spacing

indent = font.size * spec['indent']

tabs = abs(indent or font.size)

head, tail = map(self._to_px, spec['margin'])

top, bot = [font.size*font.leading*d for d in spec['spacing']]

graf.setTabStops_([])

graf.setDefaultTabInterval_(tabs)

graf.setParagraphSpacingBefore_(top)

graf.setParagraphSpacing_(bot)

graf.setTailIndent_(-tail)

if indent > 0:

graf.setFirstLineHeadIndent_(indent + head)

graf.setHeadIndent_(head)

else:

graf.setFirstLineHeadIndent_(head)

graf.setHeadIndent_(abs(indent) + head)

if not spec['tracking']:

# None means `kerning off entirely', 0 means `default letterspacing'

kern = 0 if spec['tracking'] is None else eps

else:

# convert the em-based tracking val to a point-based kerning val

kern = (spec['tracking'] * font.size)/1000.0

# build the dict of features for this combination of styles

return dict(NSFont=font._nsFont, PDColor=color, NSParagraphStyle=graf, NSKern=kern)

def _colorize(self):

"""Updates the TextStorage, rewriting Colors as rgb or cmyk NSColors based on the current output mode"""

at = 0

end = self._store.length()

while at < end:

clr, rng = self._store.attribute_atIndex_effectiveRange_('PDColor', at, None)

self._store.addAttribute_value_range_('NSColor', clr.nsColor, rng)

at += rng.length

@classmethod

def _dedent(cls, attrib_txt, idx=0, inherit=False):

"""Removes first-line paragraph indentation of at the given attributed-string index.

Sets the first-line indent equal to the subsequent-line's value (unless the first line

is outdented; in which case leave it alone).

Passing inherit=True will override this logic and set the first-indent equal to the

subsequent-indent without regard for its indent/outdent status.

Note that this method *modifies* the attrib_txt reference rather than returning a value.

"""

if not attrib_txt.length(): return

attrib_txt.beginEditing()

old_graf, _ = attrib_txt.attribute_atIndex_effectiveRange_("NSParagraphStyle", idx, None);

graf = old_graf.mutableCopy()

first, rest = graf.firstLineHeadIndent(), graf.headIndent()

if first > rest or inherit: # leave negative-indentations alone

graf.setFirstLineHeadIndent_(rest)

attrib_txt.addAttribute_value_range_("NSParagraphStyle", graf, (idx, 1))

attrib_txt.endEditing()

### flowing text into new Text objects or subsidiary TextBlocks ###

def overleaf(self):

"""Returns a Text object containing any characters that did not fit within this object's bounds.

If the entire string fits within the current object, returns None."""

seen = "".join(getattr(f, 'text') for f in self._blocks)

full = self.text

if full not in seen:

next_pg = self.copy()

# delete the first page's-worth of chars and update indices

nc = len(seen)

next_pg._store.beginEditing()

next_pg._store.deleteCharactersInRange_([0, nc])

next_pg._store.endEditing()

nodes = {}

for tag, elts in self._nodes.items():

nodes[tag] = [e._replace(start=e.start-nc, end=e.end-nc) for e in elts if e.end-nc > 0]

next_pg._nodes = nodes

# if the page-break is in the middle of a paragraph, preserve the first character's initial

# indentation (since otherwise it'll be treated as a `first' line of a new paragraph)

if not seen.endswith('\n'):

Text._dedent(next_pg._store, inherit=True)

return next_pg

def flow(self, columns=all, layout=None):

"""Add as many text blocks as necessary to fully lay out the string

When called without arguments, returns a generator that you can iterate through to

set the position and size of each block in turn (starting with the second). Each block

is initialized with the same dimensions as the previous block in the sequence.

The optional `columns` argument allows you to specify the maximum number of blocks

you'd like to have at the end of the process. Note that this count *includes* the

Text object's original block, so you'll iterate over columns-1 blocks during the flow.

The optional `layout` argument can be a reference to a function which takes a single

TextBlock argument. If present, your layout function will be called once for each

block added to the stream.

"""

# sanity-check the columns arg

columns = 1e4 if columns is all else int(columns or 1)

if columns <= 1:

# no iteration necessary in the single-block case

return list(self._reflow(columns))

if not layout:

return self._reflow(columns) # return the generator for iteration

map(layout, self._reflow(columns)) # apply the layout function to each block in sequence

def _reflow(self, count):

# wipe out any previously set blocks then keep adding new ones until

# the glyphs are fully laid out

while self._blocks[1:]:

self._blocks.pop()._eject()

block = self._blocks[0]

while len(self._blocks) < count and sum(block._glyphs) < self._engine.numberOfGlyphs():

block = TextBlock(block)

self._blocks.append(block)

yield block

### Layout geometry ###

@property

def frame(self):

"""Returns the bounding box in which the text will be laid out"""

box = Region()

for block in self._blocks:

box = box.union(block.frame)

return box

@property

def bounds(self):

"""Returns the size & position of the actual text (typically a subset of the bounds)"""

box = Region()

for block in self._blocks:

box = box.union(block.bounds)

return box

@property

def metrics(self):

"""Returns the size of the actual text (shorthand for Text.bounds.size)"""

return self.bounds.size

def _get_baseline(self):

"""Returns the Text object's baseline `origin point'"""

return Point(self.x, self.y)

def _set_baseline(self, baseline):

self.x, self.y = baseline

baseline = property(_get_baseline, _set_baseline)

### Searching for substrings (and their layout geometry) ###

def __getitem__(self, index):

"""Subscripting a Text using indices into its .text string returns a TextFragment"""

match = TextFragment(self)

if isinstance(index, slice):

match.start, match.end, _ = index.indices(len(self))

else:

if index < 0:

index += len(self)

if not 0 <= index < len(self):

raise IndexError

match.start, match.end = index, index+1

return match

def __len__(self):

return len(self.text)

def find(self, regex, matches=0):

"""Find all matching portions of the text string using regular expressions

Syntax:

txt.find(re.compile(r'...', re.I)) # match a regex object

txt.find(r'antidisest.*?ism') # match a pattern string

txt.find(r'foo (.*?) baz') # match the pattern and capture a sub-group

txt.find(r'the', 10) # find the first 10 occurrences of `the'

Args:

`regex` can be a pattern string or a regex object. Pattern strings without

any uppercase characters will be case-insensitively matched. Patterns with

mixed case will be case-sensitive. In addition, the re.DOTALL flag will be

passed by default (meaning r'.' will match any character, including newlines).

Compiled regexes can define their own flags.

`matches` optionally set the maximum number of results to be returned. If

omitted, find() will return a TextFragment object for every match that's

visible in one of the Text object's TextBlocks. Matches that lie in the

overflow beyond the Text's bounds can be included however: pass the `all`

keyword as the `matches` arg.

Returns:

a list of TextFragment objects

"""

if isinstance(regex, str):

flags = (re.I|re.S) if regex.lower()==regex else (re.S)

regex = re.compile(regex, flags)

if not hasattr(regex, 'pattern'):

nonregex = "Text.find() must be called with an re.compile'd pattern object or a regular expression string"

raise DeviceError(nonregex)

return self._seek(regex.finditer(self.text), matches)

def select(self, tag_name, matches=0):

"""Find all matching portions of the text string using regular expressions

Syntax:

txt.select('em')) # find all visible `em' tag regions

txt.select('p', all) # find every `p' tag, even in the overflow

Args:

`tag_name` is a string that corresponds to one of the element names you

used when calling text() or txt.append() with an `xml` argument. Note that

any tag-attributes you defined in the xml will be available through the

resulting TextFragment object's `attrs` property.

`matches` optionally set the maximum number of results to be returned. If

omitted, select() will return a TextFragment object for every match that's

visible in one of the Text object's TextBlocks. Matches that lie in the

overflow beyond the Text's bounds can be included however: pass the `all`

keyword as the `matches` arg.

Returns:

a list of TextFragment objects

"""

return self._seek(self._nodes.get(tag_name, []), matches)

def _seek(self, stream, limit):

found = []

for m in stream:

match = TextFragment(self, m)

if not match.blocks and limit is not all:

break

found.append(match)

if len(found) == limit:

break

return found

@property

def text(self):

"""Returns the unicode string being typeset"""

return self._store.string()

@property

def words(self):

"""Returns a TextFragment for each word in the text string (whitespace separated)"""

return [TextFragment(self, w) for w in self._store.words()]

@property

def paragraphs(self):

"""Returns a TextFragment for each `line' in the text string (newline separated)"""

return [TextFragment(self, w) for w in self._store.paragraphs()]

@property

def blocks(self):

"""Returns a list of one or more TextBlocks defining the bounding box for layout"""

return list(self._blocks)

@property

def lines(self):

"""Returns a list of TextFragments, one for each line across all child TextBlocks"""

return [TextFragment(self, slug) for slug in foundry.line_slugs(self)]

### Calculating dimensions & rendering ###

def _resized(self):

"""Ensure that the first TextBlock's bounds are kept in sync with the Text's.

Called by the FrameMixin when the width or size is reassigned."""

# start with the max w/h passed by the Text object

dims = self._frame.size

block = self._blocks[0]

# start at the maximal size before shrinking-to-fit

block.offset = (0,0)

block.size = (dims.w, dims.h)

# if the rect isn't fully specified, size it to fit

if not (dims.w and dims.h):

# compute the portion that's actually filled and add 1px of extra padding to the

# calculated width (b/c believe it or not usedRectForTextContainer is buggy...)

min_w, min_h = block.metrics

min_w += block._from_px(1)

# shift the offset if not left-aligned and drawing to a point

nudge = {RIGHT:min_w, CENTER:min_w/2.0}.get(block._alignment)

if nudge and dims.w is None:

block.x -= nudge

# shrink-to-fit any dims that were previously undefined

if not dims.w:

block.width = min_w

if not dims.h:

block.height = min_h

@property

def _headroom(self):

"""Returns the distance between the Text's origin and the top of its bounds box"""

if not self._store.length():

return 0

return self._blocks[0]._from_px(self._engine.locationForGlyphAtIndex_(0).y)

@property

def _flipped_transform(self):

"""Returns a Transform object that positions unflipped beziers returned by trace_text"""

xf = Transform()

xf.translate(self.x, self.y - self._headroom)

xf.scale(1.0,-1.0)

return xf

@property

def _screen_transform(self):

"""Returns the Transform object that will be used to draw the text block.

The transform incorporates the global context state but also accounts for

the column-width/height constraints set in the constructor. If the text

has been flowed to multiple TextBlocks, dimensions are calculated based on

the union of the various bounds boxes."""

# gather the relevant text metrics (and convert them from canvas- to pixel-units)

x, y = self._to_px(Point(self.x, self.y))

baseline = self._to_px(self._headroom)

# accumulate transformations in a fresh matrix

xf = Transform()

if self._transformmode == CENTER:

# calculate the (reversible) translation offset for centering (in px)

bounds = self._to_px(self.frame)

shift = bounds.origin + bounds.size/2.0 - (x, y-baseline)

nudge = Transform().translate(*shift)

xf.translate(x, y-baseline) # set the position before applying transforms

xf.prepend(nudge) # nudge the block to its center

xf.prepend(self.transform) # add context's CTM.

xf.prepend(nudge.inverse) # Move back to the real origin.

else:

xf.prepend(self.transform) # in CORNER mode simply apply the CTM

xf.translate(x, y-baseline) # then move to the baseline origin point

return xf

def _draw(self):

with _ns_context(): # save and restore the gstate

self._screen_transform.concat() # transform so text can be drawn at the origin

self._colorize() # convert from Color to NSColor using current output mode

with self.effects.applied(): # apply any blend/alpha/shadow effects

for block in self._blocks:

px_offset = self._to_px(block.offset)

self._engine.drawGlyphsForGlyphRange_atPoint_(block._glyphs, px_offset)

# debug: draw a grey background for the text's bounds

# NSColor.colorWithDeviceWhite_alpha_(0,.2).set()

# NSBezierPath.fillRect_(Region(block.offset, block.size))

@property

def path(self):

"""Traces the laid-out glyphs and returns them as a single Bezier object"""

# generate an unflipped bezier with all the glyphs

path = Bezier(foundry.trace_text(self))

path.inherit(self)

# set its center-rotation fulcrum based on the blocks' bounds rect

origin, size = self.frame

path._fulcrum = origin + size/2.0

# flip the assembled path and slide it into the proper x/y position

return self._flipped_transform.apply(path)

class TextFragment(object):

"""Represents a substring region within a Text object (via its `find` or `select` method)

Properties:

`start` and `end` - the character range of the match

`text` - the matched substring

`path` - a Bezier object with the glyphs from the matched range

`lines` - a list of one or more TextFragments describing line-breaking within the Match

`blocks` - a list of one or more TextBlocks that fully contain the Match

Additional properties when .select'ing an xml element:

`tag` - a string with the matched element's name

`attrs` - a dictionary with the element's attributes (if any)

`parents` - a tuple with the parent, grandparent, etc. tag names

Additional properties when .find'ing a regular expression:

`m` - a regular expression Match object

Additional methods when .find'ing a regular expression with 'captured' sub-groups

`group(idx)` - returns a TextFragment corresponding to the numbered or named group

`groups()` - returns a list of TextFragment objects corresponding to captures in the regex

`groupdict()` - returns a dictionary mapping captured group names to TextFragment objects

"""

def __init__(self, parent, match=None):

self._parent = parent

self.tag, self.attrs, self.parents = None, {}, ()

self.m = None

if isinstance(match, foundry.Slug): # via line_slugs()

self.start, self.end = match.span

self._slugs = [match]

elif hasattr(match, 'range'): # NSSubText

self.start, n = match.range()

self.end = self.start + n

elif isinstance(match, Element): # xml Element

for k,v in match.__dict__.items():

setattr(self, k, v)

elif isinstance(match, TextBlock): # TextBlock

self.start, n = match._chars

self.end = self.start + n

elif hasattr(match, 'span'): # re.Match

self.start, self.end = match.span()

self.m = match

elif isinstance(match, tuple): # regex sub-match

self.start, self.end, self._group = match

def __len__(self):

return self.end-self.start

def __repr__(self):

msg = []

if hasattr(self, '_group'):

fmt = "r'\\%i'" if numlike(self._group) else "r'P<%s>'"

msg.append(fmt % self._group)

try:

pat = self.m.re.pattern

if len(pat)>18:

pat = "%s..." % (pat[:15])

msg.append("r%s" % repr(pat).lstrip("u"))

except:

if self.tag:

msg.append("<%s>" % self.tag)

if self.attrs:

msg.append("attrs=%i" % len(self.attrs))

msg.append("start=%i" % self.start)

msg.append("len=%i" % (self.end-self.start))

return 'TextFragment(%s)' % (", ".join(msg))

### Contents ###

@property

def text(self):

return self._parent.text[self.start:self.end]

def group(self, index=0, *others):

"""Return subgroup(s) of the match by indices or names. Index 0 returns the entire match.

Works just like https://docs.python.org/2/library/re.html#re.MatchObject.group

but returns TextFragment objects rather than character strings

"""

self._is_regex('group')

subs = []

for idx in (index,) + others:

rng = self.m.span(idx) + (idx,)

subs.append(TextFragment(self._parent, rng) if rng[0]!=-1 else None)

return tuple(subs) if len(subs)>1 else subs[0]

def groups(self, default=None):

"""Return a tuple containing all the subgroups of the match, from 1.

The default argument is used for groups that did not participate in the match

Works just like https://docs.python.org/2/library/re.html#re.MatchObject.groups

but returns TextFragment objects rather than character strings

"""

self._is_regex('groups')

indices = list(range(1, len(self.m.regs)))

if not indices:

return ()

return tuple([m if m else default for m in self.group(*indices)])

def groupdict(self, default=None):

"""Return a dictionary containing all the named subgroups of the match,

keyed by the subgroup name. The default argument is used for groups

that did not participate in the match

Works just like https://docs.python.org/2/library/re.html#re.MatchObject.groupdict

but returns TextFragment objects rather than character strings

"""

self._is_regex('groupdict')

indices = self.m.groupdict().keys()

subs = [self.group(nm) for nm in indices]

return {nm:sub if sub else default for nm, sub in zip(indices, subs)}

def _is_regex(self, method):

# an assert to ensure the match supports the group* methods

if not self.m:

badmatch = '%s() can only be used with regex-based TextFragment objects (see Text.find)'

raise DeviceError(badmatch % method)

### Geometry ###

@property

def slugs(self):

"""A list of one or more line-fragment Slugs describing text layout within the match"""

if not hasattr(self, '_slugs'):

rng = (self.start, self.end-self.start)

self._slugs = foundry.line_slugs(self._parent, rng)

return self._slugs

@property

def lines(self):

"""A list of one or more TextFragments splitting the current object across line-breaks"""

return [TextFragment(self._parent, lf) for lf in self.slugs]

@property

def blocks(self):

"""The list of TextBlock objects that the match spans"""

rng = (self.start, self.end-self.start)

return foundry.text_blocks(self._parent, rng)

@property

def frame(self):

"""Returns the bounding box for the lines containing the match"""

box = Region()

for slug in self.slugs:

box = box.union(slug.frame)

return box

@property

def bounds(self):

"""Returns the bounding box of the matched characters"""

box = Region()

for slug in self.slugs:

box = box.union(slug.bounds)

return box

@property

def metrics(self):

"""The size of the rendered text"""

return self.bounds.size

@property

def baseline(self):

"""The origin Point of the first glyph in the match (or None if empty)"""

for slug in self.slugs:

return slug.baseline

return None

@property

def path(self):

"""Traces the laid-out glyphs and returns them as a single Bezier object"""

# generate an unflipped bezier with all the glyphs

path = Bezier(foundry.trace_text(self._parent, (self.start, len(self))))

path.inherit(self._parent)

# set its center-rotation fulcrum based on the blocks' bounds rect

origin, size = self._parent.frame

path._fulcrum = origin + size/2.0

# flip the assembled path and slide it into the proper x/y position

return self._parent._flipped_transform.apply(path)

class TextBlock(FrameMixin, Grob):

"""Defines a layout region for a Text object's typesetter.

Most Text objects have a single TextBlock which holds the width

and height of the layout region. You don't need to deal with it

directly since you can just set the x/y/w/h attributes on the Text

object itself.

You can create a multi-column layout by iterating over a Text

object's .flow() method and manipulating the TextBlocks it returns.

You can also inspect the existing TextBlocks without adding new ones

through the Text object's `blocks` property.

Read/Write Properties:

`offset` - a Point with the block's position relative to the parent Text's.

`size` - a Size with the maximum width & height of the layout region

`x`,`y`,`w`,`h` - shorthand accessors for offset & size components

Readable Properties:

`text` - the substring that is visible in the block

`idx` - a counter marking the block's place in the sequence

`metrics` - the size of the used portion of the block's w & h

`lines` - a list of LineFragments contained in the block

`path` - a Bezier object with all the visible glyphs in the block

"""

def __init__(self, parent):

# inherit the canvas-unit methods and a _frame

self._frame = Region((0,0), (None,None))

self.inherit()

# create a new container

self._block = NSTextContainer.alloc().init()

self._block.setLineFragmentPadding_(0)

if isinstance(parent, TextBlock):

# either piggyback on an existing block...

self._parent = parent._parent

self.offset, self.size = parent.offset, parent.size

else:

# ... or become the first block of a parent Text object

self._parent = parent

# add ourselves to the layout flow

self._parent._engine.addTextContainer_(self._block)

@trim_zeroes

def __repr__(self):

return "TextBlock(%r, %r)"%(tuple(self.offset), tuple(self.size))

@property

def idx(self):

"""An integer marking this block's place in the flow sequence"""

return self._parent._engine.textContainers().index(self._block)

@property

def text(self):

"""The portion of the parent Text object's string that is visible in this block"""

return self._parent._store.string().substringWithRange_(self._chars)

@property

def frame(self):

"""The position & size of the block in canvas coordinates"""

bbox = Region(self.offset, self.size)

bbox.origin += self._parent.baseline

bbox.y -= self._from_px(self._headroom)

return bbox

@property

def bounds(self):

"""The position & size of the block's text in canvas coordinates"""

self._parent._engine.glyphRangeForTextContainer_(self._block) # force layout & glyph gen

origin, size = self._parent._engine.usedRectForTextContainer_(self._block)

origin.y -= self._headroom # adjust for the ascent above baseline

origin += self.offset + self._parent.baseline

return self._from_px(Region(origin, size))

@property

def metrics(self):

"""The size of the rendered text"""

return self.bounds.size

@property

def lines(self):

"""A list of TextFragments describing the line-layout within the block"""

slugs = foundry.line_slugs(self._parent, self._chars)

return [TextFragment(self._parent, slug) for slug in slugs]

@property

def path(self):

"""Traces the laid-out glyphs and returns them as a single Bezier object"""

return TextFragment(self._parent, self).path

@property

def _headroom(self):

if not self._parent.text:

return self._parent._font.ascender

fnt, _ = self._parent._store.attribute_atIndex_effectiveRange_("NSFont", self._chars.location, None);

if not fnt:

return self._parent._font.ascender

return fnt.ascender()

def _eject(self):

idx = self._parent._engine.textContainers().index(self._block)

self._parent._engine.removeTextContainerAtIndex_(idx)

self._parent = None

def _resized(self):

# called by the FrameMixin when the w or h changed

dims = [d or self._from_px(10000000) for d in self._frame.size]

self._block.setContainerSize_(self._to_px(Size(*dims)))

def _get_offset(self):

return Point(self._frame.origin)

def _set_offset(self, dims):

if numlike(dims):

dims = [dims]*2

self._frame.origin = dims

offset = property(_get_offset, _set_offset)

def _get_size(self):

return self._from_px(self._block.containerSize())

def _set_size(self, dims):

if dims != self._frame.size:

self._frame.size = dims

self._resized()

size = property(_get_size, _set_size)

@property

def _alignment(self):

if not self._parent._store.string():

return LEFT

graf, _ = self._parent._store.attribute_atIndex_effectiveRange_("NSParagraphStyle", 0, None)

return {_TEXT[a]:a for a in _TEXT}.get(graf.alignment(), LEFT)

@property

def _glyphs(self):

# NSRange of glyphs in the block

return self._parent._engine.glyphRangeForTextContainer_(self._block)

@property

def _chars(self):

# NSRange of chars in the block

rng, _ = self._parent._engine.characterRangeForGlyphRange_actualGlyphRange_(self._glyphs, None)

return rng

def draw(self):

# we inherit from Grob for the methods, not drawability

codependent = "TextBlocks can't be drawn directly; plot() the parent Text object instead"

raise DeviceError(codependent)